Ben Lowry: Why I think Donald Trump is marginal favourite to win the United States presidency

I have been in the US at some point during every presidential election year of my life except 1992 and 2020, albeit not always on polling day. This visit is the first time I have not been in the country for more than five years, due to the pandemic.

In the 1970s I was far too young to notice elections, so the first contest I remember is 1984 when the Republican Party President Ronald Reagan beat the Democratic Party challenger Walter Mondale in a landslide. The first race I followed closely was 1988: that heatwave summer, age 16, I stayed with cousins in Boston, working in McDonalds. I was pro the radical liberal candidate Jesse Jackson (my politics later moved rightward!) and went to the emerging Democratic candidate Mike Dukakis’s house (then governor of Massachusetts) the day after he had his fellow Democrat, but rival for the nomination, Rev Jackson to a dinner that included clam chowder, a dish – it transpired – Jackson didn’t like.

This year’s election is tight. Yesterday I spoke to two foreign affair experts, one of whom is confident that the Democrat Kamala Harris will win, the other says the betting markets have it right, and that the Republican ex President Donald Trump will be easy victor.

My hunch is that Donald Trump is marginal favourite. I have an informal way of interpreting UK and US opinion polls, based on the most recent 20 such polls. I total up the percentage vote that the main candidates are being forecast to get, and then assume that support for conservative or hardline candidates has been under-estimated. I then assess that average polling data against scenarios in which the under-estimate is one of 1%, then 2%, then 3%, etc.

This method does not work in NI elections, where there is too little polling data to make meaningful deductions from surveys. Having said that, the limited polling prior to this last general election did under-estimate both the total unionist vote, and the Sinn Fein vote. Support for Alliance was again over-estimated.

In the UK, polling data generally under-estimates the Tories, and did so again ahead of this year’s general election. In the end the Conservatives had a very bad night, but it wasn’t as disastrous as the forecasts which had them down to a few dozen MPs. They got almost 3% more votes than polls forecast.

But while the presence of a ‘shy vote’ (people who don’t admit to pollsters that they support candidates seen as unfashionable or hardline) is a well-known one when it comes to assessing support for right wing politicians, the pollsters also under-estimated the leftist Jeremy Corbyn in 2017, then Labour Party leader. It was one of the worst polling errors of modern decades, under-estimating his support by almost 5%.

There are still pundits who have failed to learn the lesson of that Corbyn result, saying – quite wrongly – that Britain would never vote for a firm leftist to become prime minister. But that election proved that the UK public might well do so (and I think that inequality in things such as home ownership means that one day they probably will). Under Corbyn, Labour got one of the highest election vote shares ever got, 40% – more than the centrist Tony Blair did when he was re-elected in 2005 on 35%, and far more that the victorious Sir Keir Starmer did this year or the former Labour leaders Gordon Brown, Neil Kinnock, Michael Foot or Jim Callaghan ever did.

So we know that it is not just the right wing candidates who are under-estimated. But Donald Trump is so controversial that he is just the sort of candidate that people will be shy to admit supporting. In America, the polls under-estimated Donald Trump in 2016 and again in 2020. We are told that pollsters are trying hard to weight their findings to allow for that, but I have read a lot about what went wrong four years ago, and even the pollsters aren’t sure. It is to do with all sorts of issues such as the difficulty in reaching a proper sample of voters in an age in which fewer and fewer people have landline phones. So I am yet to be convinced that the weighting of poll results will work.

Another complication is this: in 2016 the polls under-estimated Hilary Clinton’s lead over him by about 1% (she was 2.1% ahead in the overall vote, but lost on the weighted ‘electoral college’ which is weighted to give a slightly bigger say to smaller states). And in 2020 the polls under-estimated him against Joe Biden by a mammoth 3%.Thus, my ‘average of the polls’ has Kamala Harris ahead by under 3%. Even if Tump is only being under-estimated by 1%, bringing her lead down to 2016 levels of around 2%, she will struggle to win in the electoral college. But if he is under-estimated by 2020 levels, Trump might even win most votes, and handsomely in the college.

I will say more in coming weeks about the merits of the candidates. In summary, neither is a leader of the free world.

• Ben Lowry (@Benlowry2) is News Letter editor