Statistical learning methods and cross-cultural fairness: Trade-offs and implications for risk assessment instruments.

The use of statistical learning methods has recently increased within the risk assessment literature. They have primarily been used to increase accuracy and the area under the curve (AUC, i.e., discrimination). Processing approaches applied to statistical learning methods have also emerged to increase cross-cultural fairness. However, these approaches are rarely trialed in the forensic psychology discipline nor have they been trialed as an approach to increase fairness in Australia. The study included 380 Aboriginal and Torres Strait Islander and non-Aboriginal and Torres Strait Islander males assessed with the Level of Service/Risk Needs Responsivity (LS/RNR). Discrimination was assessed through the AUC, and fairness was assessed through the cross area under the curve (xAUC), error rate balance, calibration, predictive parity, and statistical parity. Logistic regression, penalized logistic regression, random forest, stochastic gradient boosting, and support vector machine algorithms using the LS/RNR risk factors were used to compare performance against the LS/RNR total risk score. The algorithms were then subjected to pre- and postprocessing approaches to see if fairness could be improved. Statistical learning methods were found to produce comparable or marginally improved AUC values. Processing approaches increased several fairness definitions (namely xAUC, error rate balance, and statistical parity) between Aboriginal and Torres Strait Islanders and non-Aboriginal and Torres Strait Islanders. The findings demonstrate that statistical learning methods may be a useful approach to increasing the discrimination and cross-cultural fairness of risk assessment instruments. However, both fairness and the use of statistical learning methods encompass significant trade-offs that need to be considered. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

The cross-cultural fairness of the LS/RNR: An Australian analysis.

OBJECTIVE: Cross-cultural research into risk assessment instruments has often identified comparable levels of discrimination. However, cross-cultural fairness is rarely addressed. Therefore, this study explored the discrimination and fairness of the Level of Service/Risk, Need, Responsivity (LS/RNR) within a sample of Aboriginal and Torres Strait Islander and non-Aboriginal and Torres Strait Islander males. HYPOTHESES: We hypothesized that discrimination would not be significantly different for Aboriginal and Torres Strait Islander individuals and non-Aboriginal and Torres Strait Islander individuals. We further hypothesized that some fairness definitions would be unsatisfied. METHOD: The study included 380 males (Aboriginal and Torres Strait Islander, n = 180) from Australia. Discrimination was assessed with the area under the curve (AUC) and cross AUC (xAUC). To determine fairness, error rate balance, calibration, predictive parity, and statistical parity were used. RESULTS: The discrimination of the LS/RNR was not statistically different (p = .61) between groups. The xAUC identified disparities (p < .001), with the LS/RNR being unable to discriminate between Aboriginal and Torres Strait Islander nonreoffenders and non-Aboriginal and Torres Strait Islander reoffenders (xAUC = .46, 95% CI [.35, .57]). Disparities among certain fairness definitions were identified, with Aboriginal and Torres Strait Islander individuals scoring higher on the LS/RNR (d = 0.52) and nonreoffenders being classified as high risk more often. CONCLUSIONS: The findings suggest that the LS/RNR may not be a cross-culturally fair risk assessment instrument for Australian individuals, and standard discrimination indices with comparable levels do not imply that a risk assessment instrument is cross-culturally fair. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Exploring Differences in the Experiences, Perceptions and Reporting of Violent Incidents in Australia by Country of Birth.

Australia's fast-growing migrant population encompasses many groups from culturally and linguistically diverse backgrounds. It is well documented that these groups experience varying pre- and post-migratory challenges. Despite this knowledge, little is known about the extent to which these groups experience and perceive violence and how or whether they seek assistance after such incidents. It is important to identify any potential discrepancies to ensure that services can provide the most targeted supports to victims. Data were collected from the 2016 Australian Bureau of Statistics Public Safety Survey. Using chi-square tests, prevalence, experiences, and perceptions of violence occurring after the age of 15, postincident support, reporting behaviors, and health variables related to violent incidents were compared across three cultural groups arranged by region of birth: born in Australia (BIA), born overseas in main English-speaking countries (BNMESC), and born overseas in non-English-speaking countries (BOC). BOC individuals reported much lower rates of violent victimization compared to BIA and BMSEC individuals. More than two-thirds of each cultural group did not report their most recent experience of violence to police. Violence was most commonly experienced at home, although a higher proportion of BIA individuals experienced violence at an entertainment venue, and a higher proportion of BOC individuals experienced violence outside (i.e., in the street). The contribution of Alcohol/Substances was much higher for BIA and BMESC compared to BOC individuals. BOC individuals experienced more anxiety post-incident, while more BIA individuals sustained physical injuries. A similar proportion of each group sought assistance post-incident, however, more BOC individuals had never told anyone about the incident. Perceptions of the violent incident were generally similar across groups, though fewer BOC individuals perceived the incident to be a crime. Some differences were apparent across cultural groups regarding the prevalence, experiences and reporting of violent incidents. Implications and future research directions are discussed within.

Closing the (incarceration) gap: assessing the socio-economic and clinical indicators of indigenous males by lifetime incarceration status.

BACKGROUND: Approximately 1 in 5 to 1 in 6 Indigenous Australian males are currently imprisoned or have previously been imprisoned. Recent work has also pointed to a widening socio-economic gap within the Indigenous population. Given the myriad social, wellbeing and environmental risk factors associated with justice-involvement, it is conceivable that incarceration may contribute to the increasing disparities found within the Indigenous population. This study aimed to explore the presence and extent of an 'incarceration gap' within the Indigenous population and to uncover which social factors characterise the disparity. METHODS: The study utilised data from the 2014-5 National Aboriginal and Torres Strait Islander Social Survey (NATSISS). A number of socio-economic, environmental and clinical factors were compared by life-time incarceration status. Chi-square tests were used to examine the association between incarceration status and each of the comparison variables. RESULTS: Disparities were observed within the Indigenous Australian population across a number of important health and socio-economic markers by incarceration status - the most pronounced being for educational obtainment - year 10 completion (Never incarcerated 73%, Ever incarcerated 50%), labour force participation (Never incarcerated 56%, Ever incarcerated 26%) and drug/alcohol problems (Never incarcerated 7%, Ever incarcerated 29%). Never-incarcerated Indigenous males yielded aggregate proportions across numerous variables that approximated or matched general Australian population estimates. CONCLUSIONS: There appears to be evidence for a substantial 'incarceration gap' within the Indigenous Australian population.