Skip to content

Kamala Harris is getting close to a durable lead over Trump

My rule of thumb is that Kamala Harris needs to win the popular vote by 4-5% in order to win the election. According to the Economist tracker, she's there:

If Harris can expand her popular vote lead by another point or two, I think she'll be the very likely winner. If she expands by three points or so, she'll win by enough of a margin that even Donald Trump will have a hard time spurring his minions into a repeat of January 6.

48 thoughts on “Kamala Harris is getting close to a durable lead over Trump

    1. KenSchulz

      The relation is purely stochastic — the greater the margin, the more likely to win the EC, but there is no point-value threshold. But the factors that sway voters tend to correlate across the states — so the shift toward the Democratic ticket after Harris became the candidate has been seen virtually everywhere.

      1. FrankM

        While states tend to move together, they don't do so in lock-step so that the relationship between overall vote totals and EC count has plenty of uncertainty. I pretty much ignore overall vote totals and just look at the seven battleground states.

        Harris has small but consistent leads in WI and MI. The other five are all within 1 percentage point.

        1. mudwall jackson

          slight disagreement. if the national margin is one or two points, harris is in trouble. extend that to four or five and it's likely she's the winner. you're right that there's no guarantee either way but the odds do tilt in one direction or the other.

          btw i've seen a poll as of tuesday that has harris up 3 in pa., the keystone state in more ways than one. harris is spending a ton of time and money and has far more staff and volunteers there than trump, so i'm optimistic.

          1. FrankM

            The relationship between national vote margin and EC votes has plenty of uncertainty, but it's not infinite. Certainly it's possible to get a national vote margin so large that an EC victory would be assured. Unfortunately, I very much doubt that's going to be the case this year. I'm not convinced it's going to be the squeaker everyone is expecting, but there isn't going to be a landslide, either. I've still got my eyes on the seven battleground states.

            The Suffolk University poll had Harris up +3 in PA, but the InsiderAdvantage had Trump +1. It's still a tossup, but there's still 7 weeks to go. IMHO, it's all about turnout and that's the one thing polls can't measure. If all those who voted for Biden turn out and vote, Harris wins (I'm pretty confident only an infinitesimal number of them will switch to Trump).

          2. Batchman

            You wouldn't know it from the press coverage. Trump is still dominating the daily feed, between attempts on his life and the still-unfolding pet-eating controversy. All I see about Harris is complaints about her unimpressive performance during interviews.

    2. Altoid

      It very much does come down to individual states because of the EC, but there seems to be a general educated consensus the past several cycles that Ds need about a 5% polling edge nationally in order to be considered likely to win.

      In 2020 Biden won the national vote count by about 7 million, which was a 4.5% margin, but a shift of less than 100,000 across three states (maybe a lot less, I'm not finding the number now) would have lost him the EC. Popular vote margins for Ds tend to get padded by lopsided counts in NY and CA, solid blue states with big populations where the in-state margin has no effect on EC totals. Conversely Rs win several states by really enormous percentage margins but they're much smaller so they don't affect the popular vote count the same way.

      The main reason for tying EC outcome to polling margins is, as I understand it, that there has been a trackable relationship between national trends and outcomes in individual states, and especially in the purple ones. WV and UT will be hugely R no matter what happens nationally, but MI, PA, WI etc are closely split enough that national polling can indicate which way they seem likely to end up going.

      I think there's been discussion this year about how some new patterns might affect the percentage margin that Ds need. The recent surge in new-voter registrations, for example, might mean (if they end up voting) more first-time voters in key states, who aren't likely to get included in polling and are considered likely D votes in aggregate. Things like this are a big part of what makes polling hard to do well.

    3. Jasper_in_Boston

      CW says Trump's a fairly strong favorite within 2 points of Harris. Her chances increase as she expands her margins beyond that, and seals victory if she gets much above 3 points.

      Harris is obviously headed in the right direction, though I reckon Kevin's overstating things a bit given the margins of error of the polling and Trump's formidable Electoral College advantage (thanks, Madison!). Also, we're surely now beginning to see the effects of the debate: Harris is getting a nice bounce from that. But it's certainly possible that will fade, and polls will tighten.

  1. dante

    zero percent chance harris wins by more than 5, much less 8. very likely she'll be in the 2-3 point range, which may or may not be enough to overcome the electoral college bias

    1. KenSchulz

      Certainly greater than zero. Democrats underperformed the polls in 2016 and 2020, but overperformed since Dobbs. Compared to the last two elections, TFG is running a much less effective campaign. He can’t focus — now he’s dabbling in cryptocurrency?!

      1. Jasper_in_Boston

        Democrats underperformed the polls in 2016 and 2020, but overperformed since Dobbs.

        Not so. The polls were highly accurate in 2022, and if anything Democrats very lightly underperformed what the polling suggested their congressional vote would be.

