Predicting Risk to be Sexually Abused: A Comparison of Logistic Regression to Event History Analysis

Abstract

The primary purpose of this study was to compare a model of risk to be abused deriving from logistic regression analysis to one deriving from event history analysis. Logistic regression analysis tests for the unique and combined effects of independent variables on risk to be abused. Event history analysis, in addition, tests whether and how the independent variables differentially affect risk, over time, to be abused. This comparison was done to determine whether information gained from event history analysis improved the ability to predict risk to be abused and involved a secondary data analysis of Russell's community prevalence survey on child sexual abuse. Although most results of event history analysis and logistic regression analysis were similar, event history analysis uniquely allowed for a test of interaction effects with age at the time of abuse; this interaction added significantly to the model.