Assessing Supplemental Instruction Using Mixed Effects Modeling

• Activity Number: 45
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Social Statistics Section
• Author(s): Daniel Yanosky*+
• Companies: Kennesaw State University
• Keywords: Higher Education ; Supplemental Instruction ; Program Evaluation ; Assessment ; Mixed Effects Modeling

Abstract

Supplemental Instruction (SI) is a collaborative learning program implemented by many universities across the United States to promote student success in historically difficult courses. While the literature base for SI is substantial, most of the approaches to program evaluation are simplistic and unable to provide accurate estimates of the program's effect due to reliance on simple two-group comparisons while leaving many potential confounding variables uncontrolled. In this study, we apply mixed effects modeling to model longitudinal student outcomes data. The data consist of all undergraduate students enrolled in classes from Fall 2008 through Fall 2011 at a large university in the Southeast. Each student's average GPA for the semester was specified as the response variable. Fixed explanatory variables included high school GPA, age, sex, race, SAT or ACT scores, course subject, scholarship status, etc. Random-effects for student, instructor/SI leader combination, and course were also included. Results demonstrate that SI has a statistically significant effect on student academic performance even after controlling for many alternative explanations for student success.