Racial Bias in Police Stops and Searches

Leftists often argue that police disproportionately stop and search black people and that this is evidence of racial bias in our criminal justice system and our society more broadly. In this post I will argue that this conclusion is not justified by the available evidence.

New York Pedestrian Stops

The research most often cited by the left involves “hit rates” or the proportion of those stopped found to be engaging in criminal activity. In a racially biased system the hit rate will be higher for whites. In other words, among those searched, white people will have higher rates of having committed a crime.

The reason for this involves how high the police require the probability of someone being a criminal to be in order to justify stopping or searching them. For instance, if police only search whites when there is evidence that makes it at least 60% likely that they are committing a crime, but requires a lower bar of evidence to search blacks and so will search them if there is only a 40% probability of them committing a crime, then the “hit rate” will be .60 for whites and .40 for blacks. Thus, a higher hit rate for whites implies that police are requiring a higher bar of evidence to search whites than to search blacks.

Anyway, the left seems particularly preoccupied with pedestrian stops in New York City. According to an article published by Vox which looked at data on New York’s stop and frisk policy, only 1% of stopped blacks are found to be carrying a weapon or contraband compared to 1.4% of stopped whites. Thus, the white hit rate was 40% greater than the black hit rate. 

Lopez (2016)

The left likes to concentrate on Stock and Frisk searches and that is likely because pedestrian stops taken as a whole display less evidence of bias.

Ridgeway (2007) analyzed data on all pedestrian stops the took place within New York City between 2005 and 2006. This included “stop and frisk” encounters as well as others. The paper found that, among those who were stopped, 5.1% of whites and 4.8% of non-whites were arrested. This suggests that hit rate for whites was only 6% greater than the non-white hit rate once all police encounters were taken into account.

The paper also found that “black pedestrians were stopped at a rate that is 20 to 30 percent lower than their representation in crime-suspect descriptions.” On the other hand, “Hispanic pedestrians were stopped disproportionately more, by 5 to 10 percent, than their representation among crime-suspect descriptions would predict.” This suggests that if any bias is taking place against non-whites, it is against Hispanics and not blacks.

A limitation of this research is that it does not control for variation in the stop and arrest rates of each precinct. Minorities tend to live in precincts with police who are more aggressive in their stop and arrest patterns regardless of race. Once you control for this variation, racial bias in hit rates disappears.

For instance, Coveillo et al. (2015) analyzed data on 2,947,865 police stops in New York City from the years 2003 to 2012 and they found that “after accounting for the fact that different precincts have different baseline rates of arrest conditional on search, African Americans are no longer less likely to be arrested conditional on being stopped”. The same was true of Hispanics.

Thus, a close look at the data suggests that the NYPD are not stopping pedestrians in a racially biased way.

Car Stops and Searches

Of course, in most areas of the country people are stopped by police when they are driving, not while walking around. Perisco and Todd (2006) summarized 15 studies on the hit rate for car searches in various parts in the US.

Perisco and Todd (2006) 

There was a great deal of variation across locations, but, on average, the white hit rate was 15% higher than the black hit rate and 47% higher than the Hispanic hit rate. In absolute terms, the gap between whites and blacks was 2.45 points while the gap between Hispanics and whites was 6.9 points.

This lends plausibility to the idea that there is a small bias against blacks and a moderate one against Hispanics. It does seem that police require less evidence of wrong doing to pull over minorities.

That being said, a few things are worth noting. First, the bias for blacks is small and cannot account for the most of why it is that black people are pulled over more often than white people.

Second, if police are slightly over-estimating the criminality of black people, it does not follow that this over-estimation is based directly on their race as opposed to some other variable that correlates with race. Actually, there are a few reasons to think that race is not the variable directly responsible for this.

First, Grogger and Ridgeway (2006) analyzed data on 7,607 police stops that took place in Oakland, CA and found that black people accounted for a greater proportion of those pulled over during night hours when the race of the driver would not be visible to police until the driver was pulled over.

This suggests that police bias in police stops may be due to differences in the sorts of cars black and white people tend to drive, or the way in which they tend to drive them.

A similar conclusion is suggested by research on black police officers. For instance, Smith et al. (2001) analyzed data on 2,673 traffic stops that took place in Richmond, Virginia. As shown in table 3, the probability that someone being pulled over was black did not differ between white and non-white police officers.

Similarly, Baumgartner et al. (2018) analyzed data on 164,322 police stops that took place in Charlotte, NC and found that anti-black bias in police stops was greatest among black police officers.

So, again, it could be something about the cars black people drive, or how they are driving them, that accounts for black people being disproportionately pulled over. It could also be that black people live in more criminal areas that are policed more aggressively in-order to push the general crime rate down to an acceptable level. As we saw, this was true of pedestrian stops and, unfortunately, I haven’t been able to find a study on pull over rates that addresses this issue.

Conclusion

The bias found in hit rate studies is exhibited by white police officers even when they probably can’t tell the race of drivers until they pull them over, and it is exhibited to an equal or greater degree by black police officers. Given this, it seems pretty unlikely that police are using a lesser evidentiary standard when pulling over black people because they are black. Granted, black people are being treated differently than white people. But it isn’t clear that this is because they are black, and it is not clear that this is even a bad thing if this reflects differences in how high and low crime areas are policed. It certainly does doesn’t constitute compelling evidence for racism and, as I’ve argued elsewhere, the assumption about racist behavior among whites should be that none exists unless evidence shows otherwise. In this case, it does not.