“Work-Life Balance” and “Empathizing” Do Not Explain Women’s Career Choices

viral letter by then-Google employee James Damore has renewed the conversation about diversity in Silicon Valley. One thread of the ensuing debate has focused on the scientific validity of Damore’s claims that men and women do in fact differ in their preferences. An unspoken assumption has been that differences in preferences—if such differences exist—would go a long way toward explaining why women have remained underrepresented in tech and similar fields, despite efforts to increase diversity.

This assumption can be put to the test. Damore claimed men and women differ on two major preferences: maintaining work-life balance and empathizing with others. If we knew how much various jobs allow people to maintain a good work-life balance and empathize with others, we would be able to see whether jobs that do not satisfy these preferences have fewer women in them. If this was not the case, Damore’s argument would be undercut. For instance, if women aren’t generally underrepresented in fields where you routinely have to work 80 hours a week, why would that be a factor in tech?

Fortunately, the data to weigh this key assumption are available. They were collected for a study that I published a couple of years ago with my collaborators Sarah-Jane Leslie, Meredith Meyer, and Edward Freeland. We asked members of 30 fields across the academic spectrum how many hours they worked per week and how much their field involved empathizing versus systemizing (which is the ability to reason about and analyze systems, thought to be higher in men).

We then used these responses to predict women’s participation in a field. If Damore is right, we should see fewer women in fields in which the workload is particularly heavy, or in fields that require more systemizing than empathizing.

The relationships we found were too weak to support Damore’s claims. For example, workload explained an underwhelming 0.09 percent of the distribution of women across fields. It is most definitely not the case that women are scarce in fields that require long hours.

What did predict the distribution of women across careers? The extent to which a field was believed to require “innate talent”—a special aptitude that can’t be taught.

Initially, the second prediction seemed more promising. Empathizing–systemizing showed a statistically significant relationship with women’s participation across the 30 fields in our sample. However, we then tried to look just at the fields in science and technology, fields that vary quite a bit in their gender composition. Perhaps some sciences have been more successful than others in attracting women because they allow women to express their preference for people and empathizing (while also requiring less systemizing). This wasn’t the case. Only about seven percent of the variability in gender gaps across the sciences was explained by this empathizing–systemizing factor—far lower than what you’d expect if women’s scientific careers were guided by a desire to empathize.

What did predict the distribution of women across careers? The extent to which a field was believed to require “innate talent”—a special aptitude that can’t be taught. In other words, brilliance. This single variable explained 41 percent of the variability in women’s participation across science and technology fields. Beliefs about brilliance pose an obstacle for women because our society—including software engineers, presumably—still perceives women as less intellectually gifted than men. And this is, of course, only one of the many unfair obstacles that women in tech have to face daily.

I don’t have field-by-field data for the other personality dimensions Damore lists, such as extraversion and neuroticism. Perhaps they might do a better job predicting women’s career paths, but I doubt it. The more general point here is this: Even if gender differences in personality or preferences exist, we shouldn’t assume that they are a good explanation for gender differences in career outcomes. No matter how intuitive it may seem that women don’t pursue software engineering because they prefer being around people, intuitions are an unreliable guide to the truth. We should ask to see the data.

This article has been updated.

Further Reading & Resources

  • Leslie, S. J., Cimpian, A., Meyer, M., & Freeland, E. (2015). Expectations of brilliance underlie gender distributions across academic disciplines. Science347(6219), 262-265. (Link)
  • Stevens, S., & Haidt, J. (2017, August 22). The Google Memo: What Does the Research Say About Gender Differences? Heterodox Academy. (Link)
  • Halpern, D. F., Benbow, C. P., Geary, D. C., Gur, R. C., Hyde, J. S., & Gernsbacher, M. A. (2007). The science of sex differences in science and mathematics. Psychological science in the public interest8(1), 1-51. (Link)