Universal Screening of Students' Classroom Behaviors: An Investigation of the Validity and Classification Accuracy of the Behavior Screening Checklist III with Kindergarten Students

Abstract

The purpose of this study was to investigate the Behavior Screening Checklist III (BSC) as a universal behavior screening measure with a sample of kindergarten students. A total of six research questions were examined using bivariate correlation analyses, multiple regression analyses, and receiving operating characteristic (ROC) curve analyses. This was the first study conducted using a ROC curve analyses to investigate classification accuracy indicators of the BSC. Preliminary evidence of the BSC indicates it is a reliable and valid screening measure to use in the schools for identification of students at-risk for later behavior problems. However, the current literature is limited to correlation coefficients and multiple regression coefficients.

Results from the present study concluded the BSC yields consistent teacher ratings over time. Convergent and predictive validity were also supported, using the SDQ as a criterion measure. Multiple regression analyses revealed BSC scores accounted for a statistically significant amount of variance in end of year SDQ scores over and above student ODRs and number of absences. This finding occurred across all three multiple regression models. Winter ORDs were also found to be a statistically significant predictor variable of Spring ODRs. As for classification accuracy, the BSC was able to differentiate between those at-risk from those not at-risk based on AUC. The BSC exhibits good to excellent sensitivity but poor specificity. Of the predictor variables that were statistically significant, Spring BSC scores resulted in the strongest classification accuracy based on the AUC statistics. As this was the first study to explore classification accuracy of the BSC, it provides emerging support for utilizing the BSC as a universal behavior screening measure.