Bold claim, I know, but I bring multiple receipts from 2016, 2020, 2022, and 2024 (even a 2019 Israeli election!) that were more accurate than most of the mainstream press analysis even when I was not fully on target. So here is your one-stop for the big-picture for 2024! In a race where all seven swing state averages are within the margin of error and are therefore statistical ties, is there other data we can use to predict winners under such conditions? Voter registration data, my dear readers. Also, will there be a Puerto Rican x-factor after that horrific MSG rally?? My state-by-state deep dive.
(Traduce en español/translate to Spanish) By Brian E. Frydenborg (Twitter @bfry1981, Threads @bfchugginalong, LinkedIn, Facebook, Substack with exclusive informal content) October 31, 2024; UPDATED November 5: I noted below that my Electoral College predictions are rough, but am restating today on Election Day that I believe the most likely outcome will be my “strong” for Harris prediction (3rd map); Updated November 4 to cover Florida Senate race; *UPDATED November 3rd: added discussion of Iowa’s move into the swing state category, courtesy of the peerless Ann Selzer, and Electoral College maps have been updates in turn; also slight update to voter registration modeled partisanship; Update November 1st: fixed section on Florida and Virginia with missing text and fixed NC registration graphic that was GA; see related article on this election from September 6, 2024, and one on the 2022 midterms from November 7, 2022; because of YOU, Real Context News is approaching two million all-time content views, but I still need your help, please keep sharing my work and consider also donating! Real Context News produces commissioned content for clients upon request at its discretion.
SILVER SPRING—We’re at the point in this election where we should able to start making predictions, because it is hard to believe anything dramatically different will emerge in polling in the next week, and I am here making mine (which I may or may not update). But first, some housekeeping.
Methodology
In general, a good rule of thumb is to go by weighted polling averages. One of the weaknesses of Real Clear Politics’s averages are that they basically put any poll in there without weighting for quality in an election cycle in which there have been a lot of low-quality polls using less-accurate sampling methods as well as a high number of biased partisan polls, a fair amount of both driving the discussion and most of the latter by far are from Republican-favoring sponsors (I have not been able to find an explanation of its methodology on the RCP site). But FiveThirtyEight the past several presidential election cycles has had more accurate polling averages than especially RCP but other sites like The New York Times, whose polls this cycle, conducted with Sienna College, seem to be oversampling Republican-leaning voters significantly (meaning disproportionately including Republican representation in their estimation of who will turn out and vote). Thus, my main source of information for polling is always FiveThirtyEight’s polling averages. I will be using their national and state polling averages here.
In addition, I will be also combining polling data with modeled partisan voter registration modeled partisanship from Tom Bonier’s excellent outfit TargetSmart, the data presented from which I used during the 2022 midterms to predict, accurately, when very few others would, that there would be no “red wave” and that polls were undercounting Democratic support in the aftermath of the Supreme Court’s infamous Dobbs decision that overturned Roe v. Wade, women especially coming out strongly to support Democrats for a historic overperformance for a party in power of the same party as a sitting president in his first term. And newly registered voters are not only more likely to vote than previously registered voters in a soon-to-happen election, but they are indicative also that others in the groups they represent are also more likely to vote. While in many cases, the absolute numbers of voters registering are less than in 2020, the portions overall of all new registrants shifting significantly net towards Democrats is a very telling dataset considering 1-how insanely close the results of the key swing states were in 2020 and 2-how close in polling it is today it is in key states where there is a disproportionate portion of new voters modeled as likely to vote for Democrats.
In a moment of statistical ties in swing state polling, that that is as good a sign as any that Democrats will win in states statically tied—that is, within the margin-of-error range of what usually 95% of outcomes would be expected to fall under—or very close. For this reason, in these close races in which the polls are a statistical tie, I am going to let modeled partisanship based on voter registration data be one of the decisive factors in how I make my predictions. So for me, the more of a net shift towards Democrats in modeled partisanship share of new registrations, the more likely Democrats are likely to outperform their polls, i.e., the less polling will capture this shift. My “net shift” is calculated by the getting difference between the point shifts for each side, so if GOP went down 3 points and Democrats up 2, that’s a 5-point shift in favor of Democrats in my scoring system.
I will also very much be relying on my own analysis from my last article—based on voter registration and early voting data as presented by Bonier’s outfit and its (and my) excellent track record from 2022—that very confidently concluded state and national polls across the board are generally off, missing those hugely disproportionate swings compared to 2020 in the share of new voter registrations towards Democratic leaning populations and the effect of new abortion and reproductive rights restrictions in many states across the country since the fall of Roe. As I noted in my last piece and will reinforce here, when races are extremely close, polling with its inherent errors is not a sufficient tool in trying to analyze who will win and who will lose and analysis will have to take into account other factors, as I have before and will here.
Finally, based mainly on my discussions of the individual states and national polling data along with early voting data, I will make a range of Electoral College map predictions as well as a win probability prediction for both Kamala Harris-Tim Walz and Donald Trump-JD Vance (rough on the odds part).
