Blog/In print

Donald Trump was reduced to talking about immigrants eating pet cats and dogs during Tuesday night’s presidential debate. His own supporters would admit it didn’t go well. Lacking message discipline and straying too far into the bizarre, a CNN poll found that 63 per cent of viewers judged the debate a win for Kamala Harris. Winning debates and elections aren’t the same thing, though. The polls — and votes — that really matter in the race for the White House lie in the swing states. And currently things are close — but favour Trump.

How on earth did we get there? How does Trump, a convicted felon, still stand as the marginal favourite to win? The Democrat eco-system was high on a sugar-rush of relief after President Biden withdrew his candidacy. Harris’s polling honeymoon put her up four points nationally in the afterglow of a successful Democratic National Convention in late August. I attended this activist boondoggle with a couple of other pollsters and the atmosphere was feverish. Talk of a convention bounce, of a “joyful” candidate comfortable with herself and country, was taken as serious analysis.

The autumn has seen much of that initial optimism fade in light of less favourable polling. My firm interviewed 16,000 voters as part of an MRP poll (multi-level regression and post-stratification) out this week. It suggests the Democrats are on course to lose the critical state of Pennsylvania, along with Arizona, North Carolina, Georgia and Wisconsin. This would leave Harris with 247 electoral college votes, 23 short of the 270 required to win the presidency. Crucially our MRP model says Harris needs to be 3.5 to 4 per cent ahead in the national vote to be winning the swing states — or this could be the third time since 2000 Democrats win the popular vote but lose in the electoral college. She needs a second honeymoon. The debate is a start.

• US election poll tracker

There’s a question of whether her initial bounce in the polls was ever real. I think the phenomenon of “non-response bias” fuelled her rise. This is when voters become reluctant to take part in polls when their party is doing badly in the media, and are more inclined to take part when news is better — a bit like avoiding the league table when your football team is losing and looking every day when they are winning.

Even if the honeymoon was real there are some structural factors that should alarm Democrats. First, the voters Harris has won back — young and highly liberal — since Biden’s exit are disproportionately less likely to turn out and more likely to be living in states that matter less. Our polling suggests Harris has increased the Democratic lead among voters aged 18-24 by 19 points, and by 11 points among those who identify as “very liberal”. But there are fewer of the very young and very liberal in moderate swing states.

Second, America’s political realignment over the past 25 years, in which the university educated have drifted left and the non-university educated drifted right, has tilted the electoral college towards the Republicans. State demographics mean the Republicans have thus traded away Colorado and Virginia (together worth 23 electoral college votes) while banking Florida, Ohio and Iowa (worth a collective 53 votes).

• What will happen if Harris wins?

Problematic too for Harris is the economy. Rampant post-Covid inflation has helped turf out or hobble governments across the western world. Our polling shows the cost of living to be the top issue for more than 50 per cent of voters across the swing states — and on this Trump leads by about 25 per cent as the person best placed to handle it. Trump also has a 72 per cent lead on immigration, the second most important issue to Americans — one reason he kept coming back to it during the debate. It’s highly salient — and the Democrats have yet to find a line that works on the issue.

Perhaps that explains why the Democrats are looking over the pond to Starmer’s election victory (perhaps less on how he’s governing). Back in July Labour won a 174-seat majority with just 33.7 per cent of the vote. To construct this extraordinarily thin but wide coalition, Labour zeroed in on voters who were more suburban, working class, socially conservative, patriotic and immigration-sceptic than its typical supporters. In targeting them, Labour embraced values far from its core identity because it recognised the truth that some votes simply matter more than others in a first past the post system. This is also true in the US.

Sitting in the convention hall in Chicago last month I was struck by how little the Democratic party sought to stretch its coalition. How little it tested the patience of the hyper-progressive America and tilted towards middle America. Sure, the convention is for the party faithful, but it’s also a chance to talk to all Americans. Speakers performatively noted the contributions of American society that lean right. Veterans, rural and Christian America were there in body but not really in spirit.

This election will be won and lost in moderate suburbia and specifically the state of Pennsylvania. With huge Democrat votes in its cities — Philadelphia and Pittsburgh — big Republican support in its southern counties that border Maryland and 50/50 suburbs, it’s a composite of moderate, conservative America.

Pollsters’ forecasts of a Trump victory (the Olympian pollster Nate Silver has given the Republican a 64 per cent chance of winning) flow in large part from which way Pennsylvania will go. Our model had Harris 1.4 per cent behind Trump here, and other polls in the state also have her trailing. Harris could have improved her chances of winning here by choosing its popular governor Josh Shapiro as her vice-presidential nominee. He won the state by 15 points in 2022.

To win, Harris needs to speak to voters who care about immigration and the border and who want to know how the government is going to help them survive pay cheque to pay cheque. She must nod more towards rural, religious and business-minded America. Irrespective of Silver’s probabilities — he gave Trump a 29 per cent probability of winning in 2016 — Harris has time before November 5. Our modelling suggests some 60,000 voters will determine who becomes president. That’s like one person’s vote determining the decision of a group of two and half thousand people. Probabilities don’t fully capture just how small the battleground is to determine the leader of the free world. There’s a lot more talking to be done.

Article originally appeared in The Times.