Interpreting and Using Statistics in Psychological Research. Andrew N. Christopher

      Figure 2.1 Sample Screen Displays in Terrell et al.’s (2008) Research

      After learning the status of the other supposed participant, the real participant was then informed that he or she would be competing with this person on an upcoming task that involved learning and reaction time. Here are the directions that the researchers gave the participant:

      The task (in the experiment) is to find a target letter among four letters on the computer screen. As soon as you press the response key, the screen will change. If there is no response, the screen will change approximately every 2 seconds. If the target letter is in the layout of letters, you will press 1. If the target letter is not in the layout of letters, you will press 2. (Terrell et al., 2008, p. 819)

      You can see an example of one of these computer screen displays in Figure 2.1.

      After practicing the task for two minutes, participants were informed:

      [F]or the competitive trials, both of you (the actual participant and the supposed participant down the hallway) will be wearing headphones, and you will have the option of administering noise blasts to your competitor in order to distract (him/her). Pressing the space bar produces the noise blasts. These noise blasts are very unpleasant, and will disrupt performance, causing errors in the competitive task. (Terrell et al., 2008, p. 819)

      Of course, the real participant was told that the other (bogus) participant could also give noise blasts. Indeed, each real participant received 30 randomly spaced noise blasts while performing the task, which lasted 10 minutes. The real participant could administer noise blasts, too, which again, would impair the performance of his or her supposed high- or low-status competitor. Terrell and her colleagues (2008) wanted to see how aggressive the real participant would be toward his or her supposed competitorbased on the real participant’s sex and personality and the competitor’s supposed status.

      Learning Check

      We will take some of our knowledge from the third section of Chapter 1 and apply it to Terrell et al.’s (2008) research. It may help to keep in mind the title of Terrell et al.’s research, “Gender Differences in Aggression: The Role of Status and Personality in Competitive Interactions,” as you answer these questions.

      1 What is (are) the independent variable(s) in this study?A: Status of the other (nonexistent) participant.

      2 What is (are) the quasi-independent variable(s) in this study?A: Participant’s sex and participant’s personality. Remember that people cannot be randomly assigned to have a sex or a particular type of personality. That’s why these are quasi-independent variables. The status of the other participant is a “real” independent variable because random assignment was used to determine whether a participant thought he or she was competing with a low- or high-status other participant.

      3 What is (are) the dependent variable(s) in this study?A: Aggression is the dependent variable. It is the outcome of the status of the other participant, as well as the participant’s sex and personality; that is, aggression depends on these factors.

      Variables

      Let’s dissect this research study and introduce some basic statistical considerations that are critical for any research study to consider. First, there are, as you saw in the last Learning Check, several variables in Terrell et al.’s (2008) study. As mentioned briefly in the previous chapter, a variable is a quality that has different values or changes among individuals. For instance, qualities such as height, age, personality, happiness, and intelligence differ among people; hence, each is a variable. Variables can also be environmental features, such as noise level in the room. We all probably have some idea of what is meant by variables such as height and age. However, research in psychology specifically and social sciences more broadly needs to be able to measure variables that are not as readily apparent. In Terrell et al.’s research, they measured personality. What exactly does “personality” mean? I bet we all have an idea of what it means, but those ideas may not be the same for all of us. We need to know precisely what “personality” means in this example. Three aspects of personality were examined: tendency to act aggressively (Buss & Perry, 1992), narcissism (Raskin & Terry, 1988), and venturesomeness (Eysenck, Pearson, Easting, & Allsopp, 1985).

      These sorts of “internal” variables, such as personality, are called constructs; they cannot be observed directly but instead must be inferred in some way. We know what specific aspects of “personality” Terrell and her colleagues (2008) measured, and I bet you have an idea of what a “tendency to act aggressively,” “narcissism,” and “venturesomeness” each mean. For instance, when we say someone is a “narcissist,” we might think of someone who is self-absorbed and thinks highly of his or her abilities. But this is not good enough in research. We need to know precisely how each one of these aspects of personality was measured.

      Constructs: variables that cannot be directly assessed but must be inferred in some way.

      Operational Definitions

      Researchers must state explicitly not only what variables and constructs they examined, but also precisely how they measured them. An operational definition specifies the process used to measure a variable quantitatively in a research study. For instance, in Terrell and colleagues’ (2008) research, they measured narcissism using a 40-item survey that was constructed in prior research (Raskin & Terry, 1988). Table 2.2 contains four sample items from this survey. For each item that a person answers “TRUE,” he or she gets a point toward his or her total narcissism score. Scores could range from 0 to 40, with higher scores indicating higher levels of narcissism. This is the operational definition of narcissism in Terrell et al.’s (2008) research.

      Table 2.2

      Each item to which a person answers “TRUE” earns them one point toward their overall narcissism score. Each item to which a person answers “FALSE” earns them nothing toward their overall narcissism score.

      To get a score for the variable of “narcissism,” we sum the scores for each individual item (1 for TRUE; 0 for FALSE).

      Operational definition: specification of precisely how a variable is measured in a research study.

      Operational definitions should be perfectly clear in any research study. They allow readers to understand precisely what was done in the study, and they allow other researchers to conduct additional research to build on prior research. Let’s take another, potentially more difficult variable to operationalize in Terrell et al.’s (2008) study. Specifically, the variable of aggressive behavior is one that could be operationalized in many ways. When psychologists discuss aggressive behavior, they focus on a person’s intention of trying to harm someone else (Baron & Richardson, 1994). Here, aggressive behavior was operationalized as the numbers of noise blasts that the participant gave an unmet other participant to harm his or her task performance. How else could aggressive behavior be operationalized? Hitting another person could qualify as aggressive behavior if the intent is to harm the person being hit. However, in this research, such an operationalization was not possible because there was no actual other participant.¹ Alternatively, aggressive behavior could have been operationalized by asking participants how much they would like to harm the other participant. Doing so would get around the fact that there was no other participant. Regardless, the important point to remember here is that however a variable is operationally defined, it must be made explicitly clear. People reading about the research study must be able to judge how good or poor the operational definition