sushihorn

The direct comparisons are infrequent but here is one: https://www.cbsnews.com/news/trump-biden-economy-gdp-wages-inflation/ You can infer from right track, wrong track polling since the economy is the voters’ biggest concern: https://www.realclearpolling.com/polls/state-of-the-union/direction-of-country

I can’t find educational attainment stats for registered voters so I used population data that was available. I understand your argument for higher voting propensity; I doubt it can double that group’s representation in an electorate skewing rural. I’m using 2020 final as a smell test for poll’s demographic breakdown. I think that is the best case scenario for the NYT getting close on their demographics. It looks implausible. I introduce the 2024 EV so I can see what the remaining votes would have to look like for the poll to be close to the 2024 final electorate. I see that I did not explicitly state my reasoning and I apologize for that oversight.

That’s not how the majority of Americans see it. They think they were better off under Trump than Biden by 20 point margins. This where Maslow’s Hierarchy comes in. Most of us easily meet our survival needs. We are not living paycheck to paycheck, for the most part. That allows us to pursue self-actualization like philosophical pursuits or social justice. But please realize that we are not representative of the country or even of college grads as a whole. We are TEXAS; that means we are not like other people or other colleges. Even within flagship public universities, we are right near the top. That means we are elite, with elitist perspectives that we have to struggle to move beyond when analyzing others.

NY Times released a Cruz +4 poll this morning. Their cross tabs are screwed though. There are 2 large, fundamental errors: whites are under sampled and college is oversampled. They have a sample that is 55% white and 42% non-white. The 2020 electorate was 62.4% white the 2024 EV so far is 68% white. Their poll has 26% with a BA and 16% advanced degrees. Texas is actually 18 and 9% in those categories. There is also a large tactical issue. The people actually voting skew much older than their sample. NYT has 25% of the sample from seniors (65+). The actual EV is 37.2% from this group. The sample is 15% from the 18-29 cohort. Current EV is 8.5% young adults. This poll is describing an electorate, but it’s not the Texas electorate. It’s certainly nothing like what we are seeing this year. They will be wrong on race and education. Age might change as more votes come in but we are close to 40% of the total votes already cast. I don’t think that it can move enough to look like their sample. The remaining electorate would have to be <15% seniors citizens and >20% young adults for that to happen.

This is where the internal conflict in the GOP becomes critical. I doubt someone like McCain could hold them. No way Mitt Romney could. JD Vance might accelerate the trend. Trump is a billionaire who can speak to the working class. Vance actually comes from the working class. So add cultural affinity to the protectionist policies. OTOH, I have no idea if Vance can match Trump’s appeal to the non-white blue collar voter. He not an icon of success in business, nor has he had a high pop culture profile for 30+ years.

Historically, black turnout lags white turnout by 7-8 points. Obama actually managed to get black turnout higher than white turnout. We have seen movement back towards prior trends since he left the WH. Ten points higher turnout in 13% of the population with an 80% margin covers a multitude of sins.

I’m a statistical analyst. ToMayto, ToMahto.

White voters have flipped party affiliation by class. Obama won easily by gaining college educated white voters while losing fewer working class whites and then adding high turnout with enthusiastic black voters on top of that. Trump really juiced the reverse flow of white blue collar voters to the GOP. That’s not a good trade for the Democrats since there are more white voters without a college degree than those with one. It’s an especially bad trade in the Midwest. Look at what’s happened to Iowa and Ohio. Those used to be swing states.

There’s still a chance that the electorate will change significantly. I expect changes but not enough to alter the trend. The underlying issue is that the partisan composition of the electorate seems to be drifting R in many states where voters register by party. In 2004, FL had a D+400k registration advantage. Today that is R+1M. The Democrat’s registration advantage in PA is half what it was in 2020 and a quarter of the difference from 2012. Ruy Texiera’s axiom that demographics is destiny assumes that demographic change is transformative. Well, the PA and WI electorates aren’t cooperating. PA EV is well north of 80% white and WI is over 90% white. It also assumes racial identity = political identity. That seems to be under challenge as well with working class minorities drifting away from the Obama coalition. I don’t expect a paradigm shift but class is become more influential and race less with regard to voting behavior.

It’s far from ideal but it’s what we have. Can you see a better alternative?

Incumbents are usually vulnerable when they can’t hit 50% and the challenger is within the margin of error. Emerson shows Cruz under 50 with a one point advantage. Prior polls show 50 or higher and significant margins. So we either have a late breaking race or an outlier poll. Not enough recent data to tell. The issue for the 4 Democrat incumbents in the Great Lakes states is that all of the are polling under 50% consistently. Baldwin in WI has been under 50 in the last 8 polls and half of them have her between 46 and 48. I think Brown in OH hit 50 in one out of the last 8 polls. 6 of those show his support under 48%. With the advantages of incumbency, under 48 is red flashing lights with sirens thrown in. Voters know the incumbent. If half of them can’t support the devil they know, most of them will break for the challenger.

Trump was not on the ballot in 2022. The electorate in 2022 was almost R+3 based on votes for House candidates. R+3 this year probably would result in an electoral landslide at the top of the ticket, which would make this discussion moot. I think Nate Silver models over 90% chance of a Trump victory if the popular vote splits 50/50. R+3 is something like 330 electoral votes. In a few demos, the early electorate is even more R-friendly than 2022. For example, NC went from 45.4% rural EV in 2020 to 49.2% in 2022 to 51.0% as of yesterday. However that is the exception. Most slices are more R biased than 2020 but less so than 2022.

