SAGE Musings: Implicit Bias in STEM

Carol Ormand, SERC, Carleton College
Author Profile
published Nov 8, 2018 11:08am

Unconscious, negative associations or stereotypes are called implicit biases. Extensive research has documented that implicit bias is pervasive in STEM, that it negatively impacts the participation and success of people who belong to demographic groups underrepresented in STEM (women, underrepresented minorities, people with disabilities, and others), and that even those of us who belong to underrepresented groups have implicit biases. Fortunately, there are evidence-based strategies and resources we can all use to identify and mitigate implicit biases. In October, 2018, I attended a series of four webinars, produced by the CIRTL network, about implicit biases. This Musing outlines how pervasive implicit biases are in STEM; I'll follow up with a future Musing about how to identify and address implicit biases.

How Pervasive is Implicit Bias in STEM?

It's pervasive. Study after study after study documents implicit bias in STEM. A few examples:

Student bias toward other students

  • In a study of more than 1700 undergraduate biology students at the University of Washington, students were asked to name other students they thought had a strong understanding of class material. Male students underestimated their female peers (Grunspan et al., 2016). "Put it this way: To the men in these classes, a woman would need to get an A to get the same prestige as a man getting a B" (Yong, 2016).

Faculty bias toward students

  • Grades assigned to students differ according to demographics. When instructors remove identifying information prior to grading assignments, grades of female and minority students go up. Significantly. (Ashburn-Nardo, 2018).
  • How faculty members evaluate prospective students or job candidates exposes significant bias against female students and against all students of color (e.g. Moss-Racusin et al., 2012; Milkman et al., 2015). Both male and female faculty members exhibit this bias. For example:
    • "In our experiment, professors were contacted by fictional prospective students seeking to discuss research opportunities prior to applying to a doctoral program. Names of students were randomly assigned to signal gender and race (White, Black, Hispanic, Indian, Chinese), but messages were otherwise identical.... We found that when considering requests from prospective students seeking mentoring in the future, faculty were significantly more responsive to White males than to all other categories of students, collectively, particularly in higher-paying disciplines and private institutions" (Milkman et al., 2015).
    • "In a randomized double-blind study (n = 127), science faculty from research-intensive universities rated the application materials of a student—who was randomly assigned either a male or female name—for a laboratory manager position. Faculty participants rated the male applicant as significantly more competent and hireable than the (identical) female applicant. These participants also selected a higher starting salary [$3500 more, on average] and offered more career mentoring to the male applicant. The gender of the faculty participants did not affect responses, such that female and male faculty were equally likely to exhibit bias against the female student" (Moss-Racusin et al., 2012). The application materials were identical except for the name of the candidate ("John" or "Jennifer"). Relatively small differences in starting salary can have profound impacts on lifelong earnings (e.g., Loudenback and Gold, 2017).

Bias in the hiring process

  • When job candidates "Whiten" their c.v.'s, they are more likely to be interviewed. Whitening their names has a small impact; Whitening their experience has a larger impact; Whitening both their name and their experience has the most impact. Here's the extra icky part: when job advertisements include a statement about encouraging a diverse pool of applicants, the candidates were less likely to Whiten their resumes, but the search committees were just as likely to discriminate against candidates with names or experiences indicating that they are people of color. Thus, adding a phrase about encouraging a diverse pool of applicants actually decreases the likelihood of making diverse hires (Kang et al., 2016). This study was not restricted to STEM.
  • In a study of more than 300 letters of recommendation accompanying applications for medical faculty positions, letters of recommendation for female applicants were shorter, used language more likely to raise doubts about the candidates' abilities, and "reinforce[d] gender schema that tend to portray women as teachers and students, and men as researchers and professionals" (Trix and Psenka, 2003).

Bias toward colleagues

  • In a study of 3,652 colloquium talks across six academic disciplines -- not limited to STEM -- at 50 prestigious colleges and universities, Nittrouer et al. (2018) found that "Men were more likely than women to be colloquium speakers even after controlling for the gender and rank of the available speakers." The authors wondered whether this was because men were more likely to accept invitations to speak, and found that was not the case (Nittrouer et al., 2018).
  • Williams et al. (2016) analyzed more than 3000 responses to the Workplace Experiences Survey launched by the Society of Women Engineers and the Center for WorkLife Law at the University of California, Hastings College of the Law. Engineers reported many kinds of implicit bias in the workplace, including:
    • "68% of engineers of color (men as well as women) reported having to prove themselves repeatedly, as compared to 35% of white men."
    • "61% of women vs. 35%1 of white men reported that they have to prove themselves repeatedly to get the same levels of respect and recognition as their colleagues."
    • "Women (33%) were more likely than white men (16%) to report pressures to let others take the lead; were more likely to report doing more "housework," such as finding a time everyone can meet, taking notes, or planning parties (55% vs. 26%); and were less likely to report having the same access to desirable assignments (65% vs. 85%)."
    • "Engineers of color were more likely than white men to report pressures to let others take the lead (39% vs. 16%) or do office housework (52% vs. 26%) and were less likely to report having the same access to desirable assignments (55% vs. 85%)."

Impacts of Bias

There are many predictable consequences of these forms of bias. The most obvious one is that students who are underrepresented in STEM tend not to persist in STEM, and the data bear this out (e.g. Grunspan et al., 2016; National Science Board, 2016).

What is perhaps most disturbing about the impacts of implicit biases is how pervasive they are: "Even among highly educated people. Even among very well-intentioned people. Even among women" (Ashburn-Nardo, 2018). This means that solving the problem will require going beyond improving the representation of underrepresented humans in STEM. We have to tackle our attitudes -- including, specifically, our unconscious attitudes -- and we have to tackle everyone's attitudes. But it can be done. In countries where gender stereotypes about STEM performance are weaker, the gender gap in STEM is much smaller (Nosek et al., 2009). My next Musing will be about strategies for identifying and addressing implicit bias. Stay tuned.

Comment? Start the discussion about SAGE Musings: Implicit Bias in STEM