The Predictive Validity of a Computer-Adaptive Assessment of Kindergarten and First-Grade Reading Skills

This study examined the predictive validity of a computer-adaptive assessment for measuring kindergarten reading skills using the STAR Early Literacy (SEL) test. The findings showed that the results of SEL assessments administered during the fall, winter, and spring of kindergarten were moderate and statistically significant predictors of year-end reading and reading-related skills, and they explained 35% to 38% of the variance in a latent variable of word-reading skills. Similar results were observed with a subsample of 71 participants who received follow-up assessments in first grade. End-of-kindergarten analyses indicated that, when added as predictors with SEL, paper-based measures of letter naming, letter-sound fluency, and word-reading fluency improved the amount of explained variance in kindergarten and first-grade year-end word-reading skills. Classification-accuracy analyses found that the SEL literacy classifications aligned with word-reading skills measured by paper-based assessments for students with higher SEL scores, but less alignment was found for students with lower SEL scores. In addition, SEL cut scores showed problematic accuracy, especially in predicting outcomes at the end of first grade. The addition of paper-based assessments tended to improve accuracy over using SEL in isolation. Overall, SEL shows promise as a universal screening tool for kindergarten reading skills, although it may not yet be able to completely replace paper-based assessments of early reading.