Caveats, source, and presentation notes:
- The TargetSmart Voter Registration Dashboard shows you when each state last reported its registrants and they are not uniform. The dashboard lets you adjust to match the time when voter registration data was last reported to data from the same time in the previous presidential election in 2020 and the 2022 midterms as a frame of reference. Obviously, when not looking at individual states, there is no way to get a fully accurate national or all-swing-state picture with combined states with different dates, but most are within a few weeks of each other so the combined outputs can give a quite rough but still quite useful general sense of what’s going on. You can play around yourself with different data and different variables—race, age, gender, rural, urban, etc.—and I recommend that you do, but I found modeled partisanship to be most useful.
- All polling averages are what FiveThirtyEight has posted as of 8:15 PM on October 30; if numbers don’t quite add up (0.1 off or tied) in terms of the FiveThirtyEight numbers given, it might be because they rounded using more precise data from crosstabs.
- In my presentation here and in general, polling data/shifts are generally rounded to nearest tenth of a percent/point.
- Herein, polling figure is given as a percent. I will use “points” to describe a number-level shift, and a percent increase would be the increase by a factor, as in a 2.5-point increase from 5% to 7.5% would be a 50% increase, by a factor of 1.5 times the original.
- FiveThirtyEight uses likely voter iterations of a poll in its weighted averages whenever possible.
- Numbers given for Puerto Ricans in each state are from 2023 from the U.S. Census Bureau’s American Community Survey, except Alaska’s are from 2022.
- All polling chart images from FiveThirtyEight and all registration chart images from TargetSmart
- Coming up with overall predictions is more an art, not a science, but one using scientific tools that are inherently flawed and yet still pretty-damn good in the aggregate.
On Polling
In my previous article, I really went into the weeds on some polling issues and why I was so confident in 2022 and why I am even more confident now in the polls undercounting Democratic support in 2024 (and remember, I was one of the only people in 2022 making such a prediction, so take that for what it’s worth). But on average, according to FiveThirtyEight, overall from 2000 to 2020, presidential general election polls three weeks out from Election Day—both state and national—have been off by 4.3% since 2000. Harry Enten noted in a FiveThirtyEight 2016 classic that the a 2-point error was the average presidential national polling error a week before the election from 1968-2012, which would fall in most cases within the margin of error of each poll, or a “standard polling error,” noting that Trump was just such a standard polling error away from winning. Even with 2022 being far more accurate a year for polling that usual, the polls in close House races in the midterm year incorrectly favored Republicans (also known as GOP, for Grand Old Party) in most of those races, with Democrats winning in most of them in part because of women being fired up about their rights being taken away by the Supreme Court and their state legislatures. Things have only gotten worse for women since then, with new restrictive laws literally killing women across the country. So do not make the mistake in thinking that this issue has faded for women.
And with amazing new registration proportions for Democratic-leaners compared to Republican-leaners, it is key to note that pollsters often do not incorporate newly registered voters into their models and sampling or find it hard to do so accurately. So I am very confident that polls are undercounting the level of Democratic vote share for the 2024 election and you can explore my detailed discussion here. As a result of this gap, I am putting significant weight behind new voter registration share by likely-party vote. Therefore, you can add a few points to most polls in favor of Harris in most cases, both in state polls and nationally, and you will likely get a better sense of where things stand.
THE NUMBERS AND PREDICTIONS
The Seven States that Most Mainstream Press Agrees Are the “Swing States” (UPDATED: November 3) *AND Iowa
Ratings system: Arrange by worst for Harris to Best for Harris and reflecting likelihood/probability of winning, not necessarily margin of victory: Edge Trump, Edge Harris, Advantage Harris, Strong Harris
All of these states are currently within the aforementioned margin of error and are considered statistically tied, essentially jump balls. *UDPATE November 3: Voter registration modeled partisanship has slightly increased for Democrats and slightly decreased for Republicans in Pennsylvania, North Caroline, and Georgia with new batches of voter registration data released by those states since October 31 and new modeling on that data being released by TargetSmart; therefore, the gaps highlighted below for those three states are now slightly higher in favor of Democrats than the figures given here, having just increased in the final days of the campaign, another data point in favor of a momentum swing for Harris.
