I may be missing the linked information. How many elections did that ranking include?
A year ago, Nate Silver wrote that Rasmussen’s accuracy is average:
I can’t find the ranking list anymore. Rasmussen criticisms here and under that. Rasmussen Reports - Wikipedia
Since this article bases many of its arguments on 2022 (mid-term) and 2024 polling, I’ll cross-post this here.
Looking at the OP of this thread … this article makes more the case that “it’s the news” – that the problem is coverage about poll results and an apparent unwillingness to dig below surface interpretation of the numbers.
Nate Silver today posted a 66 per cent chance for Trump to win, while 538, using some of the same intellectual property, has Biden with a 51 percent chance to win.
I would say that fundamentals (incumbency and objective economic conditions) favor Biden, while swing state polls, and 2020-2024 same day polling comparisons, favor Trump.
And popular vote polling is a tie.
Curious to see how/if this changes over the next week or two. All of the feeder polls but one came before the debate.
Who Is Favored To Win The 2024 Presidential Election?
538 uses polling, economic and demographic data to explore likely election outcomes.
Biden wins 51 times out of 100
in our simulations of the 2024 presidential election.
Trump wins 49 times out of 100.
They have polls that match a good number of candidates. If you present it the right way, the voters will probably say what they think, Kamala vs Trump for instance. However, the voters know the actual situation. If Biden drops out, you will get a realistic answer about two weeks later. Now, the Democrats may be planning something. And Biden will be presented with a successor. We will quickly have one candidate, despite the fact that the convention will need to vote. So if done well, it may work and calm the Democratic voters. May even need to have Biden give a couple of speeches. But then he is out. The new candidate will need to take an agressive stand to Trump immediately.
Full acknowledgement that small fluctuations in aggregate poll numbers don’t mean much ** . All the same, Trump’s lead in RealClear Polling’s aggregate dropped 20% this week.
Gosh … that was weaselly. More straightforwardly: Trump led Biden by 3.4% on Sunday, July 7th and leads by 0.7% less today (2.7%). A 20.58-percent reduction in the size of the lead.
All 14 of the polls used to calculate the average are post-debate polls. The NPR/PBS/Marist poll added today is the first Biden-plus poll added to the aggregate in two weeks.
** OK, so maybe my fingers are crossed behind my back. Harumph.
This came up in another thread but the inside baseball methodology question is maybe best asked here.
Past cycle underestimates of Trump support had been explained not by “shy Trumper” but by those who supported Trump being somewhat less likely to participate in polls.
Have the pollsters adjusted methodologies to attempt to compensate for that effect? Some yes some no?
Another question as I try to talk myself down from panic in the case Biden sticks it out - there have been several articles, and some discussion here, about Trump’s lead being greatest in the disengaged voter group. I seem to recall that in general polls tend to narrow their definitions of likely voters as the election gets closer. Is that so? And if so maybe the numbers look better as that happens? Of course if so I am sure aggregators like Silver are already aware.
(my bolding)
The post below contains indirect support for the idea that “yes, they have compensated”. Or else are getting more representative sampling of Trump voters.
Namely: Trump’s 2024 polling results – in both national and battleground-state polling – are in line with his actual vote percentages in 2016 and 2020.
It’s possible that that change is meaningful. I’m not a enough of a statistician to say one way or another, but I do know that the larger the data set, the smaller the error bar. Which means that a change in a poll with a larger sample size doesn’t have to be as big to be meaningful. Aggregation is a way of artificially increasing the sample size, so that change could be significant.
Polling rated. Rasmussen does not get listed.
What are the best pollsters in America? - ABC News.
I agree. It is tempting for me to use the brief vivid “shy trumper” phrase, since the real explanation is wordy.
Some discussion here:
How 2024 pollsters are trying to avoid their 2020 mistakes
Despite the headiline, the real push was done in 2020. The fix was to weight by education, since we know Trump voters are less educated. And it only worked a little.
The article below covers just one recent poll (NPR/PBS/Marist), and for better or worse it took place before the assassination attempt. While I don’t think this particular poll result is determinative, I do think it makes a lot of things that people think are determinative seem less, um, determinative.
The poll asked: Which is more concerning in a president, someone who doesn’t tell the truth, or someone who might be too old to serve? The results were lopsided: By 68 to 32 percent, respondents were more concerned about the lying than the aging. Given the relentless media focus on presidential age of late, that’s simply remarkable.
While the poll doesn’t directly compare Trump and Biden on that particular question, it also finds that 52 percent of Americans say Biden has the “character to serve as president,” whereas only 43 percent say this about Trump. Fifty-six percent say Trump lacks the character to serve, which surely reflects public perceptions of Trump’s dishonesty.
Wikipedia article on 2024 polls.
Wikipedia is a good “see it all” tool for polling. You can also see dedicated Wikipedia articles about polling in past election cycles – although if you go back far enough, Wikipedia roll the polling information into a sub-section of the article about a given year’s general presidential election (1988 polling, for example).
Ah-yup.
Every other Trump-Harris poll in the RCP aggregate since Biden dropped out (6 polls): Each within margin of error, Trump up in four, Harris in two.
The seventh poll since Biden dropped out, Rasmussen? Trump +7
(link below changes frequently)
Helpful summary — thanks.
Mainly good news for Democrats, but two observations are worrisome:
-
RFK’s polling has plummeted — just as he (according to other data I’ve read) seemed to be drawing disproportionately from would-be Trump voters.
-
“Harris is still short of Biden’s 2020 numbers among young voters and voters of color, and the former president is still running well ahead of his 2016 and 2020 numbers among those groups” — despite Harris’ significant gains for these demographics over post-debate (and even pre-debate) Biden.
I’ll play those both as positives.
RFK numbers dropped some after the assassination attempt as Trump leaners there moved back.
Now they are dropping as D leaners move back. I think those left still overall lean Trump.
The second is the opportunity for a much higher ceiling than Biden had. She has strong potential to get back to Biden ‘20 levels or beyond; Biden was stuck. She has already mostly caught up without doing that. The possibility of getting comfortably ahead with that makes a heart flutter!
The offset per The New Republic:
Former President Donald Trump has a serious problem with white men: They don’t want to vote for him. At least, not the way they did in 2016.
…
In April, an NPR/PBS NewsHour/Marist survey found that the demographics that historically supported Trump have begun to shift and the groundswell of support from white male voters that Trump had experienced in 2016 was dissipating.
…
Things only got worse from there. According to NPR, between 2020 and 2024, Biden saw a 24-point bump among white men with college degrees, a group that has historically backed Republicans.
Trump has begun to experience a slip in swing states, as well.