NO PLAGIARISM DUE THURSDAY January 16, 2020. RESOURCES ARE ATTACHED TO ASSIST WITH ASSIGNMENT

05CH_Applied_Harper.pdf

Home >Psychology homework help >NO PLAGIARISM DUE THURSDAY January 16, 2020. RESOURCES ARE ATTACHED TO ASSIST WITH ASSIGNMENT

Learning Objectives

After reading this chapter, you should be able to

ሁ Understand the need to evaluate published research. ሁ Evaluate the quality of research questions and hypotheses. ሁ Identify variables used in research. ሁ Understand basic types of qualitative and quantitative research designs. ሁ Interpret visualized data. ሁ Understand basic types of statistical analyses. ሁ Understand the basics of inferential statistics. ሁ Demonstrate ethical research integrity. ሁ Explain career applications involving research and statistics.

Research Methods and Statistics Jeral Kirwan—Ashford University 5

gevende/E+/Getty Images

Section 5.1Introduction: Why Is Research Important?

Social scientists often need to collect and analyze data to test a research question or hypoth- esis. However, professionals in many other fields also often need to solve real-world problems. For instance, Stephon is a program evaluator who works for a non-profit organization. One day, he is asked to evaluate several new programs that are competing for a grant offered to small businesses. Stephon’s supervisor asks him to set up an evaluation model to assess these proposals and to find the top three candidates for that grant. It would be easy for Stephon to just read them and pick the three that he likes best or finds most interesting, but he knows that would not be the most fair and equitable approach. Instead, he follows some basic research principles that he learned in his research methods course and designs an evaluation that summarizes, assesses, and reports on the applications.

Stephon starts by determining the goals of the grant and using those to create a rubric with which to evaluate the proposals. Then he uses that rubric to objectively assign a score to each of the proposals. With that data, he creates a chart that shows how each proposal compares to the others based on the established criteria. He presents his results to his supervisor with his recommendations for the winners of the grant. Through this process, Stephon has used the scientific method to collect data and make informed decisions based on that data. This shows that research methods and statistics are valuable areas of expertise beyond the laboratory.

After considering this scenario, review the following questions:

ሁ What ethical considerations might have encouraged Stephon and his supervisor to use an evaluation model rather than simply selecting their personal preferences?

ሁ How, in this scenario, was the creation of a rubric useful? ሁ Are there ways that you, in your everyday life, utilize the scientific method to make

decisions? If so, when and how?

5.1 Introduction: Why Is Research Important? Being able to understand and evaluate published research is essential for anyone in a psychology field. For instance, an industrial/organizational psychologist must keep up with current trends in orga- nizational development by reading pub- lished articles in that psychologist’s par- ticular area. Even if you never conduct a research study yourself, it is vital that you become an active consumer of research related to your area of interest and prac- tice. As an informed consumer, you are able to be evaluate whether research has been done well and avoid falling victim to unsubstantiated or exaggerated claims or poorly designed research.

People in all areas of psychology use data in different ways. A clinical psychologist may read articles about new methods for treating depression and interpret the data to help inform how patients are treated locally. A forensic psychologist might use data on the current trends of

Shironosov/iStock/Thinkstock ሁ It is critical for those working in psychology to

stay abreast of current research in their field.

Section 5.2Research Questions and Hypotheses

recidivism rates to better understand certain criminal behaviors. A program evaluator uses data to determine if a program is working as it should. A research psychologist will use data to try and prove a hypothesis. Almost every career path, including most related to psychology, uses data to some degree.

In research, the type of design that you choose influences the type of conclusion that can be drawn from the research. It is important to evaluate the needs and circumstances of your research project before selecting a design. This chapter begins with a basic overview of research questions before discussing the major types of research designs, and then the rest of the chapter is focused on how data is visually represented and evaluated. Throughout, a spotlight is kept on ways in which this information can be useful to you in your professional and daily life.

5.2 Research Questions and Hypotheses Developing a good research question and hypothesis is essential to research. They are used to determine the methodological approach and help to focus the research design. When devel- oping a research question or hypothesis, or evaluating a published study, consider how well the question addresses the problem, how testable it is, and if it contributes to the body of knowledge regarding the issue.

What Is a Research Question? Students often confuse general philosoph- ical questions with research questions. Questions such as “What are the symp- toms of ADHD?”, “What is the difference between psychology and psychiatry?”, and “How does one treat depression?” are good questions but not testable with research.

A research question must be some- thing clear and testable such as “What is the relationship between learning styles and memory recall?”, “How are men and women different with regard to perfor- mance on math exams?”, and “Does drink- ing coffee improve student performance on final exams?” These questions are clear and show direct connections to the variables. Having clear, specific, and testable questions is essential to developing a research design to assess variables.

Typically, research questions that are based on how much, to what extent, when, or who are best tested with a quantitative approach. Qualitative approaches are better for measuring why questions and things that are more exploratory in nature.

A research question is based on the general topic, but narrows down the scope of what the study will seek to accomplish. You do not want the question to be too general or too narrow. Some examples might be:

YakobchukOlena/iStock/Thinkstock ሁ Good research questions must be testable.

Section 5.3Types of Variables in Psychology Research

1. What is the rate of depression in Colorado? 2. What are the characteristics of a good human resource manager? 3. Does education play a role in juvenile defenders’ successful rehabilitation?

When developing or evaluating a research question, consider if it is clear and testable. A ques- tion that is too vague, or that has too many factors, cannot be easily tested. A good research question is the guide to the rest of the research design.

What Is a Hypothesis? A hypothesis is a testable statement that proposes relationships or differences between two or more variables. It is your prediction of what will happen in your research study. Develop- ing a testable hypothesis is the first step in the scientific method approach to research; a hypothesis must be developed before data is collected. In psychological research, hypotheses commonly focus on how some aspect of the environment affects a particular behavior. Below are some basic hypothesis examples:

1. Students who eat breakfast score higher on statistics exams than students who do not eat breakfast.

2. Ninth grade girls have higher memory recall on algebra exams than ninth grade boys.

3. Medicine A will lower blood pressure more than Medicine B. 4. The personality factor conscientiousness is significantly related to self-direction

in learning. 5. The experimental group that receives a special tutorial will demonstrate better recall

than the control group that receives no intervention.

Here is an example of a more complicated hypothesis: “Conscientiousness will be uniquely, positively related to learner self-direction after controlling for the other Big Five and narrow traits” (Kirwan, Lounsbury, & Gibson, 2014, p. 5).

How do you develop a hypothesis? Before developing a hypothesis statement, you have to do some background research on the topic. Look at published research and see what other people have done and what hasn’t been done. Once you complete your literature review, con- sider what variables interest you and how you might test a relationship between them. After you have written a few hypotheses statements, consider which one could be readily tested. Think about how you might go about confirming or disproving that hypothesis. Developing a good hypothesis statement is an iterative process, and it takes time. Once you have your hypothesis, the next step is to define your variables.

5.3 Types of Variables in Psychology Research Imagine that you are a psychologist, and you are interested in conducting a study to deter- mine the effects of eating breakfast on memory in young children. You assume, based on pre- vious studies, that children who eat a healthy breakfast will perform better on exams than children who do not eat breakfast at all (this is your hypothesis). How do you begin your research study and test your hypothesis?

