Today’s entry: Despite the polls looking good for Biden, they still aren’t adding up to 100% in national or state contests. That suggests there are still undecided voters unaccounted for in these polls. Even if we’re to believe them, I remain skeptical. You’ve discussed how state polls were off by more than the national polls last time. What about these unaccounted-for voters in these polls. Is there any indication as to how they might break?
Bottom Line: Late breakers have long been a tell in close election cycles because yes, while it’s surprising to many, there are always undecided, or at least undeclared voters who show up in even the final Election Day polls. In the 2016 election, for example, there were 5.7% of voters who were unaccounted for on Election Day. They broke 2.3% for Clinton, 3.1% for Trump, and 0.3% for third party.
Notably, it was Trump’s ability to win most of the late breakers in 2016 which led to his victory. While .8% doesn’t sound like much on the surface, it exceeded the size of his wins in Michigan, Pennsylvania, and Wisconsin. But there’s another factor key to this conversation, 3rd party candidates. In 2016, 4.7% of voters voted for a third-party candidate, with former Republican turned Libertarian Gary Johnson pulling close to 4% of the vote. The third-party candidates are showing only 2.6% support this time around. When they are sampled, which is rare, but still, that is largely unaccounted for in state polling which represents a wildcard.
Factoring in the third-party candidates there appears to be 4.2% of voters who are undecided or undeclared according to national averages. The lack of polling of third-party candidates in state polls along with these undeclared voters represent the potential for significant variance between state polls and outcomes once again. So, back to your question, how do late breakers usually break? I believe in using data from like cycles. That means I’m looking at just presidential reelection bids. If we use the last three reelection bids by a Republican President here’s what happened. In 1984, Reagan did 0.21% better than final polling averages. In 1992, Bush did 6.44% better than final polling averages. And in 2004, Bush did 2.4% better than final polling averages.
Notice a trend? Incumbent Republican Presidents all outperformed their final polling averages. The average final variance is incumbent Republicans faring three points better than the final polls. I wouldn’t be at all surprised to see something similar during this cycle, especially within states where third party polling is largely absent. What you notice is a history of pollsters underrepresenting the Republican incumbent President, especially with late breakers. Moving away from numbers there’s a degree of logic to it and it’s not unique to Republicans. If one is still undecided or persuadable at this stage of the cycle, the question, which needs to be addressed. Clearly, the incumbent didn’t win them over but at the same time neither did the challenger. In these instances, in which a voter doesn’t feel generally convinced they tend to stay with the known over the unknown. This is at the crux of the incumbency advantage and is a key reason why 65% of Presidents who run for reelection win.
Submit your questions using one of these methods.
Email: brianmudd@iheartmedia.com
Twitter: @brianmuddradio
Facebook: Brian Mudd https://www.facebook.com/brian.mudd1
Photo by: Getty Images