Analysis

Social_Research_Methods_----_PART_1_THE_FOUNDATIONS_OF_GOOD_SOCIAL_SCIENCE_RESEARCH.pdf

Home >Social Science homework help >Sociology homework help >Analysis

The Foundations of Good Social Science Research Social research is an exciting, active endeavour. For me, the thrill of seeing a set of my data for the fi rst time is right up there on the list of life’s good moments. For most new social researchers, and a good many old ones, our enthusiasm at the start of a new research project can be almost overwhelming. Th e impetus to get out there and start the real research, that is, collecting and analysing our data, is very strong.

But social science research is a process and a practice that involves much more than data collection and analysis. To achieve the results we want and to ensure that our fi rst forays into our data are thrilling, not devastating, all elements of that process and practice need to be fully understood and embedded into the research project in a coherent and structured way. As is made clear in Chapters 2 and 3, in most projects, data collection and analysis will come towards the end of the research project, not at the start. Good quality data, of whatever type, can only be collected from a strong foundation. Th is is where most of the real research work of any project is actually done.

Th is foundation includes a solid theoretical and personal understanding of social science research process, a dedication to thorough preparation and planning within that process, an awareness and a commitment to addressing the moral and ethical dimensions of our research, an extensive and critical review of the empirical and theoretical literature around our topic, and ensuring that we have the technical and theoretical expertise to undertake all aspects of the research process in a competent and rigorous way. While fi gurative outlines of this process tend to be linked linear boxes, I prefer a more metaphorical vision of these aspects as a nest, supporting, enveloping and nurturing the data and their analyses and interpretations. It is these foundational elements of social science research that are covered in this fi rst section.

Th e opening fi ve chapters set the context for the next eleven. Each details a diff erent core element of good social science research. Chapters 1 to 4, on the nature, design, and process of social science research are individually focused, but work together to set out, in practical terms, the whats, whys, and hows of developing a quality social science research project, one that is methodologically clear, theoretically informed and empirically and conceptually rigorous. It is at this developmental stage that the ethical and moral dimensions of the research must be addressed. Th e foregrounding of what makes research practice ethical in this fi rst section is purposeful. Because most of our research concerns people, ethics sits at the core of social science research. Ethical research goes further than ensuring that we do not expose our respondents, wittingly or unwittingly, to harm. Ethical research is research that is also diligent in its accuracy of results, acknowledgment of sources and the inputs of others. Finally in this section, Chapter 5 addresses the key, all-research method: theoretical, empirical and technical skills needed for quality and robust sampling and measuring.

For impatient personalities, such as myself, the large body of preparatory and foundational work required before the data phase can be daunting. But I also know that such work is critical to good research, and that there is no point doing research that is not good. To remind myself to always take the time to ensure thoroughness of process and

1PART

01_WAL_SRM3_20170_TXT_SI.indd 101_WAL_SRM3_20170_TXT_SI.indd 1 7/08/13 8:52 AM7/08/13 8:52 AM

Walter, Maggie. Social Research Methods, Oxford University Press, 2014. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/uow/detail.action?docID=5449900. Created from uow on 2022-12-13 10:14:20.

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

practice I have framed the following anonymous maxim on the importance of planning. It reads:

Th e really nice thing about not planning is that failure comes as a complete surprise and is not preceded by long periods of worry and depression.

As you get started on your journey into social research, remember that, while social research is exciting, even adventurous, it is also an expensive undertaking, in time, money and personal commitment, from ourselves, our institution and, more importantly, our respondents. We cannot just do it again if we make a mistake. So, engage with these fi rst fi ve chapters to ensure that you are never surprised.

2 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

01_WAL_SRM3_20170_TXT_SI.indd 201_WAL_SRM3_20170_TXT_SI.indd 2 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

The Nature of Social Science Research MAGGIE WALTER

ASKING SOCIAL QUESTIONS, SEEKING SOCIAL ANSWERS Despite enormous social and political changes in the way that Australian women and

men engage with the labour market, the caring occupations remain largely feminised.

Researchers continue to find that caring and femininity remain culturally linked in

Australian society, theorising that gendered meanings are strong explanators for why

female workers seek caring related employment in over-representative numbers. Kate

Huppatz (2010), while concurring with the importance of gender, queried whether

the research focus on gender was obscuring other differences between women in their

motivation to pursue this type of work. She asked: Is class also a significant factor in

the analysis of women’s participation in the paid caring field? Using Bourdieu’s concept

of habitus as her theoretical frame on social class, Huppatz interviewed thirty-nine

Australian women working or studying in the fields of nursing and social work, asking

each to self-identify as either working class or middle class. Huppatz chose these

particular professions as both are seen as middle class caring occupations. Her results

indicate that class location is associated with differing motivations for undertaking

caring work. For the middle class participants, the primary motivation was caring, and

these respondents reported that their families were generally supportive of their chosen

field of study or work. The women from working class backgrounds, while also citing a

strong interest in caring as an occupation, were also motivated by employment stability,

relatively high salaries and professional job social status. Moreover, many of these

women reported that their families had not understood their career or study decisions.

Huppatz concludes that ‘[E]conomic motivations, class identities and aspirations are

therefore as significant as gender in these women’s career choices’ (130: 2010).

KEY TERMS axiology

descriptive research

empirical data

epistemology

explanatory

research

exploratory

research

method

methodology

ontology

paradigm

quantitative–

qualitative debate

research question

scientifi c method

social aggregates

social meanings

social patterns

social theory

standpoint

triangulation

01_WAL_SRM3_20170_TXT_SI.indd 301_WAL_SRM3_20170_TXT_SI.indd 3 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

WHAT IS SOCIAL RESEARCH? Th e term ‘research’ evokes a popular imagery of a scholarly endeavour pursued using complicated formulas, and uninterpretable language and techniques. Research seems far removed from our everyday lives and our social world. But appearances here are deceptive. Social research makes the social world go around. Research, and especially social research, is everywhere, and it touches many aspects of our social lives. Essentially, social research is about investigating and seeking answers to the social questions that we and others ask about our social world. ‘Investigation’ is the key word here. To be good social researchers, we need to be keen social investigators, or even social sleuths. Th e constantly changing nature of our social world means that we will never run out of social questions to ask or social issues and phenomena to investigate. For example, results from the current Growing Up in Australia: Th e Longitudinal Study of Australian Children (AIFS 2002) conducted by the Department of Family and Community Services, is already adding immeasurably to our understanding of contemporary childhood in Australia. Th is research infl uences government policy decisions around families and children, and some of the results are publicly disseminated through the documentaries Life at One, Life at Th ree and Life at Seven, shown on ABC television in 2006, 2008, and 2012. Similarly, the Longitudinal Study of Indigenous Children, the Footprints in Time project, also currently underway, shows how Indigenous children in Australia grow up strong. Th ese two projects demonstrate the components of good social research: an important social question, a well thought-out and theoretically informed research plan, the use of appropriate research methods rigorously applied, valid analysis and interpretation, and broad dissemination of results and fi ndings.

RESEARCHING THE SOCIAL As social scientists, we compare ourselves directly with other scientists, oft en using many of the same methods and techniques. Yet researching the social world is oft en more complicated than researching the physical world. Social science research is research on, and with, real people in the real world, one of social research’s exciting elements. Th e social experience and understanding we bring to our research as members of our society are also important ingredients of the research process.

Th e social nature of our fi eld of study also means that much social research involves direct communication with our research respondents. Th is essential diff erence between social science and other science research, such as physics, biology or geology, is not given the emphasis it deserves. Eff ective people skills, that is, a genuine liking of interacting with others, ease in verbal and written communication, and, perhaps most importantly, listening skills, are vital but oft en underrated attributes for good social research. Here is a hypothetical example: Does it matter to an amoeba, or the research project, if the physical scientist investigating its properties cannot hold a coherent conversation with another human being to save herself? Probably not. Would the outcome be the same if a social science researcher were affl icted with the same defi cit in the social skills department? How eff ective would that researcher be in conducting an unstructured, indepth interview, facilitating a focus group, or even designing an eff ective survey? Crucially, not only can the lack of appropriate people skills reduce the value of data gathered by whatever social research method, but poor people skills can also jeopardise the social research project itself.

As shown in Box 1.1, the human facet of social research can act as both an aid and a barrier to social research. On the one hand, our personal lived experience combined with

Language: A performative activity encompassing words, texts and other expressive behaviours.

Social research: The systematic study of society, the patterns in it and the processes that shape what people do.

Method: The research technique or practice used to gather and analyse the research data.

Indepth interview: Indepth interviews are guided by general themes rather than preset questions. They are also less formal than structured interviews, and explore issues as the interviewee raises them.

Focus group: A research method that involves encouraging a group of people to discuss some social or political issue.

Data: The information we collect and analyse to answer our research question. Data come in all manner of forms, such as survey forms, documents and secondary data.

4 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 401_WAL_SRM3_20170_TXT_SI.indd 4 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

our education and training as social scientists enables us to bring to our research a complex understanding of our social world. On the other hand, the fact that we are enmeshed in our social world means that we can oft en fail to see the social and cultural assumptions that inform our own worldviews, through which we perceive social questions and social issues.

BOX 1.1: RESEARCHING THE SOCIAL: A COMPLEX PROCESS The human aspect of social research adds not only to the excitement but also to the complexity of our research endeavour.

ETHICS We cannot research people or societies the way we study inanimate objects such as minerals or energy waves, no matter how useful that might be to our research. As detailed in Chapter 4, ethical constraints, from a moral perspective and, increasingly, from formal ethics bodies such as human research ethics committees (HRECs) set boundaries and limitations on how we approach and undertake our research. These are important to protect our human subjects from us as researchers, and perhaps as enthusiastic social researchers, to protect us from ourselves.

HUMAN AMBIGUITY, IRRATIONALITY AND SOCIAL AWARENESS People and society are not always rational or predictable. The motives and rationales of people are not always clear, sometimes not even to themselves. This means that, although we can ask the questions, the answers we obtain from our respondents cannot necessarily be regarded as fact or unambiguous. Our social awareness also means that we are not always prepared to be frank in our discussions of our behaviour, attitudes and belief systems. We would not be surprised if, for example, a study of weight-loss program participants found a discrepancy between the self-reported eating behaviour of the respondents and the associated weekly weight loss. As researchers, we must recognise the essential subjectivity of much of our data, but not be paralysed by it.

SOCIAL AND PERSONAL CHANGE People and societies are not static. Social and personal change are the norm, not the exception, and this is an important consideration in social research. While our research might provide a plausible explanation for today’s social phenomena, this does not mean these same explanations can be directly applied to the social phenomena of tomorrow. Research on the career aspirations of married women in 1980, for example, would no doubt produce very different results to a similar study carried out in 2014.

CULTURAL FACTORS AND ASSUMPTIONS Cultural factors and assumptions, our own rather than those of our respondents, can operate to blind us to some social questions and to some social answers. Until very recently most social researchers operated under the unquestioned assumption that Western science and Western society were the norm. Inherent, but undeclared, in this were gendered and culturally exclusive perspectives. More recently, other ways of knowing, such as Indigenous and feminist research paradigms, have successfully challenged these assumptions and shown that there are other ways of being in and making meaning of the social world. As a result, social research has become a richer and more valid enterprise.

HAWTHORNE EFFECT The humanness of our research subjects can lead to particular social research dangers around validity. We need to be very careful to ensure that what we think we are measuring

Ethics: The establishment of a set of moral standards that govern behaviour in a particular setting or for a particular group.

Human research ethics committee (HREC): A committee established by an institution or organisation for the task of viewing research proposals and monitoring ongoing investigations with the aim of protecting the welfare and rights of participants in that research.

55CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 501_WAL_SRM3_20170_TXT_SI.indd 5 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

is what we are actually measuring. In some cases, research results can be affected by the subject’s interpretation of what the research is about—the Hawthorne effect. The Hawthorne effect was first identified in a study undertaken in the 1930s by Elton May at a Western Electrics plant in Hawthorne, Illinois, USA. The research was aimed at establishing if different independent working environments—related variables, such as lighting, length of meal breaks and how the workers were paid—would have an effect on the dependent variable, the workers’ productivity. The problem was that every independent variable had a positive effect on productivity (to the initial delight of the researchers), but so too did a return to the original working conditions. The researchers finally concluded that the workers at the plant were interested in the research, enjoyed participating, and so tried to ensure that the researchers achieved the effects they were looking for. As noted by the Hawthorne investigator, unlike inanimate objects, people tend to ‘notice that they are being studied and form feelings and attitudes about being studied, which may in turn influence the outcome of the research’ (Dooley 1990: 212).

THE COMPLICATING SOCIAL CONTEXT As social scientists, we use social theories to explain the phenomena we observe in the social world. This seems a fairly straightforward exercise, but we need to remember that social phenomena are not stand-alone events. Social phenomena are entwined within political and moral belief systems or ideologies, and this complexity leads to social research often having political and cultural dimensions. Social research that seeks to explore and explain rising rates of sole parenthood in Australia as a social phenomenon, for example, can clash with belief systems that view marriage and two-parent families as the only legitimate form of family. Alternatively, different ways of positioning, understanding, and interpreting a social phenomena (see Methodology) can result in different social researchers coming up with very different theoretical explanations or interpretations of the same topic, as shown in Case study 1.1.

CASE STUDY 1.1: WHO AND WHAT IS POOR? Social research into poverty in Australia has, in recent years, been mired in a sometimes acrimonious debate about how poverty should be measured. This argument is more than just a contestation about where the poverty line should be set or how poverty should be defined. Different ways of measuring poverty deliver very different results. As outlined below, the preferred models of different groups tend to reflect differing ideological stances on whether poverty is a growing problem in contemporary Australian society.

In 2001, a study commissioned by the Smith Family, Financial Disadvantage in Australia: 1990–2000, was published. The main findings were that • poverty had increased steadily from 11 to 13 per cent of the population • unemployment was the key generator of poverty • having a job was no longer always an effective protector against poverty, with the risk

of poverty for those in part-time jobs rising. This report’s findings were immediately challenged by the Centre for Independent

Studies (CIS). CIS argued that the study exaggerated the extent and nature of poverty in Australia. The major problem, argued CIS, was that the Smith Family study had used the mean (average) income rather than the less volatile median as the base for its poverty line measure. Also, CIS countered, the Smith Family study confused poverty and inequality. Rising inequality does not automatically increase poverty. In CIS’s own estimates, only around 5 per cent of the Australian population were living in ‘chronic’ poverty in 2000.

Source: Senate Community Affairs References Committee 2004: 35–40

Mean: The average score for a set of cases.

6 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 601_WAL_SRM3_20170_TXT_SI.indd 6 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

THE ‘SCIENCE’ IN SOCIAL SCIENCE Because our social world is all around us, it is oft en assumed that explaining social phenomena is just a matter of common sense. Yet being a member of a society defi nitely does not equate to an automatic understanding of our social reality. Indeed, as outlined in the previous paragraph on cultural factors and assumptions, being a member of society can, and does, act as an impediment to even asking some social questions. What sets social science apart from social commentary or opinion is our use of scientifi c method. Scientifi c method is traditionally defi ned as being about observation, classifi cation, and interpretation. As mathematician Karl Pearson (1900) stated: ‘Th e man who classifi es facts of any kind, who sees their mutual relation and describes their sequences, is applying the scientifi c fact and is a man of science’ (cited in Mann 1985: 19).

While today we would most defi nitely quarrel with Pearson’s presumption of a social scientist as male, and substitute the term ‘analysis’ for ‘classifi cation’, we recognise the essential sequence of tasks involved in social science research. Just as a geologist might explore rock formations, analyse what is found, and then theorise, based on this analysis, that the area was previously the site of volcanic activity, so social science researchers follow their social observations, whether they be hard quantitative data or soft er qualitative research material, with analysis and theoretical interpretations of those social phenomena. Social science research is a planned methodical activity built around a solid, well-formulated research design.

Scientifi c method is also about the way we conduct our research. Neuman explains that scientifi c method is not just one thing, but that it ‘refers to the ideas, rules, techniques and approaches that the scientifi c community uses’ (2004: 8). Th ese include professionalism, ethical integrity in how we go about the social research process, and ensuring that the social research we conduct is rigorous in method and techniques as well as transparent in research methods and interpretation. Th ese aspects of scientifi c method mean that we endeavour to conduct our research, through all its phases, in a professional manner that abides by ethical principles. ‘Transparency and rigour’ refer to making explicit, at all stages of the research, the specifi c research method we use, the reasons for our choice and how we use our data to develop our theory or interpretations. Th e strength of these standards on how we conduct our social science research is that they are shared. Acceptance of, and adherence to, these standards within our research practice is a central element of being an active social science researcher, and a core defi ning element that sets social science apart from everyday thinking or other ways of knowing about our social world.

As social scientists, one of the key ways we ensure professionalism, integrity and transparency is by making our research public. As expanded in Chapter 16, publication of our results, usually in a recognised journal, makes our research open to public scrutiny. As an additional safeguard, most published research is subjected to a peer review process, in which anonymous, to the authors at least, social scientists review the research before it is published.

SOCIAL PATTERNS AND SOCIAL MEANINGS Put into simple terms, our primary aim in social research is to identify, investigate, and seek to understand social patterns and social meanings. It is the persistent patterns in social life, as well as the social meanings inherent in these, that we are endeavouring to uncover.

Scientific method: Planned methodical research based around observing, analysing, and interpreting our research data, conducted with professionalism and ethical integrity, and transparent and rigorous in its approach.

Social patterns: Persistent patterns in social phenomena that occur repeatedly in the social world.

Social meanings: How people(s) make sense of aspects of their social lives and the understandings that they develop of these.

77CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 701_WAL_SRM3_20170_TXT_SI.indd 7 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Exploratory research: Research undertaken to explore or open up new areas of social enquiry.

By social patterns, we mean those phenomena that occur repeatedly in social life. For example, research data fi nding that Australians with strong religious beliefs are consistently underrepresented among those who reside in a cohabiting relationship (Dempsey & de Vaus 2003) demonstrate an enduring social pattern. By social meanings, we mean how people make sense of aspects of their social lives and the understandings they make of these. Natalier (2001), for example, interviewed motorcycle riders to try to develop an understanding of the social meanings of motorcycling risk. Her results suggest bike riders downplay their risk by aligning adverse events of others and themselves with lack of technique rather than inherent risk in riding a motorcycle.

In its analysis of social patterns and social meaning, social research also has a debunking role: to test the veracity and sometimes expose the inaccuracy of our everyday assumptions about our social world. When we test these beliefs empirically using scientifi c method, we oft en fi nd that the social reality and the social belief are not a good match. Health is a good example here. In Australia, which is generally perceived to be an essentially egalitarian and wealthy society, our health is perceived as an essentially individual aspect of our lives. Yet, analysis of health data shows that in Australia, as in other Western countries, health is not shared equally. Social research consistently fi nds that health status and socioeconomic status are strongly linked, and that there is a clear and widening health gap between low-income and higher-income groups. Th e poorer you are, the more likely you are to get sick and to die at a younger age; this applies right along the social gradient, rather than just to those at the extremes (Walker 2000).

‘But’, you’ll always hear someone say if you discuss the social gradient of health, ‘I know somebody who came from a very poor family who lived till 105 and was never sick a day in her life.’ Th e question here is whether an exception such as this challenges the social theory we have developed from our identifi cation of social patterns or meanings, as those pointing them out oft en assume they do. Th e answer is a huge no. Exceptions—and there are always exceptions—are not a threat to social science fi ndings nor, indeed, unexpected. Our interpretations or theories of social phenomena are not predicting what every single outcome for every single person within a society will be. Rather, as social scientists we deal with social aggregates, that is, the ‘collective actions of and situations of many individuals’ (Babbie 2002: 12), richer Australians, in aggregate, will have better health than poorer Australians, in aggregate. For social scientists, a single case, or even a group of cases, is just that until shown by rigorous analysis that a group of cases actually forms a social pattern. But once we have established that a social pattern exists, we need to look for a social explanation for that pattern, a theory of why and how this pattern manifests itself.

Social patterns also alter along with social life. If, for example, you are researching the family in Australia, one of the fi rst things you will fi nd is that the picture is one of continual change. Th e average age at fi rst marriage, the likely number of children a couple will have, and the social meanings that are ascribed to marriage and children have changed repeatedly over time. Critically, you can expect that such changes in social patterns and social meanings will continue.

SOCIAL RESEARCH: WHY DO WE DO IT? Why do we do social research? We do it because we want to know, and because knowing is important. For our social world to function well we need to understand it, and social research is the way we gain social understandings. Th e level of understandings we seek

Social theory: An idea or a set of ideas that explain social phenomena.

Social aggregates: The collective, aggregate social outcomes or circumstances of individuals or groups.

8 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 801_WAL_SRM3_20170_TXT_SI.indd 8 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

will vary according to our question. Th ree core levels of social research are commonly identifi ed (see, for example, Babbie 2002: 83–5, Glicken 2003: 14–15, Neuman 2004: 15).

Th ese are • exploratory research • descriptive research • explanatory research.

In reality, social research oft en does not fall neatly into one category of research or another. Rather, exploratory research can also be used to describe the social phenomena under investigation, and may also develop at least tentative explanations for what is found.

EXERCISE 1.1: EXPLORING ILLICIT DRUG CONSUMPTION AMONG UNIVERSITY STUDENTS

Consider the following hypothetical research scenario. You have been asked to be a research assistant in a project that is investigating the topic

of the level of illicit drug-taking among university students. The respondents will be adult (aged 18 years and above) university students attending the three universities in a certain capital city. Although still in the development phase, the initial aims of the project are • to examine illicit drug consumption behaviour • to explore the social meanings that individuals ascribe to their illicit drug consumption.

These data will be used to generate a picture of illicit drug consumption among the student body and form the baseline for comparison data gathered every four years. The research method for the project plans to use a two-phase data collection. In the first phase, a telephone survey will collect the data from a random representative sample of students. The three universities have provided access to their student database as a sampling frame, thereby providing access to a combined student population of 36 000. In the second phase, a series of indepth interviews will be undertaken with twenty-five individual respondents.

TASK Answer the following questions, either as individuals or in groups. 1 Which aspects of illicit drug consumption might you want to investigate? As an example,

you might want to determine the influence of peer pressure on drug-taking behaviour. List five dimensions of this social phenomenon that you think would be interesting to explore.

2 How would your own experience of illicit drug consumption (or lack of it) influence how you went about your task of determining what aspects of the topic to investigate?

3 Can you think of any immediate ethical concerns that might limit or constrain the way you conduct the research?

4 What factors around ambiguity and social awareness would you need to keep in mind when designing the study?

5 How might external events affect your study? Would a study that occurred before a series of items on high profile illicit drug consumption deaths on current affairs shows have different outcomes from one undertaken immediately following their screening?

THE LANGUAGE OF SOCIAL SCIENCE Like other scientifi c endeavours, social research uses a set of key terms and concepts. Many of these are unfamiliar or have meanings that diff er from their everyday usage, but they are not inherently diffi cult. Rather, they form part of the language of social research,

Descriptive research: Research that has as its major purpose to describe social phenomena.

Explanatory research: Research that seeks to provide or develop explanation of the social world or social phenomena.

Sample: A sample is a set of cases or elements that are selected from a population.

Ambiguity: Vague or imprecise terms that have more than one meaning.

99CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 901_WAL_SRM3_20170_TXT_SI.indd 9 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

and most are just shorthand for broader ideas. Th e meaning of commonly used social science terms and concepts are highlighted and defi ned throughout each chapter, and set out as a combined set in the book’s glossary.

METHOD AND METHODOLOGY: UNDERSTANDING THE DIFFERENCE Th e previous sections emphasised the scientifi c, methodical aspects of social science research. Our approach, as social scientists, is clear, scientifi c, and objective—or at least this is the way social science research is oft en presented. But if social science is neutral, how and why are some social research projects prioritised over others? And why do diff erent researchers interpret social phenomena so diff erently?

Our methodology is at least a part explanation of this complex and sometimes ambiguous terrain.

Th e distinction between method and methodology is an important one, and one that is oft en misunderstood. Very oft en, the term ‘methodology’ is used when people really mean the method. Understanding the diff erence between the two is an essential element of understanding social research. Put simply, method refers to a technique for gathering information, such as an interview, questionnaire, or documentary analysis, methodology is the worldview-infl uenced lens through which the research is understood, designed, and conducted. Our methodology includes our method, but the method is a component of our methodology and not even the most important.

Understanding the diff erence between a method and a methodology is important in understanding research for three reasons. 1 It enables us to see where values, theories, and worldviews interact with social

research. 2 It enables us to understand how specifi c methodologies emerge and why understanding

our methodology is vital to our research practice. 3 It enables us to view methods as tools, tried and tested ways, and techniques for

gathering our data, rather than the research itself.

WHAT GOES INTO MAKING A METHODOLOGY? Understanding methodology as the lens through which we view, undertake, and translate our research provides some level of explanation, but fails to adequately convey what a methodology actually is. Yes, our methodology has multiple components, method among them. But what else is included? How do we recognise a component? Th e specifi cs of methodology are less straightforward and more debated than method. My own defi nition is that methodology is the worldview lens through which the research question and the core concepts are viewed and translated into the research approach we take to the research. I include the following as core components. In practice, these elements are oft en inextricably entwined, but it is helpful to clarify each separately: • our standpoint • our theoretical conceptual framework and paradigm • our method.

My conceptualisation of methodology is detailed in Figure 1.1 below.

Method: The research technique or practice used to gather and analyse the research data.

Methodology: Methodology is the worldview through which the research is designed and conducted. It is comprised of our standpoint, theoretical and conceptual frame, and our method.

Research question: Research questions state the major aim of the research in question form, specifying the key idea that the research seeks to investigate and/or explain, it also identifies the key concepts of the research.

Conceptual framework: The theoretical frame that we use to conceptualise the collection, and to analyse and interpret our data.

Paradigm: A shared framework of viewing and approaching the investigation and research of social phenomena.

10 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1001_WAL_SRM3_20170_TXT_SI.indd 10 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

STANDPOINT INFLUENCED THEORETICAL FRAME

RESEARCH METHOD

SOCIAL POSITION Gender, race, culture,

Economic status

EPISTEMOLOGY Theory of knowledge

AXIOLOGY Theory of values

ONTOLOGY Theory of being

STANDPOINT

OUR STANDPOINT Our standpoint is the most important aspect in defi ning our methodology, because it infl uences all other components. But it is also an aspect that, in most social research, is poorly addressed. Our standpoint is basically our own position, who we are and how we see ourselves in relation to others and in relation to society. Th is means understanding that the researcher’s position is highly relevant to the way they approach and understand the research. How we see the world is not a neutral, objective understanding, but is inevitably infl uenced by the fi lters and frames of our life experiences and circumstances and our social, cultural, economic, and personal identity location. Th is means that female researchers will have a diff erent worldview on many topics than their male colleagues, younger people will likely see the social landscape diff erently to older people, and an Aboriginal researcher will see society and social research in very diff erent terms to a non- Indigenous researcher. Our standpoint—who we are socially, economically, culturally, even politically—underpins the questions we see, the answers we seek, the way we go about seeking those answers, and the interpretation we make, the theoretical paradigms that make sense to us. Our standpoint is theoretically summarised as the way research is guided by researchers’ social position and epistemological, axiological, and ontological frameworks: • social position: shaped by gender, culture, race, economic position • epistemology: theory of knowledge • axiology: theory of values • ontology: theory of being.

SOCIAL POSITION Our social position sets the frame for our standpoint. Who we are socially, economically, culturally and racially, and who we think we are across those dimensions, underpins the research questions we see, the answers we seek, the way we go about seeking those answers and the interpretations we make, as well as the theoretical paradigms that

Standpoint: The way we see the world and our position in it in relation to others and society. Our standpoint recognises the filters and frames that have an impact on our approach to our research.

Social position: Who we are and how we see ourselves, socially, economically, culturally and racially. Our social position shapes how we understand the research topic.

Figure 1.1: Conceptualisation of a methodology

Source: Adapted from Walter and Andersen 2013

1111CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1101_WAL_SRM3_20170_TXT_SI.indd 11 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

make ‘sense’ to us. Social position, therefore, not only substantially prescribes our life circumstances and experiences, it also shapes the worldview through which we understand these. In turn, these experiences and life circumstances infl uence how we comprehend our social position. Our gender, social, cultural, economic, racial identity, therefore, forms a central aspect of our methodology. Yet mostly its infl uence is largely invisible to us, and this is especially the case if our various identities are socially dominant, that is male, middle class, Euro-Australian, middle aged. Most of the infl uence of social position is internalised, it is how we understand and make sense of the world, and if our identities are mainstream it is more likely that we will perceive our own understanding of the world as ‘normal’ or, even more problematically, as ‘natural’.

A researcher, therefore, can be consciously and genuinely egalitarian, libertarian, non-sexist, and non-racist but this does not equate to nullifying the impact of their class, culture, race, and gender on their worldview. Being a young middle class man, for example, will inevitably infl uence how a male researcher will approach social research with older female respondents from working class backgrounds. Or being a white Euro-Australian older woman will unavoidably methodologically impact on research practice with young Aboriginal men. We embody our social position and so, as researchers, it covertly or overtly, actively and continuously, shapes our research practice. We can and, of course, should, always actively try to understand the world view of our research participants, but such engagement blunts, not removes, the impact of our social position.

EPISTEMOLOGY What we regard as knowledge has a strong cultural component. Th is concept is encapsulated by the term epistemology, which refers to a theory of knowledge—ways of knowing. Epistemology is concerned with understanding how the (mostly unwritten) rules about what is counted as knowledge are set, that is, what is defi ned as knowledge, who can and cannot be knowledgeable, and which knowledges are valued over others (Dooley 1990). As with social assumptions, dominant ways of knowing and the dominance of some knowers over others are embedded into our society. Social research is conducted against a background of these dominant ways of knowing.

Gender provides a good example of how our epistemology is infl uenced by our social location. Feminist social epistemology challenges the assumed objectivity and rationality of traditional ways of designating and valuing knowledge. Th is epistemology seeks to understand how the social relations of gender shape knowledge in our societies and investigates how socially constructed norms of gender and gendered experiences infl uence the production of knowledge and valid knowers. A feminist social epistemology also challenges the abstract individualism of social theories and theorists. Until the 1970s, most social theories and theorists were uncritically perceived as universal, a positioning that ignored that these knowledges were essentially all produced by white middle and upper class European and North American males. A feminist social epistemology illuminates that the experiential diff erences of knowers leads to diff erences in perspective, and that these diff erences have epistemic consequences. Th erefore, the knowledges produced and the valuing of those knowledges become entwined with and infl uenced by the identities, social positions, and social locations linked to the attributes of the knowledge producers (Stanford Encyclopedia of Philosophy 2006).

For example, German philosopher Karl Marx’s theories of capital essentially relate to men’s experience of the system, and the inequality and exploitation he exposes is also

Epistemology: Theory of knowledge concerned with understanding how knowledge is defined, valued, and prioritised.

12 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1201_WAL_SRM3_20170_TXT_SI.indd 12 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

that of men. Women’s position and experience are not considered, except in their role as producers of the next generation of workers or as a reserve army of labour for the bourgeoisie to exploit. Yet, women’s experience of the capitalist system is very diff erent to men’s, and much of that diff erence, in both experience and consequent knowledge, is based on gender.

Additionally, institutions of knowledge production, such as the stock market, the judicial system, and universities, can be dominated by the perspective of one type of knower without that perspective being recognised (Stanford Encyclopaedia of Philosophy 2006). Underrepresentation of women, younger people, and non-whites, or an overrepresentation of those from the upper and upper middle classes, means that knowledge is likely to be shaped by the epistemological perspective of those groups, perhaps even positioning such a perspective as the only way of knowing. Th e judiciary provides a clear example. With older, white Anglo- heritage upper class males dominating nearly all infl uential positions, the perspectives of those from non-English speaking backgrounds, the poor, women, and Indigenous peoples are only refl ected, for the most part, in rulings as they are, for the most part, understood by those in positions of power. Th is is not to suggest that the judiciary is biased, but that without lived experience of the social location of immigrants, Indigenous people, poorer people and women, such understandings are inevitably limited and incomplete.

AXIOLOGY Axiology refers to the theory of values, extrinsic and intrinsic. Applying this concept to social research and seeing how it fi ts within our methodology means that we need to understand our own value systems and those of the groups and institutions that have an impact on and are intertwined within our research approach.

Th is link between a researcher’s axiological position and their research raises the contested issue of values in research. A traditional perspective holds that researchers must aim to produce value-neutral knowledge based on observed objective facts. Th e feasibility and desirability of such an aim has been substantially challenged, and is mostly rejected by, contemporary social science researchers. Th e stronger argument in current social research debates is that social research cannot be value-free. Th e reasoning here relates to two key aspects of the social context of social research. 1 Social science is part of the social world

Social phenomena occur in the real world, where moral, political, and cultural values are an integral but oft en unseen part of the social landscape. Th is social context of our fi eld of study means that being value-free is next to impossible, claiming a value-free perspective is just another value statement.

2 Social context is central to our social science Th e specifi c social, cultural, personal, and moral milieus of the social phenomena we study are inextricably entwined with those social phenomena. For a social researcher to ignore the social context of the research is similar to a physical scientist ignoring the laws of physics: you might still generate results and theories, but the value of these are highly suspect. To gain some insight into our own axiological framework, we need to ask ourselves

some refl exive questions, which can include the following. • Why have we chosen the topic we have? • What is our particular research question and why have we settled on that aspect? • How did we decide that the topic—as opposed to others—was worth researching?

Axiology: The theory of values that inform how we see the world and the value judgments we make within our research.

Context: The settings in which texts are situated.

1313CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1301_WAL_SRM3_20170_TXT_SI.indd 13 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Th e critical point is that the questions we ask and the research decisions we make are not innate. Social scientists are embedded members of society too, and the non-acknowledgment of personal and/or social and institution values in the research does not equate to value- free or objective research. Rather, social research is about the real world in which moral, political, and cultural values are central to the things we examine. Th erefore, being value- free is impossible, and those who say this are kidding themselves or disingenuous.

Acknowledging our axiological frame does not mean that it is valid to try to make our research deliver particular results. Th e research project relies on open and professional practice. We must always adhere to research rigour and the scientifi c process. Not to do so renders our research invalid, if not fraudulent. Rather, understanding our own axiology and recognising that values are implicitly, at least, embedded in all research enables us to read our own research and that of others with an eye to the values informing it. Indeed, in some research, such as discourse analysis (see Chapter 12), unearthing explicit and implicit values is the core focus of study.

ONTOLOGY Ontology can be defi ned as theories related to the nature of being. Th erefore, our ontological framework refers to our understanding of what constitutes reality, how we perceive the world around us. At its most concentrated it is about how the world is understood: what reality is.

As with axiology and standpoint, ontology tends to be little discussed, mostly because the nature of reality tends to be taken for granted, especially within predominantly Western cultures such as Australia. But as social scientists we know that reality is not quite as concrete and immutable as we might usually think. As demonstrated in the classic Australian fi lm Th e Castle, the meaning and reality of a house is very diff erent if you are the home owner as opposed to being the engineer planning a runway extension in your area. Similarly, perceptions and understanding of time completely change if we are talking about our own activities, lifetimes, specifi c events, or the theory of relativity. From inside a Western framework, with its taken-for-granted assumptions of reality, it can be hard to come to grips with ontology. But for other cultures, those that hold diff erent understandings of reality, ontological frameworks are very clear because of the likelihood of a clash between their own ontology and dominant Western understandings. As with epistemology and axiology, it is easier to perceive ontological diff erences from outside the dominant culture.

In Australia and New Zealand, this is most clearly seen in the ontological positions of Indigenous peoples. Aboriginal people, Torres Strait Islander people and Māori hold ontological understandings of the nature of reality that diff er from Western norms. While unique to each Indigenous people, these ontologies tend to be more holistic in the way they view reality and less wedded to the Western presumption of humans as separate from other life and the earth itself. Scholar Karen Martin demonstrates this when she talks about her own Aboriginal people’s ontology of relatedness. In Quandamoopa ontology, all experiences are anchored to relatedness, which is the set of conditions, processes, and practices that occur among all entities—human, animal, spiritual, and ancestral— and all aspects of nature—animate and inanimate. In Martin’s words, ‘relatedness occurs across context and is maintained within conditions that are: physical, spiritual, political, geographical, intellectual, emotional, social, historical, sensory, instinctive and intuitive’ (2008: 69). Th e core message is that all social science researchers, whether from the

Ontology: The understandings of reality and the nature of being that inform our view of the world.

