August 2007 – Tackling bias in recruitment

How biased are you towards others? The question matters to all of us. To those in recruitment or management, regularly undertaking appraisals or selection, it matters a lot.

Beyond the obvious legal implications of discriminatory behaviour, it makes sense to ensure that the best people are doing the jobs where they are needed most. But nearly all of us display an unconscious bias in almost everything we do and often the bias is greater than we might imagine.

If you don’t believe me, can I suggest you visit the Harvard University web site and undertake one of its Implicit Association tests*. The online tests are part of a long-running research programme in to unconscious preferences called Project Implicit, shared between a number of universities in the US.

I came across these tests some time ago but didn’t pay much attention to them possibly because I was reluctant to believe my scores on one of the tests that showed a bias against people with disabilities.

Returning to the site this week it didn’t help to find that these results are quite normal. The data collected so far indicate that a third of those undertaking this particular test displayed a “strong automatic preference” for people without disabilities. Some 27 per cent of those undertaking the tests registered a moderate preference and 16 per cent, as I did, a slight automatic preference.

This means that more than three-quarters of people in the research to date have shown some unconscious bias in favour of those without disabilities. Some 15 per cent revealed little or no bias and the rest - less than 10 per cent - showed some bias in favour of the disabled.

In a similar test looking at bias in regard to black and white people, some 70 per cent showed some preference towards white people against 12 per cent in favour of black people and 17 per cent with no preference. I registered a slight automatic preference towards whites.

I had revisited the tests after a discussion about heuristics over lunch last week with Binna Kandola, visiting professor at Sheffield University’s Institute of Work Psychology and senior partner at Pearn Kandola, Oxford-based occupational psychologists.

In psychology heuristsics are sometimes described as inherent rules, either learned or innate, that influence our decisions. What some might describe as “gut feel” is a product of heuristics and, in some cases, can result in bias.

One example of such hidden influences is a predisposition in some people to buy things in shops that are expensive, based on a belief that if something carries a high price tag it must be desirable. Such items are sometimes referred to as “Veblen goods” in acknowledgement of Thorstein Veblen’s theories on status seeking through what he called conspicuous consumption.

As the Veblen goods phenomenon demonstrates, heuristics can lead to distorted judgements. Prof Kandola recalled a study that looked at training approaches in the Israeli air force. Officers deliberately abandoned congratulating the trainees who made good landings because they noticed a drop in performance among those who had done well previously.

The assumption was that giving people a pat on the back encouraged complacency when, in fact, the trainee flyers were simply regressing to the mean. The ones that had landed well had been fortunate and could not repeat their success every time. But this did not mean they would not benefit from encouragement.

The Implicit Association tests appear to expose our hidden prejudices and preferences over a whole range of subjects. But they have attracted some controversy in the past few years, partly because their feedback often runs counter to individual perceptions.

It’s not easy for someone who regards himself as an open-minded liberal to find that his behaviours may be guided by unconscious influences drawn from past experiences.

If you remain sceptical, however, try to observe your seating choices the next time you board a train. In my experience people on commuter trains are exceptionally choosy about where they sit. All kinds of factors influence their decisions, primarily ones of space – a double-seat is prized and often defended by placing a bag or newspaper on the adjoining seat.

The need for space means that passengers are looking at the size of the person they will be sitting next to; cleanliness is a factor too, as is body language and physical appearance. Does race come in to it? We hope not but I suspect sometimes that it does even if the result is positive when a conscious action overrules any unconscious bias.

One problem with these tests is that they do not measure our capacity for compensating behaviours.

Using the racial bias tests as a basis for some further research, Pearn Kandola compared the results from three groups of people. One group was simply asked to do the tests and reflected the kind of bias measured across the broader population. A second group was asked to be as fair as they possibly could be in their reactions to black and white faces and the bias was halved.

A third group was asked to react as quickly when they saw a black face as they did when they saw a white face and the bias all but disappeared. A self-knowledge of bias, therefore, seemed to enable people to introduce successful compensating behaviours.

This means we’re not blinded to the colour of our skins or any other apparent difference, and should not fool ourselves in to thinking we are. We cannot ignore these differences, but we can see beyond them. What is more difficult, perhaps, is to ignore assumptions that might accompany our initial observations.

The first time I went to New York, about 20 years ago, I remember how difficult it was for black people to hail a cab. Most of the cabs were driven by white taxi drivers who associated a black face with potential trouble. Today New York taxis are driven by people from a wide variety of ethic groups and black people are represented throughout the social strata. Racial bias is no longer a limiting factor in finding a cab.

Diversity is happening in the workplace. But it must happen in a way that allows people to take notice of differences and deal with them sensibly them rather than adopting some pretence that differences do not exist.

Prof Kandola is suspicious of diversity policies that try to reflect the composition, be it age, gender or ethnic background, of the customer base. “That seems twisted. It isn’t promoting a diverse workforce,” he says.

It may be that the most important point of these tests is that acceptance of existing bias is, in itself, a step in trying to eliminate the possible negative consequences of such bias. The test findings suggest that the vast majority of recruiters who declare impartiality are not only lying to the candidates, they are lying to themselves. The way forward if you seek to recruit the best people is to face up to your bias and deal with it.

*https://implicit.harvard.edu