One of the first things you do as a researcher is define your variables (also commonly described as factors). These factors are chosen by the researcher before collecting data and

Section 5.3Types of Variables in Psychology Research

they are characteristics that vary from individual to individual. For instance, with regards to breakfast, each person may eat at different times of the morning, eat various types and amounts of food, and perform differently on their exams (even from one day to the next). There are four main scales for the measurement of variables, and each variable will fit into one of them depending its characteristics.

• Nominal – The variable is divided into categories that have no relative order. For instance, a questionnaire might ask participants to choose their gender identifica- tion from a list.

• Ordinal – The variable is divided with categories that have a relative order. An example might be high school levels: freshman, sophomore, junior, or senior.

• Interval – The scoring of the variable has equal distances between points on a con- tinuum. An example would be time of day. The distance between 8 a.m. and 9 a.m. is the same as the distance between 12 p.m. and 1 p.m. For our example, we might measure the time between eating breakfast and taking an exam.

• Ratio – The variable represents the relationship between two numbers and is char- acterized by an absolute zero (meaning zero is the lowest possible value). For our example, grade point average (GPA) could be used.

It is important to understand the different types of scales of variables as that is directly con- nected to the method of analysis. For instance, a study that only contains nominal variables would be analyzed with non-parametric analyses such as chi-square (covered later in the chapter).

Independent and Dependent Variables Quantitative research designs utilize independent and dependent variables to try and explain a phenomenon. An independent variable is a variable that the researcher controls to see what effect it has on the dependent variable. An example would be if a researcher wants to see if students perform differently on a math exam depending on the tempera- ture in the room. The researcher controls who will be taking the tests (the students can be chosen such that they are all the same age, or gender, etc.) and the temper- ature in the room to see how tests scores may be affected (the measurable out- come). So, the differences in the indepen- dent variables (students and tempera- ture) are controlled by the experimenter, and the dependent variable (test scores) only changes in response to those inde- pendent variables.

Extraneous and Confounding Variables When investigating factors in social sci- ence research, there are usually many possible variables that can be connected

Wavebreakmedia Ltd/ Wavebreak Media/Thinkstock ሁ Researchers must be aware of extraneous

variables that may impact the results of their research.

Section 5.4Types of Research Designs

to the independent and dependent variables, and that can affect their relationship. For instance, if we want to find out why people perform a certain way on a math test, some pos- sible variables that affect performance are motivation, study time, IQ, personality, time of day, and even the temperature in the room where the testing takes place. Variables that are not a part of the research design are considered to be confounding, or extraneous, variables. These are the unconsidered or undesirable variables that influence the relationship between the variables the researcher is investigating. For example, take a psychologist who wants to examine the effect of showing romantic movies to a group of men and women on how they display affection the following week. One confounding variable could be that the women had higher levels of romantic idealism than the men did before the study started. A comparison of the two groups may not show an accurate picture of the outcome of the treatment (show- ing the movie). In this instance, the extraneous variable is important because it was present in one group and not in the other. Possible extraneous variables should be considered before conducting such a study. While researchers cannot consider all possible variables related to the outcome they are interested in, they should at least be aware that many possible variables may be affecting it. A skilled researcher will consider what they can, and note limitations when they cannot. Next, we will look at research designs that researchers use to focus in on the variables of interest.

5.4 Types of Research Designs As explained at the beginning of this chapter, and as the frequently cited research expert Cre- swell (2013) explains, research scientists and people working with data in all fields must understand that the research question and hypothesis guides the research design. In the next section, we will examine several different research designs and strategies for both quantita- tive and qualitative approaches.

Quantitative Research Methods In the natural and social sciences, quantitative research is an empirical investigation of phe- nomena based on mathematical models and theories. Hypotheses are developed and mea- sured based on statistical questions that address aspects of an observed phenomenon such as what, where, when, how much, and to what extent? Before data can be collected and analyzed, a representative sample of a population must be collected based on a variety of sampling strategies.

Sampling Strategies Psychology researchers seek to examine traits, behaviors, and other human characteristics (as do marketing and advertising analysts). However, it can be difficult to collect data if the populations being investigated are large or difficult to reach. Sampling strategies are used to find a subset of the population that accurately represents the population. Random sampling is when everyone in the population has the same chance of being selected. In experimental research, the sample must be randomly selected. Stratified sampling occurs when a popula- tion is divided into subgroups (the strata) and then each subgroup is randomly sampled from. Systematic sampling involves a set protocol for selecting participants, such as every 20th person to enter a clinic, for example. Cluster sampling involves dividing the population into clusters and then randomly choosing a certain number of people in each cluster. Each of these fall under the category of probability sampling because they use randomization.

Section 5.4Types of Research Designs

Non-probability sampling does not require the use of randomization and includes several strategies, such as the fol- lowing. A convenience sample is one in which the participants are chosen based on availability. An example would be giv- ing undergraduate students extra credit to participate in a local study. Purposive sampling involves selecting participants based on some characteristic they may have, such as individuals diagnosed with a rare disease. Proportional sampling divides participants into subgroups and samples each subgroup randomly; it might be used when the researcher needs to have a particular number of people with a certain characteristic, such as having equal numbers of participants that identify as women or men. The strategies go beyond these few examples and greatly depend on the nature of the research design.

Experimental Research Designs Experimental designs are scientific studies that rely on random assignment of subjects and strict control over the research environment and variables of interest. They are the only approaches that are considered rigorous enough to address cause-and-effect relationships between variables. While there are many experiments in psychology research, they are less common than other approaches due to the nature of most studies in the social sciences and the cost and limitations of the laboratory settings that experiments require.

Here is an abstract from a study that used an experimental design:

We present a novel experimental design to measure honesty and lying. Par- ticipants receive a die which they roll privately. Since their payoff depends on the reported roll of the die, the subjects have an incentive to be dishon- est and report higher numbers to get a higher payoff. This design has three advantages. First, cheating cannot be detected on the individual level, which reduces potential demand effects. Second, the method is very easy to imple- ment. Third, the underlying true distribution of the outcome under full hon- esty is known, and hence it is possible to test different theoretical predictions. We find that about 20% of inexperienced subjects lie to the fullest extent possible while 39% of subjects are fully honest. In addition, a high share of subjects consists of partial liars; these subjects lie, but do not report the pay- off-maximizing draw. We discuss different motives that explain the observed behavioral pattern. (Fischbacker & Föllmi-Heusi, 2013, p. 525)

Quasi-Experimental Designs Quasi-experimental designs involve studies that are close to experimental in nature but do not have participants that have been randomly selected.

bogdandreava/iStock/Thinkstock ሁ Types of random sampling include stratified,

systematic, and cluster sampling.

