Winning where it counts: Trump managed to find votes in all the right places

After losing the 2012 presidential election, the Republican Party conducted a so-called post-election “autopsy”, known as the Growth and Opportunity Project.

The GOP determined that a key cause of the death of Romney’s 2012 campaign was a lack of appeal to anyone who wasn’t an older white male.

The five would-be coroners wrote:

The Republican Party must focus its efforts to earn new supporters and voters in the following demographic communities: Hispanic, Asian and Pacific Islanders, African Americans, Indian Americans, Native Americans, women, and youth.… Unless the RNC gets serious about tackling this problem, we will lose future elections.

The proportion of the white vote in the US is shrinking at about two percentage points each election, down from 88 per cent in 1980 to 70 per cent this year. The GOP report predicted it would fall to 47 per cent by 2050, with Hispanics growing to 29 per cent of the electorate.

With the white vote in decline, the report found, the Republican Party would need to reposition itself or else find itself forever after on the wrong side of the nation’s shifting demographics.

Then along came Donald Trump. Trump not only appeared to ignore this advice; he seemed actively determined to mock it.

He called Mexican migrants criminals and rapists, accusing an American-born judge of bias purely on the basis of his Mexican heritage. He demanded the suspension of Muslims entering the United States, suggesting they were more likely to present a national security threat. He routinely belittled and objectified women.

And yet, compared to the urbane Romney, Trump grew the Republican vote among every ethnic group, including – most stunningly – Hispanic voters. In spite of everything, 29 per cent of Hispanics voted for Trump.

The only major demographic group among whom Trump failed to improve on Romney’s effort was women. Even there, Trump essentially did no worse than Romney. At the same time, he did far better than Romney among men.

Women split 54 to 42 in favour of Hillary Clinton at this election. This was evenly offset by the male vote, which Trump won 53 to 41.  That represents a whopping 24 point differential between the genders.

The gap between the genders, many have suggested, says something about the reluctance of men to elect a female president. That said, since the results between the sexes were a mirror image, one could argue (somewhat cheekily) that it is just as likely to say something about women’s bias in favour of a female candidate.

Exit polls did not ask whether people’s votes were influenced by Clinton’s gender, though they did ask if Trump’s treatment of women bothered them.

Although Trump managed to improve on Romney’s performance among minorities, the core of his support was still, unsurprisingly, the white vote. But it was where he won the white vote that won him the election.

While Trump broadly replicated Romney’s 20-point margin among white voters nationally, he won over a much bigger share of white voters in key states like Wisconsin and Michigan. In both these states, he improved a full 10 points on Romney’s margin among whites.
It was here, in the old industrial “rust belt”, that Trump smashed holes in Clinton’s supposed “blue wall”. These former manufacturing states of the North-East and Midwest include Pennsylvania, Michigan and Wisconsin – all must-win states for Clinton – and the perennial bellwether state, Ohio.

All were suspicious – or even outright hostile – towards free-trade agreements, with exit polls showing that they overwhelmingly believe that international trade costs Americans their jobs.

Prior to the election, Wisconsin had been pencilled in as a certain win for Clinton. The election eve RealClearPolitics polling average in the state had her ahead by 6.5 points and, until this week, Wisconsin had not gone red since 1984. Indeed, Clinton was so sure of holding the state that she did not visit it once after securing her party’s nomination.

This will, no doubt, be a focus of the Democratic Party’s post-election “autopsy”.

It will also be the source of a lot of head-scratching among US pollsters. They can join those in the United Kingdom, who failed to predict either a Conservative majority at the last British election or, dramatically, the subsequent Brexit vote.

It is true that US polls are generally taken at a national level and reflect the “popular vote”, which Clinton may indeed end up winning by a whisker. But the vast majority of polls had Clinton ahead by several percentage points, a sufficient margin, it was assumed, to assure her of victory in the electoral college.

One theory of where the US pollsters got it wrong revolves around a reverse Bradley effect of sorts. In 1982, the African-American mayor of Los Angeles, Tom Bradley, comfortably led the race for California Governor according to most polls, yet ended up losing. Commentators suggested that some voters had felt a social pressure to tell pollsters that they planned to vote for the more politically fashionable candidate but, in the privacy of the polling booth, had voted for the other candidate.

It is certainly plausible that a small percentage of voters felt reluctant to admit to pollsters that they intended to vote for Trump.

As Trump’s incremental gains among minorities and among white voters in key states proved, small percentages in the right places can turn an election.