| Statistic | pre | post |
|---|---|---|
| mean_ratio | 0.419 | 0.310 |
| median_ratio | 0.350 | 0.252 |
| arrest_per_year | 109310 | 65909 |
| crime_per_year | 226875 | 198883 |
| n_tracts | 1117 | 1217 |
| n_dropped | 14 | 7 |
Policing Spatial Concentration Under LAPD’s Post-PredPol DICFP Program
Background
LAPD ran PredPol, a place-based predictive policing program, from 2011 to 2020 (Mohler et al. 2015). It was discontinued that April, driven by COVID-era budget cuts and by criticism that it concentrated enforcement in already over-policed neighborhoods regardless of underlying crime rates (Stop LAPD Spying Coalition 2021; The Markup 2021; Lum and Isaac 2016). LAPD replaced it that same month with Data-Informed, Community-Focused Policing (DICFP), whose stated goals were building trust, reducing violent crime, and assisting victims (Los Angeles Police Department 2020). Optimists read DICFP as a genuine break from PredPol’s targeting logic; pessimists read it as the same logic under a new name. This project tests which reading the tract-level enforcement data supports.
Findings
The median arrest-to-crime ratio dropped from 0.35 in the PredPol era to 0.25 in the DICFP era, and the mean dropped from 0.42 to 0.31. Both crime and arrests fell in annualized terms too (crime: ~226,875/yr to ~198,883/yr; arrests: ~109,310/yr to ~65,909/yr), but arrests fell faster, so enforcement intensity declined relative to crime rather than simply tracking a drop in crime itself. A small number of tracts (14 pre, 7 post) had zero recorded crime and were excluded, since the ratio is undefined there.
This is only a citywide baseline; it doesn’t yet address whether the decline is uniform across neighborhoods regardless of racial composition, which is the next layer for the full paper. It’s also worth noting that the DICFP-era window (2022–2024) still captures lingering pandemic and post-2020 effects on policing that aren’t cleanly separable from the DICFP transition itself.
References
Los Angeles Police Department. 2020. Data-Informed, Community-Focused Policing in the Los Angeles Police Department. Los Angeles Police Department. https://lapdonlinestrgeacc.blob.core.usgovcloudapi.net/lapdonlinemedia/2021/12/data-informed-guidebook-042020.pdf.
Lum, Kristian, and William Isaac. 2016. “To Predict and Serve?” Significance 13 (5): 14–19. https://rss.onlinelibrary.wiley.com/doi/full/10.1111/j.1740-9713.2016.00960.x.
Mohler, G. O., M. B. Short, Sean Malinowski, et al. 2015. “Randomized Controlled Field Trials of Predictive Policing.” Journal of the American Statistical Association 110 (512): 1399–411. https://www.tandfonline.com/doi/full/10.1080/01621459.2015.1077710.
Stop LAPD Spying Coalition. 2021. Automating Banishment: The Surveillance and Policing of Looted Land. Stop LAPD Spying Coalition. https://automatingbanishment.org.
The Markup. 2021. How We Determined Crime Prediction Software Disproportionately Targeted Low-Income, Black and Latino Neighborhoods. December 2. https://themarkup.org/show-your-work/2021/12/02/how-we-determined-crime-prediction-software-disproportionately-targeted-low-income-black-and-latino-neighborhoods.