Gender gaps and gendered action in a first-year physics laboratory

[This paper is part of the Focused Collection on Gender in Physics.] It is established that male students outperform female students on almost all commonly used physics concept inventories. However, there is significant variation in the factors that contribute to the gap, as well as the direction in which they influence it. It is presently unknown if such a gender gap exists on the relatively new Concise Data Processing Assessment (CDPA) and, therefore, whether gendered actions in the teaching lab might influence—or be influenced by—the gender gap. To begin to get an estimates of the gap, its predictors, and its correlates, we have measured performance on the CDPA at the pretest and post-test level. We have also made observations of how students in mixed-gender partnerships divide their time in the lab. We find a gender gap on the CDPA that persists from pre- to post-test and that is as big as, if not bigger than, similar reported gaps. We also observe compelling differences in how students divide their time in the lab. In mixed-gender pairs, male students tend to monopolize the computer, female and male students tend to share the equipment equally, and female students tend to spend more time on other activities that are not the equipment or computer, such as writing or speaking to peers. We also find no correlation between computer use, when students are presumably working with their data, and performance on the CDPA post-test. In parallel to our analysis, we scrutinize some of the more commonly used approaches to similar data. We argue in favor of more explicitly checking the assumptions associated with the statistical methods that are used and improved reporting and contextualization of effect sizes. Ultimately, we claim no evidence that female students are less capable of learning than their male peers, and we suggest caution when using gain measures to draw conclusions about differences in science classroom performance across gender.

Figure 1

Schematic representation of the theoretical framework adopted for this study.

Figure 2

CDPA pretest (solid columns) and post-test (hatched columns) results for the populations studied. The raw CDPA score is out of a maximum possible of ten points. Our students are learning, but the status quo remains. Uncertainty bars represent the 95% confidence interval. These data consist of 191 female and 280 male students (the ones for which we have paired pre- and post-test data).

Figure 3

Histogram of Hedges’ $g$ effect sizes for the FCI, FCME, BEMA, and CSEM results (white) (summarized in Ref. [1]) and the CDPA (black). Pretests are given by the solid columns and post-tests by the hatched columns. None of the effect sizes from the studies involving the FCI, FCME, BEMA, and CSEM are as large as the effect sizes associated with this study of the CDPA.

Figure 4

The “gain” by female and male students on the CDPA, as defined by five different estimators. The asterisk indicates a statistically significant difference below the $p=0.05$ level from an independent-samples $t$-test. Uncertainty bars represent the standard errors of the mean. From left to right: (1) the average normalized change $⟨c⟩$; (2) the average absolute gain normalized by the total test score $⟨g_{\mathrm{abs}}⟩$; (3) the course average normalized gain $⟨g⟩$; (4) the absolute gain normalized by twice the average of the pre- and post-test $⟨g_{2\mathrm{av}}⟩$; and (5) the relative change $⟨g_{\mathrm{rel}}⟩$. That different estimators lead to different stories suggests that the gain results are suspect.

Figure 5

Schematic representation of the modified theoretical framework, adapted to better align with the results of this study.

Figure 6

Summary of post-test performance on each item of the CDPA, by gender. Uncertainty bars represent the standard errors of the mean. These data consist of 191 female and 280 male students (the ones for which we have paired pre- and post-test data). Item by item, the male student population outperforms the female student population.