Implicit Biases & Evaluating Job Candidates (updated)

(Updated 11/7/13 to add link to Trix and Penska paper below. Thanks to Joanne Kamens for providing the link in the comments!)

(Note: this is the second in an occasional series* relating to job searches and hiring, though this one applies more broadly as well. The first post dealt with “illegal” questions, and appeared here.)

As much as we like to think we are all completely fair and unbiased, there is abundant evidence that we all have biases that influence how we think and act. These are known as “schemas”, and provide us with a framework for interacting with others. Schemas can be good – they allow us to more efficiently and rapidly process information – but they also can cause problems, in that sometimes our schemas lead us to treat people differently based on age, gender, sexual orientation, race, etc., in a way that we would not wish to.

We all do this – no one is immune to these implicit biases that influence how we act. This point was driven home to me clearly when I was a postdoc. I was a postdoc working on theoretical ecology projects, and viewed myself as someone who was strongly supportive of women in science. Yet, one day, I was at a seminar given by a woman who was presenting a lot of theory. I found myself wondering who she collaborated with on the mathy stuff. As soon as that thought popped into my head, I was shocked. How could I think that she needed to collaborate with someone in order to do theory? I was a woman doing a theory postdoc, for goodness’ sake. I was kind of horrified. But it fits in with what lots of studies have shown: everyone has schemas that affect how they perceive and treat other people.

What is the evidence? I’ll go into some of it here, because I think it’s important to cover. Then, in a later post, I’ll cover some related topics, including stereotype threat, and what we can do to try to overcome our biases.

A large set of evidence of how our schemas influence our evaluations of others comes from CV/resume studies. To summarize a few of these:

• Race: One study (by Bertrand and Mullainathan) sent out resumes in response to help-wanted ads in Boston and Chicago. The study found that, in order to get a call back for an interview, applicants with typically black names (e.g., Jamal, Lakisha) had to send out 50% more resumes than did applicants with typically white names (e.g., Emily, Greg).
• Gender: A study by Moss-Racusin et al. sent out applications for a lab manager position that had either a male or a female name. They found that the applications with male names were viewed as more competent and hireable, and were offered higher starting salaries.
• Gender: Steinpreis et al. (pdf link) found that psychology professors (male and female) were more likely to hire someone named “Brian” as compared to someone named “Karen” for an assistant professor position.
• Sexual orientation: A study by Tilcsik involved sending out resumes that were identical except that one indicated the applicant had been a treasurer in a gay student organization, whereas the other indicated that the applicant had been a treasurer in an environmental a progressive**** student organization. The “gay” applicant received 40% fewer call backs for interviews.
• Parental status: Correll et al. (pdf link) found bias against mothers, but not against fathers. They sent out a pair of resumes of applicants with the same qualifications, but where one indicated parenthood and the other did not. Non-mothers received call backs twice as often as mothers did. There was no difference for fathers vs. non-fathers.

There are also plenty of studies that do not use the paired CV/resume approach. One well-known example is from Wennarås and Wold, which looked at the success of applicants to the Swedish Medical Research Council. They found that women had to be 2.5 times as productive to be viewed as equally competent. Similarly, a study by Ginther et al. looked at success rates for applicants to the US NIH for R01 awards. They found that “compared with NIH R01 applications from white investigators, applications from black investigators were 13.2 percentage points less likely to be awarded (P < .001), and those from Asian investigators were 3.9 percentage points less likely to be awarded (P < .001).”

These schemas also influence letters of recommendation that are written for applicants. Trix and Penska (pdf link) looked at over 300 letters of recommendation that had been written for successful applicants for faculty positions at medical schools. Among other things, they found letters for women tended to be shorter and tended to use more “grindstone” adjectives (e.g., “hard-working”, “conscientious”, “diligent”, etc.). A short letter can indicate there isn’t much positive to say, and tends to be viewed as a negative for that candidate. And while being hard-working or diligent is a good thing, as Trix and Penska say, “There is an insidious gender schema that associates effort with women, and ability with men in professional areas. According to this schema, women are hard-working because they must compensate for lack of ability (Valian, 1998: 170).” They also found that letters for men tended to repeat “standout” adjectives (e.g., “superb”, “outstanding”, “exceptional”) more often than do letters written for women.

Back to bias in terms of how we evaluate people: A survey of managers by McKinsey & Company found that “women are often evaluated for promotions primarily on performance, while men are often promoted on potential.”

Finally, success of men and women tends to be attributed to different things. A “citation classic” by Deaux and Emswiller (pdf link) found that success of men tends to be attributed to skill, while success of women tends to be attributed to luck. This particular study is older (from 1974), so hopefully some of these attitudes have shifted by now!

What can be done about this? I’ll cover this more in a future post. But, for now, I will just say that, when evaluating applications, I find it important to keep in mind that we all have these biases.

*By “occasional series”, I mean “your guess is as good as mine as to when the next post will appear.”

***Note: I first got interested in this literature as a postdoc, and have followed it as much as I can since then. But I was reminded of some of these studies, and found new ones, thanks to a really informative workshop run by the Strategies and Tactics for Recruiting to Increase Diversity and Excellence (STRIDE) committee at the University of Michigan.

****Updated 11/18/13 to correct mistake about control group. Thanks to Lirael for pointing this out!