DB #3 Qualitative Data Collection

1

Zina, O. (2005).  Researching real-world problems: A guide to methods of inquiry. Sage.

CH. 5

The Quest for ‘Respondents’

 

Chapter Preview

Who Holds the Answer?

Cases: Delving into Detail

Key Informants: Working with Experts and Insiders

Samples: Selecting Elements of a Population

 

‘I might not know who holds the answer – but I do know you can’t ask just anyone, and you certainly can’t ask everyone.’

– L. B. O’Leary

 

WHO HOLDS THE ANSWER?

I gave up asking my father questions a long, long time ago. He was one of those dads who said, ‘Well, you’re going to have to find that out for yourself’ (a straight answer just once in a while would’ve been nice!). But he did give me advice about finding answers. He was the one who said ‘You can’t ask just anyone, and you certainly can’t ask everyone’. So I learned quite young that getting the right answers depends an awful lot on the effort you put in to figuring out who to ask.

Now although my dad is a self-professed expert on many, many things, he’s no expert on research. But when it comes to gathering data in a research context, his advice is spot on. Figuring out who might hold the answer to your questions, and how you will open up opportunities to gather information from those in the know, is absolutely fundamental to collecting credible data. Finding answers is reliant on finding those who hold the answers.

So let’s think about this for a minute. In most models of social science research, what we are after is answers that are held by some population. We want to know what the ‘masses’ do, think, or feel. And this could certainly be the case when researching real-world problems. Your answers may rest with a broad segment of society. But what if you think your answers are held by the ‘few’ rather than the ‘many’? What if you think your answers are held by experts and insiders, or within the experiences of a particular individual? What if you believe your answers are held within the practices of a setting such as a school or workplace?

Well, when researching real-world problems you may need to look at several strategies for finding those with the answers. Delving into the experiences of an individual or a setting (by defining an appropriate case) and working with those in the know (by selecting key informants) can be as crucial to answering a research question as is seeking broad societal representation (sampling a population).

There are, however, some challenges. Whether you decide to work with cases, key informants, or samples, finding ‘respondents’ who are appropriate, representative, open, honest, knowledgeable, have good memories, are not afraid to expose themselves and do not need to present themselves in any particular way, might be more difficult than you expect. At times you will need to be systematic. For example, you may decide to use a defined sampling strategy in order to locate a representative sample that can be generalized to a broader population. At other times you will need to be strategic. For example, you may decide to turn to where you know you have an ‘in’ and can call on pre-existing relationships. And at all times, you will need to be aware of the complexities of working with others in a bid to fulfil your own research agenda. Whether you decide to work with cases, informants, or samples, there are plenty of issues you will need to work through.

CASES: DELVING INTO DETAIL

In the social and even applied sciences, we tend to have a bias towards ‘representative samples’. Because we can make arguments about generalizability (see Chapter 4), we tend to think that this is where we need to go in order to gather credible data. But my goal in the conduct of rigorous research (and a goal I strongly advocate) is to determine the best possible means for credible data collection. And researching real-world problems often means delving into detail, digging into context and really trying to get a handle on the rich experiences of an individual, community group or organization. Answers to your research question(s) may lie in the rich history of an event, or in the day-to-day practices of a workplace. In other words, when researching real-world problems, legitimate, valid and worthwhile answers may be held by or within a particular ‘case’.

 

Case: A bounded system, or a particular instance or entity that can be defined by identifiable boundaries.

 

Case study: A method of studying elements of the social through comprehensive description and analysis of a single situation or case. For example, a detailed study of an individual, setting, group, episode, or event.

Opportunities in working with cases

Applied research into real-world problems, undertaken through the window of cases, is more common than you might realize. Practitioners working in the field or within an organization often limit their methodological design to a particular context in a bid to maximize both relevance and practicality. Case studies concentrate research efforts on a particular ‘case’, and often on one site. This can minimize travel, facilitate access and reduce costs. On a more strategic level, case studies attempt to build holistic understandings through prolonged engagement and the development of rapport and trust within a clearly defined and highly relevant context. The goal is richness and depth in understanding that goes beyond what is generally possible in, for example, large-scale survey research.

While case studies might not always be ‘representative’ or ‘generalizable’, they can add a tremendous amount to a body of knowledge. Cases can:

 

Have an intrinsic value – cases might be extremely relevant, politically ‘hot’, unique, interesting, or even misunderstood, for example, exploring a cult undergoing high media scrutiny.

