Effectiveness of guided group work in graduate level quantum mechanics

We investigate the effects of guided group work sessions on graduate student performance on a quantum mechanics assessment. Data from a single large Midwestern university were taken over a five-year period, during which guided group work sessions were offered to accompany the graduate-level quantum mechanics course. Students were pre- and post-tested using a set of mostly conceptual items that we call the graduate quantum mechanics assessment. The reliability and validity of this assessment are addressed. A mixed linear model is used to analyze the dependence of post-test scores on factors such as group work attendance, pretest scores, GRE Physics scores, and others. We find a statistically significant effect of group work attendance on post-pre gains, specifically that attendance of one 60-min group work session improves performance on a related post-test item by 6.4%, administered 2–10 weeks after the session. We discuss the lack of a randomized control group and address possible confounding effects such as student self-selection, and attitudinal and motivational factors. Overall, the results of this study indicate that guided group work sessions at the graduate level can be feasible and effective. We note preliminary observations of differences in group interactions and classroom logistics compared to group work at the undergraduate level.

Figure 1

Mean pretest and post-test scores on GQMA, by item for the 24 continuum-scored items used in this study. Scores are averaged over the 133 student participants.

Figure 2

Mean student pretest and post-test scores on the set of all continuum-scored GQMA items for which they attended $N$ relevant GGW sessions, for $N=0$, 1, 2 or more. Error bars represent standard error. The numbers of student items in the three treatment categories are 607, 107, and 37, respectively, and these originate from 133 students’ work.

Figure 3

$\theta$ resulting from the mixed linear model (1) for quartiles of the 133 total students. Here, the 4th GRE Physics quartile is the 25% of students with the lowest GRE Physics scores, and so on. The error bars indicate standard error.

Figure 4

Probability of correctly answering a binary (right or wrong) item on the post-test, given that students answered incorrectly on the pretest. Results are averaged over different students and items, and are separated by the number of item-relevant GGW treatments students received. $N$ is the number of student items in each category. Error bars represent standard error, although there are correlations within the data.