Home :: Research Focus Boxes :: Research Issues :: The Dependent Variable

The Dependent Variable: When Is a Stereotype Simply a Good Guess, and When Is It More?

We all know that it’s wrong to react to others on the basis of stereotypes. Yet how often do we do so? Stereotypes lead us to expect one type of behavior or another in individuals, based simply on their group membership. Adolescents live in a world of well-defined social groups, each of which is associated with characteristic behaviors. “Jocks,” for instance, are thought of as noisy and rowdy, “preppies” as well-to-do and college bound, and “techies” as spending more time with their computers than with peers. How likely are adolescents to make judgments about the behavior of other adolescents, in the absence of any other information, based on stereotypes such as these?

Stacy Horn, Melanie Killen, and Charles Stangor (1999) wanted to know as well. These investigators were interested in the way high school students would evaluate the appropriateness of punishing other students, in the absence of supporting evidence, when the actions of which they were accused were either consistent or inconsistent with stereotypes for their groups. Would adolescents’ decision making, in other words, be affected by commonly held stereotypes about members of other groups?

Stacy Horn and her associates asked ninth-graders to read a scenario in which some students had too much to drink at a dance and committed an act of vandalism. The students subsequently were told by the student council that they must pay for the damages despite the absence of evidence indicating they were responsible. The students were described either as “jocks” (football players) or as “techies” (computer club members), and the vandalism they committed was either damaging the sound equipment at the party or breaking into the school computer system. Thus, the action could be consistent with students’ stereotypes (e.g., computer club members who broke into the computer system) or inconsistent (e.g., football players who broke into the computer system). After reading the scenario, students were asked to evaluate the action of the student council, to give the reasons on which they based their evaluation, and then to indicate whether they believed the student council’s behavior was justified.

Why did these investigators use more than one measure of stereotyping? If the measures don’t all show the same relationships, how are we to evaluate which one is more accurate? The answer is that different measures pick up different aspects of behavior. Three criteria distinguish accurate measures: reliability, validity, and sensitivity.

The first consideration with any measure of behavior, or dependent variable, is its reliability: It should give you the same value each time you use it. If a student takes an intelligence test, for instance, and retakes it in three weeks, one expects the score to be about the same on both occasions. Differences in IQ from one testing to the next reflect factors other than intelligence—that is, error. Reliable measures have little error. Second, measures must have validity: They must measure what they are designed to measure. Some of the very first intelligence tests were highly reliable but not very valid. Some, for instance, measured how rapidly people could tap their fingers, something that can be measured with little error but that turns out to have little to do with actual intelligence. Third, sensitivity is a characteristic of good measures: They are able to detect even small differences where these exist. Current measures of intelligence do more than sort individuals into categories of, say, bright, average, and dull. They offer numerous distinctions within each.

Returning to adolescents’ stereotypes and their decision making, let’s consider what Horn and her associates found. They discovered that adolescents evaluated the actions of the student council as wrong, indicating that it would be unfair to punish a group simply because the type of damage that had been done fit the stereotype of the group. However, they also found that the types of reasons adolescents brought to bear in supporting this position differed depending on whether the actions were consistent with their stereotypes of either group. When their stereotypes were confirmed, adolescents were less likely to bring moral arguments to bear, leading to fewer concerns about fairness in the absence of supporting evidence. Only by using several response measures could the investigators sort out these differing relationships.



S. Horn, M. Killen, & C. Stangor. (1999). The influences of group stereotypes on adolescents’ moral reasoning. Journal of Early Adolescence, 19: 98–113.