T8-MD-3 - MIDFIELD Special Session; A Primer on Novel Methodologies in Longitudinal Analysis of Student Data3. Research Work In Progress
1 University of Indianapolis
The past decade has opened up MIDFIELD to the use of novel methodologies. Three particular phenomena – grade variance due to course size and enrolled section, and degree program change – are of particular interest. This work demonstrates three useful methodologies for analyzing these phenomena, how they are used, and preliminary results. The first, Markov chains, can examine links between success (graduation) and changing majors. The second, Hierarchical Linear Models (HLMs) can show nested relationships between different courses, including the effect of course size and section on grade variances. The third, mutual information, can uncover relationships between sets of students who switched majors and those who did not, relative to a variable such as GPA. Some novel results include: course size has no effect on grade variance, but section does; students who switch majors graduate at a higher rate; and the effect of switching majors is more random for those who leave college than those who graduate.