14 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1401_WAL_SRM3_20170_TXT_SI.indd 14 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

dominant or another culture, must try to recognise the ontological assumptions that frame their research topic and practice.

As Case study 1.2 shows, undertaking research developed from one culture’s perspective can be an activity fraught with danger.

CASE STUDY 1.2: WHEN SYSTEMS COLLIDE Australia’s five-yearly census asks around forty questions about items such as household composition, cultural background, and key demographic details. From a mainstream Euro- Australian perspective, the questions appear clear, unambiguous, and basically value- free. Such assumptions come seriously unstuck when these apparently straightforward questions are asked outside their cultural frame.

Observations of the conduct of the 2001 census at an Aboriginal settlement in the Northern Territory highlight some of the pitfalls. Morphy (2002: 40) reports that both the Aboriginal interviewers and interviewees found the process strange—‘they were Indigenous actors in a non-Indigenous scenario’. More critically, the quality of the data was compromised by the mainstream cultural assumptions of the questions, and the corresponding Indigenous lack of understanding of the mainstream culture and the intent of the census.

Examples of particular problems included the following. • How old are you?

Almost no one knew how old they were by the exact date or year of birth. Rather, local terms are used which designate degrees of maturity or stages of life, such as baby, child, circumcised boy, pubescent girl, young man. While there are clinical records for some of the younger people, for many older people dates of birth are guesstimates, and often use 1 January or 1 July for official documents.

• Place of residence ‘Place’ was interpreted to mean ‘community’ rather than ‘dwelling’. The community is seen as home, and the answers reflected identity rather than physical presence. Most answered ‘Yes’ to the question of whether they lived there most of the time, even if they were highly mobile. ‘Staying’ and ‘Living’ were viewed differently.

• What is a household? Community residents saw themselves as part of a family, but this family rarely mirrored the nuclear model of parents and children, and was often spread across more than one dwelling. Kinship relationships within households also did not fit the mainstream Australian model, and could not be adapted to the census questions around relationships. Overall, the Aboriginal interviewers took the task very seriously and endeavoured to

complete the forms as best they could, but the quality of the data collected was doubtful. For a significant number of questions, Morphy (2002: 40) notes that a jocular approach was taken to ease the awkwardness, with ‘jokes made at the expense of white people for wanting to know these things’.

THEORETICAL FOUNDATIONS In acquiring social research skills, the concentration on research methods and the collection of data means that it is easy to forget that social research involves two linked elements. Th ese are: 1 empirical data 2 social theory.

Empirical data: Information that is the result of observing and/or measuring social phenomena.

1515CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1501_WAL_SRM3_20170_TXT_SI.indd 15 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Social theory and empirical data have separate but interdependent roles. Each is an essential aspect of social research, and each is relatively useless on its own. While data fi nding, for example, a low rate of breast screening/mammograms among Aboriginal women are informative, alone these data are relatively meaningless. Th ey lack a theoretical framework in which they can be understood. Alternatively, proposing that Aboriginal women are disempowered by the medical model and are reluctant to seek non-urgent medical care is an interesting theory. However, without empirical supporting data, it is essentially an unproven speculation. Th ere is no way to tell whether the theory is correct. Put the empirical data and the social theory together, and what we have is social research.

THEORETICAL CONCEPTUAL FRAMEWORK Th is theoretical terrain is our conceptual framework. A conceptual framework is just that: a theoretical map for how we will conceptualise our data, its analysis, and its interpretation to answer our research question. Th e conceptual framework is the theory or theories that will guide our analysis and understanding of the empirical data. Which theory, or more oft en theories, we will use to provide this map emerges from our literature review. What theories have others used, or developed, when researching our or an aligned topic? Do they make sense to us and do they meet the requirements of our research question? An example helps illustrate what many new researchers fi nd a challenging task. In my honours research I was interested in the topic of child support, and decided to focus on private collection, that is, collection of child support monies directly arranged between the separated parents, rather than by the child support agency. My question was: How satisfi ed are payee parents with their private child support collection arrangements (Walter 2002)? Th e topic and developing a clear research question was relatively unproblematic, but I struggled with defi ning my theoretical conceptual framework. My area was (then) new, and there was little existing literature, or theories, around the topic. Aft er much reading I fi nally realised that feminist theories around the delegation of the family to the private sphere in social policy and family discourse were theoretically central to the topic and the question.

I am also oft en asked whether a theoretical conceptual framework is always necessary in social research. Can’t we just identify the issue, such as low rates of take-up of breast screening services by Aboriginal women, then move straight into developing strategies to address the problem? Th e answer is most strongly no. As Babbie (2002) points out, no matter how practical or idealistic our aims, unless we have a theoretical understanding of the social terrain we are traversing, our research is likely doomed to failure. Even worse, the lack of an acknowledged theoretical base can disguise the unacknowledged concepts and understandings that inform our work. Operating without a clearly established conceptual framework signifi cantly constrains the value of our work and undermines its validity and its rigour.

THEORETICAL PARADIGMS Our conceptual framework is likely to be made up of a number of theories that infl uence our understanding of the topic. Th ese individual theories can also oft en be aligned with a larger theoretical category, or paradigm, that encompasses a broad theoretical fi eld that emerges from the same perspective. A paradigm is essentially a macro theoretical frame

Social theory: An idea or a set of ideas that explain social phenomena.

16 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1601_WAL_SRM3_20170_TXT_SI.indd 16 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

of reference. Th us, for example, in my honours research noted above, the theories I used fi tted within a feminist paradigm.

Within the social sciences, a set of established paradigms exists, within which individual social theories are oft en drawn and developed. A range of these is outlined below, but this is by no means an exhaustive list. Nor is it fi xed. Social science paradigms are an evolving, developing landscape with new paradigms emerging, or re-emerging in their infl uence, or fading in impact. Refer to your social theory texts for a broader explanation of the key social theories that inform and infl uence social science research.

FUNCTIONALIST PARADIGM

Functionalism was the dominant sociological paradigm through the mid twentieth century. Associated with the work of Emile Durkheim (1858–1917) and later developed by US sociologist Talcott Parsons, functionalism starts with a basic question: How is social order possible? It sees the answers in terms of stability, social order and consensus. Th e organic analogy, developed by Durkheim, is used to explain how society works by comparing the social world to a biological entity in which all parts are separate but interdependent. If one part is not functioning well, it aff ects the operation of the others and, therefore, the wellbeing of the whole social system. With its emphasis on the objective nature of social norms and values, functionalism in social research is associated with the social science research conducted within objective scientifi c frameworks.

CONFLICT PARADIGM

From a confl ict perspective, social relations are based on exploitation, oppression, and confl ict. Th e work of Karl Marx (1818–83), who focused on the struggle of the economic classes, especially the exploitation of the working class or proletariat by the capitalist class, forms the basis of confl ict perspective. However, the application of a confl ict paradigm now encompasses more than just economic oppression, and is used to examine social struggles based around class and ethnic divisions or wherever a confl ict of interest occurs among diff erent social groups (Babbie 2002:30). A confl ict perspective begins from the notion of society as inherently unequal, and engaged in ongoing confl ict around the competing interests of diff erent social groups. It is this confl ict and the consequent relations of power that exist between diff erent groups that determine a society’s social arrangements and drive social change. Social research framed or developed within a confl ict paradigm tends to examine social phenomena in terms of who benefi ts from this set of social arrangements or this social change, and who is disadvantaged. A confl ict perspective provides a big-picture, macro-perspective of society or larger social groups. Research using a confl ict perspective frame, therefore, tends to be associated with large-scale, oft en quantitative, research methods, such as surveys (see Chapter 6).

INTERPRETIVIST PARADIGM

An interpretivist paradigm concentrates on social agency, and is concerned with the way we, as social beings, interrelate and interact in society. Developed initially from the work of German sociologist Max Weber (1864–1920), the interpretivist perspective emphasises the meanings individual actors give to social interactions, and the use of symbols, such as language, in the creation of that meaning. From an interpretivist perspective, the human world is a world of meaning in which our actions take place on the basis of shared understandings. To understand society, we need to understand people’s motives and interpretations of the world. Th e meanings actors give to their circumstances are the

1717CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1701_WAL_SRM3_20170_TXT_SI.indd 17 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

explanation of what they do. Th e micro-level emphasis of the interpretivist paradigm and its focus on the role of meanings in how individuals interpret social life mean that the interpretivist paradigm is fundamentally associated with qualitative social research methods, such as indepth interviews (see Chapter 10).

FEMINIST PARADIGM

Feminist paradigms are developed by feminist scholars to counteract what has been an overwhelming male-centric approach to the study of our social world. Until relatively recent times, women were largely ignored in the social sciences, both as researchers and as social subjects. Women’s diff ering experiences of social reality were essentially invisible in mainstream social science research and theory. Although feminist paradigms see gender as a fundamental social division and signifi er of life chances, there exist a number of feminist paradigms rather than a single perspective. Th ere is no single feminist approach to social research, and feminist paradigms have been used as the theoretical frame for a diverse range of social research methods and projects. Rather, a feminist perspective is more likely to inform the social question that is posed and how the topic is defi ned. However, because of the feminist challenge to traditional social research paradigm claims of objectivity and reason, feminist paradigms are oft en associated with qualitative research methods such as ethnography, life histories, and memory work.

INDIGENOUS PARADIGM

Here, the theoretical framework that directs the questions, the choice of methods, the way of studying, and what is valued as knowledge is determined from an Indigenous perspective. Th e Indigenous paradigm directly challenges many of the traditional Western ways of thinking about and approaching social research, what the research process should look like, and what the research outcomes should be. Th e research techniques used within an Indigenous research paradigm have generally been those that can more easily admit Indigenous agendas and Indigenous community interests to their purpose and practice (Tuhiwai Smith 1999) but can include any method. Th e applied research framework of participatory action research (see online module), which emphasises the relocation of the power in the research relationship from the researcher to the researched, means that this social research method is frequently employed by researchers approaching their work from an Indigenous paradigm.

POSTMODERNISM

Th e postmodernist paradigm has had a strong infl uence on the social sciences in recent years. Associated with the work of French philosophers such as Jean-François Lyotard and Jean Baudrillard, reality, from a postmodern perspective, is always a subjective experience and, essentially, constructed. Rather than observable social phenomena, social reality is fragmented and diverse, and all human knowledge and experience is relative. As a result, there are no absolute values or truths, and it becomes impossible to study objective realities. While a postmodernist paradigm brings into sharp relief many of the unsupported assumptions of a positivistic model, such as the provability of an hypothesis about the social world or the objectivity of social data, it also raises a social research dilemma. If all reality, or social phenomena, are fundamentally subjective then, from a postmodern paradigm, what is the point of undertaking any social research at all?

Ethnography: A research method that involves conducting ethnographic fieldwork.

Memory work: Memory work is a collaborative technique used to generate stories that are based on personal memories among a group of co-researchers: see also Triggers.

Outcomes: The specific consequences of a particular course of action.

Participatory action research: A cyclical research process aimed at providing feedback into a cycle for problem solving. It is a practical research method that requires an equal and open collaboration between the researcher and the research community.

Hypothesis: Hypotheses are prescriptive forms of research question that state a particular scenario that the research will confirm or refute.

18 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1801_WAL_SRM3_20170_TXT_SI.indd 18 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

METHOD Th e fi nal component of your methodology is the research method you choose, the core topic of this book. As stated, our research method is the technique or practice we use to gather our research data, such as an indepth interview, survey or discourse analysis. Indeed, social science researchers are spoilt for choice in their selection of a social research method. Th is book and the online chapters cover a wide selection of the methods in common use by Australian social science researchers, but this range is by no means complete. How to go about selecting your research method is covered in the next chapter, but it is important to remember two key premises in relation to method selection: 1 Th e method must suit the research topic and question, not the other way around. 2 All methods have strengths and weaknesses. Th ere are no such things as good

methods, bad methods or even methods that cannot be used within particular types of research.

VOICES IN THE FIELD: KAREN MARTIN

Karen Martin is an associate professor in the School of Education and Professional Studies at Griffith University. The following paragraphs outline how she developed an Indigenous methodology for her now completed doctoral research.

My professional background is in early childhood education, particularly Aboriginal education. It is through my teaching that I developed a keen interest in knowledge acquisition and how transfer occurs in Aboriginal teaching–learning contexts (in homes and classrooms). It is also through the experiences of a native title application that procedures for research caught my attention, particularly in the way the knowledge and realities of Aboriginal people were represented, misrepresented, distorted, and sometimes stolen. Thus began a journey of understanding research and how this misrepresentation occurs, but equally how it could be different and have better outcomes for Aboriginal people.

I theorised an Indigenist research methodology informed by an Aboriginal worldview, knowledge, and ethics. Underpinning this methodology is relatedness, a term I developed that theorises the essential and core condition by which Aboriginal people have lived, do and will continue to live. This is articulated within the research through the use of ‘traditional devices’ whereby Aboriginal ways of knowing, ways of being, and ways of doing are centred and strengthened.

This Indigenist research methodology was used in a PhD study regarding the regulation of outsiders by rainforest Aboriginal peoples of far north Queensland. The findings reveal the multiple forms of agency that exist and are used to regulate outsiders occurring from the past, to the present, and for the future. Therefore, the necessity for researchers to regulate their own behaviours in respect of and in accordance with Aboriginal terms of reference and in relatedness is paramount. When research is regarded as an interface of Aboriginal people, research, and researchers, Aboriginal voices are not erased, silenced, or diminished. This research becomes a vehicle for getting our stories back.

1919CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 1901_WAL_SRM3_20170_TXT_SI.indd 19 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

QUANTITATIVE AND QUALITATIVE In a return to the fi rst message of this section, qualitative or quantitative methods are not the same as qualitative or quantitative methodologies, although we will almost certainly use a qualitative method within a qualitative methodology and a quantitative method within a quantitative methodology. Also remember that there are many more social research methods available to you than those listed in this text. Social research methods is a dynamic fi eld, within which new methods are emerging, and the use of others waxes and wanes. Th e website accompanying this text, www.oup.com.au/orc/walter, provides core information on a range of other methods. On quantitative methods, the website contains two extension segments, quantitative analysis using SPSS 1: correlations, and quantitative analysis using SPSS 2: comparison of means. Qualitative modules include conversation analysis, action research, memory work, and the emerging fi eld of sensory and visual research methods.

QUANTITATIVE METHODS

Quantitative research involves the collection and analysis of data that can be presented numerically, or codifi ed and subjected to statistical testing. Its primary role is to allow the collection, analysis and development of understandings and interpretations of data on social phenomena from large groups or large data sources. Quantitative research is commonly associated with standard Western research scientifi c methods, but this association is limited to the shared usage of statistical analysis to demonstrate and measure associations between diff erent concepts. Quantitative methods can be incorporated into a range of methodologies. Major quantitative methods include gathering data through surveys, questionnaires, and structured interviews.

QUALITATIVE METHODS

Th e key task of qualitative research is meaning making, a process that does not usually require statistics or large-scale data. Instead, the key focus in qualitative research tends to be on smaller units of people and society, with the method and analyses drawing out the meanings, perceptions, and understandings that individuals and groups attach to behaviours, experiences, and social phenomena. Qualitative research is adaptable to a broad range of methods and data sources. It is a subjective approach whereby the researcher aims to understand and interpret experiences by viewing the world through the eyes of the individuals being studied. Methods include the various forms of observation, focus groups, and unstructured interviews.

FROM METHOD TO PRACTICE To understand what a research methodology is, we must recognise that all research and all researchers are embedded in their cultural milieus and steeped in particular standpoints. Does the cultural relativity of our own ways of knowing, the acknowledgment of the influence of our value system, and our understandings of the nature of reality leave us, as social researchers, in a quandary about the worth of our research? Not really. We just need to understand and acknowledge that our research process, our research findings, and the theories we develop are not core truths, but that they are shaped and influenced by our particular values and understandings.

From this perspective, all research is a cultural product. As you will have gathered from the preceding sections, the combination of standpoint, conceptual framework, theoretical paradigm, and method are unique to the individual researcher. But this does

20 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 2001_WAL_SRM3_20170_TXT_SI.indd 20 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

not mean that methodologies are a random combination, or that there are as many methodologies as there are researchers. Rather, there tend to be similarities across key underpinnings that shape our methodologies, as well as the need for scholarly rigour, the need to open our research process and practice, and the need for incorporating research ethics (see Chapter 4). This means that we can usually categorise individual frames of research reference into umbrella types of methodologies, such a feminist or Indigenous.

THE QUALITATIVE–QUANTITATIVE DEBATE At the centre of this point is what is commonly known as the quantitative–qualitative debate. In this somewhat vexed and pointless argument, sides are taken as to whether quantitative methods (that is, methods that produce data relating to social phenomena that are amenable to statistical analysis) or qualitative methods (methods that concentrate on drawing on the detail and social meaning of social phenomena) are superior. Th e origins of the debate are buried in the time when the dominant use of quantitative, statistically based social science was challenged by emerging qualitative methodologies and approaches. With the place of qualitative research within the social sciences fi rmly established, this debate is now past its use-by date. Qualitative and quantitative research methods are now regarded as forming diff erent, but equally vital, aspects of the social science research endeavour. Each methodological approach is just an element of the same whole: research. Th is debate also demonstrates the previously described misunderstanding between method and methodology. Many research designs now combine quantitative methods, which provide reliable results that can be generalised to the wider population under study, with qualitative methods to produce results that are rich in meaning and understanding of social processes (triangulation).

EXERCISE 1.2: FOOD IN PRISON In 2006, Di Heckenberg and Danielle Cody undertook a study of food in prison. The research was triggered by a siege at the prison the year before, in which a central demand was the delivery of fifteen pizzas and there were strong protests over new processes and practices around prison food. The research question was: What is the place and meaning of food in the prison experience of inmates? To answer this question, the researchers conducted semistructured interviews with six former inmates. The interviews revealed concerns about the following areas: • the closure of mess rooms • being expected to eat in cells • reheating food that had been prepared earlier • the move to plastic utensils and foil containers rather than plates and cutlery • the replacement of hot weekday lunches with sandwiches.

Respondents also tended to reflect positively of the ‘old days’. One former inmate said:

Years ago, I used to look forward to our food. What I mean, it was, it wasn’t anything spectacular, but there was plenty of it. Back then, you could always get extras and that, like, they used to bring down whatever was left from the kitchen, they would bring it down on plates … Just say you had a steak for tea, whatever steak was left in the kitchen they would just bring it … divide it around the yards … That was back then.

Quantitative– qualitative debate: Debate in which sides are taken by researchers as to whether quantitative methods (that is, methods that produce data relating to social phenomena that are amenable to statistical analysis) or qualitative methods (that is, methods that concentrate on drawing on the detail and social meaning of social phenomena) are superior.

Triangulation: Triangulation describes the combining of different research methods. The value of this practice is that the researcher can gain the advantages of each method used while also reducing the limitations of a single method.

2121CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 2101_WAL_SRM3_20170_TXT_SI.indd 21 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

YOUR TASK 1 What explanations (theories) can you think of that might explain why the changes

around food have been so unpopular with prisoners? 2 Based on your social understanding of the meanings we make of food, theorise why

food might be such a significant topic for inmates that they would riot over changes. 3 How might we interpret the data around the unpopularity of the closure of the mess

rooms and prisoners having to eat their meals in their cells? 4 Why do you think prison authorities made the described changes to the food regime in

this prison?

22 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

MAGGIE WALTER

01_WAL_SRM3_20170_TXT_SI.indd 2201_WAL_SRM3_20170_TXT_SI.indd 22 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

CONCLUSION Th e key message of this fi rst chapter is that social research is a very broad undertaking. Good social science research is based on a sound understanding of the scientifi c method, the specifi c complexities of studying the social realm, and the key interrelationship between data and theory. From this base, there is an endless number of social questions to be asked and social phenomena to be studied, and many diff erent methods and frameworks for conducting the social research questions we pursue. To be an eff ective and enthralled social researcher, you need a core familiarity and understanding of a wide range of these methods. Th e specifi c research method you select for your social research project is dependent on the topic of your study, the methodological framework from which you are approaching your topic, the specifi c research question that you ask, and the practical and resource constraints.

MAIN POINTS

• Social research is about investigating the social questions we have about our social world.

• Th e humanness of social research means that social research is oft en a more complicated endeavour than other scientifi c research.

• Social science research is distinguished from social commentary or opinion by its use of scientifi c method.

• Th e primary aim of social research is to identify, to investigate, and to try to understand social patterns and social meanings.

• Th e social context of social phenomena is an essential element of social science research.

• Social scientists deal in social aggregates, individual exceptions do not challenge social fi ndings or explanatory theories.

• Th e two core elements of social science research are empirical data and social theory. • Social science research is informed and infl uenced by our worldviews and

perspectives. In social science terms, these can be classifi ed as our standpoint, our epistemologies, axiologies, and ontologies.

• Methodologies are made up of these, as well as our method and theoretical framework.

FURTHER READING Th ere are many social research books available, varying in quality, depth, level of coverage, and accessibility. Your library should contain a wide variety of such texts. For more detail on the process of research, the following Australian text is useful. Bouma, G. D. & Ling, R. (2004). Th e Research Process, 5th edn. Melbourne: Oxford

University Press.

2323

MAGGIE WALTER

CHAPTER 1: THE NATURE OF SOCIAL SCIENCE RESEARCH

01_WAL_SRM3_20170_TXT_SI.indd 2301_WAL_SRM3_20170_TXT_SI.indd 23 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

REFERENCES Australian Institute of Family Studies. (2002). ‘Introducing the Longitudinal Study of

Australian Children. Longitudinal Study of Australian Children’. Discussion Paper No. 1. Melbourne: Australian Institute of Family Studies.

Babbie, E. (2002). Th e Basics of Social Research, 2nd edn. Southbank: Th omson Wadsworth.

Barraket, J. & Henry-Waring, M. (2008). ‘Getting It On(Line): Sociological Perspectives on E-Dating’, Journal of Sociology, 44 (2): 149–66.

Dempsey, K. & de Vaus, D. (2003). ‘Who Cohabits in 2001? Th e Signifi cance of Age, Gender, Religion and Ethnicity’, Journal of Sociology, 40 (1): 157–78.

Dooley, D. (1990). Social Research Methods, 2nd edn. Englewood Cliff s: Prentice-Hall.

Glicken, M. D. (2003). Social Research: A Simple Guide. Boston: Pearson Education. Heckenberg, D. & Cody, D. (2006). ‘Food Matters: Issues Surrounding Food in Prison’.

Criminology Research Unit, Occasional Paper No. 3. Hobart: School of Sociology and Social Work, University of Tasmania.

Mann, P. H. (1985). Methods of Social Investigation. New York: Basil Blackwell. Martin, K. (2008). Please Knock Before You Enter: Aboriginal Regulation of Outsiders and

the Implications for Researchers. Teneriff e: Post Press. Morphy, F. (2002). ‘When Systems Collide: Th e 2001 Census at a Northern Territory

Outstation’. In Martin, D. F., Morphy, F., Sanders, W. G. and Taylor, J. (eds), Making Sense of the Census: Observations of the 2001 Enumeration in Remote Australia. CAEPR Research Monograph No. 22. Canberra: Centre for Aboriginal Economic Policy Research, Australian National University: 29–73.

Natalier, K. (2001). ‘Motorcyclists’ Interpretations of Risk and Hazard’, Journal of Sociology, 37 (1): 65–80.

Neuman, W. L. (2004). Basics of Social Research: Qualitative and Quantitative Approaches. Boston: Pearson Education.

Senate Community Aff airs References Committee (2004). A Hand Up Not a Hand Out: Renewing the Fight Against Poverty. Canberra: Commonwealth of Australia.

Stanford University. (2006). Stanford Encyclopedia of Philosophy. Stanford: Stanford University Center for the Study of Language and Information, Stanford University.

Tuhiwai Smith, L. I. (1999). Decolonizing Methodologies: Research and Indigenous Peoples. London and New York: Zed Books.

Walker, A. (2000). ‘Measuring the Health Gap Between Low Income and Other Australians, 1977 to 1995: Methodological Issues.’ Discussion Paper No. 50. Canberra: National Centre for Social and Economic Modelling, University of Canberra.

Walter, M. (2002). ‘Private Collection of Child Support: Back to the Future’, Just Policy, 26: 18–27.

MAGGIE WALTER

PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

01_WAL_SRM3_20170_TXT_SI.indd 2401_WAL_SRM3_20170_TXT_SI.indd 24 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Research Design KRISTIN NATALIER

THE RESEARCH QUESTION: THE HEART OF THE RESEARCH DESIGN The research question is the key aspect of the research process. Once you know precisely

what it is that you want to discover, designing the rest of the research project becomes a

lot easier. After developing a well-defined and clearly articulated research question, you

will find it easier to determine the best way to approach the topic, the most appropriate

data collection method, and the most effective analysis techniques.

Research questions come in diff erent shapes and sizes. Some are simple and open

in terms of what fi ndings they can generate. One recent project was guided by this

research question: ‘Why do people take out private health insurance?’ (Natalier & Willis

2008). Others specify a more bounded focus, such as another project where our question

was ‘How does the payment and receipt of child support, along with the receipt of

government income and housing assistance, aff ect the housing outcomes of parents

apart (both resident and non-resident parents)?’ (Natalier et al. 2008). Some questions

incorporate concepts that will need to be further defi ned. All specify exactly what the

particular study is investigating, and do so in question form. The research question sits at

the heart of research design.

KEY TERMS conceptualisation

deductive theory

inductive theory

literature review

measurement

method

methodology

nominal defi nition

operationalisation

research design

research proposal

research questions

02_WAL_SRM3_20170_TXT_SI.indd 2502_WAL_SRM3_20170_TXT_SI.indd 25 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

INTRODUCTION When you look at your world with creativity and imagination, research is an exciting idea. It is a tool for making intriguing, confronting, and complex discoveries. But how can you explore society in ways that let you discover something that is not only new or unexpected, but is also meaningful? Research design is the fi rst, necessary step. It is a comprehensive task that draws together all aspects of the research: the topic, initial question, and aims, through to deciding how to collect and analyse your data, and how you will communicate your result. Without a strong research design, your research is likely to be swamped by choices and options that are diffi cult to consistently and logically respond to within the research process.

In practice, the amount of detail in a research design partially depends on your initial approach. Some qualitative work can unfold over time and may change in response to questions and themes that emerge through the collection and analysis of data. In contrast, quantitative work requires a prestructured and iterative approach to the specifi cation of research questions, and the conceptualisation, operationalisation, and measurement of variables. But even researchers who plan to stay open to where the data lead them need clear statements about the focus and aims of their project and how decisions will be made.

Th e process of research design usually results in a formal document, the research proposal, which specifi es and justifi es the research plan. Th is is typically reviewed before the researcher can begin the data collection. Even when a research proposal is not formally required, thinking through the design is a vital prerequisite for a successful project. Investing your time and intellectual energy in this early part of the research process means you will have created a blueprint for an exciting and (relatively) stress- and confusion-free research experience.

Th is chapter details the process of designing a social science research project through to the research proposal stage. Emphasis is given to the key elements of a thorough and useful research proposal: the research question/s and project aims, a comprehensive and critical literature review, a description and justifi cation of the most appropriate research method and analysis, and defi ning and measuring the key concepts, along with the pragmatic prerequisites of developing a timeline and a resource budget.

RESEARCH QUESTIONS A research question drives your study. Strong questions can lead you to generate new knowledge and theories about the world, resolve disciplinary controversies or empirical contradictions, identify and contribute to solving social problems, and debunk common sense understandings about a range of social phenomena (Sandberg and Alvesson 2011). Research questions don’t take up much space when you write them out but they are overwhelmingly important in your research design.

Developing good research questions is hard work and sometimes takes a surprising amount of time. Given the eff ort required, perhaps it is not surprising that it is also a step that is oft en missed, as people move straight from the research focus (the general topic) to research method selection. Can you spot the sometimes fatal fl aw in this process? Th e research question shapes every other component of the project. It places boundaries on a project, giving it coherence and direction. It will determine which methods are appropriate, the focus of analysis, and the choices we make when writing up research.

Research design: The planning process that generates a plan of all aspects of the research project and how they will fit together to result in rigorous research.

Research questions: Research questions state the major focus of the research in question form, incorporating the key idea that the research seeks to investigate and/or explain, and the key concepts of the research.

Operationalisation: The process of defining how concepts work.

Measurement: How you are going to identify your key concepts in the social world. Measurement is a term more commonly used in quantitative research.

Method: The research technique or practice used to gather and analyse the research data.

26 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 2602_WAL_SRM3_20170_TXT_SI.indd 26 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

In short, a research question will keep you focused. It is important to get it right, and to see the development of a research question not as wasted time (so few words for so much eff ort) but as the necessary foundation of a fi nal successful outcome. At the end of the process, your research might have generated many interesting fi ndings, but without a clearly articulated research question, these are likely to be data more or less relevant to a topic but not coherent social analysis.

Bouma and Ling (2004: 14) specify two properties of strong questions. First, they are limited in scope. No one project can address every dimension of a topic. Second, research questions are related to an empirical phenomenon, that is, they can be answered through studying observable and tangible things. Metaphysical questions such as ‘Does God exist?’ or moral questions such as ‘Is terrorism ever justifi ed?’, while important, are not questions social scientists seek to answer through their research.

GENERATING RESEARCH QUESTIONS As Robson (1993: 25) notes, there is no guaranteed way of generating good research questions. Sometimes, your research question(s) will be reasonably clearcut, arising cleanly from your research focus and review of the existing literature. More oft en, formulating a precise research question is a more diffi cult task. Th ere are so many aspects of so many social issues that you could investigate. Selecting one aspect necessarily excludes others, but a decision has to be made. As detailed in Chapter 1, such decisions are infl uenced by your interests, your values, and how you see the world; they are also dictated by practical considerations, such as time, money, and your skill set.

THE DOS OF RESEARCH-QUESTION GENERATION When it comes to developing a research question, here are some useful strategies. • Read. Th rough reading, you can identify what is yet to be known about a topic and

refi ne what you want to know. • Think in terms of concepts, not just topics. A topic may be very broad (for example,

fi nances in couple households), but exploring it through the lens of a particular concept (such as individualisation or gender) can help focus on what it is you want to ask.

• Identify puzzles. Th rough the literature review (discussed in more detail, below), you can oft en identify contradictions in the existing research. Th ese contradictions may inform your own research question.

• Identify existing gaps and limitations in existing knowledge. Th ere may be particular social groups that have not been studied in relation to your topic, either because their conceptual, political, or empirical importance has not been recognised or because previous samples did not include them. When topics tend to be studied using one type of method, a diff erent focus can be pursued through an alternative methods approach. Sometimes, too, your reading will highlight unexamined assumptions within the discipline, critically engaging with these assumptions can be the basis of an interesting project.

• Acknowledge feedback loops. David Silverman (2005: 90) points out that the move from research questions to fi ndings is not always one way, and that emergent fi ndings may shape research questions. But it is a balancing act. A tight focus is not the same as adopting an unyielding approach to emerging issues. Nor is it useful to create a question that is so unfocused that it leads you to chase down every potential issue at the expense of following through analysis and conclusions.

2727CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 2702_WAL_SRM3_20170_TXT_SI.indd 27 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

• Write a question. Rather than making a statement (for example, ‘I will be exploring what activities divorced fathers do with their children’), rewrite your words to end with a question mark (‘What activities do divorced fathers do with their children?’).

• Talk to people. Research is oft en a collaborative process. Even if you are not working in a research team, talking about your ideas and hearing other people’s thoughts can clarify what it is you want to ask.

THE DON’TS OF RESEARCH-QUESTION GENERATION Th ere are also some things to avoid when developing research questions. • Avoid rushing to your research question. A good research question requires critical and

creative skills, and takes time and work to get right. • Avoid problems of misinterpretation by unambiguously conceptualising all of the core

concepts contained in your research question. Th is will militate against questions that are too broad or vague (for example, those dealing with poverty or social problems). Although these defi nitions are not included in your research question, they should be presented in the proposal.

• Avoid posing long and complicated research questions. Your research question should be clearly and simply articulated. Use subquestions (also clearly and simply articulated) to include the other important facets of your research question.

• Avoid tying your research question to a particular method. Th e research question should always be the precursor to your selection of a method—not the other way around.

• Avoid asking questions that have already been answered. Replicating a research study or question to verify a fi nding or to test if it is applicable in a diff erent social context is diff erent from being ignorant of previous studies in your fi eld (Robson 1993: 28).

EXERCISE 2.1: DEVELOPING RESEARCH QUESTIONS Below, three current social issues are listed. Develop a research question for each of them, then check to see if you have inadvertently done any of the things you should avoid. 1 high school truancy in regional towns 2 housing affordability for Generation Y 3 ageing population and the labour market.

RESEARCH AIMS When researchers undertake a project, they do so for a reason. Th ese reasons are described as your research aims. Research aims explain why you have chosen to pursue a particular issue—what is your motivation for undertaking the study? Another way of thinking about it is to ask what you hope the outcomes of the study will be. Most obviously, you hope to fi nd an answer to the research question, but beyond that, a good research design is clear about why you are seeking that answer. Th is clarity is important in creating a coherent and focused plan. It can also play a strategic role for those researchers whose research proposal is part of an application for research funds or institutional approval, because the aims are one indication of the importance and value of the proposed research. In such circumstances a convincing statement of aims can help to answer the question: ‘Why should this research be supported?’.

Aims: What you want to achieve through your research.

28 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 2802_WAL_SRM3_20170_TXT_SI.indd 28 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Researchers pursue many diff erent aims. Th ese may include a contribution to social scientifi c knowledge (testing a theory or exploring an underresearched phenomenon), a wish to inform public policy (developing more eff ective ways of addressing a social issue), a desire to improve practices in a particular fi eld (helping people understand the social world and respond to it), or highlighting the experiences of social groups whose worldviews or experiences may be ignored in the political or academic realms. You may have more than one aim for your study. In common with the research question, it is best to keep aims modest, so that you can meet them. But it is also important to have faith in the relevance and strength of your research: rigorous and strategically communicated research can contribute to social change—and stability.

BOX 2.1: AN EXAMPLE OF CLEAR RESEARCH AIMS It can be difficult to find a clear statement of aims, even in the work of experienced researchers. Often, aims are a rewording of the research question. Others are clearer in specifying what they are trying to do with their work. In a piece of research on the housing experiences of young people leaving state care (that is, foster care), Johnson et al. (2010) described aims relating to policy and aims relating to practice.

In light of the importance of housing, this project aims to inform policy and service delivery and practices to promote positive and sustainable housing outcomes among a group where homelessness and tenancy breakdown have been widely reported for almost two decades. It also aims to establish a basis for policy development that has the capacity to enhance the wellbeing and the social opportunities of young people who ‘age out’ of the state out-of-home care system.

LITERATURE REVIEW A literature review is a vital early step in any project (with the exception of projects adopting a pure grounded theory approach). When you conduct a literature review you fi nd out what is and is not known, how it has been studied, and the key questions, debates, concepts, tensions and gaps in existing knowledge. Such a review should be, in the words of Neuman (2006: 123), ‘comprehensive, selective, current and critical’.

A comprehensive literature review shows familiarity with existing knowledge and establishes your credibility as a researcher in the fi eld. Rather than just reviewing everything related to your topic, you need to be selective in your focus so that your discussion links the proposed research with previous work. Being current is also important, because it helps you and your readers make sense of what is known in the current context. A literature review must also be more than a report of previous theories and empirical fi ndings, it critically analyses these in relation to the topic. When writing your literature review, aim to present an original perspective on what has already been written about the topic. Th is perspective will inform the rationale for your study, either by identifying gaps in the literature or by demonstrating that a new or diff erent approach is needed on an existing topic (Jesson et al. 2011). In so doing, the literature review also off ers implicit justifi cation for your choice of methods and analysis.

Theory: Existing explanations of social processes and patterns.

2929CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 2902_WAL_SRM3_20170_TXT_SI.indd 29 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

APPROACHING A LITERATURE REVIEW It can feel quite daunting to approach a literature review for a new project. Aft er all, there is so much information. Th e key is to draw boundaries around the mass of material by thinking clearly about what it is you want to do, and therefore what it is you need to read about and comment on. Some people may begin a literature review with a clearly defi ned research question, but as a literature review will stimulate new ideas, a question needs to be fl exible enough to change as you read. Other people will develop a question only aft er a lot of reading. Regardless, you will need well-developed information literacy and retrieval skills.

