Based on the latest RealClearPolitics summary of polls, without looking at the partisan split and simply removing the partisan bias per FiveThirtyEight as well as the time-specific bias in polls by state that was evident in 2016 versus the actual results, I get some interesting results.
When I say time-specific bias, I think I owe an example. In Florida in 2016, in October, the RCP polls had Clinton at 48.2% and Trump at 42.4% support. The actual election had Trump garner 49% and Clinton at 47.8%. So Clinton support was overstated in October in the polls by only 0.4%, but at the same time, president Trump got 6.6% more of the total vote than polls projected. Combined, the pro-Clinton effect of Florida polls in October was the Clinton 'victory' margin was overstated by 7%.
It's true that the polls were not as far off in 2016 by the time November rolled around, but prior to that, they were either very wrong, deliberately very misleading or else something big changed in the last week to sway a massive shift in voter preference. I don't think the latter scenario was the case, even though Comey re-opened the Hillary Clinton email investigation in late October. Why do I think that? Because in a good number of the swing states the Hillary bias did not disappear in November polls.
What's more likely is that the polls were being used to push a narrative and that as election day closed in, they had to become more accurate or be viewed as untrustworthy in the future. The bias often persisted right up to the election in 2016, but it was more muted than it was in November.
Polls today have Biden outperforming Hillary's 2016 polls versus Donald Trump. Is that a concern? Definitely - remember that Hillary Clinton had almost as high unfavorables in polls as did Donald Trump. Biden is, inexplicably, more likeable than Hillary Clinton to many. That's why he's in hiding and not wanting to debate Trump again (ignore the lies that it is about Trump having COVID-19).
But it's possible the polls are biased towards Biden as much or more (or less) than they were towards Hillary Clinton. The bias is not easy to gauge. Maybe by 2024 I'll be able to do that. For now, I'm going to assume that the bias is the same for Biden as it was for Clinton.
With all that said - here's what my electoral college map looks like, removing the polling bias (or mistake):
No comments:
Post a Comment
Disagreement is always welcome. Please remain civil. Vulgar or disrespectful comments towards anyone will be removed.