Abstract
This special issue is devoted to J. Stephen Wormith, who passed away March, 2019. Through a series of research papers, we honor Steve’s contribution to criminal justice. The first inclination is to highlight Steve’s contribution to criminal justice “psychology.” But his impact was much greater than that, which in part is reflected by the breadth of articles in this issue.
True to a Polanyi view of science, there was a perceived fallibility, expressed by healthy exchanges (i.e., support of the Aggression and Violence conferences), taking risks (considering machine learning), and a responsibility (a “this better be right” attitude in research) in the development of our understanding. Steve understood this well. In addition to seeing this impact on his work colleagues throughout the years, I had the honor to repeatedly witness this as an external examiner in numerous Ph.D. defenses. Present was a kindness and gentle mentorship style, yet hanging onto a strong academic standard. Through this commitment to a strong academic standard, Steve motivated and compelled others to a similar vision of research and scholarship. The articles in this special issue reflect the impact of his contribution to research and scholarship.
The Desmarias et al. (this issue) article acknowledges previous pretrial risk assessment reviews (one of which Steve was a co-author), but points out that these reviews have not directly examined predictive validity of the measures. This study examines predictive validity by instrument across outcomes of new criminal activity, new violent criminal activity, failure to appear, and technical violations. These results are also conducted by gender and race/ethnicity. Overall, solid predictive validities were found across outcome and groups. Of interest, the evaluation of risk assessment studies scores the RAGEE on each included article (actual scores are in the supplemental materials). This represents an increase in the standardized evaluation of risk assessment studies.
The Orton et al. (this issue) paper examines overrides in administering the LSI-OR. The typical finding is that overrides result in lower predictive validities. This paper, though, examines overrides according to gender, race/ethnicity, and index offense categories. Being male, white, and having a sexual offense conviction the odds of an override. More telling was that, after controlling for initial risk level and time-at-risk, override change scores did not predict general recidivism. This use of overrides is reminiscent of the Hilton and Simmons study (2001) - when actuarial risk assessment is overlooked the decision making is influenced by the good looks of the client!
Within the framework of a gender-responsiveness approach, the Jones et al. (this issue) article examines gender differences for criminal thinking in a day reporting center. Accounting for the total criminal thinking score, there were no gender differences. But when the items were associated with criminal justice outcomes, only 11% of the items were considered gender-neutral. This suggests that a one-size-fits-all approach is short-sighted in assessing and interventions addressing criminal thinking.
The Olver et al. (this issue) article explores treatment change of criminal attitudes among those who have sexually offended. Using pre- and post- CSS scores, these changes were relatively independent of cognitive ability and other need areas. Knowing that attitude change robustly predicts negative criminal justice outcomes should spur us on to further examine individual mechanism of change in multiple criminal justice groups and settings.
Continuing to extend the efficacy of treatment and its impact on criminal justice outcomes, the Schafers et al. (this issue) article studies the results of an outpatient treatment program for intimate partner violence. Not only did the pre-post-change score of treatment targets predict future criminal justice outcomes, these results remained after the prediction models controlled for the assessment of risk. As these authors note, the next step is to consider treatment change prediction taking into account the individual baseline risk level.
Huang and colleagues (this issue) research differences between indigenous and non-indigenous youth on the YLS/CMI risk assessment instrument. Using Item Response Theory, differential item functioning results found 7 of 42 items to differ between the two groups. Even with these differences, the items’ relationship to latent risk were similar between the two groups.
Reflecting Steve’s willingness to consider the contribution of new areas, the Ghasemi et al. (this issue) article investigates the contribution of machine learning to risk assessment. Overall, the poorest efficiency of risk assessment occurs within the mid-range levels. Using machine language learning techniques, specifically random forest procedures, improvements were made in predicting this mid-range of risk scores. The use of machine learning research highlights the possibility of the multiple combinations of risk item scores that result in a specific level of risk.
The role of self-esteem has been contentious in both criminal justice theory and practice. The Thapa et al. (this issue) article examines the role of self-esteem, criminal attitudes, and gender on criminal justice outcomes among youth. This refinement to this literature is that high self-esteem magnified the relationship between criminal attitudes and criminal justice outcomes for only females. In addition, these authors suggest that high self-esteem may be falsely inflated self-esteem. The Motiuk and Keown article (this issue) explores the application of conducting intake assessments with a more streamlined measure via a hand-held device. Solid predictive validities of criminal justice outcomes were found, suggesting successful application of new technologies.
This issue is just a small reflection on Steve’s insistence for scholarship. This occurred alongside of his athletic prowess (he did play professional football), liking a little flair (super-hero cape wearing at conferences), some frugalness (my first meeting with Steve was while he was filling his pockets with snacks at an early 90s conference), his faith, his administrative responsibilities, his humane outlook (demanding the use of the term “correctional clientele”), and his strong liking of food. It was good to have known Steve on multiple levels. This was good, in and of itself. More importantly, he is one that beckons to be emulated.
