In his usual eloquent way, Walker said, “We can’t blame no one.”

I didn’t make that up.

One thing we saw in the 2022 mid-terms was that the death of polls had been, if not greatly, at least somewhat exaggerated. In general, the pollsters did a pretty good job this year. In this particular race, RCP summarized the polls with 51.0 for Warnock – with a 3.7% edge and 1.7% undecided. 538 said 50.2 for Warnock, with an edge of 2.0% and 1.8% undecided. Splitting the undecideds in the same proportion, that would result in a 51.0-49.0 win for Warnock using 538’s numbers, or a 51.9-48.1 Warnock victory using RCP’s numbers.

If you average those two summaries, you get Warnock with 51.45 and Walker with 48.55.

The actual result was Warnock 51.4, Walker 48.6. That’s about as accurate as polling can ever get. (In fact it’s MORE accurate than polling can really get. Getting that close is just a matter of luck.)

26 thoughts on “Herschel Walker fumbles

  1. To quote Georgia radio host Erick Erickson, “A red state now has two blue senators because of one orange man.” Every other statewide GOP candidate in GA won by at least 5 points. Kemp who was probably the GOP politician most hated by Trump who wasn’t named Cheney won by 8. Secretary of State Brad Raffensperger, who refused to find the Trump votes that weren’t there won by 9.

    I’ve been a registered Republican my entire adult life. It is my fervent hope that the combination of the last 3 elections (2018 – 2022) finally convinces enough of the GOP base that following Trump and Trumpism is not the key to electoral victories, but will instead only lead to defeat. If that happens the GOP will have the chance to retake the Senate (especially with the 2022 Senate map), hold the House and capture the presidency. I happen to think that would be best for the U.S., provided we are talking about victories by mainstream Republicans. But regardless of whether you agree with the GOP ideologically, this country is far better off with 2 functioning political parties rather than having one implode.

    1. Keep hope alive! 😛

      Trump gave the medal of freedom to Jordan/Nunes 😮 so they pretty much are obligated to continue to kiss his ass, eh. And it is fun watching the party who usually falls in line implode. ok, ok, they have, for the most part, fallen in line behind their lord and savior, Trump!

      It will be fun if by some miracle someone other than Trump gets the 2024 Rep nomination watching what Trump will do. Katy bar the door …

      Either way the party is probably screwed!

  2. If you want someone who does election coverage using hard data analysis, then start following Chris Bouzy, he’s on twitter @cbouzy, he was more accurate than Nate Silver, any other big name pundit or any of the polls by miles.

    He got a large portions of the midterms correct by analyzing the data from the past few election cycles, and used that in his analysis.

    He blew 538 out of the water. So much so that many of the big names were coming at him for making them look bad this year.

    He just called the GA runoff for Warnock by 3 points (he gave the spread 3-4). So while he doesn’t hit them all (Boebert is one he missed, and he didn’t account for the NY & CA election fiascos, due to their special issues associated with them)

    He does a pretty good job with taking the raw data and coming up with solid results and not tea leaves, phone a friend and guesses like Nate and co. seem to pull out every cycle.

    1. “He got a large portions of the midterms correct by analyzing the data from the past few election cycles, and used that in his analysis.”

      Yes, historical analysis helps, but major polling services, as a rule, don’t poll House races because swing races are a crap shoot. The Trump effect pro and con foreshadowed everything and was hard to determine, especially since it has continued diminishing returns.

      Re: Bouzy if he accurately predicts say (3) election cycles in a row folks will probably start to pay attention. Zogby was the only “major” pollster to predict the 2000 potus election, but “they” were wildly inaccurate after 2000.

      Even a blind squirrel occasionally finds a nut …

      Voter enthusiasm er GOTV is always hard to predict.

      Interesting Obama won Ohio twice, but has been solidly Rep the last two potus cycles. Probably had to do with 2008 Cheney/Bush effect and the Robert Alphonso Taft III effect. The yin/yang of politics. Digressing.

  3. I watched Walker’s concession speech and it was probably his best of this whole cycle. Gracious, positive and uplifting, and surprisingly concise and coherent. Had we seen this guy sooner the election might’ve turned out differently.

    1. I don’t agree with that at all. The one positive I took from it is that he showed a little humility, but even in that process he humble-bragged about his “athletic awards, business awards, Heisman.”

      In general he was rambling, trite, repetitive and (as always) nearly illiterate.

      Warnock’s victory speech, on the other hand, was brilliant. I know he’s a preacher and he’s trained to move an audience, but even with that understanding, I was wowed. His speech was powerful, with more than a touch of poetic resonance. Maybe he was just reading from the prompter, I don’t know, but however he did it, it was impressive.

      1. I saw it on CNN and some talking head said it was tremendous, what a great guy, etc. Jake Tapper responded, “He doesn’t get brownie points for doing something that’s customary.” I thought it was an awful speech for all the reasons Scoopy listed. He lost, so his bruised ego made him mention his football & other accomplishments, which are irrelevant in terms of why he’s standing at that podium. He was just a wannabe mini-Trump who got exterminated before he could do much damage, so it didn’t make any sense for him to give a concession speech repeating phony cliches like “Never give up hope! Keep believing!” Never give up hope in what – Trump zombies getting elected by morons?

      2. Maybe I was overgenerous that he didn’t say he had been cheated and the whole thing is rigged. I figured he’d give such a terrible speech that I was surprised that it was at least acceptable and didn’t get into werewolves and vampires.

