Predictive Validity Test of the Adolescent Domain Screening Inventory

Abstract

Objective: This study assesses the Adolescent Domain Screening Inventory (ADSI) to identify adolescents at high risk of substance use. Method: The sampling frame consisted of 26,781 surveys, and a secondary analysis was conducted. A random 25% sample was used, leaving 6,661 cases. Binary logistic regressions were run to determine the predictive validity of the total ADSI and its 4 domains. Results: The ADSI correctly predicted high risk 93% of the time and problematic use 84.3% of the time. Conclusions: The results indicate the ADSI should be considered for use to assess adolescents for high risk status for substance use, or to identify those already engaged in substance use.

Keywords:

• assessment
• predictive validity
• prevention
• risk and protective factors
• substance use

Notes

Note. High Risk: Specificity = 5,133/(5,133 + 194) = 96.4%, Sensitivity = 1,031/(269 + 1,031) = 79.3%, False Positive = 194/(194 + 1,031) = 16%, False Negative = 269/(5,133 + 269) = 5%. Problem Use: Specificity = 4,586/(4,586 + 306) = 93.7%, Sensitivity = 643/(668 + 643) = 49%, False Positive = 306/(306 + 642) = 32%, False Negative = 668/(4,586 + 668) = 13%.

Note. High Risk: Cox and Snell R ² = .484, Nagelkerke R ² = .770. Problem Use: Cox and Snell R ² = .295, Nagelkerke R ² = .458. Dependent reference categories: Low Risk, No Problem Use.

Note. All relationships are significant at the p < .001 level.