Multiple Imputation for the Supplementary Homicide Reports: Evaluation in Unique Test Data

The purpose of this research is to advance methods addressing missing data by evaluating multiple imputation strategies. Multiple imputation (analyzing multiple, varying completed data sets) provides a means of evaluating missing data strategies by comparing imputed and actual data in various ways. This project will multiply impute values for missing data and compare results to known actual values in homicide incident reports using secondary data from Wadsworth and Roberts (2007) test population which includes five years of homicide incidents from Chicago, Philadelphia, Phoenix, and St. Louis. Analyses will use multiple imputation routines available in general and specific software (particularly SAS and IVEWARE) to compare imputed and actual distributions of victim-offender relationship, substantive models for relationships of criminological interest, and the impact of technical choices (e.g. particular multiple imputation methods, number of imputations used). ca/nfc