02.10 – Stereotypes in Performance Evaluations

In addition to rater errors, stereotypes can significantly affect the accuracy with which we evaluate other’s performance. And I’m not talking here about deliberate, willful discrimination. I’m talking about much more subtle, and yet very pervasive and potent effects. That often operate at a subconscious level. So the goal of this discussion would be to show you some data on how powerful these effects can be, learn to recognize those effects, and perhaps make us pause and think about their impact on performance evaluations.

What I’m showing you here is results of a meta-analysis. That compares women’s performance ratings relative to men’s. Meta-analysis is a type of study, again, that aggregates results across a range of empirical studies. So it’s a combined effect across many, many different samples. It further breaks down the results by the percentage of female membership in the group. The d-score, the difference score is women’s performance ratings minus men’s performance ratings. So you can see that negative scores indicate that women receive lower ratings that men. That difference is further divided by a group standard deviation. We need to do that to be able to compare across different scales, performance scales that companies use. So take a look at these data.

What do they tell you?

Well, you can first notice that on average, women are disadvantaged in performance evaluations but only very slightly compared to men. The overall effect is -0.07.

But a more alarming finding perhaps is that women tend to be penalized in those groups where they’re in the minority. You can see how they receive the lowest scores in those groups when they represent under 20% of group membership. -0.4, -0.5, that’s a base significant effect. That’s a half of a standard deviation difference, compared to men’s. And you can see that they continue to receive these low scores up until they reach about 50% in group membership. The effects of group composition hold even after we control for such things as your general cognitive ability. Verbal ability. Numerical ability. Psychomotor ability such as manual dexterity, finger dexterity. Motor coordination skills. Levels of education. Levels of experience at the firm. So pretty powerful.

Some of you may have followed the high profile gender discrimination case in 2015, which involved one of the leading firms in venture capital in Silicon Valley. A female employee of this firm, who didn’t make partner, accused the firm of gender discrimination. And one of the claims in that lawsuit was a claim about biased performance evaluations. Well, based on the numbers I’ve shown you, she may have had a reasonable case there. Because the number of female partners in venture capital is at about 6% now. And it’s been steadily declining since the late 90s when it was around 10%.

Coming back to the overall effect, again I want us to recognize is that men are favored in workplace evaluations, but only very slightly. The overall effect ranges from -0.07 to -0.01 in terms of the standard deviation. By comparison for instance, the effect on hiring is about seven times stronger in terms of men being favored over women for exact same positions. But there are very specific sets of circumstances where this gender bias becomes particularly acute, and I would like for us to be aware of those circumstances.

To get there let’s first understand where does this bias come from. See would develop diffuse expectations of others. Which are based in part of our beliefs on what are appropriate behaviors for a given gender. Our stereotypes about women lead us to expect women to have a strong degree of what we call communal attributes. That they’re friendly, unselfish, concerned for others, emotionally expressive, nurturing. Our stereotypes about men, allude us to expect men to have strong agentic qualities, which are they’re independent assertive powerful confident able to control others, and let’s consider two broad variations in leadership styles. One is what I would label as democratic, which is very participative, where the leader actively encourages others to participate in decision-making.

And the second is more autocratic or more centralized. Where the leader controls the decision-making process. As you can see, our stereotypes about the communal attributes of women clash particularly strongly with this autocratic leadership style, where the decision-making is centralized. And not surprisingly, that’s when we tend to penalize females in performances evaluations most strongly. The difference is about -0.3 in terms of the standard deviation. Notice too, that men are not comparably penalized for adopting a democratic leadership style. So, men have many more degrees of freedom in terms of choosing from repertoire of available leadership styles. This gender bias in evaluation carries into compensation as well.

So a study by Emir Castilla at MIT shows that salary growth is about 0.4% lower for women than it is for equally performing men. There’s also evidence for racial bias and evaluations, where in implicit association tests people are more likely to produce negative evaluative associations with black faces relative to white faces. So, must faster to associate unpleasant words with faces of African Americans, for example. This bias carries into rewards and compensation. Castila reports that African Americans and Hispanic employees receive a salary increase that on average is about half a percent lower than equally performing white employees. When you decide to raise the subject of equality in workplace.

Research by Brown, Lowery and colleagues out of Stanford suggest that there are two ways you can frame this conversation. And it may be beneficial to frame the conversation not just in terms of one group’s disadvantage, but also the other group’s advantage. So for example, one way to frame this conversation is to say that Group X, say women, are disadvantaged in workplace when they happen to be in the minority. But another way to frame that conversation is to say that Group Y, men, enjoy an unfair advantage in a workplace when they happen to be in the majority.

And what the research shows is that the second type of framing of a conversation that elicits a particularly positive, productive response by the advantaged group because it makes the costs and burdens of this undeserved status more salient. All right, at this point, I wanna shift gears and talk about this very unusual study that looked at the relationships between the facial symmetry of professional quarterbacks playing in the National Football League between 1995 and 2009 and their salaries. A quarterback is arguably the most important position in American football. It’s the person that directs the offense of the team. And software scan multiple, multiple, pictures. Analyzing both vertical and horizontal symmetry of quarterback’s faces by counting the number of pixels.

It turns out that the more symmetrical of find the face to be, the more attractive we find the person. The finding of the staple was that a change in facial symmetry of a quarterback, from one standard deviation below the mean to one standard deviation above the mean, result in a salary increase of about 11.8%. These effects are not limited to sports of course. A study by Christian Pfeiffer shows that workers who get one point higher physical attractiveness rating. In this case an 11-point scale, or an average 3% more in monthly income. Foreseeing one of your questions, the average of physical attractiveness in the sample was 8 with standard deviation of 1.7.

So pretty attractive group overall, but these two studies that I mentioned are just some of the examples in a very vibrant line of work on Beauty-is-Good Bias. It shows that physically attractive people are strongly favored in evaluations and rewards. And you can see that the overall effect ranges from 0.49 to 0.61 in terms of standard deviations. It’s about an order of magnitude higher than the comparable gender bias effect. The reason for this bias in evaluations and it’s typically discussed in psychology under the rubric of implicit personality theory, is that we tend to assign all sorts of positive attributes to physically attractive people. We believe that they’re more intelligent. We believe that they have high levels of integrity.

The ability to be more socially competent and even have higher status. And you can see that those attributions translate into bias evaluations. And bias in rewards. The goal of this discussion was to draw attention to and raise awareness about the key stereotypes that permeate performance appraisal processes. We discuss those that relate to gender, race, and physical attractiveness. Keep those in mind next time you’re filling out a performance appraisal form.