No doubt. I had not examined the NY Times/Siena PA poll, because it was clearly an outlier with all the other PA polls coming in tied or Trump +1. So that makes sense. I was satisfied with NY Times/Siena showing Harris +1 nationally, because frankly, that's a decisive loss for her in the EC and that's assuming Trump doesn't outperform the poll. Georgia has gone hard Trump since the debate. Arizona as well. Even if she wins PA, she's teetering so close in WI and MI and Trump always fights hard down the stretch. I think Trump will sweep WI, MI and PA. Her lead (if any) is so razor thin and WI and MI have a history of severely undersampling Trump voters.
You did not presume correctly, in the "Methodology " at the end it becomes clearer that there were two separate polls. Also they adjusted the Philly corcern. "The margin of sampling error for the national poll is plus or minus 3 percentage points for the likely electorate. For the Pennsylvania poll, the margin of sampling error is plus or minus 3.8 percentage points for the likely electorate. Among registered voters, the margin of error is plus or minus 2.8 percentage points nationally and plus or minus 3.6 percentage points in Pennsylvania. The Pennsylvania and Philadelphia polls were funded by a grant from the Lenfest Institute for Journalism. The polls, which were designed and conducted independently from the institute, were weighted so that in the end, respondents in Philadelphia represent the proper proportion of the poll of Pennsylvania. Similarly, the national poll was weighted so that respondents from Pennsylvania represent the proper proportion of the country."
That is a really interesting article. You see a lot of “person who predicted X now predicts Y” stories, and for the most part these can be disregarded, as individuals say all kinds of crap and are bound to hit on a prediction once in a while. What makes me take this guy a bit more seriously is that he is focused on what is probably our best predictor of any event: prediction markets. Still, I would urge some caution with this model. Again, last cycle could have been luck. It is definitely possible to reliably beat a prediction market, but very few have done it and usually only for a short period. The challenge in beating a market is that the market subsumes all other information. So if this guy’s model is that good, the market will actually end up altering its price based on his predictions. This could cause an odd feedback loop which untethers his model from reality, as his model is based on the market prices that his model is influencing. I’ll be curious to see how his predictions hold up this cycle.
This is also a really interesting article! It seems that some are focused on the Pennsylvania result and others on the nationwide result, but taken together they suggest something odd. Here’s Cohn: Now let’s consider our puzzle: a clear lead for Ms. Harris in Pennsylvania, but a tie nationwide? This is unexpected. Four years ago, President Biden won the national vote by 4.5 percentage points, but won Pennsylvania by just 1.2 points. Similarly, our poll averages have shown Ms. Harris doing better nationwide than in Pennsylvania. This poll is nearly the opposite. Cohn’s (and my) first thought is that the result is an artifact, but he finds reason that it shouldn’t be dismissed so summarily: This is becoming a trend among high-quality pollsters. Yes, our poll averageshows Ms. Harris doing better nationally than in Pennsylvania, but if you focus only on higher-quality polls (which we call “select pollsters” in our table), the story is a bit different. Over the last month, a lot of these polls show Ms. Harris doing relatively poorly nationwide, but doing well in the Northern battleground states. Finally, Cohn considers an interesting hypothesis to square this circle: that the Republican electoral advantage is shrinking: What’s clear is that recent results from higher-quality polls are very different from those of the last presidential election. If true, it would suggest that Mr. Trump’s advantage in the Electoral College, relative to the popular vote, has declined significantly since 2020. This wouldn’t come from nowhere: Almost exactly one year ago, I wrote that there were signs that Mr. Trump’s Electoral College advantage edge was fading, including in the 2022 midterm elections. In fact, today’s poll result is reminiscent of our polling ahead of the midterms, which found Republicans leading nationally but Democrats running strong in Pennsylvania and other battlegrounds. It was hard to believe given recent history — I didn’t believe it, and neither did others pollsters I spoke with — but it turned out to be right. Now I’m not sure this is what is happening, but I have long argued one thing that seems overlooked by right leaning supporters of the electoral college: there is no inherent reason that the EC should favor republicans. There will likely come a day in the future where the democrats enjoy the EC advantage, and republicans will need to win the popular vote to remain competitive. I will give them credit if on that day they still defend the EC so staunchly, but I’m skeptical that that will be the case.
