Skip to main content

Do female and male role models who embody STEM stereotypes hinder women’s anticipated success in STEM? (Cheryan et al. 2011)

Citation

Cheryan, S., Siy, J., Vichayapai, M., Drury, B. & Kim, S. (2011). Do female and male role models who embody STEM stereotypes hinder women’s anticipated success in STEM? Social Psychological and Personality Science, 2(6), 656-664.

Highlights

  • The study’s objective was to examine the impact of interacting with an upper-level college role model on female students’ beliefs that they could succeed as a computer science major.
  • The study used a randomized controlled trial to assign female students majoring in a discipline other than computer science to meet with one of four different types of upper-level students who varied by gender and whether they embodied perceived stereotypes of computer science majors. The authors administered a questionnaire to estimate impacts of the different intervention conditions.
  • The study found that women who interacted with a stereotypical computer science role model believed they would be less successful as a computer science major compared with women who interacted with a nonstereotypical computer science role model and with women who did not interact with a role model.
  • The quality of causal evidence provided in this study is low because the study had high attrition and the authors did not demonstrate baseline equivalence between the samples or include adequate controls to ensure the resulting groups were similar on all relevant characteristics. This means we are not confident that the effects estimated in this study are attributable to role model interaction; other factors are likely to have contributed.

Features of the Study

This randomized controlled trial assigned 85 female undergraduates with a major other than computer science to meet with one of four different types of upper-level students intended to serve as role models: (1) male stereotypical computer science majors, (2) female stereotypical computer science majors, (3) male nonstereotypical computer science majors, or (4) female nonstereotypical computer science majors. Before conducting the experiment, the authors developed these prototypes using results from an undergraduate survey asking about styles of dress; hobbies; and preferences for television, movies, and magazines associated with computer science majors versus typical college students. The authors recruited upper-level students to pose as role models embodying these stereotypes for the purpose of the experiment. A fifth group of study participants was randomly assigned to the control group; they were told that their assigned role model was unable to attend the meeting as planned.

After interacting with their role models (or not, in the case of the control group), study participants answered two questions about their expected performance if they were to major in the same field as their role models. The authors used analysis of variance to estimate differences in attitudes across the various study conditions.

Findings

  • The study found that women who interacted with a role model embodying computer science stereotypes expected to be less successful in computer science than women who interacted with a nonstereotypical computer science role model.
  • Women who interacted with the stereotypical computer science role model expressed lower success beliefs than women who did not interact with any role model.
  • There was no difference between the nonstereotypical treatment group and the control group, and no difference by role model gender.

Considerations for Interpreting the Findings

The authors did not provide sample assignment or attrition information by study condition to enable CLEAR to calculate differential attrition rates; however, they did report overall attrition. CLEAR estimated worst-case-scenario differential attrition by attributing all study attrition to one condition or the other, which resulted in high differential attrition in either case. Although the study was a randomized controlled trial by design, the high attrition made it ineligible to receive a high causal evidence rating; thus, it was treated as a nonexperimental design for this review. The authors did not demonstrate baseline equivalence between the samples or include the control variables required by the protocol for nonexperimental designs, so the study could not receive a moderate causal evidence rating either.

Causal Evidence Rating

The quality of causal evidence provided in this study is low because the study had high attrition and the authors did not demonstrate baseline equivalence between the samples or include adequate controls to ensure the resulting groups were similar on all relevant characteristics. This means we are not confident that the effects estimated in this study are attributable to role model interaction; other factors are likely to have contributed.

Reviewed by CLEAR

September 2015