Understanding Racial Tensions in Stop & Frisk
Understanding the Racial Tensions in NYC's Stop and Frisk Policy
Despite what you might hope for or want to believe, we are not yet living in a post-racial society. We do not have a color-blind meritocracy. While levels of explicit bias have neared zero since the 1990s, implicit racial bias still exists: the majority of people who take the race implicit association task (IAT) show moderate to strong pro-white bias. While most people are no longer willing to publicly agree to the statement "black people are lazy," their unconscious actions, attitudes, and body language tend to say otherwise.
We've seen implicit bias play out in a laboratory environment. People are quicker to identify weapons when primed with black faces. In a laboratory game, both college students and police officers mistakenly shoot unarmed more black suspects than white suspects, and they do so faster and with greater accuracy. But with the recent shootings of Trayvon Martin, Michael Brown, and other young black men, there's been a renewed focus on how implicit biases affect law enforcement in the real world. To understand the potential effects of racial biases in policing, we turn to the data gathered by the NYPD during Stop and Frisk.
Thanks to an ACLU lawsuit, the NYPD released 10 years worth of data from Stop and Frisk (2003-2013). That data can be found here. Officers filled out UF250s for every stop they made, and on these, they recorded the date, sex, age, race, whether the subject was searched or frisked, if anything was found on them, whether police force was used, and more. Dan Nguyen, a Stanford Data Journalism lecturer, recently downloaded and cleaned the data and pushed it to a public repo on Google BigQuery, making the data easy to subset and test. The data tells a complicated story of the racial tensions between police officers and NYC residents:
To start, a disproportionate amount of stops are of blacks and hispanics:
This trend persists over time:
On average, your risk of being stopped as a black person is nearly 7x that of a white person. And, if you compare rates of stops for black and white males between the ages of 14 and 24, there's a similar trend:
Black people were more likely to be frisked, but searches on white people were more likely to turn up contraband and weapons:
There isn't much evidence that whites carry more weapons than minorities, so this could indicate that police might just be more accurate at picking out which white people commit crimes. If they're worse at guessing which black or hispanic people are committing crimes, they just have to cast a wider net and search more people. What's the problem with that? Well, for one, police are more likely to use force with a black or hispanic person than a white or asian, for almost every single self-reported measure of force:
But not only are the police more likely to use force with minorities, they tend to use more force with blacks and hispanics:
When you look at the likelihood of getting arrested after being stopped, there's little difference across racial categories:
That might strike you as strange given that white people were more likely to be found with contraband and weapons. This might suggest that police have a higher threshold for arresting whites. Perhaps minorities are more likely to get arrested than whites for smaller infractions.
It's important to remember that the actions of both parties matter. Police might have implicitly biased attitudes and expect more crime or negativity out of blacks. But blacks might have biased attitudes towards the police that hurt that relationship. (Yale Law School Professor Tom Tyler, for example, argues that trust between the public and the police is massively important for having compliance with the law). It's hard to measure trust between the public and police, but I picked out one measure from this data:
When a police officer stops someone, they ask for ID. On the UF250, that ID is recorded as verbal, photo, other, or refused. As a proxy for police distrust, I looked at the rate of refusing ID:
If you look at the age demographics of refusing to give ID, young people (between 18 and 30) refuse ID much more often, and young black males are much more likely to refuse to give ID.
What came first, the black distrust of police or the police distrust of blacks? It's hard to say. But the correlation between the force and distrust is moderate to strong (~.6):
As Andrew Gelman, a statistician at Columbia, put it, "In the context of some difficult relations between the police and ethnic minority communities in New York City, it is useful to have some quantitative sense of the issues in dispute." While Stop and Frisk ended, the implicit racial biases underlying the day-to-day mentality have not, and it will be interesting to see how these issues play out in future policies.