Applying Generalizability Theory for Making Quantitative RTI Progress-Monitoring Decisions

Abstract

Language in the Individuals With Disabilities Education Improvement Act (IDEIA) allows the use of response-to-intervention (RTI) methodology in the identification of specific learning disabilities. However, there is no consensus on decision rules using curriculum-based measurement of oral reading fluency (CBM-R) for defining responsiveness. The purpose of this article is to describe how to apply generalizability (G) theory for making quantitative high-stakes RTI decisions. A sample of 68 first-grade students (36 females and 32 males) was randomly administered the first-grade Dynamic Indicators of Basic Early Literacy Skills oral reading fluency probes and correct words per minute (CWPM) were calculated for conducting both a generalizability (G) study and a decision (D) study using G theory. The results showed that the average raw scores (CWPM) across 20 probes encompassed a wide range from a low of 69 CWPM to a high of 88 CWPM. In addition, 90.2% of the variance was due to student reading skill, 7.0% was due to unaccounted sources of error, and only 2.8% was due to passage or probe variability. Also, the increasing number of probes administered (from 1 to 9) considerably reduced the values of standard error of measurement and increased the reliability coefficients. An example is provided to demonstrate how to apply G theory in using CBM-R data to guide psychometrically valid RTI decisions at the individual student level.