*UPDATE: NOVEMBER 3: IOWA!!!—6 Electoral Votes
Polling: Average not given. I am surprised to be including Iowa here, but I have no choice. I know that I always caution not to put too much stock in any one poll, but I must make somewhat of an exception here (and to be fair, it’s not just one poll). To get to the point, Iowa pollster Ann Selzer is one of the best pollsters in politics, perhaps the best. He long track record speaks for itself and her approach is meticulous. She released polls in partnership with The Des Moines Register back in February and June that had Biden losing by 15 and 18 points to Trump, respectively. Then in September, that lead in a another Selzer poll had shrunken to just 4 points against Harris. And now, just before the election, Selzer has Harris overtaking Trump by 3 points, 47%-44%! Yes, it is within her margin of error, but Selzer’s poll if the only poll to show anything like this—all others have Trump up (nearly all significantly) and another poll just released has him up 9 points—but no other pollster conducting these polls is like Ann or has her record. The profound thing about this poll is, that if it is accurate, it surely captures something beyond Iowa, something regional, maybe even national, that the vast majority of polls have simply missed. It is at the heart of what I have gotten into in this article before this update, that the modeled partisanship from voter registration data tells a story that is incompatible with that narrative and narratives we have from the polls, and suggests movements and shifts in the electorate among significant parts of the population, especially with new voters likely to be missed or underrepresented especially if they are registering in historic and abnormal ways. And Nate Silver and The New York Times‘ Nate Cohn both just pointed out how bad pollsters this cycle have been with “herding,” or modifying their results to not stand out too much and not take heat for being “wrong” or an outlier. Now, none of this means that Selzer must be right, and the others wrong. And I am not going to move Iowa into Harris’s column based on just Selzer’s latest poll and the shifts from her previous poll. But Ann is credible in and of herself, I have enough respect for and deference to her, and her previous three polls were well within the realm of being favorable or very favorable to Trump, that I am going to move Iowa into battleground swing-state status, but as all the other polls have Trump winning, even though I think they are not up to the standards of Selzer, I am rating Iowa as slightly favoring Trump. Without Selzer, I wouldn’t even be discussing Iowa, but now we all should be: Harris has a real chance if Selzer’s poll is accurate, and her polls usually are. If I am right, Selzer is capturing what I am onto, what the modeled voter registration partisanship is pointing towards, and what will explain a large Harris win if that is what unfolds over the next few days. If Selzer’s poll is accurate, it might just be the most important poll in the history of American politics. As of now, it is certainly one of the most discussed and highlighted of in American history, I have certainly never before seen a reaction to a single poll as massive as the political world’s reaction to this poll.
Also, there are also over 9.000 Puerto Ricans in Iowa (see the Pennsylvania discussion below) while the Uncommitted movement received over 600 votes, or 4.5% of the vote (see the Michigan discussion below); Trump won Iowa in 2020 by over 138,000 votes, or 8.2%.
IOWA RATING: Edge Trump (but don’t be surprised if Harris wins!)
New registration modeled party share: Last updated October 20
Vs. same weeks before election in 2020: Democrats +3.1 points, GOP +0.3 points, net shift +2.8 points for Democrats
While a slight shift, given Selzer’s poll and Iowa’s small population, this could end up making a big difference…
Pennsylvania—19 Electoral Votes
Polling average: Trump 47.9%, Harris 47.5%, Trump +0.4 points
Polling is razor close, and in most scenarios I have whoever wins Pennsylvania winning it all, finding it hard to imagine a scenario in which that is not the case (though not impossible). While the polling is within the margin of error, I think in nearly all these states, the polls are undercounting democratic support. Having said that, even if it is, say, 3% off (putting her at +2.7), that is still a very close race polling-wise and within the margin of error. So, as stated, I am going to look at new registrants.
New registration modeled party share: Last updated October 20
Vs. same weeks before election in 2020: Democrats +7.6 points, GOP -3.1 points, net shift 10.7 points for Democrats
As far as Pennsylvania’s voter registration data, of the three Rust Belt battleground states, the data is most promising here for Harris and by far. Also factor in that popular Pennsylvania Governor Josh Shapiro is a rockstar in the state and is campaigning heavily for Harris there. Additionally, factor in that at Trump’s final rally at Madison Square Garden, a speaker prominently trashed Puerto Rico as a “floating island of garbage in the middle of the ocean” (all Puerto Ricans are U.S. citizens, in case you did not know that) and that there are over 472,000 Puerto Ricans in Pennsylvania, the most of any swing state (in each swing state we will now be discussing the Puerto Rican x-factor! Thanks, Tony Hinchcliffe), which stands up well against the over 61,000 write-ins for 5.6% of the vote in the Democratic primary, most representing Uncommitted protesting Israel’s war in Gaza (see the discussion on Michigan below). Remember that Trump lost in Pennsylvania by less than 82,000 votes in 2020, and I think the net picture of these factors in Pennsylvania means the state is a pretty good bet for Harris.
PENNSYLVANIA RATING: Strong Harris
Also, Democratic incumbent Senator Bob Casey is polling better than Harris against Republican challenger Dave McCormick, so I see a strong chance for Casey’s victory.
Michigan—15 Electoral Votes
Polling average: Harris 48%, Trump 47%, Harris +1 point
Even though Harris is polling better here than anywhere else not only in the Rust Belt but also better than any other swing state, this is deceptive. And yes, while Michigan has the third-highest African-American population proportionately of any swing state and fifteenth-highest in the nation, there are other factors that are not good for Harris here in Michigan…
New registration modeled party share: Last updated October 20
Vs. same weeks before election in 2020: Republicans +8.4 points, Democrats -2.1 points: net shift 10.5 points for GOP
Of the seven states dubbed by most as swing states, this is one of only two that saw a shift in favor of Republicans (and by far the larger shift) and this bodes quite ill for Democrats in Michigan as it is a significant shift that could very well overcome the fact that Harris’s polling average here is her best in the Rust Belt, given that that it is not her best by much. One x-factor? In such a close state, there are almost 54,000 Puerto Ricans in Michigan. And all this in a state a state Trump lost by less than 155,000 votes in 2020. Also, it should be noted Michigan has a something of a screwy recent history of throwing pollsters curve balls, for what that’s worth.
