Beyond named methods: A typology of active learning based on classroom observation networks

TL;DR

This study develops a network-based typology of active-learning instruction in introductory physics by analyzing classroom observation networks formed from COPUS-coded video data across 30 instructors and 27 institutions implementing four named methods. By constructing per-instructor observation networks, calculating pairwise cosine similarities, sparsifying with locally adaptive thresholds, and applying Infomap clustering, the authors identify five instruction types that cut across method names and reveal substantial within-type variability. They further show that instruction type does not significantly affect student conceptual gains, implying instructors can flexibly adapt methods without compromising learning. The approach provides a nuanced, temporally aware framework for describing instructional practice and offers a foundation for future cross-discipline expansion and professional development.

Abstract

A growing number of introductory physics instructors are implementing active learning methods in their classrooms, and they are modifying the methods to fit their local instructional contexts. However, we lack a detailed framework for describing the range of what these instructor adaptations of active learning methods look like in practice. Existing studies apply structured protocols to classroom observations and report descriptive statistics, but this approach overlooks the complex nature of instruction. In this study, we apply network analysis to classroom observations to define a typology of active learning that considers the temporal and interactional nature of instructional practices. We use video data from 30 instructors at 27 institutions who implemented one of the following named active learning methods in their introductory physics or astronomy course: Investigative Science Learning Environment (ISLE), Peer Instruction, Tutorials, and Student-Centered Active Learning Environment with Upside-down Pedagogies (SCALE-UP). We identify five types of active learning instruction: clicker lecture, dialogic clicker lecture, dialogic lecture with short groupwork activities, short groupwork activities, and long groupwork activities. We find no significant relationship between these instruction types and the named active learning methods; instead, implementations of each of the four methods are spread across different instruction types. This result prompts a shift in the way we think and talk about active learning: the names of developed active learning methods may not actually reflect the specific activities that happen during instruction. We also find that student conceptual learning does not vary across the identified instruction types, suggesting that instructors may be flexible when modifying these methods without sacrificing effectiveness.

Figures (8)

• Figure 1: Toy classroom observation network for the example COPUS coding in Table \ref{['tab:exampleCOPUS']}. Though not directly part of our analysis, node color indicates the fraction of all two-minute time intervals that the code was present: darker colors indicate larger fractions (i.e., longer time durations). Edges point from an initial code to the code that occurs in the following two-minute time interval of the COPUS observation. Edge width indicates the number of transitions occurring between those two codes normalized by the total number of two-minute time intervals observed.
• Figure 2: (a) Sparsified similarity network with nodes (i.e., classroom observation networks for each instructor) colored by cluster (i.e., instruction type) and edge weights proportional to cosine similarity. I-WC indicates whole-class implementation of ISLE, I-LO indicates lab-only implementation of ISLE (i.e., the observation networks represent lecture sections), PI indicates Peer Instruction, T-WC indicates whole-class implementation of Tutorials, T-RO indicates recitation-only implementation of Tutorials (T-RO* indicates recitation-only implementation of Tutorials but with the observation network representing the lecture section of the course), and SU indicates SCALE-UP. The attributes of each course are provided in Table \ref{['tab:individualfeatures']} in the Appendix. (b) Average node-level measures for observation networks assigned to each instruction type (see definitions in Table \ref{['tab:networkdefs']}).
• Figure 3: Example classroom observation networks for each instruction type. The active learning method and course label (as in Fig. \ref{['fig:map']} and Table \ref{['tab:individualfeatures']} in the Appendix) for each network is indicated in parentheses in the sub-captions. Though not directly part of our analysis, node color indicates the fraction of all two-minute time intervals that the code was present: darker colors indicate larger fractions (i.e., longer time durations). Edges point from an initial code to the code that occurs in the following two-minute time interval of the COPUS observation. Edge width indicates the number of transitions occurring between those two codes normalized by the total number of two-minute time intervals observed. All 30 classroom observation networks are available at Ref. github2025.
• Figure 4: (a) Segregation Z-scores for various course and institution attributes. Points indicate the Z-scores and the gray shaded area indicates where the Z-score is not statistically significant, $|Z| < 1.96$. Points falling outside of this region indicate that there is a significant relationship between the corresponding attribute and instruction type. (b) For the two attributes with significant segregation Z-scores, LPA profile and class size, the distributions of courses with these attributes across instruction types.
• Figure 5: Effect sizes for concept inventory scores by instruction type. Points represent Hedges' g values and error bars indicate 95% confidence intervals. Positive (negative) values indicate increases (decreases) in student scores from pre- to post-semester. N values indicate the number of students included in the analysis for each instruction type (i.e., the number of students with matched pre- and post-semester concept inventory scores).