Using Experiments to Test Alternative Hypotheses

Using Experiments to Test Alternative Hypotheses Once researchers identify an interesting phenomenon, such as the tendency for people to overestimate the frequency of impactful events (e.g., homicide) and to underestimate the frequency of nonnewsworthy events (e.g., death by diabetes), they seek to learn why the phenomenon occurs. As we discussed in Chapter 1, just identifying and describing a phenomenon is not enough. If we want to know the circumstances under which people make such errors, or if we want to be able to decrease their tendency to do so, we need to know how people come to make the error in the first place. In many instances, there may be multiple plausible hypotheses, and the challenge for researchers is to design studies that tease apart these alternatives. So put on your researcher-as-detective hat and think for a moment about how you would explore the two hypotheses for why people overestimate the frequency of impactful events relative to more pallid events. Is it because (1) people actually remember more impactful events (Hypothesis 1) or because (2) it seems easier to remember impactful events (Hypothesis 2)? Where would you start? First, you would want to work through the implications of the two hypotheses. What kinds of processes or outcomes do they predict? Second, you want to find someplace where they make different predictions. If both hypotheses make predictions about a certain outcome and these predictions differ, then this is a place where a welldesigned experiment might be able to tease the two hypotheses apart. For the case of estimating event frequencies, both hypotheses imply that people overestimate the frequency of impactful events and underestimate the frequency of pallid events. So this doesn’t help. Both hypotheses also predict that people try to recall previous instances of the event when trying to estimate frequencies of such events, so this doesn’t help either. The hypotheses do differ, however, in their implications for two things: First, Hypothesis 1 states that the amount of relevant information recalled is crucial, but it also suggests that the felt ease of recalling the information would be irrelevant. Hypothesis 2, on the other hand, suggests that the actual amount of information recalled should be relatively unimportant, but it states that the felt ease of recall should be crucial. The two hypotheses, then, make opposing predictions on the roles of actual memory and felt ease of memory. This brings us to the third step: We must now design a study that pits these opposing predictions against one another. Consider the reasoning of Norbert Schwarz and his colleagues (Schwarz et al., 1991). They decided to create two experimental conditions, one in which participants would successfully recall many instances but would experience doing so as difficult (Condition A), and a second in which participants would recall fewer instances but would think it easy to do so (Condition B). Why these conditions? If the actual number of recalled events is most important, participants in Condition A should estimate a higher homicide rate than participants in Condition B. In contrast, if the felt ease of recall is most important, participants in Condition B should estimate a higher homicide rate than participants in Condition A. Having established the logic of the experiment, the final step is to create such differentiating conditions. These researchers were interested not in frequency estimates of murders and diabetes, but rather in how the availability heuristic might influence judgments people make of their own personalities. Through pretesting, they discovered that people can easily recall 8 or 9 examples of their own assertive and unassertive behaviors, but that it gets increasingly difficult when people try to remember more than 10. When you ask people, then, to describe 6 examples of assertive behaviors, it will feel easy to them. In contrast, when you ask people to describe 12 such behaviors, it will feel difficult to them (although they can eventually do so). This simple finding sets the stage for a clean test of the two hypotheses. If the number of instances actually recalled determines people’s frequency estimates, people who describe 12 instances should judge themselves to be more assertive than people who are asked to describe only 6 instances. Alternatively, if the feeling of how easy it is Focus On Methods CCONTTEENTTSS IINDEEXX HEELLPP 86 Chapter 3 Social Cognition: Understanding Ourselves and Others to recall events determines frequency estimates, then people who are asked to describe 6 assertive events should judge themselves as more assertive than people who are asked to describe 12 such events. As Figure 3.3 reveals, there is a winner: the felt ease of recall appears to underlie the availability heuristic. We see, then, that a carefully done experiment can go a long way toward differentiating among alternative plausible hypotheses. Indeed, throughout this textbook, we’ll encounter many studies that do just this.