Performance Management And The Paradox Of Meritocracy

Do you believe that your organization is inclusive? Do you believe that everyone is being treated fairly and you provide equal opportunities? Do you believe that you have a meritocratic performance management system and employees are rewarded based on their contribution to the company’s success? Don’t be so sure!

Even things like performance management, where it feels like you are working with objective data, are prone to bias. Employees have clear goals and are evaluated against these goals by their supervisors. This data then translate into some numerical appraisal rating. This rating then translates to bonuses. You have a foolproof meritocratic system. Well, you don’t. The numbers may appear objective, but they are put together based on biased individuals’ subjective thoughts. Who decides what goals or tasks will each team member work on? Who chooses the success criteria? Who decides whether they were met? Who then evaluates the impact of the job? These are all things that are very subjective, prone to unconscious bias, yet they masquerade as objective.

Imagine you manage a team of salespeople. At the beginning of the month, you give each of them a hundred accounts to work on. Unconscious bias kicks in. You can’t be entirely fair, and so some got better accounts than others. Now, some are more likely to sell more and bring in a bigger number. And indeed, at the end of the month, some vastly outsold the rest. Why? Because they are better at sales? Or because they got better accounts? If you reward just the revenue they brought in, you may claim to be meritocratic, while in reality, you are being biased against those who got dead accounts. Being fair is much harder than it looks.

Unconscious bias leads to favoritism and hurts companies badly. The moment the team believes there is favoritism at play, you lost the productivity of everyone. Those who are discriminated against feel there is no point in trying hard, and those who are favored feel there is no need to try hard. Everyone loses. Not to mention that eventually, wrong people are promoted and end up doing work they are not qualified for, while those who are qualified leave the organization.

The subjectivity of performance management

The problem with performance management systems is that no matter how much based on data you try to make it, you won’t make it entirely algorithmic. Managers want to have their input. They feel that they know best their teams and know who has the potential and who doesn’t. Data doesn’t matter. Or rather, data doesn’t tell the whole story. You hired managers to manage, to exercise their judgment. And for any individual, it is impossible to be completely objective. The challenge is to find the right balance between hard data and informed manager’s decisions.

To design more diversity and gender equality into the organization, make sure you collect and analyze data. Focus on understanding patterns, correlations, and trends. Measure things to understand what needs to be fixed, and don’t be afraid to experiment. Provide your managers with a lot of data and algorithmic suggestions while allowing them to adjust the results. They will do this consciously rather than just following an unconscious bias.

Self-rating in performance management

The gender compensation gap is widely talked about. However, the gender gap shows also in an unexpected place—self-assessment during performance appraisals. Many companies ask their employees to self-assess their performance before their manager does it. Unfortunately, this automatically creates a disadvantage for less confident individuals, and since there is a gap in confidence and self-promotion between men and women, it puts women at a disadvantage. Once the managers see the self-evaluation, their own unconscious biases kick in. The person whose performance is being assessed created an anchor that the manager then unwittingly uses to guide his or her decision. Let’s say you have equally performing employees. One is overconfident and rates himself as nine out of ten. One is less confident and rates herself as seven out of ten. You then put in your assessment and maybe feel that nine is too generous for the overconfident man, so give him eight. But the seven the woman gave herself sounds about right. Why give her a better rating than she considers fair herself? You are just a human, prone to unconscious bias, and can’t help yourself not to take into account the anchors the employees created.

The way to fix it is not through educating the managers but through designing a process that removes the possibility of unconscious bias, in this case, the anchoring effect. Don’t let the managers see the self-evaluation before they make their mind. Or even better, don’t do self-evaluations as they don’t represent the employee’s performance but their confidence level.

The paradox of meritocracy

Researchers Emilio J. Castilla and Stephen Benard looked into how biases work in meritocratic organizations. They found what they called the paradox of meritocracy. As it turns out, when organizations are explicitly meritocratic, it leads to managers favoring and rewarding male employees more than equally qualified female employees. It shows that achieving true meritocracy is not as easy as it sounds, and you can’t discount gender and racial biases even in meritocratic systems.

Castilla’s research showed that women in a merit-based organization, reporting to the same supervisor, with performance rated at the same level, received a lower salary increase than their white male colleagues.

Castilla and Benard suggest that emphasizing meritocracy as an organizational value to reward employees fairly may lead to the opposite outcome. Merit-based rewards are often seen as motivating for the employees and leading to higher performance. However, they may lead to more bias, and the link between compensation and performance is also questionable. The culprit is most likely moral credentialing when people who are led to feel unbiased and fair are more likely to behave in biased ways. The unconscious monolog in people’s head goes like this, “as an organization, we are meritocratic and unbiased, that’s who we are, so how could I be biased?” With this mindset, it is easier to succumb to one’s prejudices and act on them. It seems that the meritocracy paradox is quite visible in salary increases and payout of bonuses but not necessarily in promotions. Researchers argue that promotions are highly visible and easier to detect as they occur less frequently. Bonuses fall under less scrutiny and therefore are more likely to be impacted by unconscious bias rooted in moral credentials.

Putting it all together

Pay for performance and other merit-based systems would work great if it were not for prejudices and implicit biases. In an ideal world, there is no prejudice, moral credentials wouldn’t cause trouble, and the system would be genuinely meritocratic. Unfortunately, in a society full of prejudices, the explicitly stated merit-based systems give us the internal moral permission to act on these prejudices, and therefore, they lead to inequality.

To build a genuinely meritocratic system, it is not enough to claim the value of meritocracy. You need to design processes and policies in a way that limits the opportunity for biases and prejudices to come into play, and that often means relying more on algorithms, transparency, and accountability and less on managerial discretion.

What is your take on the topic? What do you think of paradox of meritocracy? Do you work in a meritocratic organization? What processes and policies are in place to limit biases? What is the biggest obstacle to having a true meritocracy?

Photo: geralt /

Follow me on Twitter: @GeekyLeader

Categories: Diversity, Performance

Tags: , ,

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: