7 November 2018

Americans are fired up – and bitterly divided


Sometimes an election result doesn’t deliver a clear, unambiguous, narrative. Yesterday’s US midterms fit that category, with the overall picture being very close to the midpoint of what had been widely expected beforehand.

In absolute terms, these were good results for the Democrats, as is often the case for an opposition party in midterms, taking control of the House of Representatives. But is that good enough?

The current estimate of the final House popular vote tally from the New York Times is a seven-point Democrat win. That represents a shift (eight points net, or, in UK terms, a four per cent swing) of about the size the Republicans got during the first Obama term in 2010.

In terms of seats, it’s less impressive; the Democrats need a five-point lead in the popular vote for that to translate into a majority of seats. That’s partly thanks to gerrymandering, but also because they, much like Labour in the UK, tend to pile up votes in safe seats.

And while Democrats were never likely to flip the Senate given that most of the seats that were up were ones they already held, they will be disappointed to have gone in the wrong direction.

Perhaps the clearest message from these results is that Americans are newly enthusiastic, but remain bitterly divided.

Enthusiasm was clear from the turnout. The current New York Times estimate of turnout is 114 million, up more than 30 million on 2014. Sixteen per cent of voters told exit pollsters they hadn’t voted in a midterm election before, while eight per cent said they didn’t vote in 2016 (and because people tend to remember voting in the past when they didn’t, even these numbers could be underestimates).

But if this helped the Democrats, it was only at the margin – they did win decisively among new voters, but only by about enough to offset their over-representation among those that stayed home. In terms of both recalled presidential vote and party identification, the partisan makeup of 2018 midterm voters was within a point of 2016 (although the advantage the Republicans had during the Obama years was erased).

It’s also clear that divisions haven’t healed. Donald Trump was not personally on the ballot, but midterms are widely seen – including by Trump himself – as a referendum on the president, and there wasn’t much sign of widespread regret among those who voted for him two years ago. More than 90 per cent of both Trump and Clinton voters backed the same party’s candidate for Congress.

Liberals cheered an ever more diverse set of winners, including record numbers of new Congresswomen. But if 2018 was the year of the woman in terms of candidates, when it came to voters, gender didn’t explain much of the vote switching. Though the gender gap was huge, it was only about as huge as two years ago, with Republicans faring a net 23 points worse among women than men.

White graduate women, of whom much was made pre-election, swung left, but in about the same proportions as white graduate men. And white women as a whole were evenly divided, and once the exit polls are reweighted to the final result, they may narrowly have voted Republican once again.

What did seem to matter, and where polarisation increased further, were age and education. Millennials swung to the left by about three times as much as over-65s, and graduates by about three times as non-graduates. Age polarisation probably hasn’t quite reached the levels seen in Britain, but it’s not far off.

One under-fire group that had a good night were the pollsters. Elections where turnout jumps from a low starting point (at the 2014 midterms it was a derisory 37 per cent) are some of the hardest to poll, but the eventual margin is likely to land within a point or so of the final polling averages.

State and local polling and modelling appear, at first glance, to have been pretty good too. The most notable surprise, if it can be called that, was in Florida, where polls had put both eventual Republican victors a few points behind.

What does this tell us about 2020? Well, the next Presidential election is very much in play. While the data are still coming in, the results don’t seem to give a clear indication of which way the next presidential election is likely to go.

 In fact, these midterms may raise more questions than answers. The results suggest that Trump’s base cannot be counted on to desert him, much as Democrats would like that. They will have to go out and beat him in 2020 ­– so the question for them is how to do so.

While the big decision for the challengers is who their candidate should be, the more fundamental (and related) choice is what their strategy should be: target further gains among younger, more liberal graduates, or have the difficult conversations about their losses among the white working classes. As ever, the next election cycle is underway already.

Matt Singh is an election and opinion polling analyst and founder of the website Number Cruncher Politics.