Reliability of self-reported childhood physical abuse by adults and factors associated with inconsistent reporting of childhood physical abuse

Though most studies of childhood physical abuse (CPA) depend on self-report, little is known about the reliability of self-reported CPA. We estimated the reliability of self-reported CPA and identified factors associated with inconsistent CPA reporting. This study is based on a 1995 national survey of couples randomly selected from US households using multistage cluster sampling. We included 2256 individuals who responded definitely to the same five CPA questions asked in 1995 and at a follow-up survey in 2000. We used kappa to estimate the reliability of self-reported CPA. We employed predictive modeling using logistic regression for survey data to identify statistically significant (p<0.10) predictors of inconsistent CPA reporting relative to consistently reporting CPA and consistently reporting no CPA, separately. Reliability of CPA as a dichotomous measure was moderate (K=0.41, p-value<0.01); it was slightly lower for men (K=0.32, p-value<0.01) than women (K=0.45, p-value<0.01). Compared to consistent CPA reporters, inconsistent reporters were more likely to be younger (18-29 vs. ³50 years old); have <12 years of education (vs. =12); and report only one (vs. ³2) kind of CPA. Relative to respondents reporting no CPA, inconsistent reporters were more likely to be male and younger (<18-29 vs. ³50 years old). Other factors associated with inconsistent reporting included race/ethnicity, a history of intimate partner violence, and having alcohol-related problems. Self-reported CPA is not especially reliable. Several demographic and behavioral characteristics were predictive of inconsistently reporting CPA. Researchers may find these results helpful when devising data collection tools and methods to improve self-reported CPA.