But the real elephant in the room for Democrats and Michigan are the deaths of many Arabs in the fighting with Israel: the horrific mass violence, mass displacement, and mass destruction in Palestine’s Gaza in the escalation of the conflict between Hamas and Israel that erupted with Hama’s massive terrorist attack against Israel on October 7, 2023, as well as increasing violence in Lebanon and Palestine’s illegally-occupied-by-Israel-since-1967 West Bank, are all deeply affecting Michigan’s substantial Arab-American community, which consists of over 208,000 people and in Michigan represents the highest share for an Arab population of the total population in any state. As a result these events and the Biden-Harris Administration’s support for Israel, many of them are holding Biden and Harris responsible for, in their view, not stopping and even enabling all of this and are now reluctant or refusing to support her. In fact, out of all the swing states, Michigan had the most proportionately and absolutely Uncommitted National Movement votes protesting the war in Gaza in the Democratic Primary this year: over 101,000 voters for some 13.2% of the presidential primary vote. The movement includes many non-Arabs and non-Muslims and many young people.
But to be clear, not all these people—maybe even not most—would necessarily not support Harris over Trump, as the movement was meant to send a signal and was not officially a firm anti-Biden or anti-Harris vote, hence the term “uncommitted” (and I am sure plenty of Arabs in Michigan will still vote for her, just less so than if what was happening in the Middle East was not happening), while some of the Puerto Rican population are minors and all the Uncommitteds were able to vote. Thus, do no look at the Puerto Ricans and Uncommitted one-to-one as easily canceling each other out, but I am just giving all this data an context for.. context (remember the name of my site, cherished readers).
One other thing: despite endorsing Trump, former candidate Robert F. Kennedy Jr. is still on Michigan’s ballot as a separate candidate, having lost multiple legal challenges to remove his name, including just recently at the U.S. Supreme Court. Along with Wisconsin, Michigan is one of only two swing states in which he remains on the ballot.
This state really is up for grabs, but with the polls so close and statistically tied, and such a meaningful shift towards Republicans in the share of overall new registration vs 2020 and problems Democrats are having with Arab-Americans, I would have to say I see Trump as slightly more likely to win, but it could be extremely close either way.
MICHIGAN RATING: Edge Trump
On a positive note for Democrats, their Senate candidate Elissa Slotkin is polling significantly better than Harris and I give her and an advantage against her Republican opponent Mike Rodgers.
Wisconsin—10 Electoral Votes
Polling average: Harris 48.0%, Trump 47.6%, Harris +0.4 points
Wisconsin had been where Harris was polling the best in the Rust Belt for some time, but things have narrowed for whatever reasons in recent weeks. Even though the polling now is slightly better in Michigan, there are not the complications in that state has that I just discussed here in Wisconsin.
New registration modeled party share: Last updated October 21
Vs. same weeks before election: Democrats -4.1 points, GOP -6 points, net shift 2.1 points for Democrats
Still, with a statistical tie in effect with such a slight lead, modeled voter registration partisanship does not give her a huge boost: of the five swing states where there has been a shift towards Democrats, Wisconsin has the weakest such shift by far. The potential Puerto Rican factor involves nearly 65,000 Puerto Ricans. Conversely, for what it’s worth, the Uncommitted vote in the Democratic primary was over 48,000 votes for 8.3% of the vote (Trump lost by less than 21,000 votes). Also, like Michigan and as mentioned, Wisconsin is one of only two swing states where RFK Jr. is still on the ballot, most recently being denied removal from both by the U.S. Supreme Court. Based on this available data and my conclusion that polls are undercounting Democratic support, I’d say Harris is more likely than not to win here in Wisconsin and not just barely, though it will still likely be close.
WISCONSIN RATING: Advantage Harris
Incumbent Democratic Senator Tammy Baldwin is in a tight race against Republican Eric Hovde, but has consistently polled ahead of him. Along with Harris, I see her as having an advantage.
Georgia—16 Electoral Votes
Polling average: Trump 48.7%, Harris 47.0%, Trump +1.8 points
Georgia is Trump’s second-best poll-showing among swing states, but I would argue this is deceptive. It is still well within the margin of error and there are certainly some other factors that are very much not in his favor when it comes to this state…
New registration modeled party share: Last updated October 8
Vs. same weeks before election in 2020: Democrats +7.1 points, Republicans -12.2 points, net shift +19.3 points for Democrats
Folks, this is a massive shift towards Democrats and I do not think polling has captured this. And part of that is black women being incredibly enthused for Kamala Harris in a state that has the highest African-American share of the overall population of any swing state and the third-highest of any state. And let’s not forget Trump’s longstanding feuds with Republican Georgia Governor Brian Kemp and Georgia Secretary of State Brad Raffensperger (listen to my interview with him six days before Trump’s Capitol Insurrection) acting against Trump’s longstanding election overturning/denial efforts. There are also over 120,000 Puerto Ricans in Georgia (Trump infamously lost by less than 12,000 votes) and Uncommitted was not on the ballot here in the Democratic primary. Finally, do not underestimate the political machine that former Democratic gubernatorial candidate and Georgia House of Representatives Minority leader Stacey Abrams has helped put in place. Especially with the voter registration data, I am very confident Harris wins Georgia.
