Neighborhood – Banerjee et al. (2019)

Study Reference:

Banerjee, A., Duflo, E., Keniston, D., & Singh, N. (2019). The efficient deployment of police resources: theory and new evidence from a randomized drunk driving crackdown in India (No. w26224). National Bureau of Economic Research.

Location in the Matrix; Methodological Rigor; Outcome:

Neighborhood; Focused; Proactive; Very rigorous; Mixed findings

What police practice or strategy was examined?

This study evaluated the effectiveness of anti-drunk driving programs using sobriety checkpoints in Rajasthan, India. Participating police stations implemented checkpoints one to three times per week for one to three months. The checkpoints were held from 7pm until 10pm. Some stations held all their checkpoints at one fixed high-risk location, while other stations rotated their checkpoints (randomly) from day to day between three different high-risk locations.

How was the intervention evaluated?

In total, 147 police stations were randomly selected to do the checkpoints during the study, and 76 stations served as control locations that did not do the regular checkpoints. Checkpoint stations were also randomly assigned to either fixed or rotating checkpoint locations and also to implement checkpoints once, twice, or three times per week. (The types of officers conducting the checkpoints and the exact duration of the checkpoint operations also varied based on random assignment, but these aspects were not examined in the main experimental analysis.) Some checkpoint operations were conducted from September through October 2010, but most were conducted from September through November 2011. The researchers used data from August 2010 through October 2012 to assess how the checkpoints affected total traffic accidents and deaths while in operation and for 90 days after.

What were the key findings?

Overall, stations that implemented rotating checkpoints saw a 17% reduction in nighttime accidents and a 25% reduction in nighttime deaths. This was primarily due to the stations conducting rotating checkpoints, where nighttime accidents and deaths declined 29-30%. Checkpoints held at fixed locations had no impact on accidents or deaths. The rotating checkpoints also appeared to become more effective as checkpoint frequency increased - conducting two to three checkpoints per week was more effective than only doing one. However, there were also indications that daytime accidents increased in areas that did the rotating checkpoints, suggesting some temporal displacement.

What were the implications for law enforcement?

The authors suggest that randomly rotated checkpoints in areas where drunk driving violations are high are likely the best use of scarce police resources for reducing accidents and deaths, especially if these are done over extended periods of time (conducting 20 checkpoints over 82 days appeared most optimal). Extending the operations to different times of the day might prevent any temporal displacement.

Where can I find more information about this intervention, similar types of intervention, or related studies?