Be used to debunk a theory – one case can show that what is commonly accepted might, in fact, be wrong, for example, societal assumptions related to violence in prison can be called into question through in-depth case exploration that attempts to understand the phenomenon from a prisoner’s perspective.

Bring new variables to light – exploratory case studies can often bring new understandings to the fore, for example, in-depth exploration of a particular hospital emergency room might uncover new stressors yet to be identified in the literature.

Provide supportive evidence for a theory – case studies can be used to triangulate other data collection methods or to provide support for a theory, for example, a particular organization might be explored as a lived example of a twenty-first century learning organization.

Be used collectively to form the basis of a theory – a number of cases may be used to inductively generate new understandings, for example, finding empowerment as a common theme in the ability to recover from the stress of divorce, might be the basis of new insights.

Case selection

If you have determined that your research question can be illuminated by delving into cases, you will need to turn your attention to the process of case selection. Now there are two distinct processes involved in case selection. The first is to define your case, or to set the boundaries that will give meaning and characterization to the class of ‘elements’ you wish to explore. The second involves selecting an individual case or series of cases that meet your definition and sit within your case boundaries.

Defining your case

To define a case, you need to set clear and distinctive characteristics. Perhaps the broadest and easiest distinction here is to decide if your cases will be made up of individuals, institutions, events, cultural groups, etc. Will you be looking at people, places, or things? Once this is determined, more specific criteria can be applied. For example, if your cases will be made up of individuals, you might turn to characteristics such as employment status, gender, or race to narrow the case description. If you are looking at institutions, you might look at function (factory, hospital, school, etc.), location, or size. Cultural groups (groups bound together by social traditions and common patterns of beliefs and behaviours) can be further defined by things like geography, social networks, or shared hardships. Finally, for events, defining characteristics will be the nature of the event as well as things like timeframe, geography, size etc.

As shown in Figure 5.1, possibilities are wide open. The only criteria are that your boundaries are clear, and you are able to argue the importance of case exploration within those boundaries.

FIGURE 5.1 DEFINING A CASE

Case selection

Once your class of cases has been defined, your boundaries are clear and you know precisely what it is that you are trying to delve into, you will need to select the right case (or cases) from the range of possibilities. Now depending on your goals, you may decide to delve into only one case, or you may want to compare and contrast two or more cases. You might also decide to analyse a number of cases so that you are in a position to argue representativeness.

After determining the appropriate number of cases to be explored, the selection of any particular case or cases is generally done through a strategic process. Researchers often handpick cases with a particular purpose in mind. Factors that will influence case selection include:

 

Pragmatics – there is nothing wrong with being practical. Pragmatics can involve commitments such as being commissioned or sponsored to study a particular case, or timely opportunities, where you take advantage of current events and work at being in the right place at the right time, for example, studying a community recovering from a flood event, or exploring a recent sports-related riot. Pragmatics can also involve accessibility where you take advantage of access that might normally be hard to get, for example exploring a case that has connections to your own workplace, or delving into a case involving an individual with whom you have an existing relationship based on mutual trust and respect.

Purposiveness – researchers will often select cases they hope will enable them to make particular arguments. For example, if the purpose is to argue representativeness, you may select a case considered ‘typical’. ‘Extreme’ or ‘atypical’ instances may be chosen in order to debunk a theory or highlight deviations from the norm; while wide variance in cases might be used to build new understandings and generate theory. The section on non-random sampling at the end of this chapter provides strategies that can be used in purposive case selection.

Intrinsic interest – researchers might also select a particular case because it’s interesting in its own right. It might be relevant, unique, unfamiliar, misunderstood, misrepresented, marginalized, unheard, politically hot, or the focus of current media attention. In this situation, the challenge is to argue the inherent worth and value of a particular case.

It’s worth keeping in mind that a prerequisite to all case selection should be access. It is absolutely essential that researchers who wish to delve into cases will be able to reach required people and data. When working with individuals, your ability to generate rich data will depend on building high levels of trust and rapport. In an organizational setting, you may need to gain high level access to relevant records and documents or be allowed broad access to an array of individuals associated with a case. No matter what the situation, the holistic understanding and rich detail demanded in cases studies will require you to have access to what is going on ‘inside’.

