Log-Multiplicative Association Model for Allocating Homicides with Unknown Victim-Offender Relationships

• NCJ Number: 196072
• Journal: Criminology Volume: 40 Issue: 2 Dated: May 2002 Pages: 457-480
• Author(s): Steven F. Messner; Glenn Deane; Mark Beaulieu
• Editor(s): Robert J. Bursik Jr.
• Date Published: May 2002
• Length: 24 pages
• Grant Number(s): PD30 HD32041; SBR 9512290
• Sponsoring Agency: National Science Foundation; National Institute of Child Health and Human Development (NICHD)
• Publication Format: Article
• Publication Type: Research Paper
• Language: English
• Country: United States of America
• Note: An earlier draft of this article was presented at the American Society of Criminology annual meeting held in San Francisco, CA, November 2000.

Annotation

This article discusses the research problem of studying homicides with unknown victim offender relationships and proposes an alternative methodology to approximate these relationships for the purpose of criminology research.

Abstract

In homicide research, researchers often face data which does not specify a victim/offender relationship for between 27 to 40 percent of available data. This missing data impedes research and several methods have been developed for imputing the victim/offender relationship. Methods discussed in the article include Adjusted Rate (AR) allocation method, "Circumstances Adjusted Rate" (CAR) allocation method, and "Multinomial Logit Regression Model" (MLR) allocation method. The authors propose an alternative method for this estimation which is based upon an imputation algorithm derived from a log-multiplicative model. Missing data are not required to be random. The authors applied their proposed methodology to the 1996-1997 Federal Bureau of Investigation Supplementary Homicide Report (SHR) and compared the results with other existing methods. In general the authors’ method resulted in a higher rate of stranger victim/offenders. 3 tables, 16 notes and references