This is the eighth of 10 chapters on "Spatial Modeling II" in the user manual for CrimeStat IV, a spatial statistics package that can analyze crime incident location data.
This chapter, "The CrimeStat Discrete Choice Module," explains the creation of a dataset that is appropriate for the Conditional Logit (CL) model. The first of two pages of CrimeStat's discrete choice module allows the creation of a data set appropriate for the CL model, and it estimates either Multinomial Logit (ML) or CL models. The model coefficients can be saved. Using model coefficients, the second page of CrimeStat's discrete choice module calculates predicted probabilities in either the same or another dataset. The first page explains how to create a dataset for a conditional discrete choice model. Topics addressed include the input case file and alternatives file, saving the output, the estimate model, the data file, choice variable, independent variables, and the type of discrete choice model. Also addressed are the reference alternative (ML model only), the output for the discrete choice model, and saving the output and the estimated coefficients. In illustrating the process of running a ML model, the chapter models the premises chosen for Chicago residential robberies for 1997. In illustrating the process of creating a file for the CL model and then running a file to estimate the predictors of the alternatives, the chapter uses an example of predicting which traffic analysis zone offenders use to commit crime. The second page of CrimeStat's discrete choice model guides the user in applying the estimated coefficients from a discrete choice model to another dataset or a subset of the same dataset. It also explains the calculation of predicted probabilities for either the ML or the CL logit model. 3 tables and 11 figures that include sample computer screens