Not surprising. A lot of assumptions were made post election about those who shifted right. Those assumptions are being systematically disproven, and this data set seems likely to contribute to that effort.
I think Strength In Numbers is right when it states that most "swing" voters are non-ideological and go where the grass seems greenest for them. I will add that such voters might "swing" faster or harder if the grass behind them is also on fire.
This is all to say that "new republicans" weren't necessarily republican at all. They might have been republican curious in 2024, thinking the grass was greener on the GOP side. And they might have moved a bit harder to the right than they would have otherwise, because the democrats have been a collective dumpster fire for the entire Trump era.
But now ... well, the grass don't look so green anymore. So I'm not so sure how "republican" these "new republicans" really are -- and I certainly don't think they'll stay that way for long, whatever the case.
But I could be wrong, so I am also looking forward to more info. I guess we'll see.
Fascinating breakdown of the attitudinal rifts within the GOP coalition. The 29% New Entrant segment being substantialy more progressive yet voting Trump speaks to how much 2024 was a referendum on economic perceptions rather than ideological alignment. What strikes me is the temporal vulnerability here: if these voters swung once based on affordability concenrs, they're likely to swing again if the economy doesn't deliver tangible improvements by 2026. The Manhattan Institute's finding that only 56% would "definitely" vote Republican again isn't just soft support, it's a coalition held together by dissatisfaction with the alternative rather than commitment to the platform.
Love the 'Stack ~ will for sure love 😍 w/💵 as soon as I can. QUESTION for future polls? How open are "change" voters to giving their change agent(s) more than one term to fix things? Especially among infrequent voters, will they be willing to vote change in '26/'28 AND THEN COMMIT TO GIVING MORE THAN 2/4/6 YEARS TO CLEAN THIS MESS UP?
Like, the Biden economic universe was sound, and was working quickly IN ECONOMIC TERMS, but not on the speed scale required for a fickle electorate. I feel like messaging in the current cycles needs to prime people for the fact that un-fucking all this is going to be hard and it may be slow.
Another large group of voters in 2026 could be from non-white turnout. The UF Election Lab data shows all non-white ethnic groups voting at higher rates in 2018, than in 2022. Seems like the backlash to Trump could be even larger next November.
Always strange to me that every analyst will have reasons the Democrats lost, but they never want to discuss the elephant in the room. Affordability and inflation, of course. But since the majority of the largest voting population in the US, white people, have never cast a presidential vote for anyone other than a white male, it seems the fact that Kamala was a mixed race female is a key data point.
That is the case for white people, but even more so for white men. They voted for Trump by 62% in 2016, 57% in 2020, and 59% in 2024. Since this group does experience harm by loss of status, messaging that emphasizes a good living for everybody might help.
Oh, but a LOT of those white dudes genuinely feel that they are losing status in empirical terms, and that someone can reverse that so that's the whole MAGA premise right there.
Interesting essay. As a resident of a red state, I’m more used to the assumption Republicans live in a right-wing media world and would never consider voting for a Democrat. I’d like to learn more about these “new” Republicans.
I would appreciate an article talking about the polling methods of the various pollsters. In such a fragmented electronic environment, who are the people giving their opinions that polls are based on, and how are their opinions solicited? For instance, I don't respond to those texts, emails, phone calls, that ask me to participate in a survey or a poll. The 'sponsors' of the poll are never revealed by the solicitors. So who knows who is funding the opinion gathering
Someone in that org sent me their Substack post on TN, and it seems to hinge on the assumption that turnout and vote share should not be correlated at the geographic level. I do not think that is correct, since we observe different turnout rates and vote loyalties for different groups, and those groups can be clustered geographically. Their key evidence is that higher-turnout precints in Montgomery County, TN went more for the Republican candidate. I reckon the voters _within_ those low-turnout, high Dem precincts just had high shares of Black voters.
I'd prefer to see an analysis based on changes in vote share and turnout rate, since I imagine these patterns were also present in the 2020 and 2024 elections.
I believe that was addressed somewhere: what has been happening in these suspicious events has been compared to prior elections with a thorough analysis each time and charts explaining why the vote distribution seems to be unlike what historically happened. the really interesting thing is how the Dem was winning until the alleged tampering began and then votes split in a rather symmetrical way, with the same percentage of votes being subtracted from the Dem while being added to the Republican.
