The Florida State patrol stop data from Stanford University’s Open Policing project contains an abundance of categorical variables with information about officers and subjects in unique traffic stop interactions. The primary focus of these variables correspond to demographic information, such as the sex, race and age of the people involved in the stop. The dataset also includes location data on a county-level for each stop, as well as a timestamp relating to when the stop occurred. Also included are indicator variables for whether an arrest was made, a citation issued, a warning issues, a frisk performed, and/or a search conducted. Lastly, we are given a brief comment about the reason for the initial stop and the final violation, if applicable.

Our interest in exploring the impact of a mass shooting with respect to policing behaviors led us to pose the following motivating question: Was there a change in traffic stops in Orange County, Florida after the Pulse Nightclub shooting on June 12, 2016? To answer this overarching question, we categorized our analysis into the following three questions:

In first exploring the data, we wanted to understand whether the subjects and officers involved in police stops in Orange County were roughly demographically equivalent before and after the Pulse mass shooting. We constructed boxplots examining the distribution of officers’ ages for each officers’ rate and found no substantial difference in these demographics before and after the Pulse shooting.

We then created mosaic plots of the subjects’ races against officers’ races. As a result, we found the proportions of police stops by officers’ and subjects’ races appear to not change substantially from before to after the Pulse shooting.

Finally, we constructed bar graphs of search rates by subject race. For both Pre-Pulse and Post-Pulse stops, the ordering of subjects’ races by officers’ search rates does not change. While post-Pulse stops have overall higher search rates for each race, each search rate is small enough that the differences appear minimal; thereby, we do not find sufficient evidence that search rates by subjects’ races differ from before to after the Pulse shooting.

To understand whether the number of police stops per day were roughly the same before and after the Pulse shooting, we modeled a 30-day moving average (accounting for noise) of the number of stops per day. We found that while there appeared to be an increase in the number of police stops during the first half of 2016, there was a steep decline a few months following the event. As a result, we cannot conclude there exists a relationship between the number of stops and the impact of the Pulse shooting.

Finally, we wanted to see if the Pulse Nightclub shooting had an impact on the raw number of stops performed on a county-level in Florida. To do this, we used data from the year leading up to the shooting and the year after the shooting and created two Florida state maps colored by the number of stops in each county. Our resulting maps suggest that there was no significant change in the number of traffic stops conducted after the Pulse shooting as compared to the year prior.

We find much evidence that the spatial and demographic distributions of traffic stops in Orange County, FL are roughly equivalent from before to after the Orlando nightclub shooting. However, the shooting’s potential impact on traffic stops is unclear: a time-series plot of the stops in Orange County does not show a prevailing trend, even around the date of the shooting. Future researchers can thus build upon this analysis by trying to infer causality using econometric methods. We suggest they consider a 2019 paper which uses Stanford’s open policing data to determine the effects of recreational marijuana legalization on policing in Colorado and Washington [1]. Perhaps a similar framework to this could lead one to determine whether the Pulse shooting truly affected policing in Florida. Racial bias in officers’ stops is one possible outcome of interest since 90 percent of victims were Hispanic [2].

[1] https://5harad.com/papers/100M-stops.pdf

[2] https://www.theguardian.com/us-news/2016/jun/14/latino-hispanic-orlando-shooting-victims