In opinion polling, particularly in the United States, there is a phenomenon known as partisan response reflex. That is to say, the committed political partisan, when polled, will always give the answer to a polling question most favourable to his or her own side, even when they know the answer they are giving is utter nonsense. Ask 1000 Americans in an opinion poll which of the two Presidential candidates owns the nicer golf courses, and some portion will answer “Kamala Harris”. Ask them which of the two candidates wears a blouse pantsuit combination better, and some portion will answer “Donald Trump”.
When it comes to polling Presidential debates, this should be factored in. A portion of the electorate will always say that their guy (or gal) won, regardless of what happened. In June, for example, Donald Trump was held to have been the victor over Joe Biden by a margin of 67% to 33% in the CNN flash poll immediately after the event. Did one third of Americans really think that Joe Biden won that debate, or were they simply always going to say that regardless of what happened? You decide.
In the aftermath of Tuesday night’s debate, the CNN flash poll was almost as disastrous for Donald Trump as it was for Joe Biden in June: Harris was judged the winner by 63% to 37%, with Trump just 4 points short of the disastrous margin of defeat that ended Biden’s campaign. And again, the question must be asked: Did the 37% really, genuinely, think Trump won, or would they have said that in literally any circumstance?
It is important to put those two figures side by side, because it is the same poll, using the same methodology, judging two different events. If you choose only to believe the poll when it tells you what you wanted to believe, then you’re a prime candidate for partisan response bias.
That said, clearly, and unlike in June, the Tuesday debate will not end the Trump campaign or come close to doing so. The simplest reason for this is that it is now too late for Republicans to select another candidate, but even if it were, Republican voters (and indeed many right wingers around the globe) are simply emotionally invested in Trump in a way that liberals were never invested in Joe Biden. Biden was a tool to do the job for many of his supporters; by contrast, Trump, for far too many of his supporters, is the job. Were Trump dumped from the Republican ticket tomorrow by some act of God, polls have shown that many of his supporters would simply refuse to vote.
(Ironically, these are by and large the same voters who insist that all other right wingers must support their guy because the alternative is worse.)
Now, there are reasons to be skeptical, in the round, of post-debate polls. The first reason is that post-debate polls are, by definition, polls of people who watched the debate. The majority of US voters will simply not have watched the debate, which means that you’re looking at a poll of people interested enough to pay close attention.
The second reason is that debate-viewing audiences can be skewed. For example, there is reason to think that Republican voters may have been more eager to view June’s debate than Tuesday’s, and vice versa. As a glory-hunting Manchester United fan of old, for example, I am more likely to watch Match of the Day if I know the team has played Luton Town FC that Saturday than I am if I know they’ve played Manchester City. There is intuitively reason to think that Tuesday’s audience may have skewed more Democratic than June’s, as Democratic voters may have been hopeful of a better performance.
The third reason is that most people, thankfully, do not exhibit partisan response bias. That is to say, there is some portion of the 63% who told the pollster that Harris won who both know that she won the debate, and know that they’re going to vote for Trump anyway for reasons separate to the debate. This is the difference between a Manchester United fan who insists his team is better than Man City, and one who lives in reality: Both still, ultimately, support the red team regardless.
The fourth reason is simple fundamentals: As polls repeatedly show, the audience of persuadable voters in this election is tiny. As I’m writing this, for example, CNN has a politics professor on talking about “the key 200,000 voters who will decide the election”. Those voters live in six or seven key states, and anyone who tells you that they understand them is talking nonsense. If either campaign knew for certain what that tiny sliver of voters wanted, the election would already be functionally over.
However, those same fundamentals are the biggest problem for the Trump campaign: He has, in two previous Presidential elections, not once exceeded 46.9% of the vote. Already, many of his strongest supporters are insisting that his path to victory lies in a popular vote defeat, but an electoral college win – an event so rare that it has only happened five times in fifty-nine elections since the country was founded.
Harris has her own challenges with fundamentals: Voters simultaneously regard the economy as their most important issue, and the state of the economy as bad. On several other issues where voters are concerned – immigration and crime, for example – voters rate Trump well ahead of her.
Before the debate, all of the fundamentals pointed to a very tight election decided by a handful of votes in a few states. After the debate, that remains the case.
That said, it is very clear that Harris’s debate performance was a victory on a similar scale to Trump’s win over Biden in June. In June, that mattered a lot. We will have to see whether it matters as much in September. I suspect not, but we will see.