How To Free Your Decisions From Bias

It’s not easy to free yourself and others from decision bias. But the pay off for your organisation is worth it…

A CEO mentioned recently to me his frustration with a few of his Senior Leaders who play the ‘merit card’ whenever diversity is raised. In doing so, they stymie good initiatives. Each small block they construct rebuilds the wall as fast as the CEO and supportive leaders tear it down. ‘What can I do?’ he asked. I shared his pain: invoking the ‘merit card’ is a wicked, if effective, tactic for, paradoxically, subverting merit and keeping control.

The CEO and his leaders have an awareness of unconscious bias and know a bit about how it works. Until recently unconscious bias was heralded as the holy grail for achieving significant improvement in diversity and inclusion outcomes. But the value of unconscious bias training in particular, and diversity training in general, is being challenged.

Dobbin & Kalev’s influential article ‘Why diversity programs fail’ importantly identified that command and control approaches, adopted by many organisations, backfire. You can’t get people to change by telling them to. And you don’t get people to change by blaming them for doing the wrong thing.

Making training about beliefs and preferences mandatory is almost guaranteed to fail. That’s because suppressing unconscious beliefs, to ‘do what’s expected’, is  well-known to make bias more, not less, likely. And that’s the danger with these senior leaders who play the ‘merit card’; their biases may increase rather than decrease.

Unconscious bias awareness is not a silver bullet, it is however, worthwhile. It’s not easy to free yourself and others from decision bias, so what will make it worth the CEO’s effort? You can’t work with it effectively if you don’t understand it. And it’s how you work with it that counts. 

Debias by accepting your fallibility

At an individual level, part of the work is to accept your own fallibility. We are susceptible to many types of bias, that cover all sorts of decisions. Frustratingly, because these biases operate unconsciously, we can’t really know when we are in their grip. And our bias for overconfidence means that we tend to think that our decisions are much better than they are. So, we’re not actually very likely to think we’re biased. It’s bit of a Catch-22.

The most practical approach is to be aware of the tendency towards overconfidence. Be more modest, less certain, about your decisions. Whether or not you know you are biased matters less than accepting that you are likely to be biased.

Leaders who play the ‘merit card’ probably suffer certainty bias, they don’t think they are biased. They don’t like the suggestion they have a ‘weakness’ like ‘bias’. Without that openness, their decisions remain narrow. Feelings of certainty are biases themselves. It’s when we feel most certain that we are most likely to be unsystematic, think we know, circumvent objective methods, or neglect to ask for alternatives.

If you accept that you are likely to be biased you are more likely to act to mitigate against bias. And that, currently, seems to get the best results.

Biases show up in:

• What we notice
• What we expect
• What we ask, and
• What we value.

What we notice

Collectively, we are getting much better at noticing gender-participation differences by industry and occupation. When we take the time to collect and examine the data about, for example, pay, it transpires that there are often gaps that can only be attributed to gender.  When we notice the difference, we can act on the difference.

At the individual level, what we notice has a big impact on careers.

Letters of recommendation for male academics emphasise research skills, publications and career aspirations, which are the ‘get ahead’ characteristics. Whereas teaching skills, practical clinical skills and personal attributes, the ‘get along’ characteristics, are more often identified for females.

Women scientists’ early career advancement is hindered, even when they have the same qualifications as male scientists. Male and female faculty make biased hiring decisions, preferring male candidates over female. Their capabilities are noticed differently. Male candidates are seen as more competent, more worthy of mentoring and deserving of a higher salary than female candidates.

Notice what you notice

Set yourself a noticing challenge. Pair yourself up with someone of the opposite gender, with whom you will be interacting regularly throughout a designated day. Commit to taking observations during the day. Each half hour, record what you have observed in terms of interpersonal interactions.

At the end of the day, compare your notes with each other.

What do you notice about who takes what kinds of actions, and what is the impact of their actions on others? What’s similar in your observations, and what’s different?

What we expect

We expect men to be ambitious and we don’t expect women to be. This erodes women’s ability to express their ambition. In numerous professions, from policing to medicine and science, women begin with the same levels of ambition as do men. Yet, while men’s ambition increases over time, women’s decreases. Because women are constantly fighting structural barriers, their ambition often wanes.

We expect men to be competent and women supportive. A recent European study reviewed 125 applications for venture capital funding. Forty-seven percent of women’s applications, versus 62% of men’s, were funded. Women applied for and received less funding.

There were four distinct differences in the language used to assess applications:

• Women were described as needing support, men as assertive.
• Women were not described as entrepreneurs but as growing a business to escape unemployment. Superlatives were used about men’s fit with entrepreneurship and risk taking.
• Women’s credibility was questioned, men’s was not.
• Women were seen to lack competence, experience and knowledge; men to be innovative and impressive.

