Gender inequity in individual participation within physics and science, technology, engineering, and math courses

Gender inequities continue to persist within science, technology, engineering, and mathematics (STEM) disciplines, even at the undergraduate level. This has led researchers to further examine potential factors that contribute to retention and persistence of undergraduates in STEM fields. In this study using classroom observations, we examined gender equity in individual verbal participation in large introductory physics courses, and compared our results to observations in introductory courses in other STEM disciplines. We found that in introductory physics courses, men had disproportionately higher rates of individual verbal participation than women. Observation-level analysis confirmed that three-quarters (76.2%) of the physics observations had descriptively higher than expected participation by men and almost a quarter of observations (23.8%) were statistically significant for a gender imbalance in individual verbal participation. We then sought to determine if any pedagogical strategies or student behaviors correlated with a more equitable classroom to better understand what drives gender inequity in participation, and found three classroom behaviors—an increasing amount of instructor questions, group responses from students, and student questions—correspond with a more gender equitable classroom. Student-level survey data, which mirrors the observation data, also show that self-reported levels of individual participation have small, significant correlations with both course-level belonging and inclusivity. The introductory physics results were mostly replicated in the other STEM disciplines, despite their differences in course structure. The patterns of individual participation were still disproportionately higher for men, with two-thirds of observations displaying a bias towards more men participating. Student-level survey data continued to mirror the observation data, and small, significant correlations between student self-reported participation and course-level belonging and inclusion were found. However, only the number of student questions correlated with a more equitable classroom in other STEM courses. This study extends the conversation on the relationship between active learning and equity in the classroom, demonstrating a need to move beyond mere inclusion of active pedagogies towards proactive facilitation of equitable and comfortable verbal participation by all students. Practical strategies for encouraging inclusive classroom dialogue, such as transparency, growth-mindset messaging, and multiple modes of engagement, are discussed.

Figure 1

Sample segmenting for each subcategory. In the top left corner, the subcategory the timeline represents is noted (see list of categories and subcategories in Table 2). The center black bar shows the 2-min time intervals during an observation. The brackets indicate the teaching strategy used in that time interval. The combined “Demo & Clicker Activity” brackets are used to visualize clicker activities based on a demonstration, but the time spent on each individual teaching strategy was noted during the observation and used in our analyses. The asterisk on the last timeline denotes that this timeline shows the first 50 min of an 80-min observation. The corresponding full timelines for each segmented sample in this figure are in the Supplemental Material [64].

Figure 2

Examining physics observations by category and participation score. Box-and-whisker plots of the participation scores in all categories within physics. Each point represents a class observation relative to the “neutral” participation score of one (horizontal line). The boxes show the median participation score in each category (center bar), the first and third quartiles (i.e., the 25^th and 75^th percentiles; “hinges” or outer lines of box), and 1.5 times the interquartile range (“whiskers” or lines extending from boxes). The majority of observations have higher rates of participation of men (i.e., participation scores lower than 1), and the mean participation scores do not differ from each other. Category 1 is for lectures with clicker activities as its main active-learning component; Category 2 is for lectures with either demonstrations or other active learning as its main active-learning component; and Category 4 is a combination of clickers as well as demonstrations or other active-learning components.

Figure 3

Examining individual classroom observations for gender equity. Each point represents a binomial test evaluating whether the observed proportion of men individually participating verbally in class differs from the proportion of men on the roster. Cohen’s $h$ represents effect size, or the distance between the expected proportion (based on the course roster) and the observed proportion of men participating. A Cohen’s $h$ greater than zero indicates that women participated at higher levels than expected, while a Cohen’s $h$ less than zero indicates that men participated at higher levels than expected. Significance is denoted by color: the 10 (out of 42) significant tests are black, and the nonsignificant tests are gray.

Figure 4

. Women tend to individually participate more in women-majority physics classes. Scatterplot of physics observations by gender representation in the course ($x$ axis) and participation score ($y$ axis). Simple regression lines illustrate the linear relationship between proportion of men in the course and participation scores (shaded area represents the 95% confidence interval). We find a moderate, significant correlation between participation scores and the proportion of men and women enrolled in the course, meaning we tend to find higher rates of women individually participating when there are more women in the classroom.

Figure 5

An increase in the number of (a) instructor-prompted questions, (b) group responses, and (c) student questions appears to be strongly correlated with a higher participation score (regardless of category). Scatterplots of physics observations by the (time-normalized) number of instructor questions ($x$ axis) and participation score ($y$ axis). Simple regression lines illustrate the linear relationship between instructor-led questions and participation scores (shaded area represents the 95% confidence interval).

Figure 6

Participation scores across all disciplines seem to show a disproportionate amount of men individually participating. Box-and-whisker plots of the participation scores in physics courses versus all other introductory STEM courses observed. Each point represents a class observation relative to the “neutral” participation score of one (horizontal line). The boxes show the median participation score in each category (center bar); the first and third quartiles (i.e., the 25^th and 75^th percentiles; “hinges” or outer lines of box), and 1.5 times the interquartile range (“whiskers” or lines extending from boxes). While the other STEM courses had a higher average participation score ($M=0.85$) than physics ($M=0.67$), their means did not significantly deviate from each other ($W=939.5$, $p=0.48$), suggesting similar levels of gender inequity in individual participation across disciplines.