Meanwhile, brand new poll out of Florida from Florida Atlantic University’s Political Communication and Public Opinion Research Lab (PolCom Lab) (my emphasis):
Former U.S. President Donald Trump holds a slight edge over U.S. Vice President Kamala Harris among likely Florida voters, with 50% supporting Trump and 47% backing Harris. This close margin suggests Florida’s potential swing state status may return in the upcoming election.
“The narrowing gap between Trump and Harris is consistent with the tightening we have seen in other states,” said Kevin Wagner, Ph.D., professor of political science and co-director of the PolCom Lab. “If this trend holds, we may see a competitive race in Florida.”
“Young people who attend schools” works ok to get more numbers but it’s still not necessarily representative. There are plenty of people in that age range who aren’t in education or vocational training. So, some weighting is still required.
Ultimately, there’s a large excluded middle here. While not as representative as we’d like, there are poll respondents across all age ranges. Weighting is still required but where things tend to go wrong (as referenced earlier in the thread) is that the margins of error are themselves erroneous. The margins are almost certainly larger than reported, probably by quite a lot. That can lead to unjustified confidence in the results, even if the polls themselves can be quite useful.
It reminds me, to some extent, of how NASA estimated failure rates for the Space Shuttle, and how they were shown to be bogus. They had a lot of good numbers and engineers on the back end but also a lot of motivated reasoning and managers up front not making proper use of them.
In addition to @Northern_Piper’s excellent objection with which I fully agree I’ll add another.
If your logic is perfect then the popular vote will be a much underpredicted landslide.
OTOH, the EV count, the only count which matters, will be much closer because the “Swifties” are concentrated in the bluer parts of the bluer states where more blue votes don’t change the EV situation at all.
I hope you’re right as to the severe underpolling. I think you probably are. How much that changes the actual outcome is where we may diverge.
And not just geographically-neighboring states. A white 39-year-old suburban Presbyterian woman in Virginia is likely to vote similarly to a white 39-year-old suburban Presbyterian woman in Wisconsin, or Georgia, or any other state. IIRC, Silver has 20-some different demographic variables that he’s done statistical analysis on to establish correlations between states, and a lot of the differences in voting between different states is due to different demographics in them.
Then, too, there’s some degree of correllation between all states. If the nationwide average shifts five points blueward between July and August, then it’s not the worst bet to think it probably changed by about five points in an unpolled state, too. You can do better than that, of course, especially if your polls have cross-tabs, but it’s a good start.
Statisticians can do all sorts of amazing things with data, but they can’t do anything without data. There just isn’t enough data on the “Swifties”. You can try to get more data on them, but it’s hard to tell if your methods are actually balanced, because there might well be many different subgroups of Swifties, and without the data that you’re missing in the first place, how do you know which relevant groups those are? Heck, literal Swifties, people who are specifically fans of Taylor Swift, might be a relevant group. Even there, do you find them at a Taylor Swift concert? Maybe, or maybe that’s just rich Swifties you find there, which might not be the same group.
538, The New York Times, and a few other poll aggregators have added Outward Intelligence’s online-only poll to their models today. Outward Intelligence is a new polling company, founded in 2021 and independently funded. 538 has ratings for over 200 polling outfits, but did not have Outward Intelligence in their February 2024 pollster ratings.
Monmouth University Polling Institute yesterday released a multi-faceted “strength of support” poll for the 2024 presidential general election. Its results won’t get included into aggregators’ models because they don’t ask top-line horse-race questions. Instead, they ask questions in this fashion:
The “definitely nots” for both candidates tell the tale – 49% definitely won’t vote for Trump, 44% won’t for Harris. Those five percentage points are also the separation on their respective “definitely” vote-for numbers.
Cross-posting Moriarty’s post about Republican pollster Frank Lutz from the Kamala run-up thread. Cliffs Notes: Lutz is bullish on Harris and has the numbers to back it up:
Yes, and he acknowledges that. His point is that in all his years in the polling business, he has never seen a turnaround in the political dynamic like this one.
Depending on the state. I’m no Silver fan boy but that is one of his points: different states have different degrees of elasticity. Georgia for example he says, is rather inelastic, and moves less than the national numbers shift.
And extrapolating (weighting) from smaller sample size of data that actually is not representative of that subgroup can lead to errors in any direction.
I remember voting for Obama in 2008. I was in Ft Lauderdale, a predominantly blue area. Turnout was insane. We were happy to stand in line for 4 or more hours. When you saw the line, and the atmosphere, you could just tell that he had inspired a lot of people who were otherwise unengaged to vote.
In 2016, I volunteered at a phone bank, and voted absentee, but I imagine that enthusiasm from the otherwise disinterested tracked towards Trump.
Now, the swell of lethargic people who are now engaged are voting for Harris, and were not accounted for when it was Trump/Biden. That is going to tip the scales, and what I believe Luntz has identified.
It’s taken nearly two weeks … but Kamala Harris has finally surpassed a 3% lead over Trump in FiveThirtyEight’s national poll aggregate (3.3% today, previous high had been 2.9%).
Also from that link – some pollsters don’t think Trump’s support is being undercounted this time around (my emphases):
“People who told us they voted for Trump in 2020 are responding at the same rates as people who told us they voted for Biden in 2020,” said Jackson, from Ipsos, which “suggests we’re not having a really strong systemic bias.” The New York Times poll master Nate Cohn made a similar observation in a recent interview with The New Yorker: Democrats were much likelier to respond to Times polls in 2020, but this year, “it’s fairly even—so I’m cautiously optimistic that this means that we don’t have a deep, hidden non-response bias.” Another difference between 2020 and now: There is no pandemic. Some experts believe that Democratic voters were more likely to answer surveys in 2020 because they were more likely than Republicans to be at home with little else to do.
What’s clear at this point is that the election is close, and Harris is in a stronger position than Biden was. Natalie Jackson, a Democratic pollster at GQR Research, told me that if Harris’s numbers were just a result of energized Democrats being in the mood to answer polls, then Democrats would be seeing a comparable bump in generic congressional polls. The fact that they aren’t suggests that the change is real. “Trump’s numbers haven’t moved,” Jackson said. “This is all shifting from third party or undecided to Democrat.”
Pollsters tend to be pretty good at fixing the problems from the prior election. The problem lies it what changes happened in the intervening 4 year that aren’t being accounted for which could go either way.
Thought this was interesting: Nate Kohn, the NYT’s pollster, wrote an article about cohorts of voters among which Harris has gained and lost the most support compared to Joe Biden (free link via Yahoo News reprint). The two graphics serve as a useful summary – Kohn is most impressed by the gains Harris has made among voters with “a somewhat unfavorable view of Trump”:
The group that said “I dislike Biden and trump both but between the two I dislike trump less. Sleepy Joe scares me.” Now suddenly in a sense a viable 3rd party candidate, Harris, has POOF! appeared from behind the curtain in a cloud of brightly colored blue smoke and their argument that in their opinion trump is the lesser of two evils needs to be reevaluated.
The huge difference between Harris as a “3rd party” surprise and e.g. RFK Jr. as a real 3rd party candidate is that a) Harris is the candidate for a real party, not an underfunded malcontent fringe group. And b) Harris in neither insane nor unqualified. Unlike both RFK & trump.
