Psychometric Analyses of Stalking Measures from the National Intimate Partner and Sexual Violence Survey

Abstract

This study examined the psychometric properties of stalking victimization and related fear-based measures in the National Intimate Partner and Sexual Violence Survey (NISVS) to better understand the ability of these items to produce valid estimates and comparisons across gender groups. This analysis used data from 9086 women and 7421 men in the 2010 NISVS conducted by the U.S. Centers for Disease Control and Prevention. Descriptive, reliability, and dimensionality statistics were examined for seven measures of stalking behaviors and four indicators of fear. The structure of each measure was then examined using item response theory-based methods. Results showed that the NISVS measures were effective at capturing information for a broad range of stalking victimization and fear levels. Item difficulties did vary by gender, particularly for the fear indicators. Some items were more difficult for men, while others were more difficult for women. In the end, differences canceled each other out, and no meaningful gender differences were found. The measures performed identically for women and men. This study provides confidence in our ability to use the NISVS questions to estimate the prevalence of stalking across gender groups. Results have important implications for selecting stalking behaviors and fear indicators to measure stalking.