Award Information
Description of original award (Fiscal Year 2012, $560,620)
The purpose of this study is to combine and collect data from a large number of sources to study crime and crime trends across the Los Angeles-Long Beach-Santa Ana metropolitan area. This wide array of information will allow accounting for the multi-dimensional and inter-related sources of crime and crime trends in Southern California at different units of analysis including blocks and neighborhoods.
Using this data, UCI will 1) build a model to predict crime in small geographic areas; 2) assess the effect of ex-offenders in an age of mass incarceration and the issues surrounding the implementation of realignment in California; 3) examine the clustering of social problems; 4) assess the effect of neighborhood organizations and institutions on crime trends; 5) determine the effect of the spatial distribution of poverty (at both small and large scales) on crime rates; 6) test the impact of concentrated affluence and whether clustering of low income housing increases crime; 7) examine the effect of gentrification on crime in neighborhoods and nearby areas; 8) consider the relationship between immigration and crime, taking into account the neighborhood institutional context; and will 9) assess the effect of foreclosures and vacancies on neighborhood home values and crime over time.
The data for this project includes crime, offender supervision, business and employment, socio-demographic data, government expenditures, health, housing, locations of neighborhood institutions, land use, schooling, and weather. The analysis plan includes Poisson models that include an estimation of population effects. If over-dispersion is encountered, the research team will estimate negative binomial regression models. All estimations will directly model spatial processes. UCI will construct spatial weight matrices to determine the appropriate distance decay to model nearby areas. The analysis will also assess the impact of geographic characteristics on crime longitudinally, through cross-lagged or latent trajectory models. ca/ncf