Selecting respondents within cases

I know we’ve already talked about case selection, but since researching real-world problems often involves working with cases defined as organizations, workplaces, community/cultural groups and even events, it’s worth talking about how you might select and work with individuals within a particular case. Now if your ‘case’ is an individual – you need go no further. But if your case is an organization, group or event you may need to consider further strategies for sourcing ‘respondents’. The next two sections of this chapter talk about selecting key informants and sampling populations, and both of these strategies can sit under a case study approach. As you read over the rest of this chapter, keep in mind that key informants can be central to both gaining access and insider knowledge; while sampling within your case can be an effective strategy for ensuring broad representation.

KEY INFORMANTS: WORKING WITH EXPERTS AND INSIDERS

Working with key informants means you are attempting to gather some insider or expert knowledge that goes beyond the private experiences, beliefs and knowledge base of the individual you are talking to. Your goal is to find out what this individual believes ‘others’ think, or how ‘others’ behave, or what this individual thinks the realities of a particular situation might be. Working with key informants means you believe the answers to your research questions lie with select individuals who have specialized knowledge and know what’s really going on.

But who really knows what’s going on? Well, in my workplace, it’s certainly not me; I try very hard to stay out of the loop. In fact, if your case study involved my little academic world, I’d recommend you try talking to my Head of School, or the Dean – but then again they may end up giving you the party line; when you work at that level, you are sometimes forced to call on rhetoric. Wait! I have an idea; you should try talking to Joycee from administration. She’s an institution unto herself, and if anyone knows what’s going on – it’s her. While she doesn’t have official ‘power’ she does have knowledge – which, of course, is a form of power in its own right.

Now this tends to be the case in almost any institution, organization, or community group you might want to explore. There tend to be people ‘in the know’. Whether through a position of power or some less official means, some people have a knack for knowing what’s really going on. So when you’re researching real-world problems, set in real-world contexts, it is not unusual for ‘experts’ or ‘insiders’ to be precisely the right people to help you answer your research questions.

 

Key informants: Individuals whose role or experiences result in them having relevant information or knowledge they are willing to share with a researcher.

Opportunities in working with key informants

There is nothing like having an inside track, or having an expert at your fingertips. In fact, the insights you can gather from one key informant can be instrumental not only to the data you collect, but how you process that data, and how you might make sense of your own experiences as well as the experiences of others. Key informants can give you access to a world you might have otherwise tried to understand while being locked on the outside.

Now this doesn’t mean that all your data should come from key informants. Informants may end up being just one resource you call on in a bid to build understandings. In fact, there are a number of distinct ways key informants can be used in the research process. Key informants can:

 

Be instrumental to preliminary phases of an investigation – key informants can be called upon by researchers to build their own contextual knowledge. They might also be used to help generate relevant interview questions; or be called on to aid in the construction or review of a survey instrument.

Be used to triangulate or confirm the accuracy of gathered/generated data – data from interviews with key informants can be used to confirm the authenticity of other data sources such as data gathered by survey, observation or document review. Key informants might also be called upon in a less formal way to overview data to confirm credibility, or to explore researcher interpretations for misunderstandings, misinterpretations, or unrecognized bias.

Be used to generate primary data – in-depth interviews with key informants can also be a primary source of qualitative data in its own right.

Informant selection

I think the most important consideration in the selection of key informants is your ability to gather open and honest information from them. Key informants must be accessible and willing to share information. If they have the knowledge you are after, but are not willing to share it, they will not be of any use to your study.

Informant types

It is also important to recognize that key informants do not need to be foremost experts. In fact, there are a number of characteristics that might make someone a useful informant. Depending on your research question and context, any or all of the following might have something to offer:

Experts – the well respected, who sit at the top of their field

Insiders – those who sit on the inside of an organization, culture, or community who are willing to share the realities of that environment

The highly experienced – perhaps not deemed an expert – but someone with a rich depth of experience related to what you are exploring

A leader – this might be at a formal or informal level

The observant – individuals in an organization or community who have a reputation for knowing who’s who and what’s what

The gossip – similar to the observant but enjoy passing on their observations (and sometimes rumours) – it will pay to make sure your information here is accurate

Those with secondary experience – for example, if exploring the problem of youth suicide, in addition to youth, you might look to certain counsellors, teachers, or parents to provide relevant insights

Stool pigeons – individuals who want to be classic police type ‘informants’ – you’ll need to be wary of both overt and hidden agendas!

The ex – this might include someone who is disenfranchised, alienated, recovered, converted, retrenched, fired, or retired

Working through appropriate selection

There are four distinct challenges you need to face before you can work with key informants. The first is to identify the type of informant you are after, and then identify individuals who have the characteristics associated with that type. The advice here is to ask around. You can also try a snowball technique in which you generate a list of respondents through a referral process (see Figure 5.3). One person in the know is likely to lead you to a host of others.

