Statement of the Problem: Serial sexual assault is a pervasive problem: court record recidivism studies show that 10-15% of convicted sex offenders re-assault within five years, and self-report surveys suggest that 60%+ of males who have committed sexual assaults have raped more than one individual. The purpose of the current study is to document the scope of repeat perpetration using a different method of finding serial rapists: DNA evidence in sexual assault kits (SAKs). Using a sample of previously untested SAKs from Detroit, we will examine rates and patterns of serial sexual assault based on DNA matches in CODIS (Combined DNA Index System), the federal criminal DNA database. Guided by criminal career theory, we will explore serial sexual offenders longitudinal crime patterns and the extent to which their offending varies with respect to victim-offender relationship (i.e., stranger vs. non-stranger) and victim age (i.e., child vs. adult victims).
Subjects: We will examine how many CODIS hits emanate from a sample of N=7,393 previously untested SAKs. We will also obtain full adult criminal histories for all CODIS-identified offenders to model their offending patterns over time.
Partnerships: Our practitioner partners are the Michigan State Police, who have granted research access to the CODIS testing outcomes and criminal histories; and the Michigan Domestic & Sexual Violence Prevention & Treatment Board, who will advise the team on implications for survivors and victim advocacy services.
Research Design & Methods: Using a cross-sectional design, we will document the rates and types of serial sexual assault CODIS hits and how those rates vary by victim-offender relationships and ages. Then, we will shift to a longitudinal design to explore patterns of serial sexual offending, both patterns identifiable within an offenders own criminal history and similarities between the cumulative criminal histories of different offenders. We will also examine variability in the geographic locations of serial sexual offenders patterns of crime.
Analysis: We will use continuation-ratio and polytomous logistic regression models to quantify the rates and types of CODIS hits. Latent variable mixture models will be used to identify subgroups with distinct patterns of offending over time. We will apply log-linear models to transition matrices to explore geographic mobility of offenders crimes over time.
Products, Reports, & Data Archiving: Research products will include: interim and final reports; archived data pertaining to CODIS hits and adult criminal histories (de-identified); and conference presentations, scientific publications, webinars, and podcasts that summarize key findings. ca/ncf