Title

Diagnostic Accuracy of Direct Behavior Rating as a Screener of Elementary School Students

Date of Completion

January 2011

Keywords

Education, Educational Psychology|Psychology, Psychometrics

Degree

Ph.D.

Abstract

In response to controversy regarding the use of the "wait-to-fail" model toward identifying students at-risk for social behavior problems, a call has been made for schools to utilize preventive models of service delivery. It has been suggested that these approaches should include universal screening procedures that are usable and contextually appropriate, as well as technically adequate. Unfortunately, few universal screening approaches have been found to meet all requirements to a sufficient degree. As such, the current study sought to expand upon existing screening-related Direct Behavior Rating Single Item Scale (DBR-SIS) research through an evaluation of diagnostic accuracy and concurrent validity. Results of receiver operating characteristic (ROC) curve analyses indicated that each DBR-SIS (i.e., disruptive behavior [DBR-DB], academic engagement [DBR-AE], and compliance [DBR-CO]), as well as a DBR-SIS factor scale (DBR-Factor) was a moderately to highly accurate predictor of behavioral risk (as defined by the Behavioral and Emotional Screening System; Kamphaus & Reynolds, 2007a). Furthermore, each was found to be no more or less accurate than the Social Skills Improvement System Performance Screening Guide (Elliott & Gresham, 2007a) to a statistically significant degree. ROC curve analyses were also used to identify optimal universal screening cut scores on each DBR scale (DBR-DB=2, DBR-AE=8, DBR-CO=9, and DBR-Factor=0). The disproportionately high false positive rate associated with each was reduced through the simultaneous use of multiple DBR-SIS to determine student risk. In particular, the DBR-AE+DBR-CO scale combination was associated with the best balance between true positives and true negatives. Finally, correlational analyses reflected the concurrent validity of each DBR single item and factor scale. Limitations of the current study, implications for practice, and directions for future research are each discussed herein. ^

COinS