The second challenge is to confirm the status of those identified. Do they really have the expertise, experiences or insider knowledge that will inform your study in a credible way? The advice here is to seek confirmation by looking for things like a long record of involvement, direct personal experiences and detailed comments from potential informants that show internal consistency. You are after more than just broad generalizations.

The third challenge is to look for and recognize informant subjectivities. Remember, all respondents will have a particular worldview and some will have a real agenda operating. Some may want to be listened to, some may have an axe to grind, some may like the sound of their own voice, some think they know a lot more than they do (Gee, it sounds like I’m describing my family at Thanksgiving dinner…), and some think that their particular take on an experience is how the world should or does respond to the same experience. Bear in mind that you need to develop and build a strong relationship with your key informants. Not only do they need to feel comfortable so they can open up to you, you also need to end up in a position that allows you to know how to best treat the data they provide.

TABLE 5.1 OPPORTUNITIES AND ETHICAL DILEMMAS IN WORKING WITH KEY INFORMANTS

Opportunities Ethical dilemmas

Building relationships of trust to enhance flow of informationtoo emotionally invested

Having informants become too emotionally invested

Developing friendships that are one-way

Making promises you cannot or do not intend to keep

Gaining the ability to avoid or skirt around official channels and protocols

Putting informants in an unethical position

Acting unethically in regard to the organization you might be exploring

Being able to get your hands on confidential information

Asking for, expecting, or accepting illegal/unethical conduct from your informants

Acting unethically, and possibly illegally, in regard to the organization you are exploring

Being able to really dig into the emotional aspects of a topic

Asking your informant to make a large emotional investment

Having your informant relate private and personal details of others

Asking your informants to relive their own unpleasant memories

The final challenge is related to ethics. Now if you look at the list of informant types – and think about the motivations I outline above, it should be pretty obvious that ethics and integrity need to come into play when selecting and working with key informants. In addition to the challenge of managing bias (both yours and theirs), you will need to think about the power position you are in as a researcher. You have to remember that key informants can be put, and can put themselves, in very vulnerable positions. It is your responsibility to respect their needs at all times.

Table 5.1 highlights some of the ethical issues you will need to negotiate when selecting and working with key informants.

SAMPLES: SELECTING ELEMENTS OF A POPULATION

As stated at the beginning of this chapter, in most models of social science research, when we are looking for answers, what we are after is answers that are held by a population. We want to know what the ‘masses’ do, think, or feel. The idea is to get a snapshot or picture of what people really do and what they really think. Rather than delving into cases, or attempting to gather expert or insider knowledge, the goal here is capturing the reality of a ‘population’.

 

Population: The total membership of a defined class of people, objects, or events.

Now the ultimate in population research is to be able to ask everyone – in other words, to be able to gather data from every element within a population. But with the exception of in-depth research into very small, defined and accessible populations, or the conduct of a funded ‘census’, which is basically a survey of every element within a population, the goal of asking everyone just isn’t practical. Your study will probably involve a population that you cannot reach in its entirety; it will either be too large, or it will have elements that you simply cannot access.

Yet our inability to access every element of a population does little to suppress our desire to understand and represent it. This means we will have to sample. The idea here is to speak to the ‘few’ in order to capture the thoughts, knowledge, attitudes, feeling, and/or beliefs of the ‘many’.

 

Sampling: The process of selecting elements of a population for inclusion in a research study. Many samples attempt to be representative, that is, the sample distribution and characteristics allow findings to be generalized back to the population.

Opportunities in working with a ‘sample’

So why would you choose to work with a sample? Well samples can make the research process manageable. They allow you to explore groups of people, organizations and events that you simply could not access in their totality. Whether your population is too large, too widely dispersed, too difficult to locate, or too hard to access, sampling can provide you with a window for exploring an unwieldy population.

Sampling can also be used to represent a population with some level of ‘confidence’. Certain sampling strategies actually allow you to calculate the statistical probability that your findings are representative of a greater population. Sampling is therefore key to making research affordable, and if done with integrity, also credible.

Sample selection

Sampling is therefore a process that is always strategic, sometimes mathematical, and generally quite tricky. The goal is to select a sample that is: (1) broad enough to allow you to speak about a parent population; (2) large enough to allow you to conduct desired analysis; and (3) small enough to be manageable.

