Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Modeling Active Learning Classrooms

Olive Ross, Meagan Sundstrom, N. G. Holmes

Abstract

Over the past several decades, a large body of research has shown that undergraduate science students learn more and more equitably in classrooms that are not entirely lecture-based. Non-lecture activities are collectively referred to as ``active learning;" however, this term lacks definition and little research has examined which types and combinations of active learning strategies are most effective. In this study, we use a dataset that includes more than 10,000 students and spans a range of scientific disciplines, institutions, and instructors to create a predictive model that maps time spent on different classroom activities to student conceptual learning. We find that four variables -- classroom time spent on lecture, group worksheets, clicker questions, and student questions -- are sufficient to reliably predict student learning, as measured by concept inventory scores. We identify two types of classes that consistently lead to exceptional student learning gains (effect sizes greater than 2). The first are classes that spend 10-20\% of class time on group worksheets, 20-40\% of class time on group clicker questions, and average two or more student questions per hour of class time. The second is classes spending 30\% or more of class time on group worksheets. We also find that classes which do not include any group worksheets consistently have learning outcomes comparable to fully lecture classes, even when other active learning strategies are used. These results provide testable recommendations for future controlled studies to investigate effective active learning implementation.

Modeling Active Learning Classrooms

Abstract

Over the past several decades, a large body of research has shown that undergraduate science students learn more and more equitably in classrooms that are not entirely lecture-based. Non-lecture activities are collectively referred to as ``active learning;" however, this term lacks definition and little research has examined which types and combinations of active learning strategies are most effective. In this study, we use a dataset that includes more than 10,000 students and spans a range of scientific disciplines, institutions, and instructors to create a predictive model that maps time spent on different classroom activities to student conceptual learning. We find that four variables -- classroom time spent on lecture, group worksheets, clicker questions, and student questions -- are sufficient to reliably predict student learning, as measured by concept inventory scores. We identify two types of classes that consistently lead to exceptional student learning gains (effect sizes greater than 2). The first are classes that spend 10-20\% of class time on group worksheets, 20-40\% of class time on group clicker questions, and average two or more student questions per hour of class time. The second is classes spending 30\% or more of class time on group worksheets. We also find that classes which do not include any group worksheets consistently have learning outcomes comparable to fully lecture classes, even when other active learning strategies are used. These results provide testable recommendations for future controlled studies to investigate effective active learning implementation.
Paper Structure (12 sections, 1 equation, 5 figures, 1 table)

This paper contains 12 sections, 1 equation, 5 figures, 1 table.

Table of Contents

  1. Abstract
  2. Introduction
  3. Results
  4. Discussion
  5. Methods
  6. Data
  7. Quantifying Classroom Activities with the COPUS
  8. Quantifying Student Learning with Concept Inventories
  9. Modeling
  10. Model Errors
  11. Out-of-sample predictions
  12. Final Model Training and Generation of Results

Figures (5)

  • Figure 1: Heatmaps of predicted effect sizes on simulated COPUS code data by amount of group worksheets (WG), Group Clicker Questions (CG), and Student Questions (SQ). The amount of lecture (Lec) is calculated as $Lec = 1 - CG - WG$, assuming that all class time not spent on clicker questions or worksheets is spent lecturing. Student questions (SQ) are almost always co-coded with lecture (Lec) and therefore are not present in this approximations. Because student questions tend not to take up a whole 2-minute COPUS interval, the right y-axis translates from percent of class time spent on SQ to average number of student questions per hour of class. Blue hatched regions are predictions that either have high prediction error ($>$0.3) or predicted effect size values outside the range of effect sizes present in the data.
  • Figure 2: Heatmap of predicted effect sizes on simulated COPUS data by amount of Group Worksheets (WG) and Student Questions (SQ). All Group Clicker Question (CG) values are set to 0 and amount of lecture (Lec) is calculated as $1 - WG$. Blue hatched regions are predictions that either have high prediction error ($>$0.3) or predicted effect size values outside the range of effect sizes present in the data.
  • Figure 3: Results of Leave-One-Out tests for 57 previously published classes. Each class is removed from the dataset, the model is trained, and a prediction is made for that class. Classes with prediction error higher than 0.3 are shown in red.
  • Figure 4: Results of Leave-One-Out tests for 57 previously published classes by field (left) and class size (right). Each class is removed from the dataset, the model is trained, and a prediction is made for that class.
  • Figure 5: Effect size predictions on the data collected for the current study from 12 astronomy and physics classes. The model used for prediction is trained on the 57 previously published classes.