Diversity training doesn’t change people’s behaviour. We need to find out what does

The US government is training every one of its 2.8 million employees to avoid unconscious bias. Starbucks is planning to close more than 8,000 stores for the day to give its staff similar courses. In the UK, the government has already trained 100,000 civil servants, and recommends that every major employer do the same.

The only problem? It won’t work.

Unconscious bias training is often given to managers in charge of hiring and promotion. It’s meant to reveal to them their hidden prejudices and biases, usually towards women or ethnic minorities, and educate them about the impact that bias has on decision-making.

But study after study has shown that while training may raise awareness of prejudice, it doesn’t change people’s behaviour at all. (You can read more about this on Apolitical’s Gender Equality newsfeed. Some of the people working on this are on our global list of the gender equality top 100).

The reason for this is neurological, according to the CEO of hiring platform Applied, Kate Glazebrook: "it’s hard to retrain the brain not to fall prey to the prejudices you have reinforced and trained for your entire life".

Some studies have even shown that unconscious bias training can backfire. If managers are forced to do it, they can feel accused and become defensive. They can end up reporting more animosity toward minority groups. It can also lead to defeatism: if biases are so pervasive and hard to change, why bother trying?

Unconscious bias training is money ill-spent. But it is just one particularly striking example of a wider issue. There is a lack of evidence about what actually does work to improve gender equality and diversity at work.

Companies spend their diversity budgets on a range of initiatives that promise change, from women’s leadership development programmes to new assessment systems. Many of these investments are done in the dark, and their impact is rarely assessed. They are chosen because they’re easy, fast and popular, not because we know they work.

"What organisations don’t do across lots of training is...systematically measure before and afterwards to see what difference has been made", says Mike Noon, author of Pointless Diversity Training: Unconscious Bias, New Racism and Agency.

"They want to do something quickly without spending weeks on it, so they go for a two or three-hour unconscious bias session."

In the UK, all large employers were forced for the first time in 2018 to publish their gender pay gaps on an online portal visible to the public. The new transparency law has made the impetus to improve workplace diversity greater than ever. It has provoked debate throughout the country. Heads have rolled, salaries have been publicly axed and many have been named and shamed. At Ryanair, female employees earn just 33p for every £1 that men earn. At HSBC, a bank with more than 20,000 employees, the mean woman worker earns just 41p for every £1 that the mean man earns.

Many employers have already published detailed action plans about how to close their gaps and promote more women. Others are working on them. But bigger diversity budgets risk being wasted on ineffective policies. Marks & Spencer plans to "[weave] unconscious bias and inclusive leadership through learning and development at all levels".

Many employers say they will close their gap by mandating diversity on hiring shortlists. But the evidence says that only works under certain conditions. If a shortlist contains just one woman, her chance of being hired is statistically zero, a 2016 study in Harvard Business Review found. If a second minority is added to the pool, the chance of either being hired skyrockets.

The UK government has now realised that pressure without evidence-based solutions won’t help. It has engaged its world-famous 'nudge unit', or behavioural insights team, to find out what actually does work to close the gender pay gap. Over the next 18 months, the unit will run experimental field trials with five large UK employers, including the insurer Zurich and a big city bank.

In the shorter term, the unit is publishing a list of dos and don’ts, based on the patchy evidence that exists so far. Do have skills-based assessments and structured interviews for recruitment and promotion. Do have multiple women on shortlists. Do introduce transparency into pay and promotion processes. Do encourage women to negotiate their salaries by clearly indicating upper and lower limits of the salary range.

But don’t offer diversity training, unconscious bias training, or leadership training for female employees, because there’s no evidence that they work. Don’t bother with diverse selection panels. Sometimes they may help female candidates, but sometimes they harm them.

Spreading these evidence-based recommendations is critical. A new wave of money and effort could now be wasted on actions that don’t lead to change. If that happens, momentum may be lost, and employers may just give up trying.