Section 5.4Types of Research Designs

Here is an abstract from a study with a quasi-experimental design:

Many job redesign interventions are based on a multiple mediator–multiple outcome model in which the job redesign intervention indirectly influences a broad range of employee outcomes by changing multiple job characteristics. As this model remains untested, the aim of this study is to test a multiple mediator –multiple outcome model of job redesign. Multilevel analysis of data from a quasi-experimental job redesign intervention in a call center confirmed the hypothesized model and showed that the job redesign intervention affected a broad range of employee outcomes (i.e., employee well-being, psychological contract fulfillment, and supervisor-rated job performance) through changes in 2 job characteristics (i.e., job control and feedback). The results provide fur- ther evidence for the efficacy and mechanisms of job redesign interventions. (Holman & Axtell, 2016, p. 284)

Observational Research Observational research (often called field research) is a type of non-experimental design in which the researcher observes and records behaviors in the context of the study’s param- eters. For instance, a researcher may want to observe aggressive behaviors in children ages 2–4, so he gets permission to sit at a preschool playground and watch the children interact.

Here is an abstract from an observational research study:

Background Regular physical activity is associated with a range of physical and psychologi- cal health benefits. In North America the majority of adolescents are insuffi- ciently active.

Purpose The purpose of this study was to examine the prospective relationship between adolescents’ perceptions of transformational leadership displayed by their school physical education teachers and their own physical activity behaviors, both with respect to within-class physical activity (WCPA) and also leisure time physical activity (LTPA).

Method The study used a prospective observational design. Using multilevel structural equation modeling (MSEM), we examined the extent to which adolescents’ affective attitudes mediated the effects of teachers’ behaviors on adolescents’ physical activity responses. Two thousand nine hundred and forty-eight ado- lescents (Mage = 14.33, SD = 1.00, Nfemale = 1,641, 55.7 %) from 133 Grade 8–10 classes in British Columbia (Canada) provided ratings of their physical educa- tion teachers’ behaviors midway through the school year. Two months later, students completed measures of affective attitudes, WCPA, and LTPA.

Results The results indicated that adolescents’ perceptions of transformational teach- ing explained significant variance in both WCPA and LTPA, and these effects

Section 5.4Types of Research Designs

were fully mediated by adolescents’ affective attitudes (total indirect effect: b = 0.581, p < 0.001).

Conclusion The findings suggest that transformational leadership behaviors displayed by physical education teachers may be an important source of adolescent enjoyment of physical education as well as health-enhancing physical activ- ity involvement within school and outside of school. (Beauchamp et al., 2014, p. 537)

Qualitative Research Methods Qualitative research in the social sciences is a method of inquiry where the goal is to gather information to achieve a more in- depth understanding of human behavior. Popular approaches include interviews, case studies, ethnographies, narratives, and empirical data collection based on grounded theory. Qualitative research uses non-probability sampling strategies. Samples are typically convenience or pur- posive samples as the goal is to look at a particular person or group and does not follow the quantitative goal of generaliz- ing to a larger population. Some common designs and examples can be seen below.

Interpretive Study An interpretive study is used when the researcher wants to see how individuals make mean- ing of a particular situation or phenomenon. Data is collected by conducting interviews, observations, or reviewing documents (such as patient records or personal diaries). Analysis of interpretive data focuses on themes or common patterns to form a descriptive exposition connecting the data to the published literature. For example, if a researcher wanted to study how people in poverty make decisions about what food to buy, that researcher might collect information on a sample group’s food purchases over a certain period of time and also con- duct interviews with the sample population to understand their thought processes and lived experiences.

Grounded Theory Grounded theory is a systematic qualitative method used in social science research that starts with general questions to develop a theory based on emerging patterns in collected data. The early phase of this type of research is intended to be somewhat vague and leads into the development and verification of a theory based on the data collected.

Here is an abstract of a study that used grounded theory:

KatarzynaBialasiewicz/iStock/Getty Images ሁ Interpretive studies involve interviewing and

collecting qualitative data.

Section 5.4Types of Research Designs

Currently, a relatively small number of studies have employed qualitative methods to rigorously examine the experiences of health care professionals enrolled in mindfulness-based stress reduction (MBSR). This study devel- oped a working model of how participants may experience change during an adapted MBSR program for health care professionals. The model derived from the data demonstrated that participants echoed themes similar to those described by clinical populations engaged in MBSR, such as the salience of the group experience and support, discovery of acceptance as well as the realization that some degree of frustration and/or distress is part of learning and establishing a mindfulness practice. Unique themes highlighted included becoming aware of perfectionism, the automaticity of “other focus” and the “helping or fixing mode”. Findings illustrated the nuanced change processes undertaken by participants and the implications such change held across pro- fessional and personal domains. (Irving et al., 2014, p. 60)

Phenomenology Phenomenology is a qualitative research approach looking at human consciousness and self- awareness of experiences. A phenomenological approach emphasizes the subjective experi- ences people have and how they interpret the world in particular situations. The main goal of this approach is to better understand how others view their surroundings. For example, a researcher might use a phenomenological method to study the experiences of children at Disneyland by interviewing children and their parents and then analyzing those interviews to find the essential themes of their experiences.

Ethnography Ethnographic research is a broad approach in field research that focuses on participant obser- vation and involves the researcher being immersed in the culture being studied as one of the participants. Ethnography is a very broad area with many different methods associated with it. The benefits include learning about a particular culture or group from the inside, rather than being an impartial observer.

Here is an abstract from an ethnographic study:

Threat of supernatural punishment can promote prosociality in large-scale societies; however, its impact in smaller societies with less powerful deities is less understood. Also, while perceived material insecurity has been associated with increasing religious belief, the relationships between insecurity, super- natural punishment beliefs, and prosocial behavior are unclear. In this study, we explore how material insecurity moderates the supernatural punishment beliefs that promote different expectations about distant, anonymous strang- ers among a sample of villagers living in Yasawa, Fiji. We examined this rela- tionship by employing an economic game designed to measure local recipient favoritism vs. egalitarian, rule-following behavior. Using indices of three dif- ferent “punishing” agents – the Christian God (“Bible God”), the deified ances- tors (Kalou-vu), and the police – we find that increased belief in Bible God punishment predicts less local recipient favoritism at low and moderate but not high material insecurity. Punishing Kalou-vu also predicts less favoritism at low and moderate insecurity, but more favoritism at high insecurity. Police

Data Visualization

punishment poorly predicts favoritism, suggesting that secular authority has less impact on isolated communities. We discuss implications for understand- ing how different kinds of supernatural and secular agent beliefs impact pro- social behavior. (McNamara, Norenzayan, & Henrich, 2016, p. 34)

Case Study and Single-Subject Designs Case study research in the social sciences is focused on a specific group, environment, or situation and can include large groups or even single subjects. This research strategy is com- monly used when the topic of interest is not common, or when a particular local problem is the focus. The value of investigating a real-world phenomenon within a particular context cannot always be done using inferential designs. For instance, a researcher may want to look at why a person committed mass murder, which is not a common enough situation to be generalizable.