GEORGIA RATING: Strong Harris
North Carolina—16 Electoral Votes
Polling average: Trump 48.3,% Harris 47.3%, Trump +1.1 point
Once again, there is a statistical tie here. But with a Republican Governor candidate who is possibly the worst gubernatorial candidate in modern American history—an African-American extremist who called himself a “black NAZI!” on a porn website’s forum, among other disgracefully pathetic and offensive things in his history—likely dragging Trump down among undecideds, independents, and some Republicans, the closeness in the poll might be deceptive. But then there is the voter registration stuff…
New registration modeled party share: Last updated October 11
Vs. same weeks before election in 2020: Democrats +5.4 points, GOP -5.5 points, net shift +10.9 points for Democrats
North Carolina also has the second-highest portion of African-American voters of any swing state and the eight-most in America, something that is definitely quite favorable for Harris. Also, there are over 130,000 Puerto Ricans in North Carolina; on the other side, North Carolina had second-highest absolute and proportionate turnout for Uncommitted during the Democratic primary among swing states: almost 89,000 voters for 12.7% of the vote (Trump won the state in 2020 by less than 75,000 votes). On balance with all of this, and given especially the big shift in modeled partisanship and the state’s large black population, I think Harris has a clear advantage in this state, though not a strong one as Trump won by close to 75,000 votes in 2020, so that margin won’t be easy to overcome.
NORTH CAROLINA RATING: Advantage Harris* (I am putting an asterisk here because an x-factor is the intersection of the damage from Hurricane Helene that may yet lead to many people having difficulties in voting, the rampant and wildly dangerous disinformation spewed by Trump and Russia and many others about FEMA and overall government relief efforts in the state leading to actual armed threats against FEMA staff, and the actual major efforts going on there by the Biden-Harris Administration to help people there; I do not know how and/or if these dynamics will play out and affect the election here so leave some room for North Carolina to surprise).
Nevada—6 Electoral Votes
Polling average: Harris 47.5%, Trump +47.4%, Harris +0.1 points
Polls here in Nevada are really tight, though they have generally been better for Harris here than its southwestern counterpart, Arizona. As in most other cases, though, I still think polls are undercounting Democratic support. So what does voter registration data tell us?
New registration modeled party share: Last updated October 25
Vs. same weeks before election in 2020: Democrats -7.1 points, GOP -6.2 points, net shift +1.1 points for GOP
This is the only other swing state besides Michigan that shifted towards the GOP, though only slightly so, and in a state with a smaller population like Nevada, that could amount to a big deal. There are also approaching 35,000 Puerto Ricans in Nevada, while a “None of these candidates” inspired by Uncommitted received less than 7,500 votes at 5.5% of the Democratic primary vote (Trump only lost this state by in 2020 by less than 24,000 votes). This has the potential to be the closest state, but with Harris actually leading in the polls if only by the most minute amount, and with only the barest shift so far towards the GOP in registration, this comes down to a gut feeling if I am not going to award a tie, and my gut is going with Harris given how things are breaking overall late in the game and with the wisdom of Jon Ralston, who really knows Nevada politics, also giving me confidence.
NEVADA RATING: (barely) Edge Harris
Incumbent Democratic Senator Jackie Rosen is polling significantly ahead of her Republican opponent Sam Brown as well as Harris, so I would argue she has a strong chance to win.
Arizona—11 Electoral Votes
Polling average: Trump 48.7%, Harris 46.6%, Trump +2.2 points
Arizona is currently by far Trump’s best-polling state, but is still within the margin of error in a climate in which, again, I think most polls are undercounting Democratic support. So what does the modeled partisanship of new registrations based on registration data tell us?
New registration modeled party share: Last updated October 21
Vs. same weeks before election in 2020: Democrats +13.3 points, GOP -19.1 points, net shift +32.4 points net shift for Democrats
This is the most significant shift vs. 2020 for any swing state, and that almost speaks for itself, casting serious doubt as to the accuracy of the neck-and-neck polling. On top of this, the Puerto Rican factor here: almost 54,000 people and Uncommitted was not on the ballot in a state Trump lost by less than 11,000 votes. While Trump is up a few points in the poll, this massive shift in voter registration makes me think the polls in Arizona are significantly underestimating Democratic votes. Trump has a real shot to win Arizona, sure, but Harris should be favored and not just barely.
ARIZONA RATING: Advantage Harris
Also, in the race to fill the vacancy left by departing Arizona Senator and former Democrat Kyrsten Sinema, Democratic Candidate Ruben Gallego has been polling very well against his opponent, the failed Republican 2020 gubernatorial candidate Kari Lake, who refused to concede defeat and supports Trump’s election denialism. Gallego has a very strong chance of winning.
My Dark Horses
Yes, there can be some shocks!