Keep in mind that those at ETA are forensic statisticians, not conspiracy theorists. I hope you will read thoroughly what they've written and re-read anything prior. These alleged manipulations echo what has happened in overseas elections that were CONFIRMED to have experienced Russian interference and we KNOW the Russians have and continue to interfere in OURS.
I really do think the whole thing is based on flawed assumptions by the "forensic statisticians" involved. Eg, in this case, the reason Dems were ahead in TN-07 early and then lost vote share as the count went on was that the Dem won among early voters, which were reported first. We have seen this in previous elections; in 2020, for example, people used the same argument to claim the election was stolen from Trump!
so, based on what ETA said in their preliminary report:
"Core Result:
Turnout is tightly and abnormally correlated with candidate support.
Matt Epps (R):
r = 0.8855, p ≈ <.001
Turnout → +15.9% increase in vote share per 10%-point increase in turnout
Aftyn Behn (D):
r = –0.8807, p ≈ <.001
Turnout → –15.6% decrease in vote share per 10%-point increase in turnout
Interpretation:
In normal elections, vote share is expected to remain relatively stable across turnout levels. The “cross-pattern” found here—where higher turnout consistently boosts one candidate’s vote share while reducing the other’s—matches known patterns associated with turnout inflation or manipulation. Additionally, the effect size of +15–16 points per 10%-point increase in turnout far exceeds what is expected from human voter behavior."
what is "flawed" about what they found? also from their Substack report:
"Again, a clean election is expected to have a relatively consistent trend across all turnout thresholds, and abnormal results will have aggressive vote-share shifts from low to high turnout."
and
"Statistical patterns in Montgomery County show large, uniform shifts in vote share as turnout increases—anomalies associated with ballot-stuffing or digital allocation in established election forensic literature (Shpilkin, Udot, Klimek).
While statistical results alone do not prove manipulation, such large and consistent effects across multiple precincts require independent verification."
From ETA's website on the MN results in 2024:
"As part of routine election operations, a small number of Minnesota voting precincts perform hand counts instead of machine-counting paper ballots to determine how many votes were cast for each candidate. The Election Truth Alliance (ETA) conducted a detailed analysis of hand-counted (“hand count”) versus machine-counted (“machine count”) precinct data in Minnesota. Statistical ‘election integrity red flags’ present in machine count precincts were not present in hand count precincts."
Please explain to me like I'm 5 why their findings are flawed and why a true forensic audit is not warranted. It seems to me that a clear pattern is being found to add to the clear patterns of OTHER voter interference by the GOP (illegal removal of Democratic voters from the rolls with no notification, voter challenges of Black voters AT the polls forcing them to vote provisional ballot and then not counting those ballots, the continuing assaults on the rights of citizens overseas to vote by mail...) to install one party rule over the entire nation. We can't afford to think that the GOP will somehow find loyalty to the Constitution instead of to their Orange anti Christ before the midterms.
I’m curious about the reports of republicans who have “soft” support for the party. Where I live, for example, there’s a decent number (maybe it’s insignificant) of people who are registered republicans because they view that party as the ruling party and they think their vote has more impact. As state legislatures (for the most part?) seem to be more conservative these days, do you think that’s contributing to the rise of these “soft republicans.” And, secondly, is there any data on how much public sentiment swings in the month (or so?) before an election? I guess I’m asking because we get a lot of polling data for years?/several months leading up to an election but it seems like it can all change rapidly in the 3-4 weeks prior to the actual elections? I guess that’s human nature, but just wondering if there are any studies along those lines?
There's a great question hidden in here which is essentially "When do generic ballot polls become predictive?" and that's a great idea for a quick article in January!
YES! And, can we quantify the "coming home" effect of people who have been polling as unhappy w/their usual choice, but vote for them anyway once they are standing there with a ballot?
Not surprising. A lot of assumptions were made post election about those who shifted right. Those assumptions are being systematically disproven, and this data set seems likely to contribute to that effort.