Here is an abstract from a case study:

When a serial killer is present, the media and local authorities often claim that public attitudes and behaviors change sharply, usually becoming more fearful and less outgoing. Unfortunately, much of the evidence on this score is either speculative or anecdotal. This study reports quantified results from a series of surveys conducted over the course of a serial killing spree in Baton Rouge, Louisiana. The temporal trend in fear of crime is punctuated by a mod- erate increase during the serial killing spree, and a sharp decline after the apprehension of the serial killer. Moreover, post apprehension data reveal that nearly 56% of respondents report experiencing an increase in their fear of crime specifically in response to the serial killer. This was fairly evenly distributed across races and marital statuses, but, as expected, females and younger people were more likely to report increases in fear. Additionally, 46% of respondents took the extra step of implementing some sort of protective measure, with the most frequent being carrying mace or pepper spray or add- ing a security device to their home. In the latter case, respondents were moti- vated by a mix of concern over their own safety and that of their family. (Lee & DeHart, 2007, p. 1)

5.5 Data Visualization Data can be represented in graphs and figures to help show its characteristics. Data visu- alization can be helpful to researchers as they check to see whether the data are normally distributed or independent and whether other assumptions are being met. For people in the business world, data visualization can also be a quick way to show a summary of results in a business presentation, for example a pie chart can show a company’s total sales by month. Some examples of ways to visualize data are included below.

Frequency Distribution A frequency distribution shows how scores of a variable are distributed in a sample. They can be helpful in seeing the characteristics of a group in relation to some category of interest. For example, if you want to see how many people a certain disease affects based on age, you might use a frequency distribution. Another example is shown in Figure 5.1.

Section 5.5

Section 5.5Data Visualization

Figure 5.1: Frequency distribution ሁ A frequency distribution shows the frequency with which certain data points appear.

Cumulative Frequency Distribution A cumulative frequency distribution (see Figure 5.2) adds the frequency of data points at a certain score together with all of the previous data points to show the percentage as it increases from the lowest score on the variable to the highest. This might be useful for under- standing the number of observations that lie above and below a particular value. For example, in Figure 5.2 one can see that around 500 students scored 100 points or fewer on the text, while the remaining students scored above 100 points.

Figure 5.2: A cumulative frequency distribution ሁ This cumulative frequency distribution shows the proportion of students who scored at or below

each data point.

Section 5.6Statistical Analyses

Pie Chart A pie chart (see Figure 5.3) is divided into “slices” to illustrate the numerical proportions of the variable of interest. This type of illustration is often used to show different responses to a survey question or proportions of a population with a certain characteristic, such as race or age. One particular strength of pie charts is that they show proportions very clearly. Figure 5.3 shows the proportion of students receiving a particular grade on a test.

Figure 5.3: A pie chart ሁ This pie chart notes the distribution of grades in a group of students.

5.6 Statistical Analyses Descriptive statistics are simple ways of describing the characteristics of a sample or popula- tion. For example, if you want to see which major city in the U.S. treats the most people for addiction-related issues, you could look at the records of all major cities and compare. By looking at the rates of treatment, you could make decisions about where resources could be better spent. See the Career Spotlight feature on Paul Randall Gesn to learn more about sta- tistical analysis.

D, 26%

F, 10% A, 9%

B, 17%

C, 38%

Career Spotlight: Paul Randall Gesn

Name: Paul Randall Gesn

Primary job title: Statistical Analyst

Type of employer: State Health Agency

What degrees do you hold? A doctorate in experimental psychology with a specialization in social and personality psychology.

(continued on next page)

Section 5.6Statistical Analyses

Measures of Central Tendency: Mean, Median, and Mode The average is a general term that can refer to all three measures of central tendency, but typically indicates the arithmetic mean. However, it is important to have an understanding of all three measures: mean, median, and mode.

The mean (often notated as M, µ, or x̄) is the sum of all values in a group of scores divided by the number of scores (this is also known as the arithmetic average). For example: (5 + 10 +

Career Spotlight: Paul Randall Gesn (continued)

Describe your major job responsibilities. I provide research and data analytic support for a program that oversees the provision of services to women and children. Specifically, I help to develop indicators to monitor for fraud and other noncompliant behavior among the external vendors who actually provide those services. I also conduct analysis on trends in program participation and expenditures.

Provide a general overview of your career background. I have held positions, in both government and the private sector, related to healthcare, education, and the development of nationally representative standardized tests and assessments. The common theme among all my nonacademic positions has been a focus on research and data analysis. I have also taught psychology courses online and in a campus-based setting. Although my full- time work is in a nonacademic/applied setting, I enjoy maintaining my ties to academia. Many colleges and universities are hiring adjunct instructors rather than full-time instruc- tors, so the opportunity to teach at least one course is generally available.

Are there any undergraduate activities that you think helped to influence success in your type of career? I was involved in conducting my own research project during my junior and senior undergraduate years, and this really helped to solidify my interest in research and analyzing data. This experience helped to influence the direction I took in selecting a graduate program and then deciding on a career path to pursue.

What advice would you give to someone thinking about a field similar to yours? In graduate school, take as many statistics, research methods, and psychometrics courses as you can. If possible, become involved with an actual research project. You should enjoy working with numbers, extracting meaningful information from large datasets, and pro- viding that information to a variety of audiences, some of whom do not have a background in research and statistics. Research- and statistics-intensive job positions often require at least a master’s degree, but some employers will hire at the bachelor’s level, provided the applicant has previously gained the required knowledge and experience. Also, you need to have a good working knowledge of at least one statistical analysis software package, such as SAS, SPSS, Stata, or R.

What do you like most about your job? I really enjoy starting with a large database of raw data and determining if there is any meaningful information to be gleaned from that data. I think every dataset tells a story, and it is the job of the analyst to determine what that story is. I also enjoy conducting research and writing up the results in formal research reports. Finally, I enjoy communicating results to audiences who might not have a back- ground in research or statistics. The communication is generally two-way: they can also provide me with information on the policy implications of the results.

Section 5.6Statistical Analyses

15 + 20) / 4 = 12.5. The median (often notated as Mdn or Mdn) is the number that lies at the midpoint of a distribution of numbers when the numbers are arranged in numerical order. An example would be 1, 2, 3, 4, 5, 5, 5, 6, 6, 7, 8, 8, 8, 9, 20. The value 6 is in the exact middle, so it is the median. When there are an even number of values, the average of the two middle values is the median, such as: 1, 2, 3, 4, 5, 5, 5, 6, 7, 8, 8, 8, 9, 20. In this case, since (5 + 6) / 2 = 5.5, the median is 5.5. The mode (often notated as Mo) is the most frequently occurring number in a distribution. For instance: 1, 2, 3, 4, 5, 5, 5, 6, 6, 7, 8, 8, 9. Five is the mode as it occurs more than any other value.

As you critique articles, or any type of research results (such as is commonly found in news reports), be sure to pay close attention to which measure(s) of central tendency were used in the study and be careful about assuming what the “average” is.

Measures of Dispersion The measures of central tendency tell us where the relative center of the distribution lies. Researchers also want to measure how spread out the values are in the sample; this is where measures of dispersion come into play. Some of the measures of dispersion are the range, interquartile range, variance, and standard deviation. The range tells us the distance between the lowest and highest values in the distribution. The interquartile range (see Figure 5.4) tells us the difference between the lowest and highest values in the two middle quartiles (the middle 50% of the scores) when the distribution is divided into four sections, or quartiles.

Figure 5.4: The interquartile range ሁ When the data are divided into four equal quartiles, the interquartile range is composed of the

middle two quartiles, or the middle 50% of the data, as shown here.