Alaska—3 Electoral Votes
Polling: Average not given, but all polls since September have been conducted by the same pollster and all were at least +8 Trump except for one poll in mid-September that had Trump only up 5 points. Based on this, it would seem safe to rule Harris out for winning Alaska, except…
New registration modeled party share: Last updated October 21
Vs. same weeks before election in: Democrats +8.5 points, GOP -17.8 points: net shift +26.3 points for Democrats
That’s a huge deal, dear readers: Trump only won Alaska by some 36,000 votes (10% of the vote in such a low-population state), meaning a shift of a little over 18,000 votes (5%) would have given Biden the state.
And in 2022, Alaskans voted to send to the House as their one representative the amazing Mary Peltola, the state’s first Democrat to fill that seat and first Alaska Native to do so. I really expect her to help drive a big Alaska Native turnout, and the registration data is also showing this, with Alaska Natives as a share of voter registrations nearly 43% higher than in 2022, meaning they are 43% more of the share of new registrants (Native American/Alaska Native coded as NAAN in TargetSmart in the above). There was also no Uncommitted on the ballot in the Democratic primary, but there are over 10,000 Puerto Ricans in Alaska. And finally, in a state with such a small population, the fact that RFK Jr. is still on the ballot here could also make a difference. So if a state not at all on the radar is a surprise flip, I’m saying that state will be Alaska.
ALASKA RATING: Advantage Trump but Harris could really do this
Ohio—17 Electoral Votes
Polling average: Trump 51.7%, Harris 43.7%, Trump +7.9 points
Yes, Trump has a sizable lead over Harris. While it is unlikely Harris will win Ohio, incumbent Democratic Senator Sherrod Brown has been generally far ahead of her and generally also ahead of his Republican opponent, Bernie Moreno, but it’s still a close race. With my analysis telling me polls are undercounting Democrats in general, I find the best way to feel confident in specific races is corroborating those Races to increases in the share of modeled voter registration partisanship for Democrats where the races are taking place. So, without further ado…
New registration modeled party share: Last updated October 10
Vs. same weeks before election in 2020: Democrats +18.8 points, GOP -17 points; net shift +35.8 points for Democrats
These are remarkable numbers for Ohio and a larger shift than any of the swing states and seem quite dissonant with the polling. So remarkable, in fact, that they actually give Harris a real shot at an upset even if Trump is favored. You would think with sitting Ohio Senator JD Vance as Trump’s running mate that these numbers would be different, but they are not. The numbers also give Brown a strong chance to win. Maybe the debunked racist smears about legal Haitians migrants eating neighbors’ pets as well as ducks and geese touted by both Trump and Vance and many other Republicans backfired, as Democrats gained almost 0.4 points in the modeled share 9 weeks out from the election—when these vile rumors began to rise—to four weeks out (when the latest Ohio data is available) while in the same period, Republicans lost about 2.3 points… Back to the racism theme, here there are nearly 134,000 Puerto Ricans in a state where six years ago, Brown won reelection by less than 300,000 votes.
Sure, Trump is favored, but there is something in these numbers telling me you can’t rule out a Harris upset and it’s would not be a miracle to happen.
OHIO RATING: Advantage Trump
Texas—40 Electoral Votes
Polling average: Trump 51.2%, Harris 44.0%, Trump +7.2 points
It sure doesn’t look good for Harris here, but like in Ohio, the Senate is a different story, and also like in Ohio, the registration information is shocking.
New registration modeled party share: Last updated October 11
Vs. same weeks before election in 2020: Democrats +11.1 points, GOP -18.5 points: net shift +29.6 points for Democrats
Yet another massive shift for Democrats in a supposedly safe state for Republicans. And there are over 269,000 Puerto Ricans here who might make a difference here in a state in which incumbent Republican Senator Ted Cruz only won his 2018 Senate reelection against Democrat Beto O’Rourke by less than 215,000 votes. These Puerto Ricans may help the state say “adios” to Cruz, as his Democratic challenger, former NFL player Colin Allred, currently serving in the U.S. House for Texas, is polling much better than Harris, Cruz polling much worse than Trump. But given these blowout numbers on modeled registration partisanship, Harris has a real if relatively small chance, too. As for Trump, he won by over 631,000 votes (yes, Texas is a big state). When it comes to Uncommitted, it was not on the ballot in Texas.
TEXAS RATING: Advantage Trump
And I’d give Allred an edge over Cruz since I already think the polls are undercounting Democratic support and because there is such a huge shift in favor of Democrats.
All Swing States
New registration modeled party share 5 weeks out vs. same in 2020 (rough since not all states reporting on same dates, so October data not included):
Democrats +3.7 points, GOP -5.4 points, net shift +9.1 points for Democrats
Other Notes from Outside the Battleground States
Florida seemed it might be competitive in polling for a bit, but polling has now given Trump an even larger lead and modeled voter registration partisanship based on favorable voter registration partisanship data for Republicans from there as of September 30 seems to rule out a major upset there, though if somehow this changes dramatically with data in October months, maybe there’s a miracle waiting for Harris. Update November 4: Incumbent Republican Senator Rick Scott is underperforming compared to Trump, and while he is in a strong position to be expected win, Democrat Debbie Mucarsel-Powell would have a small chance if there was a shift in new registrations for October and/or polls are way off and Harris is competitive or winning in a place like Iowa, for, while a decent number of polls gave Scott a large lead, a decent number down the stretch did show a closer race, though none had Mucarsel-Powell ahead or closer than 3 points behind. Virginia had hinted at being competitive in a few polls—though I never bought that this would be the case closer to the election and when the final votes were tallied—but no longer, and with the same type of registration data and modeling current as of October 18 overwhelmingly favoring Democrats, it seems impossible for Trump to win now. And quirky New Hampshire flirted with Trump in a few polls, but Harris seems to have pulled away there, too (New Hampshire has not provided registration data, as it is wont to quirk).
