Case-by-Case Comparison of the Classification of Law Enforcement and Vital Statistics Data on Homicide

• NCJ Number: 213104
• Journal: Criminal Justice Policy Review Volume: 17 Issue: 1 Dated: March 2006 Pages: 61-82
• Author(s): Marc Riedel; Wendy C. Regoeczi
• Date Published: March 2006
• Length: 22 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This study used data from the California Linked Homicide File in order to test the validity and reliability of homicide data.

Abstract

Case-by-case comparisons of variables reported by both Supplementary Homicide Reports and vital statistics for 1992 through 1998 showed that agreement between police and vital statistics data was highest when using the variables of victim's gender and race and lowest with the variables of victim's age, manslaughters, and police justifiable homicides. A multilevel analysis that examined the types of cases that could not be linked through the two data-collection systems found that homicides with Hispanic victims; weapons other than handguns; family members other than intimate partners; and homicides that involved felonies, other nonfelonies, and negligent manslaughters had a greater likelihood of not being matched across the two data sources. Death investigation systems that used medical examiners also diminished agreement between the data sources. In the case-by-case matching, individual-level variables were multiple victims, victim race/ethnicity, victim-offender relationships, circumstances, location, and weapons. County-level variables were the death investigation system and population size. The analysis was conducted in two stages. The first addressed the amount of agreement between shared variables for the two data sources. Comparisons were made for crime status, gender, age, and race/ethnicity. These variables were also compared by county. The second analysis focused on the differences between the 24,426 matched cases and the 1,818 unmatched cases. For this analysis, the nested structure of the data required the use of a program that could incorporate the multilevel nature of the data. These analyses were conducted with the hierarchical linear modeling software developed by Raudenbush, Bryk, and Congdon. 1 figure, 5 tables, and 24 references