I'm curious about what will move the polls from this point on. We're looking at undecided and primarily independents. Seems the debate helped Kamala a bit. Not sure how much your uncommitted rank and file voters feel about the fed cut, if anything. They need to feel an improving economy, not just read about the mechanism beneath it. Otherwise, it's the words and actions of each candidate that will move the needle. I know of one group that's feeling good about the economy. Yesterday, Amazon's warehouse workers got a $1.50 and hour raise plus free Prime. That's nearly a million people across every US state. There seems to be a growing comfort with Harris but she's still not connecting well with men. At the same time, Trump can't help himself from alienating women. I can see him losing some tenths of a point in the wake of the Taylor Swift debacle. Also have to wonder about those 300k Swifties who got the voter registration link.
Once again. Another reminder that 2024 isn't 2020. A Republican hasn't won a statewide election in Michigan or Pennsylvania since 2016. In the 2022 Michigan gubernatorial election the Trump-endorsed MAGA election denier Tudor Dixon was within the margin of error according to most polls. She lost to Gretchen Whitmer by double digits and not only that but for the first time decades Democrats held every statewide office and both houses of the Michigan state legislature. In Wisconsin only one Republican candidate has won a statewide election (Ron Johnson) with a margin of less than one percent underperforming the polls. In a 2024 nominally nonpartisan election for a seat on the Wisconsin Supreme Court the liberal candidate beat a conservative former justice by a double-digit margin when the most significant issue was whether the court would upheld a century old statute outlawing all abortions. In the 2022 midterms the Democratic candidates in Pennsylvania beat the Trump-endorsed Republicans by significant margins. While I would concede that the NY Times/Siena poll had the national race essentially tied if I recall it was you who said that national polls do not really matter, it's the swing states that count (one of the few times that you were correct by the way).
Calling BS. I read what they did as including the Pennsylvania voters in the total national polling. No need to adjust something 3.8% while saying Harris leads by 4 in Pennsylvania which you know is straight up bullshit. Poll it right to start with and say it's tied in Pennsylvania. Why they didn't? Wanted the headline that Harris leads by 4.
Florida is looking real interesting the last two days… it’s passed up both Wisconsin and Michigan as more likely to flip on 538. They have it +2.9% for Trump, basically a MOE state now.
I doubt Florida actually flips blue, but encouraging for the future. Residents in this state have seen what Republican leadership has gotten them, and they’re not impressed.
Yeah, Florida polling has been terribly off the last two presidential cycles, so not sure what to make of it.
When the quote clearly says "The margin of sampling error for the national poll is plus or minus 3 percentage points for the likely electorate." and then proceeds to state the other poll (Pa,) : " For the Pennsylvania poll, the margin of sampling error is plus or minus 3.8 percentage points for the likely electorate." then you ascertain there are two distinct polls with different margins. Your desperation notwithstanding, they adjusted the sampling to the Philly polling and the felon is a loser.
So I performed my own independent survey while cycling almost 500 miles over the last 3 weeks and based upon the yard signs I counted, Kam Harris will come in third place behind Trump and Yard Sale.
I don't think anybody following this race from a non-partisan standpoint thinks Harris is up by 4 in PA. Not to mention their national poll has Harris losing the race, because if she wins the popular by 1, she's losing the race.
While the 4 may be on the high side even if she wins by 2 or only one she still gets every electoral vote from Pennsylvania. By the way Quinnipiac which is also a highly rated poll has her up by 5 with a 2.7% margin of error. https://poll.qu.edu/images/polling/mi/mi09182024_swmf64.pdf
