Reducing "Unknown" Data in Violent Death Surveillance: A Study of Death Certificates, Coroner/Medical Examiner and Police Reports From the National Violent Death Reporting System, 2003-2004

• NCJ Number: 229085
• Journal: Homicide Studies Volume: 13 Issue: 4 Dated: November 2009 Pages: 385-397
• Author(s): Joseph E. Logan; Debra L. Karch; Alexander E. Crosby
• Date Published: November 2009
• Length: 13 pages
• Publication Format: Article
• Publication Type: Report (Study/Research)
• Language: English
• Country: United States of America

Annotation

This study examined the value of using multiple-source documents (i.e., death certificates, coroner/medical examiner reports, and police reports) in order to reduce missing or "unknown" data on violent deaths.

Abstract

The finding that each of the data sources examined was complete for only 52-74 percent of cases emphasizes the potential benefits of having multiple data sources available for data collection and analysis. A common belief is that using more data sources will substantially reduce the amount of unknown data on violent deaths; however, this study found that this was true for only certain types of variables, i.e., those that captured the circumstances preceding the violent death. Using multiple documents to capture victim demographic information or event-related information (i.e., data and location of death or injury) provided little value over using only one source document. Although there was some evidence that each document source provided information that complemented the other sources, multiple data sources did not necessarily increase knowledge or reduce missing values on demographic and event information. Data on some variables may be inherently uncollectible by any source. A total of 22 demographic and event variables and 37 preceding-circumstance variables were analyzed across all documents on which they were collected. The source document that provided the most cases with "known" data (highest completion percentage) for a particular variable was used as the baseline for the analysis of the variable. The contribution of each additional source document was calculated by counting the number of unknown codes that the additional source replaced with known codes in the database. Finally, the percent increase in known data, when all documents were used, was determined in order to show the extent to which using multiple sources instead of a single source could reduce missing data for these variables. 2 tables and 8 references