Research-based assessment of students’ beliefs about experimental physics: When is gender a factor?

The existence of gender differences in student performance on conceptual assessments and their responses to attitudinal assessments has been repeatedly demonstrated. This difference is often present in students’ preinstruction responses and persists in their postinstruction responses. However, one area in which the presence of gender differences has not been extensively explored is undergraduate laboratory courses. For example, one of the few laboratory focused research-based assessments, the Colorado Learning Attitudes about Science Survey for Experimental Physics (E-CLASS), has not been tested for the existence of gender differences in students’ responses. Here, we utilize a national data set of responses to the E-CLASS to determine if they demonstrate significant gender differences. We also investigate how these differences vary along multiple student and course demographic slices, including course level (first-year vs beyond-first-year) and major (physics vs nonphysics). We observe a gender gap in pre- and postinstruction E-CLASS scores in the aggregate data both for the overall score and for most items individually. However, for some subpopulations (e.g., beyond-first-year students) the size or even existence of the gender gap depends on another dimension (e.g., student major). We also find that for all groups the gap in postinstruction scores vanishes or is greatly reduced when controlling for preinstruction scores, course level, and student major.

Figure 1

Violin plots showing the distribution of overall pre- and postinstruction E-CLASS scores for men and women in the full, aggregate data set ($N=7013$). Solid lines indicate the mean for each distribution. Differences between the distributions for men and women are statistically significant for both pre- and post-tests (Mann-Whitney U, $p\ll 0.01$).

Figure 2

Sorted plot of the difference between the average scores (points) of women and men on each item of the E-CLASS. Zero difference is marked by the solid horizontal line, and statistically significant differences are denoted by solid markers. The dotted horizontal line marks the point below which the differences between men’s and women’s scores represent moderate effect sizes ($d>0.3$). See Ref. [9] for the full list of question prompts.

Figure 3

Learning gains on each of the four different metrics for gain given in Table 5. From left to right these are average normalized change $⟨c⟩$; Hake gain $⟪g⟫$; average absolute gain $⟨g_{\mathrm{abs}}⟩$; and average percent increase over pretest $⟨g_{\%}⟩$. Error bars represent the standard error of the mean. The difference between the gains for men and women is statistically significant only in the case of normalized change $⟨c⟩$ (independent sample $t$ test, $p\ll 0.01$).