I think Strength In Numbers is right when it states that most "swing" voters are non-ideological and go where the grass seems greenest for them. I will add that such voters might "swing" faster or harder if the grass behind them is also on fire.
This is all to say that "new republicans" weren't necessarily republican at all. They might have been republican curious in 2024, thinking the grass was greener on the GOP side. And they might have moved a bit harder to the right than they would have otherwise, because the democrats have been a collective dumpster fire for the entire Trump era.
But now ... well, the grass don't look so green anymore. So I'm not so sure how "republican" these "new republicans" really are -- and I certainly don't think they'll stay that way for long, whatever the case.
But I could be wrong, so I am also looking forward to more info. I guess we'll see.
Thanks for a great post, SIN
Fascinating breakdown of the attitudinal rifts within the GOP coalition. The 29% New Entrant segment being substantialy more progressive yet voting Trump speaks to how much 2024 was a referendum on economic perceptions rather than ideological alignment. What strikes me is the temporal vulnerability here: if these voters swung once based on affordability concenrs, they're likely to swing again if the economy doesn't deliver tangible improvements by 2026. The Manhattan Institute's finding that only 56% would "definitely" vote Republican again isn't just soft support, it's a coalition held together by dissatisfaction with the alternative rather than commitment to the platform.
Comment on NYT survey a few weeks ago of Fed judges on recent rulings by the US Supreme Court?
Love the 'Stack ~ will for sure love 😍 w/💵 as soon as I can. QUESTION for future polls? How open are "change" voters to giving their change agent(s) more than one term to fix things? Especially among infrequent voters, will they be willing to vote change in '26/'28 AND THEN COMMIT TO GIVING MORE THAN 2/4/6 YEARS TO CLEAN THIS MESS UP?
Like, the Biden economic universe was sound, and was working quickly IN ECONOMIC TERMS, but not on the speed scale required for a fickle electorate. I feel like messaging in the current cycles needs to prime people for the fact that un-fucking all this is going to be hard and it may be slow.
Another large group of voters in 2026 could be from non-white turnout. The UF Election Lab data shows all non-white ethnic groups voting at higher rates in 2018, than in 2022. Seems like the backlash to Trump could be even larger next November.
Always strange to me that every analyst will have reasons the Democrats lost, but they never want to discuss the elephant in the room. Affordability and inflation, of course. But since the majority of the largest voting population in the US, white people, have never cast a presidential vote for anyone other than a white male, it seems the fact that Kamala was a mixed race female is a key data point.
That is the case for white people, but even more so for white men. They voted for Trump by 62% in 2016, 57% in 2020, and 59% in 2024. Since this group does experience harm by loss of status, messaging that emphasizes a good living for everybody might help.
Oh, but a LOT of those white dudes genuinely feel that they are losing status in empirical terms, and that someone can reverse that so that's the whole MAGA premise right there.
Interesting essay. As a resident of a red state, I’m more used to the assumption Republicans live in a right-wing media world and would never consider voting for a Democrat. I’d like to learn more about these “new” Republicans.
I would appreciate an article talking about the polling methods of the various pollsters. In such a fragmented electronic environment, who are the people giving their opinions that polls are based on, and how are their opinions solicited? For instance, I don't respond to those texts, emails, phone calls, that ask me to participate in a survey or a poll. The 'sponsors' of the poll are never revealed by the solicitors. So who knows who is funding the opinion gathering
we can do that!
To explain 2024 I think we need to add sexism among younger men and the weakness of the Democratic brand at the end of the Biden years
I am concerned about what ElectionTruthAlliance.org had to say about the TN race (as well as what they've been finding out about 2024's election)...
Someone in that org sent me their Substack post on TN, and it seems to hinge on the assumption that turnout and vote share should not be correlated at the geographic level. I do not think that is correct, since we observe different turnout rates and vote loyalties for different groups, and those groups can be clustered geographically. Their key evidence is that higher-turnout precints in Montgomery County, TN went more for the Republican candidate. I reckon the voters _within_ those low-turnout, high Dem precincts just had high shares of Black voters.
I'd prefer to see an analysis based on changes in vote share and turnout rate, since I imagine these patterns were also present in the 2020 and 2024 elections.