Literature search skills have two dimensions. Th e fi rst is familiarity with generic search procedures. Most institutions run short workshops on how to use the library, databases, and the internet, and the few hours of instruction can, in the end, save a great deal of time and annoyance. Th e second is developing search strategies for your own topic. Sometimes you will be working in a fi eld that has been comprehensively studied, but other topics may be relatively new or underresearched. Regardless of whether you are facing too much or too little information, you will need to be creative in your exploration of existing information. It is useful to consider the following. • What disciplinary boundaries are you working in? If there is too much information, can

you narrow your search to just sociology or criminology, at least as you begin your reading? Alternatively, it may be that a topic with no profi le in sociology may have received considerable attention in sociolegal research, for example, and concepts and fi ndings in that discipline can be translated into your own.

• How are you conceptualising key issues in your topic? Th is requires higher level thinking than a simple topic area. It may be that you are using your topic as a way of thinking about concepts or theories, or specifi c concepts will give you an insight into the empirical issue.

• Identify the key works. As you begin reading, you will soon fi nd that the same monographs and articles are cited again and again. Focus on these, at least to begin with, as they form the basis of the literature in your fi eld.

• Look in broadly relevant journals. Even when sophisticated searching turns up little of relevance, identify journals in the fi eld and start to fl ip through the abstracts. You may fi nd previously unconsidered synonyms or conceptualisations that point you in the direction of existing research. Similarly, one book may include a lot of relevant resources in its reference list.

• Broaden or narrow the years you are searching in. If your search turns up too many sources, consider narrowing the fi eld to the past fi ve years—any important earlier work will turn up in the references of more recent sources. If you are searching with defi ned date parameters and fi nding nothing of relevance, you may need to broaden your time period. Topics in the social sciences can, like any fashion, come and go, and your chosen focus might currently not be popular (which does not mean it is unimportant).

CASE STUDY 2.1: FINDING THE RIGHT LITERATURE My PhD thesis explored the gendered dimensions of housework in shared households. An initial search turned up almost no literature. But when I thought carefully about study, I realised I wasn’t interested in shared households so much as I was interested in the gendered dimensions of housework. Shared households were a subset of non-traditional household structures in which housework was (not) done. When I thought about the topic

30 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3002_WAL_SRM3_20170_TXT_SI.indd 30 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

in this light, I found relevant studies among the old communal living literature (largely from the 1970s), as well as more recent work on gay and lesbian relationships, couples who are separated, and those who are cohabiting. This work, as well as studies on gender and housework in marital households, formed the basis of my literature review.

Other researchers may need to narrow the literature available to them. When starting her PhD, Dr Emily Bishop wanted to look at young people’s sexual practices. This is an issue that has been framed up as a social, policy, and theoretical problem, and Emily was faced with masses of studies. But when she thought carefully about where her interests lay, she realised she wanted to explore how young people made sense of their sexual practices and, in particular, how they understood risk. Clarifying what it was she wanted to research helped her limit what she needed to read.

A range of useful social science databases is listed in Box 2.2. These databases can be located directly on the internet, or via university library websites. Librarians are an excellent resource for learning more about literature searches and accessing databases. Often, university libraries will have already collated relevant resources in subject-specific guides and indexes, which can provide a good starting point.

BOX 2.2: RELEVANT DATABASES MULTIDISCIPLINARY DATABASES • JSTOR • Web of Science • Current Contents Connect • ProQuest • Ingenta • Australia/New Zealand Reference Centre • Google Scholar

SOCIAL SCIENTIFIC DATABASES • International Bibliography of the Social Sciences (IBSS) • Project Muse • Australian Family and Society abstracts (FAMILY) • Australian Public Affairs Information Service (APAIS) • Aboriginal and Torres Strait Islander Health bibliography (ATSIhealth) • Gender Studies Database • Sociological Abstracts

EVALUATING SOURCES When you conduct a literature review, you will be faced with an array of sources of information. Each brings something diff erent to your literature review. A review based on one source only is likely to be both bland and limited. You will fi nd it helpful to explore the following types of sources.

PEER REVIEWED ARTICLES IN SCHOLARLY JOURNALS Th ese are research articles that have been read and commented on by other academics—in other words, they have been quality controlled. Th ey provide full details of the research, including the methods and research questions, as well as fi ndings and conclusions.

3131CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3102_WAL_SRM3_20170_TXT_SI.indd 31 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Th ey are an important resource for students because they present legitimate research and discussion on how to interpret that research. However, the length of the peer review process and publishing schedules mean that their discussion of social problems can sometimes be a little out of date.

CONFERENCE PROCEEDINGS Th ese are a bit like edited books, created from papers presented at gatherings of researchers and practitioners. Many conference proceedings are peer reviewed and published in conjunction with, or soon aft er, the conference. Th ey can include new research, but may not be easily accessible.

MONOGRAPHS AND EDITED BOOKS In this context, monographs are scholarly books on a particular issue, and edited books consist of chapters by a number of diff erent authors. Th ey are sometimes, but not always, peer reviewed. Th ey can be a useful source of information and analysis, but are not always up to date.

TEXTBOOKS Th ese are written with the aim of making information accessible to people who are learning about a subject. Writers develop summarised and simplifi ed accounts of research, which may be incomplete or lack the nuance of the original report. While textbooks can be a useful place to start reading about a new topic, researchers must move beyond them to understand the complexities of an issue.

MEDIA PIECES Th is category includes articles, opinion pieces, editorials, interviews, and radio and television reports, none of which is peer reviewed. While some include descriptions of social issues and social research, others are notable for presenting incorrect, discriminatory, or hatemongering commentary. Work from the popular media is perhaps most useful in illustrating the social or political context of your topic.

RESEARCH REPORTS Not all research is conducted by academics in universities, and not all research is published and easily accessible in books and articles. Other institutions also conduct and report on research and this work can be original and present recent data. Some have strong links to academics and subject reports to peer review—the Australian Housing and Urban Research Institute is one example of this. Research reports are also produced by government departments (for example, the Australian Institute of Health and Welfare or the Australian Bureau of Statistics), and are subject to internal review processes. Other work is not subject to review processes and may be diffi cult to access—this is oft en termed grey literature. Market researchers, think tanks (for example, the Australia Institute or the Melbourne Institute), government departments, and social organisations (for example, Anglicare, Brotherhood of St Laurence) produce grey literature. Th e quality and rigour of all of these sources varies. Traditionally, they have not played a large role in literature reviews, but their increased dissemination and political relevance in recent years means they are becoming a very visible element in the research landscape.

32 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3202_WAL_SRM3_20170_TXT_SI.indd 32 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

INTERNET SITES Th ere is a broad array of sources on the internet, and these overlap with all of the sources listed above. Th e information is accessible and oft en reports on very recent issues. Some sites are electronic versions of social scientifi c research (for example, the ABS or journal sites), but others—for example, interest groups, blogs, discussion forums, or online journalism—are not subject to peer review; the information they provide should be treated with caution. Some present personal opinion or report on research that has not necessarily been conducted rigorously, while others may off er you new ways of thinking about your topic.

BOX 2.3: USEFUL INTERNET SITES There are thousands of research-orientated internet sites that may be useful for you as a student. Here are some Australian sites to help you begin your searches. • Australian Policy Online www.apo.org.au. A useful collection point for new Australian

research, but be aware that not all the reports are peer reviewed, and some are more rigorous than others.

• Australian Institute of Family Studies www.aifs.gov.au. Families and relationships research from Australia.

• Australian Institute of Health and Welfare www.aihw.gov.au. Data and discussion papers on health and some social services.

• Australian Institute of Criminology www.aic.gov.au. Data and discussion on crime and criminal justice issues.

• Australian Bureau of Statistics www.abs.gov.au. Statistics on almost every component of life in Australia.

• Australian Housing and Urban Research Institute www.ahuri.edu.au. Peer-reviewed research on housing and urban planning issues.

• Australian Parliamentary Library www.aph.gov.au/About_Parliament/Parliamentary_ Departments/Parliamentary_Library/pubs. Research and discussion papers prepared to support the work of senators, members, and other parliamentary staff.

• The Society Pages http://thesocietypages.org. This is the entry site to a collection of sociology blogs and sites written by academic sociologists.

Even though evaluating available literature is an important task, it has few clear rules. It is not possible to say ‘You can always rely on peer reviewed articles’, or ‘Th ink tank reports can never be trusted’. You will fi nd this yourself as you read ‘legitimate’ research that doesn’t seem to be very rigorous, or fi nd an internet site that is full of relevant information and carefully designed research projects. So, rather than following hard and fast rules, it can be useful to consider these questions. • Is the work peer reviewed? Th is process is an important part of ensuring the rigour of

the research. • Who is the author of the work? Is the author known in your fi eld? What are the

author’s institutional affi liations? What are the aims of, and who are the personnel in, the institution or organisation? Organisations vary in their motivations for conducting research, which will shape what is studied, how it is studied, and what claims are made of the fi ndings. You need to think about the implications of such diff erences (but not prejudge the research on the basis of them).

3333CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3302_WAL_SRM3_20170_TXT_SI.indd 33 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

• Who is the author writing for? Consider how the audience may have shaped the project design and interpretation of fi ndings.

• What is the research design? Is it clear from the report? Are the methods and methodologies consistent, and do they generate the kind of data necessary to answer the research questions? What are the assumptions informing the research? No piece of research is perfect—what are the limitations and strengths of the research? How have the research questions and aims, methods, and analysis shaped the fi ndings? Are fi ndings and conclusions based in the data?

• Has the author referred to other works? Do they show an understanding of the broader context within which their work is located? Th e ways in which these are used will diff er according to the type of report. A technical report by the Australian Bureau of Statistics (ABS) will use references very diff erently from the way a monograph uses them.

• Is the material dated? Th is doesn’t necessarily mean that the source is irrelevant, but you must consider the implications of using old data given the sometimes quite dramatic and rapid changes in the world. Research conducted twenty years ago occurred before the terrorist attacks of 9/11, the invasions of Iraq and Afghanistan, the global fi nancial crisis, Rudd’s apology to the Stolen Generations, Australia’s fi rst female prime minister, and the USA’s fi rst African American president. Twenty-year-old research is likely to have identifi ed social attitudes very diff erent to those found today.

• Are the data presented comparable to other data on the issue? Research won’t generate exactly the same fi ndings, because each project is going to be diff erent, but there should be comparable results from comparable studies.

EXERCISE 2.2: EVALUATING SOURCES Find an example of each of the sources previously described. Subject each to the questions listed above. 1 What other indicators might be used to evaluate research? 2 What sources are most appropriate for your own research purposes?

MANAGING YOUR REVIEW Writing a literature review is easier if you have systematically managed your information right from the very fi rst search. Programs such as Endnote are excellent for retrieving and managing your reference database, but they are not necessary. It can also be eff ective to develop a simple template. Regardless of which option you choose, you need to have discipline and diligence, and to make a commitment to yourself to record as you go, every single time.

Ideally, you will keep a record of the following types of information: 1 key search terms you’ve used, so that you don’t repeat them 2 full bibliographic details of each work you read 3 details of each work, including key words, abstract, research questions, problems and

aims, methods, key fi ndings, arguments, conclusions 4 a list of works that look relevant and that you plan to track down.

34 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3402_WAL_SRM3_20170_TXT_SI.indd 34 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Th is information should be recorded with a view to its use in your literature review. If you make a habit of noting the page numbers on which you found not only relevant quotes, but also any type of key information, this will help you reference appropriately when you are writing a literature review, and will assist you in revisiting an author’s ideas. As you read, make notes, but limit the quotes you transcribe. Importantly, insert your own thoughts and responses to what you read: describe what you think about the claims made for the research, its limitations, its signifi cance within the discipline and for your own research project, and its relationships to other material you are reading.

Most importantly, actually read the materials you fi nd. Faced with a huge list of references and a tight deadline, it is very tempting to read only the abstract and conclusion, to skim the work, or highlight almost everything, or copy out entire passages. But the main purpose of reading is to develop a deep understanding of your subject area and fi nd tools that will be helpful in your research. You need to make time to read the work and review your notes on it, so that the issues are more likely to become embedded in your memory and linked with other ideas.

Deciding when to stop reading can be diffi cult. You are not required to read and report on everything written on your topic—you are creating a context, not a catalogue. As a rule of thumb, when your reading is revealing nothing new in terms of general fi ndings, discussion, conclusions, and concepts and theories, then it is time to begin writing in earnest.

WRITING THE LITERATURE REVIEW By the time you come to write your literature review, you will already have a lot of words and thoughts to work from. You need to be selective in terms of what you write about and how you write about it. When weighing what to include, ask yourself: Does the reader need to know this to understand my choices in the research plan?

You are aiming to demonstrate your knowledge of an area by identifying relevant information, then synthesising and evaluating it. Crucially, a literature review is about critical quality rather than simple quantity. You cannot present a critical discussion of the literature by describing books and articles. A good literature review teases out the key themes and issues within a body of literature. To do this you must organise information in ways that relate to your own research question, evaluate what is and is not known about an issue (which will include empirical data and the core methods, methodological, conceptual and theoretical approaches) and explore any issues of controversy and contradiction. Your literature review should also be organised by an argument, so that the reader is in no doubt about where you are heading in the discussion and why you have included the issues and references you have. By the end, the reader should have a strong sense of the subject area, and where your proposed research fi ts within it.

Your ideas should be expressed in your own words. Too much quoting suggests you are not confi dent enough in your understanding of the issue. Sometimes people write something so well it can be hard to fi nd a way to put it better, but the point is to let the reader know how you are interpreting others’ work for your own research project. Your ideas should also be presented in a respectful tone, even when you disagree with an author’s conclusions or are not convinced of the rigour of their work. Avoid qualifying

Literature review: A critical analysis of the existing theoretical and empirical literature, related to our research topic.

3535CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3502_WAL_SRM3_20170_TXT_SI.indd 35 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

statements such as ‘It may be the case’, ‘probably’ and emotive language, such as ‘Th is is a shocking and outrageous misinterpretation’. Also, avoid oversimplifi cation. Most social science topics are complex, and a review will oft en reveal contradictory literature. It is your role and challenge as a social researcher to fi nd a way of making sense of those tensions.

A word on plagiarism: do not do it. You must acknowledge others’ work—not just their exact words but their ideas as well. Appropriate and accurate referencing is an expected academic practice.

METHODS Once you have decided on what you are studying and why, your attention turns to the how. Th is involves decisions about the methodological basis of your study and the methods you will use. Methodological issues are covered in Chapter 1, but for the purposes of this discussion it is important to remember that your methods need to fi t the methodological framework you have adopted.

SELECTING A RESEARCH METHOD So how does a social scientist go about selecting a research method? Th ere are three key points to be made here.

1 UNDERSTAND A WIDE RANGE OF SOCIAL RESEARCH METHODS Just as in a restaurant, in order to choose the dish of your preference you fi rst need to know what is on the menu, so it is with selecting a social research method: to be able to eff ectively decide on an appropriate research method, you fi rst need to have a good understanding of the range of social research methods available and the key elements that they can bring to your research project. Otherwise, you might select a method that, while adequate for your research purposes, is not the method you would have chosen if you were aware of other, perhaps more appropriate, options.

Again using the restaurant menu analogy, you might just keep on selecting the same dish or research method every time because you know about it, you’ve tried it before and it’s worked, without considering whether another might suit you better. While such an attitude might produce reasonable quality research, you will never know what it is, from a research perspective, that you are missing. Th is point does not mean you need to be an expert in all research methods, but you should hold a basic working knowledge of a broad range of research methods, which is the key objective of this text.

2 ALL RESEARCH METHODS HAVE STRENGTHS AND WEAKNESSES All social science research methods are valid—all have strengths and all have value. As importantly, all social science research methods have weaknesses. Th ere is no best method, or even best category of method. All researchers develop preferences but it is important not to be partisan about your method. As noted in the previous chapter, triangulation of methods, oft en combining qualitative and quantitative method, is increasingly used in the social sciences.

36 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3602_WAL_SRM3_20170_TXT_SI.indd 36 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

RESEARCH QUESTION

CONSIDERATIONS What is the best method for answering this research question?

PRACTICAL PERSPECTIVE

RESOURCE PERSPECTIVE

SOCIAL SCIENCE

PERSPECTIVE

SOCIAL SCIENCE RESEARCH METHOD

3 SELECT THE RESEARCH METHOD TO SUIT THE RESEARCH PROJECT Th e third important point to emphasise is that you must select the research method for the project, not adjust your project or question to suit a selected method. Th e research method is a tool of the social research project. Remembering this key point will help you avoid the trap of basing your research project around a method or, more problematically, seeing the research method as the research project. Using your selected research method to collect data is just one part of the overall research endeavour.

MOVING FROM YOUR RESEARCH QUESTION TO THE RESEARCH METHOD As shown in Figure 2.1, the starting point in selecting your method is the research question. Your choice is guided by the essential aim of selecting the best research method to answer the question you are asking; however, there are several other considerations that mediate this process. In selecting the best method for your research question, consider these points. • Social science considerations: Which methods will provide the most rigorous and valid

way of answering this research question from a social science perspective? • Practicability considerations: Which research methods are feasible for this particular

research question and for you as a researcher? Researchers and participants in a study work under resource, skills, and personal constraints, which need to be acknowledged if a research project is to be successfully completed. Remember also the wide range of social research methods available and that modules

on additional qualitative and quantitative methods are included on this book’s website at www.oup.com.au/orc/walter.

Figure 2.1: Selecting your social science research method

3737CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3702_WAL_SRM3_20170_TXT_SI.indd 37 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

VOICES IN THE FIELD: PATRICIA SHORT

Patricia Short is a senior lecturer in sociology at the University of Queensland, and director of the Bachelor of Social Science program. Her research interests are broad- ranging but focus upon vulnerable householders in contexts of economic and political change in developing and overdeveloped worlds. Recent research in the Australian context has concentrated on the housing experiences of low-income householders in the private rental sector. In conjunction with postgraduate research students, she currently is engaged in studies of vulnerable householders and urban change in Australia, and studies in Bangladesh and Vietnam centred on gender, social change, and livelihoods. Patricia, an advocate of mixed methods research, pursues her interest in fundamental methodological issues in the social sciences by working with colleagues and students across disciplines.

From large-scale study of the impacts of natural disasters to indepth analysis of household relations and kinship economies to studies of low-income households in the housing market, and families in the developing world devising strategies to sustain their daily lives, my research has highlighted the value of understanding and respecting the standpoints of vulnerable householders. Through my work, I have developed a commitment to critical theory and an interest in mixed methods approaches to sociological analysis.

The analogy of a hot-air balloon ride that Stones (1996: 77) uses to describe a process of social enquiry that moves between macro- (and necessarily more abstract) and micro-level (closely contextualised) views of the social world rings true for me because it draws attention to how we might gain different perspectives on the same subject by employing (and mixing) a range of research methods, and also to the connectedness among them. Like the hot-air balloon ride, a journey of social enquiry requires careful, expert preparation, it demands well-informed, astute decisions along the way, and a reflexive awareness of one’s position in relation to others and the context of the journey. Moreover, to persuade onlookers that either journey is worthwhile, we will have to document our path, provide evidence to assure them of our competence and the reliability of our mode of travel, and recount the experience in detail, providing snapshots to demonstrate the insights the journey affords.

Imagining a hot-air balloon ride provides a way of understanding the value of both viewing the social world from afar (as we do when using large-scale, historical comparative or quantitative methods of analysis and statistical theory to describe our subjects) and moving in close to take in the sights, sounds, and odours we observe, the conversations and stories we hear, and the emotions we feel on the ground (as we do when adopting smaller-scale, qualitative, interpretive approaches). Mixed methods approaches enable us to reveal how patterns we observe from afar are produced by the people who animate them, how the experiences that motivate (or constrain) people to persist with, preserve, change, abandon, or succumb to their everyday/night worlds constitute such patterns. In the broader view, we can see the sum total, in close-up, what is rendered invisible from afar can be brought into the picture.

Exercising my sociological imagination is what drives my research, applying clear rules of sociological method is what disciplines it. Both, I believe, are essential for worthwhile social research.

38 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3802_WAL_SRM3_20170_TXT_SI.indd 38 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

DEFINITION AND MEASUREMENT DECISIONS Th e focus of research is clarifi ed not only through a research question, but also by defi ning what it is exactly that you are going to study, and what is the best way to measure that concept or variable. Th is process generally involves conceptualisation, operationalisation and measurement.

CONCEPTUALISATION Conceptualisation means the clear and unambiguous defi nition of the key concepts in our research. Babbie (2011) points out that many of the issues that social scientists study don’t exist in material form. While people experience poverty and gender discrimination, no one can point to an object that is poverty, prejudice, cosmopolitanism, or gender discrimination. Th ese are concepts. Concepts are constructs that do not have an intrinsic or fi xed meaning. Th is can sometimes be diffi cult to remember, particularly if you are using terms that are part of our everyday language (such as ‘poverty’, for example, or ‘gender’, or ‘housework’). But as you become more familiar with your topic, it becomes clear that social scientists have diff ering takes on how to defi ne almost any element of social life. It is important to be absolutely clear what it is you mean by a particular concept for this particular piece of research. A nominal defi nition of exactly what you mean by the particular concepts you are using can help provide this clarity.

Conceptualising something can entail breaking it down into elements, which are called dimensions. De Vaus (2001: 25) suggests three steps for arriving at a nominal defi nition. First, gather a range of defi nitions in the course of your literature review. Second, adopt a defi nition or draw together elements from various existing defi nitions. Th ird, delineate the dimensions and think about which are relevant to the current study.

Th e place to start conceptualising is by defi ning the key concepts in our research question. If, for example, your research question is: Is smoking during pregnancy more common among teenage than older mothers?, you need to fi rst defi ne what you mean by your key terms. In this case, the key terms are ‘smoking during pregnancy’, ‘teenage mothers’, and ‘older mothers’. ‘Smoking’ can be defi ned as smoking cigarettes on a regular basis throughout the pregnancy. But what is a ‘regular basis’? Based on your literature review you might decide that this means smoking more than fi ve cigarettes in a week for at least half of the pregnancy. ‘Teenage mothers’ is also pretty straightforward: mothers aged from 13 years to 19 years inclusive. But as pregnancy is a nine-month process, what happens if they are 19 when they become pregnant but are 20 years old when they have their baby? ‘Older mothers’ is more problematic. Th is comparative group is older than teenagers, but how much older? Th e answer depends on what you decide it should be, based on your literature review, other studies, and conventions. If you found that other comparative motherhood studies with an age dimension tended to class older mothers as 30 years and above, this might be a good defi nition to adopt. Also keep in mind conventions of data sources, such as that of the Australian Bureau of Statistics, to report data in fi ve-year age groups, such as 19–24 years and 25–29 years. For the purposes of your project, you decide that ‘older mothers’ means aged 30 years or more at the time of their baby’s birth. So, to summarise, the core concepts are conceptualised as • ‘smoking during pregnancy’: smoked fi ve cigarettes or more per week for at least half

of the pregnancy • ‘teenage mothers’: mothers aged below 20 years at the birth of their child • ‘older mothers’: mothers aged 30 years and above at the birth of their child.

Conceptualisation: The process of defining concepts that are important in the research.

Concepts: Concepts are components of theories that can be drawn from theory or developed through induction from the data. They describe key aspects of the processes or patterns examined in the research.

Nominal definition: A definition of exactly what we mean by the particular concepts we are using.

3939CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 3902_WAL_SRM3_20170_TXT_SI.indd 39 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

OPERATIONALISATION, OR MEASURING Operationalisation, or measuring (terms most used in quantitative research), is about deciding how to identify and measure concepts and the means we will use to determine their presence or absence. How we operationalise depends again on the type of study we are undertaking. In quantitative work conceptualisation, operationalisation, and using determined measures while collecting data are strictly applied and guided by rules specifying how data can be measured and how they are linked to concepts. Th e process is necessary for each variable. In qualitative work, decisions may be made about operationalising the diff erent concepts as they emerge during data collection and analysis.

As with all elements of the research process, deciding how to measure is a creative enterprise. Sometimes you will create your own indicators and measurements and sometimes you will use pre-existing ones. Th e tack you take will depend on your research question and the existing body of knowledge. Adopting existing indicators can be very useful: there is no point reinventing the wheel, and using agreed-upon or pre-existing indicators means your research can be compared to that of others. But if the topic has been understudied, or studied through diff erent methods, if you have identifi ed signifi cant weaknesses in existing approaches or your data set does not have the variables you need to adopt existing defi nitions, then you may need to develop your own indicators.

Regardless of your approach, you need to fi t your measure to the conceptual defi nition you have adopted. In the example above, you were to gather the data needed to answer your research question by surveying all new mothers at a busy maternity ward over a period of twelve months. You might operationalise your concepts by dividing the mothers into two groups by age, as per the conceptualisation, and then count them as either smokers or non-smokers during pregnancy, again, according to your conceptualisation of that concept. To answer the research question, you would then compare the number of smokers among the group of teenage mothers with the number of smokers among the older mothers. Similarly, if you were to conduct a follow-up qualitative study to try to gain an understanding of why some women continue to smoke during pregnancy and others do not, in your interviews with new mothers you would still need to classify their level of smoking activity before asking more indepth questions about their smoking or non-smoking decisions and behaviours.

THE THEORETICAL DIRECTION: DEDUCTIVE OR INDUCTIVE? As detailed in Chapter 1, thinking through the place of theory in your research project is important to the overall design, your aims, and your research questions. In some cases, the theory is developed fi rst, and the researcher then collects empirical data to test if the evidence supports the theory. Th is order—theory followed by data—is referred to as a deductive theory development. A social scientist, for example, might deduce from a theoretical study of the class system that children of working-class parents are less likely to attend university than those of the middle class. Th e researcher would then undertake an analysis of the parental background of university students to test the veracity of this theory.

In other cases, and perhaps more commonly, it works the other way: fi rst a pattern in social phenomena is identifi ed, and a theory is then developed to explain that pattern.

Data set: The entire set of observations recorded in the course of data collection.

Deductive theory: Deductive theory is a way of developing theory that begins with the idea (theory) and proceeds to collect data to test the validity of the theory.

40 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4002_WAL_SRM3_20170_TXT_SI.indd 40 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Deductively developed theory

Data

Inductively developed theory

Th is order—empirical data to theory—is referred to as inductive theory development. Using the same example, an analysis of the demographic background of university students might reveal that students from families where the father has a blue-collar occupation are signifi cantly underrepresented within the university’s undergraduate population. A theory is then developed to explain this social phenomenon.

In practice, the trajectory does not always follow the precise, preset path of models. Rather, the sequencing is oft en neither defi nitively inductive nor deductive, tending to a more circular process. Th eory development leads to empirical data collection, which in turn leads to a redefi nition or refi nement of the original theory, which oft en again leads to more data collection, and so on.

Inductive theory: Inductive theory is a way of developing theory that begins with the identification of a pattern in a social phenomenon, and then proceeds to the development of a theory to explain that pattern.

Figure 2.2: Developing theory: inductive or deductive

TIMELINES Research, as with most other activities in our lives, operates within time periods. Whether you are an honours student, a PhD candidate, or an internationally recognised expert in your fi eld, your research will always need to be planned, implemented, and completed within a set period of time. In many cases, that period of time is strictly regulated. Th e research must be completed and the fi nal thesis or report submitted by the deadline. To manage your research within the time allowed, you need a tightly organised and feasible timeline. Developing timelines requires breaking down the research into discrete tasks, which can in turn give a sense of exactly what needs to be done and if the project can be completed in the time available. It will also help to plan for upcoming tasks, and gives an early indication if things aren’t going according to plan. If you are working in a team or with a supervisor, a shared timeline lets people know what to expect and when they need to make time available to the project.

CASE STUDY 2.2: AN EXAMPLE OF A TIMELINE The following timeline was developed for a funded project that investigated the relationship between the payment and receipt of child support and housing outcomes of resident and non-resident parents. In this project, the initial work around aims, research questions and methods had been completed for the funding application. In the end, recruiting

4141CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4102_WAL_SRM3_20170_TXT_SI.indd 41 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

participants took far longer than the research team envisaged and a contract extension was needed because the extra time needed for data collection bumped out all the tasks that relied on the data.

Table 2.1: An example of a timeline

Task 2013 2014

M A M J J A S O N D J F

Literature review Positioning paper Statistical data analysis Develop interview schedule Ethics clearance Pilot interviews Project status report Recruitment and interviews Transcription Analyse interview data Final report Research bulletin

Students need also to remember to check that their own timetable matches the availability and expectations of their supervisors. Acknowledge that some elements of the process are out of your control, and build in some leeway; recruitment and analysis can be unexpectedly time consuming. Making changes to ethics applications also bites into the available time, especially if you are required to resubmit the application to a committee that meets only once a month, for example. Do not plan to have everything fi nished on the last day: leave time for editing, last-minute changes, and space for things that do not run according to your schedule.

RESOURCES Developing a research budget can be crucial. For academics and professional researchers, budgets are a required element of research proposals. Funding contracts are negotiated in accordance with institutional costs, industry standards, and competitiveness in mind. Th is level of detail is not necessary for students, but estimating the cost of the project is still vital. Knowing likely costs in advance will help you decide if your plan is doable or if it will need to be changed to conform to the resource constraints you are working under. It may be that, in the end, it will cost too much to travel to thirty interview sites, and then have the interviews professionally transcribed, but you may be able to aff ord to do twenty. Such calculations require a back and forth between practical considerations and the aims and methods of the project. If conducting fewer interviews means you can’t answer your research question, you may have to alter your question or, alternatively, fi nd a less expensive way of getting interview data (telephone rather than face-to-face interviews, for example, or doing your own transcribing).

42 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4202_WAL_SRM3_20170_TXT_SI.indd 42 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

As with timetables, talk about the likely costs with people who have worked in comparable projects, as well as with your supervisor. List all of the project elements, and be realistic about what they cost and if you can aff ord to meet those costs. Also make sure you are familiar with the resources available to you as a student: it might be the institution covers photocopying, or telephone calls, or travel, all of which can aff ect what you can cover in your project.

DESIGN IN PRACTICE: THE RESEARCH PROPOSAL A research proposal is the outcome of pulling together all elements of the research design. It is a substantial written document that details and justifi es the proposed research project, and is usually submitted for review and approval by an institution or funding body. It is a requirement of honours and research higher degree level study, and forms the basis of most applications for research grants and contract research. Research proposals require you to set down a clear statement of your proposed research and your plan for proceeding. Th ey clarify points you might have taken for granted and show up any ‘gray areas or loose thinking’ (Kelly 1998: 114). Well-written proposals ensure that the whole project hangs together. Th ey also act as a reminder of what you are trying to achieve, which is useful when you get lost in the fog of data collection or analysis. In short, research proposals are as important as the fi nal written documents arising from your study.

GETTING STARTED Take note of the institutional requirements for the research proposal: the word-length limitations, necessary documentation, and due dates. Also take note of where the relevant people will be when the proposal is due—it can be a nerve-wracking experience to chase a signature when no one knows where the relevant person is or if they’ll be back before your submission deadline. Th is information will help you timetable your work plan for this initial component of your research, and ensure that your project is not rejected because you did not cross your t’s and dot your i’s. It can also be useful to read other research proposals, to get a sense of the language used, and to learn what format and level of detail is required. Finally, think about your planned research. A research proposal can take a lot of time to write, but if you have carefully thought through your research design your groundwork is already done.

GENERAL PRINCIPLES Th e structure of a research proposal and the information it includes will diff er according to institutional requirements and level of study. However, a set of general considerations is relevant regardless of the context. It is important to provide as much information as you can. Regardless of the headings you are required to work under, Silverman (2005: 144) uses Marshall and Rossman (1989: 2) to describe the questions a research proposal must answer. Th ese provide a useful focus, irrespective of the headings you are required to work under. • What is the research about and what is it trying to achieve? A proposal must clearly

state parameters of the project by specifying the research questions, problems, and hypotheses (if relevant). It also needs to specify its aims.

• Why should anyone be interested in the research? Th e research proposal must justify the time, eff ort, and other resources that will be expended on the project. It needs to be clear that the research has a relevance and importance beyond the immediate interest of the researcher.

Research proposal: A formal written document that fully details all aspects of the research you intend to undertake. A research proposal is usually required before academic research can proceed.

4343CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4302_WAL_SRM3_20170_TXT_SI.indd 43 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

• Is the research design credible, achievable, and carefully explained? Th e proposal must show that the project is coherent and the elements are logically related. It must also provide detail on the methods and methodology that informs the project: are they appropriate and are they realistic?

• Is the researcher capable of doing the research? A researcher does not necessarily need to have existing expertise in the proposed methods, and the project might off er a valued opportunity to develop the relevant skills. But it should be clear from the proposal that the researcher has a full understanding of the implications of undertaking the project and the skills and resources that will be needed. Write clearly, avoid jargon, and try to avoid presumptions about what your readership

should know. Th e proposal may be judged by people who are not experts in your fi eld. Th is does not mean they are not able to critically and usefully read your research plan and off er insights, but it does mean you should not take their understanding of key issues for granted. (Th is is not a bad thing, as it militates against your own presumptions undermining the rigour of the research.)

Be honest and realistic about the scope of your project and the issues it raises. Th e research proposal is not the blueprint of a perfect study. It is written to defi ne the project and show that the researcher will be able to carry out the project. Every piece of research has its strengths and weaknesses, being clear about the limits of your proposed research indicates you have thought carefully about what you plan to do, and why. Avoiding the issues may be read as a failure to understand the implications of your approach.

Selling your project is also important. A research proposal should be more than a statement about what you are going to do. Be persuasive about the value of the proposed research, our chosen research topic oft en comes to mean a great deal to us, but we cannot assume that it is self-evidently relevant and exciting. You need to make an argument to convince the reader that your project is worthwhile and doable. Th is is most obviously the purpose of any discussion justifying why—why this question or why these methods—but it should also be evident in the insight you show throughout the research proposal, and in the way it hangs together.

FROM METHOD TO PRACTICE STRUCTURE OF THE RESEARCH PROPOSAL While usually set out under headings, the proposal should be written as a coherent whole. The sections should relate to each other and to an overarching argument about the value of the proposed research. Avoid repetition, and make your points in a logical order.

Research proposal formats and headings are not consistent across institutions, but all tend to have some combination of the following (not necessarily in the following order):

TITLE The title succinctly and clearly sums up the project so that its focus is evident to a non-expert reader. Avoid puns, jargon, and very long titles. In the words of Punch (2000: 69), it ‘should convey as much information as possible in as few words as possible’.

INTRODUCTION This is a concise preview of the key issues of the research project. Write the introduction that will appeal to the reader, spark their interest. You need to capture the attention of people who might not know a lot about the area or be interested in the topic. That said, don’t oversell and don’t be defensive about the value of the project.

44 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4402_WAL_SRM3_20170_TXT_SI.indd 44 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

RESEARCH QUESTIONS AND AIMS These are explicitly stated near the beginning of the proposal so that the readers know exactly what you are proposing to do and can read the document with this in mind. There should be no mystery or confusion in the mind of readers. Remember, the research questions and aims need to be fully justified so that the question: Why should we care about this issue?, is answered. Some proposals also require a separate statement of significance, but even if this is not required, your literature review and/or background discussion should contribute to the justification. You may find it useful to be clear about the importance of your project, using wording such as ‘This research is significant because …’.

LITERATURE REVIEW AND/OR BACKGROUND TO THE ISSUE Punch (2000: 43–4) suggests three questions when determining the scope of the literature review 1 What literature is relevant to the project? 2 What is the relationship of the literature to the proposed study? 3 How will the literature be used in the proposal? Ideally, your discussion will make clear how the questions, problems, and methods arise out of the literature.

METHODOLOGY AND CONCEPTUAL FRAMEWORK A proposal includes a statement of your methodology and theoretical and conceptual framework, and how these are related. They should be justified, rather than merely stated, but any justification needs to be brief and written in plain English. A word on your statement of methodology: this is not the time to develop an extended response to methodological debates. Rather, the aim of this section is to make your starting point clear, so that the reader can judge how your initial stance shapes your project.