The standard deviation tells us how much the scores deviate from the mean scores and is just the square root of the variance. Another way of putting it is the average amount of vari- ability in a set of scores. The standard deviation is usually notated as s, SD, or σ. You often see distributions expressed as the mean ± the standard deviation, such as 14.83 ± 2.51. A standard deviation can tell us how varied a data set is. For instance, if the standard deviation of a set of grades on a particular test is small, then we can conclude that the grades were very similar. If the grades were mostly high, this might mean the test was too easy; if they were mostly low, it might have been too difficult. Looking at the distributions of scores can help teachers adjust their tests over time to be appropriately challenging.

Section 5.7Inferential Statistics

Standard Normal Curve: The Bell Curve In a normally shaped distribution, half of the population will be above the mean and half will be below it. In psychology research, most populations are normally distributed and the bell- shaped curve helps us understand how this might look. If we averaged the math test scores of all the students in the 8th grade at a school, we might find that the average score is 50. Half of the students would have higher scores, and half lower, in a normally distributed sample (see Figure 5.5).

Figure 5.5: The bell curve ሁ This normal curve is based on the distribution of grades in a sample.

In a normal curve, about 68% of the data is within one standard deviation of the mean, about 95% is within two standard deviations, and about 99.7% is within three standard deviations. This fact makes it easy to quickly assess ranges and frequencies of certain outcomes across a population. Additionally, the standard normal distribution (a distribution with a mean of 0 and a standard deviation of 1) can be used to calculate the probabilities involved in any other normal distribution. Due to these benefits and others, normal distributions are a popular statistical analysis tool.

5.7 Inferential Statistics Researchers often want to see if two variables are connected. For instance, we might want to investigate if the amount of time a person studies is connected to how well they do on a math exam. In this case, we want to use an inferential test called a test of relationships to measure that connection. Tests of relationships are designed to show how groups or variables are related. Non-researchers also need to have an understanding of how variables are connected, and how those connections are verified with scientific methods. For example, say the results of the previously suggested research on studying show that more studying is related to a

Section 5.7Inferential Statistics

higher grade, but only to a point; after a certain number of hours of studying, there is nothing more to be gained. Being able to interpret these results might influence your study habits.

Correlation Correlation analyses are used to look at the mutual relationship between two vari- ables, such as how IQ is related to GPA. A positive correlation indicates the extent to which the two variables move up together or down together. IQ and GPA are posi- tively correlated, so as IQ gets higher, GPA gets higher. A negative correlation indi- cates the extent to which one variable moves up while the other variable moves down. If that were the case with IQ and GPA, as IQ moved lower, GPA would move higher. When there is no clear relationship, the variables are considered to have zero correlation. For instance, IQ may not be a factor in the grades in a subject such as wood shop. While in cases of positive and negative correlations, the movement of one variable can reliably predict a similar move in another variable, that correlation does not imply causation. There may be an unknown factor that influences both variables similarly. For instance, amount of ice sold in grocery stores may have a positive correlation with sunburns, but it is obvious that ice does not cause sunburns, nor do sunburns cause people to buy ice. The root of the relationship may be that a third fac- tor, weather, influences both sunburns and ice consumption in similar ways, causing them to be correlated.

In a similar vein, it is important to note that a correlation coefficient does not address the issue of cause and effect. Do not make the mistake of thinking that a high correlation implies that one variable causes variation in the other variable. For example, there might be a nega- tive correlation between the number of pirates in the world over the last 200 years and the pounds of ice cream sold each year (the number of pirates has gone down while the amount of ice cream sold has gone up). However, it is not likely that the reduction in pirates each year has caused ice cream sales to go up. This is often referred to as a spurious relation- ship or spurious correlation. This shows how essential it is to carefully interpret results from correlational research and to not make causal assumptions without considering all possible variables that affect a relationship. For those not in the research field, this skill can come in handy when reading the news; for example, when you hear about some connection between eating a certain food and experiencing a certain health concern, it is important to be critical of assumptions that one caused the other to happen. Cause-and-effect assertions cannot be substantiated outside of true experimental designs (see Figure 5.6).

Some examples of variables that are correlated include:

• Height and time to run a mile • Hours of studying and grade on a statistics exam • Income and happiness • Caffeine intake and alertness

Dmitrii_Guzhanin/iStock/Thinkstock ሁ Tests of relationships measure how two

variables interact.

Section 5.7Inferential Statistics

Figure 5.6: Scatter diagram of a positive correlation ሁ This plot shows the correlation between the monthly price of car insurance and the annual

income of a new college graduate.

The scatter diagram in Figure 5.6 is a visual display of data which allows you to see an asso- ciation between two variables. For example, you can see there is a strong, positive relation- ship between the annual income of new college graduates and the price they pay monthly for car insurance. The scatter diagram illustrates the strength of the correlation between the variables along the slope of a line; the straighter the line, the stronger the correlation. This correlation can point to, but does not prove, a causal relationship. It may be that people who make higher wages purchase more expensive cars, and that is why their insurance premiums are higher.

The Correlation Coefficient While a scatter diagram is an excellent way to look at the relationship of two variables, many journals do not publish graphs due to the large amounts of space they take up. Numerical summaries of results appear in research reports much more frequently than do visual rep- resentations. Correlations are also often presented in the news and on social media in the following manner: “a new study suggests there is a correlation between heart health and consuming meat.” Claims like that need to be carefully evaluated on the nature of the research design.

The correlation coefficient (see Figure 5.7) is the numerical summary of a bivariate (two- variable) relationship. It is usually symbolized as r. The correlation coefficient has a range of –1.0 (a perfect inverse relationship) to +1.0 (a perfect positive relationship), with 0 being no correlation at all.

Section 5.7Inferential Statistics

Figure 5.7: Correlation coefficient range ሁ The correlation coefficient ranges from –1.0 to 1.0.

When an r falls on the right side of the continuum (from 0.0 to 1.0), it indicates that there is a “direct” or positive relationship between the two variables. As the score on one variable increases, so does the score on the other variable. Here is an example of a positive correlation from a real research project:

Results indicated that increases in self-compassion were significantly related to the number of days a week that participants meditated (r = .42, p < .05), as well as the number of times per day they informally practiced self-compas- sion (r = .43, p < .05). (Neff & Germer, 2013)

When an r falls on the left side of the continuum (from –1.0 to 0.0) it indicates an “indirect” or “inverse” relationship between the variables. As the score on one variable increases, the score on the other variable decreases. For example, a hypothetical study might have looked at school attendance and GPA:

Results indicated that there is a significant inverse relationship between the number of days a student misses in school and their GPA (r = .72, < .01). So, the more days a student misses of school, the lower their GPA tends to be, compared to the average.

When a correlation is strong, it is said to be more predictive than when a correlation is weak. How do we know if a correlation is strong? The common conventions for the effect size of a Pearson correlation coefficient are (note that these are based on absolute values) based on Cohen (1992, p. 157):

Trivial: 0 to < .1 Small to medium: .10 to .30 Medium to large: .30 to .50 Large to very large: .50 to 1.0

Be aware that though an effect size can tell us the general strength of the relationship between the two variables, the practical value greatly depends on the research conducted. For instance, when looking at human behavior, there can be many variables that contribute to a particular outcome, so a relatively small effect size can be significant. But a medical study looking at the effects of a particular medicine may require a much higher effect to be considered viable.

