Guest Post: The enthusiasm gap

A guest post by Stephen Russell:

In my previous posts I described how the polls were wrong in 2016;  what can be said about the accuracy of polls in 2020; and whether polls are being distorted by a “shy Trumper” effect. 

There is of course, a lot more in the polls than just the headline numbers of how many are voting for who. There are other factors which may provide clues to deeper currents or to hidden truths. Two often-cited factors here are the related issues of relative enthusiasm and how many of the polls’ respondents will actually vote.

For example, a national survey, conducted in early June by SSRS for CNN, found that 7 in 10 Trump voters are voting to re-elect him primarily because they support him. But only 37% of Biden voters plan to cast a pro-Biden vote. Sixty percent are simply anti-Trump.

YouGov has similar results. They report 68% of Trump voters being “enthusiastic” about Trump. Only 31% say the same about Biden. That might result in more Trump voters being motivated to actually get out and vote than Biden voters.

The candidate who has led on enthusiasm (or a closely related question) has won every presidential election since 1988 (when Michael Dukakis had the more enthusiastic supporters, but lost to George H Bush).

But seizing on that as a predictive factor may be confusing cause and effect. Ultimately, a candidate is elected because he or she is more popular than their opponent. The very things which make them popular also make their supporters enthusiastic. And popularity creates enthusiasm: people like winners and will be less enthusiastic about their party’s candidate if it looks like they are going to lose.

What we see in 2020 is an unusual situation, where popularity and enthusiasm are out of sync.  But an enthusiasm gap is normal when an incumbent president is being challenged. Trump has had four years to unite his party behind him and fire up his base. Biden has barely had four months.  As the divisions of the primaries fade into the past and the general election looms, there is every chance that the “enthusiasm gap” will narrow.

It is also not clear that the huge enthusiasm among Biden’s supporters to eject Trump from his office is any less powerful than enthusiasm for Biden himself. There has probably never been an election in which so many voters were so powerfully fired up against one of the candidates.

Michael Tesler, at Fivethirtyeight, says the enthusiasm gap is mostly a myth. That’s because “while Biden voters may not be all that excited about voting for Biden, they’re very enthusiastic about voting against Trump. And that gives Biden a pretty strong edge, because Trump supporters don’t despise Biden the way they despised Hillary Clinton in 2016.” In fact, according to his research, when you combine the favourable ratings and the unfavourable for each candidate to get net ratings, it is Biden who has a huge advantage.

He goes on: “What’s especially notable here is that Biden’s net enthusiasm rating is near zero, which is similar to most major-party presidential candidates’ ratings from 1980 to 2012. Trump’s current score of around -20, on the other hand, has only one historical comparison other than his own campaign four years ago: Hillary Clinton in 2016.”

Research by Abramowitz and McCoy suggests (sadly) that hatred is a more powerful political motivator than liking.

To some extent, tepid enthusiasm can also be the flipside of broad support. Biden has the latter, because he does not polarise people as much as Trump, and doesn’t repel them. Bernie Sanders’ supports were very enthusiastic, but he still got trounced, because Biden was more broadly acceptable. Political columnist Ed Kilgore observes: “Any ‘enthusiasm’ beyond that which is necessary to get a voter to the polls is pretty much wasted … you don’t get extra votes for being psyched out of your skull about your candidate.”

The question of enthusiasm feeds directly into the issue of the general adult population vs registered voters vs likely voters. Historically, those who support Republicans have been more reliable in getting registered and then actually voting than those who support Democrats.

To capture that effect, a good pollster will want to focus on likely voters, since the support of those who don’t vote doesn’t count.  But even the potential voters themselves cannot reliably say how likely they are to vote when an election is far away. So while LV polls are best, they only develop that value as the polling day nears.

Fivethirtyeight elections analyst Nathaniel Rakich observes: “Likely-voter polls tend to be a few points better for Republicans than registered-voter polls, and … we don’t have a ton of these polls right now.”

Two of the factors underlying that trend are the higher voting rates of Americans who are older, and those who are wealthier. Historically, these have been Republican-leaning groups. But Trump has changed that: polls have shown that old voting allegiances are inverting. Low-education blue-collar workers have flocked to Trump. But wealthier white-collar suburbanites are flocking the other way. For the first time in twenty years the Democrat is beating the Republican among older voters, while youth support has weakened. That might make Biden’s present support less vulnerable to the shift in polling towards those most likely to vote.

The third big factor in voting likelihood is colour – and Trump’s support remains overwhelmingly white.  But recent events, and Trump’s reaction to them, could well reduce the turnout gap. It would not be at all surprising if Black Americans were to feel extra-motivated this year.

Curiously, however, CNN analyst Harry Enten reports  that recent events seem to have made no difference to the level of Black support for Trump. While that is small – Biden leads here 83-8, that is slightly smaller than Clinton’s lead in 2016 polls, and was unchanged from the position three months before.

This brings us to another point: turnout. In 2016 this was 55.7% – very low by New Zealand standards, but not by US ones. It was the second-highest turnout in 50 years.

In November 2018, the BBC reported: “Preliminary figures for nationwide turnout in the 2018 mid-term elections are in, and they’ve reached a mark not seen in more than a century. Across the US, 49.2% of the voting age public cast ballots. In 2014 that number was 37%, and the average over the last few decades has hovered around 40%.” Final figures put the turnout at 50.3%.

The Atlantic reports that “…strategists in both parties and academic experts are now bracing for what Michael McDonald, a University of Florida political scientist who specializes in voting behavior, recently called a voter turnout storm of a century in 2020.’.”

That suggests (probably thanks to Trump himself) that the chances of people with an opinion simply staying home on November 3 might be a little less than it has been in the past. 

In summary, there are plenty of reasons to imagine that current polling might be underestimating support for Trump. But none of those reasons are certainties. The track record of polling in the US is actually very good: even in 2016 the polls were far closer to the results than they are given credit for and there has also been no consistent anti-Republican bias.

Polls in 2020 might be wrong. Indeed, mathematics alone predicts that they will be! But they are unlikely to be dramatically wrong. That means that Trump will almost certainly need to close the present gap in order to get re-elected.  The final post in this series will examine Trump’s potential advantage in the Electoral College – which could effectively reduce that gap.