A Longitudinal Study of Students Academic Achievement Change: a Multilevel Structural Equation Model From a Multiple Group Perspective.
Date of Award
Doctor of Philosophy (PhD)
Educational Leadership, Research and Counseling
Chad D. Ellett
The focus of this study was three-fold: First, to determine if the growth parameters (intercepts and slopes) in mathematics and language were related within domains and ethnicity. Second, to determine if the pattern of interrelationships among the individual achievement growth parameters were the same for African American and White students. Third, to establish whether there existed discernible patterns variability in academic growth parameters within each ethnicity over time. The study employed a three-wave longitudinal panel design, with data drawn from Louisiana State Department of Education. The data were then analyzed utilizing both a multilevel structural equation methodology and hierarchical linear modeling in conjunction with individual growth trajectories. The subsets of students involved were African American and White students who had complete records for grades 4, 6 and 7. Based on this criterion, 26,051 (African Americans = 11,627, Whites = 14,424) students were sampled. The grade 4 test scores, recorded in Californian Achievement Test (CAT/5) were converted to Iowa Test of Basic Skills (ITBS) so that scores for all grade levels were in one scale and vertically equated, to enable students' achievement performance to be compared over time. The major findings of the study showed that: (1) students vary significantly in knowledge of mathematics at entry into grade 4 and that White students overall initial status in mathematics was higher than that of African American students, (2) language intercepts for the two groups were statistically significant, signifying language knowledge differences at grade 4, (3) mathematics overall slope for the two groups of learners were positive and significantly different from zero, (4) language overall rates of learning within ethnicity were significantly different from zero, (5) the correlation coefficients of the slope and initial status for each domain and within each ethnicity were not statistically significant and (6) variance estimates for language and mathematics slopes were significantly different from zero and that variances increase at lower grade levels as students advance in school from grade 4 through grade seven. Major findings and conclusions of the study are discussed in view of their implications for future research, measurement theory, research design methodology and practice.
Rugutt, John Kipngeno, "A Longitudinal Study of Students Academic Achievement Change: a Multilevel Structural Equation Model From a Multiple Group Perspective." (2000). LSU Historical Dissertations and Theses. 7225.