Nice. Good to see two of my favorites, Harry Enten and Amy Walter, on there (along with luminaries like “the Nates” Cohn and Silver).
Frustratingly, they do not include all the data, and I couldn’t find it elsewhere (unless it’s behind the WSJ paywall). But if you read the whole article, there is clearly another category not mentioned in this section: “have reservations”. So Trump partisans can on the one hand spin this as half full: “Look, our guy is only three points behind Biden on the combined ‘comfortable/enthusiastic’ categories!” But they are 21 points worse in the “very uncomfortable” box. By inference, we can say there is probably something close to a tenth of the electorate who “has reservations” about voting for Trump, in addition to over half being “very uncomfortable”; whereas Biden has a much larger “reservations” group of about one-fourth of the electorate. (That almost certainly includes a lot of hardcore Democrats who, like me, are not enthused about Biden, but will drag ourselves over broken glass to vote for him in November 2020 if he is the nominee, because we are so intent on getting rid of Trump.)
It seems to me that would be most meaningful are head-to-head comparisons. “If the election were held today, and were Biden vs. Trump, whom would you vote for?” (and repeat for several other Democratic candidates). If you really want to get fancy, you could start by asking which Democratic candidate the answerer likes best, and then (if it’s not one of the ones you’ve already asked for) “If the election were Trump vs. your preferred Democratic candidate, whom would you vote for?”.
I wouldn’t put much stock in enthusiasm data this early in the campaign. That becomes more of a thing in the general, once the parties ask voters to unite behind a single candidate.
Head to heads or “trial heats” are notoriously inaccurate this far out. But I believe the polls that show 52-55 percent unwilling to even consider Trump really do mean something this time, because we have never seen something like this before. They reflect that a majority, albeit a modest one, sees the trainwreck for exactly what it is. Five years ago, if a time traveler from 2019 described everything that has happened, the only real surprise would be that these numbers weren’t at least 70%.
There have been other polls that have showed similar numbers and the issue is the same: being uncomfortable voting for Trump is not the same as voting for the person running against him. The biggest winner last time was as usual not bothering to vote.
It is consistent with Trump having a ceiling of support. But he’s had that ceiling. He’s never been popular and the honeymoon after an election was for him very brief. But his ceiling is, if he wins the right states, just enough for him to win.
A D candidate needs more than voters not voting for Trump. They need voters to also vote for them.
Wouldn’t hurt, but this time I think it’s actually the best bet for the nominee to be the closest thing to a bowl of plain oatmeal in a bowl labeled “Generic, Inoffensive Democrat Who Speaks In Vague Platitudes”.
Are we trusting these polls? I know people say the polls in 2016 were within the margin of error, but I also know everyone thought Clinton was essentially a lock. If it was the media’s faulty analysis of the polls, do we know they’ve learned their lessons? I honestly don’t know what or who to trust. Have the pollsters learned from the shy Trumper effect?
The polls were all reasonably accurate, and the media almost all misinterpreted them. The solution is to continue to trust the polls, but to not trust most of the media.
And let’s please note that Nate Silver and 538 did NOT call Hillary a lock. They gave Trump a one in three chance of winning (and took a lot of flak for it from the “lock” crowd). 1 in 3 is Russian roulette with two bullets in a six-shooter, and one of them rolled into the chamber.
Let’s also remember that the polls were pretty accurate last fall.
What about the SDMB? This tends to be a community of people who look into the granular details and don’t accept what they’re fed by the media, but a poll taken in 7/16 had a 82% chance of a win for Hillary (not sure if things changed later). And like others have said, fivethirtyeight was relatively conservative and they had Hillary at 70%. I recall seeing a bunch of battleground state electoral analysis that was calling for a Clinton landslide, so I can’t imagine it was all down to over-reliance on national polls.
Do you really think 70% (which was as low as 60% earlier in the fall) is the chances one would give to a likely landslide? If I saw that for a Senate election (which does come down to popular vote), I would expect that the candidate with the 70 percent chance would probably win by one or two points. Hillary won by 2.1 points.
Even 86% is probably more like four or five points. A ten point win is going to look more like >95%.
I’m not sure how a poll of the SDMB translates into an 82% chance of a Clinton victory. Was the question just “Who do you think is more likely to win?” Because if everyone thought that she had a 2/3 chance of winning, that would have resulted in a 100% for Clinton in the poll.
As for the folks saying 99% or more based on individual state polls, the problem there was that they were incorrectly assuming that states are uncorrelated. In reality, in a case where Trump outperforms his polls in Ohio, he probably also does so in Pennsylvania and Michigan, and likely does so in Florida and Iowa and other states. Which is what in fact happened.
Poor choice of words on my part.
I’ve been imprecise in my wording. The question was “Who do you think our next president will be.” While it may not ask “What is probability of a Clinton win?” it still illustrates that 82% of people in a relatively analytical and detail-oriented population thought she would win and, iirc, most people thought it would be by a pretty safe margin. I think even pollsters said they didn’t take into account the Shy Trumper effect. Im just wondering if they’ve made changes to their methods after 2016, who I can trust, and how to get an accurate assessment for 2020.
The polling in 2016 was actually quite good, on a national level, and even on a state-by-state level was quite good aside from a few states. Only Michigan (or maybe Wisconsin) had a greater than “normal” level of error in the polling aggregates, IIRC.
I don’t think polls are going to be of much value at any point during this race. As the last election proved, enough people have zero problem lying to pollsters if they believe their sincerely held beliefs are problematic to make any poll results questionable. Best I can tell, it seems to work a bit like this.
A) Mainstream commentators decide amongst themselves that right-of-center positions are only held by a bad fringe.
B) Voters then lie to pollsters in response.
C) A vote is held.
D) Same commentators switch to complaining about dangers of “populism”.
Did you miss the 2018 election results?
A new poll has found some vulnerabilities for Trump among his base, particularly regarding taxes and health care:
https://www.politico.com/story/2019/06/24/democratic-poll-trump-health-care-1377517
I like the savvy exhibited in their stated goal: “We’re trying to go from losing these segments 85-15 to maybe 75-25”. That sounds like it’s still a calamitous loss, but it would actually checkmate Trump in swing states.
If Donald actually loses a couple of points in his base, he’s toast next year because I don’t think he can make up the deficit elsewhere… I wish I had confidence in that happening.
Repeating myself here, I will never understand people lying to pollsters. I can not wrap my brain around the desire to do this.