        I think many people confuse "Dems had a very good midterm relative to what typically happens to the White House party" with "Dems over-performed the polls." But the latter isn't borne out by the numbers:

        https://fivethirtyeight.com/features/2022-election-polling-accuracy/

        1. Joseph Harbin

          Leaving aside what polls said about the 2022 elections, a lot of political forecasters predicted a red wave in the midterms, and they were wrong.

          The red wave was never likely, however, and the reason for that was what happened in 2020. Biden's victory over Trump did not translate into a large D majority in Congress. Without many blue reps in red districts, a red wave was hardly possible.

          (D's lost 13 House seats in 2020 despite winning the national pop vote by a margin of 3.1%. In the 2022 midterms, they lost only 8 seats despite losing the pop vote by 2.8%.)

        2. jdubs

          538 and the polling in general were predicting a large red wave through the summer. Polls did narrow a bit closer to the election, but 538 was still predicting a GOP capture of the Senate and a much larger takeover of the House.

          Also lots of similar results in the one-off, individual elections that have happened the last several years.

          538 has a strong interest in presenting 'analysis' that shows how accurate they were after the fact.

          But its certainly accurate to say that Dems overperformed.

          1. Jasper_in_Boston

            But its certainly accurate to say that Dems overperformed.

            It's definitely not accurate to say Democrats over-performed the polls, in 2022, no—at least as these things are usually analyzed (the article I linked to above from 538 presumably cites polls on the eve of the election; it's possible in say, August, the polls were more positive for Republicans. I'd have to double check).

            But sure, pundits and media outlets will frequently make non-evidenced predictions.

            1. KenSchulz

              I stand corrected. What happened in 2022, according to the 538 article, was that the Democratic bias of the polls was much smaller than in 2016 or 2020 (as measured by weighted average absolute error). By the way, ‘bias’ to a statistician simply means nonrandom error, not animus.
              On the other hand, the percentage of poll leaders who ultimately won their elections was lower than in previous contests. This wasn’t broken down by party, perhaps because outcome is a lousy measure of accuracy, as the article notes. And it’s not clear to me if this was based on district or generic-ballot polls, or both.
              From the article: “In fact, 55 percent of the polls we analyzed for last cycle were closer than 6 points.” So, 45% were off by 6 points or more. Oy.
              Lastly, I take strenuous issue with Nathaniel Rakich’s lauding of several polls for their accuracy in 2022. We know that high scores on performance measures are likelier than not to have been inflated, and low scores deflated, by random error. From one cycle, we can’t know the relative contributions of actual accuracy vs. luck. Next time out, random error is equally likely to help or hurt performance, that is what the term ‘regression to the mean’ describes; and it is why an investment prospectus carries the caution “Past performance is no guarantee of future results.” Rakich is on more solid ground using letter grades, which 538 bases on multiple cycles.

    2. kahner

      i think kevin is saying if polling shows her 5-8 point ahead nationally then it is very likely she wins, but not by than margin. he's factoring in the possibility that trump outperforms polling and/or national and swing state polling don't correlate much.

  2. tomtom502

    Nate Silver gives lower odds. I know he is reviled by many but I think he is both most rigorous and has the best track record.

    You have to subscribe to see it but he has a full statistical breakdown of Trump's electoral college advantage.

    1. Joseph Harbin

      Whatever you think of his record, his methodology this cycle appears to be problematical and his relationship with Thiel raises many questions.

        1. Joseph Harbin

          His arbitrary dinging Harris a couple of points to cancel a convention bounce that was never there to the degree he compensated for it. His poll averages including crap like Patriot and Trafalgar. His prediction model that had Trump's chances well into the 60s even when Harris was leading by about 3% nationally in poll averages and a slight lead in enough swing states to give her an edge in EC votes.

          I don't pay much attention to the guy but enough to see him dismissive of criticism from others who he impugns as dumb liberals who just don't like his results. I don't think he's an honest broker or a disinterested data nerd who just wants to understand the race. Maybe he was once upon a time. But he seems to me very interested in making news, in being provocative, and that may explain why his analysis at times seems way out of line with where other polling experts see the race.

          1. tomtom502

            The convention bounce was programmed before Biden stepped aside, based on historical patterns. Part of being fair is that you stick with the model and don't make changes on the fly.

            The polls you mention are adjusted for their lean.

            if he was systematically biased it would show in his record, but his record is good.

            1. Jasper_in_Boston

              but his record is good

              Yep. And Silver is also the most Democratic-friendly, if Democrats knew what was good for them. I want Democrats scared (they should be, given what's at stake) in late October.

              Nobody had the slightest fear of a Trump victory in 2016!

      1. dante

        he has no relationship with thiel, this has been an asinine accusation and a rare example of left-wing misinformation. frankly it's an accusation transparently based on motivated reasoning. liberals don't like that his model shows trump winning, so they fabricate shit about his integrity even though he has made it clear that he thinks trump is unfit for office and that he is voting for kamala.

