Predictive validity of curriculum-based measurement and teacher ratings of academic achievement

Abstract

Two alternative universal screening approaches to identify students with early learning difficulties were examined, along with a combination of these approaches. These approaches, consisting of (a) curriculum-based measurement (CBM) and (b) teacher ratings using Performance Screening Guides (PSGs), served as predictors of achievement tests in reading and mathematics. Participants included 413 students in grades 1, 2, and 3 in Tennessee (n=118) and Wisconsin (n=295) who were divided into six subsamples defined by grade and state. Reading and mathematics achievement tests with established psychometric properties were used as criteria within a concurrent and predictive validity framework. Across both achievement areas, CBM probes shared more variance with criterion measures than did teacher ratings, although teacher ratings added incremental validity among most subsamples. PSGs tended to be more accurate for identifying students in need of assistance at a 1-month interval, whereas CBM probes were more accurate at a 6-month interval. Teachers indicated that (a) false negatives are more problematic than are false positives, (b) both screening methods are useful for identifying early learning difficulties, and (c) both screening methods are useful for identifying students in need of interventions. Collectively, these findings suggest that the two types of measures, when used together, yield valuable information about students who need assistance in reading and mathematics.