Responsiveness to Citizen Input: Topic Model Analysis of Public Comments to the San Francisco Police Commission

Abstract

Police commissions are uniquely positioned to funnel critical issues raised by citizens to police departments, thereby increasing accountability and potentially improving police/community relations. However, no empirical study to date has analyzed the responsiveness of a civilian policing oversight entity to the voiced needs and concerns of its local community. This paper details a dictionary-based topic model analysis conducted on the minutes from the last 20 years of San Francisco Police Commission (SFPC) meetings. Across the time period, the San Francisco community voiced issues related to crime, accountability, and community most commonly at meetings. Further analysis using a logit regression with a distributed lag model indicated public comment on a topic was strongly associated with the topic being discussed by the SFPC at the same meeting. However, SFPC discussion of the topics voiced by the public waned in subsequent meetings. The results suggest a potential failure of the SFPC to meaningfully act on the concerns raised by the public.