Header Image for Print

Program Monitoring and Research Division Working Paper Series


Authors:   

Posted Date:   
August 26, 2009



Bayesian Hierarchical Model Estimates of Local Crime Perceptions, by Brent D. Mast.

This study uses survey data to estimate a Bayesian hierarchical model of local crime perceptions. Data are employed from two sources.

Top level prior hyperparameters are based on crime perception responses from the American Housing Survey (AHS). The AHS is actually two surveys, metro and national, taking place in different years. I employ data from the national AHS for 2001.

Data was aggregated to the county level from HUD's survey of Section 8 Housing Choice Voucher (HCV) households. Dubbed the Customer Satisfaction Survey (CSS), it was a three year national survey conducted between 2000 and 2002.

Nearly one-half million households returned questionnaires, answering a wide variety of questions regarding the condition of their housing and neighborhoods. The large sample was stratified by public housing agency and year. This paper focuses on responses to a question regarding neighborhood crime and drug problems.

Results indicate that the Bayesian approach yields more robust local estimates. Compared to estimates solely based on CSS data, the Bayesian estimates have lower variance and correlate more highly with published county crime rates.

The data are described in more detail in the following sections. The Bayesian Hierarchical model is then described. Estimates are presented next. Correlation of survey estimates with county crime rates is then explored. The final section summarizes the analysis.


Program Monitoring and Research Division Working Papers



Related Categories: 




Most Recent Publications



Note: Guidance documents, except when based on statutory or regulatory authority or law, do not have the force and effect of law and are not meant to bind the public in any way. Guidance documents are intended only to provide clarity to the public regarding existing requirements under the law or agency policies.