Female Role Models: Protecting Women’s Math Test Performance

The presence of a female role model can buffer women’s math test performance in the face of gendered stereotypes that can cause stress and lead women to underperform men in this field.

Introduction

Women are historically underrepresented in math-related undergraduate and graduate programs. One of the many reasons for this disparity is that women consistently underperform on standardized math tests compared to men, despite similar educational backgrounds. A potential explanation for this underperformance on test day is the existence of a “stereotype threat”. This phenomenon occurs due to the stereotype that women are less adept at math, and therefore women face additional anxiety not only to perform well but also to disprove pre-existing and widely-held stereotypes.

The authors propose that female role models in math-based fields could serve to protect women’s performance from this stereotype threat by offering stereotype-disconfirming information. They posit three factors critical to role models: their success needs to appear attainable, they need to seem similar to the student, and the student needs to care about their performance in this domain. Using three separate experimental studies to measure the impact of female role models on highly motivated and math-identified female students’ test performance and self-reported confidence, this article addresses how role models may buffer students’ performance in the face of negative stereotypes.

Findings

Study 1

  • Women perform as well as men on a math test when a female experimenter administered the test (14.46 vs. 13.56 on 25-question test), but underperformed compared to men when a male experimenter administered the test (9.88 vs. 15.02).
    • There was no significant difference in the number of problems answered.
    • There was a significant difference in performance accuracy (number of correct problems divided by number of total problems answered). Women with a male experimenter had the lowest performance accuracy (51%) compared to women with a female experimenter (69%), men with a male experimenter (68%), and men with female experimenter (62%).
  • Women report self-esteem scores comparable to men after taking a math test administered by a female experimenter (29.19 vs. 27.35 on scale of 7 to 35), but report lower self-esteem scores than men after taking a math test administered by a male experimenter (23.76 vs. 27.40).

Study 2

  • Women score higher on a math test when their female experimenter is perceived to be highly competent in math, compared to when their female experimenter is perceived to be less competent in math (12.18 vs. 9.23).
  • Men score lower on a math test when their female experimenter is perceived to be highly competent in math, compared to when their female experimenter is perceived to be less competent in math (12.79 vs. 15.90).
    • There was a trend for men to answer slightly more problems when their female experimenter is perceived to be less competent in math compared to students in the other three groups.
  • Women score lower than men on a math test when the female experimenter is perceived to be less competent at math (9.23 vs. 15.90).
    • Women and men had comparable levels of performance accuracy when the female experimenter is perceived to be highly competent (63% vs. 59%).
    • Women had lower accuracy and men had higher accuracy when the female experimenter is perceived as less competent (69% vs. 46%).
  • Men did not have any differences in reported self-esteem whether taking a test with a highly math competent female experimenter or a less math competent female experimenter (28.52 vs. 28.50), while women reported a trend in lower self-esteem in the high competency scenario and higher self-esteem in the low competency scenario (24.14 vs. 28.74).

Study 3

  • As before, women score higher on a math test when their female experimenter is perceived to be highly competent in math, compared to when their female experimenter is perceived to be less competent in math (8.02 vs. 5.63 on 15-question test).
    • As before, there was no significant difference in the number of problems answered.
    • As before, there was a significant difference in performance accuracy, with women in the high math competence scenario performing more accurately than women in the low math competence scenario (63% vs. 48%).
  • Women have higher scores for self-appraised math abilities when the female experimenter is perceived to be highly competent in math compared to when under the female experimenter is perceived to be less competent at math (18.61 vs. 16.82 on a scale of 5 to20). 
  • Women have higher self-esteem when the female experimenter is perceived to be highly competent in math compared to when under the female experimenter is perceived to be less competent at math (27.17 vs. 22.85 on a scale of 7 to 35).

The perception of a female experimenter who is highly competent in math buffers highly motivated and math identified women’s performance on math tests, and leads to scores that are comparable to equally talented men. These same women perform worse when the same test is administered by a male experimenter or a female experimenter who is perceived to be less competent in math. Additionally, women report higher self-esteem and higher self-appraisals of their math abilities when the female experimenter appears highly competent in math, effectively protecting them from the negative effects of gender stereotypes. The authors suggest these findings could inform educational policy by identifying the importance of a female role model for female college students in fields where negative gender stereotypes exist.

Methodology

Across all 3 studies, participants were required to have a minimum math SAT score of 650, have taken at least one math course in college, and have responded highly (≥3 on a 5-point scale) on 4 screening questions gauging their interest and ability in math (e.g., “How important is it for you to do well on your math exams?”). Each participant reported to a laboratory where an experimenter presented them with a math test. To appear like a competent mathematician, the experimenter told the participant they had written the exam and that they intended to provide the student with feedback on their results, although no feedback was ultimately provided.

Each participant has 25 minutes to a complete a 25-question math test resembling a section of the standardized Graduate Record Exam (GRE). After the test, participants were asked to rank 7 statements regarding their current feelings of confidence on a 5-point scale (the performance subscale from the State Self-Esteem Scale). The results of the math test, performance state self-esteem scale, gender of the participant, and gender of the examiner were tested using analyses of covariance (ANCOVA) to establish statistical significance, after adjusting for their SAT scores.

Study 1 involved 20 female and 20 male participants and both male and female examiners to evaluate whether participants’ test scores differed based on the gender of their examiner.

Study 2 involved 20 female and 20 male participants to evaluate whether the perceived competence of a female examiner impacted participants’ test scores even when the examiner was not in the room. The participants never met the examiner, but were each assigned a short biography designed to signal that the female examiner was either a peer who was very competent at math or a peer who was less competent at math.

Study 3 included 40 female participants to evaluate the role of self-appraised math ability. The study was identical to Study 2 with a slightly shorter math test (15 questions), but prior to the test participants additionally completed a set of 4 questions about their self-appraised math abilities (e.g., “I deal poorly with challenges in math”) on a 5-point scale.

Related GAP Studies