The optimal dynamic treatment rule superlearner: considerations, performance, and application to criminal justice interventions.

The optimal dynamic treatment rule (ODTR) framework offers an approach for understanding which kinds of patients respond best to specific treatments - in other words, treatment effect heterogeneity. Recently, there has been a proliferation of methods for estimating the ODTR. One such method is an extension of the SuperLearner algorithm - an ensemble method to optimally combine candidate algorithms extensively used in prediction problems - to ODTRs. Following the ``causal roadmap," we causally and statistically define the ODTR and provide an introduction to estimating it using the ODTR SuperLearner. Additionally, we highlight practical choices when implementing the algorithm, including choice of candidate algorithms, metalearners to combine the candidates, and risk functions to select the best combination of algorithms. Using simulations, we illustrate how estimating the ODTR using this SuperLearner approach can uncover treatment effect heterogeneity more effectively than traditional approaches based on fitting a parametric regression of the outcome on the treatment, covariates and treatment-covariate interactions. We investigate the implications of choices in implementing an ODTR SuperLearner at various sample sizes. Our results show the advantages of: (1) including a combination of both flexible machine learning algorithms and simple parametric estimators in the library of candidate algorithms; (2) using an ensemble metalearner to combine candidates rather than selecting only the best-performing candidate; (3) using the mean outcome under the rule as a risk function. Finally, we apply the ODTR SuperLearner to the ``Interventions" study, an ongoing randomized controlled trial, to identify which justice-involved adults with mental illness benefit most from cognitive behavioral therapy to reduce criminal re-offending.

Estimators for the value of the optimal dynamic treatment rule with application to criminal justice interventions.

Given an (optimal) dynamic treatment rule, it may be of interest to evaluate that rule - that is, to ask the causal question: what is the expected outcome had every subject received treatment according to that rule? In this paper, we study the performance of estimators that approximate the true value of: (1) an a priori known dynamic treatment rule (2) the true, unknown optimal dynamic treatment rule (ODTR); (3) an estimated ODTR, a so-called "data-adaptive parameter," whose true value depends on the sample. Using simulations of point-treatment data, we specifically investigate: (1) the impact of increasingly data-adaptive estimation of nuisance parameters and/or of the ODTR on performance; (2) the potential for improved efficiency and bias reduction through the use of semiparametric efficient estimators; and, (3) the importance of sample splitting based on the cross-validated targeted maximum likelihood estimator (CV-TMLE) for accurate inference. In the simulations considered, there was very little cost and many benefits to using CV-TMLE to estimate the value of the true and estimated ODTR; importantly, and in contrast to non cross-validated estimators, the performance of CV-TMLE was maintained even when highly data-adaptive algorithms were used to estimate both nuisance parameters and the ODTR. In addition, we apply these estimators for the value of the rule to the "Interventions" study, an ongoing randomized controlled trial, to identify whether assigning cognitive behavioral therapy (CBT) to criminal justice-involved adults with mental illness using an ODTR significantly reduces the probability of recidivism, compared to assigning CBT in a non-individualized way.

Psychopathic Personality in Early Childhood: A Critical Comment on Lopez-Romero et al. (2021).

A recent article published in the Journal of Personality Disorders (López-Romero et al., 2021) described the identification of "putative psychopathic personality" in a school cohort of 3-6-year-old children from Spain. This comment offers cautionary considerations of the original article on scientific grounds and critical comments on policy grounds. We caution researchers, policymakers, attorneys, judges, and the general public about the dangers of using this label given present knowledge about the antecedents, early indicators, and stability of the adult disorder of psychopathic personality when assessed in childhood.

Community Behavioral Health Service Use and Criminal Recidivism of People With Mental, Substance Use, and Co-occurring Disorders.

