Date of Graduation
College of Education and Human Services
Learning Sciences and Human Development
Sebastian R. Diaz
Gayle A. Neldon
This research compares and contrasts two approaches to predictive analysis of three years' of school district data to investigate relationships between student and teacher characteristics and math achievement as measured by the state-mandated Maryland School Assessment mathematics exam. The sample for the study consisted of 3,514 students taught by 99 teachers in a small Appalachian school district in western Maryland. The first analytic approach, standard multiple linear regression, produced a model in which each of the predictors is statistically significant: student gender, prior math achievement, student performance on school district mathematics benchmark exams, teacher years of experience, and advanced teacher certification. In the second approach---multilevel modeling with students as the level-1 unit of analysis and teachers as the level-2 unit of analysis---student characteristics are significant predictors of math achievement, and teacher characteristics are insignificant predictors. The study is set within a context of an exploration of relationships among society, education, and technology. Implications of the study's results for K-12 mathematics education practice and policy are discussed including: the need to define teacher effectiveness and to identify teacher characteristics that contribute to student achievement; the promise of benchmarking exam systems; the necessity of effective math education, minimally from early education through Algebra II; the need to evaluate teacher certification criteria and the efficacy of teacher preparation programs; the importance of using appropriate statistical modeling approaches in education research; and a call to put students back into the education equation through student-centered funding models.
Deering, Pamela Rose, "Implications of Interactions Among Society, Education and Technology: A Comparison of Multiple Linear Regression and Multilevel Modeling in Mathematics Achievement Analyses" (2014). Graduate Theses, Dissertations, and Problem Reports. 123.