![handle is hein.journals/crmjusbhv44 and id is 1 raw text is:
INTRODUCTION TO SPECIAL ISSUE
STATISTICAL ISSUES AND INNOVATIONS IN
PREDICTING RECIDIVISM
L. MAAIKE HELMUS
Forensic Assessment Group
KELLY M. BABCHISHIN
Royal Institute ofMental Health, University ofOttawa
Risk assessment is one of the most common tasks in the criminal justice system, yet most professionals in this field receive
little to no formal training in statistical techniques for predicting dichotomous outcomes, such as recidivism. The purpose of
this special issue was to help fill this gap in training and resources. We wanted to make some of the latest statistical issues
and advances in predicting recidivism accessible to the readership of Criminal Justice and Behavior. In this introductory
paper, we briefly describe the seven articles in this issue. The first three articles provide primers on topics (statistics to assess
predictive accuracy, the Expected/Observed [E/O] Index, and mediation analyses, respectively) in a way that is meant to be
understandable to clinicians and researchers. The next two articles describe and compare different statistics for assessing
change over time. The last two articles explore limitations of currently used recidivism analyses (area under the curves
[AUCs], Harrell's C, Cox and logistic regression). We hope this issue will serve as a helpful resource for those who conduct
or consume research on predicting recidivism.
Keywords: risk assessment; prediction; recidivism; statistics; effect sizes
The more I live, the more I learn. The more I learn, the more I realize, the less I know.
Michel Legrand
R isk assessment is arguably the most ubiquitous task in the criminal justice system-
every decision impacting an offender is typically (or at least should be) based on an
evaluation of the offender's likelihood to reoffend. Consequently, forensic journals are full
of papers examining recidivism, whether it be predictors of recidivism or the impact of
something on recidivism. Our confidence in these research studies hinges on their method-
ological rigor, which includes choosing statistical analysis methods that are appropriate to
the research questions and the types of data involved, and executing and interpreting these
analyses appropriately.
AUTHORS' NOTE: Correspondence concerning this article should be addressed to L. Maaike Helmus,
Forensic Assessment Group, 2-215 St. Patrick Street, Ottawa, Ontario, Canada KIN 5K2; e-mail:
Lmaaikehelmus@gmail. com.
CRIMINAL JUSTICE AND BEIAVIOR, 2017, Vol. 44, No. 1, January 2017, 5-7.
DOI: 10.1177/0093854816678897
© 2016 International Association for Correctional and Forensic Psychology](ViewImageLocal?handle=hein.journals/crmjusbhv44&div=1&collection=&method=preview&ext=.png&size=3)
What Is HeinOnline?
Contact us for annual subscription options:
Already a HeinOnline Subscriber?
profiles profiles most
Contact us for annual subscription options:
Already a HeinOnline Subscriber?
profiles profiles most
Contact us for annual subscription options:
Already a HeinOnline Subscriber?
profiles profiles most