Predictive Validity of a Computerized Adaptive Test in Early Mathematics

Children who perform poorly in mathematics continue to do so over time without intervention. Given the importance of early skill development for later achievement, there is a need for research investigating the predictive validity of mathematical assessments used to identify students at-risk for later difficulties, including curriculum-based measurement (CBMs) and computer adaptive tests (CATs). We examined the predictive validity of i-Ready, a CAT, and the relative value of adding early numeracy CBMs (beginning quantity discrimination, number identification fluency, and next number fluency) in predicting broad math performance at the end of kindergarten and first grade. Data from 256 kindergarten students in a Northeastern suburban school district were collected over the span of two years. Results support the predictive validity of fall and winter kindergarten i-Ready scores in predicting performance on the mathematical reasoning subtest on the Stanford Achievement Test at the end of kindergarten, computational fluency at the end of grade 1 and earlyMath composite scores from FastBridge Learning at the end of grade 1.The proportion of the variance accounted for in regression models did not meaningfully change after adding CBM performance when predicting outcomes on the SAT (Δ R2 = .03 to .04). Early numeracy CBMs were helpful in improving predictions of computational fluency (Δ R2 = .15 to .22), while mixed results were observed when predicting earlyMath composite scores (Δ R2range = .07 to .17). Results are discussed in the context of underscoring considerations for increasing the efficiency of early mathematics assessment procedures in school settings.