Underreporting of Crime-Related Content and the Prediction of Criminal Recidivism Among Violent Offenders Daryl G. Kroner Pittsburgh Institution Jeremy F. Mills Bath Institution and Carleton University Robert D. Morgan Texas Tech University The aim of this study was threefold: (1) to examine offenders’ underreporting of crime-related content, (2) examine explanations for underreporting, and (3) inves- tigate if accounting for underreporting increases predictability of recidivism over a standardized risk assessment instrument. Participants consisted of 89 adult male offenders incarcerated for violent offenses. Analysis revealed that when relying on offenders’ self-report of crime-related content, only 10% of information is lost because of underreporting. Correlation analyses indicated that underreporting was not explained by impression management, arrogant/deceitful interpersonal style, or number of past convictions. Finally, logistic and Poisson regression analyses indicated that accounting for underreporting in the prediction of recidivism did not increase predictive validity over a standardized risk assessment instrument. Impli- cation of these results for offender assessment and criminal risk assessment are discussed. Keywords: violent offenders, criminal prediction, self-report, underreporting One of the most basic tasks within the crim- inal justice system is to make decisions. Specif- ically, dispositional decisions regarding inca- pacitation are made by the courts, hospitals, prisons, and releasing boards. Gathering accu- rate information about offenders is essential for equitable decision making. Two common sources for such information are self-report and professional judgment/ratings. When self-report and professional ratings agree, our confidence level for the accuracy of the information is increased. But there are situations when self- report and professional ratings do not agree. This article examines three aspects of a discrep- ant situation. First, we ask a practical question, “How much crime-related information is lost through underreporting with client self-report?” Then, a second question, “What can account for underreporting?” Third, we evaluate underre- porting in the context of criminal risk prediction assessment. Early research on self-reported crime rates suggested that respondents are willing to dis- close their delinquency and criminal behavior (Wallerstein & Wylie, 1947). Among college students, Porterfield (1947) found that all of the 337 surveyed had committed at least one of- fense from a list derived from a sample of delinquents’ convictions. These initial findings of disclosure have been substantiated with sub- sequent research showing self-reported delin- quency across all socioeconomic stats (SES) categories (Short & Nye, 1957). However, knowing that respondents are willing to disclose their crime and delinquency does not address the correspondence of self-report with official records. Daryl G. Kroner, Department of Psychology, Pittsburgh Institution; Jeremy F. Mills, Department of Psychology, Bath Institution and Carleton University; and Robert D. Morgan, Department of Psychology, Texas Tech Univer- sity. Gratefully acknowledged are Roberto Di Fazio and Jeanne Clark for their assistance with the study, and Jennifer Walsh for her comments. The views expressed in this article are those of the authors and do not necessarily reflect the views of the Correctional Service of Canada. Correspondence concerning this article should be ad- dressed to Daryl G. Kroner, Pittsburgh Institution, P. O. Box 4510, Kingston, Ontario, K7L 5E5, Canada. E-mail: KronerDG@CSC-SCC.GC.CA Psychological Services Copyright 2007 by the American Psychological Association 2007, Vol. 4, No. 2, 85–95 1541-1559/07/$12.00 DOI: 10.1037/1541-1559.4.2.85 85