As part of our new project on border-mapping, I’ve started poking through the database of traffic stops by North Carolina police agencies in 2006 (provided by the thoroughly-helpful North Carolina SBI Crime Statistics Unit). Our eventual goal is to track how 287(g) enforcement, immigrant populations, and traffic stops interact and vary geographically across the state but in order to get into the data I first wanted to answer some broader questions about how race, ethnicity, and age influence the rate at which folks are stopped by the police in this state.
The data was provided to us in text-delimited format; I dumped it into a database using the highly-flexible (and recommended) sqlite3 database engine. For each stop, we have access to information about the age and perceived race, ethnicity, and gender of the driver (in addition to stop location, agency, time, etc.):
–next step: to the spreadsheets!–
First, I grouped the data by age, by perceived ethnicity (hispanic and non-hispanic) and by perceived race (white and non-white, since those are the categories which correspond with the 2006 state population-by-age data I had). Then, I grouped distinct ages into age classes to match up with the age classes used by the census, from 15 and 16-17 through 85-94 and 95+, so that I could calculate the number of stops of folks in each age and race or ethnicity class as a percentage of the total state population belonging to that class. The result would be more-or-less a likelihood of being stopped by the police if your age is such and you are (white/non-white/hispanic/non-hispanic). Following another analysis I also grabbed census data on 2006 NC car ownership-per-person, and used that to rescale the statewide population counts to get an estimated count of drivers statewide (note that this would not be as accurate as using drivers license data, if anyone has a lead on getting it).
The resulting graph is pretty clear, but there are still lots of questions to ask (click for a bigger view)…