When at or close to r = 0.0, there is no evident correlation, the pattern of association is ran- dom, and it shows non-systematic variation. No line, straight or otherwise, can be fit to the

– 1.0 0.0 1.0

Positive correlationNegative correlation

Section 5.7Inferential Statistics

relationship between the two variables. The two variables are said to be uncorrelated (see Figure 5.8).

Figure 5.8: Scatterplot ሁ This scatterplot shows no evident correlation between the variables income and shoe size.

One issue to watch out for when analyzing data for correlation is the possibility of outliers. You have outliers when one or more of the data points are located away from the bulk of the scores. These can cause the size of a correlation coefficient to underestimate or exaggerate the strength of the relationship of two variables. Often when performing statistical analyses, tests will be done to identify whether or not a data set contains outliers. If outliers are pres- ent, the researcher will note that in the findings and potentially run another analysis on the data without the outliers to compare the results against the previous findings. Outliers should be carefully considered in correlational research.

Regression Analysis In some situations, we may want to use more than one variable (often called predictor vari- ables) to predict some other variable (often called the criterion variable); this is called regres- sion analysis. Using more than one predictor variable can usually help increase the amount of variance accounted for in the criterion variable. Put another way, we can usually increase predictive validity by using multiple predictors rather than just one.

As an example, suppose that the variables are as follows:

Y Annual salary as a college professor

X1 Number of years since PhD was obtained

X2 Number of publications

Section 5.7Inferential Statistics

X3 Age (in years)

X4 Student ratings of teaching performance

Suppose that we do a correlational study and obtain the following correlation matrix (Table 5.1):

Table 5.1: Correlation matrix

Y X1 X2 X3 X4

Y ––– .62 .46 .26 .51

X1 ––– .68 –.15 .46

X2 ––– .05 .30

X3 –– –.01

X4 ––

Next we run a multiple regression analysis, and obtain the following summary of results (Table 5.2):

Table 5.2: Summary of results

Step Variable R R2 R2 change

1 Years since PhD (X1) .62** .38 .38**

2 Student ratings (X4) .67** .45 .07**

3 Age (X3) .70** .49 .04*

4 # of publications (X2) .71** .50 .01

*p < .05 **p < .01

This can be interpreted as follows: The four predictors X1 – X4 display a multiple correlation of .71 (R; significant at the p < .01 level) which accounts for 50% of the variance in Y (R2). The most important variable in predicting annual salary appears to be the number of years since the professor got his or her PhD (X1), which accounts for 38% of the variance in annual salary. After controlling for X1, student ratings of the professor’s teaching (X4) account for an additional 7% of the variance in annual salary. After controlling for X1 and X4, we find that the professor’s age (X3) accounts for an additional 4% (p <.05), and after controlling for the other 3 predictors, number of publications (X2) adds an additional 1% to the predictable variance of annual salary. The moral of this regression example is if you want to make more money as a professor, get your PhD early, teach well, and age.

Statistical Tests of Differences Sometimes researchers want to test a hypothesis that groups are different from each other. Different from the tests of relationships, the tests of differences look to see how variables or groups are different from each other. This set of tests include t tests and ANOVA analyses.

Section 5.7Inferential Statistics

The t Test We use t tests to assess if the means of two groups are statistically different from each other. For instance, a researcher might want to see if a group that gets a depression medicine (the experimental group) is statistically different from the group that gets no medical treatment for depression (the control group). There are two main types of t tests, dependent and inde- pendent. A dependent t test, also known as a “within-subjects” or “repeated measures” test, is used when the same sample is measured twice. The goal is to evaluate if there is a change from the first measurement to the second. The other type of t test is the independent samples t test. This is used when comparing two different samples on only one measurement.

Here is an example of a study using a t test:

Children in our sample were generally very familiar with electronic games. Of our sample, 84.4 % reported playing video games on a computer, 81.2 % on a console and 50.4 % on a handheld device in the previous 6 months. Only 6.1 % reported playing no games at all during that time. Similarly, only 11.4 % of our sample had no exposure to violent video games. Boys had consider- ably more exposure to violent video games than did girls [t(189.24) = 9.07, p < .001, r = .46, 95 % CI = .38, .54] (Ferguson & Olson, 2014).

Analysis of Variance (ANOVA) A one-way analysis of variance (ANOVA) makes it possible to compare data for three or more groups, similar to running all possible t tests between each pair of groups at the same time. Let’s say you want to do a study to compare three study groups to see how study style affects grades on a final exam. The first group studies each day for two weeks, the second group studies each day for one week, and the third group only studies on the day before the exam. The ANOVA looks at the differences between the grades of the three groups (called between-groups variation), as well as the variance between the grades of each individual in each group (called within-groups variation). If the variances between the three teams of stu- dents are significantly larger than the average variation found in each person’s performance, then you may have a significant difference in student grade by group. For instance, the aver- age score for three groups might be 90, 85, and 72, making the overall average 82.33 (the three numbers added together and divided by three). In the end, the ANOVA (also commonly referred to as an F test) gives you one value that quantifies whether or not there is a signifi- cant difference between your populations, such as F = 3.43, p = 0.012. In this case, because the p-value is less than 0.05, we know that one of the groups is significantly different from the others.

Post Hoc Analysis Let’s say that you ran an ANOVA and the results suggest that there are some differences between the groups—how will you know which of the groups are different from the average? You have to run a post hoc (meaning “after the fact”) analysis. This analysis dives deeper into the single-value result of the ANOVA so that you can figure out which group study strategy is better than the others, rather than just saying that there is definitely a difference between the three groups. We know from our ANOVA analysis that the average score for the three groups is 82.33, and we know that one group is significantly different from the others, but not which one. A post hoc analysis can then be performed to find out that the third group, with an

Section 5.7Inferential Statistics

average score of 72, is significantly different from the others (different meaning, in this case, they performed significantly worse on the test).

Non-Parametric Tests: Chi-Square Tests The ANOVA procedures we discussed in the previous section allow researchers to compare group means, but sometimes we might want to simply compare the frequencies associated with different groups or with some expected values. Suppose, for example, that we choose to study the number of men and women that are managers in a major company. If the distribu- tion of managers is random with respect to gender (based only on the two categories of male and female), we would expect close to half males and half females. However, if we observe that 85 of the 100 managers are female and only 15 are male, we would be tempted to con- clude that this distribution is non-random and that the company has an unfair distribution of women to men based on our expectation of a 50/50 ratio. A chi-square analysis would be used to try and verify that assumption that the ratio is 50/50.

The chi-square tests reviewed below are called nonparametric tests, because we do not have to make the assumption that the data are normally distributed. They test nominal data, which are measured in terms of frequencies. The key question asked in a chi-square test is whether the frequencies observed in a group (or levels of a variable) are significantly different from what is expected.

One Sample Chi-Square Goodness of Fit Test We sometimes want to compare observed frequencies to see if they are a good fit compared to the frequencies we expect to see. For example, we might want to compare the observed frequencies of extroverts versus introverts in a classroom of 100 undergraduates to see if fits a 50/50 distribution. An alternative hypothesis might be that there are a significantly unequal number of introverts and extroverts in the class. Suppose that there were five rows of 20 seats each and we used the Myers Briggs Type Indicator (MBTI) to determine if each of the 100 students in class is an extrovert or introvert. The expected frequencies of extroverts for each row would be .5 × 20 = 10. The following distribution is observed where the column marked O refers to the observed frequencies of extroverts (Table 5.3).

