I’ve heard that sentence in different forms hundreds of times over the past five years in working with diversity in tech. The belief in meritocracy in tech is pervasive. A meritocratic system is one where equal merit results in equal rewards.
It’s pleasing to think that everyone is on equal grounds, and that only the results of your work influence your success. Pleasing, but naive.
In this blog post, we’re going to explain, through scientific, peer-reviewed evidence, why and how meritocracy, in the absence of systems ensuring transparency and accountability, is a myth.
Specifically, we are going to talk about two specific places in one’s career where merit has been shown to be overruled by bias in multiple studies: in hiring, and in promotions and pay raises.
The thing that prevents merit from being the most important factor in both cases, is bias. Bias is defined as a “prejudice in favor of or against a thing, person or group, compared with another.” Biases can be conscious or unconscious. Everyone has bias, including the authors of this blog. In fact, we challenge you to take this well known, simple bias test from Harvard to test your own biases, particularly if you don’t believe that you’re biased.
A note before we get started: We’re not arguing here that the ideal of trying to create a meritocratic organization is a flawed goal. The myth of meritocracy that we’re talking about comes into play when people assume (or believe, or expect) that organizations are simply meritocratic by default.
The first place that merit fails is in recruiting. In one study, from 2012, the same resume and application materials for a science lab manager job were sent out to 127 biology, chemistry and physics professors at American universities. The researchers sent out the same application to all the professors, but changed the names on the application so that half had a male student’s name, and half had a female student’s name.
The professors, who were both men and women, had to rate the applications on a scale of 1-7 in terms of competence, hireability, how much mentoring they’d give the candidate, and starting salary. The results were staggering: the application with the woman’s name was on average rated lower in competence, hireability, and how much mentoring the professors would give. The “female” application package received, on average, an average wage that was more than $3500 less than the “male” applicant per year.
The bias appeared in both female and male professors who were reviewing the packages. That is, women and men are equally biased.
The second place that merit fails is in promotions and salary increases. There have been multiple studies on this. One study focused on the link between performance evaluations and wage determination.
In that study, the researcher, Emilio Castilla, studied a research and IT company with 20,000 employees in the US. What’s interesting was that this company prided itself on being diverse, and it was, in fact, quite diverse, with over 30% minority employees, and 67% female employees. The study features data from just under 9,000 employees.
The company, was of course, convinced that it was a meritocracy, saying “performance is the primary basis for all employee salary increases.”
The way promotions worked was that first, a supervisor completed a performance appraisal of each employee by meeting with them annually. Then, a second manager, usually higher in rank than the supervisor, would recommend (or not) the employee for a salary increase or bonus. (Castilla 2008) The performance review needed to be completed for any employee to receive a salary increase.
HR had to then also approve any salary raises, as a double check after the managers and supervisors. However, in practice, HR was simply approving all the requests made, as long as the written paperwork was there.
In this system, bias in salary increases was prevalent.
In cases of the exact same merit, same job, same unit and same manager, and entirely equal performance review ratings, salary growth was still 0.4% lower for women than for men. African-American employees got a salary increase that was 0.5% lower than equally performing white employees. Hispanic workers received salary increases that were 0.5% lower than equally performing whites. Non-US-born employees received 0.6% lower salary increases than native employees, all else being equal.
Thus, employees who performed equally well at the exact same job, under the exact same manager, got different salary increases even after they received the same performance review scores.
Interestingly, bias did not occur in the performance reviews themselves, when managers were held accountable because they knew that a higher manager would be reviewing their work. Bias only occurred when the higher managers were allowed to give salary increases without being held accountable regarding the amount.
According to the study, “bias is more likely to occur when the structural conditions are such that there is more discretion, less accountability, and less transparency.”
When decision makers know they will be held accountable to someone else for their decisions, bias is less likely to occur. This also means that having formal systems where someone else checks the work of the first person will be less biased.
That being said, though, it only works if the second checker actually checks. HR simply approving all requests for raises was not an efficient enough check to prevent the bias.
Secondly, transparency prevents bias by making gaps noticeable. The most visible parts of the performance review processes (who got salary increases, who received promotions) were not affected by bias. The process that was not transparent (the amount of the raise) was affected by bias. It was hard for people to compare their raise amounts with others, which in turn, allowed bias to go unchecked. This lack of transparency was especially true since the salary increases were low, never more than 8% of the base salary.
Because of these factors, many supervisors and administrators remained unaware (during the study) that their behavior was profoundly biased.
The researcher finds that women and minorities “need to work harder and obtain higher performance scores in order to receive similar salary increases to white men.” This double standard would not be characteristic of a truly meritocratic organization.
There is hope, though! At the same company, after implementing a series of changes, such as creating a performance-reward committee, and consistently keeping management updated with pay data statistics disaggregated for all genders, races and nationalities, the wage gap was gone after five years.
For blind recruiting, there’s a start-up helping companies with blind applications processes, called GapJumpers, where you can post jobs with blind applications processes.
It’s not impossible for an organization to become a true meritocracy, as this study shows. Through careful planning, transparency and accountability, and an awareness of bias, unbiased promotions and raises can happen. Through blind hiring, true meritocracitocractic recruiting can in fact also exist. But the belief, assumption, and expectation that plagues this industry, that a company just somehow, by default, “is” a meritocracy without carefully planning out exactly how that outcome will be achieved, is a myth.
Thank you to our reader Galina Shubina for her suggestion to cover this topic!
Follow the GETBlog here.