A Re-Examination of the Subscale Factor Structure of the Service Needs Inventory and the Development of Clinical Cut Scores



Journal Title

Journal ISSN

Volume Title



Optimal services for reducing the criminal recidivism risk for justice-involved persons with mental illness should target both criminogenic risk factors and mental health functioning (Gross & Morgan, 2013; Morgan et al., 2010; Skeem et al. 2011); however, prior to the development of the Services Needs Inventory (SNI), there was no singular measure that assessed both dimensions. The goal of the current study was to re-examine the factor structure of the SNI given the previous inconsistent findings (Olafsson, 2020, Olafsson et al., 2021), and then identify the optimum cut scores for subscales that display evidence for acceptable model fit. To achieve this, a clinical sample and a nonclinical sample were collected. The clinical sample was comprised of justice-involved individuals recruited from 3 jails and 1 residential treatment center in southwest United States. The nonclinical sample was comprised of individuals recruited through Texas Tech University’s SONA system and undergraduate psychology classes who reported no current or past justice involvement. McDonald Omegas were calculated for each SNI subscale within both samples. A total series of 16 independent confirmatory factor analyses (CFA) where a CFA was conducted on each of the SNI subscales within both the clinical and nonclinical samples. Support for acceptable model fit was found across the clinical and nonclinical samples for seven out of the eight SNI (Substance Use, Criminal History, Negative Affect, Psychiatric Symptomology, Social Functioning, and Trauma History). A series of receiver operator curve analyses were conducted to identify an optimum cut score and provide evidence for the validity of this cut score. Cut scores with a specificity close to 90% were considered optimal (Sweet et al., 2021). Cut scores at this specificity reflect the risk principle of the Risk-Needs-Responsivity framework (Andrews et al., 1990) by identifying those at the higher service need for risk reduction interventions. Additionally, it would minimize the financial strain on the criminal justice system by minimizing false positives and maximizing true negatives. These cut scores were validated on a replication sample, a mental health-only sample, and a justice-involved-only sample. The findings of this study support the use of seven subscales of the SNI as a research tool to identify individuals who are at elevated risk of criminogenic risk and mental health needs, and for limited clinical use (e.g., for treatment planning). However, additional examination is needed before the SNI has psychometrical properties (e.g., reliability, model fit, and predictive validity) sufficient for clinical determinations (e.g., treatment placement determinations).

Embargo status: Restricted to TTU community only. To view, login with your eRaider (top right). Others may request the author grant access exception by clicking on the PDF link to the left.



Recidivism, Risk Assessment, Factor Structure, Cut Score, Service Needs Inventory