Fig 1.
a. Race differences in past year stop-and-frisk prevalence of past year stop-and-frisk* among P18 participants over cohort follow-up (2014–2019; N = 591 at baseline). *P18 questionnaire item: “How many times have you been stopped, questioned, and/or frisked in the past year?” We coded a dichotomous indicator of any prior stop-and-frisk in the past year. In the sample overall, 32.3% reported being stopped-and-frisk at baseline and 21.5% reported being stopped-and-frisked at least once over the course of the follow-up. b. Race differences in past year stop-and-frisk rate* among P18 participants over cohort follow-up (above bar chart) and unadjusted rate ratios (RRs) and 95% confidence intervals (CIs) for associations between race and stop-and-frisk rate (below table) (2014–2019; N = 591 at baseline). *P18 questionnaire item: “How many times have you been stopped, questioned, and/or frisked in the past year?” P18 rates were calculated as the race-specific number of stop-and-frisk events reported by P18 participants/race-specific P18 study population times 1,000. Note: Rate among Black P18 participants at the 36-month visit was driven by a high stop rate among one participant (66 stops in the past year); when this participant was excluded from the analysis the stop rate was reduced to 0.5 from 0.9 among the remaining participants.
Fig 2.
Stop-and-frisk rate per 1,000 population* reported by P18 participants vs recorded by the New York City Police Department (NYPD), by race.
*P18 rates were calculated as the race-specific number of stop-and-frisk events reported by P18 participants/race-specific P18 study population. These were compared to rates in the general population, calculated as the race-specific number of stop-and-frisk events recorded by the New York City Police Department/race-specific American Community Survey New York City population estimates. Note: Rate among Black P18 participants at the 36-month visit was driven by a high stop rate among one participant (66 stops in the past year); when this participant was excluded from the analysis the stop rate was reduced to 58.6 per 1,000.
Fig 3.
Geographic distribution of past year stop-and-frisk among P18 participants at cohort baseline, by race: P18 participants are mapped by participant ZIP code of residence.
Those with a history of stop-and-frisk are indicated by red circles (White race), green circles (Hispanic/Latinx race/ethnicity), blue circles (Black race) with increasing circle size indicating increasing number of stop-and-frisk events, while those with no stop-and-frisk history are indicated by Black circles.
Fig 4.
Race differences in past year stop-and-frisk rate among P18 participants at cohort baseline in 2014 by neighborhood-level stop-and-frisk rate (counts/1000)*.
*P18 participants were categorized according to their neighborhood-level stop-and-frisk rate, which were calculated as followed. Participant’s ZIP code was translated to latitude and longitude, which was then used to identify the New York City Police Department (NYPD) precinct in which the respondent lived (total of 77 possible precincts). NYPD precinct-level stop-and-frisk rates were calculated as the 2014 NYPD recorded number of stop-and-frisk events divided by the precinct population of male residents aged 20–29 years old; restricted given the younger age range of the P18 sample. The population estimates for New York City were calculated from 2014 estimates from the 2010–2014 5-year American Community Survey; population estimates for each census tract of New York City were calculated after restricting the sample to males aged 20–29 years. Each of the 77 precincts were categorized into group groups based on the precinct-level stop-and-frisk tertile. Within each tertile indicating neighborhood-level stop-and-frisk, stop-and-frisk rates were calculated at the race-specific number of stop-and-frisk events reported by P18 participants/race-specific P18 study population.
Table 1.
Odds ratios (ORs) and 95% confidence intervals (CIs) for associations between participant socio-demographic, psychosocial, and behavioral factors and past year stop-and-frisk at cohort baseline in 2014.