How 'Desirability Bias' Weakens Democracy

If you just believe what you want to believe, it's a problem.

A failure to communicate.

Photographer: Jimmy Anderson/Getty Images

When people get new information – about immigration, about President Donald Trump, about climate change – will they change their minds?

It’s common to say that if they don’t, the reason is “confirmation bias," which means that people are far more likely to accept information that supports their current beliefs. But in some situations, what really matters is something different and even more insidious: “desirability bias.”

When people display desirability bias, they find information more credible when it pleases them – even if it fails to confirm their pre-existing beliefs.

An ingenious study by the University of London’s Ben Tappin, Leslie van der Leer and Ryan McKay shows how desirability bias works and how, in important contexts, it can prevail over confirmation bias.

Surveying 900 Americans in late 2016, Tappin and his colleagues took advantage of the fact that some supporters of Hillary Clinton believed that Donald Trump would win, while some supporters of Donald Trump believed that Hillary Clinton would win. They asked people to disclose which candidate they preferred, and also to predict the likely winner.

After gathering that information, the experimenters randomly sorted participants into two different groups. The first group was presented with nationwide polling results suggesting that Trump would win. The second was presented with results suggesting that Clinton would win. All participants were then asked to update their beliefs about who would win.

The main result was clear: Both Trump and Clinton voters were far more likely to change their beliefs when the new information fit with their desires. They found the polls credible, and a reason to change their views, only if they suggested that their preferred candidate would win.

Participants even showed “disconfirmation bias.” When the polls favored their preferred candidate, they were more likely to revise their beliefs on the basis of those polls, even if they originally expected their candidate to lose.

This finding fits well with growing evidence that in multiple domains, people are more likely to believe good news than bad news. If people tell you that you are smarter or better-looking than you think, you might well think, “They’re probably right!” – but if you hear that you are not quite as brilliant or as beautiful as you thought, you might respond, “What do they know?”

With Tali Sharot of University College London and other collaborators, I have found something similar for political issues, such as climate change. People who are not greatly concerned about that problem, and think that the planet is not going to get much hotter, show a form of “asymmetrical updating”: When they get information suggesting that there’s going to be less warming than they expected, they become even less concerned. But when they get information suggesting that there’s going to be a lot of warming, they don’t become more concerned.

For democratic self-government, desirability bias is a major problem. For one thing, it makes learning a lot less likely. For another, some people deplore news that other people love -- which means that whenever citizens receive conflicting information, they can become highly polarized, even if they’re not living in echo chambers.

For anyone who hopes to persuade people, there is a clear lesson: Try to find a way to get people to want to agree with you, or at least not to want to disagree with you.

If you seek to persuade people that a problem is serious, it’s wise to try to convince them that the solution isn’t too costly or burdensome -- and that they might even like it. In politics, Bill Clinton was a master of that strategy – emphasizing, time and again, that we can “both protect the environment and grow the economy.”

And if you want to get people to ignore or dismiss a problem, it’s smart to emphasize that any solution would turn out to be unbearably painful. Donald Trump is skilled at that – arguing, for example, that gun control would not only take weapons away from law-abiding citizens but also cost lives on balance.

The good news is that on many political issues, most people don’t have an emotional stake in any particular set of facts – for example, the number of deaths on the highways, the rate of diabetes, or whether nanotechnology creates serious environmental risks. But when emotions are running high, people’s receptivity to new information often depends on whether it pleases them.

For those who occupy public office, or seek to do so, that’s a stark warning – and also an invitation to think creatively about how to make their own proposals seem more desirable to those inclined to oppose them.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

    To contact the author of this story:
    Cass R Sunstein at csunstein1@bloomberg.net

    To contact the editor responsible for this story:
    Katy Roberts at kroberts29@bloomberg.net

    Before it's here, it's on the Bloomberg Terminal.