There is no reason to think Harris can win Montana, but that state’s embattled incumbent Senator Jon Tester is a different story. Until recently, he was trailing significantly, every poll from August 12 through October 8 having his opponent, Republican Tim Sheehy, ahead 6-8 points (save one that had Tester up 5 points from August 14 and several were GOP partisan polls). But the last two polls in October have Tester tied or only down three, a decent sign the race may very well be tightening significantly, suggesting Tester has at least a decent chance to hang on. Yet the voter registration shifts up to October 21 there were massively favoring Republicans, making it less likely than in other states that the polls are significantly underestimating Democratic support. Thus, without any significant data favoring Democrats here, Tester is easily the most vulnerable incumbent Democrat. Yet don’t count him out yet—momentum does seem to be in his favor now and recent polling is likely a reflection of multiple serious scandals that have embroiled Sheehy—a relative newcomer to the state—in the final weeks of this race, calling into question his credibility and character. If future polling before the election comes out showing a tighter race or even tester up again, I would favor Tester in a state where even slight shifts can make a huge difference in state with such a small population. Still, with the available data, Sheehy has an edge for now, but one he is in danger of losing in these final days.
And finally, polling in congressional districts and voter registration data in the two states, Maine (registration data as of October 15) and Nebraska, that award electoral votes the winners of the districts, do not seem competitive: Maine’s 2nd Congressional District (1 electoral vote) seems to be safely going to Trump and Harris seems a safe bet to win Nebraska’s 2nd Congressional District (also 1 electoral vote). A big shift towards Democrats in the share of new registrants in Nebraska statewide (with data as of September 30) doesn’t just help Harris in the split district favoring her: it bodes ill for incumbent Republican Senator Deb Fischer against the surprisingly very-competitive-in-polling independent candidate, Dan Osborn (+12 points for Democrats, -11.2 for Republicans, a net shift of +32.2 points for Democrats as of September 30!). My money would definitely be on a strong chance for Osborn here.
NATIONALLY
Popular vote: Harris 48.1%, Trump 46.7%, Harris +1.4 points
I don’t think for a second that Harris only wins the popular vote by 1.4% unless maybe she loses the Electoral College and even then she might beat that polling. As I have stated throughout, I think polls are generally undercounting Democratic support, whose coalition runs from Republican President George W. Bush’s Vice President, Dick Cheney, and his daughter and former third-highest Republican leader in the House, Liz Cheney, and former Republican Congressmen Joe Walsh and Adam Kinzinger on one side, to Nancy Pelosi and AOC and Bernie Sanders on the other, to include never-Trumper Republicans and pro-lifers who do not want to impose their views on others and still others keeping their vote a secret for various reasons (including many married people from their partners). But as we have seen in 2016 and 2000, the popular vote winner does not always win. Yet as far as the popular vote, feel free to add a few points to her total here.
New registration modeled party share vs 2020, 3 weeks out (rough, because of different reporting times and because New Hampshire and North Dakota are not reporting information):
Democrats +3.2 points, Republicans -4 points, net shift +7.2 points for Democrats nationally
It seems almost impossible for Trump to win the popular vote with a shift like that from 2020 to 2024. So expect Harris to win the popular vote no matter what and to win by more than 1.4% and more than just a little more.
ELECTORAL COLLEGE
The way I see this, and this is a rough probability breakdown, there’s a 20% chance of Trump edging out Harris, a 30% of Harris edging out Trump, and 30% chance of a strong defeat of Trump, and a 20% chance of a huge win for Harris (all maps from 270toWin) *UPDATE NOVEMBER 3: Iowa!
Edge Trump Map: 277-293* Trump, 245-261* Harris (roughly 20% chance)
*I am really bullish on Georgia and Pennsylvania for Harris and think she would win both even if Trump wins, but could see Trump picking off one or the other in theory, though I think this is less likely than what I chose to show in the map.
Edge Harris Map: Harris 272-288*, Trump 250-266* (roughly 30% chance)
*I can see Nevada and Wisconsin going either way, hence the range, but the map is my best guess within these possibilities.
Strong Harris Map: Harris 319-328*, Trump 210-219* (roughly 30% chance^ and this is what my head believes will really happen; UPDATE November 5: ^I am sticking with this as my most likely scenario out of the four maps I have presented herein)
*Even in this scenario, am giving her a shot at Alaska. The range is for Alaska and Hawaii going either way.
