You worked hard to get that job. Really hard. It’s all there on your CV, from top grades to promotions. Psychologists and economists call these markers “competence signals” and they’re what organisations use to draw inferences about your ability when they’re hiring. Hooray, you’re thinking. All that effort pays off.
Well, yes and no. Because, if you’re a woman, it’s a little more complicated. Once you’re in the job something curiously unrewarding can happen. Those same glorious accomplishments can work against you. Specifically, when it’s time for your review.
“Although competence signals can help during the hiring process, those same signals can have a detrimental effect on later performance evaluations,” says Ena Inesi, Associate Professor of Organisational Behaviour at London Business School (LBS).
This may seem like a stark claim, but it’s backed up by research that Dr Inesi carried out in conjunction with Dan Cable, Professor of Organisational Behaviour at LBS. Their starting point was a data set of real performance evaluations in one branch of the US military. “Given that the military tends to be hierarchical and male, we thought it would be a good place to test our ideas,” she explains.
They theorised that past competence signals, which are not relevant to evaluations of current performance, might nevertheless have a negative impact for high-performing women. They thought this might be the case because such information is threatening to the traditional gender hierarchy (in which men are at the top and perceived as more competent than women). On analysing the US Army data, they found a negative relationship between competence signal strength and performance evaluations, when the woman was performing at a high level in her present job.
“There were other studies suggesting this might be the case but to see it so starkly was surprising to us,” says Dr Inesi. “Women know they are going to encounter challenges: people are going to make assumptions about your competence based on your gender so you are going to do everything you can to prove yourself. What this research shows is that these accomplishments are looked upon in a threatening way by some individuals, which has a negative downstream effect.”
But is this down to gender?
Could there have been other factors at play? The researchers wondered about this too. “We controlled for everything we could think of but we still wondered, was there something we were missing? So we brought this hypothesis into the lab, where we could keep everything constant and then manipulate the gender of the individual and the signals, to find out. We needed to make sure there wasn’t something we didn’t know about these real women and men in the military. They might be actually different from one another in ways that we were not capturing.”
If they could show, through pure experiments, using profiles of imaginary people, where literally the only difference was that they were “John” or “Jane”, with either strong past competence signals or relatively weak ones, then they could safely conclude that, yes, this is about gender bias.
To detail one of the experiments: the researchers recruited 271 college-educated adults via the online marketplace Mechanical Turk to participate in an online study. Participants were asked to act as managers in a company that develops creative ideas to solve their clients’ problems. Before working on a task, they were told they would be collaborating with a subordinate, whom they would evaluate at the end of the study. These “subordinates” didn’t exist: they were invented profiles so that the researchers could manipulate the variables they wanted to test while keeping performance constant.
The subordinate’s name was blocked out but their age (26), gender, highest educational achievement and employment status were shared with the participant. Next the group was introduced to the project plan and given a task (involving them having to figure out what to do with 20,000 surplus pipecleaners that a client had bought but couldn’t use).
They were then shown a list of 10 ideas, ostensibly generated by their subordinate, and given time to develop one solution and submit it to their client. Finally (this is the important part), they were asked to evaluate their imaginary subordinate, on a scale of 1 (extremely bad) to 7 (extremely good). They were also asked a deliberately subjective question: “How happy were you with your subordinate’s performance?” and these items were combined to form a composite measure of performance evaluation.
For the exact same performance, female subordinates were given worse evaluations when they had a university vs high –school degree (i.e., a strong vs weak past competence signal). This was not true for male subordinates. This is striking, and a good indicator of bias, because this wasn’t framed as something relevant to the task. Also – notably – this was not true for all evaluators. The participants had also completed a pre-task scale measuring individual differences in preference for hierarchy and group-based domination and discrimination. It contained statements such as, “To get ahead in life it is sometimes necessary to step on other groups.”
As predicted, the pattern described above only emerged for male evaluators who tended to agree with the hierarchy-supporting statements. This makes sense when you put yourself in the shoes of such evaluators, says Dr Inesi. “If you are of the mindset that there’s a “right” hierarchy out there, with men in roles that have more power and status than women, then you are more likely to feel that a high-performing woman with great past credentials is somehow not right. This feeling of malaise ends up being reflected in things like performance evaluations.”
Steps towards a fairer system
“This may be an additional dynamic that prevents women from breaking through the glass ceiling,” says Dr Inesi. Our results suggest that being penalised for high competency may increase as women become more senior in their organisations and their track record becomes increasingly evident and increasingly threatening. By nature, competence signals build across an employee’s career, meaning positive performance ratings can become harder and harder for women to obtain as their careers progress.”
So what can be done? “The most obvious takeaway is that organisations should make efforts to make these evaluation processes as free of subjectivity as possible by trying to solicit as much feedback about the individual as possible from a variety of different sources, to get a more holistic and hopefully less biased evaluation. The old way of just one senior individual having to state how well a person performed over the past year on a scale of 1 to 5 is clearly inadequate.
“Organisations should have clear metrics for progress – objective markers of accomplishment, not subjective opinions on, say, whether this person seems smart. Did they deliver on x and y?”
They can also educate people about stereotypes and prejudices, she suggests. They should also watch out for partiality – a phenomenon explored in depth by Simon Keller, Professor of Philosophy at Victoria University Wellington. Essentially, we tend to like certain people more than others and end up treating them differently from those we like less – selecting them for projects, say, or giving them more encouragement. It can become a self-fulfilling prophecy that those people we like end up doing better while the people we’re less keen on perform poorly.
“The same can be applied to these gender issues,” says Dr Inesi. “Do I feel a greater affinity for someone of my own gender vs another gender – and, as a result, am I creating the reality where this man is performing better than this woman? You need to be aware of the possibility not just during appraisals but more generally: does the fact that you more naturally get on with someone who is similar to you mean you are giving individuals different opportunities?”