Minorities: are we qualified to say ‘less qualified’?
- Amy Nelson-
“We hire only the most qualified people”: this is perhaps the commonest defence of companies and institutions whose demographic makeup strays far from population patterns. Which seems fair enough, given that positive discrimination—hiring based on race, gender, or other baseline characteristic, without demonstrating equal qualification—remains illegal in the UK. Let’s stop to consider that hiring on meritocracy might not be the fair (or useful) strategy that it at first appears.
What do we mean when we say most qualified? Our exemplar likely attracts the most relevant qualifications and awards, the lengthiest work experience, the closest fit to job description, or the highest average of all three. Unfortunately, merit—association with top institutions, best schools, or competitive internships—remains the bed partner of privilege, and privilege biases us against certain types of people. The causal network linking qualification, wealth, race and gender is entangled such that we prefer candidates with a career path reverse engineered to meet the prima facie ideal, a process accelerated by wealth, rather than those with unproven skills and potential for growth.
This leads us to a key question: is past performance—measured by qualification—the best predictor of future success? Yes and no. Should our candidate be required to perform tasks very similar to previous experience, or to score highly in examinations that are biased in the same way as entrance tests, he or she will likely continue to perform predictably well. Dynamic, flexible competence and collaborative disposition, however, are not guaranteed, and may indeed have been selected against. Highly competitive people do not necessarily translate comfortably to the labour market—see medical school entrance criteria that create shortages of future GPs. The idea of meritocracy is not wrong: our ‘objective’ measures of qualification are flawed and yet followed without due caution.
The issue of why the ‘most qualified’ people are not necessarily the best hires can be framed as a statistical problem known as the bias-variance tradeoff, which explains why some predictions look great in sample data but generalise poorly to real life situations. When a group’s output faces the public—as in most tech, healthcare, and business situations—generalisability describes how faithfully our output translates to the real world. In his book, ‘The Wisdom of Crowds’, James Surowiecki observes that a bull’s weight, or the number of jellybeans in a jar, is more accurately estimated by groups than single experts only when the crowd is diverse (mathematically known as high variance). We need people who guess in either direction to create a balanced picture of truth: we need adequate variance to ensure generalisability. If hiring the most qualified people results in a room full of identikit characters, we limit our team’s capabilities.
The most convincing evidence against measures of non-diverse ‘most qualified’ lies in financial data. According to McKinsey analysis:
- Companies in the top quartile for gender diversity are 15% more likely to have financial returns above national industry medians, increasing to 35% for racial diversity
- In the UK, for every 10% increase in senior-executive gender diversity, earnings rise by 3.5 percent
Although correlation here does not equal causation, these observations warrant serious consideration of what ‘most qualified’ really means: are traditional measures sophisticated enough for hiring in non-diverse areas like tech, engineering and business leadership? Alternative solutions should be high dimensional, built to avoid bias, and matched to both individuals and teams (such as in collaborative filtering); our focus must shift from ‘most qualified’ to a fair and effective metric of ‘best hire’.
Amy is an FY3 Doctor and Senior Research Associate at UCL Institute of Neurology
Do pop Amy a line if you have any comments, thoughts or responses! She can be reached on: