How Psychology Can Reduce the Achievement Gap for Women in STEM

by Dr. Sarah Herrmann, Weber State University

In her New York Times Op-Ed, Eileen Pollack said this of her experience in an introductory Physics course at Yale in 1978:

The boys in my introductory physics class, who had taken far more rigorous math and science classes in high school, yawned as our professor sped through the material, while I grew panicked at how little I understood. The only woman in the room, I debated whether to raise my hand and expose myself to ridicule, thereby losing track of the lecture and falling further behind.

Pollack’s experience accurately describes the phenomenon of numerical underrepresentation in a classroom; namely, awareness that you are the only (or one of the only) people like you in a space. This can suggest that people like you don’t belong or can’t be successful in that setting. Indeed, we see that while today, 50.3% of STEM (i.e., science, technology, engineering, mathematics) degrees are awarded to women, this varies by field. Women receive over half of bachelor’s degrees in the biological sciences, but receive far fewer in the computer sciences (17.9%), engineering (19.3%), physical sciences (39%) and mathematics (43.1%). However, there is an increasing demand for STEM graduates in the United States, as STEM careers are forecasted to increase by 13% between 2017 and 2027. Research in the psychological sciences can help us to understand some of the psychological phenomenon underlying these educational disparities, and can suggest ways to change individual perceptions that can increase belonging, performance, and persistence in these fields.

The cornerstone of the research on women’s experience in STEM environments relates to stereotyping: the perceptions of a group of people (e.g., women) as having less academic ability as a function of their group identity. This can ultimately lead to disidentification and dropout from these areas as a function of stereotype threat—a “threat in the air”—that stems from fear of confirming a negative stereotype about one’s group (Spencer, Steele, & Quinn, 1999; Steele & Aronson, 1995). Research by Inzlicht and Ben-Zeev (2000) revealed that women performed more poorly on a math section of the GRE when taking the test with two men, as compared to two women. There were no such performance differences on a verbal section of the GRE, as there are not stereotypes about women’s ability. What could produce these differences in performance? Schmader and Johns (2003) demonstrated that the effect of gender imbalance on performance on a math task is explained by reduced working memory. Essentially, when one is afraid of confirming a negative stereotype about their group, they have reduced working memory to devote to difficult tasks, thereby impugning their performance.

 Luckily, research points to several ways that we can reduce the effects of stereotype threat. For example, Ben-Zeev and colleagues (2017) demonstrated that just learning about stereotype threat can reduce performance gaps. Additionally, if a task is reframed as not being indicative of ability, the effect goes away (Good, Aronson, & Harder, 2008; Quinn & Spencer, 2001; Spencer, Steele, & Quinn, 1999; Steele & Aronson, 1995). Another strategy for reducing the effect of stereotype threat is having participants make external, rather than internal, attributions for difficulty; specifically, if a participant explains their performance in terms of poor preparation (i.e., I didn’t study enough), there will be less effect of stereotype threat than an internal attribution (i.e., I’m not a “math person”; Ben-Zeev, Fein, & Inzlicht, 2005; Johns, Inzlicht, & Schmader, 2008; Johns, Schmader, & Martens, 2005).

Another factor that accounts for underperformance among women and underrepresented groups in STEM are feelings of not belonging. We see, for example, situational cues such as numerical underrepresentation of fellow students, graduate students, or professors can indicate who belongs and who can be successful in a given environment (Abrams & Hogg, 1999; Crocker et al., 1998; Inzlicht & Ben-Zeev, 2000; 2003; Inzlicht & good, 2006; Major & O’Brien, 2005; Murphy, Steele, & Gross, 2007; Sekaquaptewa & Thompson, 2003; Tajfel & Turner, 1986). These assumptions begin early; for example, we see that by first grade, children believe that boys are better than girls in math, science, programming, and engineering (Master, Cheryan, & Meltzoff, 2017). One study by Murphy, Steele, and Gross (2007) had STEM majors view a video depicting an upcoming summer program that was either gender balanced or unbalanced (1 women to 4 men). Results revealed that women STEM majors had higher heart rates, skin conductance, and lower feelings of belonging wen watching the gender unbalanced video, accounting for a physical side effect of underrepresentation.

Several psychological interventions have successfully reduced the performance gaps between men and women related to belonging. Walton, Logel, Peach, Spencer, and Zanna (2015) demonstrated that hearing about the challenges that senior engineering students (both men and women) faced at the beginning of college significantly increased women’s engineering GPA. Additionally, exposure to a similar role model for women increased identification with math and reduced concerns about gender stereotyping (Ramsey, Betz, & Sekaquaptewa, 2013). Similarly, my colleagues and I found that reading a brief, online story from a woman graduate student about the challenges she experienced in introductory chemistry increased women’s chemistry course grades and reduced their failure and withdrawal rate (Herrmann et al., 2016). Role models are most effective if they are successful in an area of mutual interest, accessible, and if they communicate past challenges. In fact, encountering a woman role model who is not identifiable can actually decrease women’s interest and perceived success in computer science (Cheryan, Siy, Vichayapai, Drury, & Kim, 2011; Cheryan, Drury, & Vichayapai, 2013). Importantly, direct contact with a role model, while beneficial, is not necessary.

These findings suggest that the performance gap between men and women in STEM, especially in fields where underrepresentation is more apparent, has as much or more to do with the psychological experience of underrepresentation, rather than any inherent differences in ability. Additionally, these sorts of interventions could easily be incorporated into educational programs by balancing faculty and graduate student or teaching assistant gender, encouraging external attributions for poor performance, and providing formal or informal mentoring structures and organizations for women in STEM.