The Danger Assessment Instrument: Modifications Based on Findings From the Intimate Partner Femicide Study

Jacquelyn C. Campbell, Johns Hopkins University
Judith McFarlane, Johns Hopkins University
Daniel Webster, Johns Hopkins University
Susan Wilt, Johns Hopkins University
Carolyn Rebecca Block, Illinois Crim Justice Info Authority
Phyllis Sharps, George Washington University
Doris Campbell, University of South Florida
Carolyn J. Sachs, California State University - Los Angeles
Jane Koziol-McLain, Johns Hopkins University

This paper will present new findings on the psychometric properties of the Danger Assessment (DA) intimate partner homicide risk assessment instrument from the NIJ/NIH/CDC funded 12 City case control study of risk factors for intimate partner femicide. A sample of femicide/attempted femicide (380) victims (cases) were recruited from police, medical examiner, shelter and trauma center records with interviews conducted with proxies for the actual femicides and the victims of attempted femicide. Abused controls (N = 384) and non-abused Controls (376) were recruited by telephone survey. Internal consistency reliability for the DA was .72 in the cases and .74 in the abused controls with a significant mean difference between the two groups (6.3 vs. 3.2). Risk factors for femicide from the DA supported by bivariate analysis were increased severity and frequency of physical violence, threats to kill, perception of capability of killing, choking, gun in house, forced sex, abuse during prenancy, extreme controlling behaviors, extreme jealousy, perpetrator suicidality, violence outside of home, and child abuse. Adjusted multivariate analysis resulted in major risk factors of: increased severity and frequency, threats to kill, gun access, forced sex and extreme controlling and jealousy in situations of estrangement (interaction effect). At a score of 8 or more on the original DA, Positive Predictive Value (PPV) was 90% but Negative Predictive Value was less acceptable. ROC analysis demonstrated good specificity at a score of 8 or more and acceptable sensitivity at a score of 4. Based on these findings, a modified DA will be presented with scoring instructions to weight the strongest risk factors and to take into account stalking and estrangement as risk factors under certain conditions.

(Return to Program Resources)

Updated 05/20/2006