Meeting these goals will require you to think through a number of sampling issues, including the need to define your population; determine appropriate sample size; and select a suitable sampling strategy.

Defining your population

It is absolutely crucial you have a very clear and well-defined population in mind before you do any sampling. You don’t want to fall into the trap of asking just anyone, and figuring out who you were trying to target after the fact. This means you’ll need to go into your study knowing the total class of ‘elements’ you want to be able to speak about.

For example, say you want to present findings that will be representative of 13–18-year olds in the United States. Your population here is made up of individuals (the most common type of population in social/applied science research) with a particular set of defining characteristics, in this case both age (13–18) and geography (in the United States). Keep in mind that in a study of individuals you might have used other defining characteristics, such as, gender, marital status, race etc.

Now just like with cases, populations are not always made up of individuals. Depending on the nature of the research question, the ‘elements’ of your population might be households, workplaces, or even events. For example, your population might be hospital emergency rooms across Europe. In this case, it is a particular type of organizational setting that makes up the population. Defining characteristics include both geography (across Europe) and type of setting (hospital emergency room). Other possibilities for defining ‘organizations’ might include number of employees, years of operation, public or private etc. An example of a population made up of events might be professional soccer games held in Sydney in 2005. Defining characteristics here are type of activity (professional soccer matches), geography (Sydney) and time period (2005).

Determining sample size

Once your population in clearly defined, you will need to figure out how many elements of that population should be in your sample. I tend to get asked, ‘How many do I need?’ And it’s a tough question because the answer really is ‘It depends’. There are no hard and fast rules. Sample size is highly dependent on the shape and form of the data you wish to collect, and the goals of your analysis. For example, the in-depth nature of collecting qualitative data will generally limit your sample size; you simply can’t collect that type of data from thousands. But fortunately you don’t have to. Qualitative data analysis strategies aren’t generally dependent on large numbers. On the other hand, the statistical analysis of quantitative data will require a minimum number. Statistics and the ability to work with probabilities rest on adequate and appropriate sample size.

The following guidelines might help you work through the intricacies of determining appropriate sample size:

WORKING WITH QUALITATIVE DATA While not necessarily a prerequisite for qualitative data collection, the goal of generating a representative sample can pose a real dilemma for researchers, since the nature of collecting qualitative data generally limits sample size. There are two strategies you can call on here. The first is to ‘handpick’ a small sample using criteria chosen to assure representativeness. For example, selecting your sample based on a clearly defined population profile, for example individuals with the average age, income and education of the population you are studying. Rather than relying on numbers, you will need to logically argue that your sample captures all the various elements/characteristics of your population.

The second strategy is to select a sample large enough to allow for minimal statistical analysis (see next section). This will give you the option of quantitatively summarizing some of your findings in order to make more mathematical generalizations about your population. While we tend to dichotomize qualitative and quantitative research, the best studies are not afraid to cross this constructed boundary.

WORKING WITH QUANTITATIVE DATA When working with quantified data, the basic rule of thumb is to attempt to get as large a sample as possible. The larger the sample, the more likely it is to be representative, hence generalizable. But most researchers working in real-world settings struggle at the other end, and need to know minimum requirements. Now, as highlighted in Table 5.2 below, the most basic statistical analysis requires a minimum of about 30 respondents; anything smaller and it can be difficult to show statistical significance – particularly if findings are widely distributed (have a large standard deviation – see Chapter 11).

As you move to more sophisticated analysis, the use of any ‘subdivisions’ within your sample will require approximately 25 cases in each category. For example, you may have a sample of 100, but if you want to compare men aged 18–35 with women of the same age, you will need to be sure you have at least 25 in each of these categories. Similarly, if you want to conduct multivariate analysis (the analysis of simultaneous relationships among several variables) you will need at least 10 cases for each variable you wish to explore.

Another way to approach sample size is to use the following formula:

Personally, I don’t believe in working formula unless I have to, so I tend to use a ‘sample size calculator’ where the only things you need to know are: the population size; the confidence interval – what range you will accept above and below the mean, say ± 5%; and the confidence level – how sure you want to be that your findings are more than coincidental, generally 95% or 99% (see Chapter 11).

Table 5.2 was generated using a calculator from www.surveysystem.com/sscalc.htm, and gives you some idea of the required sample size for more commonly used confidence levels. Note that as the population increases, shifts in sample size do not increase as dramatically. What does require a significantly increased sample size, however, is a desire for higher levels of confidence.

