Date of Award


Document Type

Open Access Dissertation


Criminology and Criminal Justice


College of Arts and Sciences

First Advisor

Robert J. Kaminski


Occupational risk of violent victimization is a serious concern for law enforcement officers. However, there have been virtually no studies that examined the relationship between the incidence of police officer homicide victimization and the deinstitutionalization movement during which large number of persons with mental illness were released back into communities, often without adequate support systems. Research has shown that persons with certain types of mental illness have a greater propensity for violent behavior if they fail to take prescribed medications and/or abuse illicit substances. Since police are most often the first responders to persons with mental illness in crisis, increases in police encounters with such subjects may increase officer risk of injury and death. The present study will test whether or not increases in the number of mental health patients released from psychiatric hospitals is positively associated with murders of law enforcement officers. State-level data on police officer homicide victimization for the years 1972-2003 are used to test this hypothesis. The study takes advantage of a Bayesian-based hierarchical spatio-temporal analysis, a relatively new analytic technique in Criminology, to simultaneously account for spatial autocorrelation across states as well as over time. The results indicate that the change in the hospitalized mentally ill population had no statistically significant effect on the fatal victimization risk for police in general, but showed some temporal variations when a random slope model was employed. Meanwhile, this study finds negative effects of residential stability, residual incarceration rates, and age structure on police homicides, and positive effects of economic deprivation, female headed households, and percent black on the fatal risks for police. A hot spot of high-risk areas for police consisting of Georgia, Louisiana, Mississippi, and South Carolina is identified by exceedance probability mapping of the estimated relative fatal risks. Elevated residual risks for police due to unmeasured risk factors are found in several southern states, western states, and midwestern states.


© 2016, Xueyi Xing