METHODS Provide a lot of detail in your methods discussion. Do more than state, for example, that you are doing semistructured indepth interviews or bivariate analysis. Be specific on how the methods will be used in this particular piece of research. Be clear about your sample and recruitment processes, conceptualisation, operationalisation, variables and measurement (where relevant), instruments and data collection processes, and analysis. Remember that all projects have their limitations, so it is best to present an informed discussion about what you can and cannot do with your sample and methods.

Qualitative proposals often involve evolving procedures, questions, theories, operationalisation, and conceptualisation as themes emerge through the data. This is acceptable when it is part of the logic of the research design (rather than evidently arising through sloppy thinking and planning). In these instances, it is important to explain the basis of the flexibility and how decisions will be made as the study evolves.

ETHICS Institutions will often ask that you acknowledge ethical implications of the research. Some projects may give rise to difficult ethical issues, which does not mean they will be rejected out of hand. Rather, the researcher should be open about potential issues and their strategies for dealing with them. This clarity shows that you have thought seriously about the issues and ways of protecting participants in the research. Trying to hide ethical challenges by failing to acknowledge them is likely to be read as a lack of understanding and competence to undertake the research.

4545CHAPTER 2: RESEARCH DESIGN

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4502_WAL_SRM3_20170_TXT_SI.indd 45 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

TIMETABLE Be realistic with key deliverables or accomplishments, and focus on what can actually be done in the time rather than trying to impress the reader with promises of delivering so much for so little. This is an important piece of information for people assessing your research proposal because it helps them judge if the project is doable and for funding bodies deciding if they should take a chance on you.

BUDGET Students are likely to be dealing with very limited budgets. The resource implications of this may still be used to judge if the project is workable.

DISSEMINATION OF RESULTS Be clear about how you are going to share your findings. In some projects it may be relevant to explore the possibility of communicating results to the people who took part in a study.

REFERENCE LIST A reference list is often an afterthought, popped in at the last minute. But a research proposal is an academic document and referencing is an expected part of academic practice. Referencing enables the reader to return to your sources to make up their own mind, and gives an appropriate acknowledgment to other authors when you work with their ideas.

SUMMARY In sum, the research proposal needs to be • succinct • detailed • coherent • honest • justified.

46 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

KRISTIN NATALIER

02_WAL_SRM3_20170_TXT_SI.indd 4602_WAL_SRM3_20170_TXT_SI.indd 46 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

CONCLUSION Research design is essential for research that is rigorous and relevant. It can be confusing to think your way through a new topic before you have actually done the research, but doing this groundwork will guide your research project and your own development as a skilled and insightful researcher. Th is chapter has presented the basics but we cannot properly convey the iterative nature of research design. In a book, one topic follows another in a linear fashion, but in real life, things are messier. You will fi nd yourself going back and forth between your research question and aims, conceptualisation and operationalisation, methods and literature. You may start with a clear question that does not change throughout the research design process, but it is just as likely you will rethink what you want to know, what you need to know, and what you can know as you read existing research and come to some hard decisions about your focus and the methods you can use. Sometimes you will foresee these issues and strategically respond to them, but other times your planning will be subject to ‘false starts, blind alleys, mistakes and forced changes’ (Bryman 2012: 16). Th e key is give yourself time to work through the back-and-forth process, balancing fl exibility with the need for structure and boundaries as you develop, and then conduct, your research. You will also learn how to balance your intellectual curiosity with practical constraints. Th ese are skills that aren’t easily taught in a textbook—they require judgment, confi dence, and a willingness to learn from your mistakes, and we all make mistakes, even the most seasoned researchers. But everyone, also, is thrilled when they fi nd and understand something meaningful about the world.

MAIN POINTS

• Th e process of research design is a necessary foundation for successful research outcomes.

• Research design involves generating research questions and aims, undertaking a literature review, conceptualising and operationalising key terms, and specifying methodologies and methods.

• Designing research properly allows the project to hang together, so that each element of the research is informed by the others.

• Research questions form the basis of a project. Th ey need to be limited in scope and empirically answerable.

• Conceptualising and operationalising key terms assists in clarifying exactly what is being studied and how it will be identifi ed in the data collected.

• A literature review provides the context for the proposed study. Ensure that it synthesises rather than describes relevant literature.

• Choose methods with a view to answering the research question. For the purposes of research design, make them as detailed as possible; it is also important to remember that you will sometimes need to be fl exible once the research begins.

• A research proposal is a document that clearly explains and justifi es the research project.

4747

KRISTIN NATALIER

CHAPTER 2: RESEARCH DESIGN

02_WAL_SRM3_20170_TXT_SI.indd 4702_WAL_SRM3_20170_TXT_SI.indd 47 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

FURTHER READING Babbie, E. (2011). Th e Basics of Social Research, 5th edn. Belmont: Cengage Learning. Bryman, A. (2012). Social Research Methods, 4th edn. Oxford: Oxford University Press. de Vaus, D. (2001). Research Design in Social Research. London: Sage. Fink, A. (2005). Conducting Research Literature Reviews: From the Internet to Paper.

Th ousand Oaks: Sage. Punch, K. F. (2000). Developing Eff ective Research Proposals. London: Sage.

REFERENCES Babbie, E. (2011). Th e Basics of Social Research, 5th edn. Belmont: Cengage Learning Bouma, G. D. & Ling, R. (2004). Th e Research Process. Melbourne: Oxford University

Press. de Vaus, D. (2001). Research Design in Social Research. London: Sage. Johnson, G., Natalier, K., Mendes, P., Liddiard, M. Th orenson, S., Hollows, A. & Bailey, N.

(2010). ‘Improving Housing Outcomes of Young People Leaving State Out Of Home Care’, AHURI Positioning Paper. Melbourne: Australian Housing and Urban Research Institute.

Jesson, J., Matheson, L. & Lacey, F. (2011). Doing Your Literature Review. Traditional and Systematic Techniques. London: Sage.

Kelly, M. (1998). ‘Writing a Research Proposal’. In Seale, C., Researching Society and Culture. London: Sage, 111–22.

Marshall, C. & Rossman, G. (1989). Designing Qualitative Research. Newbury Park: Sage. Natalier, K., Walter, M., Wulff , M., Reynolds, M. & Hewitt, B. (2008). Child Support

and Housing Outcomes: Final Report. Melbourne: Australian Housing and Urban Research Institute.

Natalier, K. & Willis, K. (2008). ‘Taking Responsibility or Averting Risk? A Socio- cultural Approach to Risk and Trust in Private Health Insurance Decisions’, Health, Risk and Society, 10 (4), 399–411.

Neuman, W. L. (2006). Social Research Methods. Qualitative and Quantitative Approaches. Boston: Pearson.

Punch, K. F. (2000). Developing Eff ective Research Proposals. London: Sage. Robson, C. (1993). Real World Research: A Resource for Social Scientists and Practitioner-

researchers. Oxford: Blackwell. Sandberg, J. & Alvesson, M. (2011). ‘Ways of Constructing Research Questions: Gap-

spotting or Problematization?’, Organization, 18 (1), 23–44. Short, P., Seelig, T., Warren, C. M. J., Susilawati, C. & Th ompson, A. (2008). Risk

Assessment Practices in the Private Rental Sector: Implications for Low-Income Renters. Final Report No. 117. Melbourne: Australian Housing and Urban Research Institute.

KRISTIN NATALIER

PART 1: THE FOUNDATION OF GOOD SOCIAL SCIENCE RESEARCH

02_WAL_SRM3_20170_TXT_SI.indd 4802_WAL_SRM3_20170_TXT_SI.indd 48 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Sidhu, R. & Taylor, S. (2007). ‘Educational Provision for Refugee Youth in Australia: Left to Chance?’, Journal of Sociology, 43 (3), 283–300.

Silverman, D. (2005). Doing Qualitative Research. London: Sage. Stones, R. (1996). Sociological Reasoning: Towards a Post-modern Sociology. Basingstoke:

Macmillan.

4949

KRISTIN NATALIER

CHAPTER 2: RESEARCH DESIGN

02_WAL_SRM3_20170_TXT_SI.indd 4902_WAL_SRM3_20170_TXT_SI.indd 49 7/08/13 8:52 AM7/08/13 8:52 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

The Research Process DOUGLAS EZZY

TESTING THE TRADITIONAL, QUESTIONING THE ACCEPTED Why are men more likely to be leaders than women? What do you think? Even though

Australia has had its first female prime minister in Julia Gillard, men are still more likely

than women to be ministers in parliament, leaders in business, or senior academics,

and to hold many other leadership positions (Moon & Fountain 1997, Pini 2002, 2005,

White 2003). Is this because men are better at being leaders than women, or is there

some other process at work here?

One social process that causes men to be more likely to be leaders is tradition.

Some people still believe that a woman’s place is in the home. Indeed, for much of

the twentieth century, Australian society was organised around a family structure in

which the father worked and the woman stayed at home and cared for the children.

Until 1966, married women were not allowed to work in the public service in Australia

(Probert 1980). Traditional answers to questions such as this are often based on moral

or religious beliefs; however, these traditions are becoming less important. There are

now a number of women in parliament, some of whom are ministers, some of whom

have young children. Economic pressures mean that few families can aff ord for women

to stay at home (McDonald 1998). More and more women are delaying having children

so they can build their careers, and when they do have children, they are choosing to stay

in employment. In other words, men used to be in positions of leadership because most

women stayed at home to look after the children. But now that women are staying at

work, why are there still more men than women in positions of leadership?

For social researchers the question of why men are more likely to be in leadership

positions has to be answered in terms of social process and cultural assumptions. We

cannot understand the question—let alone answer it—unless we consider the social

context and the infl uence of social processes on our assumptions about the topic. It is the

research process that ensures that we understand what it is we are examining and the

contexts in which these social phenomena occur, as we investigate the social question.

KEY TERMS concept

conceptualisation

ideology

interpretation

measurement

population

qualitative research

quantitative

research

reductionism

reliability

sample

sampling frame

textual analysis

theory

unit of analysis

validity

variable

03_WAL_SRM3_20170_TXT_SI.indd 5003_WAL_SRM3_20170_TXT_SI.indd 50 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

INTRODUCTION A key role of social research is to investigate what is actually going on in our social world. As discussed already in Chapters 1 and 2, social investigation requires rigorous, innovative, scientifi c, and questioning research activity. Social research needs to be undertaken in such a way as to ensure that it meets these prerequisites. Just as the research design enabled us to develop a map of our research project, the research process provides a map for how social research, in general, needs to be undertaken.

In its questioning role, social research looks anew at social phenomena rather than relying on tradition, or accepting social assumptions or beliefs. In the text that opens this chapter, social researchers have looked at some of the social processes involved in determining why there are fewer female than male leaders in our society. Others have taken this question further, asking what evidence there is that women perform better or worse as leaders. Kabakoff (2002), for example, conducted a study of 172 US corporate managers, asking all their employees to rate these managers on their eff ectiveness as leaders. He found that there was no diff erence in the ratings for male and female leaders. Similarly, Eagly and associates (1995) reviewed a wide range of psychological studies of leadership and found that there was no overall diff erence in the eff ectiveness of male and female leaders. In other words, social research results indicate that female leaders are just as good as male leaders.

If women are just as good leaders as men, why are there still more male leaders? When Pini (2002: 45) studied women in leadership in the sugar industry in Australia, she found that elected leaders of the sugar industry claimed that the election process was ‘open, objective and gender neutral’. However, when Pini studied how these same leaders described their own success, they described a system that was closed, subjective, and clearly biased against women. In other words, Pini argues, there are more men in leadership because the system is biased against women.

Th e question ‘What is a “better” leader?’ is also important. Studies suggest diff erences in the leadership style of men and women (Rosser 2003). How do you measure ‘better’? Is a leader of a church with more parishioners than its counterpart church in the next suburb a better leader than a counterpart church with less people? Is it more important that church leaders are trustworthy and compassionate? Is a manager who looks aft er the welfare of their workers a better manager, or is it more important that the company makes a high profi t? What if the company happens to sell chemicals that make people sick when they use them, such as cigarette companies? Does this make the manager better or worse? Th e choice about how to defi ne ‘better’ must, eventually, appeal to some sort of ethical criteria as well as the conceptual criteria aligned with our research (see Conceptualisation, below).

THE PROCESS OF SOCIAL RESEARCH Social sciences, like sociology, politics, or anthropology, involve social research. Social research is the process of examining social life to obtain a better understanding of what people are doing and why they are doing it. Before an engineer builds a dam, that person fi rst examines the rocks and geology of the area in which the dam is to be built. Similarly, it makes sense that before a politician introduces a new program to, say, prevent the spread of HIV/AIDS, that politician fi nds out what people are doing and why they are doing it. It doesn’t make much sense to have a television advertising campaign aimed at heterosexuals in the general population, as the early Australian AIDS programs did, when the people most at risk are gay men, injecting drug users and prostitutes.

5151CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5103_WAL_SRM3_20170_TXT_SI.indd 51 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Many of the more pressing world problems include a signifi cant social dimension, which require social solutions. Global warming, for example, will not be solved by new technology alone. Changes in patterns of behaviour and in cultural beliefs are also required. Zero-emission cars, say, will only be developed and sold widely if policy makers and the general population believe that they are necessary and start to purchase such cars. Th ese changes in beliefs and patterns of buying are cultural and social changes. HIV/AIDS will not be cured by medical advances alone. Policy makers and people in the general population need to change their willingness to talk about safe sex, condom use in particular. To have this conversation is very diffi cult in cultures where talk about sex is taboo. Violent confl ict associated with terrorism will not be resolved by military power alone. Rather, an examination of the cultural beliefs and social conditions that lead people to be willing to participate in terrorism, including economic and political contexts, is required. A more sophisticated understanding of the social and cultural conditions that generate terrorism may help to prevent the development of terrorism in the fi rst place, better than will the attempts to try to crush it once it has arisen.

Th ese issues require social research. Why is it that men are still more likely to be leaders? Why is it that many people still think that global warming is not an important issue? How can we encourage people to practise safer sex and safer injecting practices to minimise the spread of HIV/AIDS? Why do people become terrorists, and what social, political and economic changes might lead them to choose diff erently? Th ese are the sorts of questions that social researchers ask.

Th is chapter has two parts. In the fi rst, a set of issues that all researchers must address as part of the process of social research is explored. Th e second part examines the research process in action through an analysis of these core steps in three recent social research projects: a survey of who does housework in Australia, a qualitative interview study of teenage witches, and a discourse analysis of racism in newspapers.

TEN ISSUES FOR SOCIAL RESEARCHERS Th e process of social research can be usefully divided into ten key issues. 1 Identifying the topic. 2 Focusing on a research question. 3 Specifying the theoretical framework that will inform the research. 4 Developing a research methodology. 5 Delineating the main variables or processes. 6 Defi ning and measuring. 7 Selecting a research sample and collecting data. 8 Addressing the political and ethical dimensions of the research. 9 Analysis of data and interpretation of results. 10 Writing up the results.

Th ese steps are sequential and interrelated, with decisions and developments in one task leading to and informing the next step. Because the practice of social research is a very varied one, not all researchers will need to address all these steps to the same degree or in the same order. While there is variation in how a social researcher addresses each step of the research, skipping or ignoring a step can jeopardise the validity of the whole research project. Some of these steps have already been touched on in the fi rst two chapters. Th e focus of this chapter is the critical importance of how these elements operate as a process,

Variable: A characteristic of data that has more than one category. A variable varies between different values, and is measured in social research, for example, the variable gender usually has two categories, male and female. But other categories such as intersex and transexual could be included.

52 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5203_WAL_SRM3_20170_TXT_SI.indd 52 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

IDENTIFYING THE TOPIC

THE RESEARCH QUESTION

THEORETICAL FRAME

RESEARCH METHODOLOGY

DELINEATING VARIABLES AND PROCESSES

DEFINING AND MEASURING

SELECTING A SAMPLE AND COLLECTING DATA

POLITICAL AND ETHICAL DIMENSIONS

ANALYSIS AND INTERPRETATION

WRITING UP AND COMMUNICATION OF RESULTS

rather than as individual aspects; they all work together to create valid and rigorous social research. Th is sequential nature of the research process is demonstrated in Figure 3.1.

1 IDENTIFYING THE TOPIC Most research begins with a topic deemed important. Th e research question, as outlined in Chapter 2, fl ows from the activities undertaken during the design process. Others have a topic, but may not fully articulate the research question until some way into their research. Th is is a product of the diff erent reasons for conducting social research. When research is funded by an organisation with a specifi c concern then, clearly, the topic is already given. In 2006 WorkCover Tasmania approached me and sociology colleagues with the topic ‘Th e outcomes for longer-term workers’ compensation claims in Tasmania’. We began with a scoping study of the research problem, and from this developed the research question related to the claimants’ longer-term health, social and economic outcomes. With this larger project now complete, a report has been presented to WorkCover Tasmania (Ezzy, Walter & Welch 2009), which in turn has informed the development of workers’ compensation policy. In contrast, some research is motivated by fascination. When I began my PhD research, I was fascinated with the topic of the meaning of life. It was only aft er quite a bit of thinking and writing that I narrowed my research down to a question concerning how people make meaning of their lives when they become unemployed.

Figure 3.1: The research process

5353CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5303_WAL_SRM3_20170_TXT_SI.indd 53 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Th e topic of a research project should be clear enough to specify exactly what it is that you want to research. Th is specifi cation is an essential part of developing your specifi c research question. ‘Injecting drug users’ would be too vague and general. Although it might serve as a starting point, it requires further clarifi cation. ‘Dental hygiene and the health of injecting drug users’ is probably too specifi c, although in certain circumstances it may be appropriate. Th is is not to minimise the signifi cance of this issue, but the health of injecting drug users is shaped by many factors other than dental care, and narrowing the question to this one aspect may limit the usefulness of the research. Similarly, a topic such as global warming is too broad, and requires more specifi cation. A more realistic topic is probably something like ‘Representations of global warming in newspapers’. Th is specifi cation will lead directly to the development of your research question, which, in this case, it might be ‘How is global warming represented in mainstream Australian newspapers?’

2 FOCUSING ON A RESEARCH QUESTION As already outlined in Chapter 2, clearly specifying the research question takes the researcher beyond the topic of interest to nailing down exactly what it is that they want to know. Specifying the research question can avoid many problems later in the research. Research questions that are too general or vague can be very diffi cult to answer, and research projects easily become bogged down when the topic is not translated into a clearly defi ned research question. A PhD student in sociology at the University of Tasmania began her PhD research with the topic ‘What is a good life?’ (Verdouw 2008). Th is is a good sociological research question, but it proved to be a very big question that was not focused enough. Th e student embarked on a qualitative study of the topic, interviewing forty-one young adults about what they understood to be a good life. As she began to analyse her data, Verdouw identifi ed many diff erent aspects of the good life and found it diffi cult to try to discuss them all. She eventually decided to narrow her question down to the role of money in the good life. Her specifi c research question was, ‘How do the orientations of 41 young adults towards what is “good” in life shape their money meanings, and consequently their motives, values, behaviours and social interactions?’ (Verdouw 2008: 8). Th is research question still enabled Verdouw to examine her more general topic of the good life, while at the same time provided a narrower focus that was manageable and practical for a PhD project.

Research questions relate closely to the theory (step 3) and the methodology (step 4) of the research. It is oft en useful to move back and forth between these steps to ensure that the research question, theory, and methods are compatible. Research questions that focus on meanings and experience are typically best studied using qualitative methods and interpretive theory. Research questions that ask ‘How many?’ are best studied with quantitative research methods, perhaps using a survey.

3 SPECIFYING THE THEORETICAL FRAMEWORK Research is like a conversation. Th e researcher talks to what is happening out there in society, and listens to what is said in reply. A researcher might wonder if newspapers are accurately representing what scientists say about global warming. When they do some research on this, the research talks back about what newspapers say about global warming. Th eory can be thought of as a record of previous conversations on the same topic. Research brings the ideas from these previous conversations and uses them to shape the way research is conducted.

54 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5403_WAL_SRM3_20170_TXT_SI.indd 54 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

As outlined in Chapter 1, the theory or paradigm that a researcher uses will shape the way in which the researcher understands the research topic. Marxist theory, which is associated with a confl ict perspective, argues that the primary inequalities in society are a product of economic processes. Marxist-infl uenced studies of global warming argue that coal and oil companies are using their infl uence on governments and policy makers to ensure that global warming is not taken seriously as an issue (Newell & Paterson 1998). In a fascinating study of global warming sceptics, Jacques and colleagues (2008) found that 92 per cent of global warming sceptic books published were linked to conservative think tanks, which have a clear political agenda that serves the interests of the elite and wealthy. If a social researcher were to study newspaper representations of global warming from a Marxist or confl ict perspective, they would probably point out that the leaders of big business own most newspapers, and that it is in the short-term interests of the wealthy and powerful to ignore, or at least defl ect, the global warming issue.

Th eory is an essential part of all social research, but theories can be—and are—changed and developed as a consequence of research. Marx began with a model of two classes in society: those who own the means of production, such as factories, and those who work in them. Later in his life, aft er he had researched these issues in more detail, he developed a more complex model that included other classes, such as people who inherit wealth and people who we would now describe as the middle class.

Also, as detailed in Chapter 2, the placement of theory within social research can vary. Deductively conducted research begins with theories. In the case of Newell and Paterson’s (1998) study, for example, they fi rst deductively developed a Marxist theory that the coal and oil industries have infl uenced government policy to protect their own interests. Th ey then go on to test this theory empirically by examining oil and coal lobbying strategies and how these have shaped particular government decisions.

Inductively conducted research begins with data. I developed an inductive analysis of environmental change by examining four popular journals (two from business, two from science) and comparing the way they described climate change (Ezzy 2001). I found that the business journals were not worried about climate change because the writers seemed confi dent that science would solve any problem that arose. In contrast, the science journals were worried about climate change because they did not think that science would be able to stop global warming once it was started. My analysis started from the data and built up to the theory, whereas Newell and Paterson started with a theory, then tested it against some data. Although one side is usually emphasised, all research contains both deductive and inductive elements.

4 DEVELOPING A RESEARCH METHODOLOGY Research methodology is the frame of reference for social research. As previously outlined in Chapter 1, this frame is infl uenced by our worldview (standpoint), and the paradigm in which our theoretical perspective is placed or developed. Within this defi nition, methodologies can be subcategorised into one of two main types: quantitative (or statistical) research and qualitative (or interpretative) research.

In quantitative research, the main research questions are ‘What?’ and ‘How many?’. A quantitative approach perceives that we can best understand the particular research question being asked by using statistical methods. Th e focus of statistical research is measuring or counting. Knowing the answer to ‘What?’ and ‘How many?’ questions can

Quantitative research: The study of social processes through the collection of data that is amenable to statistical analysis. Quantitative research is often used to identify and establish relationships between research variables. Major quantitative social research methods include surveys, structured interviewing, and secondary data analysis.

5555CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5503_WAL_SRM3_20170_TXT_SI.indd 55 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

explain a lot about how or why people do things. Quantitative research tends to examine social issues from a big picture perspective, gathering larger-scale data that can then be analysed using statistical techniques to establish patterns. In the workers’ compensation research discussed in the last section, for example, we surveyed a representative sample of longer-term workers’ compensation claimants to establish the parameters of their current and previous fi nancial, social and health status.

In qualitative research, the main research questions are ‘What meaning?’ questions. Th e focus is on how people think about and interpret what they are doing. Knowing the answers that people give to ‘what meaning’ questions can explain a great deal about how or why they do what they do. In the second phase of the workers’ compensation research, for example, we conducted indepth interviews with twenty longer-term claimants to delve into the meanings and understandings they had of their experience of the workers’ compensation process. Qualitative approaches tend to investigate social issues from a more micro perspective, concentrating on individual experiences or those of a small group, involving interviews and thematic analysis.

Research approaches, such as quantitative or qualitative, are the theories or general research paradigms that justify the particular research method chosen. If the research question is: At what age do women have their fi rst child?, then it probably makes most sense to use statistical methods to study this question. If you want to know why women are having their fi rst child at an older age than was previously the case, this question could be examined using either quantitative or qualitative methods. Statistics may reveal that more women are working than ever before, and qualitative interviews would reveal the meanings that women are giving to having children. It is important to match the research question correctly with the appropriate method.

Although quantitative and qualitative are the two main research approach categories, there is a variety of other approaches and methods, some discussed in this book—either within the text or on the accompanying website at www.oup.com.au/orc/walter—that do not fi t neatly into this distinction. Th ese include discourse analysis, semiotics, and some forms of content analysis. Further, the distinction between these two types of research is in many ways an oversimplifi cation. It is possible to construct statistical surveys that provide answers to questions about what meanings people give to things. It is also possible that some qualitative research can answer how many questions, particularly if the group being studied is a small one. Nonetheless, the distinction between statistical and qualitative research is a useful way to begin thinking about research methods.

5 DELINEATING THE MAIN VARIABLES OR PROCESSES Once the research topic, question, and methodology are identifi ed, the next question to ask is: What are the main dimensions of the topic? Th e key point is that the identifi cation of the dimensions of the research topic is closely related to the choice of research method, and that these three aspects need to be considered together. A study of the health needs of injecting drug users (IDUs), for example, could focus on how many IDUs have particular sorts of medical conditions. Th e main dimensions would be variables that measure the particular medical conditions IDUs have. One way to examine these sorts of variables would be through a survey distributed to IDUs or to medical practitioners involved with IDUs. Bammer and associates (2002) report HIV status from a survey of injecting drug users who attend needle-exchange clinics. Here, the key variable is whether a person

Qualitative research: Qualitative research is concerned with exploring the understandings, meanings, and interpretations that people or other groups attribute to their social world. Major qualitative methods include participant observation studies and indepth interviews.

56 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5603_WAL_SRM3_20170_TXT_SI.indd 56 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

attending a needle exchange is HIV positive. In this case, IDUs are either HIV positive or HIV negative. Counting up the results from recording this variable, Bammer and associates found that 3 per cent were HIV positive and 97 per cent were HIV negative. Such a fi nding enables a larger picture of the level of HIV infection among IDUs to be developed.

In contrast, other researchers have examined the social processes that generate these sorts of health issues. Why is it, for example, that some IDUs use clean injecting equipment or take medication for medical conditions such as HIV, when other IDUs don’t maintain their health in these ways? Such research could be conducted qualitatively, with long interviews with a smaller number of IDUs. Demas and associates (1995) interviewed fi ft een injecting drug users who had HIV to try to identify what motivated or prevented them taking their HIV treatments. Th ey found that social factors such as stigma, uncertainty, and social support, among other factors, infl uenced their actions. Th ese social factors could be measured as variables. Th e degree of social support could be measured and examined through a statistical survey. Th e main point of the study, though, is not to measure social factors, but to show how the experience of social support, as a social process, infl uences the way that people with HIV feel about taking their medication, and therefore changes their behaviour. Increased social support, reduced stigma, and increased certainty shape the way IDUs feel about themselves, and this in turn increases the likelihood of them taking their medication. Th e focus of this qualitative study is on social processes. Th e research is not a ‘How many?’ question, but a question about what meaning people give to their actions. Why do some IDUs take HIV medications when others do not, and what meanings and understandings shape these decisions?

In summary, the dimensions of a research project can be variables that can be statistically measured through surveys. Th ey can also be social processes that are usually described qualitatively through interviews. Social researchers also use a variety of other research methods, such as content analysis and discourse analysis, to examine dimensions and variables. Th e important point is that the dimensions of the research are worked out in conjunction with the research topic and the method.

CASE STUDY 3.1: SHOPLIFTING Cameron (1964) conducted an inductive qualitative study to identify what meanings people gave to being caught shoplifting. She conducted her research in a large department store, observing the way that different people responded when they were caught by store detectives. From her observations, she inductively identified two main types of shoplifters: pilferers and professional thieves. 1 Pilferers are people who have typically never been caught shoplifting before, so they

do not think of themselves as thieves. When they are arrested, they are told that they could go to court and to jail, which forces them to realise that they are a thief, which typically results in emotional outbursts. Pilferers are also very concerned about what their family and friends will say and think once they find out, fearing they will be ashamed and disgraced.

2 In contrast, the professional thief is someone who has stolen before and knows the risks. Professionals already think of themselves as thieves, and while they may be frustrated at being arrested, they are typically not overly disturbed by the arrest. They are also not worried about the reactions of their friends, who already know about their shoplifting, and they may already have contacts with a lawyer who acts for them. The main dimensions that Cameron identifies in this study are the differences between first-time offenders, or pilferers, and professional thieves.

5757CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5703_WAL_SRM3_20170_TXT_SI.indd 57 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

6 DEFINING, MEASURING, AND OBSERVING Once the main dimensions are identifi ed, the next step is to determine how to defi ne, measure, and observe the variables or processes. As outlined in Chapter 2, defi ning, also called conceptualisation, involves carefully and unambiguously defi ning research variables or processes. Th is specifi cation of concepts is important for both quantitative (statistical) and qualitative (meaning-orientated) research. Much confusion arises in social research, of various types, because researchers do not clearly defi ne the concepts they are using, then how they are measuring or observing these concepts.

For quantitative researchers, once terms have been conceptualised, these concepts are operationalised through survey questions. Two key terms here are validity and reliability. Th e general health questionnaire (GHQ), for example, has been established as a valid and reliable set of questions that measure a person’s mental health (Ezzy 1993). To say that a measure is valid means that it actually measures what it is supposed to measure. If a person has a high score on the GHQ, it probably means that person has some mental health issues; however, the GHQ might not be a valid measure of whether people are delusional. Someone who is delusional might think that everything is wonderful. Th e GHQ is also a reliable measure. Th is means that the GHQ consistently provides similar results for people with similar sorts of mental health conditions. People who are healthy, for example, consistently score well on the GHQ. Questions of validity and reliability are clearly central to the ‘How many?’ questions that quantitative researchers typically ask.

Qualitative researchers prefer to use the term ‘rigour’ instead of ‘validity’ and ‘reliability’ (Ezzy 2002). Meanings and interpretations are oft en complex, and historically and socially located. Th e aim of the qualitative research is to provide a rich and nuanced description of the meanings that people give to their experiences. Rigour refers to the methods that qualitative researchers have developed for ensuring that they faithfully represent the stories and experiences of the people being studied. Elements of rigour include close scrutiny of the detail of the lives of the people being studied, ‘thick description’, a focus on process and subjective meanings, and a tolerance for complex and nuanced explanations (Ezzy 2002: 54). Selecting concepts that adequately represent people’s experiences and meanings is central to conducting rigorous qualitative research.

Each of the strategies for measuring or observing has advantages and disadvantages. Th e important thing is to ensure that the correct approaches are used for the methods and research questions for the study.

CASE STUDY 3.2: WORKING SOLE MOTHERS Maggie Walter (2002) conducted a quantitative statistical study to examine if Australian sole mothers were better off when they were on welfare or when they obtained work. Walter conceptualises ‘better off’ as material wellbeing, focusing on economic issues. She then identifies three measures, or ways of operationalising, material wellbeing. I will discuss the first two measures.

The first measure examines the respondents’ earnings from employment. Clearly, this improves if someone obtains paid work when she previously had not had any paid work, although Walter points out that this is not a very good measure of material wellbeing.

Second is a measure of the total annual household income, including government benefits. Walter shows that if a sole parent obtains part-time work in a low-paid job, such as in sales or a manual occupation, her total household income does not improve. This is a consequence of the loss of various government benefits and the additional costs, such as

Validity: Validity is the extent to which our data or results measure what we intended them to measure.

Reliability: Reliability is the consistency of our data or results. If we repeated the data collection or analysis, will we consistently get the same results?

Interpretation: Data analysis results must be interpreted in light of our conceptual frame, that is, not only what our results indicate, but what they mean in the context of our topic.

58 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5803_WAL_SRM3_20170_TXT_SI.indd 58 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

childcare, that the sole parent has to pay. Household income only improves if the person can obtain higher-paid work, such as professional or administrative work.

The important point here is to see that different ways of measuring, or operationalising, variables have a significant influence on the results of a study. Walter uses the results of a large survey to test whether sole mothers are indeed better off, as measured by these two indicators.

7 SELECTING A RESEARCH SAMPLE A sample is a subset of the population or range of elements that you want to study. To choose a sample, the researcher must fi rst choose who or what is to be observed, that is, your units of analysis. In Nissani’s study of global warming, for example, the units of analysis are newspaper stories (see Box 3.1).

Th ere are many diff erent ways of sampling, and each has diff erent justifi cations. Th is topic, detailed in Chapter 5, is overviewed here to make sense of how sampling fi ts within the research process. Quantitative or statistical research typically seeks to fi nd a representative sample. A statistically representative sample should have the same make-up as the population from which it is drawn. Gurlich and associates (2003) conducted a survey of Australians about their sexual and drug-use practices. Th e sampling method was to dial telephone numbers randomly and interview a preset number of respondents over the phone. Random digit dialling includes people with silent numbers who would be missed if the phone book were used as the sampling frame, and excludes people who don’t have phones at all. While the researchers recognised the problems inherent in this method they considered that it was a better way to conduct their survey than other sampling alternatives. Th ey argue that random digit dialling gives a reasonably representative sample of the Australian population with minimum bias. On the basis of this study, Gurlich and associates (2003: 245) found that 4 per cent of men and 1.8 per cent of women in their survey reported ever having injected drugs, more than twice as many men as women. As the sample was representative of the Australian population, the results can answer quantitative questions about the population from which they are drawn. Th erefore, the authors can be confi dent that when 4 per cent of men in the survey say they have ever injected drugs, that this is also the case for the Australian population as a whole.

Other strategies can also be used to gain a representative sample. For example, there is no list anywhere of the people in Australia who are IDUs, but as most IDUs in Australia attend needle exchanges to obtain clean injecting equipment, sampling people who attend needle exchanges provides a reasonable indicator of the health of IDUs in Australia. Of course, it excludes those who do not attend needle exchanges. Th is is not ideal, but this group of people is even harder to access, and researchers try to take this bias into account in interpreting their results.

Th e aim of qualitative research is to be representative of the processes that are studied. Th ere are a number of ways to do this (Rice & Ezzy 1999). Cases that are typical may be selected and studied in depth. Alternatively, extreme cases may be selected to provide the greatest range. Vincent (2004), in his qualitative study of the health needs of IDUs, selected typical injectors who attended a needle exchange. His study included equal numbers of men and women because he wanted to see if the same social processes were at work for men and women who inject drugs. Vincent does not identify any major diff erences between the way women and men talk about risk, which suggests that similar social processes are operating in both groups. One of Vincent’s most interesting fi ndings

Sample: A sample is a set of cases or elements that are selected from a population.

Population: The collection of all the units that we want to study.

Unit of analysis: A particular instance of what or who we are researching. A unit of analysis can include individuals, groups, social artefacts such as newspaper articles or policy documents, or anything related to social life that can be investigated.

Sampling frame: A list that represents, as closely as possible, an abstract population, the researcher uses this list to draw a probability sample.

5959CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 5903_WAL_SRM3_20170_TXT_SI.indd 59 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

is that when people who inject drugs become socially isolated and depressed, they are unable to look aft er their health.

Samples that are representative of processes are used to answer ‘What meaning?’ questions. Vincent wanted to know what meanings IDUs give to the risks they face and health issues they deal with. Because Vincent studied typical injecting drug users, the processes he describes are likely to be similar for many other IDUs, that is, lack of social support and higher social isolation can lead IDUs to ignore basic health issues such as eating well, exercise and dental hygiene.

Sampling techniques sometimes involve more impressionistic justifi cations. Nissani (1999), in her study of US newspapers, selected four newspapers that she argues are typical of US newspapers. She then found all the news stories that mentioned global warming and analysed them. In this case, while the units of analysis are news stories, the sampling technique involved selecting representative newspapers. Nissani’s argument for the representativeness of these newspapers is not statistical, but rather is justifi ed on the grounds that they are typical of US newspapers. For a range of common sampling techniques used in qualitative research projects, see the section in Chapter 5 that relates to non-probability sampling.

8 ADDRESSING POLITICAL AND ETHICAL DIMENSIONS Social researchers and the research reports they produce are part of political processes. Research is inherently—and unavoidably—political in nature and has ethical dimensions. Research about IDUs can help or harm the lives of IDUs. Research about global warming will involve the researcher in an ongoing political debate about how to respond to this issue.

