FACULTY Q&A

Justin Frake is interested in cause-and-effect relationships in real-world data and the hidden dynamics that shape workplace behavior and equality—or inequality, as the case might be.

His curiosity has led to research whose findings challenge some popular beliefs as well as published studies related to women in the workforce. One study shows that firms promoting flatter hierarchies inadvertently discourage female applicants and another counters several recent studies that claim women CEOs negatively impact career outcomes of other women.

The assistant professor of strategy at the University of Michigan’s Ross School of Business discusses his work as well as future research into other forms of workplace behavior and discrimination.

Two of your recently published studies either turn some conventional wisdom on its head or reveal an unintended bias in the hiring or advancement of women in the workplace. What led you into these subjects?

When I worked at a large public accounting firm, most of the leadership were men. I saw the male and senior managers were more likely to choose other men to work on their teams.

It often wasn’t explicitly because they didn’t like women. Instead, it was because they felt more comfortable traveling with other men than women. I also saw that when asked why they choose so many men to be on their team, they didn’t even realize it was happening and couldn’t articulate a reason.

All of this made me interested in how inequality in the workplace happens without conscious bias against women.

What surprised you the most?

In the study about hierarchies and applicant diversity, I was somewhat surprised to find that, relative to men, women perceive flatter hierarchies as offering fewer career progression opportunities and higher workloads, contrary to the popular belief that flatter structures are universally appealing.

In the study examining whether women CEOs affect the career outcomes of other women, we show that these studies suffer from a serious statistical issue called collider bias. Once we correct the statistical issue, we show women CEOs do not seem to hurt other women’s careers.

This finding was particularly striking as it challenged published studies that purported to show that female leadership may hurt gender equality. These studies underscore the importance of scrutinizing widely held beliefs and using rigorous statistical methods in research.

Can you describe collider bias and what harm it may cause in business settings?

Collider bias occurs when researchers control for or select a sample based on a variable that is affected by both the independent and dependent variables, leading to the illusion of a connection between variables that might not actually exist. The potential harm stemming from the previous findings comes from the misconception that research suggests female CEOs detrimentally impact the careers of other women within the firm.

If corporate boards read the previous research, they might conclude that they should refrain from promoting women to CEO positions under the guise of supporting other women within the firm.

Where is your research leading you now? Are you building on either of these studies or moving into new directions?

I’ve started a new project to understand the role of partisanship in the workplace. For example, do workers select into firms where there are employees that share their partisanship? Do managers hire people who share their partisanship?

This is related to the questions on gender, but it is not against the law to discriminate on the basis of political partisanship, so this kind of discrimination may be similar or even more prominent than gender or racial discrimination.