Expectations about how men and women should behave were carried over into evaluations which then affected their relative success.

Disrupt your expectations

What happens if you disrupt your expectations regarding ambition and competence? What if you spent a day imagining all the women you engage with are ambitious, competent and want to get ahead? Imagine the men with whom you engage want to provide support and take a back seat.

If our Senior Leaders imagined that the men in their teams wanted to leave work to pick up the kids from school and prepare dinner, how would they think about their next career move?

What we ask

The group of researchers involved in the VC funding example above observed the full application process. They concluded that the questions that were asked undermined women’s potential, but underpinned men’s.

A recent US study found a similar kind of bias. In a start-up funding competition, venture capitalists (VCs) were much more likely to ask male entrepreneurs promotion-oriented questions. They focused on ideals, achievements and advancement. By contrast, VCs asked females entrepreneurs prevention-oriented questions. These questions focused on vigilance, responsibility, risk and safety. Male-led start-ups raised five times the funding of females. Consistent with what we know about unconscious bias, the research found that male and female VCs displayed the same questioning biases.  It is often assumed that men favour men and women favour women; increasing the number of women on selection panels is routinely seen as the solution. Yet unconscious biases about gender are held as commonly by women as by men. While simply increasing the number of female decision makers does make balanced decision making more likely, it does not guarantee it. However, when panels have gender balance, or are female only, bias tends to disappear.

Question what you ask

 How might you disrupt the kinds of questions you ask men and women? Do you ask men and women the same questions? What happens when you do?

Imagine our Senior Leaders asked men and women the same questions they ask women. What would they learn?

What we value

Johnson & Johnson, which fields about 1 million job applications for over 25,000 job openings each year, now uses Textio to debias their job ads. When they first started using it they found that their job ads were skewed with masculine language. They were disproportionately valuing male characteristics. Their pilot program to change the language in their ads resulted in a 9% increase in female applicants.

Even when managers and decision-makers espouse a commitment to gender equality and a desire to promote more women into leadership positions, they are prone to evaluate women less positively

By deliberately analysing and structuring how information is conveyed and options are presented, it can become easier to make fairer decisions.

Women are commonly demoted to traditional gender roles. Forty-five percent of women in one study have been asked to make the tea in meetings. Some were CEO at the time. Female doctors are often mistaken for nurses, female lawyers for paralegals and female professionals of many kinds for personal assistants.  We do not expect women to hold senior roles, despite the fact that, increasingly, they do.

Student evaluations of teaching appear to be influenced similarly. Even in an online course where the gender of the instructor was manipulated so that identical experiences were provided to students, those students who believed they had a female teacher provided significantly lower teaching evaluations. While these lower ratings misrepresent actual competency, they nevertheless may create a self-fulfilling prophesy where women’s career advancement choices begin to conform to the stereotype. And erroneous beliefs about women’s competency levels limit the opportunities that are provided to them; the misrepresentations are perpetuated.

Put the value back into evaluation

Debias evaluation by using blind, automated processes. Take human bias and error out, and increase the value of the decisions you are making.

Would our Senior Leaders be prepared to do this? Would they be prepared to take themselves out of the equation? Would they believe an objective merit-based process could occur for a decision in which they have an interest, but in which they were not involved?

Put it all together

People are responsible for their own minds. Our CEO has provided opportunities for his senior leaders to engage with curiosity, respect and candour in their diversity programs. There are some wonderful stories emerging.

The challenge for those who don’t yet get it, is to agree to the overarching purpose that people decisions are based on merit. If merit is what we are aiming for, we should all be prepared to sign-up for practices and tools that increase and uphold it. Will they do this?

But merit is both more and less than it seems. It is more complex and difficult to define than most people think. It is less objective and rigorous, particularly in knowledge work and leadership roles. It is ripe for bias. Paradoxically, invoking merit is perhaps the most powerful way to subvert it.

It’s time for Senior Leaders to throw away the ‘merit card’; their people deserve a fairer hand.

Eroding merit corrodes culture, and culture is where the CEO leaves his biggest legacy. What can he do? To leave a lasting legacy, the CEO knows he needs to call out the fallacy of the ‘merit card’ and hold his Senior Leaders to account for fair people decisions. He can help them to exit the organisation if they are not prepared to play a fair hand.

If the Senior Leaders are prepared to admit to fallibility, to be aware that they may notice and value the behaviours of different groups of people in different ways, there are many practices that will make sure bias is minimised and fairer decisions are made.

We can all keep working to debias our decisions.

What to do if you believe in merit:

1. Accept your fallibility – be more modest, less certain about your decisions.
2. Notice what you notice – record what you notice and assess it for fairness.
3. Disrupt your expectations – imagine women are ambitious and men supportive.
4. Question what you ask – ask the same questions of everyone.
5. Put the value back into evaluation – by using blind processes.