"Broken windows" theory is an influential model of neighborhood change, but there is disagreement over whether public disorder leads to more serious crime. This article distinguishes between public and private disorder, arguing that large-scale administrative data provide new opportunities to examine broken windows theory and alternative models of neighborhood change.

We apply an ecometric methodology to two databases from Boston: 1,000,000+ 911 dispatches and indicators of physical disorder from 200,000+ requests for nonemergency services. Both distinguish between disorder in public and private spaces. A cross-lag longitudinal analysis was conducted using two full years of data (2011–2012).

The two databases provided six dimensions of physical and social disorder and crime. The cross-lag model revealed eight pathways by which one form of disorder or crime in 2011 predicted a significant increase in another in 2012. Although traditional interpretations of broken windows emphasize the role of public disorder, private conflict most strongly predicted future crime.

Our results describe a social escalation model where future disorder and crime emerge not from public cues but from private disorder within the community, demonstrating how "big data" from administrative records, when properly measured and interpreted, represent a growing resource for studying neighborhood change.