The great myth of the 2020 US elections: split voting

For America’s Democrats, Joe Biden’s inauguration is not quite the celebration they were expecting; and it’s not just because Washington’s National Mall is closed. On election day eleven weeks ago, they hoped not just to win the presidency but to tighten their grip on Congress. The 538 website’s central prediction was that they would win 239 House seats, four more than in the mid-term elections two years ago and enjoy a majority of 43. Instead, they lost seats. Their majority almost disappeared.

This unexpected setback hardened initial impressions into conventional wisdom. It holds that many centrist voters wanted to see the back of Donald Trump but feared left-wing Democrats wielding power in Washington. To prevent this, they split their votes in unprecedented numbers, sending Biden to the White House but more Republicans than expected to Congress. That is not all. Split voting on this scale, runs the argument, sends a warning to the Democrats that Biden’s victory was more precarious than the party’s seven-million vote lead in the popular vote suggests.

Here’s the problem. The figures tell a different story. A false consensus emerged when only incomplete voting numbers were available. In the US, it takes weeks for final votes to be counted. Only now can the results be examined in detail. They show that the split-voting hypothesis, which has done much to unnerve the Democrats, is simply wrong.

Let’s start with the national figures. Biden defeated Donald Trump by 81-74 million, or by 4.5 per cent in the nationwide two-party vote. In elections to the House, the margin was 3.3 per cent.  The gap between the two was 1.2 points. So, yes, there was some vote-splitting. However, there was actually less than in 2016 (when the gap was 1.6 points) and the same as in 2012 (1.3).

When we delve deeper, we find that the decline in split-voting is even more marked.  Not all House districts are contested by both Democrats and Republicans. The number left uncontested by one or other party varies from election to election – generally between 30 and 60 of the 435 seats in the House. They are mainly in states that where one party enjoys a strong lead. The point is that these non-contests slightly distort the national totals.

In the swing states, both parties contest almost all districts. They show more accurately how split-voting evolves from one election to the next. Here are the aggregate figures for the past three elections, combining the results from Arizona, Florida, Michigan, North Carolina, Pennsylvania and Wisconsin.*

Those figures show that there was much less vote-splitting than last time or the time before. Biden’s share of the two-party vote in these six key states was just 0.5 points more than Hillary Clinton’s, while Democratic candidates for House seats, far from losing ground, increased their average share by 2.1 points.

As a result, the difference between Trump’s overall lead over Biden in these swing states (0.6 per cent) and the Republican lead over Democrats in House races (2.0 per cent) was just 1.4 points. Compared with the latest election, the impact of vote splitting was more than twice as high in 2012 (when the gap was 3.4 points) , and more than three times as high in 2016 (when it was as much as 4.6).

None of this should make the Democrats complacent. They should have done better. However, the lessons the party needs to learn are found elsewhere. This time, vote splitting was a relatively minor issue. Here are three lessons that do need attention.

  1. As in Britain and, indeed, much of the post-industrial democracies, the traditional voting base of centre-left parties – manual workers with unionised jobs in large factories and mines – has crumbled in recent decades. A second source of votes in the US and UK – ethnic minorities – is no longer as solidly loyal to centre-left parties as it used to be.

These trends have been partially offset by the huge expansion in higher education: graduates tend to have more liberal social values than non-graduates, and veer away from more conservative parties. The result has been a growing tension between the conservative values of many traditional blue-collar voters and the liberal outlook of white-collar graduates. So far, neither America’s Democrats nor Britain’s Labour Party have successfully reconciled this tension. Across the Channel, it has come close to wrecking a number of centre-left parties.

2. A universal truth for western democracies is that, in government, performance matters more than ideology. In the US, “liberal” policies that succeed, such as Obamacare, end up far more popular than “conservative” policies that fail, such as Trump’s stance on Covid-19. As a pollster, I would urge Biden to be wary of survey data when contemplating any idea advanced by left-wing Democrats. What matters is not what voters think today, but what but will work tomorrow.

3. America’s political geography is tilted towards the Republicans. They have done well enough in state elections to control the districting process in most of the country. This gives them an inbuilt advantage. In 2016, when they enjoyed a one per cent lead in the aggregate vote for Congress, they secured a 47 seat majority. Two months ago, the Democrats’ lead in the popular vote was three times as much, but with one seat still to be decided, the Democrats’ majority is likely to be only nine. An important reform, which would serve the interests of both democracy and the Democrats, would be to hand the task of deciding boundaries to an independent commission.

As is well-known, the Republicans also have an advantage in the electoral college. But the reason is different. Biden and Trump each won two of the four biggest states. But whereas Biden won California and New York by a combined majority of seven million, Trump won Texas and Florida by just one million in total. Biden (like Clinton) piled up millions of surplus votes in two safe states; Trump did not.

Beyond those four states, the two-party presidential vote this time divided Biden 50.4 per cent, Trump 49.6 per cent; and Biden won 222 electoral college votes, compared with Trump’s 165. No Republican bias there: it’s all to do with the four biggest states.

Here time might solve the problem. Texas’s growing population has been steadily eroding the Republican majority. There is a real chance that it will switch to the Democrats before too long, as other once-solid Republican states have done, such as Arizona, Colorado, Georgia and New Mexico. If the Democrats can gain Texas and also hold the line in Michigan, Pennsylvania and Wisconsin, they can fight future presidential elections with confidence.

Those are, though, very big “ifs”.

* Sources:  Official statistics compiled by the clerk of the House of Representatives, for 2012 and 2016; Cook Report for 2020. Georgia has been omitted; although Biden gained the state in November and both parties contested every district this time, they fielded only partial slates in 2012 and 2016, before Georgia was regarded as a swing state. However, from the data that are available, including Senate election  results, Georgia conforms to the pattern of the other states in the table.

This blog was first published by The Article website