Even how a research topic or question is framed has political and ethical implications. A research topic might, for example, examine the frequency of dangerous injecting practices by measuring how oft en IDUs share injecting equipment. Th is may be useful information to have. Such a research topic begs at least two questions: Are IDUs sharing needles because they are ignoring advice, or is it because it is diffi cult to obtain clean injecting equipment, as is the case in the USA? It is important to design research in which all aspects of an issue are examined. While the research might be technically correct, with a representative sample and well-constructed questions, it has been designed to examine only a limited part of the issue and, as such, is ethically suspect. Some politicians and policy makers oft en selectively use research to justify reducing the public resources spent on minority groups such as IDUs. However, if the research also examined the availability of and obstacles to access to clean injecting equipment, then the research could also examine what social policies and organisational practices help or hinder safer injecting practices. Service providers and advocates of minority groups have used such arguments to advocate for services that improve the lives of members of these minority groups.

As detailed in Chapter 4, social researchers are routinely required to obtain ethics approval for their research from ethics committees. Many large institutions involved with research, such as universities and hospitals, have ethics committees. Some of the key concerns of ethics committees are that the people who participate in research know what they are participating in (this is called informed consent), that they remain anonymous, and that people are not harmed by participating. Most surveys are anonymous, and people can choose whether to participate. Many qualitative studies are more detailed. For these studies, ethics committees typically require that the participant be provided with an information sheet about the research and sign a statement of informed consent.

Information sheet: A short written document that provides respondents with a summary of key aspects of the research for the purposes of informed consent. This summary includes what the research is about, what participation involves, and any potential risks arising from participation.

60 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6003_WAL_SRM3_20170_TXT_SI.indd 60 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

9 ANALYSIS OF DATA AND INTERPRETATION OF RESULTS Data analysis is the process of analysing the evidence that is produced from the research. In quantitative research, this involves various statistical techniques, some as simple as counting the number of responses and calculating percentages. In qualitative research, this involves identifying patterns of meanings and interpretations. Methods and techniques for data analysis vary hugely (see Chapters 8 and 14 for more detailed discussion of qualitative and quantitative analysis).

Data analysis can be one of the most exciting parts of the research. Analysis can reveal surprising fi ndings that contradict expectations and shed new understandings on your research topic. Cameron’s (1964) realisation that there were two types of shoplift ers is an example of this. It might seem obvious once you discuss it, but only good social research can provide this sort of insight. Similarly, data analysis can be socially valuable when it is demonstrated that there is little economic incentive for Australian sole mothers to obtain part-time, poorly paid employment (Walter 2002). Methods of data analysis vary considerably, depending on the research method, this is discussed in detail in various chapters later in the book.

It is not enough to only analyse data and obtain results. Th is is only part of the task, and not necessarily the most important aspect. Th e real question is: What do these results mean? It is the theoretical and contextual interpretation of the results that provides this information, and this interpretation sits at the heart of the research process. It is here that all the elements of the research process come together. When the data have been collected by the most appropriate method, and analysed in the most rigorous and relevant way, these results must be interpreted within their social and political context using the theoretical framework. It is only when the results are analysed against and within your theoretical base, and the political and ethical context taken into account, that their meaning can be assessed and formulated. It is here that we use our theoretical base to move from the ‘What?’ to the ‘Why?’ or ‘How?’ and to theorise our fi ndings. It is not enough to say longer-term workers’ compensation claimants have poorer social outcomes post injury or illness, although this is a very important fi nding in itself. Rather, using our theoretical framework relating to stigma and primary role disruption, we can theorise that aspects of a longer-term involvement in the workers’ compensation system can operate to negatively aff ect injured workers’ relationships and marginalise them from previous support networks.

BOX 3.1: UNDERSTANDING UNITS OF ANALYSIS UNITS OF ANALYSIS Our unit of analysis is what or who we are researching. While much social research uses the individual respondent as the unit of analysis, our unit of analysis can be much wider. Households, organisations such as universities, social artefacts such as newspaper articles or policy documents, or indeed anything related to social life that can be investigated, can be our unit of analysis for social research. Correctly specifying our unit of analysis is important if we want to avoid two common errors in social research: the ecological fallacy and reductionism.

THE ECOLOGICAL FALLACY The ecological fallacy occurs when we incorrectly draw conclusions about one unit of analysis from social research conducted with a different unit of analysis. Primarily, the

Percentage: A number expressed as a proportion of 100.

6161CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6103_WAL_SRM3_20170_TXT_SI.indd 61 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

problem occurs when we move findings from an aggregate unit of analysis, such as a group or an organisation, to the level of the individual. If we were, say, to investigate the entertainment venues in the inner city (the aggregated unit of analysis) and were to find that for most the average age of patrons is twenty years younger (25 years) than that of suburban venues (45 years), and that inner-city entertainment venues sell proportionately far more peanuts as bar snacks than suburban venues, we might conclude that younger patrons like peanuts more than do older patrons. Such a conclusion could be an ecological fallacy, because we cannot make a conclusion about any link between the two disaggregated variables (patron age and level of peanut consumption) from data from an aggregate unit of analysis (entertainment venues). We do not know, from our unit of analysis, which patrons are eating the most peanuts, and there are likely to be many other explanations for this phenomenon. What other explanations can you think of?

REDUCTIONISM Reductionism is also related to an incorrect use of the unit of analysis, but the mismatch occurs in the other direction, that is, data from disaggregated units of analysis are used to make inferences about aggregated units of analysis. The result is the reduction of a complex social issue or phenomenon to a specific explanation. While research on, for example, unemployed young people (the disaggregated unit of analysis), might find many are ill-prepared for the workforce in terms of job skills, we cannot then claim that the cause of youth unemployment (the aggregate unit of analysis) is poor labour market skills. This is reductionism, in that the complex issue of unemployment has been reduced to a single cause from findings that related to individual unemployed youth. The tendency to reductionism can derive from our worldview. Under the ideology of economic rationalism, social behaviour tends to be explained in exclusively economic terms (economic reductionism). As social scientists, we must try to ensure that we do not commit the error of social reductionism.

10 WRITING UP THE RESULTS Th e fi nal stage of the research process involves writing up the results and communicating the results of the research. Again, this aspect of the process is the subject of a separate chapter (see Chapter 16), but it is overviewed here to place it into the research process context. Most researchers advise that writing should begin as early as possible in the research process, and continue throughout the research. It is much easier to write up your fi ndings if you have a collection of notes and ideas that you have already begun to write (Ezzy 2002).

Writing is the act that brings a research project together as a whole. Research reports include discussions of the whole research process, from the topic and research question, through the theory, method, and methodology, to the ethics and political signifi cance of the fi ndings. One of the most diffi cult skills to acquire as a social researcher is that of clear expression. Writing with clarity and precision demonstrates respect for both the reader and those who have participated in the research.

It is also important to identify the audience for whom you are writing. Th e writing style for an academic journal article is diff erent to that of a popular book, which is diff erent again for a report for a business, government adviser, or community lobby group.

Reductionism: Reducing complex arguments to simplistic assertions.

Ideology: Shared values held by groups and societies.

62 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6203_WAL_SRM3_20170_TXT_SI.indd 62 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

VOICES IN THE FIELD: VIRGINIA WATSON

Virginia Watson is an academic at the University of Technology, Sydney, where she teaches across the disciplines of sociology, history, and political and cultural studies. Her research, which covers Australian history, politics, and public policy with a specific focus on Indigenous affairs, is similarly cross-disciplinary in approach. She is especially interested in using political theory to examine and explain the relationships between ideas and political and social institutions with a view to better understanding public policy and its effects. This has resulted in several academic articles and a book, Mining Australia’s Northern Frontier: Government, Development and Indigenous People.

Another aspect of Virginia’s research has involved her in some work being undertaken by two Aboriginal land councils and a non-government organisation that looks at the provision of telecommunications services to remote Indigenous communities. This research has involved fieldwork across some of those communities to ascertain the viewpoints, ideas, and understandings of community residents. It has been centrally informed by the research method, ethnographic action research. You can read more about this work in an edited collection, Communication for Social Change and Development.

Virginia is currently interested in researching the ways in which mining developments on Aboriginal land interlink with socioeconomic advantage and disadvantage.

EXERCISE 3.1: THE RESEARCH PROCESS 1 Identify an issue on which you would be interested to conduct some social research.

Make sure you specify your topic clearly. 2 What is it, specifically, that you want to know about this topic? What is your research

question? 3 What theory do you think might inform your research? Some examples of theoretical

traditions include feminism, Marxism, functionalism, and symbolic interactionism. (Refer back to Chapter 1 if you are unsure of the main features of these traditional theoretical social science paradigms.)

4 What sort of research method is best suited to answer this question? Are you interested in ‘How many?’ questions, or are you interested in finding ‘What meaning?’ questions? Is your question best answered by conducting a survey, focus groups, or perhaps a textual analysis of newspapers or television programs?

5 What are the main variables or processes that you are interested in? Try to identify at least two of the key dimensions of your research topic.

6 How would you conceptualise and operationalise these variables or processes? What sorts of questions would you ask?

7 What are your units of analysis: people, stories, organisations, or events? How would you sample these? Would your sample be statistically representative or representative of processes?

8 What political and ethical issues might come up in the research? Do you think your participants might become distressed while participating in your research?

9 Would the analysis involve statistical methods, or would it utilise qualitative techniques that look for patterns of meaning and interpretation? How would you use your theoretical frame to interpret your results?

10 How would you communicate the findings of your research? Would you write a policy document for government advisers or business leaders, or would you write an academic journal article or book?

Textual analysis: Analysis of the vocabulary, grammar, cohesion, and structure within a text.

6363CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6303_WAL_SRM3_20170_TXT_SI.indd 63 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

FROM METHOD TO PRACTICE THE RESEARCH PROCESS IN PRACTICE The following three studies provide examples of how the research process has been worked out in practice. They have been chosen because they provide examples of three very different types of research. At the end of each case study, I provide a list of how they have addressed the ten issues discussed in Exercise 3.1.

A SURVEY OF WHO DOES HOUSEWORK IN AUSTRALIA Are partnered men doing more of the housework than they used to a few decades ago? What do you think? Do you think your father does more housework than your grandfather? Do you think your mother does less housework than your grandmother?

Feminist reformers in the 1960s and 1970s were optimistic that changes in the labour force participation rates for married women, in combination with increased awareness of the value of women’s unpaid work in the home, would lead to an increased involvement of men in domestic labour and a more equal domestic division of labour between men and women (Baxter 2002: 400).

How do we find out if there has been some change in who does the housework? The way to answer this question is through research. Janeen Baxter examined three

Australian national surveys—1986, 1993 and 1997—that included questions about the amount of housework performed by the men and women living in the household. Each of these surveys was designed to be representative. This means that even though they only asked questions of 1000 or so people, we can be confident that their responses are very similar to the responses of the general population (see the later chapter on sampling for more detail). Baxter’s research question is well defined. Has the gender distribution of housework changed in recent decades? This is clearly a ‘How many?’ question, and, as such, a quantitative statistical analysis of survey data is appropriate.

The results of her analysis are fascinating. There is almost no change in the distribution of childcare tasks. In answer to questions about who has responsibility for getting children to bed, men had, on average, responsibility for this 43 per cent of the time in 1986 and in 1997. Women had responsibility for getting the children to bed 68 per cent of the time in 1986 and 62 per cent of the time in 1997 (the numbers add up to more than 100 per cent because of the particular method of calculation Baxter used). The difference between 68 per cent and 62 per cent was not large enough to be statistically significant. In other words, there is no significant change in who puts the children to bed. In the past it was mostly done by women, and this has not changed.

There were some significant changes in the hours spent on housework. With respect to the number of hours spent on preparing meals and cleaning up afterwards, men spent 7 hours a week on average in 1986 and 5 hours a week in 1997. Women, on the other hand, on average spent 14 hours a week in 1986 and 10 hours a week in 1997. Both of these changes were statistically significant. In other words, men and women are spending less time on food preparation. Baxter suggests that this is probably a result of people making greater use of takeaway and preprepared food. There are similar patterns in a variety of other household tasks. When asked about the hours spent on cleaning the house and washing, men spent, on average, 3 hours a week in 1986 and 1997. Women spent, on average, 10 hours a week in 1986 and 8 hours a week in 1997. The change in hours for women is statistically significant. In summary, men are not spending any extra time doing household duties, but women are spending less time on them. There are a number of possible reasons for this, but one might simply be that less housework is being done. Perhaps we don’t iron so many clothes or vacuum the floors as much as was done in the past.

64 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6403_WAL_SRM3_20170_TXT_SI.indd 64 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Baxter (2002: 401) identifies two possible theories: that of the economic exchange model that argues that women perform housework in exchange for economic support, and a feminist perspective that emphasises power differences and gender roles. Using a statistical regression analysis to assess the level and type of change in housework activity across time and gender, Baxter (2002: 419) interprets these against her theoretical framework to conclude that the economic exchange model and the feminist analysis of gender roles are both important for explaining patterns of housework in Australia.

The interesting thing about Baxter’s findings is that they do show a change in housework, but not in the way that was expected. As Baxter (2002: 420) puts it: ‘The gender gap in domestic labour involvement is getting smaller, but mainly because women are doing much less, rather than men doing much more.’ The ethical and political dimensions of Baxter’s research are not explicitly discussed by Baxter. The research findings clearly have political and ethical import. If aspects of the economic exchange model and the feminist model are correct, then equality in the home will not be achieved simply by women entering the labour market and becoming economically independent. Rather, if household duties are to be distributed equally between men and women, then changes will also be required in attitudes towards gender roles.

BOX 3.2: THE RESEARCH PROCESS: A SURVEY OF WHO DOES THE HOUSEWORK

1 Topic and question Gender distribution of housework in Australia 2 Question Has gender distribution of housework changed between

1986 and 1997? 3 Theoretical framework Economic exchange model and feminist theory 4 Research method Secondary analysis of survey data 5 Main variables or

processes Gender, and amount of housework

6 Defining and measuring Concepts derived from pre-existing research and theory

‘How many hours per week do you do housework duties?’

7 Sampling Representative national sample 8 Ethics and politics Feminist perspective identifies private and political

dimensions 9 Analysis and

interpretation Quantitative analysis and interpretation of results against theories

10 Writing up Journal article

A QUALITATIVE INTERVIEW STUDY OF TEENAGE WITCHCRAFT Witchcraft is a contemporary Pagan religion that began in Britain in the 1940s and in Australia in the 1970s (Hume 1997). Witchcraft focuses on venerating the earth and nature as sacred. In the 1990s, around the same time as the release of movies such as Th e Craft and television shows such as Buff y and Charmed, large numbers of teenagers

6565CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6503_WAL_SRM3_20170_TXT_SI.indd 65 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

became interested in Witchcraft . At the same time, popular ‘how to’ books on Witchcraft became available in mainstream bookstores (Berger & Ezzy 2007). Why are young people so interested in Witchcraft ? What meanings and interpretations lead young people to become Witches?

Helen Berger and I (2007) interviewed ninety young Witches, thirty each from Australia, the USA and Britain. Th e questions we asked were not ‘how many’ questions, so we did not use a survey. Rather, we studied the meanings and interpretations of the young Witches. In order to do this, we interviewed the young Witches for approximately an hour each about their practice and experience of Witchcraft .

Unlike Baxter’s research on housework, our study does not examine variables, such as gender, hours in childcare, or food preparation. Rather, we examine the meanings or social processes that led young people to become Witches. We utilised the method of long qualitative interviews with young Witches to examine the meanings and interpretations that people give to Witchcraft . In particular, we were interested in how Witchcraft related to their sense of self-identity. A survey would not have given us as close an understanding of why people participate in Witchcraft .

Th e theories that inform our research are drawn from sociology, anthropology, and religious studies. In our discussion of Witchcraft and self-identity, we draw on the sociological theory of Anthony Giddens (1991). He discusses the centrality refl exivity to contemporary identity. We show how young Witches are refl exive about their spirituality. Th ey had oft en spent substantial time researching various religions and refl exively thought through the appropriateness of Witchcraft as a religion for them.

Qualitative researchers aim to tell a good story about what is going on. If you do not know much about the teenage Witches, when you hear a bit of their story it might feel like walking in on a movie when it is half way through. News reports about teenage Witches are somewhat like events in those movies that don’t make sense because the background information is missing. However, if you have read the book behind the movie, or if you borrow the DVD or download it and see the fi rst half of the movie, then the meaning of the events may become clear. In the same way, qualitative research aims to tell the whole story about why people do things. If researchers do their job well enough, then the actions of the people being studied will make sense, because the ethnographer has clearly told the story of what is going on.

Meanings cannot be measured. Rather, they are stories to be heard and retold by the researcher. Th rough our research, we recorded the stories of young Witches. Th e central skill of the qualitative researcher is to be able to step inside the skin of the people they are studying, so that they can genuinely listen to the stories and make sense of the meanings the participants give to their actions.

Berger and Ezzy’s sampling method was to advertise for participants on various internet forums and through newsletters and other established networks of Witches. Th e sampling logic here is not statistical, because the questions are not ‘How many?’ questions. Rather, the sampling logic is one of signifi cance. Our interviews with young Witches provide a clear picture of the meanings and signifi cance of Witchcraft for young people.

Th e politics of our research is not explicitly stated, but it is still discernible. We are arguing that the religion of Witchcraft is typically a positive and life-affi rming practice for young people. We describe how young people choose Witchcraft as a way of refl exively helping themselves deal with the challenges of contemporary life. Even this short quote

66 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6603_WAL_SRM3_20170_TXT_SI.indd 66 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

from one of the young Witches we interviewed may seem like coming in halfway through a movie, but it gives you a taste of our argument.

Probably the main reason why I like being Pagan [a Witch] is it’s equal. It’s got Gods and Goddesses … It’s not male dominated or anything … And you can be yourself … You don’t have to believe in this, this, and this, and you don’t have to go to church on this day, or you don’t have to do this ritual in this way. I like being an individual [laughs] … Th e essence of Witchcraft ? For me, it’s just being myself (Berger & Ezzy 2007: 19).

BOX 3.3: THE RESEARCH PROCESS: A QUALITATIVE INTERVIEW STUDY OF TEENAGE WITCHCRAFT

1 Topic and question Teenage Witches and identity 2 Question How do young people become Witches? 3 Theoretical frame Sociological identity theory 4 Research method Long qualitative interviews 5 Main variables or

processes Meanings and interpretations of young Witches

6 Defining and measuring Inductively derived concepts

Telling the participants’ stories 7 Sampling Self-selected sample in response to advertisements 8 Ethics and politics Witchcraft is portrayed as a valid religion 9 Analysis and

interpretation Thematic analysis and Interpretation using identity theory

10 Writing up Book

A DISCOURSE ANALYSIS OF RACISM IN NEWSPAPERS On 29 April 1996, the Australian newspaper proclaimed the headline ‘33 slain in our worst massacre’ (the fi nal death toll was thirty-fi ve people). Th e Australian went on to report on the Port Arthur massacre by Martin Bryant, saying, ‘Th e Tasmanian massacre surpasses any other documented in Australia’s history, and is believed to be one of the worst mass shootings of all time’ (Banerjee & Osuri 2000: 268). Th ese few lines of text, and those from the headlines of a few other newspapers, form the basis of Banerjee and Osuri’s research. Th ey ask a simple question of these headlines, that could also be a description of their research topic: ‘Why … was this event termed the worst massacre in Australian history when the Australian nation seemed to have been founded on a history of [Aboriginal] massacres?’ (Banerjee & Osuri 2000: 264). Th e diff erence between Banerjee and Osuri’s research and that of Berger and Ezzy is not simply in the very small piece of empirical evidence that Banerjee and Osuri examine. It is also that they are as much interested in what the text does not say as they are in what the text does say. Th e main dimension of

6767CHAPTER 3: THE RESEARCH PROCESS

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6703_WAL_SRM3_20170_TXT_SI.indd 67 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Banerjee and Osuri’s research is the endemic racism in Australian news, and they use these few headlines as a way of further examining this issue.

Th eory is central to their research question, and in fact, theoretical argument takes up the bulk of their article. Th ey are particularly interested in theories of racism, drawing on Stuart Hall’s distinction between overt racism, which does not apply to the Port Arthur massacre news reports, and ‘inferential racism’, which can be applied to the news reports. In particular, they argue that the news reports implied that ‘the history of Australia is the history of Anglo-Celtic Australians … [and] that the histories of Aboriginal peoples are not a part of Australian national history’ (Banerjee & Osuri 2000: 272).

Ethics and politics are central to this form of research. Th is is because research such as Banerjee and Osuri’s oft en aims to use social research to demonstrate that politics and power are central parts of social life. Cultural studies research is oft en informed by theories, such as feminism, socialism or race relations theory, that explicitly incorporate political dimensions. Banerjee and Osuri (2000: 279) refer to Foucault, arguing that their research aims to ‘unmask the power relations behind knowledge production’.

Sampling and measurement, or observation, are not central issues to this sort of research. Some cultural studies reports do not provide any justifi cation for the texts they analyse, while others use more systematic methods for selecting their sample.

Measurement: How we are going to measure our key concepts. How we measure and what we measure differs between quantitative and qualitative methodologies.

BOX 3.4: THE RESEARCH PROCESS: DISCOURSE ANALYSIS OF RACISM IN NEWSPAPERS

1 Topic Racism in newspapers 2 Question Why are massacres of Aboriginal people silenced? 3 Theoretical frame Theories of racism and power 4 Research method Cultural textual analysis 5 Main variables or

processes Derived from analysis of inferential racism

6 Defining and measuring Deductively derived concepts

Examination of language, presence and absence 7 Sampling Case study 8 Ethics and politics Politics of Indigenous history as invisible 9 Analysis and

interpretation Discourse analysis and interpretation

10 Writing up Journal article

68 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DOUGLAS EZZY

03_WAL_SRM3_20170_TXT_SI.indd 6803_WAL_SRM3_20170_TXT_SI.indd 68 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

CONCLUSION Th is chapter builds on the material discussed in Chapters 1 and 2 to provide a solid foundation of the why, what, how, where, and when of contemporary social research. Within this, the importance of good social research to our lives and society has been stressed. Social issues are some of the most pressing and diffi cult issues that confront contemporary society. Many of our major problems, such as global warming, terrorism, pollution, mental health, and the spread of HIV/AIDS, are primarily social issues, requiring responses from policy makers that address the social dimensions of these issues. Social research provides essential background information so that policy makers can respond appropriately to such issues. But good policy responses rely on solid research evidence, and well-constructed, valid, and rigorous social research follows a process. Th is research process guides researchers and ensures that each vital step is undertaken, to the right depth and at the right time, for the specifi c research project. In this chapter, ten key steps have been identifi ed and explained. Understanding and using the research process allows us to have confi dence in our research and in ourselves as researchers.

MAIN POINTS

• Social research includes a variety of diff erent types of research. • Social research methodologies can be categorised into two main groups: quantitative

and qualitative. • Quantitative, or statistical, research primarily aims to examine ‘How many?’

questions. More complex forms of statistical research can be used to examine questions of causality.

• Qualitative, or interpretative, research is used to examine ‘What meaning?’ questions that focus on interpretations and understandings.

• Th ere is a variety of other forms of research that do not fi t into either of these categories. • Th is chapter has identifi ed ten key issues that social researchers must address. • Social researchers aim to provide a better understanding of what is happening in the

social world. • Many of the biggest challenges faced in contemporary society are primarily social

issues. • Social researchers cannot escape from issues of ethics and politics. • Social researchers hope that their research will provide a better understanding of

social processes so that policy makers and political leaders can make better policy decisions to benefi t society.

FURTHER READING If you want to learn more about social research, the following books are recommended. Bouma, G. & Ling, R. (2004). Th e Research Process, 5th edn. Melbourne: Oxford

University Press.

6969

DOUGLAS EZZY

CHAPTER 3: THE RESEARCH PROCESS

03_WAL_SRM3_20170_TXT_SI.indd 6903_WAL_SRM3_20170_TXT_SI.indd 69 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Kellehear, A. (1993). Th e Unobtrusive Researcher: A Guide to Methods. Sydney: Allen & Unwin.

May, T. (1997). Social Research: Issues, Methods and Process. Buckingham: Open University Press.

REFERENCES Bammer, G., Hall, W., Hamilton, M. & Ali, R. (2002). ‘Harm Minimization in a

Prohibition Context’, Annals of the American Academy of Political and Social Science, 582 (1): 80–93.

Banerjee, S. & Osuri G. (2000). ‘Silences of the Media: Whiting Out Aboriginality in Making News and Making History’, Media, Culture and Society, 22 (3): 263–84.

Baxter, J. (2002). ‘Patterns of Change and Stability in the Gender Division of Household Labour in Australia, 1986–1997’, Journal of Sociology, 38 (4): 399–424.

Berger, H. & Ezzy, D. (2007). Teenage Witches. Piscataway: Rutgers University Press. Cameron, M. (1964). ‘Identity and the Shoplift er’. In Cameron, M. O., Th e Booster and

the Snitch. New York: Free Press. Demas, P., Schoenbaum, E., Wills, T., Doll, L. & Klein, R. (1995). ‘Stress, Coping, and

Attitudes toward HIV Treatment in Injecting Drug Users: A Qualitative Study’, AIDS Education and Prevention, 7 (5): 429–42.

Eagly, A., Karau, S. & Makhijani, M. (1995). ‘Gender and the Eff ectiveness of Leaders: A Meta-analysis’, Psychological Bulletin, 117 (1): 125–45.

Ezzy, D. (1993). ‘Unemployment and Mental Health: A Critical Review’, Social Science and Medicine, 37 (11): 41–52.

Ezzy, D. (2001). ‘Reading for the Plot, and Not Hearing the Story: Ecological Tragedy and Heroic Capitalism’. In Mills, A. and Smith, J. (eds), Utter Silence: Voicing the Unspeakable. New York: Peter Lang.

Ezzy, D. (2002). Qualitative Analysis. Sydney: Allen & Unwin. Ezzy, D., Walter, M. & Welch, A. (2009). WorkCover Tasmania Long Term Benefi ts Study

Final Report. Hobart: University of Tasmania, School of Sociology and Social Work. www.workcover.tas.gov.au/__data/assets/pdf_fi le/0006/165147/Long_term_benefi ts_ study_phase_fi nal_report.pdf.

Giddens, A. (1991). Modernity and Self-Identity. Stanford: Stanford University Press. Gurlich, A., de Visser, R., Smith, A., Rissel, C. & Richters, J. (2003). ‘Injecting and Sexual

Risk Behaviour in a Representative Sample of Adults’, Australian and New Zealand Journal of Public Health, 27 (2): 242–50.

Hume, L. (1997). Witchcraft and Paganism in Australia. Melbourne: Melbourne University Press.

Jacques, P., Dunlap, R. & Freeman, M. (2008). ‘Th e Organisation of Denial: Conservative Th ink Tanks and Environmental Scepticism’, Environmental Politics, 17 (3): 349–85.

DOUGLAS EZZY

PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

03_WAL_SRM3_20170_TXT_SI.indd 7003_WAL_SRM3_20170_TXT_SI.indd 70 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Kabakoff , R. (2002). ‘Gender and Leadership in the Corporate Boardroom’. www.mrg. com/Publications/articles/APA2000.pdf.

McDonald, P. (1998). ‘Contemporary Fertility Patterns in Australia’, People and Place, 6 (1): 1–13.

Moon, J. & Fountain, I. (1997). ‘Keeping the Gates? Women as Ministers in Australia, 1970–96’, Australian Journal of Political Science, 32 (3): 455–66.

Newell, P. & Paterson, W. (1998). ‘A Climate for Business: Global Warming, the State and Capital’, Review of International Political Economy, 5 (4): 679–703.

Nissani, M. (1999). ‘Media Coverage of the Greenhouse Eff ect’, Population and Environment, 21 (1): 27–43.

Pini, B. (2002). ‘Gender and Power: Th e Exclusion of Women from Agri Political Leadership: A Case Study of the Australian Sugar Industry’, Sociologia Ruralis, 42 (1): 45–60.

Pini, B. (2005). ‘Th e Th ird Sex: Women Leaders in Australian Agriculture’, Gender, Work and Organization, 12 (1): 73–88.

Probert, B. (1980). Working Life. Melbourne: McPhee Gribble. Rice, P. & Ezzy, D. (1999). Qualitative Research Methods. Melbourne: Oxford University

Press. Robinson, A. & Robinson, Z. (1997). ‘Science Has Spoken: Global Warming Is a Myth’,

Wall Street Journal, 4 December, 1. Rosser, V. (2003). ‘Faculty and Staff Members’ Perceptions of Eff ective Leadership: Are

Th ere Diff erences Between Women and Men Leaders?’, Equity and Excellence in Education, 36 (1): 71–81.

Sarantakos, S. (1993). Social Research. Melbourne: Macmillan Education. Sony (2004). EverQuest. http://eqlive.station.sony.com/library/faqs/faq_pop.jsp. Verdouw, J. (2008). ‘For the Love of Money’. Unpublished PhD thesis. Hobart: School of

Sociology, University of Tasmania. Vincent, A. (2004). ‘Health, Risk, and Injecting Drugs’. Honours thesis. Hobart: School of

Sociology, Social Work and Tourism, University of Tasmania. Walter, M. (2002). ‘Working their Way out of Poverty? Sole Motherhood, Work, Welfare

and Material Well-being’, Journal of Sociology, 38 (4): 361–80. White, K. (2003). ‘Women and Leadership in Higher Education in Australia’, Tertiary

Education and Management, 9 (1): 45–60.

7171

DOUGLAS EZZY

CHAPTER 3: THE RESEARCH PROCESS

03_WAL_SRM3_20170_TXT_SI.indd 7103_WAL_SRM3_20170_TXT_SI.indd 71 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

4 KEY TERMS academic fraud

anonymity

confi dentiality

covert research

ethical research

ethics

experimental

human research

ethics committees

(HRECs)

information sheet

informed consent

longitudinal study

low-risk research

plagiarism

Ethics and Social Research DAPHNE HABIBIS

THE CARNEGIE MELLON STUDY ON CYBER PORN In 1995, a Carnegie Mellon University student, Martin Rimm, gathered data on the

reading habits of 4227 users of a university computer system’s newsgroup service

(Usenet) (Rimm 1995). The controversial finding was that 83.5 per cent of the 900 000

images accessed by the Usenet groups were pornographic. The paper implied that those

individuals who used blocking software to screen their internet use from surveillance

could be paedophiles.

The results were published as the cover story in Time magazine (Elmer-Dewitt

1995), unleashing a storm of public controversy about the moral values of US society.

There were also concerns about Rimm’s methods, as well as ethical issues of informed

consent and invasion of privacy (Thomas 1995). These included

• accessing personal information about the university’s internet users, including their

names, age, sex, nationality, marital status, and employment

• misrepresentation of data

• the unsubstantiated imputed allegation that those using blocking software were

paedophiles

• the probable use of covert methods to obtain the data

• the publication of potentially identifying information, such as the names of the cities

from which the user called, including small cities. Source: Thomas 1995

04_WAL_SRM3_20170_TXT_SI.indd 7204_WAL_SRM3_20170_TXT_SI.indd 72 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

INTRODUCTION Ethics is concerned with the establishment of a set of moral standards that govern behaviour in a particular setting or for a particular group. Ethical research is concerned with ensuring that ethical principles and values always govern research involving humans. John Barnes (1979: 16) explains that ethical decisions in research are those decisions that arise when we try to decide between one course of action and another, not in terms of expediency or effi ciency, but by reference to standards of what is morally right or wrong.

Whereas in every other aspect of research the core focus is on the project and the requirements fl owing from the research questions, when it comes to ethical issues, the focus shift s from the needs of the project to those of the key stakeholders: the participants, the researcher, the research organisation and the scientifi c community, as well as the broader public.

WHY IS ETHICS AN ISSUE IN SOCIAL RESEARCH? Public outcry and academic debate about studies that violated the human rights of participants led to the requirement that all social research conducted by government organisations fi rst be submitted for approval to a human research ethics committee (HREC) before funding.

Understanding the uneven power relations between the researcher and the researched is critical to understanding the ethical concerns that accompany social investigation. Undertaking research carries a mystique and prestige. Researchers possess expert knowledge and are concerned with the discovery of truth. Because of the cultural value attributed to scientifi c knowledge, and the qualifi cations and high status of researchers, there is a glamour associated with that role. Researchers belong to the new middle class. Most enjoy secure, well-paid jobs with considerable workplace autonomy. Th ey possess high levels of social and cultural capital, and infl uence public debate and social policy.

Th e power and infl uence of researchers oft en contrasts with that of their study population. Social science places a particularly intense gaze on vulnerable populations, such as young people, criminals, women, ethnic minorities, and the sick. Researchers use participants with a particular objective in mind, but participants may have only a partial understanding of this. With important exceptions, such as action research and some feminist research, it is the researcher who controls the production of knowledge, including the design, data collection, interpretation, and reporting of the project. Participants have very little infl uence over this process even though they have contributed to it.

Diff erent methods place diff erent emphases on the level of participants’ involvement. In general, interpretative methods give more space to participants’ views than do quantitative approaches. But, in most cases, it is the researcher who sets the agenda. Feminist theorists argue that the relationship between researcher and researched is exploitative because the researcher benefi ts directly from the research through career progression (Oakley 1981). While the direct benefi ts received by participants are usually limited to some small fi nancial recompense, the research is justifi ed in terms of its contribution to knowledge and its anticipated benefi ts to the population group to which participants belong.

Ethics: The establishment of a set of moral standards that govern behaviour in a particular setting or for a particular group.

Ethical research: Ensuring that ethical principles and values always govern research involving humans.

7373CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7304_WAL_SRM3_20170_TXT_SI.indd 73 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

BOX 4.1: RESEARCHER AND RESEARCHED: AREAS OF POTENTIAL CONFLICT

The needs of the researcher The needs of the researched

Inexpensive and fast data collection

Protection from exploitation through HREC approval prior to commencement Adequate information and time to make an informed decision about participation

High response rate The right to refuse participation Low attrition rate The right to withdraw participation at any time Minimal missing data The right to refuse any question Use of a wide choice of methods, including covert investigation

Protection from methods that potentially violate rights, such as informed consent Protection from acts that might cause harm or discomfort beyond what is normally experienced in everyday life

Freedom to use the findings in the most effective way, including informing other studies

Protection of anonymity and confidentiality

Timely data analysis and the freedom to interpret the data

Control over how the individual contribution to data is used through access to draft publications and the right to withdraw contribution

Timely and widespread dissemination of findings

Control over who receives copies of the report and where it is published

Th ere are many examples in which researchers have threatened the safety of participants or abused their trust in some way (see Case study 4.1). Th e reasons researchers might disregard the rights and needs of participants are infl uenced by the institutional context within which research takes place. Far from being disinterested observers motivated by pure considerations of knowledge, researchers have careers that oblige them to be successful researchers. Paying careful attention to the moral and ethical requirements of research can get in the way of successful projects that deliver results in a timely manner (see Box 4.1).

Feminist research and some types of action research try to overcome what they see as the inherently exploitative nature of research by involving participants in some or all stages of the research process. Th is may include establishing a reference group and involving participants in data collection, as well as giving them control over what and how the results are reported.

Ultimately, the most important mechanism to ensure ethical research is the researcher’s own sense of integrity and moral values. Th e solitary and autonomous nature of most research means that, ultimately, researchers must be responsible for the ethical conduct of their projects. Th e rarity of reported abuses suggests that most researchers are committed to ensuring the wellbeing of their participants.

74 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7404_WAL_SRM3_20170_TXT_SI.indd 74 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

EXERCISE 4.1: EMPOWERING PARTICIPANTS Think about the respective roles of the researcher and the researched, and answer the following questions. 1 What measures can researchers put in place to empower participants? 2 What are the methodological advantages and disadvantages of bringing participants

into the research process?

THE HISTORY OF ETHICS AND RESEARCH Contemporary approaches to ethical issues in research date back to the Second World War and the inhumane abuses perpetrated by physicians and scientists, in the name of scientifi c investigation, on Jewish and other inmates of Nazi concentration camps. Th is led to the Nuremberg Code (1949) (see Box 4.2), which set out what were to become the basic principles for ethical research on human beings.

