Using curriculum-based measurement of reading to inform practice and improve student achievement

Abstract

An integral component of a successful system of prevention and early intervention is timely, informative assessment. Curriculum-based measurement (CBM) provides an approach to assessment that is focused on individual progress over short periods of time. CBM is used increasingly for progress monitoring, benchmarking, and informing eligibility decisions. The purpose of the current studies was to investigate two additional uses of CBM of reading (R-CBM) – documenting summer learning loss and predicting performance on a state reading test. Study 1 documented change in R-CBM scores from spring to fall with specific attention to individual factors such as grade level, family income level, special education (SPED) status, and English language learner (ELL) status. For this study, 317 students in Grades 2 – 5 were assessed using the Dynamic Indicators of Basic Early Literacy Skills (DIBELS) Oral Reading Fluency (ORF) measure. Repeated measures-analysis of variance (RM-ANOVA) modeled overall change in ORF scores by grade, as well as change based on individual factors. Students in Grades 2 and 3 evidenced overall summer loss in ORF scores. In addition, students in Grade 2 showed differential loss based on family income level and SPED status. Study 2 evaluated the relationship between R-CBM and the Georgia reading test by examining the accuracy of predictions for different ORF benchmarks and the bias of predictions based on subgroup membership. Scores on DIBELS ORF for 1,374 students in Grades 2 – 5 were used to predict outcomes the Georgia reading test. Cut scores for ORF were generated using logistic regression and receiver operator characteristic (ROC) curve analysis, following the procedure outlined in Silberglitt and Hintze (2005). The generated cut scores were compared to the published DIBELS ORF benchmarks based on diagnostic efficiency. The generated cut scores were lower than the suggested DIBELS ORF benchmarks and had improved diagnostic efficiency. The potential for bias based on family income level, ethnic minority status, and ELL status was investigated using regression and RM-ANOVA. These results support the broadening application of R- CBM and the need for additional research. Findings are discussed relative to prevention and intervention efforts and implications for educational policy.