Table 5.3: Frequencies of extroverts

Row O

1 19

2 17

3 15

4 9

5 5

There are a total of 65 extroverts in the class. A calculation of the chi-square would show that this is significantly different from the expected value of 50 extroverts. So, it appears there is not an equal distribution of extroverts and introverts in the classroom.

Section 5.7Inferential Statistics

Chi-Square Test of Independence We may want to compare several groups on some type of classification variable. For example, we might want to gather data on Ashford University psychology majors who indicate that they are a) going to graduate school after getting their bachelor’s degree, b) getting a job after getting their degree, and c) not sure what they are going to do after graduation. Then, perhaps, we want to compare them based on a four-category variable indicating the degree status of their father: 1) high school or lower, 2) some college, 3) bachelor’s degree, or 4) some graduate work or graduate degree. The results from a hypothetical study of 200 Ashford University psychology majors are shown below (Table 5.4):

Table 5.4: Data from hypothetical survey of psychology majors

Degree of education of father

High school

Some college

Bachelor’s degree

Graduate school

Row total

Post- graduate plans

Grad school 4 14 22 40 80

Job 25 25 25 5 80

Not sure 10 15 10 5 40

Column total 39 54 57 50

What trends do you see in these data? A statistical analysis of these data would show that there is a significant difference between the father’s degree status and what the child may do. But, further analysis would be needed to determine which factor level is most significant. However, a look at the observed table will often help the researcher determine what is signifi- cant. For instance, you can see just looking at the table that a student whose father went to graduate school is more likely to go to graduate school (the value is 40) than other students.

T E S T Y O U R K N O W L E D G E : D E S C R I P T I V E V E R S U S I N F E R E N T I A L T E S T S Which of the following is a good example of using inferential statistics?

1. Calculating the average age of each person in the U.S. for the Census.

2. Maintaining a spreadsheet of all the NFL player scores for each year.

3. Asking random people at a local mall about their shopping preferences to inform store owners about current trends.

The first two are examples of descriptive statistics in that the data describes everyone in a population. The third choice is inferential, as data from a random sample is used to make inferences about a larger population.

Section 5.8Ethics in Research

5.8 Ethics in Research What is research integrity and why is it important from an ethical standpoint? What are the costs of poor integrity in research? Research integrity refers to using honest and valid methods in design- ing, conducting, and reporting research based on ethical standards. These meth- ods are often based on rules, regulations, guidelines, and professional norms that are regulated by professional organiza- tions. For instance, all research in the United States that involves human sub- jects must adhere to the guidelines of the Office of Human Research Protections of the Department of Health & Human Ser- vices, which can be found at https://www .hhs.gov/ohrp/. Poor conduct in research can result in negative outcomes such as harm to participants, loss of credentials (such as a professional license), and legal issues related to fraud and scientific misconduct.

Researchers must consider everyone involved in the process of collecting and analyzing data, from the researchers to the participants. Having strong research integrity is vital to ensuring all people (and sometimes animals) involved are carefully considered and protected from undue harm. For instance, if a researcher provides false data regarding a new medicine to treat schizophrenia, patients that are prescribed the medicine could be exposed to harm- ful side effects, or even death. Psychologists and researchers in this area are governed by the guidelines proposed by the American Psychological Association, as discussed in prior chapters.

While there are overarching ethical guidelines involved in the practice of psychology, there are some considerations that are specific to data and research methods. Some ethical pitfalls to avoid in research include the following:

• Data falsification: This occurs when researchers falsify data that they collected to show the results that they are looking for even when nothing significant was found.

• Data fabrication: This occurs when researchers claim to have data even when none was actually collected.

• Plagiarism: Unintentional plagiarism occurs when a researcher utilizes too much information gathered from previous studies during the literature review stage. Intentional plagiarism occurs when researchers publish others’ results as their own.

• Conflict of interest: A conflict of interest can occur when a researcher has another obligation related to a study. For instance, a researcher that works for a pharmaceu- tical company may choose to show only positive gains when reporting the results so that the company (and the researcher) can receive financial benefit.

skynesher/E+/Getty Images ሁ Research integrity requires conscientiousness

and an adherence to established ethical guidelines.

https://www.hhs.gov/ohrp/

Section 5.9Career Applications Involving Research and Statistic

5.9 Career Applications Involving Research and Statistics

Many students ask, “why do I need to take research methods and statistics?” Statistics is not most people’s favorite topic, but it is essential to practice in many areas of psychology, even in non-research settings. To be viable in any field, it is essential that one is able to keep up-to- date with current practices and research. For instance, a clinical psychologist that works with veterans must understand the newest research on PTSD to be able to utilize the latest treat- ments. To do that, he or she must read the current research articles on that treatment, which requires an understanding of the research methods and principles of how the statistical test- ing was applied. An industrial/organizational psychologist utilizes many different tests of fac- tors like personality and career strengths, and a marketing analyst may be required to mine user data to develop a report about the demographics of the people that have used a certain social media tool and looked at their product. An understanding of statistical processes, such as testing for reliability and validity, is essential to effectively utilize research tools. But indi- viduals who do not intend to conduct research still need to know the main ideas behind the most common strategies.

There are also many careers that are research-focused, even for individuals with only a bach- elor’s degree. Most universities employ graduates that have a bachelor’s degree as analysts in institutional research and assessment. These individuals are responsible for data entry, designing studies, and analyzing data using basic and advanced statistical procedures. Gradu- ates with skills in research design and statistics can also find work in many fields, such as marketing, clinical research, and military organizations. As data becomes more and more available in online modalities, there is also a need for graduates who are skilled at analyz- ing information from social media sites and online behavior. Whether you want to conduct

Career Spotlight: Sandra G.

Sandra Gibson is a human resources analyst for a Fortune 500 company. She uses research and data to help find the best job candidates for certain positions in the organization. This may start with a job analysis for a particular position to determine the most desir- able characteristics and qualifications for that specific job. For instance, if the company is looked to hire a sales manager, her research may show that the best candidate would have previous sales experience and be outgoing, conscientious, versatile, and able to work under tight deadlines. Once the profile is created, the job is posted and people apply. Sandra can use psychometric analyses, such as personality tests, along with performance measures, to help find the best candidate for the sales manager position. While Sandra is not a tradi- tional researcher, she uses research methods and data to perform her job and to find the best candidates for her organization.

Additionally, Sandra must keep to an ethical code set by the government and by the com- pany she works at, as well as her own moral compass. Every day she handles employees’ and applicants’ personal data, and it is her duty to keep that information confidential, even from her other coworkers in human resources. She also has to remain impartial during the job selection process, and refrain from letting bias interfere with her decision making. Keeping these ethical considerations in mind, Sandra is always looking for new, data-based ways to better predict who would be the best person for a certain position.

Summary

research or work in a particular area related to psychology, you need to have a good under- standing of basic research methods and statistical analyses.