OBJECTIVE: This study assessed the relationship between community behavioral health service (CBHS) use and criminal recidivism in a broad sample of potential beneficiaries and by diagnostic group. METHODS: Among a cohort of people on probation with any mental and/or substance use disorder (N=772), the study estimated the effect of CBHS use on rearrest with Cox proportional hazards models. RESULTS: Service use significantly predicted reduced recidivism among people with any mental disorder (hazard rate=0.36, p=0.008), but not among those with any substance use disorder or co-occurring disorders. CONCLUSIONS: CBHS use in a given week predicted a 64% reduced recidivism risk during the following week among people with any mental disorder. However, CBHS use had no clear relationship with recidivism among people with co-occurring disorders or any substance use disorder. CBHS use may reduce recidivism, depending on recipient and service characteristics.

Testing assumptions about peer support specialists' susceptibility to stress.

Peer support specialists (PSs) have mental health recovery experience and are hired to assist others with similar challenges. This study is among the first to characterize stress among PSs, compared to data on other groups. Seven hundred and thirty-eight PSs working in U.S. mental health settings completed a cross-sectional online survey that included the Maslach Burnout Inventory, the Secondary Traumatic Stress Scale, and the Perceived Stress Scale (PSS). Participants' scores were compared with those obtained in prior samples of nonreferred adults and nonpeer clinicians. Comparisons are stratified by PSs' level of current symptoms, assessed with the Brief Symptom Inventory. As a group, PSs experienced low to moderate levels of general and work-related stress. PSs endorsed modestly lower levels of general stress (d = -.25) than a normative sample of community residents. Although PSs endorsed lower levels of secondary trauma (d = -.15) than social workers and greater emotional exhaustion (d = .13) than nonpeer clinicians, effect sizes are "smaller than small." A small subgroup (21.6%) of PSs with significant current symptoms experienced substantially greater general stress, secondary trauma, and emotional exhaustion than comparisons (d = 0.77, 1.04, and 1.12, respectively); despite having work conditions similar to other PSs respondents. Overall, PSs appear no more susceptible to general stress and work-related stress than relevant comparison groups of community residents and clinicians. A small subgroup of PSs experience both significant stress and symptoms-as is the case in other populations, given the well-established association between these constructs. Implications for supporting PSs and other clinicians with periods of work stress are discussed. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Neighborhood Risk Factors for Recidivism: For Whom do they Matter?

Justice-involved people vary substantially in their risk of reoffending. To date, recidivism prediction and prevention efforts have largely focused on individual-level factors like antisocial traits. Although a growing body of research has examined the role of residential contexts in predicting reoffending, results have been equivocal. One reason for mixed results may be that an individual's susceptibility to contextual influence depends upon his or her accumulated risk of reoffending. Based on a sample of 2218 people on probation in San Francisco, California, this study draws on observational and secondary data to test the hypothesis that individual risk moderates the effect of neighborhood factors on recidivism. Results from survival analyses indicate that individual risk interacts with neighborhood concentrated disadvantage and disorder-and these factors increase recidivism among people relatively low in individual risk, but not those at higher risk. This is consistent with the disadvantage saturation perspective, raising the possibility that some people classified as low risk might not recidivate but for placement in disadvantaged and disorderly neighborhoods. Ultimately, residential contexts "matter" for lower risk people and may be useful to consider in efforts to prevent recidivism.

Clarifying Conceptions Underlying Adult Psychopathy Measures: A Construct Validity Metric Approach.

Although the Psychopathy Checklist-Revised (PCL-R) and Psychopathic Personality Inventory (PPI) ostensibly measure the same construct, they seem to emphasize different conceptions of psychopathy. This study was designed to clarify these differences by testing how well the PCL-R and PPI map alternative conceptions of psychopathy. Construct validity metrics were used to compare patterns of associations between psychopathy measures and 14 theory-relevant criterion variables that were observed in a sample of 1,281 offenders-with patterns of associations that were predicted based on alternative psychopathy conceptions. PCL-R total scores were most consistent with Karpman's affective dysfunction-centered secondary conception, and PPI total scores were most consistent with the McCords' lovelessness-based conception. Although similarities emerged at the factor level, the PPI demonstrated higher levels of consistency between theory-based predictions and observed relations than did the PCL-R. These results provide direction for refining measures in future research and interpreting PCL-R and PPI scores in current practice.