I believe that was addressed somewhere: what has been happening in these suspicious events has been compared to prior elections with a thorough analysis each time and charts explaining why the vote distribution seems to be unlike what historically happened. the really interesting thing is how the Dem was winning until the alleged tampering began and then votes split in a rather symmetrical way, with the same percentage of votes being subtracted from the Dem while being added to the Republican.
Keep in mind that those at ETA are forensic statisticians, not conspiracy theorists. I hope you will read thoroughly what they've written and re-read anything prior. These alleged manipulations echo what has happened in overseas elections that were CONFIRMED to have experienced Russian interference and we KNOW the Russians have and continue to interfere in OURS.
I really do think the whole thing is based on flawed assumptions by the "forensic statisticians" involved. Eg, in this case, the reason Dems were ahead in TN-07 early and then lost vote share as the count went on was that the Dem won among early voters, which were reported first. We have seen this in previous elections; in 2020, for example, people used the same argument to claim the election was stolen from Trump!
so, based on what ETA said in their preliminary report:
"Core Result:
Turnout is tightly and abnormally correlated with candidate support.
Matt Epps (R):
r = 0.8855, p ≈ <.001
Turnout → +15.9% increase in vote share per 10%-point increase in turnout
Aftyn Behn (D):
r = –0.8807, p ≈ <.001
Turnout → –15.6% decrease in vote share per 10%-point increase in turnout
Interpretation:
In normal elections, vote share is expected to remain relatively stable across turnout levels. The “cross-pattern” found here—where higher turnout consistently boosts one candidate’s vote share while reducing the other’s—matches known patterns associated with turnout inflation or manipulation. Additionally, the effect size of +15–16 points per 10%-point increase in turnout far exceeds what is expected from human voter behavior."
what is "flawed" about what they found? also from their Substack report:
"Again, a clean election is expected to have a relatively consistent trend across all turnout thresholds, and abnormal results will have aggressive vote-share shifts from low to high turnout."
and
"Statistical patterns in Montgomery County show large, uniform shifts in vote share as turnout increases—anomalies associated with ballot-stuffing or digital allocation in established election forensic literature (Shpilkin, Udot, Klimek).
While statistical results alone do not prove manipulation, such large and consistent effects across multiple precincts require independent verification."
From ETA's website on the MN results in 2024:
"As part of routine election operations, a small number of Minnesota voting precincts perform hand counts instead of machine-counting paper ballots to determine how many votes were cast for each candidate. The Election Truth Alliance (ETA) conducted a detailed analysis of hand-counted (“hand count”) versus machine-counted (“machine count”) precinct data in Minnesota. Statistical ‘election integrity red flags’ present in machine count precincts were not present in hand count precincts."
Please explain to me like I'm 5 why their findings are flawed and why a true forensic audit is not warranted. It seems to me that a clear pattern is being found to add to the clear patterns of OTHER voter interference by the GOP (illegal removal of Democratic voters from the rolls with no notification, voter challenges of Black voters AT the polls forcing them to vote provisional ballot and then not counting those ballots, the continuing assaults on the rights of citizens overseas to vote by mail...) to install one party rule over the entire nation. We can't afford to think that the GOP will somehow find loyalty to the Constitution instead of to their Orange anti Christ before the midterms.
I’m curious about the reports of republicans who have “soft” support for the party. Where I live, for example, there’s a decent number (maybe it’s insignificant) of people who are registered republicans because they view that party as the ruling party and they think their vote has more impact. As state legislatures (for the most part?) seem to be more conservative these days, do you think that’s contributing to the rise of these “soft republicans.” And, secondly, is there any data on how much public sentiment swings in the month (or so?) before an election? I guess I’m asking because we get a lot of polling data for years?/several months leading up to an election but it seems like it can all change rapidly in the 3-4 weeks prior to the actual elections? I guess that’s human nature, but just wondering if there are any studies along those lines?
There's a great question hidden in here which is essentially "When do generic ballot polls become predictive?" and that's a great idea for a quick article in January!
YES! And, can we quantify the "coming home" effect of people who have been polling as unhappy w/their usual choice, but vote for them anyway once they are standing there with a ballot?