Reducing the Methodological Heterogeneity ("Noise") in the Literature Predicting In-Prison Interpersonal Harm in Male Populations.

Interpersonal harm is a preventable public health problem. A growing body of literature shows persistently elevated exposure rates to physical and sexual victimization during incarceration. Yet how to prevent interpersonal harm during incarceration has proven far more elusive. A public health approach to prevention offers promise. To develop effective prevention strategies, the public health approach begins with defining and measuring the problem, followed by identifying risk and protective factors for the problem. The dynamic literature on in-prison interpersonal harm includes both parts of the public health approach but theoretical and methodological "noise" in this literature limits its instrumental utility to build effective prevention strategies. Herein, we critically review this evidence base (15 peer-reviewed articles published since 2000 with samples of 1,000+) to isolate the noise and the substance. We, then, minimize the methodological noise by testing for risk factors using self-report data that is representative of an entire U.S. state prison system for men and best data collection practices. Multilevel logistic regression is used to predict four types of interpersonal harm using theoretically grounded individual and prison-level covariates that are supported by the empirical literature. We conclude with recommendations for building an evidence base from which to develop prevention strategies that would create and sustain custodial conditions for people to be safe and healthy while incarcerated. [ABSTRACT FROM AUTHOR]