BOX 4.2: THE NUREMBERG CODE: DIRECTIVES FOR HUMAN EXPERIMENTATION 1 The voluntary consent of the human subject is absolutely essential. This means that

the person involved should have legal capacity to give consent, should be so situated as to be able to exercise free power of choice, without the intervention of any element of force, fraud, deceit, duress, overreaching or other ulterior form of constraint or coercion, and should have sufficient knowledge and comprehension of the elements of the subject matter involved as to enable him to make an understanding and enlightened decision. This latter element requires that before the acceptance of an affirmative decision by the experimental subject, there should be made known to him the nature, duration and purpose of the experiment, the method and means by which it is to be conducted, all inconveniences and hazards reasonably to be expected, and the effects upon his health or person which may possibly come from his participation in the experiment. The duty and responsibility for ascertaining the quality of the consent rests upon each individual who initiates, directs or engages in the experiment. It is a personal duty and responsibility, which may not be delegated to another with impunity.

2 The experiment should be such as to yield fruitful results for the good of society, unprocurable by other methods or means of study, and not random and unnecessary in nature.

3 The experiment should be so designed and based on the results of animal experimentation and a knowledge of the natural history of the disease or other problem under study that the anticipated results will justify the performance of the experiment.

4 The experiment should be so conducted as to avoid all unnecessary physical and mental suffering and injury.

5 No experiment should be conducted where there is an a priori reason to believe that death or disabling injury will occur, except, perhaps, in those experiments where the experimental physicians also serve as subjects.

6 The degree of risk to be taken should never exceed that determined by the humanitarian importance of the problem to be solved by the experiment.

7 Proper preparations should be made and adequate facilities provided to protect the experimental subject against even remote possibilities of injury, disability or death.

Experiment: A research method in which pretest measurements are taken of a target group or area, before the implementation of an experimental intervention (introduction of a particular program, strategy or project), followed by post-test measurements after the intervention has had time to take effect.

7575CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7504_WAL_SRM3_20170_TXT_SI.indd 75 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

8 The experiment should be conducted only by scientifically qualified persons. The highest degree of skill and care should be required through all stages of the experiment of those who conduct or engage in the experiment.

9 During the course of the experiment the human subject should be at liberty to bring the experiment to an end if he has reached the physical or mental state where continuation of the experiment seems to him to be impossible.

10 During the course of the experiment the scientist in charge must be prepared to terminate the experiment at any stage, if he has probable cause to believe, in the exercise of the good faith, superior skill and careful judgment required of him that a continuation of the experiment is likely to result in injury, disability, or death to the experimental subject.

These principles were enshrined in the World Medical Assembly’s Declaration of Helsinki in 1965.

Source: Reprinted from Trials of War Criminals Before the Nuremberg Military Tribunals under Control Council Law, No. 10, vol. 2: 181–2. Washington, DC: US Government Printing Office, 1949

THE HISTORY OF ETHICS AND RESEARCH IN AUSTRALIA In Australia, a Statement on Human Experimentation, based on the Declaration of Helsinki, was initially produced in 1966 by the National Health and Medical Research Council (NHMRC). Th e application of these principles was at fi rst left to researchers, but reports of abuses led to the external surveillance of research by human research ethics committees (HRECs). In 1973, the NHMRC made it a requirement that all grant applications must have received ethical approval from a human research ethics committee (Loblay & Chalmers 1999: 9).

In 1992, the NHMRC released the National Statement on Ethical Conduct in Research Involving Humans in accordance with the NHMRC Act of that year. Th is established the NHMRC as a statutory body with functions, powers, and obligations to issue human research guidelines as developed by the Australian Health Ethics Committee. Th ese guidelines are regularly updated in a process of community consultation; the current statement was released in 2007 (see Box 4.3) (NHMRC 2007). All human research undertaken within a government organisation that involves the potential risk of harm or inconvenience to humans must receive ethical approval before it can be conducted. Professional organisations such as the Australian Sociological Association (TASA), the Sociological Association of Aotearoa New Zealand (SAANZ), and the Australian Association of Social Workers (AASW) include sections that require their members to conform to the standards established by the NHMRC.

CASE STUDY 4.1: AN ETHICALLY CONTROVERSIAL RESEARCH STUDY HUMPHREYS’ TEAROOM TRADE Humphreys’ study of casual homosexual encounters caused considerable debate about whether the ends can ever justify the means. This influential study, which was conducted in the late 1960s, employed covert research methods to discover the social background of men who engage in homosexual acts (Humphreys 1970). Humphreys’ method involved observing homosexual encounters in public places, including truckstop restaurants (tearooms). Having gained the trust of the unwitting participants, Humphreys acted as their lookout—the watchqueen. In this role he was able to make a note of their car registration numbers. A year later, he used this information to track down the addresses

Covert research: The inclusion of respondents in a research project without their awareness of and/or agreement to their participation in the project.

76 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7604_WAL_SRM3_20170_TXT_SI.indd 76 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

of the ‘participants’. Under the guise of collecting data for another study, he then visited the men and obtained detailed information about their socioeconomic background. His key finding was that the men were quite ordinary citizens. Many were married and many held respectable jobs.

Humphreys’ study challenged negative stereotypes about gay men and so made an important contribution to the destigmatisation of homosexuality. However, its violation of fundamental ethical principles, such as informed consent, the safety of participants, and the use of deception, compromised its contribution to knowledge. Although Humphreys had been careful to disguise who his participants were, and kept details of their identities confidential, he was criticised for his invasion of their privacy and the possibility that the information might fall into the public arena (Warwick 1982). Humphreys defended his methods on the grounds that the study demystified a critical social issue and that it was only through covert techniques that the knowledge would have become available.

WHEN IS ETHICAL REVIEW NEEDED? Research that involves negligible risk to humans may be exempt from submission to an HREC. Such research must carry no foreseeable risk of harm or discomfort, and any foreseeable risk should be no more than an inconvenience. Th e use of existing data or records that contain only non-identifi able data, and that are publicly accessible, is an example of research that would meet this criterion.

Research that is of low risk may be reviewed under expedited processes involving a lower level of scrutiny by the HREC or others familiar with the legislation on ethical research. Research is low risk when the only foreseeable risk is one of discomfort. Low-risk research may involve a lower level of scrutiny by HRECs. To be designated low risk, the project must not involve • deception of participants • participation of people without their prior consent • withholding from one group treatment or methods of learning from which they may

benefi t • access or use of medical records from which participants can be identifi ed or linked

to their records • the use of personal data obtained from a Commonwealth or state government

department without the consent of the participants • potential disclosure of illegal activities or criminal behaviour • specifi c targeting of vulnerable population groups, including children and young

people, Aboriginal and Torres Strait Islanders people, who are unable to give informed consent because of diffi culties in understanding an information sheet

• people who may be involved in illegal activity • socially or emotionally sensitive topics, including ethnic identity, eating disorders,

parenting practices, sensitive personal issues, suicide, grief, and substance abuse • risk to the researcher • politically sensitive information.

PROTECTING PARTICIPANTS Th e National Statement on Ethical Conduct in Research Involving Humans (section 1) (NHMRC 2007) identifi es four values that govern research involving humans. Th ese are research merit and integrity, justice, benefi cence and respect.

Negligible risk research: Research that carries no foreseeable risk of harm or discomfort to respondents.

Low-risk research: Research where the only foreseeable risk to respondents is one of discomfort. Low-risk research may involve a lower level of scrutiny by HRECs.

Informed consent: A document or agreement that ensures, first, that research respondents are fully informed about what the research is about and what participation will involve, and second, that they make the decision to participate without any formal or informal coercion.

7777CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7704_WAL_SRM3_20170_TXT_SI.indd 77 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

RESEARCH MERIT AND INTEGRITY Unless research has merit through its potential benefi t and is carried out with integrity and honesty, then the involvement of participants cannot be ethically justifi ed.

JUSTICE Th is involves a regard for the human sameness that each person shares with every other, and requires distributive and procedural justice. Distributive justice is expressed in the fair distribution of the benefi ts and burdens of research, procedural justice involves fair treatment in the recruitment of participants and the review of research.

BENEFICENCE Th is requires researchers to assess and take account of the risks of harm and the potential benefi ts of research to participants and to the wider community, to be sensitive to the welfare and interests of people involved in their research, and to refl ect on its social and cultural implications.

RESPECT FOR HUMAN BEINGS Th is value binds the three values of research: merit and integrity, justice, and benefi cence. Recognition of the intrinsic value of human beings through respectful treatment underpins the ethics framework. Th is means having due regard for the welfare, beliefs, perceptions, customs, and cultural heritage of those involved in research.

Implicit in these values are the principles of • informed consent • anonymity • confi dentiality • protection from harm.

BOX 4.3: NATIONAL STATEMENT ON ETHICAL CONDUCT IN RESEARCH INVOLVING HUMANS: EXCERPT

The relationship between researchers and research participants is the ground on which human research is conducted. The values set out in this section—respect for human beings, research merit and integrity, justice and beneficence—help to shape that relationship as one of trust, mutual responsibility and ethical equality. For this reason, the National Statement speaks of research ‘participants’ rather than ‘subjects’ …

The values of respect, research merit and integrity, justice and beneficence have become prominent in the ethics of human research in the past six decades, and they provide a substantial and flexible framework for principles to guide the design, review and conduct of such research. This National Statement is organised around these values …

Among these values, respect is central. It involves recognising that each human being has value in himself or herself, and that this value must inform all interaction between people. Such respect includes recognising the value of human autonomy … It also involves providing for the protection of those with diminished or no autonomy, as

78 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7804_WAL_SRM3_20170_TXT_SI.indd 78 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

well as empowering them where possible … Reference to these values throughout the National Statement serves as a constant reminder that, at all stages, human research requires ethical reflection that is informed by them.

SEEKING AND GIVING CONSENT Th e guiding principle for researchers is that a person’s decision to participate in research is to be voluntary, and based on suffi cient information (see Box 4.4 below) and adequate understanding of the proposed research, and the implications of participation in it.

Participation that is voluntary and based on suffi cient information requires an adequate understanding of the purpose, methods, demands, risks, and potential benefi ts of the research.

Th is information must be presented in ways suitable to each participant. Th e process of communicating information to participants and seeking their consent

should not be merely a matter of satisfying a formal requirement. Th e aim is to achieve mutual understanding between researchers and participants. Th is aim requires an opportunity for participants to ask questions, and to discuss the information and their decision with others if they want to do so.

Consent may be expressed orally, in writing, or by other means, such as return of a survey or conduct implying consent, depending on • the nature, complexity, and level of risk involved in the research • the participant’s personal and cultural circumstances. See Box 4.5 for an example of a

consent form.

BOX 4.4: INFORMATION SHEET IMPROVING HOUSING RESPONSES TO INDIGENOUS PATTERNS OF MOBILITY Researcher: Dr Daphne Habibis

Purpose of the study This is a national study to improve social housing responses to Indigenous patterns of mobility. Indigenous people are more geographically mobile than non-Indigenous people in the way they move between places that are culturally distinctive. Current housing policies and practices take little account of these mobility patterns; this is associated with poor housing outcomes for Indigenous people. This study seeks to identify what services are available to meet these kinds of population movements, how effective these are, and how they can be improved. The study has been funded by the Australian Housing and Urban Research Institute.

How will I benefit? Improving Indigenous housing outcomes is a national priority and therefore a key objective of government-funded housing services. This study will assist housing services to achieve this goal by • identifying policies that will reduce the high rate of Indigenous tenancy failure • providing a model for distinguishing between different forms of mobility and homelessness • providing information about patterns of mobility that will help housing services to improve

their planning to meet the needs of Indigenous tenants.

7979CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 7904_WAL_SRM3_20170_TXT_SI.indd 79 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Why have I been chosen to be in this research? You have been identified as a potential respondent for this research through a process of consultation with academics and housing personnel. They believe you have expert knowledge arising from your position as a policy maker or service provider in one of the following areas: • state housing • community housing • non-government organisations that provide human services.

What will my participation involve? You will be asked to participate in a face-to-face interview that will focus on some or all of the following areas: • your views on whether or how Indigenous population movement influences service delivery • what provisions exist within your service to respond to these, how effective you believe

these are, and how they might be improved • any examples of good practice you are aware of that could be nationally disseminated. The interview will take place at your workplace at a mutually convenient time and will last about one hour. The interviewer will make notes of the interview.

Are there any risks or discomforts to me? There are minimal risks associated with participation. To ensure that you are comfortable with your participation the following measures will be taken • You will be informed about your right to withdraw consent at any time or not to answer

specific questions. • You will be given the opportunity to review and amend any material you provide.

Will I be identifiable by being involved in this study? Your individual contribution will not be identified but will be reported by position and service category. You may also access the part(s) of the report in which your contribution appears to ensure that you are satisfied with the content.

How private is the information I give? Because the study is concerned with the dissemination of good practice it is anticipated that confidentiality will not be a concern. Safeguards to ensure confidentiality of research data will include • storage of data in locked cabinets and on password-protected computers • storage of raw data for at least five years from the date of publication of reports, after which

computer files will be deleted and hard copies will be shredded • all data will be stored in offices at the University of Tasmania.

Can I withdraw from the research? Prior to the interview your consent to participation in the study will be evidenced by your signing of the consent form. Participation in the study is entirely voluntary. You can withdraw at any time regardless of the effect and without the need to provide any explanation. You may also withdraw any data you have supplied.

Contact person Should you have any questions about the study at any stage, contact Daphne Habibis by phone on (05) 8322 3536 or by email at [email protected].

Has this research been approved by an ethics committee? This project has received ethical approval from the Human Research Ethics Committee (Tasmania) Network, which is constituted under the National Health and Medical Research Council. The committees under the HREC (Tasmania) Network use the National Statement on Ethical Conduct in Research Involving Humans Guidelines to inform their decisions.

Consent form: A short, written statement that respondents are asked to read, and, if they agree with it, sign it before participation in an investigation. It states that respondents have been fully informed about the study and that their participation is voluntary.

80 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8004_WAL_SRM3_20170_TXT_SI.indd 80 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

BOX 4.5: STATEMENT OF INFORMED CONSENT TITLE OF PROJECT: IMPROVING HOUSING RESPONSES TO INDIGENOUS PATTERNS OF MOBILITY 1 I have read and understood the information sheet for this study. 2 The nature and possible effects of the study have been explained to me. 3 I understand that the study involves a one-hour telephone or face-to-face interview

covering the following topics: • my experience of finding and keeping accommodation when living away from

my usual home • my experience of housing services for temporary accommodation • my views on how helpful these housing services are and how they could be

improved • the reasons why I, or other Indigenous people I know, decide to move away

from the country to make a new home in town. I understand that the interview will be recorded in note form and that I may be asked to have my home photographed and reproduced in the published findings of the study. 4 I understand the risks associated with participation and have been advised of my right to

• choose not to answer any individual questions or to terminate my participation at any time

• refuse consent for my home to be photographed. 5 I understand that all research data will be securely stored on the University of Tasmania

premises for at least five years, and will be destroyed when no longer required. 6 Any questions that I have asked have been answered to my satisfaction. 7 I agree that research data gathered from me for the study may be published, but that

my individual contribution will remain anonymous. 8 I understand that any information I supply to the researcher(s) will be used only for the

purposes of the research. 9 I consent to have my home photographed and reproduced in the published findings.

Yes ☐ No ☐ (tick one) 10 I agree to participate in this investigation and understand that I may withdraw at any

time without any effect, and if I so wish, may request that any data I have supplied to date, including photographs, be withdrawn from the research. I understand that whether or not I agree to participate in this study, neither decision will have any effect on any services I receive.

Name of participant

Signature Date

Statement by Investigator I have explained this project and the implications of participation in it to this volunteer and I believe that the consent is informed and that s/he understands the implications of participation.

Name of investigator

Signature Date

Th ere are some circumstances in which written consent is not required, for example, where it is not practicable, as in telephone surveys, or where the level of intrusion is minimal, such as observing everyday behaviour. Consent may be implied by participation, for example, through return of a survey. Limited disclosure may be approved if the research is low risk. On completion of the research, participants must be informed about the limited disclosure, if possible. If the research involves active concealment or explicit

8181CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8104_WAL_SRM3_20170_TXT_SI.indd 81 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

deception a full explanation of the aims of the research, why deception or concealment was necessary and the reasons for non-disclosure must be provided to participants.

ANONYMITY Anonymity refers to the protection of participants from identifi cation as participants in the research. Th is ensures that their rights to privacy are respected and that any information they provide cannot be used to identify them. Th e most eff ective way of ensuring anonymity is not to collect identifying data. Th is is easily achieved in some cross- sectional studies, such as mailout surveys, and whenever personal information such as names or addresses are not necessary for data analysis.

Th ere are many investigations in which the anonymity of participants is not possible. Th e method of recruitment may require researchers to have access to the contact details and names of participants. Th e usual method of preserving anonymity in these circumstances is to use a pseudonym for the participant and to keep the code breaker in a secure place, such as a locked fi ling cabinet, separate from the collected data. Th e same technique can be used in longitudinal studies in which participants will be recontacted at intervals.

Th ere are some studies in which anonymity is diffi cult or impossible to achieve. Studies of small towns close to where the researcher lives may be diffi cult to disguise. In Smalltown (1990) Dempsey dealt with this by changing the name of the town and including some misleading information. Th ere are other circumstances in which the ethical requirement for anonymity may be breached, for example, if there is a legal or moral requirement for identifi cation of participants, as in cases of child abuse or sexual assault.

Anonymity can also be compromised once it enters the public domain. Th is is especially so when the topic is of intense public interest and the issue is politically contentious (Warwick & Pettigrew 1983).

CONFIDENTIALITY Th e principle of confi dentiality overlaps with that of anonymity, since both are concerned with maintaining the privacy of participants. Anonymity is concerned with the identifi cation of individual participants, while confi dentiality is concerned with ensuring that the information they provide cannot be directly linked to them. Even if the participants in a study are identifi ed, the principle of confi dentiality means their specifi c contribution cannot be identifi ed. Th e most common method of ensuring confi dentiality is to aggregate the data by grouping it into categories such as percentages or themes.

Th ere are a number of ways in which confi dentiality may be breached. As well as the immediate data collector, research team members and administrative staff are likely to see their information. Th ose in authority over participants may only consent to research taking place within their organisation if they have access to the data. In this case, the researcher needs to negotiate what information will be made available and whether it will be provided in a way that identifi es individual participants. Potential participants must be informed of this arrangement.

Th ere are situations in which participants want their contribution to be recognised. Th is is most likely to occur in types of qualitative research such as ethnographic, action, or feminist research where the aim is to give a voice to participants. In Sidewalk (1999), Duneier used ethnography, including participant observation, to provide a detailed account of how race and class shaped the lives of male African-American street vendors in New York’s Greenwich Village, who were oft en homeless. Identifying participants gave them a public voice.

Anonymity: The protection of respondents from identification as participants in the research.

Longitudinal study: A study that collects data on the same phenomena over an extended period of time.

Confidentiality: The principle of ensuring that the specific contribution of respondents cannot be identified.

Participant observation: The main data-collecting technique used by ethnographers. It involves the researcher observing first hand in the research setting. The researcher is free to be an active participant in the normal routines of the research.

82 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8204_WAL_SRM3_20170_TXT_SI.indd 82 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

EXERCISE 4.2: WRITING AN INFORMATION SHEET Imagine you are a researcher seeking to undertake a qualitative investigation of intravenous drug use among young people. You plan to use indepth interviews as your research method. Think about the following. • What are the ethical issues you will face? • How would you deal with them?

Now write an information sheet to accompany your application to an HREC. Use the example in Box 4.4 as a guide.

DATA MANAGEMENT Th e principles of anonymity and confi dentiality also have implications for the way data are recorded and stored. In addition to keeping identifying information such as contact details separate from the raw data, researchers need to be sensitive to their legal—or illegal—use by third parties. Th e Australian Research Council, for example, requires recipients of its grants to deposit their data in an archive that is accessible to other researchers, such as the Australian Social Science Data Archive (ASSDA). Th is raises questions of informed consent, confi dentiality, and ownership of material (Richardson & Godfrey 2003), and of how long the original researcher must keep the data.

DATA STORAGE According to the NHMRC National Statement on Ethical Conduct in Research Involving Humans, researchers should • remove identifying material from records at the earliest possible time • only collect personal information where it is absolutely necessary for the research • use coding procedures • store separately any data containing personal information • not store personal information on their hard drive • dispose of personal information as soon as possible, for example, destroy contact

information once data collection is complete.

SAFETY Issues of participant safety are especially relevant to medical research, which may involve placing research participants in situations that carry some physical risk. Some forms of psychological research, particularly experimental psychology, in which participants can be deliberately placed in stressful situations in order to see how they behave, also have a safety aspect. Th e most famous example of this is Milgram’s study on authority. One group of participants took on the role of teacher and were asked to apply what they incorrectly believed were potentially lethal electrical shocks to pupils if they failed to answer questions correctly (Milgram 1963, 1974). Th e pupils were another group of participants who had been instructed to fake signs of physical distress when the shock treatment was applied. Despite the apparent distress of the pupils, most teachers continued to apply the shock treatment. Th e study provided important insight into how perceptions of authority can lead people to physically harm others, but also became an example of the unethical use of deception. Th e study’s use of covert research techniques placed the teachers in a situation of moral

Raw data: Data that has been recorded but not manipulated or analysed.

8383CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8304_WAL_SRM3_20170_TXT_SI.indd 83 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

discomfort that had potential to cause ongoing distress. In Behind the Shock Machine, Perry argues (2012) that some participants remained traumatised years aft er the experiments.

Most social research does not expose participants to the risk of physical or psychological harm, but it has the potential to cause psychological discomfort by, for example, asking them to recall past or current experiences about which they have strong feelings, such as personal relationships or their experience of illness. Some research may also uncover information among a network of people that could be disruptive to their relationships.

Safety also relates to protection from legal harm. Researchers must take care not to place participants in legal jeopardy. If participants are asked to disclose involvement in illegal behaviour, such as illegal drug-taking or motor vehicle theft , researchers need to take special care to protect participants’ anonymity and the confi dentiality of any information provided. To prevent this, participants should be warned not to provide information about illegal activity that has not been dealt with by the judicial system.

Qualitative methods tend to involve greater ethical dilemmas than quantitative methods (Hopf 2004). Field research in particular can lead to diffi cult legal issues. A researcher may unintentionally observe criminal behaviour, making that researcher liable to becoming a witness in criminal proceedings. Or researchers may place themselves in situations where they deliberately witness criminal activity.

Stephen Tomsen’s research on the relationship between alcohol and violence in public places involved participant observation in pubs and clubs, some of which were chosen because of their reputation as places where violent fi ghts occur (Tomsen 1997). Tomsen was a witness to the violence, but sometimes had limited contact with perpetrators, victims, and others at the scene. Had someone been seriously injured, Tomsen could have found himself being subpoenaed as a witness and his research notes treated as legal evidence.

RESEARCH MERIT When potential participants are invited to participate in a research project, the request is justifi ed through its potential benefi ts to society. Ethical guidelines therefore require the research design and methods to meet discipline standards and be acceptable for publication in a peer-reviewed academic journal. It should also avoid researching a topic where a strong knowledge base already exists.

VULNERABLE PARTICIPANT GROUPS Incidences of abuse of some study populations, such as prisoners or children with a disability, by medical researchers has led to National Statement guidelines that specifi cally protect vulnerable populations. Vulnerable populations include • children and young people • people in dependent or unequal relationships including

– carers and people with chronic conditions or disabilities – health care professionals and their patients or clients – teachers and their students – prison authorities and prisoners – governmental authorities and refugees – service providers and especially vulnerable communities – employers and employees

84 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8404_WAL_SRM3_20170_TXT_SI.indd 84 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

• Aboriginal and Torres Strait Islander peoples • people highly dependent on medical care • people with an intellectual or mental disability • people who may be involved in illegal activities • people in other countries.

All vulnerable groups, such as the elderly, people with ill health or on low incomes, should be treated with special sensitivity by the social researcher. Special care is required to ensure that requirements such as informed consent are observed. In my research on people receiving treatment in an acute psychiatric ward, for example, obtaining informed consent was diffi cult, because potential respondents were oft en quite unwell and their medication sometimes meant they did not understand what they were being invited to be involved in. Th rough the HREC, special protocols were established to work with psychiatric staff on when and how to approach them.

ABORIGINAL AND TORRES STRAIT ISLANDER PEOPLES Submissions and personal representations from Indigenous peoples, and health and research organisations to the AHEC in the mid 1990s led to ethical guidelines for health research with and by Aboriginal and Torres Strait Islander peoples (NHMRC 2002, NHMRC 2003, 2004). Th ese guidelines introduce additional requirements to develop and conduct research in a manner that is respectful and inclusive of Aboriginal and Torres Strait Islander values and cultures (see Box 4.6). A key message to researchers is that there is a great diversity across the many Aboriginal and Torres Strait Islander cultures and societies. Although there is a set of common values shared, each society has the right to express how these common values and their own unique values should be addressed in research (NHMRC 2007).

BOX 4.6: NHMRC NATIONAL STATEMENT GUIDELINES ON RESEARCH WITH ABORIGINAL AND TORRES STRAIT ISLANDER PEOPLES

Research with Aboriginal and Torres Strait Islander peoples should • be conducted in a way that reflects the six core values of reciprocity, respect, equality,

responsibility, survival and protection, spirit and integrity, identified as being important to Aboriginal and Torres Strait Islander Peoples

• provide evidence of support from relevant Indigenous communities • ensure that its methods provide for mutually agreed mechanisms for such matters as

appropriate recruitment techniques, suitable information about the research, notification of participants’ consent and research progress, final reporting

• identify any potential negative consequences of the proposed research, to design processes to monitor them and to advise steps for minimising them

• adopt methods and processes that provide opportunities to develop trust and a sense of equal research partnerships

• provide fair opportunity for involvement of Aboriginal and Torres Strait Islander Peoples • ensure that benefits include the enhancement or establishment of capabilities,

opportunities or research outcomes that advance the interests of Aboriginal and Torres Strait Islander Peoples, that these should have been discussed with them, and that any realisable benefits should be distributed in a way that is agreed to and considered fair by these participants

• should demonstrate evidence of respectful engagement with Aboriginal and Torres Strait Islander Peoples and that the research approach should value and create opportunities

8585CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8504_WAL_SRM3_20170_TXT_SI.indd 85 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

to draw on their knowledge and wisdom through their active engagement, including interpretation of the research data

• that national or multicentre research should take care to gain local level support for research methods that risk not respecting cultural and language protocols.

Source: NHMRC 2007: 69–71

M ̄AORI PEOPLE In New Zealand, the Māori Health Committee of the Health Research Council of New Zealand has produced guidelines that all researchers are required to meet when conducting research projects on a topic relevant to a Māori health issue or Māori are to be involved as participants (Health Research Council of New Zealand 2010). Th is has developed against a background in which some Māori communities and groups have felt overresearched and disempowered, and have received few benefi ts from participation in the research. Building research capacity within Māori is part of the ethics framework. Researchers are expected to respect the principles of partnership and sharing that are implicit in the Treaty of Waltangi and, where applicable, incorporate these into all health research proposals. Th ey should develop collaborative research partnerships between Māori communities and non-Māori researchers, which may include providing training partnerships for emerging Māori researchers. Research institutions should be able to demonstrate the quality and extent of their partnership with Māori and researchers are expected to discuss policies and procedures for consultation with Māori within their host organisations. Before developing the proposal for any project involving Māori, researchers must fi rst consult and discuss it with a variety of Māori and Māori groups. Considerations should include how the project will eff ect Māori health, how it will benefi t Māori, how Māori will be involved, and whether a Māori researcher can be the lead researcher.

Recognition of Māori knowledge is emphasised. Th e term kaupapa Māori research refers to the ‘philosophy, theory, methodology and practice of research for the benefi t of Māori which is also produced by Māori’ (Health Research Council of New Zealand 2010: 7). Questions about who defi nes the research problem emphasise Māori ways of knowing, for whom the study is worthy and relevant, and who is making this claim; all issues fundamental to its conduct. Th is emphasis enables mainstream cultural assumptions about research problems and benefi ts to be repositioned so that research agendas can be integrated with, and informed by, Māori cultural realities and philosophies.

SOCIAL MEDIA Social media such as Facebook and Twitter are valuable research data sources. Th e data are inexpensive, accessible, do not require transcription and can usually be easily transferred to a database for analysis. Th e material is unmediated by the researcher and so avoids the Hawthorne eff ect, in which the behaviour of respondents is infl uenced by the research process itself. But research using social media raises questions of informed consent and copyright. When someone posts personal material on a publicly accessible site, does this mean that the researcher can analyse and reproduce it in public documents without seeking the poster’s consent? What about the intellectual property contained in their postings? By posting it publicly, does this mean that authors consent to whatever the researcher may choose to do with it?

Kaupapa M ̄aori research: The ‘philosophy, theory, methodology and practice of research for the benefit of M ̄aori which is also produced by M ̄aori’ (Health Research Council of New Zealand 2010: 7).

86 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8604_WAL_SRM3_20170_TXT_SI.indd 86 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Hookway (2012) was faced with these questions in his study of modern morality, which used blogs to examine everyday moral constructions and practices. He identifi ed three diff erent positions in relation to the question of informed consent (2008). • Informed consent is not required because online postings are using public spaces in a

way that is similar to any other publicly available material, such as newsprint articles or artworks in public galleries.

• Informed consent is required because even though they are publicly accessible, online postings are written with an ‘expectation of privacy’ (2008: 105).

• Th ere is no clear-cut position. Online postings are simultaneously public and private so researchers must be careful not to assume that, because material is accessible, its authors have constructed it as available for public use. In his own research, Hookway took the fair game–public domain position, arguing

that if bloggers wanted their postings to be limited to a particular audience, they would have applied the necessary settings. Since they chose not to do so, it can be assumed that, while these blogs are personal, they are not private (2008: 105).

Copyright and author anonymity are additional ethical issues. Online postings are subject to copyright law but so long as the purpose of their use is for research rather than commercial purposes, and meets criteria for fair dealing, such as the use of small excerpts then the material may be used without infringing legislation. Author anonymity presents a more diffi cult ethical issue since there is a tension between research ethics of preserving participant anonymity and the moral rights of attribution for authors in copyright law. Hookway resolved this tension by using pseudonyms for his bloggers and disguising potentially identifying information, drawing on provisions within copyright legislation in which non-attribution may be judged ‘reasonable’ (2008: 106).

ETHICS AND THE RESEARCHER Th e second stakeholder in any research project is the researcher, for whom, in a number of areas, ethical issues are important. Th ese include • personal safety • truthful reporting • acknowledgment of sources (as opposed to plagiarism).

PERSONAL SAFETY Personal safety of researchers is regarded by the NHMRC as an ethical issue (see, for example, Collyer 2004: 18–19). Concerns include • safety of interviewing in participants’ homes • cold interviewing (that is, with no formal introductory process), especially in settings

that may be unsafe, such as nightclubs • observation of, or involvement in, illegal activities • the psychological impact of researching sensitive issues.

TRUTHFUL REPORTING Th at fi ndings should be truthfully reported is axiomatic in social research. Yet bias in the reporting of results can be introduced because of attachment to a particular view. As discussed in Chapter 1, researchers do not undertake their work in a social vacuum, but

Plagiarism: The unacknowledged use of the work of another person so that the ideas or information appear to be one’s own.

8787CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8704_WAL_SRM3_20170_TXT_SI.indd 87 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

with a particular set of beliefs and assumptions that infl uence what they choose to study and how they go about their work, including the interpretation of their fi ndings. Th ere is always the possibility that attachment to a particular perspective will lead to judgmental attitudes towards the social group under investigation, thereby introducing bias into the study. Th e need to guard against such value bias is acknowledged in the Australian Sociological Association’s Code of Ethics. Article 10 states:

Members have a responsibility to consider the interests of those involved in their research, those aff ected by their work, and those utilising their fi ndings. Th is broad responsibility implies that in conducting research, and presenting the results, due consideration is given to the plurality of the social world, the diversity of beliefs and values, especially in the context of culture, age, gender and other aspects of social complexity, and to the requirements set out in affi rmative action and anti-discrimination legislation and policies (TASA 2008).

Some researchers deal with this issue by being explicit about their own values in the write-up of the research so that the research community can assess their contribution in an informed way.

Academic fraud is another aspect of truthful reporting. Most professional ethical codes within the social sciences include a statement about the unacceptability of presenting false or misleading results. Any researcher found to have engaged in academic fraud faces severe disciplining, professional ostracism and probable loss of employment. While blatant fabrication of fi ndings is rare in social research, it does occur. An Australian example is the case of William McBride. William McBride achieved fame for being the person who fi rst publicly identifi ed the link between the drug thalidomide and the occurrence of birth defects in babies. In a development quite unrelated to this he was later accused of systematically falsifying data in a paper published in the Australian Journal of Biological Sciences. An inquiry by his university confi rmed that this had taken place. Despite this, McBride was permitted to return to the board of the foundation that had conducted the research, however, his academic reputation was irredeemably tarnished.

Less obvious and understood are more subtle modifi cations of fi ndings such as the tweaking of responses to fi t an anticipated pattern or the misleading interpretation of fi ndings.

ACKNOWLEDGMENT OF AUTHORSHIP Authorship or ownership of research also have ethical dimensions. Th is is especially so when the research and ownership are embedded in the relationship between academics and honours and postgraduate students. What are the implications for authorship of the contribution supervisors make to their students’ production of knowledge? How this is acknowledged in any publication will vary, depending on the extent of the supervisor’s contribution: possibilities range from no acknowledgment to the supervisor being fi rst author. Most research institutions and professional codes have guidelines on how the contribution of students and supervisors should be refl ected in authorship.

BOX 4.7: GUIDELINES ON AUTHORSHIP OF PUBLICATIONS The Statement and Guidelines on Research Practice addresses the related area of authorship of publications. Each person who has made a substantial contribution to the research leading to a publication must be given the opportunity to be included as an author.

Academic fraud: The misrepresentation or falsification of research findings by the researcher.

88 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8804_WAL_SRM3_20170_TXT_SI.indd 88 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Someone who has not participated in conceiving, executing or interpreting at least part of the relevant research should not be included as an author. This may be a sensitive issue when the contributions of a candidate and his or her supervisors are considered in determining authorship and the order in which the authors’ names are listed. Some candidates are disinclined to acknowledge any contributions made by their supervisor, just as some supervisors are inclined to claim a result as theirs alone even though their last real contribution was two years ago. The question of authorship should be discussed at the earliest possible stage in a research project, and reviewed whenever there are changes in participation.

Candidates are encouraged to publish their research findings, if appropriate, throughout candidature. The Board strongly advises candidates, in consultation with their supervisors, to prepare a statement of authorship at the time each paper is submitted for publication. This statement should provide the details of the contribution of the candidate, the proportionate authorship of each of the coauthors and, where practical, should be endorsed by all of the coauthors. These statements should be kept on file in the school. If co-authored papers are to be included in the thesis, these statements will be needed to verify authorship.

University of Tasmania 2006: 31

ETHICS AND THE ORGANISATION Social research is usually practised within an organisation, oft en with more than one organisation bearing some responsibility for the research. Most university research is funded by an external body. Researchers apply for grants from government-funded bodies, such as the Australian Research Council or the National Health and Medical Research Council to assist them to conduct their research, or they tender for research consultancies with groups that require research, such as community sector or industry organisations. Th ere are a number of ethical issues aff ecting the funding sources of research activity.

DISCLOSURE OF FUNDING SOURCES How social research is funded raises ethical issues because of the potential for research bias in favour of the interests of the funding source, whether government or private. Th e potential for bias in biomedical research is an area of current debate, because of the potential profi ts for and the large fi nancial investment in research and development made by biomedical companies. Pharmaceutical companies in particular have a documented history of ethical violations in medical research (Braithwaite 1985).

Although less common in the social sciences than in medical research, there have been cases in which sponsors have compromised the independence of social sciences researchers (Blumer 1967). It is more likely, though, to be ideological rather than commercial interests that threaten the requirement for honesty in the conduct of social research. Th e role of social research in informing social policy means that fi ndings oft en have implications for government support for the funding body or for the population under investigation, raising potential ethical dilemmas about the reporting of fi ndings. Th e political nature of research is also apparent in debates about what projects public money should support, with claims sometimes being made that some projects were not suffi ciently in line with the nation’s strategic goals, and, as such, were a waste of public money. One of the ways in which governments seek to prevent such criticisms is to ensure that researchers identify

8989CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 8904_WAL_SRM3_20170_TXT_SI.indd 89 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

how their research contributes to identifi ed national objectives. Th e Australian Research Council requires all research proposals to classify their research under one of four socioeconomic objectives: defence, economic development, society, and environment. ‘Society’ covers ‘research and development directed towards the improvement of society’, and has four subdivisions: health, education and training, law, politics, and community services, and cultural understanding (ABS 2008).

