Same Batchannel — analysis courtesy of Washington Post, however:
The Washington Post has a battleground state polling average that makes a significant effort to exclude the right-aligned polls. It shows Harris in a much stronger position than some of the other forecasters and aggregators. I’ll take this map for I still believe we are going to win AZ and NC.
I know some here believe Simon Rosenberg is full of it … but I think his takes are being misunderstood (and, yes, maybe goosed a bit by posters like myself reporting second-hand). See his next paragraph below — nothing here is unreasonable:
The bottom line my friends is all that polling can tell us now is that the election is really close, within margin of error everywhere. It is close now and will be close on Election Day. I will keep reporting on what I see but all that matters now is that we put our heads down and go out win this thing, together, for our democracy, our freedoms and our future!
The WaPo aggregation is interesting. Reading up on their methodology there are two major differences between their aggregation and “the others”: one is indeed that they use only the very select best polling houses (not excluding partisan only, excluding others that don’t meet all their requirements); the other is that the others have more recency weighting, while they are intentionally slow to move to polling changes.
Not sure which explains the differences such as they are.
Compared to 538 they are the same for AZ, GA, and NV. Essentially the same NC. The big differences are the Blue Wall states, basically even on 538 and all H+2 for WaPo. NYT aggregation very similar to 538 except leaning slightly H<1 in NC.
I’m looking at PA and biggest impact I see is 538 and NYT include AtlasIntel as a quality pollster which gave a T+3 recently, weighted heavily for recency and quality. WaPo excludes them. They are not on the list of alleged Red Flood pollsters.
Some new polls showing up for MI. Bullfinch group is probably off, as it gives Harris +8. She could, though, end up at +4 to 5. The Atlas group poll in the list gets poor “transparency” rating from 538. I take that to mean they do not reveal methods.Michigan : President: general election : 2024 Polls | FiveThirtyEight
This is also likely the reason the Washington Post excludes AtlasIntel from their aggregate.
Some on liberal Substack do their own aggregate with a ground rule that excludes all overseas polling outfits. This also excludes the Brazil-based AtlasIntel.
Now he believes that the aggregations saying it is toss up (or “close”, which is not actually quite the same thing) are pretty much right.
But previously he was saying they were successfully “gamed” in 2022 (weren’t, were scarily closer than usual) and not to trust them. That NC aggregations were successfully being gamed when the aggregate results were the same when only looking at the few top level pollsters.
Partisanship and optimism does not have to include delusionality.
ETA yes those are good reasons to exclude AtlasIntel, but they do not define them as Red Wave biased pollsters. They are also rated as minimally biased by 538 and have, unfortunately, been historically very accurate. I’m willing to personally believe they are wrong this time and their outsized impact on the results in the Blue Wall state aggregations IS problematic. Unless they’re right.
I think we are getting close to the point where Democrats are going to need a “typical” polling error in their favor for Harris to win. She just can’t win the EC if she only wins the national vote by 1-2%, barring a pretty major realignment in states like NC and GA relative to national averages.
Historically, this type of error has happened, and has happened in the Democrats favor in many of the primaries and ballot initiatives leading up this election. But, with Trump on the ticket, the error has tended to be in the other direction.
We are certainly past the days when polls were providing comfort and optimism to Harris supporters.
IDK it feels like since Harris has been in the race it’s been somewhere between 60-40 one way or the other at least according to the poll-based forecasts, and these relatively small changes in polling are nothing but rohrsharch tests for peoples biases one way or the other.
It’s hard for me to believe that anything within an average polling error could be shifting the odds that little (eg if we webt from “Trump needs at least a standard polling error to win” to “Harris needs at least a standard polling error”. Is something so simple like national polling narrowing meaning that the southern states must be probable trump states something the experts just can’t see?
Yup, whether its a 54% chance of a Harris win or a 54% chance of a Trump win according to the polling has absolutely no real significance. Its a coin flip that is going to be decided by which of those polled in these surveys actually vote.
I am cautiously optimistic that Harris will have an advantage there. I’m hoping that the MAGA tendency to replace the old guard who actually knew how to run a campaign with whose sole qualifier is loyalty to Trump’s delusions will allow reality to kick them in the pants.
But this optimism has no empirical evidence to back it up so I’m still expect to be biting my nails until mid November.
Well she does in the WaPo aggregation referenced above. Up by two nationally but sweeping the Blue Wall and winning NV albeit barely. Losing NC also barely. Which is a win.
The Republican EC advantage may not be as large this time.
WP is a paywall site, you can look for the poll. In Threads, a summary, via Newsweek (mostly a clickbait site, link to Threads below):
The just released large Washington Post Poll (5000+ voters) shows good news for Harris. If accurate, she would win with 286 Electoral College votes. GA: Harris up 51-49, MI: Harris up 49-47, PA: Harris up 49-47, WI: Harris up 50-47. Nevada: Tied 48-48. Trump up in North Carolina - 50-47, Trump up in Arizona - 49-46. More in comments.
Right, this is one of the unknowns. The theory is that losing some Black and Hispanic voters (largely men) but gaining lots of well-educated suburban white women might be a good tradeoff from at EC perspective (they are more evenly spaced across multiple states). But that depends on holding enough of the Black vote in places like Detroit and Philly.
I personally would be pretty shocked if she wins the EC with only a 1-2% national win, but it could happen.
If Americans would all stay home and not vote and let Canadians do the voting for them, Harris would win in a landslide. In a “trivia poll” in Canada amongst presumably 100% ineligible voters in the US election, Harris would garner 64% of the popular vote, Trump with 21% and undecided with 15%.
Something I currently thinking about the polling. The models pollsters use to predict who’s voting may vary in accuracy over the course of a campaign. That is, a model that was perfect in September, may be wrong in October. And the reverse. Because not only are the electorate and their responses to polling changing, but also the available information is different.
Besides statistical noise, changes in the polls could be changes in polling accuracy and changes in the models. And of course changes in people’s voting intentions.
It is IMHO even likely that reality of who intends to vote for whom is essentially unchanged from very soon after Harris became the nominee. The small (but in the case of polling being so close in the group of possible tipping point states, impactful) polling changes, are changes in the mix of even high quality polls, and changes in models, than actual changes.
Knowing who has not been reached in polls, who is actually going to vote, so on? End of day they have to make best guesses. And they really don’t want to make the same mistake they made last time. Better a brand new one!
We remain with a prediction based on past cycles of the likely tipping point states being anywhere from roughly 54 to 46 one way to the other and combinations in between that account for third party shares of anywhere from 1 to 4%.
Odds are best I think of results not far off from 2016 and 2020 with the key question being exactly where the votes are. Trump’s past topping off at about 47% seems like a solid pattern. He’s won with that before and he could this time.
Staring at the heading of this thread long enough conjured up a song parody: Polling has broken,
Like the first polling,
Back when poor people,
Had no home phones.
Polling has broken,