Summary We began this chapter discussing why an understanding of research is essential to many dif- ferent career paths psychology graduates might pursue. An understanding of how research is conducted and reported is vital to being able to keep up with current trends in any field. For instance, a person that works in marketing may need to learn about new methods of utilizing social media to promote a product based on psychological factors. This chapter also described some basic qualitative and quantitative methods that are common in psychological research. In this chapter, common methods of visualizing data were reviewed, and a simple overview of statistical analysis was given. This should be familiar from your previous statis- tics class, but hopefully this served as a useful refresher. Lastly, the chapter discussed ethical considerations, as being able to understand the ethical implications of research affords you the chance to critically evaluate research and professional practices in many of the areas you may find yourself in.

Concept Check 1. Which of the following might a research psychologist use research data for?

a. to create a new medicine b. to learn more about other fields c. to attempt to prove a hypothesis d. to evaluate a program

2. A hypothesis does not have to be clear or specific to be testable. a. True b. False

3. Grouping people by their ethnic group membership constitutes data of what scale? a. nominal b. ordinal c. interval d. ratio

4. In statistics, a population refers to which of the following? a. all of the values in a particular sample b. all possible members of a defined group c. a theoretical group d. any group based on a large, random selection

5. Pie graphs have what particular strength? a. they indicate each value individually b. they list individual data points in order c. they make proportions very clear d. they make calculating descriptive statistics easy

6. Which of the following is not a measure of central tendency? a. mean b. range c. median d. mode

Summary

7. Studying the characteristics of a population by analyzing a sample describes which? a. descriptive statistics b. inferential statistics c. statistical analysis d. parameter estimation

8. Which of the following is not considered an ethical pitfall related to research? a. data falsification b. plagiarism c. conflict of interest d. making a lot of money

9. An understanding of research methods and results is important in many careers, even if the professional is not conducting research. a. True b. False

Answers 1. c. The answer can be found in Section 5.1. 2. b. The answer can be found in Section 5.2. 3. a. The answer can be found in Section 5.3. 4. b. The answer can be found in Section 5.4. 5. c. The answer can be found in Section 5.5. 6. b. The answer can be found in Section 5.6. 7. b. The answer can be found in Section 5.7. 8. d. The answer can be found in Section 5.8. 9. a. The answer can be found in Section 5.9.

Questions for Critical Thinking 1. Your colleague asks you to be an author on her research article that she is submit-

ting for publication, but you have not been a part of that research project. Discuss some ethical considerations that might be related to this situation.

2. An understanding of published research is important for a researcher, but it can also be valuable to non-researchers. Describe some ways that you might need to review and evaluate published research in the career that interests you.

3. A colleague approaches you with a research idea that involves investigating the relationship between job satisfaction and employee selection at the organization you work for. Describe what type of sampling methods and inferential statistics might be most appropriate for this study.

Key Terms analysis of variance (ANOVA) A set of statistical models used to test the mean dif- ferences on a variable of interest when there are more than two groups.

average A general term that can refer to the mean, median, or mode, but is most often used to indicate the arithmetic mean.

chi-square analysis A set of statistical analyses used to test if there are statistical differences between expected and observed frequencies.

cluster sampling A sampling strategy that involves dividing the population into clus- ters and then randomly choosing a certain number of people in each cluster.

Summary

confounding variable Variables that influence the independent and dependent variables of a study but are not a part of the research design. Also referred to as extrane- ous variable.

convenience sample A sampling strategy in which the participants are chosen based on availability. See also non-probability sampling.

correlation analyses A statistical test to determine the strength of relationship between two variables.

correlation coefficient The numerical summary of a bivariate (two-variable) rela- tionship. It is usually symbolized as r. The correlation coefficient has a range of –1.0 (a perfect inverse relationship) to +1.0 (a perfect positive relationship), with 0 being no correlation at all.

cumulative frequency distribution A table of frequencies of a data class that is the sum of the data elements in that class and all the previous classes.

dependent variable In scientific research, it is the variable that represents the out- come a researcher is attempting to measure; it is directly affected by the independent variable.

frequency distribution A statistical table that displays the frequency of outcomes in a sample.

hypothesis A testable statement that pro- poses relationships or differences between two or more variables.

independent variable In scientific research, it is the stand-alone variable that the researcher controls.

interquartile range The difference between the lowest and highest values in the two middle quartiles (the middle 50% of the scores) when the distribution is divided into four sections, or quartiles.

interval scale A scale of measurement where the distance between data points is equal and known.

nominal scale A level of measurement that is categorical and has no inherent numerical value.

non-probability sampling A method of gathering a sample from participants who are chosen based on the judgement of the researcher rather than at random.

observational research A type of non- experimental design in which the researcher observes and records behaviors in the con- text of the study’s parameters. Also referred to as field research.

ordinal scale A level of measurement where the order of responses is important, but the difference between them is not quantifiable.

outliers Data points that are located away from the bulk of the scores.

pie chart A circular statistical graphic which is divided into slices to illustrate the numerical portions of a variable.

probability sampling Any method of gathering a sample from a larger population using some form of randomization.

proportional sampling A sampling method that divides participants into subgroups and samples each subgroup randomly.

purposive sampling A sampling strategy that involves selecting participants based on some characteristic they may have that is of interest to the researchers. See non-proba- bility sampling.

quartile In descriptive statistics, the quartiles of a ranked set of data values are the three points that divide the data set into four equal groups, each group comprising a quarter of the data.

Summary

random sampling The sampling method where everyone in a population has an equal chance of being selected.

range The distance between the lowest and highest score in a set of values.

ratio scale A scale representing the math- ematical relationship between two numbers indicating the number of times the first number contains the second. This scale of measure is also characterized by having an absolute zero as the lowest possible value.

regression analysis A set of statistical measures to assess the strength of the rela- tionship between several variables.

research question A specific, clear, and testable question that guides the research design.

spurious relationship An assumed causal relationship between two correlated vari- ables that are almost certainly not caus- ally related. Also referred to as spurious correlation.

standard deviation A statistical value of the variation in a set of data values.

stratified sampling The sampling strategy where a population is divided into sub- groups (the strata) and then each subgroup is randomly sampled from.

systematic sampling A sampling strategy that involves a set protocol for selecting participants.

tests of differences Tests that examine how variables or groups are different from each other.

tests of relationships Tests that exam- ine the relationships between groups or variables.

t test A statistical test to assess if the means of two groups are significantly different from each other.

variable Any factor, in relation to research, that varies in quantity or quality.

Professional Resources Professional Organizations that Post Research Careers in Psychology http://www.apa.org/careers/index.aspx

http://www.siop.org/

Research and Ethics Training https://about.citiprogram.org/en/homepage/

Psychological Research Blog http://bps-research-digest.blogspot.com/

http://www.apa.org/careers/index.aspx

http://www.siop.org/

https://about.citiprogram.org/en/homepage/

http://bps-research-digest.blogspot.com/

_Hlk522011150
_Hlk510769925
_Hlk522013031
_Hlk528668325
_Hlk528668381
_Hlk528668391
_Hlk528668414
tea
_Hlk528668427
_Hlk528668445
_Hlk528668452
_Hlk528668468
_Hlk522012187