Modelling Assessment Data with a Hierarchical Approach

Assessment data were collected from multiple institutions in fall 2013 and winter 2014 to measure the conceptual understanding and perspectives of statistics for students taught under a randomization-based curriculum. We concentrated on identifying student and instructor characteristics that were associated with student gains and determining whether there was a disparity in student gains based on level of instructor’s experience with the curriculum. A hierarchical modelling approach was employed to analyze the data, because multiple students were in each section taught by the same instructor and variables were collected on students and instructors. Our exploratory data analysis included graphical analysis and instructor classification based on their experience with the curriculum, teaching and school background, and the idiosyncrasies of their students. Subsequently, we ran hierarchical models to model student improvements in their conceptual understanding of statistics with student-level and instructor-level predictors.