I’m not sure I see “a lack of continued increase” as a tightening. It’s, well, staying the same.
At this point it would take a really huge swing in polls for the 538 prediction to change, since it’s a more-or-less Bayesian model. That did happen in 2016, thanks to James Comey’s memo. It doesn’t appear likely to happen this year, as the Hunter Biden Laptop bullshit story isn’t really catching fire.
Hmm… I think it’s somewhere in between. The model assumes a certain mount of uncertainty, and that uncertainty decreases as you get closer to election day. At election day, the only uncertainty should be polling error.
The model used to have a “now-cast”, but they don’t publish that anymore. Nate Silver has been tweeting it from time to time. The last one I remember was like 6% or so Trump odds of winning the EC. So, one would assume that between the day Silver tweeted that and election day the Trump odds would decrease from 13% (or whatever it was that day) to 6%. But if it stays at 13% that would imply a weakening in Biden’s position in the underlying data, no?
I don’t think they give much in the way of “this input caused this output change” other than in the various commentary pieces. I agree it’s frustrating, and it’s probably best to just check it once a day or whatever and not be as hyper-aware of it as many of us are… there really isn’t any practical difference between an 88% and an 86% chance of Biden victory.
538 has an “uncertainty factor” that decays day by day, and in its priors includes the swing from the Comey factor (along with other historical swings). So if the polls themselves stay exactly the same, the probability envelope would decrease (from a previous comment from Nate from a couple days ago, it would take an 87 to about a 91 on election day if the polls stayed the same). So if the 87 is still an 87 on election day, that’s a tightening in the polls counteracting the decreasing uncertainty.
I’d argue that some (probably most) of the observed tightening is really just reversion to the norm as the temporary decline in Trump’s standing triggered by the Trump has COVID! and Trump Cancels Debates! has worn off. The polling response to any news for/against either candidate has a delay before it fully reflects in the polls then a decay half-life as it becomes old news and interest fades away. There need not be any deeper significance than this cause-wise.
In terms of impact, I’ll argue that most of the recent big news for either candidate has been fully decayed and incorporated into the polls. No big news moves in the last 10-ish days.
So absent a new surprise, all we’ll see in the next few days is a narrowing of the error bars as the several posters just above me have noted.
And if we do get a surprise, well, it’ll do whatever it does and by definition we can’t decide now how much things will move then.
I agree that the current 86-88% range is likely a combination of wobble and a return to the longer-term average of before the first debate (across the board and reflected in the national polling average). But you could also get a drop like that if a few state polls come in that say “omg the lead in WIS is half what we thought it was.” My family’s phone banking and still donating to key races, so that’s the kind of change I’ve got an eye out for.
Harry Enten on polls tightening…
Part of Nate Silver’s reply…
and
To contrast, New Poll Shows Biden's Lead Over Trump Grows to 9 Points in Wisconsin, Beyond the Margin of Error
Damn, it would be so nice to turn into the news on the 3rd and see Biden has turned GA blue.
Well, regardless of what happens next week, I was heartened to see this happening and I hope the people there stay the course and keep fighting the suppression machine.
Asking for a friend?
If Trump is ahead in ANY of these swing states at some point after the polls close on November 3, (and I’m going to include Arizona, Colorado, Florida, Georgia, Iowa, Maine, Michigan, North Carolina, Ohio, Pennsylvania, Texas, and Wisconsin in this list), then watch for:
Republican lawyers/operatives to move into action and petition to have the counting shut down (due to “voter fraud”, “irregularities”, “bad juju” or whatever), and have Trump declared the winner.
Look for confiscation of mail ballots, destruction of mail ballots, whatever it takes.
Trump will be declared the winner, LOUDLY and OFTEN. Whatever it takes to “win” these states.
At 70+ I was not even aware of an “Electoral college” until Hillary won the vote count; and yet lost the election because of 270+ Electoral College votes. These are people I have never heard of.
Even today I could not give you a name of any of them.
The Electoral College is not a permanent body. Every four years new people are elected to it, they meet in their respective state capitals, vote for president and VP, and then they’re done being EC members. Hardly anyone knows who they are or really care, as long as they vote for the candidates they’re pledged to. They only get in the news if they’re unfaithful.
Are you saying that some electoral college members voted against the wishes of their constituents>?
There are often a couple. In 2016 there was a record seven - five of Clinton’s electors and two of Trump’s. None of whom, I might add, voted for either candidate.
“Faithless” electors have happened, though pretty rarely. Since the electors are selected by the political parties, they try very hard to choose people who will be loyal to the party, and the party’s candidate. As Wikipedia notes:
There were three more. Some states now have laws that allow an elector to be replaced on the spot if they attempt to vote for someone they’re not pledged to. This happened to two electors, one from Colorado and one from Minnesota. One from Maine tried to vote for Sanders, but was ruled out of order. So he changed his vote to Clinton. There’s a summary of all the faithless electors in this NPR article: