Date of Award




Document Type


Degree Name

Doctor of Philosophy (PhD)


School of Criminal Justice

Content Description

1 online resource (xv, 286 pages) : illustrations (some color), color maps

Dissertation/Thesis Chair

Alan J Lizotte

Committee Members

Terence P Thornberry, Marvin D Krohn, Steven Messner


crime clusters, hotspots, offender routines, spatial analysis, Crime prevention, Criminal behavior, Prediction of, Geographical offender profiling, Crime analysis

Subject Categories



Research confirms what many people witness everyday--levels of crime are higher at and around some places compared to others. Explanations surrounding the spatial clustering of crime incidents, or hotspots, typically focus on characteristics of the criminal event and devote little attention to the role of the offender. Building from ideas set forth in routine activities and crime pattern theories, the first goal of this dissertation is to address this missing element. The presence of crime hotspots are estimated using geocoded crime incidents. The resulting maps, along with bivariate and multivariate analyses, examine offender-based explanations for the development of crime hotspots while taking into account social-structural and place-based explanations. The second goal of this dissertation is to examine particular hypotheses about how these three types of explanations interact to influence the spatial clustering of crime. Offense data for a sample of subjects participating in the Rochester Youth Development Study (RYDS) are combined with interview data, data from the 1990 Census, and the geographical locations of criminogenic places. This person-incident file is divided into four samples based on offender age. Results of logistic regression models indicate some support for offender-based explanations for the clustering of crime incidents, but only among offenders 12-15 years old. On the other hand, the effect of criminogenic places influences the spatial clustering of crime across the three age groups examined in logistic regression models. Age mediates the relationship between some measures and clustering, but otherwise there are no two- or three-way interaction effects among variables across the three levels of analysis.

Included in

Criminology Commons