• Is there evidence of (racial) inequity in the amount of traffic stops by police?
• Attach this analysis with broader demographics in each state/region - how likely is a traffic stop for a random person based on demographics? For what reason?
• Working paper: https://5harad.com/papers/traffic-stops.pdf
• Data source: https://openpolicing.stanford.edu/data/
• Visualization Inspiration: https://www.nytimes.com/interactive/2018/03/19/upshot/race-class-white-and-black-men.html

• More than 60 million (63.7) state patrol stops
• 20 States
• Between 2011-2015

1. Quantify racial disparities in stop rates and post-stop outcomes Black drivers stopped more often than White relative to driving-age population, but Hispanic drivers stopped less often than White... Black and Hispanic drivers more likely to be ticketed, searched, and arrested

2. Investigate the degree to which these differences may result from discrimination - focusing on search decisions Bar for searching Black and Hispanic drivers is lower than searching White drivers

3. Effects of drug policy on stop outcomes Legalization of recreational weed in WA and CO reduced search and misdemeanor rates across the board, though relative race gap still exists

Aggregated cleaned data set

• stop_rate: "the rate at which white, black, and Hispanic drivers are stopped, relative to their share of the driving-age population" (p. 4). driving-age population is readily available, accurately estimated by US Census bureau, doesn't take into account race-specific differences in driving behavior (I am ok with this) (Figure 2, pg. 4) ⋅⋅* https://github.com/5harad/openpolicing/blob/master/results/data_for_figures/fig2_stop_rate_county.csv.

• speeding citation rate?: "the probability a driver stopped for speeding is given a citation as opposed to a warning (or no penalty at all) - (p. 6)"

• search_rate: Likelihood of getting searched for drugs, weapons, contraband after you're stopped. 2% white, 3.5% black, 3.8% hispanic (p. 6)

• consent searches: officers must seek permission from drivers to search their vehicles - probable cause searches don't need consent but have a higher standard of evidence.

• arrest rates: % of stops resulting in the arrest of the drivers (p. 7)

• hit rate: The proportion of searches that successfully turn up contraband (p. 8) - nine states include race, location of stop, search, and contraband found fields. Hispanic hit rates lower than whites (22%), white=black rates (28%) (a lower hit rate means more discrimination, since this suggests double standard of searching more on less evidence)

• threshold test: In effect, what is the "threshold" of suspiciousness that an officer has when evaluating whether to search a driver. A higher threshold = more trust, so a lower threshold shows more bias, which will require some though on how to actually visualize this in the context of more drivers - maybe using that threshold.