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

NA (site), NA (city),

Abstract:

From the abstract: "This study examined the predictive validity of Star Early Literacy (SEL), a computer-adaptive assessment for measuring kindergarten reading skills. Findings revealed that the results of SEL assessments administered during fall, winter, and spring of kindergarten were moderate and statistically significant predictors of year-end reading and reading-related skills, and 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. Additionally, SEL cut scores demonstrated 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."

Citation: Clemens, N. H., Hagan-Burke, S., Luo, W., Cerda, C., Blakely, A., Frosch, J., Gamez, B., & Jones, M. (2015). The predictive validity of a computer-adaptive assessment of kindergarten and first-grade reading skills. School Psychology Review, 44(1), 76-97.

The full report is available online.

Publication Date:
10/21/2014

Updated: 04/06/2015



For information about research not available electronically, please email us at research@renaissance.com.