Jason Brennan is the Flanagan Family Chair of Strategy, Economics, Ethics, and Public Policy at Georgetown University.

Residents of Prince George’s County cast their votes at Eleanor Roosevelt High School in Greenbelt. (Sarah L. Voisin/The Washington Post)

Each week, In Theory takes on a big idea in the news and explores it from a range of perspectives. This week we’re talking about voter rationality. Need a primer? Catch up here.

Jason Brennan is Flanagan family chair of economics, ethics and public policy at Georgetown University. His most recent book is “Against Democracy.”

Many are scratching their heads: How can someone like Donald Trump win the presidency? To win elections, politicians must push policies that appeal to voters. But most voters are systematically misinformed about the basic facts relevant to elections, and many advocate policies they would reject if they were better informed. We get low-quality government because voters have little idea what they’re doing.

Worst of all? There’s little we can do about it. The problem is built into the democratic system. More education won’t help. Voters are more educated now than in the past, but they are as ignorant now as they were 65 years ago. We could require voters to take an exam to earn the right to vote, but that would be unconstitutional, and few people support such an undemocratic policy. Perhaps, at best, we could change the culture surrounding voting. We could push the idea that citizens should not just vote, but also be informed voters.

Sixty-five years ago, researchers began studying what voters know and how they think. The results are depressing. The median voter knows who the president is, but not much else. Voters don’t know which party controls Congress, who their representatives are, what new laws were passed, what the unemployment rate is or what’s happening to the economy. In the 2000 U.S. presidential election, while slightly more than half of voters knew that Al Gore was more liberal than George W. Bush, they did not seem to know what the word “liberal” means. Significantly less than half knew that Gore was more supportive of abortion rights, was more supportive of welfare-state programs, favored a higher degree of aid to blacks or was more supportive of environmental regulation.

High-information voters have systematically different policy preferences from low-information voters, even controlling for whatever impacts our demographic differences have on our ideologies. For instance, as political scientist Martin Gilens found in one study, high-information Democrats are more in favor of free trade, abortion rights, civil liberties, gay rights and less hawkish foreign policy than low-information voters. Political scientist Scott Althaus and economist Bryan Caplan get the same results using different sets of data. But the American voting public as a whole shares the preferences of low-information voters, simply because there are far more of them.

When these voters do bother to seek out information, they turn to news sources and experts who share their bias. They almost never talk to the other side, whom they regard as stupid and evil. When they read material that suggests they’re mistaken, they dig in their heels and conclude they’re still right.

It’s no mystery why most voters are like this. Imagine you are in a thousand-person college class. The professor tells the class on the first day that she plans to average all the grades together and that every student will receive the same grade. Most students would not bother to study, and the average final grade would likely be an F.

So it goes with voting. How the entire electorate votes does matter, but how individual voters votes does not. As a result, for most individual voters, the costs of acquiring political information exceed the expected benefits. It’s not that voters are stupid individually; it’s that they just don’t care. They respond rationally to the bad incentives democracy creates.

This mirrors the problem of air pollution. As an individual, each of us can pollute to our heart’s content. Buying a Prius won’t save the world and driving a gas guzzler won’t end it. But if we all act without regard to the consequences, we get dangerous climate change. As with air pollution, voting pollution is a collective action problem.

Some theorists say voters can use information shortcuts to make voting easier. They’re partly right. Sure, the two-party system makes the choice easier, because voters have to choose between only two platforms. Still, most voters don’t know much about what the two parties will do or have done — and many parties push the policies they do only because they’re trying to appeal to low-information voters. Similarly, voters can turn to information leaders and experts for advice on how to vote, but most voters just seek out “experts” who agree with them.

Voters are badly informed because they have no incentive to be informed. Perhaps governments could do something to change that. Suppose that two weeks before the election, the government administered some sort of “basic political knowledge test” and gave anyone who passed a $500 tax credit. This would incentivize voters to learn the basic facts, though perhaps the marginal benefit would not be worth the expense.

In a country made up entirely of high-information voters, Trump would have never stood a chance, and the Democrats would have probably fielded someone far better than Hillary Clinton. But voters are more like sports fans waving towels than they are like dispassionate social scientists. We can’t fix the problem. It’s built into democracy.

Read more:

Robert Kurzban and Jason Weeden: No, Trump voters were not irrational

David Greenberg: In defense of political spin