For these reasons, ethical guidelines for research invariably carry a statement that all publications should acknowledge who has provided funding and how much it amounts to. In addition, they should declare any affi liation with the sponsor as well as any fi nancial interest.

EVERYDAY ETHICAL DILEMMAS Ethical dilemmas are routinely experienced, so the researcher must assess what is the appropriate course of action to achieve the research goals without compromising ethical standards.

VOLUNTEERS Th e ideal of voluntary participation sounds straightforward, but in practice there are many ways in which individuals may be subtly encouraged or manipulated into participation. A doctor asking a hospital patient to complete a consent form for participation in a research project may state that participation is voluntary, but the setting and the relationship may infl uence the patient to agree. One way of ensuring that potential participants are not in any way coerced to participate is for contact with the researcher to be instigated by the participant rather than the other way around. Th is usually means reliance on the dissemination of information about the project through the media, posters in strategic locations, postal surveys, and the use of a third party.

Postal surveys, for example, can work well, because they rely on participants to complete the form and mail it back to the researcher. But there are many projects in which reliance on participants to contact researchers is not realistic. Th is is especially true of many qualitative data collection methods, for which the populations being researched are oft en small and diffi cult to locate.

Some populations are also unlikely to contact researchers. Th is is especially true of those drawn from marginalised social groups. Participants may fi nd the idea of participation in research unfamiliar, unnecessary, and uncomfortable, and so be unwilling to engage in it. Some low-income groups, including Aboriginal, Torres Strait Islander and Māori peoples, and groups experiencing ill health, legal problems, or disabilities, may see involvement in a research project as just another demand, with no obvious benefi t. From the researcher’s perspective, the best way of reaching participants is to contact them to explain what the project is about and to discuss their possible involvement in it. Th is can be achieved indirectly through a third party, such as a service used by the research population. An employee of the service can make the initial contact and provide the potential participant with an information sheet, an outline of the research project, and can explain what involvement would mean. If the person expresses interest in participation, they can be asked to contact the researcher or asked if their contact details can be provided to the researcher.

90 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 9004_WAL_SRM3_20170_TXT_SI.indd 90 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

PAYMENTS TO RESEARCH SUBJECTS When research subjects agree to participate in a research project, they are donating their time, knowledge, and experience. Th ey may incur expenses, such as travel to and from an interview, arranging childcare or taking time off work. It is reasonable that they should receive some fi nancial remuneration for the inconvenience. Th is payment needs to be made at a level that cannot be seen as coercion or an inducement that transgresses the principle of voluntary consent. For this reason, HRECs provide guidelines on payments to research volunteers.

UNOBTRUSIVE, COVERT RESEARCH AND DECEPTION One of the ways in which codes of ethics work is to prevent ethically dubious projects from ever going beyond the stage of the researcher’s imagination. One contentious method is covert research, which is the inclusion of subjects in a research project without their awareness of, and/or agreement to, their participation. It usually involves a qualitative method, including the use of some kind of direct observation, especially participant observation.

Political involvement in marginal political organisations is one of a number of sensitive topics that can be diffi cult to investigate with the full knowledge of participants. Other areas include sexual behaviour, membership of marginal religious organisations, criminal involvement, and violent behaviour. Th ere is a long tradition of studies using covert techniques (see Case studies 4.2 and 4.3), which in itself suggests that covert methodologies are not always unethical. Unobtrusive methods need to be distinguished from deception, which involves deliberate misleading of participants by the researcher (Spicker 2011: 119). Tomsen’s use of unobtrusive observation is an example of covert research that does not involve deception (see Case study 4.3).

STUDIES USING COVERT METHODOLOGIES CASE STUDY 4.2: FIDDLING AND PILFERAGE IN A BREAD FACTORY Paul Ditton was a PhD student who had experienced employment in a factory. His study in industrial relations was investigating whether there is a difference in the output of workers according to whether they are paid by piecework or by the hour. He returned to the factory as a worker without revealing to the organisation or his co-workers that he was using his presence there as an opportunity to collect data. Working 12-hour shifts, he used the breaks to write up his observations of his co-workers’ behaviour. The title of his book, Part-time Crime: An Ethnography of Fiddling and Pilferage (1977), proclaims his main findings. It is likely that Ditton’s co-workers would not have cooperated with the research if they had been given the choice. On the other hand, the extent of fiddling in the factory would not have been revealed by conventional techniques.

CASE STUDY 4.3: A TOP NIGHT OUT Stephen Tomsen was interested in the relationship between crime and masculinities. He decided to explore this by looking at pub violence (Tomsen 1997). This involved spending time in pubs that had a reputation for violent confrontations between customers. Tomsen never revealed his identity to customers or staff, but appeared as a paying customer. He witnessed a number of fights between customers and between customers and bouncers.

Covert research: The inclusion of respondents in a research project without their awareness and/or agreement to their participation.

9191CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 9104_WAL_SRM3_20170_TXT_SI.indd 91 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Many of the concerns about covert research apply principally to deception rather than to methods that involve undeclared or undisclosed research in informal settings (Spicker 2011: 118) of the kind illustrated by Tomsen. Th ese include • violation of the personal liberty of the unwitting informants • betrayal of their trust • contamination of the research environment for subsequent researchers • discrimination against the defenceless and powerless • damage to the behaviour or interests of subjects • infl uencing the behaviour of the unwitting participants and thereby undermining

the integrity of research fi ndings (the Hawthorne eff ect) (Homan, in Spicker 2011: 119–20). Spicker argues there are strong practical and methodological reasons for non-disclosure

of research that does not raise ethical concerns. Th ese include avoiding the Hawthorne eff ect and conducting research in settings that involve brief and anonymous contact. He also points out that the ethics framework is project-based, but researchers develop their ideas over a period of time across projects and gain knowledge in informal ways. Th ey may change their agendas in the course of research and use information obtained fortuitously (2011). Formal consent cannot apply in these circumstances, instead researchers have to draw on their own integrity and moral frameworks, and ensure that these are fully informed by relevant research ethics guidelines.

VOICES IN THE FIELD: HANNAH GRAHAM

Hannah Graham is an associate lecturer in criminology and sociology and a PhD candidate at the University of Tasmania. In 2007, she evaluated a new mental health court diversion initiative with the Magistrates Court of Tasmania. Her study illustrates how differences in perspective regarding ‘vulnerable people’ in criminological research can emerge (see Graham 2011).

This study involved stakeholder consultation meetings, interviews, court observation, and statistical data collection, with the voluntary participation of practitioners (magistrates, police prosecutors, barristers, forensic mental health, advocates) and defendants with a mental illness. Ethical safeguards for defendant participation included: anonymity, confidentiality, advanced notice without the researcher present, an independently written statement of their rights and access to advocate support, forensic mental health support before and after interviews, and, for court observation, verbal clarification of informed consent in court.

The Human Research Ethics Committee (HREC) of Tasmania approved the research but, controversially, they stipulated that all defendants appearing in this open court had to read a detailed information briefing and sign a consent form to simply allow me to attend. Arguably, my observation yielded little sensitive information because the media could attend, and the names and charges of defendants were published in court lists on the internet and court foyer walls. We requested the HREC’s condition be changed to independently witnessed verbal consent—a request they rejected.

The Ethics Committee didn’t seem to fully recognise the difficulties of orchestrating this graciously in the crowded court foyer 15 minutes beforehand. It meant I had to approach every person in the foyer to ask if they were appearing in this special court. Defendants were often anxious, and being seen filling out these forms could identify them to others as ‘mentally ill’. I was mindful of catering for defendants with low literacy and impaired cognitive function, plus the added sensitivities of asking people

92 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 9204_WAL_SRM3_20170_TXT_SI.indd 92 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

with paranoid psychosis to sign forms. So we devised strategies to make the process more sensitive, including asking about ‘Court 4’ instead of the mental health list, and getting forensic mental health staff to approach defendants first (i.e. identifying them only to me, not others), then introducing them (and often their families and lawyers) to me, before giving them space to decide and to approach me if and when they were ready. Despite some early challenges, the research progressed well with full defendant participation.

Do you agree with the HREC’s decision? Why?

EXERCISE 4.3: ASSESSING THE ETHICAL DIMENSIONS Imagine that you are a member of a human research ethics committee (HREC) that has received an application from a social scientist seeking to use covert methods to examine the treatment of refugees in detention centres. Answer the following questions on how you might go about determining the ethical dimensions of this project. • What criteria would you use to assess the research project application? • What politico-legal considerations might influence your decision? • What conditions would you require to be met before you would be willing to consider

approving the proposal?

9393CHAPTER 4: ETHICS AND SOCIAL RESEARCH

DAPHNE HABIBIS

04_WAL_SRM3_20170_TXT_SI.indd 9304_WAL_SRM3_20170_TXT_SI.indd 93 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

CONCLUSION Attention to ethical considerations at all stages of the research process is an essential component of professional research. Because social research has an impact on people, directly and indirectly, the potential for social harm needs to be consciously, and carefully, addressed. Th is potential extends beyond research subjects; it includes the population group to which participants belong, funding bodies, the research profession, and anyone involved in the investigation, including the researchers. Because researchers lead the investigation and control most aspects of it, they have a direct responsibility to minimise the potential for harm to their participants, especially if participants are drawn from vulnerable social groups. Th ey need to operate with an awareness that their interests and those of their participants do not always coincide, and that all research must be bounded by ethical practice. Without ethical guidelines and ethical practice as an integral part of the research process, abuse and breach of participants’ trust can occur. Ethical considerations also vary according to the research method used. Ethical concerns particular to individual methods are noted within each methods section in the text (Chapters 6–14) and in the ten other methods modules that are available at www.oup.com.au/orc/walter.

MAIN POINTS

• Ethics is concerned with the establishment of a set of moral standards that govern behaviour in a particular setting or for a particular group.

• Ethical research is concerned with ensuring that ethical principles and values always govern research involving humans.

• Th ere are many stakeholders involved in research, including the participant, the researcher, the funding organisation, and the researcher’s employer.

• One of the main reasons guidelines for ethical standards in research are necessary is because of the power imbalance between the researcher and the researched.

• Diff erent methodologies place diff erent emphasis on the level of involvement of participants. In general, interpretative methodologies give space to participants’ views, quantitative approaches do not.

• Researchers are not disinterested observers motivated by pure considerations of knowledge, but have careers in which the success of their investigations is central. Paying careful attention to the moral and ethical requirements of research can get in the way of timely and successful projects.

• Feminist research and some types of action research try to overcome the potentially exploitative nature of research by involving participants in some or all stages of the research process.

• Any research undertaken or funded by a government organisation must receive ethical approval from a human research ethics committee before it can be conducted.

• Th e key Australian document that identifi es the principles for ethical research on humans is the National Statement on Ethical Conduct in Research Involving Humans (section 1).

DAPHNE HABIBIS

PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

04_WAL_SRM3_20170_TXT_SI.indd 9404_WAL_SRM3_20170_TXT_SI.indd 94 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

• Th e principles most concerned with the protection of participants are the principles of informed consent, anonymity, confi dentiality, and protection from harm.

• Th e principle of informed consent is concerned with ensuring that – research participants are fully informed about what the research is about and

what participation will involve – they make the decision to participate without any formal or informal coercion.

• Th e main mechanisms used to ensure that potential participants are fully informed and that participation is voluntary are the information sheet and consent form.

• Anonymity refers to the protection of participants from identifi cation as participants in the research.

• Confi dentiality is concerned with ensuring that the information provided by participants cannot be linked back to them.

• Qualitative methodologies give rise to more ethical dilemmas than do quantitative methodologies.

• Researchers need to take special care not to abuse the human rights of vulnerable groups, such as children or prisoners.

• Ethical considerations relating to the researcher are ensuring their personal safety, truthful reporting, and accurate acknowledgment of sources.

• Researchers must always disclose the funding sources of their research. • Covert research is the inclusion of subjects in a research project without their

awareness of, and/or agreement to, their participation. Covert research includes unobtrusive observation, which can be a valuable research method and is distinct from research involving deception of participants.

FURTHER READING American Association of University Professors. (2004). Protecting Human Beings:

Institutional Review Boards and Social Science Research. www.aaup.org/statements/ redbook/repirb.htm.

Australian Association of Social Workers. (2004). Code of Ethics. www.aasw.asn.au/ adobe/about/AASW_Code_of_Ethics-2004.pdf.

Australian Sociological Association. (2008). Code of Ethics. www.tasa.org.au/ethical- guidelines.

Grieg, A., Taylor, J. & MacKay, T. (2007). Doing Research with Children. London: Sage. Health Research Council of New Zealand. (2010). Guidelines for Researchers on Health

Research Involving Māori 2010, Version 2. www.hrc.govt.nz/sites/default/fi les/ Guidelines%20for%20HR%20on%20Maori-%20Jul10%20revised%20for%20Te%20 Ara%20Tika%20v2%20FINAL[1].pdf,

Israel, M. & Hay, I. (2006). Research Ethics for Social Scientists: Between Ethical Conduct and Regulatory Compliance. London: Sage.

9595

DAPHNE HABIBIS

CHAPTER 4: ETHICS AND SOCIAL RESEARCH

04_WAL_SRM3_20170_TXT_SI.indd 9504_WAL_SRM3_20170_TXT_SI.indd 95 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Mills, G. E. (2007). Action Research: A Guide for the Teacher Researcher. Upper Saddle River: Pearson Merrill/Prentice Hall.

National Health and Medical Research Council. (2007). National Statement on Ethical Conduct in Human Research. www.nhmrc.gov.au/PUBLICATIONS/synopses/ e72syn.htm.

REFERENCES Australian Bureau of Statistics. (2008). Australian and New Zealand Standard

Research Classifi cation Cat. No. 1297.0. www.abs.gov.au/ausstats/[email protected]/ Latestproducts/7D64C238F82B5DABCA25741800051071?opendocument. Accessed 4 October 2012.

Australian Sociological Association (TASA). (2008). Code of Ethics. www.tasa.org.au/ ethical-guidelines.

Barnes, J. (1979). Who Should Know What?. Harmondsworth: Penguin. Blumer, H. (1967). ‘Th reats from Agency-Determined Research: Th e Case of Camelot’.

In Horowitz, I. (ed.), Th e Rise and Fall of Project Camelot: Studies in the Relationship between Social Science and Practical Politics. London: MIT Press.

Braithwaite, J. (1985). Corporate Crime in the Pharmaceutical Industry. Canberra: Australian Institute of Criminology.

Christensen, P. & Prout, A. (2002). ‘Working with Ethical Symmetry in Social Research with Children’, Childhood, 9 (4): 477–97.

Collyer, F. (2004). ‘Ethics and Institutional Reform’, Nexus, 16 (2): 18–19. Craig, G., Corden, A. & Th ornton, P. (2001). ‘Safety in Social Research’, Social Research

Update, 29. www.soc.surrey.ac.uk/sru/SRU29.html. Dempsey, K. (1990). Smalltown: A Study of Social Inequality, Cohesion and Belonging.

Melbourne: Oxford University Press. Ditton, J. (1977). Part-time Crime: An Ethnography of Fiddling and Pilferage. London:

Macmillan. Duneier, M. (1999). Sidewalk. New York: Farrar, Straus & Giroux. Elmer-Dewitt, P. (1995). ‘On a Screen Near You: Cyberporn’, Time, 3 July, 38–45. Graham, H. (2011) ‘A Marriage of (In)convenience? Navigating the Research Relationship

between Ethical Regulators and Criminologists Researching “Vulnerable Populations” ’. In Bartels, L. and Richards, K. (eds.), Qualitative Criminology: Stories from the Field. Sydney: Hawkins Press.

Habibis, D., Hazelton, M., Schneider, R., Davidson, J. & Bowling, A. (2002). ‘A Comparison of Patient Psychiatric and Social Outcomes Before and Aft er the Introduction of an Extended-hours Community Mental Health Team’, Australian and New Zealand Journal of Psychiatry, 36 (3): 392–8.

Health Research Council of New Zealand. (2010). Guidelines for Researchers on Health Research Involving Māori 2010, Version 2, Health Research Council of New Zealand,

DAPHNE HABIBIS

PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

04_WAL_SRM3_20170_TXT_SI.indd 9604_WAL_SRM3_20170_TXT_SI.indd 96 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Auckland. www.hrc.govt.nz/sites/default/fi les/Guidelines%20for%20HR%20on%20 Maori-%20Jul10%20revised%20for%20Te%20Ara%20Tika%20v2%20FINAL[1].pdf.

Hookway, N. (2008), ‘Entering the Blogosphere: Some Strategies for using Blogs in Social Research’, Qualitative Research, 8 (1): 91–113.

Hookway, N. (2012). ‘Emotions, Body and Self: Critiquing Moral Decline Sociology’, Sociology: 1–17.

Hopf, C. (2004). ‘Research Ethics and Qualitative Research’. In Flick, U., von Kardorff , E. and Steinke, I. K. (eds), A Companion to Qualitative Research. London: Sage.

Humphreys, L. (1970). Tearoom Trade: Impersonal Sex in Public Places. Chicago: Aldine. Jones, D. G. (1990). ‘Contemporary Medical Scandals: A Challenge to Ethical Codes and

Ethical Principles’, Perspectives on Science and Christian Faith, 42: 2–14. www.asa3. org/ASA/PSCF/1990/PSCF3-90Jones.html.

Loblay, R. H. & Chalmers, D. (1999). ‘Ethics Committees: Is Reform in Order?’, Medical Journal of Australia, 170 (1): 9–10.

Mackintosh, N. (ed.). (1995). Cyril Burt: Fraud or Framed?. Oxford: Oxford University Press.

McCosker, H., Barnard, A. & Gerber, R. (2001). ‘Undertaking Sensitive Research: Issues and Strategies for Meeting the Safety Needs of All Participants’, Qualitative Social Research, 2 (1), article 22. http://qualitative-research.net/fqs/fqs-eng.htm.

Milgram, S. (1963). ‘Behavioral Study of Obedience’, Journal of Abnormal and Social Psychology, 67: 371–8.

Milgram, S. (1974). Obedience to Authority: An Experimental View. New York: Harper & Row.

National Health and Medical Research Council (NHMRC). (2002). Th e NHMRC Road Map: A Strategic Framework for Improving Aboriginal and Torres Strait Islander Health through Research. www.nhmrc.gov.au/research/srdc/indigen.htm.

National Health and Medical Research Council (NHMRC). (2003). Values and Ethics: Guidelines for Ethical Conduct in Aboriginal and Torres Strait Islander Health Research. www.nhmrc.gov.au/publications/synopses/e52syn.htm.

National Health and Medical Research Council (NHMRC) (2004). ‘Community Guide to the Ethics of Health Research’. Unreleased draft V. Canberra: Commonwealth of Australia.

National Health and Medical Research Council (NHMRC). (2006). Challenging Ethical Issues in Contemporary Research on Human Beings. Commonwealth of Australia www.nhmrc.gov.au/PUBLICATIONS/synopses/e73syn.htm.

National Health and Medical Research Council (NHMRC) (2007). National Statement on Ethical Conduct in Human Research. www.nhmrc.gov.au/PUBLICATIONS/ synopses/e72syn.htm.

Oakley, A. (1981). ‘Interviewing Women: A Contradiction in Terms’. In Roberts, H. (ed.), Doing Feminist Research. London: Routledge & Kegan Paul: 30–61.

Perry, R. (2012). Behind the Shock Machine: Th e Untold Story of the Notorious Milgram Psychology Experiments. Melbourne: Scribe.

9797

DAPHNE HABIBIS

CHAPTER 4: ETHICS AND SOCIAL RESEARCH

04_WAL_SRM3_20170_TXT_SI.indd 9704_WAL_SRM3_20170_TXT_SI.indd 97 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Richardson, J. C. & Godfrey, B. (2003). ‘Towards Ethical Practice in the Use of Archived Transcripted Interviews’, International Journal of Social Research Methodology, 6 (4): 347–55.

Rimm, M. (1995). ‘Marketing Pornography on the Information Superhighway’, Georgetown Law Journal, 83: 1849–1934.

Spicker, P. 2011 ‘Ethical Covert Research’, Sociology 45 (1): 118–33, Th omas, J. (1995). Th e Ethics of Carnegie Mellon’s Cyberporn Study. http://sun.soci.niu.

edu/~jthomas/ethics.cmu. Tomsen, S. (1997). ‘A Top Night: Social Protest, Masculinity and the Culture of Drinking

Violence’, British Journal of Criminology, 37 (1): 90–103. University of Tasmania, Board of Graduate Research. (2006). Graduate Research Resource

Book www.research.utas.edu.au/gr/docs/resource_book.pdf. Warwick, D. & Pettigrew, T. (1983). ‘Towards Ethical Guidelines for Social Science

Research in Public Policy’. In Callahan, D. and Jennings, B. (eds), Ethics, the Social Sciences and Policy Analysis. London: Plenum.

Warwick, D. P. (1982). ‘Tearoom Trade: Means and Ends in Social Research’. In Bulmer, M., Social Research Ethics: An Examination of the Merits of Covert Participation. London: Macmillan.

DAPHNE HABIBIS

PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

04_WAL_SRM3_20170_TXT_SI.indd 9804_WAL_SRM3_20170_TXT_SI.indd 98 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Sampling BRUCE TRANTER

BUSHRANGERS IN THE SYDNEY MORNING HERALD: A POPULATION STUDY While the focus here is upon sampling, sometimes it is possible to conduct research

using the whole population the researcher wants to study. Tranter and Donoghue

(2010) performed a content analysis to examine articles on bushrangers in a major

Australian newspaper, the Sydney Morning Herald (SMH ). They were interested in

understanding why bushrangers such as Ned Kelly remain so well known so long after

their deaths, and how their portrayal contributes to Australian foundation myths.

Using key word searches for articles on bushrangers, the authors extracted 348 articles

published between 1987 and 2004 from the SMH electronic database. Tranter and

Donoghue found that Ned Kelly was by far the most frequently mentioned bushranger,

appearing in 60 per cent of the articles, followed by Captain Thunderbolt (9 per cent),

Ben Hall (8 per cent), and Captain Moonlight (4 per cent). Appearing frequently in

1987, prior to the Australian bicentennial celebrations, articles referring to Ned Kelly

peaked sharply again in 2001 following renewed interest in historical figures during

Centenary of Federation celebrations. Tranter and Donoghue argued that the legend

of the armoured outlaw Ned Kelly, as depicted in books, film, and art, is an important

aspect of Australian culture. Journalistic accounts of bushrangers that emphasise the

mythical over more historically accurate accounts continue to feed the legend of the

antihero in Australian folklore. Importantly, this research is not based upon a sample of

bushranger articles, but on all articles (that is, the entire population) on bushrangers in

the SMH between 1987 and 2004. Given that the authors analysed population data,

they did not need to report inferential statistical tests with their results because the

results were not estimates of population characteristics based upon a sample, but the

actual population values (parameters).

KEY TERMS confi dence interval

confi dence level

convenience

sampling

non-probability

sampling

population

probability

sampling

purposive sampling

quota sampling

sample non-

response

sample

sampling error

sampling frame

self-selected

sampling

simple random

sampling

snowball sampling

stratifi ed sampling

systematic

sampling

05_WAL_SRM3_20170_TXT_SI.indd 9905_WAL_SRM3_20170_TXT_SI.indd 99 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

WHY WE SAMPLE Th e above vignette of population-based research raises the inevitable question: Why do we sample? Sampling is the process by which researchers usually select their unit of analysis (see Chapter 3), and there are four main reasons we draw samples rather than survey or interview a whole research population. Th e fi rst two are more applicable to quantitative research, the fi nal two more relevant to qualitative research. 1 It is generally not practicable to survey all members of a given group. If, for example,

we wanted to estimate how satisfi ed Australians are with the federal government’s imposition of a carbon tax, it would not be feasible to survey all Australians. Even if we limited the study to only those people living in one Australian state, the number of potential respondents would still be too large to sample.

2 If we select our sample using probability sampling techniques (as explained below), we will be able to draw inferences that apply to the whole population of interest. Using a well-designed sampling strategy and a sample of suffi cient size enables us to make general statements about our population with a known degree of confi dence about how accurate our results are. So, drawing upon the tenets of probability theory, it is not necessary to survey the entire population.

3 In qualitative research, there is no need to work with the whole group, because we are interested in meanings and understandings, not in statistics. In such work, larger samples do not necessarily off er an advantage, and non-probability samples (see below) of ten, twenty, or thirty (the size will depend on the research question) can provide the rich detail and nuanced understandings required.

4 For many units of analysis there is simply no existing complete list of elements, so it is impossible to work with a whole population. A sample of a particular group, or particular artefact, such as fi lm posters, may be the only option, even though representativeness cannot be established.

DEFINING THE POPULATION Samples are selected from our population of interest. In this context, the meaning of the term ‘population’ diff ers from its everyday use. For research purposes, a population is the collection of all the units we want to study, and it is important that we are able to delineate precisely and unambiguously who or what comprises the population. If, for example, we aim to investigate the proportion of workers who plan to stay in the workforce past 65 years of age, we need to decide whether to include all adult workers in our study, or to limit the defi nition of our target population to older workers, say those aged 50 and older. Th ere are also other considerations in determining our target population. Given that the literature suggests retirement decisions are oft en infl uenced by the retirement plans of a spouse, do we interview couples or individual workers? Making these sorts of decisions is usually relatively straightforward, and is guided by our research question and our resources. Th e important thing is to defi ne our population clearly and unambiguously before we begin selecting our sample.

SELECTING OUR SAMPLING METHOD Th ere are two main types of sampling: • probability sampling • non-probability sampling.

Probability sampling: Samples selected in accordance with probability theory. Probability sampling relies on the fact that we have a list of all elements or cases in the population we are studying, and that we are able to randomly select elements or cases from this list and that all cases or elements have an equal chance of being selected.

Confidence: The quantitative researcher is, more often than not, very interested in specifying the level of certainty with which a relationship between variables within a data set can be generalised beyond the confines of a specific sample to some wider population.

Non-probability sampling: A sampling process that does not use probability sampling techniques to select a sample. Examples include convenience sampling, snowball sampling, and self-selected sampling.

Representativeness: What quantitative researchers want their sample to achieve with respect to a population of interest.

Population: The collection of all the units that we want to study.

100 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10005_WAL_SRM3_20170_TXT_SI.indd 100 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Probability sampling is exclusively connected to quantitative research methodologies, although some qualitative studies do incorporate an element of representativeness in their sampling. Surveys (Chapter 6) mostly use probability sampling and quantitative analysis techniques (see Chapters 8 and 9). Th ese chapters, as well as the quantitative analysis extension modules on the website, presume the use of probability sampling. Non-probability sampling is mostly associated with qualitative methodologies, but can be utilised in quantitative projects (usually with conditions attached) if a probability sample is not feasible. In this text, methods in which non-probability sampling would be used include indepth interviews (Chapter 10) and discourse analysis (Chapter 12). Also, narrative research, conversation analysis, memory work, and sensory and visual methods are detailed in the online modules available at www.oup.com.au/orc/walter. Because of their alignment with diff erent methodological approaches, probability and non-probability methods are discussed separately in this chapter, although some researchers may combine both methods.

BOX 5.1: SOME SAMPLING DEFINITIONS Sample: A set of cases or elements selected from a population. Population: The total collection of all the cases or elements that we want to study (such as Australians aged 18 and over). Study population: The finite set of individuals (or other unit of analysis) that we intend to study, such as Australians aged 18 and over registered on the Australian electoral roll at 1 October 2010. Population parameter: An aspect of our population that we want to measure or estimate (such as the actual proportion of Australian adults who voted for the Greens at the 2010 federal election). Sample statistic: Used in probability sampling, and is an estimate of a population parameter generated from sample data (such as an estimate of the proportion of Australian adults who voted for the Greens at the 2010 federal election). Sampling frame: The list of all of the cases or sampling units from which we will select our sample, such as the Australian electoral roll.

PROBABILITY SAMPLING In probability sampling, samples are selected in accordance with probability theory, using a procedure that gives every member of the population a known probability of selection (Maisel & Persell 1996: 5). Probability sampling, an extremely powerful technique, is the primary method of selecting large, representative (and ‘representative’ is the key word) samples for social and political research. Probability sampling enables researchers to use comparatively small samples to draw accurate inferences about a large population. Th is is especially useful in cases in which the population is too large to be observed directly, such as the population of Australian voters or the population of Vietnam War veterans. Another advantage of probability sampling is that it prevents the researcher from consciously or unconsciously choosing a sample in a biased way.

Probability sampling relies on two core assumptions. First, we need a list of all of the elements in the population that we want to study: our sampling frame. In order to

Sampling frame: A list that represents as closely as possible an abstract population, the researcher uses this list to draw a probability sample.

101101CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10105_WAL_SRM3_20170_TXT_SI.indd 101 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

survey Sydney high school students, for example, we would need a list of all high school students in Sydney. But such lists are rarely complete. Some students may have left school, changed schools, or changed their address, while others may have commenced studying in the time between the preparation of the sampling frame and the administration of the survey to students. Such situations, while imperfect from an absolute accuracy perspective, still fulfi l our requirement for a list of all elements in our population. By defi nition, populations that are social, whether people or other social artefacts, are never static. Th e result is a slightly imperfect sampling frame, a usually unavoidable reality of selecting probability samples.

Second, probability sampling assumes that we are able to randomly select elements from our sampling frame according to probability theory. Th is means that each element of the sample has a known chance of being selected in the sample, independent of any other event in the selection process. Probability sampling has a mathematical justifi cation in probability theory, and enables us to determine how accurately our sample characteristics (sample statistics) estimate our population characteristics (population parameters) by quantifying the degree of sampling error (see Box 5.2 for an explanation of sampling error).

SIMPLE RANDOM SAMPLING Simple random sampling (SRS) is the sampling strategy upon which probability sampling theory and inferential statistics are based. SRS requires an accurate sampling frame, in other words, a complete list of all of the elements from which the sample will be drawn. Th e basic idea here is that a sample will be drawn from the population at random, free from any bias or pattern in the selection process. In the past, SRS was accomplished by means of the tedious process of selection using tables of random numbers. Luckily, many databases and other computer programs will draw samples at random, again assuming the researcher has access to an appropriate list of sampling elements. Computer programs can be used to assign numbers to the sampling elements, and then draw a sample of the required size at random from the sampling frame. If we know, for example, that all Tasmanian private telephone numbers fall between 6200 0000 and 6499 9999, we could program a computer to randomly dial from this number range to select a sample. SRS can generate unbiased estimates of population parameters when a complete list of the sampling elements is available, although in practice systematic sampling will produce samples of similar quality, which also enables researchers to calculate inferential statistics.

SYSTEMATIC SAMPLING While SRS seems like a relatively straightforward approach to probability sampling, in practice many researchers tend to employ systematic sampling. As is SRS, systematic sampling is a probability sampling approach. It enables the calculation of sampling error, and therefore also enables the researcher to make inferential claims about the population based on the sample from which it was drawn.

To systematically draw a sample, the researcher fi rst needs to decide on an appropriate sample size. Once again, in systematic sampling it is necessary to have an accurate list of sampling elements of a known size. Next, the researcher divides the number of elements in the sampling frame by the required sample size. Th is provides the ‘sampling fraction’ (de Vaus 2002: 72). If, say, the sampling frame comprised 200 000 cases and we require a

Simple random sampling: Sampling randomly using probability, but with no additional techniques such as stratification.

Systematic sampling: A probability sampling approach that uses a sampling fraction to determine the system of drawing a random sample.

102 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10205_WAL_SRM3_20170_TXT_SI.indd 102 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

sample of 1000, we would have a sampling interval of 200 (that is, 200 000 ÷ 1000 = 200). Th e next step is to take a random starting point and select every 200th case from the sampling frame until a sample of 1000 cases is drawn.

Systematic sampling is an effi cient, relatively straightforward method of sampling so long as the sampling frame is not ordered in a way that is at all cyclical. Assume we wanted to systematically sample houses from streets in a particular suburb. We have a list of the street addresses of the houses, and in our list the houses are listed as odd number, even number, odd number, and so on. Aft er taking a random start, for example on an odd street number, then selecting every tenth house, we would end up with a sample that consisted of all odd numbers. Odd and even numbers usually coincide with a particular side of a street, so we would end up with a sample that contained only houses from one side of the street. In some suburbs, houses at the top of hilly streets tend to have better views, and so are likely to be more expensive than those on the lower part. In such a case, our systematic sample could result in sample bias because we have selected higher-income households.

Th e problem of selecting a biased sample from a sampling frame that is itself cyclically ordered can be avoided by using a computer program to fi rst randomise the elements in the sampling frame, then systematically sampling from the randomised list of elements. Alternatively, the fact that the sampling frame is cyclically ordered or stratifi ed can be used to the researcher’s advantage if we want to capture a particular subsample, such as those living on the upper side of streets.

STRATIFIED SAMPLING Stratifi ed sampling is not a probability sampling method, but can be used in conjunction with probability sampling to enhance the representativeness of a sample. Stratifi ed sampling is an attempt to reduce the level of sampling error by splitting the population into subgroups or strata, and then drawing a sample using a probability selection process such as SRS or systematic sampling from within the strata. Th e example in Case study 5.1 (below) demonstrates how stratifi ed sampling is oft en used in Australia to make sure that a large enough sample is obtained from each state and territory to enable comparisons to be made.

CASE STUDY 5.1: DISPROPORTIONATE AND PROPORTIONATE STRATIFIED SAMPLING In 1999 a group of political scientists conducted the national Australian Constitutional Referendum Survey (ACRS), a mailout, mailback administration with the sample drawn from the Australian electoral roll. The referendum asked Australians if they wanted to change the Australian head of state from the constitutional monarch to a president appointed by the parliament. It also asked whether the preamble to the Australian Constitution should be altered. The sample used was stratified on the basis of the states and territories. A simple random sample or a systematic sample drawn from the electoral roll would tend to include more people from the most populous states, such as New South Wales and Victoria, and very few from Tasmania and the territories. As the researchers wanted to make comparisons between the states with large enough subsamples for meaningful statistical comparisons, they sent 1000 questionnaires to potential respondents in each of the states, 100 to potential respondents in the ACT and fifty to the Northern Territory. This is known as a disproportionately stratified sample, as the smaller states were deliberately oversampled as a proportion of the population of Australian adults.

Stratified sampling: Often used with probability sampling to ensure that certain subgroups (such as an equal number from each state and territory) are included at an appropriate level within the sample.

103103CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10305_WAL_SRM3_20170_TXT_SI.indd 103 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

The table below is reproduced from the ACRS codebook. It shows the population in each state and territory in thousands, the size of the sample drawn, the numbers in In scope, which is the number of surveys sent after refusals and return to sender questionnaires are removed, the actual number of responses, and the response rate for each state and territory.

State Population (000s)

Sample size In scope Response number

Response rate %

New South Wales 4133 1000 948 507 53.5 Victoria 3161 1000 941 557 59.2 Queensland 2222 1000 926 517 55.8 Western Australia 1172 1000 944 548 58.1 South Australia 1028 1000 945 607 64.2 Tasmania 326 1000 949 588 62.0 Australian Capital Territory

212 100 95 57 60.0

Northern Territory 108 50 45 22 48.9 Unknown 28 28 -

Total 12 362 6150 5821 3431 58.9

Source: Adapted from Gow et al. (2000)

SAMPLING ERROR, CONFIDENCE LEVELS AND CONFIDENCE INTERVALS Th e use of a sample instead of the whole population usually involves a degree of error. In practice, the true value may vary by a few percentage points above or below our sample statistic. To work out how much our sample value varies from the whole population value, we need to determine our sampling error, confi dence levels, and confi dence intervals.

SAMPLING ERROR In probability sampling, the sampling error is the measurement error or diff erence between a sample statistic and the actual population parameter it represents. Th e method for calculating sampling error is outlined in Box 5.2. For the main purposes of this chapter, just be aware that in all studies undertaken using probability sampling there will usually be some degree of diff erence between the sample estimate and the population value. Th e sampling error enables us to estimate the size of that diff erence.

