U.S. flag

An official website of the United States government, Department of Justice.

Multiple Imputation for Missing Values in Homicide Incident Data: An Evaluation Using Unique Test Data

NCJ Number
254018
Journal
Homicide Studies Volume: 22 Issue: 4 Dated: 2018 Pages: 391-409
Date Published
2018
Length
19 pages
Annotation

Since incident-level homicide datasets, such as the Supplementary Homicide Reports (SHR), commonly exhibit missing data, the current study evaluated multiple imputation methods (that produce multiple completed datasets, across which imputed values may vary) via unique data that included actual values from police agency incident reports of seemingly missing SHR data.

Abstract

This permitted evaluation under a real, not assumed or simulated, missing data mechanism. The study compared analytic results based on multiple imputed and actual data; multiple imputation rather successfully recovered victim-offender relationship distributions and regression coefficients that hold in the actual data. Results are encouraging for users of multiple imputation, though it is still important to minimize the extent of missing information in SHR and similar data. (publisher abstract modified)

Date Published: January 1, 2018