Differential Suspicion: Theory Specification and Gender Effects in the Traffic Stop Context

Michael Smith Matthew Makarios and Geoffrey AlpertJustice QuarterlyVol. 23 No. 22006pp. 271-295,

Smith et al. attempt to develop a theory of racial profiling using data from Miami-Dade County Florida police department. The authors examine more than 66,000 traffic stops from April to October 2001. The focus of the theory is gender because gender has been a key factor in past police decision making research. The authors note that their data does contain 9/11 and that could be a possible history effect against their internal validity, but the race and gender of those stopped remains constant both before and after 9/11. The theory draws from the literature on police suspicion and script theory. Smith et al.explain that police suspicion is a corner stone of the police subculture and is part of their job, and script theory explains the development of cognitive rules used to evaluate unfamiliar situations.

The theory says that police have a war on crime mentality and dealing with criminals on a continual basis may cause suspicion scripts to be formed about all members of the group in the larger society. The authors note that the police work itself may be the necessary condition for the formation of these group stereotypes. Smith et al. present three hypotheses:

• 1.

  police will have a greater suspicion of males than females;

• 2.

  there will be a positive correlation between suspicion and likelihood of arrest; and

• 3.

  suspicion will mitigate the impact of gender on the probability of arrest.

Smith et al. (2006) used three indicators of officer suspicion:

• 1.

  filling out a field interview (FI) card;

• 2.

  record check on a person; and

• 3.

  record check on a vehicle.

These were the criteria used because they represent more than just a normal stop and were relatively uncommon. Logistic regression was used, with gender as the independent variable of interest. Age, race, whether the stop was at night, the age of the vehicle, whether the stop was for a moving violation,whether the stop was for an equipment violation, and census tract were entered into the model as controls. To test the second hypothesis, the authors used two logistic regression models with arrest as the dependent variable. In the first model they do not include the suspicion variables with gender and they do include suspicion variables in the second model.

Smith et al. found that police were significantly more suspicious of men compared to women across all three of suspicion indicators, but that the FI card was the strongest predictor of suspicion. Police were also more suspicious of younger and nighttime drivers, and much more suspicious of minority drivers compared to whites. Both moving violation and equipment violation variables were weaker than the investigative-based stop variable, which indicates that police are much more likely to request a record check or do a FI when investigating something.

Support was also found for the second hypothesis, which pertains to the correlation between suspicion and the likelihood of arrest. Smith et al.found that when the suspicion variables were entered into the model the probability of arrest variable increased. Suspicion variables were the second strongest predictor of the likelihood of arrest, the first was the presence of contraband. Suspicion variables also reduced the coefficient for the contraband variable as well. The third hypothesis, that suspicion variables would mitigate the effect of gender, was also supported, although the support was weaker than the other two hypotheses. Smith et al. found that the introduction of suspicion variables into the model reduced the coefficient for gender slightly. Smith et al. also note that the introduction of the suspicion variables turned the race variable to insignificance in the second model; the black driver variable was significant in the first model. While these results are promising for the theory, the authors do note that some important correlates of arrest were missing, specifically, the presence of bystanders, demeanor of the suspect,social class, and evidence strength, and this means that further examination of this topic is necessary.