Late deciders typically break 3:2 or 2:1 the same direction across states, and given how close the race is now, that will decide all the swing states.

The challenger generally has the advantage with late deciders, and I think Harris is the challenger in this situation since Trump has already been President.

For all of Trump's "momentum" the last week or two, the non-aligned polls have not moved at all in months. The momentum is mostly MSM stories and Red Wave polls.

I remember it from 2004 as evidence Kerry was going to win since Bush was under 50%. They actually went for Bush.

Similarly in 2012, the same talking points were used for Romney.

They broke Obama.

In 2016? They broke Trump, who was the challenger really of that race. Of those who picked in the final week, Trump won 49-41 over Hillary.

And in 2020, as the incumbent, he won those who decided in the last week 64-31.

I just don't know how you can predict this stuff tbh

The basic idea is that if the more well-known candidate does not make the sale by the final week or so, the undecideds will break the other direction. I assumed everyone understood this basic concept which is generally true for all political races from President of the United States down to local Zoning Board.


2004 - Bush had a small but solid lead the last month of the campaign. Other than a few polls and the early exits, Bush was up around 2 points for a month. He won by 2.4%. You could argue that undecided broke roughly even.

2008 - Undecideds broke big for Obama, who was effectively the challenger as an unknown.

2012 - I agree that this is an example of the undecideds breaking to the incumbent.

2016 - Trump was clearly the challenger. He had never held a public office before. Hillary Clinton had been in the public eye as a political person for decades.

2000 - I don't know which way the undecideds broke. Turnout was huge, with the 2020 election having the third highest turnout in American History as a % of voting age population (61.5%), surpassed only by 1960 (62.8%) and 1952 (61.6%). I think the turnout on both sides so dramatically exceeded anything the polls expected that the likely voter screens were off on both sides, which means the undecided screens were off too.

Your own post shows it's a mix of both. It's not really about the challenger - it's about the candidates.

I don't think it's safe to presume Harris will carry most the undecideds with a candidate like Trump. You absolutely could make the argument that if someone is undecided about voting Trump literally to the last moment, after everything he's said and done, they're probably not really all that undecided anyway.

And that's why polling has underestimated his support - a lot of supposed undecideds could not admit they supported him.