Improving the Measurement of Socially Unacceptable Attitudes and Behaviors With Item Response Theory

Assessment of socially unacceptable behaviors and attitudes via self-report is likely to yield skewed data that may be vulnerable to measurement non-invariance. Item response theory (IRT) can help address these measurement challenges. This paper illustrates application of IRT to data from a teen dating violence intervention study. Three factors reflecting teens’ attitudes about dating violence were identified, and items from these 3 scales were subjected to IRT calibration and evaluated for differential item functioning (DIF) by gender. The IRT scores displayed superior measurement properties relative to the observed scale scores, and in 1 of the 3 factors, inferences about group mean differences were impacted by the presence of DIF. This application demonstrates how IRT analysis can improve measurement of socially unacceptable behaviors and attitudes.