Arbitrary decision making in the absence of evidence: an examination of factors related to treatment selection and recidivism for female sexual offenders

ABSTRACT

As criminal justice professionals are asked to assign risk levels and treat females who sexually offend, identifying risk factors for recidivism prediction and treatment targets is important. Although the majority of risk and treatment studies have focused on males, general female offender research and developing research with female sex offenders both provide some evidence for possible factors related to sexual offending behaviour in women. The purpose of this study was to explore what possible factors were related to treatment selection in a sample of 506 females serving prison sentences for a sexual offense, and, in turn, to examine which factors predicted non-sexual and sexual recidivism. Results demonstrated that exploratory variables were significantly related to treatment selection, as opposed to variables previously supported in research. Furthermore, living with a significant other for more than two years, and prior sexual offending arrests predicted sexual recidivism, while decreased age and treatment participation predicted non-sexual recidivism. These findings indicate that while similarities between male and female sexual offenders exist, females still demonstrate gender-specific risk factors.

KEYWORDS:

• Female offenders
• female sexual offenders
• recidivism
• risk assessment
• sexual offenders

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 In the State of Texas, this item is coded differently for females who sexually offend, in order to capture the same victim-offender gender dynamic as captured by the coding method for males.

2 Jones and colleagues (2006) separated offenses into “used force in crime”, “female child victim”. “male child victim”, “female child porn”, and “male child porn”. This method was not followed exactly since information on male or female child porn was not available in these data and victim gender was already accounted for in the Static 99-R (Phenix et al., 2016) item on victim gender, which could introduce multicollinearity issues.