2/20 Traffic Stops 1 - Gelman, Fagan, & Kiss

Key Notes on the Article

Main Points:

  • The article analyzes racial disparities in police stop-and-frisk practices in NYC, using data from 125,000 pedestrian stops over 15 months.

  • Black and Hispanic individuals were stopped at higher rates than White individuals, even when controlling for local crime rates and precinct demographics.

  • The study does not definitively prove racial discrimination but shows patterns that suggest disparities in policing tactics.

  • Hit rates (arrests per stop) were lower for Black and Hispanic individuals, meaning police stopped them more frequently but with a lower likelihood of actual arrests.

  • The study raises questions about whether stop rates are justified by crime rates or if they reflect racial bias in police decision-making.

Interesting Discussion Points:

  • Neighborhood Policing and Racial Profiling – Police claimed that stops reflect crime rates, but the study suggests stop rates were disproportionately high in minority neighborhoods.

  • "Racial incongruity" stops – People were more likely to be stopped when their race didn’t match the neighborhood’s racial demographics (e.g., Black individuals in White neighborhoods and vice versa).

  • Police efficiency debate – Stops of White individuals resulted in more arrests, raising questions about whether the police used a lower threshold of suspicion for minorities.

  • Policy impact – The aggressive stop-and-frisk strategy was credited with reducing crime in NYC but also increased racial tensions and led to legal challenges.

Specific Cases & Evidence:

  • The study references the NYPD’s Street Crimes Unit, which was heavily criticized for racial profiling and later disbanded.

  • High-profile police brutality cases like Amadou Diallo (1999) and Abner Louima (1997) fueled public distrust in stop-and-frisk tactics.

  • Court rulings like Terry v. Ohio (1968) and Illinois v. Wardlow (2000) helped shape legal standards for stop-and-frisk.

  • Statistical models showed that even in high-crime areas, minorities were stopped more frequently than crime data would predict.

The article does touch on how the data used to justify stop-and-frisk might itself be influenced by biased decision-making. Specifically, it discusses how police stop rates are often defended by pointing to crime data, but those very stop rates might be shaped by where and whom police choose to stop in the first place.

For example:

  • The study found that stop rates did not always align with crime rates, meaning that policing patterns could be reinforcing racial disparities rather than simply responding to them.

  • It also highlights how local precinct strategies play a role in determining who gets stopped, which means that policing data is not a neutral measure of crime—it is shaped by decisions made by police departments about where to focus enforcement.

  • The "hit rate" paradox (where Black and Hispanic individuals were searched more but less likely to be found with contraband) suggests that police were making stops based on criteria that were not actually leading to more crime detection—raising questions about the reliability of stop data as an indicator of criminal activity.

So while the article doesn’t use the exact phrase “biased data,” it does imply that stop-and-frisk statistics are shaped by the choices police make rather than being an objective reflection of crime itself. Your question about whether data-driven policing can be skewed by the biases built into the data is definitely relevant to the article’s findings!