Using algorithms to address trade-offs inherent in predicting recidivism.

Although risk assessment has increasingly been used as a tool to help reform the criminal justice system, some stakeholders are adamantly opposed to using algorithms. The principal concern is that any benefits achieved by safely reducing rates of incarceration will be offset by costs to racial justice claimed to be inherent in the algorithms themselves. But fairness trade-offs are inherent to the task of predicting recidivism, whether the prediction is made by an algorithm or human. Based on a matched sample of 67,784 Black and White federal supervisees assessed with the Post Conviction Risk Assessment, we compared how three alternative strategies for "debiasing" algorithms affect these trade-offs, using arrest for a violent crime as the criterion . These candidate algorithms all strongly predict violent reoffending (areas under the curve = 0.71-72), but vary in their association with race (r = 0.00-0.21) and shift trade-offs between balance in positive predictive value and false-positive rates. Providing algorithms with access to race (rather than omitting race or "blinding" its effects) can maximize calibration and minimize imbalanced error rates. Implications for policymakers with value preferences for efficiency versus equity are discussed.

Lost in translation: "Risks," "needs," and "evidence" in implementing the First Step Act.

In this article, we focus on two highly problematic issues in the manner in which the First Step Act of 2018 is being implemented by the Bureau of Prisons: an uncritical separation of "dynamic risks" and "criminogenic needs"; and a spurious reliance on "evidence-based" interventions to reduce recidivism risk. We argue that if the Act is to live up to its promise of being a game-changing development in efforts to reduce crime while simultaneously shrinking mass incarceration, "needs assessment" must be subject to vastly increased empirical attention, variable and causal risk factors must be identified and validly assessed, and interventions to reduce risk must be rigorously evaluated both for their fidelity of implementation and impact on recidivism. Rather than further proliferating programs that ostensibly reduce risk, we believe that serious consideration should be given to the Bureau of Prisons offering one signature, well-established cognitive-behavioral program that can simultaneously address multiple risk factors for moderate and high-risk prisoners.

The limits of human predictions of recidivism.

Dressel and Farid recently found that laypeople were as accurate as statistical algorithms in predicting whether a defendant would reoffend, casting doubt on the value of risk assessment tools in the criminal justice system. We report the results of a replication and extension of Dressel and Farid's experiment. Under conditions similar to the original study, we found nearly identical results, with humans and algorithms performing comparably. However, algorithms beat humans in the three other datasets we examined. The performance gap between humans and algorithms was particularly pronounced when, in a departure from the original study, participants were not provided with immediate feedback on the accuracy of their responses. Algorithms also outperformed humans when the information provided for predictions included an enriched (versus restricted) set of risk factors. These results suggest that algorithms can outperform human predictions of recidivism in ecologically valid settings.

Impact of risk assessment on judges' fairness in sentencing relatively poor defendants.

OBJECTIVE: Use of risk assessment instruments in the criminal justice system is controversial. Advocates emphasize that risk assessments are more transparent, consistent, and accurate in predicting re-offending than judicial intuition. Skeptics worry that risk assessments will increase socioeconomic disparities in incarceration. Ultimately, judges make decisions-not risk assessments. This study tests whether providing risk assessment information interacts with a defendant's socioeconomic class to influence judges' sentencing decisions. HYPOTHESES: Tentatively, socioeconomic status was expected to have a main effect; without an interaction with risk assessment information. METHOD: Judges (N = 340) with sentencing experience were randomly assigned to review 1 of 4 case vignettes and sentence the defendant to probation, jail, or prison. Information in the vignettes was held constant, except the defendant's socioeconomic status and whether risk assessment information was provided. RESULTS: Risk assessment information reduced the likelihood of incarceration for relatively affluent defendants, but the same information increased the likelihood of incarceration for relatively poor defendants. This finding held after controlling for the sex, race, political orientation, and jurisdiction of the judge. CONCLUSIONS: Cuing judges to focus on risk may re-frame how they process socioeconomic status-a variable with opposite effects on perceptions of blameworthiness for past crime versus perceptions of risk for future crime. Providing risk assessment information may have transformed low socioeconomic status from a circumstance that reduced the likelihood of incarceration (by mitigating perceived blameworthiness) to a factor that increased the likelihood of incarceration (by increasing perceived risk). Under some circumstances, risk assessment information may increase sentencing disparities. (PsycINFO Database Record (c) 2020 APA, all rights reserved).

