Doctor of Philosophy (PhD)
Longitudinal high-stakes mathematics test scores were studied to determine the stability of individual and skill-level group (e.g.,Failing, Proficient, Advanced) scores across time, the contribution of high-stakes reading and mathematics scores when predicting later high-stakes mathematics scores, latent growth trajectories, the number of latent class growth trajectories present from third to sixth grade, and decision statistics. Participants were 1,414 students, comprised of three cohorts attending a suburban public school district in Southeastern Pennsylvania. Results indicated earlier high-stakes reading and mathematics scores predicted later high-stakes mathematics scores well. Skill-level
trajectories demonstrated failing students improved significantly between third and sixth grade but were unable to catch up to proficient-and advanced-level
students. Earlier high-stakes reading scores accounted for little variance over and above earlier mathematics scores when predicting later high-stakes mathematics scores. Latent growth analyses suggested participants improved their scores across time and that gender, third grade reading level, IEP
status, and school attended were important predictors of mathematics growth trajectories. Latent class analyses yielded three prototypical growth patterns, roughly high, medium and low. Decision statistics produced appropriate sensitivity and specificity statistics but also produced a high number of false-positives and an important number of false-negatives. The current study is one of the first studies to use longitudinal high-stakes tests scores outside of the value-added literature, which has failed to publish basic findings about students’ latent growth and latent class trajectories using high-stakes data. Results of the current study extend previous results found between preschool and third grade using high-stakes test scores.
Mack, Thomas C. "Predicting and Tracking Mathematics Growth Using High - Stakes Test Scores: An Examination of Latent Growth and Latent Class Models." PhD diss., Bryn Mawr College, 2014.