Surveillance of Internalizing Behaviors: A Reliability and Validity Generalization Study of Universal Screening Evidence

Abstract

Due to early onset and rising prevalence rates, there is an established need for early identification of youth demonstrating symptoms of internalizing disorders (e.g., anxiety and depression). Accordingly, calls have been made for the use of universal screening tools to identify internalizing behavior risk. Current research has spoken to the adequacy of individual screeners, but researchers have yet to quantitatively synthesize the broader internalizing screening literature. The purpose of the current study was to systematically review the literature and conduct a reliability and validity generalization meta-analysis of available psychometric evidence. This meta-analysis identified 40 studies in which 18 distinct internalizing behavior screeners were examined. Of interest were both (a) broadband screeners, which included internalizing items along with items specific to other areas (e.g., externalizing problems), and (b) narrowband screeners, which included internalizing items alone. Across examined studies, mean coefficient alpha reliability estimates were equal to .84 for broadband screeners and .78 for narrowband screeners. For broadband screeners, mean validity correlation coefficients were equal to .66 when predicting other internalizing measures and .51 when predicting other criteria (e.g., extant academic and behavioral data). For narrowband screeners, mean validity coefficients were equal to .45 when predicting other internalizing measures and .26 when predicting other criteria. Interestingly, moderator analyses indicated neither reliability nor validity was moderated by any study or measurement characteristics. Taken together, results supported the psychometric defensibility of both broadband and narrowband internalizing screeners. Limitations of the study and implications for research are discussed, including recommendations for additional diagnostic accuracy research.