Using Positive Emotion Training With Maltreated Youths to Reduce Anger Bias and Physical Aggression.

Maltreated youths often overinterpret anger in others' emotional expressions, particularly expressions that are ambiguous, and this "anger bias" is associated with aggressive behavior. In the current experiment, we tested the effect of an emotion-training intervention on anger bias and subsequent aggression. Eighty-four youths, ages 8 to 17, who had been removed from home because of maltreatment and had screened positive for aggressive tendencies, served as participants. Over 4 days, youths completed positive emotion training, a computerized program in which youths classify emotional expressions. Youths in the treatment condition received feedback to encourage their recognition of happiness over anger in ambiguous expressions. Physical aggression up to 1 week posttraining was assessed on the basis of self- and staff reports. The intervention was effective in reducing youths' anger bias and somewhat so in reducing aggression-the latter of-which occurred infrequently, limiting power. Results offer direction for developmental research and cost-effective interventions for maltreated youths at risk for aggression and future justice-system involvement.

Examining hypothesized interactive and curvilinear relations between psychopathic traits and externalizing problems in an offender sample using item response-based analysis.

Fearless dominance (FD) generally manifests null to small relations with externalizing problems, leading some researchers to propose alternative paths by which FD features may relate to these problems. The current study provides a test of two possibilities, namely that FD (a) interacts statistically with self-centered impulsivity (SCI) such that FD is associated with externalizing problems only at high levels of SCI and (b) demonstrates curvilinear relations with externalizing problems such that FD is more strongly associated with these problems at high levels. We used a large correctional sample and item-response theory-related statistics to precisely estimate individuals' scores at the extremes of each major psychopathic trait. FD was not significantly associated with externalizing problems in interaction with SCI or at higher levels of FD, suggesting that psychopathic traits linked to boldness are not especially relevant to generalized externalizing behavior. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

An exploration of the symmetry between crime-causing and crime-reducing factors: Implications for delivery of offender services.

Both the Risk-Needs-Responsivity (RNR) and Structured Professional Judgment (SPJ) risk assessment approaches assume that a strong relationship exists between crime-causing and crime reducing factors. Using a probation sample, the present article examines whether crime-causing and crime-reducing factors correspond. Probationers completed questionnaires where they were asked what factors were crime-causing and what factors were crime-reducing. Overall, the relationship between the crime-causing and crime-reducing factors was very weak-even after ruling out potential measurement and methodological artifacts (i.e., internal consistency, item stability, and acquiescent responding). Applied to an individual offender, the results suggest that conducting assessments and recommending interventions need not be bound by assumptions that risk factors for past crime must be targeted to reduce crime. New endeavors to develop causal and idiographic crime-reducing strategies warrant consideration. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Efficiently assessing firm, fair, and caring relationships: Short form of the Dual Role Relationship Inventory.

