Applications of Hierarchical Linear Modeling in Chemistry Education Research

Abstract

The utility of hierarchical linear modeling (HLM) in the analysis of nested data is well established. As data in chemistry education research is often considered nested, HLM is a beneficial statistical model and analytical tool within that context. We present an overview of HLM and use an example absorbance spectroscopy study within the context of general chemistry to introduce HLM. To show the benefits of HLM with various types of educational data and its usefulness within various contexts we discuss three published chemistry education research studies. The self-assessment study aimed to characterize changes observed in chemistry students’ self-assessments of their understanding when engaging in instructional activity. HLM allowed for the identification and quantification of student and instructional variables related to observed changes in self-assessment. The inequity study aimed to assess inequities between sexes and races/ethnicities in a traditional chemistry course compared to a reformed active learning course. HLM was used to identify and compare inequities between sexes and races/ethnicities. Additionally, HLM was used to assess several possible causes of the observed inequity between sexes in one of the courses and identified the grading system as a major cause. The grading study looked at variations in the approaches graduate student instructors and faculty instructors take when evaluating student work. Using HLM, the approaches that correlated with assigned grades for instructors of differing experience levels were specified. The presentation of these studies demonstrates how HLM is uniquely useful within several chemistry education research contexts.