Gender-based measurement invariance of the Substance Use Risk Profile Scale

Abstract

Gender may influence the psychometric properties of psychological and affect-related measures. The American Psychological Association has made recommendations to conduct tests of group-level measurement invariance (MI) before comparing scores between groups. Gender-based measurement invariance has been examined in many well-known psycho-social scales such as the CES-D and the Big Five Inventory. The Substance Use Risk Profile is a 23-item scale measuring affective- and personality-related traits known to increase risk for substance use, with 4 dimensions: anxiety sensitivity, hopelessness, sensation seeking and impulsivity. Despite similarities in the constructs assessed by the SURPS, the CES-D and the Big Five Inventory, gender-based measurement invariance of the SURPS has not yet been published. Multi-group confirmatory factor analysis was used to assess the measurement invariance of the four dimensions of the SURPS across gender. MI was conducted with M-Plus 6.2 using a 2-step analysis for ordinal variables suggested by Muthén and Muthén, and model fit was assessed using the comparative fit index (CFI) criteria recommended by Cheung and Rensvold. A single group confirmatory factor analysis (CFA) was also conducted. The sample was composed of 1352 adolescents (56% female, mean age of 14years) participating in the BC Adolescent Substance Use Survey, an online survey capturing substance use and psychosocial trends in secondary students across British Columbia, Canada. Measurement invariance across gender was demonstrated for the SURPS (ΔCFI=0.003), and the single group CFA supported a four-dimensional structure for the SURPS items (CFI=0.92, RMSEA=061, 95% CI=0.058–0.065).