Please click HERE to see my dissertation!
My first question is whether black and white criminal offenders are equally likely to be arrested. To visually assess agencies' treatment of black and white offenders, I plot the percentage of black offenders arrested by the percentage of white offenders arrested in each agency. The x-axis is the percentage of white offenders arrested, and the y-axis is the percentage of black offenders arrested. The red line shows where these percentages are equal, so if most agencies fall above this line, then agencies usually arrest a larger proportion of black than white offenders.
It appears that within most police agencies, black offenders are not more likely than white offenders to be arrested. (Please note that I have done many additional analyses to confirm this finding.) In addition, there is considerable variation in arrest rates across agencies, and black and white arrest rates often differ greatly within agencies.
Another question I ask is how racially diverse police agencies compare to agencies with relatively more white officers. Are black and white criminal offenders equally likely to be arrested across all agencies, regardless of their racial demographics?
I find that police agencies with relatively more white officers are not more likely to arrest black rather than white offenders. The figure below shows agencies' percentages of black and white offenders arrested by the percentage of their officers who are white. The orange and purple dots represent the percentage of black and white offenders who are eventually arrested, respectively, and lines are locally weighted regressions fitted to the points.
In general, black offenders are less likely than white offenders to be arrested. In addition, for some offenses, police agencies with relatively more white officers arrest a greater percentage of offenders. However, police agencies with relatively more white officers are not relatively more likely to arrest black, rather than white, offenders (the orange line does not become relatively more steep as agencies become less diverse). This is further confirmed by logistic regression and multilevel models predicting offenders' likelihood of arrest, which show no significant interaction of police agencies' racial diversity and offender race on the likelihood of arrest.
Please note that the figure above shows individual-level percentages, while the figure on the left shows agency-level percentages. For a comparison of arrest rates at the individual and agency levels, please see this plot, which shows that for certain offenses, black offenders are less likely to be arrested than white offenders at the individual level but have comparable arrest rates at the agency level. This suggests that agencies with fewer offenders are relatively less likely to arrest black offenders.