        1. Joseph Harbin

          "he has no relationship with thiel."

          He works for a company funded by Thiel.

          Does that prove he's corrupt? No. As I said, it raises questions.

          I didn't "fabricate shit about his integrity" so don't fabricate shit about me. Fuck off.

          1. Joseph Harbin

            In the interest of comity (bc that's the kind of place this is), I'll say one thing in favor of Thiel. He believes the election won't be close. I think he's right about that. I happen to think it's more likely to be Harris by a good margin. Recent poll movement in her direction jibes with that.

            My guess is not based on crunching numbers so much as taking a longer view on where we are historically and how the two (or four) candidates are campaigning.

            Could be wrong, could be right. We'll know in November.

        2. jdubs

          Recognizing that a relationship and dependacy form when you work for someone is an accusation transparently based on motivated reasoning? Lol, sure.

          You seem to be quite the motivated reasoner yourself.

    2. kahner

      i don't revile silver and think he has some good insights and rigorous methodology, but i do think some aspects of his model are probably causing it to underestimate harris' odds.

  3. Joseph Harbin

    KD:
    "My rule of thumb is that Kamala Harris needs to win the popular vote by 4-5% in order to win the election."

    G. Elliott Morris (formerly of The Economist, now chief poll dude at 538):
    "The current gap between the national popular vote and Electoral College majority in 538's forecast is about 2.1 points. That means Harris needs to win by 2.1 to be favored to win the election."

    Other Morris tweets say Harris is now +4.2 in national polls, and less in swing state polls (which lag), and she has a "slight lead" in the race.

  4. D_Ohrk_E1

    If you look at this election through a different lens, then the goal is to win the election by an Obama-level electoral college landslide. 332 - 206.

    It is possible to reach over 300 electoral college votes if Harris is able to sweep Arizona, Georgia, and North Carolina while holding onto the Blue Wall (=319 EV), even if she can't win Iowa, Ohio, and Florida.

    And I know no one believes Texas and Alaska are reachable -- that they're fools gold -- but surely a few billionaires can pour some money into these states to at least win statewide offices and then maybe enough to tip the election to Harris. Use Arizona as the template.

    And maybe the Harris campaign can explain to voters that 2024 isn't about handouts to different demographics, but a separation of visions of who we are and what American values are.

    1. lawnorder

      Alaska should be reachable; they've twice elected a Democrat to their only House seat. On the other hand, it's only three votes. The election would have to be VERY close for those three votes to matter.

      1. D_Ohrk_E1

        But would the Trump campaign spend what meager $ they have to defend Alaska? It could be a relatively cheap way for the Harris campaign to win a state that was thought to be solid red, and it'd nudge Alaskan representatives and senators further to the center in a post-Trump world.

  5. Steve V

    We are about to enter "October surprise" season. Who knows if something one of the campaigns springs will stick and change the direction one way or another.

    1. Batchman

      With the growth of early voting (which has already started in some states), the impact of an October surprise is less than it used to be. I don't consider this a good thing, btw.

  6. raoul

    Trump will get his 45-46%. I doubt he will reach 47% as he has in the past for several reasons including abortion, youth vote, race, old people dying off (Covid), and his own tiresome personae. He may not lose a lot but his ceiling is lower. The only question is how far north of 50% Harris will get. My current very realistic prediction is 51-45. This should be enough to carry the swing states. However, Trump could still win with a Comey letter type event that would lead to a 50-46 loss in the popular vote but still enough to win the EC. Recall that Biden won by 4.5% but a swing of 50,000 votes in three states would have cost him the election.

  7. ScentOfViolets

    Put me down for Trump playing Godfather with an oxygen mask in a wheel chair if the spread widens to a sustained six percent or more. I don't know about anyone else here, but I firmly believe that Trump's overwhelming motivation for running is to stay out of jail. The grift bit is just his nature.

  8. Batchman

    Don't forget that, unless Harris wins in a landslide, Trump will certainly contest the election, very likely by trying to force it into the House of Representatives to resolve an Electoral College deadlock. Unless the Democrats win the House in November, this will benefit Trump greatly. Whether Mike Johnson will go along with such a move is being kept deliberately vague.

  9. pjcamp1905

    Also, to my eye, the uncertainties no longer overlap so this is likely a real lead and not a statistical artifact.

    I don't think a new January 6 is very likely with any sort of win for Harris. Last time, I'm pretty sure everyone involved in the insurrection believed they were patriots and nothing would happen to them. Now they've seen a thousand or so go to prison, and I suspect that speaks pretty loudly to them. Being a convicted felon carries lifetime disadvantages and anyone who has ever filled out an employment application knows it.

Comments are closed.