Abstract
Drawing on script theory and related research from cognitive social psychology, this paper suggests that police may develop unconscious, cognitive schemas that make them more likely to be suspicious of population subgroups that they repeatedly encounter in street‐level situations involving crime and violence. Drawing on more than 66,000 traffic stop records from the Miami‐Dade County Police Department, this article presents an initial test of this theory using gender as the principal variable of interest. Police were found to be significantly more suspicious of men than of women in traffic‐stop encounters, and suspicion was strongly associated with the decision to arrest. Consistent with the specified theory, suspicion had a modest, attenuating affect on the relationship between gender and arrest. However, gender remained a statistically significant predictor of arrest even after controlling for suspicion, suggesting that other factors associated with gender continue to operate and drive police decision‐making.
Notes
1. To illustrate their concept of a script, Schank and Abelson (Citation1977) describe a visit to a restaurant. They argue that we know what is likely to occur in a restaurant because we have developed restaurant scripts. For example, without really thinking about it, we know that we will be shown to a table, that we will be given menus, that a waiter will take our order, and that he or she will bring us our food when it is ready. Although the details of any visit to a restaurant will vary, the restaurant script will remain fairly stable across situations.
2. We note that the data‐collection period spanned the 9/11/2001 terrorist attacks in New York and Washington, DC. A before‐and‐after analysis revealed that although stops dropped off slightly during October 2001, the racial and gender makeup of drivers detained remained consistent.
3. We agree with one of the reviewers of our paper that community context can play an important role in police decision‐making (see Smith, Citation1986). We note only that our analysis was theoretically driven, rather than exploratory, which mitigated against the inclusion of community‐level variables in the models. For the same reasons, we, like others who have examined police discretion (Smith et al., Citation1984; Visher, Citation1983), also did not include officer characteristics in our analysis. Moreover, we did not control for officer characteristics in the models (like we did for census tracts) because no single officer accounted for more than 1 percent of the total number of stops, and the sheer number of officers represented in the data (more than 1,600) suggested that officer characteristics were not true level 2 variables (see Raudenbush & Bryk, Citation2002). We confirmed this by re‐running the models without the high‐stop officers included (greater than 2 standard deviations away from the mean), which produced no substantive changes to our findings.
4. In the interest of brevity, we do not report the 257 census tract‐based dummy variables in the tables below. The full results of the regression analyses are available from the authors upon request.
5. Illusory correlation refers to the cognitive tendency to overestimate infrequent or otherwise unique events that occur among small (e.g., minority) groups.