For many liberals, it is simply intuitive that the police stop more black motorists than white. Conservatives tend to dismiss this intuition as mere liberal bias. Intuitions and ideology are not a good source of information about this matter, so it is fortunate that research has been done on this matter.
When considering the data from the 2015 Stanford study, it is important to note that only about half of the states responded to the request for data and many of the states that did respond, did not track race. Despite these problems, the study does provide useful data—if its limits are kept in mind. While doubt about the accuracy of the study can be reasonable, to take its limits as evidence for denying its findings would be an error in logic. After all, even if a study is flawed, it does not follow that the denial of its conclusion is thus true.
As liberals predicted, in 17 states where data is available, black motorists are 1.2 (Connecticut) to 2.8 (Cook County, Illinois) times more likely than white motorists to be pulled over. Interestingly, there are 3 states (for which there is data) that whites are more likely to be stopped than blacks. These states are Texas (1.2 times more likely), Colorado (1.3) and Tennessee (1.4). This evidence supports the claim that blacks are, in general, more likely to be stopped than whites. A critical question is, of course, why this disparity exists.
One stock explanation is that blacks commit more crimes than whites and hence they are stopped, searched and arrested more. However, the data shows that blacks are no more likely than whites to have contraband in their possession when stopped. Also, there is the standard problem with the usual argument used here: if one argues that blacks are arrested more because they commit more crime and the evidence is that they are arrested more, then one has run the argument in a circle. This is a fallacy known as begging the question or circular reasoning in which one assumes as true what needs to be proven. As such, the claim that blacks commit more crimes than whites does not explain the disparity.
Another stock explanation is racism—the police discriminate against minority drivers. A study in Connecticut used an interesting approach to assess the role of race in stops: it compares day and night stops. It was found that Hispanic drivers were stopped 2.3 times more often during the day than at night. The reasonable explanation for the disparity is that the police could see the driver better during the day than at night, thus making race more of a factor during the day. If stops were not influenced by the race of the driver, the day and night statistics should have been the same. There do not seem to be plausible alternatives, unless one wants to claim that Hispanics drive better at night.
This study also found that when white drivers were stopped and searched, 42.9% of them had contraband. Blacks who were stopped had contraband only 8.3% of the time. While this could be taken as evidence of racism, there is a reasonable alternative explanation: blacks know that they are likely to be targeted by the police, so they are less likely to carry contraband. Whites know that they are not a prime target, so they are more likely to risk carrying contraband. Of course, this explanation serves as indirect evidence of possible racism since it is based on the claim that people know how the police target their stops and searches.
A third explanation is to claim that the results are due to chance—it just so happens that in 17 states for which there is data blacks are 1.2-2.8 times more likely to be stopped. Proponents of this view will point out that, as noted above, in 3 states whites are 1.2-1.4 times more likely to be stopped than blacks—and surely this is not because the police in Texas, Tennessee and Colorado are racist towards whites. Some states, they would say, stop more blacks. Some stop more whites. So, there is clearly no racism.
This argument rests, of course, on the fact that disparities can be due to chance and one can always raise that possibility. The challenge is, obviously, sorting out whether the difference is due to chance or some other factor, such as racism. Without getting into the technical details, the difference between chance and causal connections is sorted out by considering the difference in the results, the size of the study population, and the methodology of the study. To us a simple example, imagine someone wanted to determine whether their 20-sided die (a d20 in gamer jargon) rolled randomly or had a flaw that caused it to roll certain numbers more often. If the person rolled it 20 times and rolled some numbers twice (or more) and some not at all, this could still be due to chance rather than a flaw in the die. After all, a small sample size will tend to not be very representative. As the person rolls more and more, the data gets better and better—as the sample size increases, the more likely it is that a flaw would be revealed. So, rolling 20 times without rolling a 20 would be unlikely, but not unexpected. Rolling 200 times without a single 20 would not be impossible but would be good evidence that there is something wrong with the die. After all, as the number of rolls increase, the results should match the statistical expectation of one 20 for every 20 rolls.
Likewise, a disparity between whites and blacks over a small number of stops could be purely a matter of chance; but as the numbers increase, a persistent disparity would be evidence of some other factor at play, such as racism. Given the size of the study, it would be unreasonable to attribute the results to chance and bias would be a likely factor. It should not be concluded that all or even most police are racist. It has been claimed that everyone is affected by their implicit biases, so it is certainly possible that this accounts for some of the disparity in stops. That is, some officers are not explicitly racist, but their implicit bias leads them to believe that minority motorists are more likely to be breaking the law. Of course, racism should not be dismissed entirely and is likely to be a factor in some cases.
While the disparity in stops is a problem, the fact that some departments are modifying their training to address the matter is encouraging. Assuming, of course, that such training works.