It is hard to imagine a presidential election with higher stakes than the present one: Hillary Clinton and Donald Trump offer radically different ideas about the country’s direction, and many people believe that one or the other would be catastrophic. Yet new evidence raises the possibility that we will see an unusually low voter turnout.

To explain that puzzle, here’s a quiz. Since 1972, when do you think the United States had the biggest turnout in a presidential election, measured by the percentage of the voting-age population that actually voted?

The answer is 2008, when 58.2 per cent of potential voters cast ballots. A big reason was Barack Obama, an electrifying candidate who turned “hope and change” into a national mantra. And of course John McCain had his own strong core of enthusiastic supporters, ensuring a fierce intergenerational battle that appealed to people’s competing ideals and aspirations.

By contrast, the lowest turnout in recent years was in 1996, when President Bill Clinton, chastened and tarnished by struggles with the Republican Congress, defeated Bob Dole, who proved a hopelessly ineffectual candidate. Only 49 per cent of potential voters cast ballots.

Turn now to the present, when only 72 per cent of registered voters say they are highly interested in the presidential elections — a 15 per cent reduction from 2008 (and 4 per cent lower than in 2012).

The number of highly interested voters matters because it is closely correlated with the ultimate turnout. Those numbers suggest a drop this year, despite the high stakes.

The biggest decline is among younger voters and blacks. In 2008, a whopping 83 per cent of voters under 35 said they were highly interested, compared with 54 per cent today. That same share of blacks, 83 per cent, said in 2008 that they were highly interested, compared with 65 per cent this time.

What’s going on? One lesson is that if you want to predict voter turnout, you should ask whether at least one candidate is attracting high levels of enthusiasm — not whether the stakes are high or even perceived to be high.

That fits the historical pattern. The charismatic John F. Kennedy was a forerunner to Obama, helping to produce an unusually high turnout in 1960 (62.8 per cent). By contrast, the 1988 election, in which the unexciting George H.W. Bush opposed the even less exciting Michael Dukakis, produced one of history’s lowest participation rates — just 50.2 per cent.

Many Republicans lack enthusiasm for Trump, and they are not going to turn out for him. A much higher percentage of Republicans reported being keenly interested in the 2012 campaign than in the present one. Among Democrats, the level of high interest is about the same as what it was in 2012 — an excellent sign for Hillary Clinton — but that is still far lower than in 2008.

Although many Democrats think that Trump would be a terrible president, hope was apparently a much stronger spur in 2008 than fear is today.

If the 2016 election produces an unusually low turnout, should we be alarmed?

To the extent that prospective voters are intimidated — for example, by the prospect of “monitoring” by Trump supporters — self-government is compromised. And if people do not vote because they lack enthusiasm for the candidates, the process is not working as it should.

From the standpoint of democratic legitimacy, it is a problem if half the electorate, or close to it, declines to vote, not least because they may not feel much of a stake in the whole process.

It is true that the existing data must be viewed with care. The election is still almost three weeks away and voter interest could jump in that time, producing unexpectedly high turnout levels.

Moreover, the correlation between interest and turnout may be weaker this time around. People who remain unexcited about the election or their own candidate may ultimately appreciate the stakes, and so end up voting.

But don’t bet on it. At least at the moment, the most reliable guess is that we will see a solid, even emphatic victory for Clinton — from a little more than half the voting-age population.

• Cass Sunstein, a Bloomberg View columnist, is director of the Harvard Law School’s programme on behavioural economics and public policy