Effectiveness of criminal attitudes-based intervention in reducing recidivism in young offenders

Abstract

Criminal attitudes are a well-known risk factor of criminal behaviour. Various corrective institutions have adopted interventions that specifically target offenders’ criminal attitudes in hopes to reduce recidivism. Although past literature supports the effectiveness of criminal attitudes-based intervention in reducing recidivism, whether this relationship can be attributed to a decrease in criminal attitudes is still unknown. A common measure of criminal attitudes is the Criminal Sentiments Scale-Modified (CSS-M; Shields & Simourd, 1991). Due to its predictive validity for the Western population, practitioners have begun to utilize the CSS-M in the Asian context. The Hong Kong Correctional Services (HKCS) translated the CSS-M into Traditional Chinese to assess offenders’ level of criminal attitudes. However, the translated scale has yet to be validated. Furthermore, the factor structure underlying the CSS-M is subjected to ongoing debate. For an increasingly adopted measure of criminal attitudes, it is imperative to further explore the psychometrics of the CSS-M. The aims of the current thesis are twofold. First, to elucidate the mechanism behind the relationship between criminal attitudes-based intervention and recidivism. Second, to validate the Traditional Chinese translation of the CSS-M and explore the factor structure underlying this measure. Study one addresses the first aim of this thesis by utilizing secondary data that was collected from young offenders by the HKCS. Recidivism records and two waves of criminal attitudes scores were used for analysis. Results demonstrate that a decrease in offender criminal attitudes significantly mediates the relationship between criminal attitudes-based intervention and recidivism. Criminal attitudes-based intervention significantly lowers offenders’ level of criminal attitudes, which then effectively reduces their likelihood of recidivism. Study two utilizes two university samples and two offender samples to address the second aim of this thesis. Both offender samples and one university sample completed the Traditional Chinese version of the CSS-M. The remaining university sample completed the original English version. Confirmatory factor analysis and measurement invariance was conducted to explore the psychometrics underlying both versions of the CSS-M. Results indicate that the correlated traits models that were proposed by previous studies do not fit our data well. Rather, a three-factor structure nested within a bifactor model achieved acceptable fit for all four samples. Results from bifactor analysis suggests the CSS-M is multidimensional in nature but can be scored and interpreted as a composite measure. Configural invariance was also achieved between the English and Traditional Chinese versions of the CSS-M. There are multiple practical implications from the current thesis findings. The continued use of criminal attitudes-based interventions in reducing recidivism is supported. The Traditional Chinese version of the CSS-M was also validated for use in the Hong Kong population. Finally, findings from bifactor analysis lend support to the continued use of composite scoring by practitioners as well as advised future researchers to consider the multidimensionality of the CSS-M in their analyses.