CrimeStat IV is available
The National Institute of Justice (NIJ) recently released version 4.0 of CrimeStat (CrimeStat IV). NIJ is the research division of the U.S. Department of Justice. CrimeStat IV was developed by Ned Levine & Associates of Houston, TX with NIJ grants and is a spatial statistics program for analyzing crime incident locations. It provides spatial statistical tools to aid criminal justice researchers and crime analysts in their crime mapping efforts. The program and documentation are free. CrimeStat IV is applicable to many diverse research fields. Although the program was designed for crime analysis, many of the spatial tools are appropriate for social and policy analysis (e.g., the spatial distribution of poverty or of non-voters or disease locations). The program handles both incident and zonal data and interacts with most GIS packages. For incidents, the geographical coordinates of the locations are input (e.g., individual residences, motor vehicle crash locations, or homicide locations) while, for zonal data, the geographical coordinates of the zones are input (e.g., the centroids of census block groups or polling districts). Various routines calculate properties of these data sets and many results can be written as graphical objects to a GIS. CrimeStat IVincludes more than 100 statistical routines for the analysis of incident data including the spatial distribution of the incidents, the density of events, the relative risk of events compared to a baseline variable, the behavior of individuals both in space and time, and a full-blown travel demand module for describing the mobility of rare events (e.g., modeling medical treatment trips or drunk driving crashes in a region). Hot spot analysis involves identifying concentrations of events or people, for example the concentration of voters by party, the concentration of the homeless, and the distribution of party voting percentages. CrimeStat IV includes the most comprehensive collection of hot spot analysis tools of any software package, from identifying very small clusters of events to documenting clustering by sub-regions of a study area. What’s New in CrimeStat IV? New to CrimeStat IV is a spatial regression module that models events for zones. The module uses both Maximum Likelihood and Markov Chain Monte Carlo simulation to estimate complex multivariate spatial functions that can fit the distribution of crime in an area. The module estimates Poisson, Normal, and Binomial functions and can include a spatial autocorrelation component as well as an ‘at risk’ variable. Other new modules include:· Discrete choice modeling of unique choices (e.g., housing choice decisions; type of weapon used in robberies; decisions to evacuate during a hurricane)· Time series forecasting for estimating the likely number of future incidents that will be committed in small areas (e.g., the number of homicides expected next month; the number of EMS trips expected next week)· Head-bang smoothing for producing estimates of risk with zonal data (e.g., estimating voter preference rates for local elections with low voter turnout)· An extensive number of routines for examining spatial autocorrelation and clustering with zonal data (e.g., the clustering of census tracts that have high numbers of households living in poverty; number of voters choosing Democratic or Republican candidates by census block group; identifying the clustering of blocks that have high numbers of multi-unit buildings).· In addition, .NET libraries allow third-party applications to use many CrimeStat routines. The documentation includes 32 chapters, three appendices, an extensive reference list, sample data sets, and documentation for the .NET libraries. The program and documentation are available from: http://www.nij.gov/CrimeStat For more information, contact Ned Levine, PhD, at: CrimeStat@nedlevine.com.