Many people involved in the justice system and people with serious mental illness are required to participate in psychosocial treatment, whether they want it or not. With these clients, case managers, probation officers, and other providers are tasked with both promoting client recovery (a helping, therapeutic role) and protecting community safety (a controlling, surveillance role). The 30-item revised Dual-Role Relationship Inventory (DRI-R) assesses the quality of provider-client relationships in mandated treatment-and DRI-R based research indicates that firm, fair, and caring relationships (authoritative, not authoritarian) predict better client outcomes. In this study, we developed and validated a short form of the DRI-R-the 9-item DRI-SF-by applying multidimensional item response theory methods to four data sets (N = 815). We simultaneously refined the measure by selecting items that cleanly assessed relationship features (i.e., minimized construct-irrelevant variance from provider traits) and performed similarly across client groups (juveniles and adults; with-and-without mental illness). DRI-SF total scores strongly predict DRI-R total scores (r = .97). The DRI-SF fully represented the DRI-R's range of item difficulties, produced the same three-factor structure, predicted theoretically relevant external covariates as strongly (i.e., groups known to differ in relationship quality, relationship satisfaction ratings, future arrests)-without item bias by sex or race. Moreover, the favorable psychometric properties of the DRI-SF were replicated in a new sample and shown to generalize across provider groups (from probation officers to treatment providers). This newly developed DRI-SF applies to a range of provider-client relationships in mandated treatment-and will benefit practitioners and researchers with ease of administration. (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Exposure to Community Violence and Self-harm in California: A Multilevel, Population-based, Case-Control Study.

BACKGROUND: Self-harm is a leading cause of morbidity and mortality. Exposure to community violence is an important and potentially modifiable feature of the social environment that may affect self-harm, but studies to date are limited in the samples and outcomes examined. METHODS: We conducted a population-based, nested case-control study. Cases were all deaths and hospital visits due to self-harm in California, 2006-2013. We frequency-matched California resident population-based controls from the American Community Survey to cases on age, gender, race/ethnicity, and year of survey/injury. We assessed past-year community violence using deaths and hospital visits due to interpersonal violence in the community of residence. We estimated risk difference parameters that were defined to avoid extrapolation and to capture associations between changes in the distribution of community violence and the population-level risk of self-harm. RESULTS: After adjustment for confounders, setting past-year community violence to the lowest monthly levels observed within each community over the study period was associated with a 30.1 (95% confidence interval = 29.6, 30.5) per 100,000 persons per year lower risk of nonfatal self-harm but no difference in the risk of fatal self-harm. Associations for a parameter corresponding to a hypothetical violence prevention intervention targeting high-violence communities indicated a 5% decrease in nonfatal self-harm at the population level. In sensitivity analyses, results were robust. CONCLUSIONS: This study strengthens evidence on the relationship between community violence and self-harm. Future research should investigate reasons for differential associations by age and gender and whether community violence prevention programs have meaningful impacts on self-harm.

Comparing Costs of Traditional and Specialty Probation for People With Serious Mental Illness.

OBJECTIVE: Specialty mental health probation reduces the likelihood of rearrest for people with mental illness, who are overrepresented in the justice system. This study tested whether specialty probation was associated with lower costs than traditional probation during the two years after placement in probation. METHODS: A longitudinal, matched study compared costs of behavioral health care and criminal justice contacts among 359 probationers with mental illness at prototypic specialty or traditional agencies. Compared with traditional officers, specialty officers supervised smaller caseloads, established better relationships with supervisees, and participated more in treatment. Participants and officers were interviewed, and administrative databases were integrated to capture service use and criminal justice contacts. Unit costs were attached to these data to estimate costs incurred by each participant over two years. Cost differences were estimated by using machine-learning algorithms combined with targeted maximum-likelihood estimation (TMLE), a double-robust estimator that accounts for associations between confounders and both treatment assignment and outcomes. RESULTS: Specialty probation cost $11,826 (p<.001) less per participant than traditional probation, with overall savings of about 51%. Specialty and traditional probation did not differ in criminal justice costs because the additional costs for supervision of specialty caseloads were offset by reduced recidivism. However, for behavioral health care, specialty probation cost an estimated $14,049 (p<.001) less per client than traditional probation. Greater outpatient costs were more than offset by reduced emergency, inpatient, and residential costs. CONCLUSIONS: Well-implemented specialty probation yielded substantial savings-and should be considered in justice reform efforts for people with mental illness.