Baseline is 2020, not 2022. But in some cases this looks more like a midterm electorate than any recent presidential year. The polls depend on representative population samples and the early voters so far look nothing like what most pollsters are modeling.

I’m going to say that the trends I’m seeing across a lot of states match the much more detailed reports from Ralston in NV. Rural turnout is massive. Urban turnout is low and Clark County is only a few points +D. Racial minorities are not showing up in large numbers. Race is a minor factor in AZ and WI. But in PA, black vote share is down by a third and that matters. There is a small in black vote share in MI but NC and GA have large black populations, which are the Democrat core in those states. A 3.4 point decline in NC is potentially decisive. I did a thought experiment: if a group that traditionally favors one party heavily starts to shift away, how would that manifest? Votes share shift would not be visible until ballots are counted. But partisan slippage would also manifest as non-participation to a degree. Would rural turnout increase dramatically, only to shift towards Democrats? It seems unlikely. What I would expect is that as some votes shift, others would stay home rather than vote D instead of R. So which demos are showing this pattern of declining participation? Young voters, non-white voters, urban voters and to a small extent, female voters are. 2020 is a terrible year to use for a baseline due to Covid and mass mail, but it’s what we have. We can build models but we also need real world historical data as a reality check. There is still time to change the trend, but the total EV is getting pretty large and that task is growing more difficult. If Ralston is estimating turnout correctly, 40-45% of the final total votes have already been cast.

https://targetearly.targetsmart.com/g2024?count_prefix=current_eav_voted_count_&demo_filters=[{"key"%3A"urbanicity"%2C"value"%3A"All"}]&state=TX&view_type=state You may have to set up the comparison years but you can access state data by various demographic slices. I don’t fully understand or trust their modeled party data so I don’t use it but known demographics like age and race aren’t malleable in modeling.

The issue is not whether more women are voting than men; they are. The issue is are women a larger proportion of the EV electorate than they were in 2020; they are not. In Georgia, the major demographic slices all are skewing more R than the 2020 electorate. The 2024 vote so far is significantly whiter, older and more rural than 10 days out from the 2020 election. That is true across many states. In Texas, the black vote share is stable but Hispanic share is down and white share is up. In Georgia, white share is up and black share is down almost 3 points. Rural share is up and urban down pretty much everywhere. It’s a 9-10 point shift in Texas and 5-6 in GA. 65+ voters are a larger share in Texas this year by 6.6 points. 50-64 is flat to slightly down. Every age group under 50 is a smaller share this year. That dynamic is also playing out widely. In the Georgia example, it’s a 5.4 point shift towards the olds. While gender shifted 1-2 points from female to male in 4 years, it’s probably less significant than the other demographic shifts in the EV electorate. If I told you the 2024 votes would be much older and more rural, significantly whiter and slightly more male than 2020, would you think those groups are turning out to vote enthusiastically for Kamala and the Democrats? Not sure I buy that.

Valmy, there’s no need to account for uncontested seats; they’ve always existed and always will. Unless there was an enormous and partisan increase in such races, it’s not going to be the cause of significant change.

MAGA in large part is a continuation of the Tea Party. Abolitionists were a minor party (Free Soil) until they allied with economic nationalists to form the GOP. Similarly, the Tea Party could tilt legislative elections (2010, 2014) but couldn’t unseat Obama. Adding economic nationalism created MAGA and elected a president. I see no indication it’s going away. We can certainly revisit in a few weeks.

A lot of people are sick of politics. That same phenomenon is the reason polling has become much less reliable. 5% is a good response rate these days. There’s a lot of room for non-response bias when 95% refuse to respond.

GOP total votes for US House increased 7% from 2018 to 2022: 50.8 million to 54.2 million. Democrat total House votes fell by 15% in that same time frame. So overall turnout dropped but the impact was confined to the Dems. I’m not sure how that could be interpreted as a rejection of Trump or the Rs.

That statement doesn’t seem to be supported by the numbers. The 2022 electorate was slightly more Republican than polls indicated: R+2.8 actual vs R+2.5 polling average. That translated into fewer seats than pundits expected because their advantage was not efficiently distributed. That was analogous to the popular vote vs Electoral College.

Bama, if your theory is correct you should be looking at changes in party registration for clues on how the new voters will trend. The moves in party affiliation is one thing that has Ralston concerned in NV. IIRC correctly, Washoe County (Reno) went from 50/50 to R+4 or 5 since 2020.

Mail in skews D and IPEV skews R everywhere I’ve seen. Historically, ED skews R relative to EV. We’re seeing voter behavior change this cycle so there’s little basis to determine how much of the R strength in EV reflects turnout of low propensity and new voter vs time shift of existing voters that will cannibalize ED turnout to a large degree. The NV blog posted stuff tonight suggesting there is a big partisan gap in turnout by the low propensity vote. OTOH the pro GOP EV gap did shrink a bit today.

It represents 20% cumulative turnout. We are analyzing the data we have, not a complete data set or even a representative sample, not yet.

Georgia EV by gender Overall vote trends female, EV even more so. This year’s EV shows a female bias but less so than historically. This seems to be true across a wide swath of states.