As with others, I have to wonder why actual voting results are so divergent from the polls. It’s crazy.
I saw one person on TV say that undecideds are prone to vote against incumbents. I don’t think that’s been true in the Trump era. I think many white undecideds are really unannounced Trump voters. They are definitely uncomfortable with Trump, in a number of ways, but they are still likely to vote red when they get in the booth.
I think we need to have a different way of talking about polls. For example: instead of saying “Biden leads Trump in state x by 6 points according to one poll,” we should say “Biden leads Trump in state x by 6 points, with undecideds are at 8%.” We then need to have a robust discussion of what the undecideds will do. For example, if an undecided lives in an area that’s pro-Trump, that’s probably not an undecided voter at all, that’s an unannounced Trump voter. In the example I give here, the poll might give Biden a lead, but actually, Biden is perhaps tied or even losing under the expectation that undecideds will break for Trump. And then the margin of error kicks in.
I myself have looked at the polls this year with the idea that undecideds would vote for Trump, and so I braced for the possibility that the election would be a lot closer than some interpretations of the polls had suggested. And here we are.