ADDITIONAL ISSUES There are two other issues you will need to keep in mind when determining sample size. The first involves the challenge of working with both qualitative and quantitative data. Unless you have unlimited time and money, there will usually be some trade-off between the collection of rich in-depth qualitative data and the level of statistical analysis that might be possible. The second issue involves remembering practicalities. While a very large sample might seem ideal, it can be an expensive option. Realistic planning from the start can protect you from setting unachievable goals.

Employing a sampling strategy

Once you have defined your population and determined appropriate sample size, you will need to adopt a strategy for gathering your sample. There are two main ways to go about this. The first is to use a strategy for random selection. The second is to use a strategy that aims to strategically select your sample in a non-random fashion. The best method will depend on a number of factors, including the nature of your question, the make-up of your population, the type of data you wish to collect and your intended modes of analysis.

RANDOM SAMPLES Random samples rely on random selection, or the process by which each element in a population has an equal chance of selection, for example, names drawn out of a hat, or computer-generated random numbers. The process attempts to control for researcher bias and allows for statistical estimations of representativeness. The process, however, demands that (1) all elements of a population are known and accessible and that (2) all elements are equally likely to agree to be part of a sample.

If this is not the case, two types of error can occur. The first is coverage error, or the situation in which the list you draw your sample from is incomplete. For example, while every name in the hat has an equal chance of being drawn, all the names need to be in the hat first. Surveys reliant on e-mail addresses can have this problem. Unless everyone in a particular population has email, coverage is likely to be incomplete.

The second type of error is non-response bias, or the situation in which those who accept an invitation to be in a sample are somehow intrinsically different from those who decline. For example, in conducting a customer satisfaction survey, it might be that those who agree to participate have an axe to grind. If this is the case, your eventuating sample will not be representative of your population.

It is therefore important to consider whether the lists you draw your sample from (the sampling frame) are complete. If this is not possible, you’ll need to think about how you can give a voice to any sector of the population that might miss inclusion. You will also need to explore issues of non-response, and come up with strategies that will ensure broad representation. Figure 5.2 highlights the random sampling strategies you are likely to call on in conducting research into real-world problems.

NON-RANDOM SAMPLES Non-random samples are just that – they are samples that are not drawn in a random fashion and are sometimes called ‘purposive’ or ‘theoretical’ samples. In order to generate a sample that is meaningful, and possibly representative, non-random sampling demands conscientious decision making. Now non-random samples are considered by some to be inferior because they cannot be statistically assessed for representativeness – they are sometimes seen as second best or last resort. There is a growing belief, however, that there is no longer a need to ‘apologize’ for these types of samples. Researchers using non-random samples can generate meaningful samples, and even credibly represent populations, if (1) selection is done with the goal of representativeness in mind and (2) strategies are used to ensure that samples match population characteristics.

Certainly, non-random selection offers researchers flexibility when working with populations that are hard to define and/or access (for example homeless women or sports people who have used steroids). There is, however, an added burden of responsibility in ensuring that eventuating samples are not biased. Specifically, researchers who are after representativeness need to be aware of unwitting bias and erroneous assumptions.

FIGURE 5.2 RANDOM SAMPLING

Unwitting bias refers to the tendency to unwittingly act in ways that confirm what you might already suspect; something that can be quite easy to do when you are handpicking your sample. For example, you may want to conduct a focus group that can help evaluate an initiative you have started in your workplace; unless you make a conscious decision to do otherwise, it’s just too easy to stack the deck in your favour.

Erroneous assumptions is somewhat different and refers to sample selection that is premised on incorrect assumptions, for example, assuming that you can go to the local McDonald’s to select a representative sample of teenagers in a small town without realizing that a certain sector of that town’s teenage population wouldn’t be caught dead there.

You might also make erroneous assumptions about the characteristics of ‘elements’ within your sample. Say, for example, you want to study teenage ‘angst’ and you select what you believe are extreme cases of angst. If your assumptions are incorrect and what you see as extreme is actually quite average, the generalizations you make will not be valid.

Now, in order to control for such biases, it is well worth brainstorming your own ideas, assumptions and expectations as they relate to both your research questions and your sample. This will put you in a strong position to work towards the development of an appropriate sampling strategy. Figure 5.3 highlights the non-random sampling strategies that you are likely to call on when researching real-world problems. Keep in mind that while they can be used to build a representative sample, these strategies can also be called upon in studies that do not rely on representativeness, for example, when the goal is to build knowledge by working with cases and/or key informants.