CONFIDENCE LEVELS All such studies are also conducted with a certain confi dence level. Th e confi dence level refers to the degree of confi dence that researchers can have when making claims about their statistical estimates—in this case, predicting an election outcome. Th is level of confi dence is expressed as a percentage or a probability (for example, the 95 per cent level corresponds to a probability or p value of .05). Social scientists oft en use the 95 per cent

Sampling error: The error in sampling statistics that result from the difference between a sample and its population.

Confidence level: Tells us how confident we can be that our population parameter lies within a given confidence interval.

104 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10405_WAL_SRM3_20170_TXT_SI.indd 104 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

or 99 per cent confi dence levels (also referred to respectively in relation to the .05 and .01 probability or p values). If, for example, we choose the 95 confi dence level and calculate statistics based upon a probability sample, we can be 95 per cent confi dent that our sample estimate will hold in the population. Th is essentially means that, while we are 95 per cent confi dent, we also consider that our sample estimates could be wrong 5 per cent of the time. Most social science research uses a confi dence level of 95 per cent.

CONFIDENCE INTERVAL Next, combining the confi dence level and the sample error we are able to estimate the confi dence interval for our results. Th e confi dence interval refers to the range, usually expressed in percentage points above and below our sample result, where we estimate that the true population value actually lies. So it is the interval associated with a particular confi dence level that encompasses the true population parameter we are seeking to estimate.

AN EXAMPLE Let’s assume that a political scientist collected a sample of voters and asked questions about who they would vote for in the House of Representatives if a federal election were to be held next Saturday. Th e political researcher draws a random sample from the most recent Australian electoral roll, and estimates the proportion of people who said they would vote for the Labor Party and the proportion who claimed they would vote for the Coalition (Liberal and National) parties. In this hypothetical example, 54 per cent said they would vote Coalition and 46 per cent for Labor (voting for other political parties is excluded here). But this result is from a sample, and we know the sample usually varies at least a little from the whole population. Working through this result from a sampling error, confi dence level, and confi dence interval base allows us to estimate from this sample result what the parameters of the actual, or true, two-party voting intentions are in the population of Australian voters.

First, the researcher needs to calculate the sampling error for this binomial voting variable (that is, the variable in which voters are counted as indicating whether they would vote Labor or Coalition). Th e sampling error in this case is calculated to be 1.8 percentage points. Without engaging in a detailed explanation of statistical concepts such as Z cores and the standard normal distribution, suffi ce it to say that the researcher then draws upon probability theory to calculate the confi dence interval, or the range that encompasses the true intended vote among our population of Australian voters. To construct a confi dence interval for this proportion, with 95 per cent level of confi dence, probability theory suggests that we multiply the sampling error by 1.96 (corresponding to a Z score of 1.96). Th e researcher then subtracts this result from the sample statistic (the estimated Coalition vote of 54 per cent) to calculate the lower bound of the confi dence interval (54 − [1.8 × 1.96] = 50.47), and adds it to the sample statistic to calculate the upper bound (54 + [1.8 × 1.96] = 57.53).

So, as the Coalition vote is estimated at 54 per cent, and the sample error at 1.8, at the 95 per cent confi dence level, the confi dence interval for the Coalition vote is 54 per cent ± approximately 3.5 percentage points, or between 50.5 and 57.5 percentage points. Th e 95 per cent confi dence interval for the Labor vote is 42.5 to 49.5 percentage points, so that the upper bound for Labor is less than the lower bound for the Coalition. As the confi dence

Confidence interval: The statistical range (interval) of our results within which we estimate the true population figure lies.

105105CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10505_WAL_SRM3_20170_TXT_SI.indd 105 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

intervals for the two parties do not overlap, the researcher can say with 95 per cent confi dence that the true Coalition vote in the population is somewhere between 50.5 and 57.5 per cent (that is, the lower bound is just over the magic 50 per cent), as a result we could predict that the Coalition would win an election based upon these results.

Th e confi dence level chosen by the researcher also has an impact upon the confi dence intervals. If our political scientist wanted to be even more certain of the election outcome, the confi dence level could be increased to 99 per cent. However, for a given sample error, if the confi dence level is increased the confi dence interval will widen, because, according to probability theory, at the 99 per cent level of confi dence the sampling error should be multiplied by 2.58 (that is, by a Z score of 2.58).

At the 99 per cent confi dence level, the confi dence interval for the Coalition increases to 54 per cent ±4.6 percentage points, or between 49.4 to 58.6, and Labor between 41.4 and 50.6, so that the intervals for the parties now overlap. In other words, the researcher would not be able to predict with 99 per cent confi dence that the Liberal/National coalition would win the election, based on these results. In order to make a more accurate prediction of the intended vote at the 99 per cent level, the researcher would need a larger sample in order to reduce the size of the sampling error.

BOX 5.2: SAMPLING ERROR In the social sciences we are often interested in measuring variables that are categorical (nominal or ordinal level variables), such as the percentage of Australians who are in favour of changing from a constitutional monarchy to a republic, how many consider themselves to be middle class, or how many participate in voluntary community organisations.

It is therefore useful to consider how to calculate standard errors (that is, sampling error) for proportions, and to construct a confidence interval around the proportion that encompasses the true population value (parameter) at a certain level of confidence. In statistical jargon, the standard error we are calculating is called the standard error for binomial variables.

SE = p x q

• SE is the standard error for the binomial distribution. • p is the percentage of the category of the variable we are interested in. • q is the percentage for the remaining category or categories of the variable of interest. • n is the sample size.

If we work though this simple formula we can see the importance of the heterogeneity or homogeneity of the variable in question and the size of the sample in the calculation of the standard error for proportions. For example, in the 2012 Australian Survey of Social Attitudes the proportion of Australians who claim to be members of an environmental group is 7.5 per cent and the number of valid cases (sample size) is 1856.

We can use this information in our estimation of the standard error. With a sample size of 1856, the standard error is the sqrt([7.5 × 92.5] ÷ 1856) = 0.61 rounded. While there is always some error associated with estimates of population parameters based on sample data, we can use the standard error to create confidence intervals around our estimates (that is, 0.61 × 1.96 = 1.2) Confidence intervals are statistics that can be used to make inferential claims about a population. Based on these estimates of environmental group members from the AuSSA data, we can be 95 per cent confident that the proportion of environmental group members in the population of Australian adults is 7.5 ± 1.2, or between 6.3 and 8.7 per cent.

106 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10605_WAL_SRM3_20170_TXT_SI.indd 106 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

DECIDING ON AN APPROPRIATE SAMPLE SIZE Th e question everyone asks at some stage is: How many cases do we need in our sample? Th e answer ‘it depends’ is unsatisfying. While a detailed discussion of the calculation of sample size is beyond the scope of this chapter, the necessary probability sample size for any project is basically determined according to four factors: the amount of variation in key variables in the population, the confi dence level set by the researcher, the amount of sampling error the researcher is prepared to tolerate, and the population size. Th ere are formulas available for calculating suitable sample size based on these contingencies.

Other things being equal, larger samples produce smaller standard errors, and therefore provide us with more accurate estimates of our population parameters. As a rule of thumb, to reduce sampling error by half it is necessary to quadruple the sample size. Randomly drawing a sample of, say, 100 university students to examine their attitudes toward an increase in student course fees would enable us to collect data on this topic, but the sampling error might be too large to enable us to make accurate estimates. Th erefore, if our budget enables the collection of a larger sample, it may be reasonable to increase the sample size to 400 cases. Th is fourfold increase in sample size would halve the sampling error.

Yet there is also a point where increasing sample size to reduce sampling error will provide diminishing returns. Other things being equal, a sample of 2500 has a sampling error of approximately ±2 percentage points. Reducing the sample error to ±1 percentage point would require a sample of 10 000 cases: a large increase in sample numbers for not a large change in sampling error. Social researchers are usually prepared to tolerate a sampling error of ±2 to 3 percentage points. Th is is why many academic researchers and public opinion pollsters tend to draw samples of between 1000 and 2000 a case (for example, the 2007 Australian Election Survey has 1783 cases). Confi dence level also plays a role. Being able to claim with 95 per cent confi dence that the true population value (parameter) lies within our confi dence intervals is generally acceptable. Being 99 per cent confi dent is more accurate, but the cost of increasing our sample size to achieve this level may not be judged worth the additional expense.

Th e homogeneity or heterogeneity of a sample, in other words the amount of variation in key variables, is also an important consideration. Because of the way sampling error is calculated (see Box 5.2), opinion pollsters who attempt to estimate voting intentions during a close election campaign where voting intention is assumed to be a homogeneous variable—when two parties each appear close to achieving 50 per cent of the vote—will require a relatively large sample to diff erentiate between the parties. Alternatively, when there is a lot of variation in a variable of interest (heterogeneity), such as the proportion of Australian adults who join environmental groups (7.5 per cent of Australian adults, according to the 2012 Australian Survey of Social Attitudes), smaller samples can be used. Again, other things being equal, the more variation in the variables we are interested in studying, the smaller the required sample size.

SAMPLE NON-RESPONSE A particular problem faced by survey researchers is referred to as sample non-response. Th is term covers the number of sample respondents, who, for whatever reasons, decide not to participate. (Th is topic is discussed in relation to surveys in Chapter 6.) Survey response rates have been declining in many countries (Porter 2004). In Australia, for example, the

Sample non-response: The number of sample respondents who do not participate in the research.

107107CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10705_WAL_SRM3_20170_TXT_SI.indd 107 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

Australian Election Study has shown a decline in recent years. Large falls in response rates may dramatically increase the costs of conducting a survey. As response rates decrease, it is necessary to make more calls before securing each telephone interview or to send out larger numbers of surveys in the mail to achieve a sample of the desired size.

To calculate response rates, the sample size is reduced by the number of potential respondents who were either ineligible (did not complete the survey correctly or did not meet sample criteria) or who were unreachable (phone number disconnected or survey returned undelivered). Th is number is then used to divide the number of completed surveys and multiplied by 100 to attain a percentage. Th e formula is

Response rate = Number of surveys completed × 100 N in sample – (ineligible and unreachable)

Sample non-response of itself is not necessarily an issue, but it becomes a problem when non-respondents diff er from respondents in a non-random manner. In studies of the general population, people who are unlikely to participate in surveys tend to be older and to have low levels of formal education (de Vaus 2002). Th e problem is that this introduces sampling bias, as samples drawn from the general population will be more highly educated and younger than the population from which they are drawn. Given that age and education are associated with many of the phenomena that survey researchers are typically interested in studying, such as social and political attitudes, the potential arises for estimates based upon such samples to be biased.

WEIGHTING TO ADJUST FOR SAMPLE NON-RESPONSE? Weighting the responses is one possible action. Weighting is most oft en done to adjust the sample to be proportionately representative, such as for the whole of Australia, with correspondingly adjusted weighted responses for each state and territory. Weighting was used in the data gathered in Case study 5.1 to adjust between states and territories and national level analyses. It is also possible to statistically adjust for sampling non-response by weighting the sample, that is, we can adjust our data by multiplying the categories of particular variables by a factor so that the sample proportions for a given variable are the same as the proportions of the population parameter. Th is approach requires the researcher to select some key variables that are not representative of the population parameters, for example, demographic variables such as age, sex, and education level that can be compared with their respective parameters in the population from which the sample was drawn. Variables are then created to adjust the sample statistics so that they resemble the population parameters. As an example, we calculate from the 2006 Australian census data that the proportion of women aged 18 and over in Australia is 51.27 per cent and the percentage of adult men is 48.73 per cent. Let’s assume we have drawn a sample where 56 per cent of respondents are women and 44 per cent are men. Clearly, compared to the population, in our sample women are overrepresented and men underrepresented.

Given that the 2006 census is eff ectively population data, we can adjust our sample data by multiplying the proportion of men and women by a factor so the sample proportions for a given variable are the same as the proportions in the population. To do so, we divide the population proportions for men and women by the sample proportions for men and women as in Table 5.1 to create a weight variable, then, using a statistical package such

108 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10805_WAL_SRM3_20170_TXT_SI.indd 108 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

as SPSS or STATA, statistically adjust our sample estimates. In this example, we would reduce the proportion of women and increase the proportion of men in our sample so they resemble the population.

Weighting our sample according to key variables such as age, sex and education is particularly useful when our aim is to estimate summary statistics. As an illustration, to estimate the median income of Australians involved in paid work from a labour force participation survey, it would be important to fi rst check the age, sex and education profi le of our sample, because all of these variables have an infl uence upon income. On average, men tend to receive higher incomes than women, tertiary graduates earn more than non-graduates, and very young or old people are less likely to be involved in paid work and therefore to be on lower incomes. Th e sample could therefore be weighted to refl ect the age, sex and education profi le of all Australian adults before estimating the median income.

Table 5.1: Calculation of simple weight variable for sex

Variable attribute

Population % Sample % Weight = (population % ÷ sample %)

Multiply sample by weight

Weighted sample

Women 51.27 56 0.9155 56 × 0.9155 51.27 Men 48.73 44 1.1075 44 × 1.1075 48.73

CASE STUDY 5.2: DECLINING RESPONSE RATES The longest running and most important academic social and political survey in Australia is the Australian Election Study (AES) series. The first academic attitudinal surveys in Australia, and precursors to the AES, were the Australian National Political Attitude surveys administered by Don Aitkin in 1967, 1969, and 1979. The AES has been administered as a self-completion mailout, mailback survey from the Australian National University by political scientists such as Roger Jones, and in more recent years, the team of Ian McAllister, Clive Bean, David Gow, and their collaborators.

The AES has been administered in a relatively similar format at the time of every federal election since 1987, and contains a range of consistently worded behavioural and attitudinal questions on topics such as voting behaviour, election issues, environmental concerns, and value orientations. By asking questions in an identical format at each election, researchers are able to examine changes in attitudes and behaviours over time, for example: How has the proportion of Australian adults who agree that abortion should be widely available changed between 1987 and 2010?, or: To what extent has the percentage of Australians who identify with a particular political party declined?

The AES achieved impressive response rates, particularly for a mail survey, as the researchers employed a range of strategies in order to maximise responses. Samples were systematically drawn from the Australian Electoral Roll by the Australian Electoral Commission. Individually addressed envelopes containing questionnaires were sent to respondents on the Monday following each election, which is always held on a Saturday. The signed letter described the study, guaranteed confidentiality for participants, and included a return postage-paid envelope. After one week a follow-up thank you/reminder postcard was sent, then after three weeks a second questionnaire, letter and return post-paid envelope were mailed to those who had not responded or had not asked to be excluded from the study (McAllister & Clark 2007: 74).

109109CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 10905_WAL_SRM3_20170_TXT_SI.indd 109 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

In recent years, the response rate to the AES has declined quite sharply (see Table 5.2). From the high points of 63 per cent in 1987 and 1993, the response rate has declined since the late 1990s, dropping markedly from 55 per cent in 2001 to 45 per cent in 2004 and 40 per cent in 2007–10. While we can only speculate as to the reasons for the decline, it seems likely that the increase in telephone marketing and the plethora of surveys employed by commercially orientated social researchers have had an impact on the response rate of surveys such as the AES.

Table 5.2: Australian Election Study response rates

1987 1990 1993 1996 1998 2001 2004 2007 2010

Response rate %

62.8 58.0 62.8 61.8 57.7 55.4 44.5 40.2 40.1

Sample size (1825) (2020) (3023) (1795) (1896) (2010) (1769) (1873) (1838)

Source: Australian Election Surveys (1987–2010)

NON-PROBABILITY SAMPLING Despite its many advantages, probability sampling is not always possible—or necessary— for social research. For research with victims of domestic violence, for example, how would you access a random sample? Although domestic violence crisis centres might have lists of people who have been victims of domestic violence, the lists won’t be close to being complete, and there is no feasible way of developing a complete list. People who mark their ballot invalid in elections is another example. In such circumstances, we can, and do, use non-probability sampling techniques, that is, select a sample without reference to probability theory. Th e most commonly used non-probability sampling methods are • convenience sampling • purposive sampling • snowball sampling • quota sampling • self-selected sampling.

CONVENIENCE SAMPLING Babbie (2005: 189) refers to this method as ‘reliance on available subjects’. It is also known as haphazard or opportunity sampling. Convenience sampling refers to sampling those who are easy to locate. Sampling the people you know, such as friends or relatives, or your classmates, are examples of convenience sampling. Conducting a vox pop by standing on a street corner and asking anyone who passes by if they can interview that person is another. Th e popularity of this approach is related to the ease and inexpensive nature of data collection. However, these advantages are outweighed by the likelihood that the sample will be extremely unrepresentative: we tend to know and mix with people like ourselves, while standing on a street corner will only allow you to meet people who happen to be walking past on that particular street at the time you are standing there. While convenience sample is an easy way of fi nding respondents for a research project, its lack of representativeness constrains the validity of the results.

Convenience sampling: Convenience sampling is making up your sample of respondents from people who are easy to locate, such as your friends, family, neighbours, or classmates.

110 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11005_WAL_SRM3_20170_TXT_SI.indd 110 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

PURPOSIVE SAMPLING In purposive sampling, the sample is selected based on what we know about the target population and the purpose of the study. Th e researcher exercises their judgment or knowledge of a population and the aims of the research to select a sample. If we wanted to conduct a study of why people own fi rearms but do not have access to the government register of fi rearm owners, we could make a list of gun clubs. We could then approach selected clubs and ask permission to interview their members. While a purposive sample of fi rearms owners may be more representative than a convenience sample, neither convenience nor purposive sampling would allow us to generalise our results to the wider population of gun owners. Th is is because we simply do not know how representative our sample is. Purposive sampling is, though, useful for certain applications. In pilot testing of survey questions, you may decide to select a wide range of respondents to test a new questionnaire (Babbie 2005: 189). In such a case the representativeness of the sample is not an important consideration.

SNOWBALL SAMPLING Snowball sampling makes use of the respondents we do have to fi nd other potential respondents. Snowball sampling is so called because, aft er making initial contact with participants, referred to as informants or gatekeepers, each respondent is asked to suggest other respondents, so that the sample grows like a snowball that expands in size as it rolls down a mountain. Th e technique is most appropriate when the aim is to collect data on a small number of respondents from diffi cult-to-reach groups among whom a sampling frame does not exist or the researcher does not have access to one. In this author’s research on environmental leaders in Tasmania (see ‘From method to practice’), environmental activists who held positions in prominent environmental organisations were initially contacted as gatekeepers. Th ese gatekeepers then provided the author with the names of people they identifi ed as infl uential in the Tasmanian environmental movement. Th ese leaders were subsequently interviewed and provided the names of other leaders, until considerable overlap occurred in the names that emerged and a sample was obtained (Tranter 1995). Snowball sampling will, by its nature, be unrepresentative of the population, however, as the technique is oft en used to locate members of a specifi c subgroup, sample representativeness is oft en not of primary concern.

QUOTA SAMPLING As in probability sampling, the aim in quota sampling is to attempt to draw a sample that is representative of the population we want to study. Rather than using probability theory, however, quota sampling is an attempt to draw a representative sample on the basis of information about key variables in the population. Th e variables of interest in the population, such as gender, age, or education, are defi ned, and a table or matrix is constructed based on the proportions for each of the categories of each key variable (for example, sex × age × education). Data are then collected, for example by ensuring interviewers fi ll the quotas for the proportions for each key variable (Barnett 1974). A big problem with quota sampling is that in many cases the population parameters are not known. Even if accurate population data are available, selection bias can still occur. Th is is a potential problem in face-to-face interviewing, where interviewers are required

Purposive sampling: Purposive sampling is selecting a sample in a systematic or purposive way, based on what we know about our target population and the purpose of our study.

Snowball sampling: Used to access respondents from hard-to-reach groups by asking respondents to suggest other prospective respondents to the researcher.

Quota sampling: The development of a sample that is representative of a population by drawing a sample on the bases of population estimations of factors such as gender, age, and education level.

111111CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11105_WAL_SRM3_20170_TXT_SI.indd 111 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

to approach individual dwellings to interview the occupants, then use a quota system to sample those with particular characteristics. Th e interviewer may, for example, need to interview thirty females aged between 18 and 30 to fulfi l a quota, although bias may occur in the selection of households that are diffi cult to reach or where the interviewer is concerned about their safety (Babbie 2002: 196).

SELF-SELECTED SAMPLING Self-selected sampling is where people are requested to make contact with the researchers to participate in a study. People self-select into the study. Television programs or newspapers that survey their audience by asking them to phone in, or SMS or tweet to express an opinion on a particular topic, are self-selected samples (Maisel & Persell 1996:5). For more specifi c target populations, researchers might use newspaper advertisements or articles, or advertise on websites requesting that people who fulfi l the criteria of the target population make contact with the researchers. Th e major limitation of self-selected sampling is the high potential for sample bias. Th ose people with a strong interest in the topic are most likely to self-select. In a recent qualitative study of workers’ compensation claim experiences (Welch et al. 2009), respondents self-selected into the study by answering newspaper advertisements. Th is sampling technique was successful in garnering a good number of willing respondents, but it also ensured that those interviewed tended to have strong feelings about diff erent aspects of the workers’ compensation process.

CASE STUDY 5.3: COSMOPOLITANISM: RELIGION AND KINSHIP In recent research published in the Journal of Sociology, Turner, Halse, and Sriprkash (2011) explored how the notion of cosmopolitanism would help ‘understand the experiences of young people living in culturally and linguistically diverse areas of Australia’s largest city, Sydney’. The study involved forty-four young people aged between 12 and 15, who were interviewed in a series of focus groups. The aim of the focus groups was to examine these peoples’ ‘perspectives and experiences of cultural diversity, specifically, their global connectedness through popular culture, internet, and media, and family and social networks’ (Turner et al. 2011: 5). Each focus group lasted approximately one hour and involved between three and seven students, with the groups divided on the basis of gender and school grade. Students were selected from three schools: an all boys school, an all girls school, and a coeducational school. The majority of students in each of the schools were from non-English-speaking backgrounds, the communities in which the schools were situated were ‘characterized by lower rates of household income and higher rates of unemployment than the national average’ (Turner et al. 2011: 4). Before interviewing the students, the researchers had to obtain permission from the students’ parents, the New South Wales Department of Education, and the principals of each of the three schools.

SELECTING SAMPLING SIZE FOR NON-PROBABILITY SAMPLING Deciding on the appropriate sample size for non-probability sampling is quite a diff erent task to the one involved in probability sampling. Unlike probability sampling, there are no formulae to calculate the right size, and no right answer to the ‘How many?’ question.

Self-selected sampling: Where people self-select into a survey or study by making contact with a researcher or by choosing to respond to questions, such as in television programs or newspaper polls that ask people to register their opinion or vote.

112 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11205_WAL_SRM3_20170_TXT_SI.indd 112 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

In qualitative research (see Chapter 10), sample sizes can legitimately vary from one hundred or more respondents to just one.

Again, how many depends on the project question, the budget and time limits, and interviewing resources. An honours student undertaking qualitative interviewing might interview between eight and fi ft een respondents, whereas a PhD candidate might interview forty or more. Th e right number also depends on the saturation point for your key concepts. Saturation point occurs when the interviews or other data collection method you are undertaking are no longer revealing new information on meanings or understandings. When this point is reached depends on the question and concepts involved. Achieving saturation is the main idea in deciding what size sample you need for qualitative research. More is not necessarily better, and indeed may actually be worse. Th is is because qualitative data such as interview transcripts tend to be quite long and require intensive analysis (see Chapter 14). Th e quality and rigour of your research is likely increased by analysing a smaller number of interviews thoroughly than a larger number with less attention to detail. When making this decision, it is oft en helpful to look at other research that has a similar method and topic or fi eld as a guide of what an appropriate size sample would be.

EXERCISE 5.1: SAMPLING STRATEGIES Imagine the following scenario. You work as the manager of a sports centre in a small town of 15 000 people. In addition to yourself as manager, there is another staff member, a receptionist, and a part-time cleaner. The opening hours of the centre are 9 a.m. to 5 p.m. Monday to Saturday. The sports centre has been operating for three years. The feedback you have had from the community is generally very positive, although you have also had some complaints about overflowing rubbish bins, about the attitudes of the staff, and about the general state of repairs of this somewhat ageing building. You decide to try to evaluate how satisfied your clients are with the sports centre so that you can understand how well your centre is meeting the needs of your community. You are considering whether a survey is the most appropriate research method for this topic.

YOUR RESEARCH QUESTION How satisfied are clients with the services they receive at the sports centre?

OBTAINING A SAMPLE You have decided that the best way to gather the information you need to assess on client satisfaction with the services they receive is to conduct a survey of your clients. The client database at the sports centre lists over 2000 names and contact details of people who have been to the centre at some time since it has been operating.

TASK 1 • What will the sampling unit for the survey be? What other options do you have? • What could you use as the sampling frame? Again, what other options are there? • How will you define your population? Will you include all clients, only those aged

over 18, or only those who have used the centre on a regular basis?

TASK 2 • Devise a sampling strategy to undertake the survey of your clients using

– a probability sampling technique – a non-probability sampling technique.

113113CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11305_WAL_SRM3_20170_TXT_SI.indd 113 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

• For the probability sampling strategy, ensure that you address the following questions – How many cases will you need in your sample? – How would you ensure randomness and representativeness for your sample?

• For the non-probability sampling approach, ensure that you address the following questions

– What type of non-probability sampling strategy would best suit your study? – How many cases would you include in your sample? – What steps would you take to ensure that your case selection is not biased?

VOICES IN THE FIELD: WALLY KARNILOWICZ

Wally Karnilowicz is a social psychologist at the College of Arts, Victoria University, Melbourne, Victoria. His major area of expertise is in social research methods (quantitative and qualitative), with interests and research activity in social psychology, cross-cultural psychology and cultural psychology. Wally’s early work was in the area of stigma and how it related to notions of disability (behavioural and physical). More recently, his emphasis is on cultural psychology, and in particular, examining notions of psychological ownership related to disease state. Consequently, this change in emphasis has resulted in a change in approach.

Wally’s earlier focus was on quantitative methods and its associated concerns, with the determination of the problem, appropriate sampling, identification of groups, and statistical modelling. This approach was also reflected in the work of a large number of his honours, masters and PhD students in that period. Given the nature of the problem at hand, the focus on notions of ownership and disease state required a re-examination of appropriate methodologies. Psychological ownership, as it relates to disease state— such as breast cancer, prostate cancer, and hepatitis C—is a unique and often individual experience. It tends to differ significantly in a qualitative sense between individuals, and considers cultural symbols and actions.

To get at the problem it is necessary to embed yourself within the experience, to inform yourself and soak in what is happening around you. It seeks a non-critical ‘I’ and the skill is in informing one of ‘What is?’ Its primary purpose is not to seek universal truths but rather to assume a relativist stance. As such, it is an approach different to dominant positivist notions and quantitative research in psychology. This is not to say that Wally has completely abandoned his quantitative roots. In fact, he is heavily and collaboratively involved in survey research with the Department of Human Services, Victoria, in its Neighbourhood Renewal Project. It is more to say that the question at hand drives the research process, and that each question needs to consider the appropriate epistemology and larger theory in harmony with the method.

FROM METHOD TO PRACTICE This next section details the sampling details of two recent Australian research projects, one quantitative, using probability sampling, the other qualitative, using non-probability sampling processes.

THE 2007 AUSTRALIAN SURVEY OF SOCIAL ATTITUDES The 2007 Australian Survey of Social Attitudes (AuSSA) is the third in a biennial series that studies social attitudes and behaviour of Australian citizens for the Australian and international research community. The AuSSA collects cross-sectional data on the social attitudes and behaviour of Australians, repeating a core questionnaire for each

114 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11405_WAL_SRM3_20170_TXT_SI.indd 114 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

cross-section and fielding specific modules relevant to the changing needs of the social research community. The 2007 survey includes questions on attitudes and behaviour organised into thirteen categories: describing Australia, role of government, risk, culture, retirement, private health insurance, financial future, Aboriginality, work, gender, IVF technology and sex selection, politics and society, and personal background.

Random sampling of people on the 2007 Australian Electoral Roll was used to access a sample of potential respondents. This sample was mailed a self-complete questionnaire to be returned by mail. The survey had three different subsamples—A, B, and C. The details of sample numbers and outcomes for each of these is detailed in the Table 5.3 below. As can be seen, all three subsamples had similar response rates of around 40 per cent.

Table 5.3: AuSSA response rates and sample sizes

Sent Subsample Out of scope Refusals Never returned

Completed Response rate (%)

6666 A 348 905 2631 2782 42 6666 B 374 837 2686 2769 42 6666 C 348 921 2814 2583 39

Source: Australian Survey of Social Attitudes 2007

ACTION TO MINIMISE LOSSES A five-stage mailout was used. In week 1, potential respondents received an introductory letter informing them of the survey objectives and the researchers responsible for the survey. In week 2 respondents were sent a survey package that included a cover letter, information on how to complete the questionnaire and the processes used to ensure confidentiality, a questionnaire booklet, and a reply-paid envelope. The letter also included contact information for the survey administrators and for the Australian National University Ethics Committee, and the number of a toll-free telephone hotline run by trained archive staff for any queries about completing the questionnaire or the survey in general. In the third week, a reminder–thank you postcard was mailed to potential respondents who had not yet replied, encouraging them to complete the survey and reminding them that all responses are confidential. The reminder also included the number of the toll- free telephone hotline. Approximately two weeks after the reminder postcard was sent, a reminder package containing another copy of the questionnaire and a letter were sent to those who had not responded, then a final reminder card after another week.

Source: Adapted from Phillips et al. 2008. For details of how to access the AuSSA data set, visit the AuSSA website at http://aussa.anu.edu.au

WHAT MAKES AN ENVIRONMENTAL LEADER? This study outlines a qualitative study of environmental leaders in Tasmania (Tranter 2009). As a follow-up study to earlier research on environmental leaders (Tranter 1995), semi-structured interviews were conducted in 2002 with a small sample (n = 13) of environmental leaders. The author used a form of snowball sampling to identify the most important environmental leaders and to examine changes that had occurred in the Tasmanian environmental movement since the earlier study. Snowball sampling was an appropriate technique in this case, because the aim was not to draw a probability sample, but to select participants for a study in which an appropriate sampling frame did not exist.

Several informants, or gatekeepers, were identified as potential study participants. Initial contact was made with five gatekeepers: two position holders in environmental

115115CHAPTER 5: SAMPLING

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11505_WAL_SRM3_20170_TXT_SI.indd 115 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

organisations, one Tasmanian Green politician, and two staff members who worked for Green politicians. This method of selecting an initial group of influential environmentalists as informants was based upon the assumption that those who occupied organisational positions were influential within and, to an extent, beyond the environmental movement because of their strategic location in organisations.

The five informants were interviewed and asked to identify other influential Tasmanian environmentalists. Environmentalists who were identified by three or more gatekeepers were, where possible, subsequently interviewed. Using Green politicians and employees in environmental movement organisations as informants was an attempt to minimise a problem sometimes associated with the selection of snowball samples, in which the use of only one or a few gatekeepers can introduce bias into the sampling process.

As a cross-check, the names of the initial positional leaders also emerged consistently in the interviews. The sampling strategy adopted in this research may have missed some influential leaders by, for example, oversampling high profile leaders and underrepresenting less visible but equally important leaders. Interviewing gatekeepers in environmental organisations, then snowball sampling, may also have resulted in a narrower focus on organisational-based leadership and downplayed the role of other movement leaders, although in the state of Tasmania, with its small population and limited number of influential environmentalists, these potential problems were viewed to be relatively low risk.

116 PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

BRUCE TRANTER

05_WAL_SRM3_20170_TXT_SI.indd 11605_WAL_SRM3_20170_TXT_SI.indd 116 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

CONCLUSION Choosing an appropriate method of sampling is a very important aspect of the research process. Sampling strategies tend to be used by social scientists when it is impractical to examine entire populations. A major advantage of the probability sampling approaches oft en used by quantitative researchers is that they enable inferences to be made about entire populations based upon the results of much smaller samples, because well-designed probability samples are highly likely to be representative of the larger populations from which they were drawn. Non-probability sampling strategies are useful when researchers are not able to select their samples from complete lists of population elements or, in cases where they do not need to make representative claims about a given population. Th e selection of an appropriate sampling strategy will also be determined by factors such as the researcher’s methodological approach, the methods they choose to employ and, perhaps most importantly, by the nature of the research question or questions the researcher wants to answer.

MAIN POINTS

• For most social science research we use a sample rather than the whole population as our data source.

• Sampling is used because it is not practicable to survey all members of a given group. • We can use probability samples to generalise our results back to our whole

population. • it is not necessary to survey a whole group to gain meaningful results because for

many units of analysis a complete list is not available. • Th ere are two main sampling methods: probability and non-probability sampling. • Probability sampling selects our sample in accordance with probability theory. All

elements in the population must be known, and each element must have a known chance of being selected.

• Sampling error, confi dence levels, and confi dence intervals enable us to account for the diff erences between our sample and the whole population in probability sampling.

• Non-probability sampling is used mostly in qualitative research designs. • Non-probability sampling includes a range of sampling techniques such as

convenience sampling, purposive sampling, snowball sampling, quota sampling, and self-selected sampling. Th e sampling method chosen and sample size depends on the project.

FURTHER READING Maisel, R. & Persell, C. (1996). How Sampling Works. Th ousand Oaks: Pine Forge Press.

117117

BRUCE TRANTER

CHAPTER 5: SAMPLING

05_WAL_SRM3_20170_TXT_SI.indd 11705_WAL_SRM3_20170_TXT_SI.indd 117 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed

REFERENCES Australian Election Study. (2008). Canberra: Australian National University. http://aes.

anu.edu.au. Babbie, E. (2002). Th e Basics of Social Research, 2nd edn. Melbourne: Th omson

Wadsworth. Babbie, E. (2005). Th e Basics of Social Research, 3rd edn. Belmont: Th omson Wadsworth. Barnett, V. (1974). Elements of Sampling Th eory. London: English Universities Press. Curtin, R., Presser, S. & Singer, E. (2005). ‘Changes in Telephone Survey Nonresponse

Over the Past Quarter Century’, Public Opinion Quarterly, 69 (1): 87–98. de Vaus, D. (2002). Surveys in Social Research. Sydney: Allen & Unwin. Gow, D., Bean, C. & McAllister, I. (2000). Australian Constitutional Referendum Study,

1999. Canberra: Social Science Data Archives, Australian National University. McAllister, I. & Clark, J. (2007). Trends in Australian Political Opinion: Results from the

Australian Election Study 1987–2004. Canberra: Australian National University. Maisel, R. & Persell, C. (1996). How Sampling Works. Th ousand Oaks: Pine Forge Press. Natalier, K, Walter, M., Wulff , M., Reynolds, M. & Hewitt, B. (2008). Child Support and

Housing Outcomes, AHURI Final Report 113, Melbourne: Australian Housing and Urban Research Institute.

Phillips, T., Mitchell, D., Tranter, B., Clark, J. & Reed, K. (2008). Australian Survey of Social Attitudes, 2007. Canberra: Australian Social Science Data Archive, Australian National University.

Porter, S. (2004). ‘Raising Response Rates: What Works?’, New Directions for Institutional Research Special Issue: Overcoming Survey Research Problems, 121: 5–21.

Tranter, B. (1995), ‘Leadership in the Tasmanian Environmental Movement’, Australian and New Zealand Journal of Sociology, 3: 83–93.

Tranter, B. (2009). ‘Leadership and Change in the Tasmanian Environment Movement’, Leadership Quarterly, 20 (5): 708–24.

Tranter, B. & Donoghue, J. (2010) ‘Outlaw Culture: Bushrangers in the Sydney Morning Herald’, Australian Studies, 2: 1–17.

Welch, A., Walter, M. & Ezzy, D. (2009). Long Term Benefi ts Study Workers Compensation Research Phase Two Research Report. Hobart: School of Sociology and Social Work, University of Tasmania.

118

BRUCE TRANTER

PART 1: THE FOUNDATIONS OF GOOD SOCIAL SCIENCE RESEARCH

05_WAL_SRM3_20170_TXT_SI.indd 11805_WAL_SRM3_20170_TXT_SI.indd 118 7/08/13 8:51 AM7/08/13 8:51 AM

C op

yr ig

ht ©

2 01

4. O

xf or

d U

ni ve

rs ity

P re

ss . A

ll rig

ht s

re se

rv ed