What features of psychopathy might be central? A network analysis of the Psychopathy Checklist-Revised (PCL-R) in three large samples.

Despite a wealth of research, the core features of psychopathy remain hotly debated. Using network analysis, an innovative and increasingly popular statistical tool, the authors mapped the network structure of psychopathy, as operationalized by the Psychopathy Checklist-Revised (PCL-R; Hare, 2003) in two large U.S. offender samples (nNIMH = 1559; nWisconsin = 3954), and 1 large Dutch forensic psychiatric sample (nTBS = 1937). Centrality indices were highly stable within each sample, and indicated that callousness/lack of empathy was the most central PCL-R item in the 2 U.S. samples, which aligns with classic clinical descriptions and prototypicality studies of psychopathy. The similarities across the U.S. samples offer some support regarding generalizability, but there were also striking differences between the U.S. samples and the Dutch sample, wherein the latter callousnesss/lack of empathy was also fairly central but irresponsibility and parasitic lifestyle were even more central. The findings raise the important possibility that network-structures do not only reflect the structure of the constructs under study, but also the sample from which the data derive. The results further raise the possibility of cross-cultural differences in the phenotypic structure of psychopathy, PCL-R measurement variance, or both. Network analyses may help elucidate the core characteristics of psychopathological constructs, including psychopathy, as well as provide a new tool for assessing measurement invariance across cultures. (PsycINFO Database Record

Comparing Public Safety Outcomes for Traditional Probation vs Specialty Mental Health Probation.

Importance: Probation is a cornerstone of efforts to reduce mass incarceration. Although it is understudied, specialty probation could improve outcomes for the overrepresented group of people with mental illness. Objective: To test whether specialty probation yields better public safety outcomes than traditional probation. Design, Setting, and Participants: A longitudinal observational study with group matching on age, sex, race/ethnicity, probation time, and offense at 2 urban agencies that exemplify specialty and traditional probation. Enrollment began October 19, 2005; follow-up data were complete January 26, 2013. Participants were 359 diverse probationers with serious mental health problems and functional impairment. Probationers and officers were assessed 3 times (for probationers, 6-month retention, 315 of 359 [88%]; 12-month retention 304 of 359 [85%]) and follow-up records were obtained. Machine learning algorithms were combined with a targeted maximum likelihood estimation, a double robust estimator that accounts for associations between confounders and both treatment assignment and outcomes. Statistical analysis was conducted from January 1, 2016 to May 5, 2017. Interventions: Specialty probationers were assigned to small, homogeneous caseloads supervised by experts. Prior data indicate that specialty officers had better relationships with probationers, participated more in probationers' treatment, and relied more on positive compliance strategies than traditional officers. Main Outcomes and Measures: Violence during 1 year, determined by probationer report, officer report, and records, and rearrest during a period of 2 to 5 years, according to Federal Bureau of Investigation records. Results: Participants were 183 specialty (73.8% of 248 eligible) and 176 traditional (56.6% of 311 eligible) probationers (205 men and 154 women; mean [SD] age, 36.9 [10.6]). Although specialty probation had no significant effect on violence (odds ratio, 0.97; 95% CI, 0.69-1.36), the odds of rearrest were 2.68 times higher for traditional probationers than for specialty probationers (95% CI, 1.86-3.84; P < .001). At 2 years, estimated probabilities of rearrest were 28.6% for specialty probationers and 51.8% for traditional probationers. Survival analyses indicate that arrest effects endured up to 5 years. Conclusions and Relevance: Although it did not specifically reduce violence, well-implemented specialty probation appears to be effective in reducing general recidivism. Reform efforts for people with mental illness could leverage probation-a ubiquitous and revitalized node of the justice system.