Did public scrutiny blunt Boston’s stop-and-frisk program?



For this analysis, I used data from Boston’s open data portal to examine trends in Field Interrogation and Observation (FIO) reports.

Did public scrutiny blunt Boston’s stop-and-frisk program?

On a cold December night in 2011, a Boston cop named Luis Anjos spotted two young Black men on Martin Luther King Boulevard in the city’s Roxbury section. He had a hunch they might be suspects in a break-in he’d been investigating, so he drove up to them and yelled out for them to “wait a minute.” They turned and jogged into a nearby park.

Two cops, alerted by Officer Anjos, later approached the men. When one of the officers said, “Hey fellas,” the men fled again. One of them, Jimmy Warren, was carrying a Walther .22 pistol when Boston police arrested him. He’d had nothing to do with the break-in Officer Anjos was investigating but was convicted of unlawful possession. 

Controversially, police can stop you on the street if they’ve got “reasonable suspicion” that you’re involved in a crime. That’s a lower bar than the “probable cause” normally required for them to search or detain you.

New York City famously built a policing strategy around aggressively increasing police-civilian encounters; Rudy Giuliani takes credit for this “stop-and-frisk” approach, even though it actually didn’t reduce crime and was later ruled unconstitutional there. Perhaps more importantly, President Trump has called to expand stop-and-frisk and seems fixated on reducing crime in “the inner city.”

For Jimmy Warren, justice came about four years later. The Massachusetts Supreme Judicial Court threw out his conviction last September, ruling that the police had confronted him without reasonable suspicion: hunches don't cut it, and Officer Anjos's was an especially weak one. The SJC—including my former boss, Judge Fernande Duffly—noted specifically that Warren's flight wasn’t evidence of guilt: instead, a young Black man “might just as easily be motivated by the desire to avoid the recurring indignity of being racially profiled as by the desire to hide criminal activity.”

The SJC relied on a 2014 ACLU report finding that Boston cops disproportionately targeted young Black men through the Field Interrogation & Observation (FIO) program—Boston’s version of stop-and-frisk. Even controlling for higher rates of gang membership and crime history (factors intended to prompt more FIOs), the report found an additional disparity in the rates at which Black men were stopped between 2007 and 2010.

In response, BPD said its FIO rates had declined substantially and that FIOs were less racially disparate. After the department released records from January 2011 to mid-2015, in response to an ACLU request, researchers and journalists questioned the validity of BPD’s figures.

I wondered how the public controversy over stop-and-frisk might have affected FIO rates in Boston, even before the 2016 Warren decision. The results of the ACLU report were released in October 2014, and while the policy had faced scrutiny earlier, the report’s findings added significant weight to claims of disparate treatment. It’s pretty easy to see how reported FIOs started to fall off after the ACLU report:

Post-ACLU report FIO trendWe can see how 2,000 stops per month, which had been around the lowest monthly rate from 2011-2014, became the new high point in 2015.

These raw findings could be deceptive, since the general trend over 2011-2014 was slightly downward and it’s hard to interpret always-low winter FIO rates. So I fit a regression model to the FIO data and used it to estimate what the FIO rates would have been after October 2014, if the ACLU report hadn’t had any effect:

Boston FIOs with pre-trendI included the red line’s path over earlier years so the model's fit is visible. It’s a basic linear regression with fixed intercepts for each month (Jan-Dec), year and the pre/post-report “treatment effect,” plus race and its interaction with pre/post.

This model helps us see the change in FIO reporting between October 2014 and May 2015. (I dropped the two weeks’ data included from June 2015 since I wanted to run a simple by-month analysis.) On this model, citywide stops were down by about 25-40% each month between December and May, except February, when they were down by 64%, compared to the prior trend.

Changes in crime don’t appear to be a major confounder: reported crime incidents through the post-report period stayed within pretty normal ranges, and adding it as a control variable doesn’t significantly improve fit.

Crime report trends BostonThese are just preliminary analyses from incomplete data and they don’t show causation. Still, the hypothesis deserves further testing. If true, officers might have turned to other methods than stopping individuals on the street; more cynically, they could mean officers found ways not to write them down.

Either way, they suggest how ratcheted-up public pressure can influence police practice.

Thanks to ACLU-MA/Harvard's Paola Villareal for pointing me to these data! Check out her visualizations of the Boston FIO data here: http://warondrugs.justicesos.org/.