Academic screening in middle school: Exploring bidirectional and intraindividual relations in student growth in reading and math

Abstract

Early adolescence is a pivotal time for academic development; however, the vast majority of research on reading and math development within a multi-tiered system of support has been conducted among elementary students. Using triannual (Fall, Winter, Spring) academic screening data, we examined the transactional development of reading and math skills among sixth and seventh grade students (N = 1,693) using Bayesian longitudinal structural equation modeling (BSEM). We find equivocal support for four of the five types of longitudinal SEMs tested (dual-change, linear change, proportional change, latent curve with structured residuals, and random-intercepts cross-lagged models). Stable between-person differences in math and reading (i.e., latent/random intercepts) are strongly correlated regardless of the modeling approach (r = .70-.77), consistent extensive prior research in this area. However, correlated growth processes at the between-person level and within-person transactional relations of math and reading were inconsistent across models. We discuss the practical implications of these findings as well as the methodological issues with detecting within-person instructional response in triannual screening.