On a wide range of issues, such as taxes on the wealthy, infrastructure spending, gun control, and corruption in politics, the will of majorities is routinely misrepresented by elected politicians. How is this possible? How can a democratically elected politician act against what he knows a majority or even supermajority of his constituents want?

One answer is that lawmakers vote according to the preferences of donors and lobbyists rather than according to the preferences of the public at large. Money talks, lawmakers listen. There is, alas, plenty of evidence for that theory. But another reason lawmakers neglect their constituents is that so many of their constituents don’t vote.

Data on elections for the seats of the U.S. Senate show the alarming consequences of low voter participation. For all the races resulting in the current Senate, I collected election results reported in the New York Times and eligible voter-population estimates from the United States Election Project. (A state's eligible voters are all its voting-age citizens not disenfranchised by state law due to felon status.) I grouped eligible voters into three categories: those who voted for the winner, those who voted for a rival candidate, and those who did not vote. The results shocked me: not one sitting senator was elected by a majority of eligible voters.

Senators Cory Booker of New Jersey and John Cornyn of Texas, for example, both won their seats with votes from less than 17 percent of their states’ eligible voters, which means five out of six voters did not vote for them. More than seventy senators won with the votes of just a third of their states’ voters. Only three senators won with more than 40 percent of voters. John Hoeven of North Dakota was closest to earning a majority of voters when he won with 47 percent. Still, ninety-seven senators won even though a supermajority of voters did not vote for them. Both Republicans and Democrats won their Senate races with small fractions of eligible voters. On average, senators from both parties needed just 29 percent of voters to win an election.

When candidates need the votes of only a minority of voters to win, they need never learn what a majority of their constituents want. Nor can they be held to account for acting against the will of the majority.

Not that senators refrain from using election results to claim democratic legitimacy. In a recently recorded conversation with a group of her constituents, California Sen. Dianne Feinstein used her recent reelection to justify her apparent unresponsiveness to constituent concerns: “I was elected by almost a million-vote plurality. I know what I’m doing. So, you know, maybe people should listen a little bit.” Here Feinstein acknowledges that she won only a plurality of votes, but at the same time claims that her election tally is strong evidence that she has listened to her constituents enough, and that now her constituents need to listen to her. Feinstein was elected with votes from just one of every four California voters.

Denial of the problem is bipartisan. Sitting senators are empowered by the current system, and have incentives not to change it. Mitch McConnell, who currently controls whether bills are voted on in the Senate, argued that proposals to make our democracy more representative are unnecessary: “People are flooding to the polls,” he claimed. “What is the problem we’re trying to solve here?” Turnout in the polls he’s referring to, the 2018 general election, was just 50 percent. Three times as many Kentucky voters didn’t vote for McConnell as did.

The problem is not confined to the Senate. Of the 416 contested House races in 2018, only three were won by voter majority. And no U.S. president has ever been elected by voter majority. Donald Trump won with the votes of just 27.7 percent of eligible voters, which was worse than Barack Obama’s first-term 32.6 percent, and better than the first-term elections of George W. Bush, Bill Clinton, Ronald Reagan, and Richard Nixon.

Increased voter participation does not necessarily benefit one party over the other. In the data I looked at, Democrats don’t win higher-turnout elections more often than Republicans. The sitting senators who received the lowest and highest percentages of their constituents’ votes are both Republicans: Kennedy and Hoeven, respectively. Of the incoming class of senators, the senators who received the lowest and highest percentages are both Democrats: Manchin and Klobuchar. But the effect on a particular party or politician is, or should be, irrelevant. Increasing voter turnout will increase the legitimacy, accountability, and responsiveness of all lawmakers.

Both parties have proposed legislation to affect voter turnout. Republicans are trying to reduce the participation of voters they expect to vote for Democrats. Democrats have proposed measures to increase voter participation among all eligible voters. Ongoing Republican efforts to tactically shrink voter turnout among Democrats have been effective. In fact they are, in words McConnell has used to describe Democratic election-reform proposals, “a naked attempt to change the rules of American politics to benefit one party.” By contrast, in H.R. 1, their first bill since taking control of the House, Democrats propose several policies that will help increase voter turnout and insure that the vote of every eligible voter is securely cast and counted. No Republican in the House voted for H.R. 1; McConnell is blocking a vote on it in the Senate. The provisions of H.R. 1 include automatic, online, and same-day voter registration; an Election Day holiday; early voting and voting by mail; independent redistricting commissions; and campaign-finance reform.

These reforms would all help correct our democratic deficit, and many states passed just such policies in the 2018 midterms. And yet even the complete implementation of H.R. 1 would not be enough to really solve the problem. Same-day registration is estimated to increase turnout by only around 5 percent. Automatic voter registration increased turnout in Oregon by not more than 4 percent. In advance of the 2018 elections, Vermont used automatic voter registration to register over 92 percent of its eligible voters, but still had turnout of only 56 percent. Voter registration is clearly not the same as voter turnout. Even if all states implemented all the proposals of H.R. 1—and they certainly should—the forecasted increases in turnout would not be enough to ensure that lawmakers are elected by a majority of voters.

If H.R. 1 doesn’t get us there, what would? The simplest and most effective policy would be to make voting compulsory. Why not mandate that Americans take their civic responsibility at least as seriously as they take buckling their seatbelts? Coupled with some sort of runoff election system, compulsory voting would result in voter participation rates high enough to guarantee that all winning candidates would win because a majority of voters voted for them. The many liberal democracies in the world that have made voting compulsory, making Election Day a national holiday and levying fines or tax penalties on voters who abstain, routinely achieve voter participation rates above 85 percent. Australia’s voter turnout averaged 65 percent before instituting compulsory voting, and now averages 95 percent.

But if compulsory voting is too bold a goal for now, we should at the very least pursue the many other effective policy tools proposed in H.R. 1 and elsewhere. We should arrange incentives so that everyone eligible to vote does so—which will in turn strengthen incentives for lawmakers to serve their constituents rather than just their donors.

Americans can fail to get the representative democracy we deserve in two ways: either lawmakers can be elected by a minority, or they can act on behalf of a minority. Right now both of these things are happening. But solving the first problem is easier, and may help us to solve the second.