Huge Harris Map: Harris 385, Trump 153 (roughly 20% chance but capturing my heart)
Sure, you laugh, but remember that there were net shifts in Texas of +29.6, Ohio of +35.8, and Alaska of +26.3 points in favor of Democrats as a share of voter registrations for this election since 2020! The polls don’t tell the whole story, and, anyway, I have 80% of outcomes being more favorable for Trump than this.
So basically, a 1 in 5 chance Trump wins, a 1 in 5 chance she blows him out, and a 2 in 3 chance she wins by between a little (1 in 3) and solidly (1 in 3). This is not meant to be precise, but a rough proclamation. I thought about tinkering a bit, maybe changing the solid Harris win to 35% at the expense of a Harris blowout, moving that down to 15%, but I want to keep it relatively simple and I was almost blown away by the modeled partisanship registration shares of Texas, Ohio, and Alaska, which I think tell us something big is going on the ground in these places that the polls are just missing. So I am sticking, officially, with 20-30-30-20 on the odds unless decisive new data pops out in the next few days, which is highly unlikely.
And if I had to pick one and a professional, as noted, it would be the third strong map.
CONCLUSION
There are other factors, such as voters who would normally not vote for Democrats but are increasingly disgusted by Trump’s election denialism and insurrectionism and recent economic developments., that may also prove to be hard-to-measure factors boosting Harris’s final vote tallies and/or margins that I did not discuss in any detail. What is clear is that the voter registration data shows a lot of drops in GOP-modeled voters’ share of registrations and increases in both Democrats and independents throughout the country, something clearly not being captured by the estimate of likely voter models that are being used to craft the poll samples.
Will Puerto Ricans be a fired-up factor in the swing states and beyond? I would at least bet on at least somewhat higher turnout from Puerto Ricans for Democrats after the aforementioned disgusting insult was recently blasted live from Madison Square Garden to the nation at a Trump rally which came off like a stylistic homage to a 1939 American Nazi rally that also took place at Madison Square Garden.
One way I could be wrong is if somehow massive amounts of voters who voted for Biden in 2020 will be voting for Trump in 2024, but color me skeptical—highly skeptical—of that being a thing. As I discussed in my last article one thing is clear: certain groups of voters are registering in highly disproportionate level compared to past elections, and these groups significantly lean Democratic. And none have surpassed their previous proportional levels of registration as have African-American women and Latinas/Hispanic women. These historic proportions and the shifts in voter registration during this cycle—the Harris Effect—they have helped lead to fruition tell a story different from the polling, one not captured by polling, and that means we should expect some surprises in the coming days as a nation and the world hold their collective breaths.
In short, the likelihood of Harris winning is definitely significantly higher than the polls would indicate, though of course this does not mean Trump does not have a still real—if significantly less likely—chance, not minuscule one but perhaps a higher chance than of losing Russian Roulette with a six-shooter.
And if she wins, from the information presented here there, it should be clear there is a good chance Democrats can hold the Senate, especially if Tester is able to continue closing the gap in Montana and the independent Osborn defeats incumbent Republican Fischer in Nebraska, helping offset the near-certain loss of the current seat of retiring West Virginia’s Senator Joe Manchin, now formerly a Democrat, to West Virginia Republican Governor Jim Justice in a race against Democrat Glenn Elliot (the Senate is currently a coalition of Democrats and independents that form 51 majority against 49 Republicans (50-50 would be a tenuous majority if Harris wins because Walz as vice president would be the tie breaking vote).
As for the House, there are a number of good polls for Democrats against incumbent Republicans, but House polls are a lot more challenging to conduct and less accurate. In the end, there is good reason to think there is a good chance that if Harris can ride a rising wave of momentum to victory, so can many House Democratic candidates in competitive races, and Democrats only need a net gain of five seats to take back control. Especially if Harris wins more than narrowly, expect Democrats to retake the House and hold the Senate, but a more narrow win for her would leave such an outcome in doubt without ruling it out. Conversely, if Trump wins, with a number of Republican Senate candidates polling well below him, there is more of a chance of divided government.
As with every election, so much is at stake, but so much more is at stake in this particular election: almost everything is up for grabs, including the very nature of our government and the survival of democracy itself.
One favor: when the dust settles, if I turn out to be as accurate as I have been in the past, please do share this profusely so we can improve the level of analysis in our fraught media landscape.
In data, we trust, so trust in a Harris victory as by far the most likely outcome.
© 2024 Brian E. Frydenborg all rights reserved, permission required for republication, attributed quotations welcome
Also see Brian’s eBook, A Song of Gas and Politics: How Ukraine Is at the Center of Trump-Russia, or, Ukrainegate: A “New” Phase in the Trump-Russia Saga Made from Recycled Materials, available for Amazon Kindle and Barnes & Noble Nook (preview here).
If you appreciate Brian’s unique content, you can support him and his work by donating here. Real Context News produces commissioned content for clients upon request at its discretion.
Feel free to share and repost this article on LinkedIn, Facebook, Threads, and Twitter. If you think your site or another would be a good place for this or would like to have Brian generate content for you, your site, or your organization, please do not hesitate to reach out to him!