. . . what some of them got right. We don't know yet, obviously. Here's a couple of posts from elsewhere that are setting the early parameters of the discussion. Of course, we won't know what worked and what didn't until we have a winner declared. Or winners, since there are a few hundred races besides the presidency.
Nope. But when it is, there will be a few pages on the polling issue. Just as there were in 2016, 2020, and the midterms.
Nate Silver had a lot to say about polling before the elections. https://www.perigon.io/news/politics/2024/10/31/nate-silver-accuses-pollsters-manipulation He has a really good point that if polls have a 3% accuracy then why are virtually all polls calling a 1% split? There should be more divergent results and it is fishy that there wasn't. As it turns out Trump is winning by 3% points in the general vote. He is winning PA and GA by 2% while WI and MI are at 1%. Polls should have picked that up.
As Nate Silver said, polls should diverge more. Given the actual results we should have had as many polls calling for a 5% Trump advantage as those calling for a 1% tie.
Not polling but a breakdown of the split ticket in Wisconsin. About 56,000 more people voted for Trump than the Republican Senator, while Harris and the Democrat got the same. CNN comentator says that likely many Trump voters just voted for President and not for down ballot races. The counties with the biggest gap, were counties with larger Latino population, but that is speculation that it was Latinos doing that. But it could be young Latinos that like Trump but do not like Republicans (or Democrats) https://www.threads.net/@nbcnews/post/DCTFaAGhLjI?xmt=AQGzKa3NdZoCZBlBPiOw4XT4jevHG1bEcG_QxzoLB6gNYw
This thread is stupid. Polling is stupid. At this point pollsters do far more harm than illumination. They should be banned by law. It's clear now that there was massive herding in all the polls because the pollsters reach so few contributors that they can't actually do their jobs with any measure of accuracy. But to keep the money rolling in, the pollsters release what is, at best, complete guesswork. And more often than not their results are NOT complete guesswork, but rather the results are nothing more than absolute propaganda, paid by the political parties in benefit to the political parties. In the past, as the election drew near the pollsters could at least take an educated guess which way the wind was blowing and follow up the complete bullshit by getting into the same zip code as the result in an attempt to maintain their reputation. They can't even do that anymore. In my home congressional district, the Republican candidate hired Republican pollsters who (surprise!) showed him winning by 8 points and the Democratic candidate hired Democratic pollsters who (surprise!) showed him winning by 7 points. It was ultimately a 2-point win. How is this useful in any way, other than lining the pockets of the Polling-Industrial Complex? If one Exxon-funded climate scientist wrote a paper that sea levels would fall one meter, another Greenpeace-funded climate scientist wrote another paper that sea levels would rise 10 meters, and a third "neutral" climate scientist wrote a paper that sea levels would rise by somewhere between -1 and 10 meters, climate science would rightly be seen as a fücking joke. Yet polling does this every goddamned election and gets treated with respect. I wish I could live in a world where there was an Ignore List for any news article that mentions polling in any way. But there isn't. So it should be banned by law. Fraud is fraud. Thanks for listening to my Ted Talk.
None of the states were shocking. 538 said that the outcome was what they considered to be the most likely outcome, albeit it was only 6 percent. The mean and median had Harris with more electoral votes than that, but if you only care about the most likely outcome, they were right. You could argue that Trump winning some states by over 11 is more surprising than him winning all seven of the tossup states. Trump became the first Republican to win Florida, Iowa, Ohio, and Texas by over 11 since Nixon in 1972. It is impossible to ignore polls and what media says about them and act like you have no idea who will win until votes are counted.
Article in NPR about how Trump dipped below 50% with 96% or so of the vote in. And a couple of interesting graphs that are not formatted for easy copy-and-paste. Higher turnout seemed to favor Dems, but all the competitive states besides AZ had normal-good turnout and still went Trump. And it reiterates how 2020 was an aberration in turnout.
A Republican linked to an AtlasIntel poll released the day before Election Day that was very accurate. You can go to https://www.atlasintel.org/polls/general-release-polls and download it. It downloaded it without giving me an option of reading it without downloading. I read that websites where experts like Sabato's Crystal Ball give ratings of safe, likely, lean, or tossup predicted better than computer models.
This is not about polling, but it is about why another predictive method turned out wrong. https://abcnews.go.com/538/americas-swing-2024-wide-deep/story?id=116639076 says that every state got more Republican than in 2024. The state with the least change was Washington, which had its Democrat win decrease from 19 to 18. Since Washington has one primary for all candidates (called "top-two"), the Washington primaries are used to predict general elections, and the Washington primary predicted a better Democrat year than 2020 and 2022. Nobody knew it at the time, but it turned out that using Washington's primary voters was predicting based on people who changed much less than the 6 percent the country moved in. To go along with Washington not getting much farther right, District 3 Democrat Marie Glusenkamp Perez was the only Democrat in the primary, and she got 5.6 percent less than the Republican combined to get. Then she won the general election by 3.8 percent, for a swing of 9.4 percent from the primary.
One common topic I keep hearing is that the Republican strategy was to specifically target low propensity voters. Those guys don't vote in primaries.
Ann Selzer of Iowa fame says she’s getting out of electoral polling following baseless accusations of voter fraud by Trump and others. https://www.thedailybeast.com/pollster-ann-selzer-mystified-by-backlash-to-iowa-prediction/
Somethings they got right, but many of us didn't want to hear it. 1869054244344717579 is not a valid tweet id