Curriculum-based measurement procedures to develop multiple-skill mathematics computation probes: Evaluation of random and stratified stimulus-set arrangements

Curriculum-based measurement (CBM) procedures have been developed to assess mathematics computation (M-CBM). Recent research has demonstrated that the psychometric characteristics of measurement outcomes vary as a function of administration procedures. This study examined the effect of an alternate set of procedures to construct more consistent multiple-skill M-CBM computation probes. Researchers compared two probe construction techniques: randomized stimulus-set arrangement and stratified stimulus-set arrangement. The results of repeated multiple-skill M-CBMs that were administered to students within five primary grades (N = 219) suggest that stratified construction of multiple-skill M-CBM computation probes is likely to yield more generalizable and dependable measurement outcomes. This result was especially robust within the upper primary grades. The convergence of these findings with previous research is discussed along with implications for practice and future research.