        1. While he was speaking, I was thinking, raise your hand if you really believe his ‘God is good!” rhetoric. This is a man who fucks anything that moves, and if she has a baby, he quietly sends a check for the abortion and moves on to the next one. But one of his biggest running stances was that he’s vehemently pro-life, so it’s unclear exactly what bible he’s reading. Probably the one by Anton LaVey, it seems.

          Bottom line, he was undone by his own stupidity & hypocrisy, as well as his family’s Twitter bombs against him. His concession was only good to the extent that he didn’t fuck anyone onstage, or claim to be a cop, or demand that the Constitution be ripped up so that he can win. I was actually pleasantly surprised that he didn’t shout as he left the stage, “OJ is innocent!”

          1. Georgia wins in two ways: Hersche isn’t their Senator, and he moves back O to Texas.

            (Also known as the Oz Bonus.)

          2. The one thing that Trump has really brought to the foreground in political rhetoric is that audience reaction of “Is he really that clueless, or is he just dishonest?” Walker kind of took that to a Trumpian level. I was never sure whether he really believed some of the things he said, because … I dunno … maybe he really is that dumb.

  4. Still scary that over 48% of Georgians thought this lying, incoherent concussion patient was the better man for the job.

    1. True dat.

      It really reflects how tribal we have become.

      To be fair, the good people of Georgia have always tried to pick the candidate with the lowest IQ.

      Herschel “why are there still apes” Walker is probably no dumber than Hank “Guam will capsize” Johnson.

      And Brian Kemp is no physicist.

      1. Their slave master once suggested that Americans drink disinfectant to cure Covid, so that explains a lot. Walker displayed a fake police badge, which is something Trump is still kicking himself for not thinking of. They’re trying to appeal to stupid, ignorant people, so the more mindless shit they spew, the better off they think they’ll be. And that was working for a few years, until Trump went so far off the map that he ruined it for all the others trying to walk in his phony footsteps, as Walker & Kari found out the hard way. Walker is his own special kind of stupid though, which is why he pronounces elections as “elecsha”.

    2. I agree completely, Nature Mom. While I am glad the Dems are ekeing out victories, the narrow margins they win by are frightening. What does that mean for the future? It’s not like Fox News and their kind are going away, and their smooth, simple lies are more appealing than the truth, which is complicated and hard to even be sure of.

  5. Repeating, polling is a science, but there are good “scientists” and bad 😮 which is why Nate Silver/538 is constantly re-evaluating pollsters 24/7 for bias/bad methodology. Nate/538 take several factors into consideration including a state’s voting history.

    btw, my above statement has a +/- 4% margin of error. 😛

    Yielding back the balance of my time …

    1. And again, repeating: the only thing that really matters in evaluating the accuracy of a polling company is their actual accuracy.

      Nate’s, or my, personal biases about proper and improper scientific methodology are utterly irrelevant, which is why, as I have pointed out, he gives A ratings and excessive weight to some of the consistently least accurate companies because he feels their methods are proper science.

      If you are truly a scientist, eventually you have to stop evaluating results based on what should work, and start evaluating based on what does work.

      If you do not make that adjustment, you become Lysenko.

      So, when he starts to develop a strong statistical correlation between the weight he assigns to pollsters and the actual performance of those pollsters, then you can claim he’s employing science. Until then he’s just a guy assigning more or less random values based on what he believes to be proper scientific metholology.

      Lysenko would be smiling from his grave.

      Oddly enough, he does employ a step that is supposed to adjust the results for bias, but then once he makes that adjustment, he still gives some pollsters more weight than others. The two steps seem contradictory to me.

      If I were doing it, I would skip step one and just weight each pollster the proper amount to produce the right result retrospectively. (In other words, in a simplified example, if I had just weighted Survey Monkey 90% and Monmouth 10%, I would have produced the exact correct result.) Then I would evaluate polling results by state to see if some pollsters are particularly ineffective in certain states, despite consistency in their methods. In other words, I might give the NY Times a quadruple weighting in Alabama and no weight at all in Minnesota, because despite using all the proper scientific polling methods, they are wildly ineffective there. (Not real examples. Just writing in the names of random states.)

      As I see it, there is a missing piece between public opinion polling and predictive likely voter modeling. It may be that people like to deceive pollsters, or it may be that there are great last-minute psychological pressures that significantly alter a voter’s thinking when he finally has to choose black or white. I don’t know. That’s not my field. But if it were my field, my question would be “How can I use my past errors to better convert polling data into a predictive model for actual voter behavior?” I think I would start by trying to develop an instrument that attempts to identify a voter’s likelihood to do any of the following (1) not actually go to the polls when he says he will (2) deceive a pollster (3) change preference at the last minute. If any voter is determined likely to do any of the above, I would then eliminate such a person’s responses from my sample group.

      1. Your above gobbledygook notwithstanding, again repeating, Nate is not a pollster ie he evaluates polls to give you a percentage on what he thinks should happen in any given race.

        What many analysts didn’t/couldn’t calculate in the 2022 midterms was the youth vote turnout, effect of Dobbs scotus decision and effect of Trump’s fascist lemmings wanting to end democracy because “they” were sore losers re: the 2022 potus election.

        Indeed, Trump has become more of a negative re: the Rep party ~ ymmv. An ironic anecdote Dems lost control of the House mainly because they lost (4) seats in both CA & NY as maybe both Newsom/Hochul were drags on the Dem ticket as well.

        Lemmings on both sides who vote a straight ticket aside, in a nutshell Kemp won GA quite easily, but Walker was a bridge too far. Candidates still matter in “swing” states.

        1. Correction ~ sore losers re: 2020 potus election, but this applies to every potus election from 1992 onward. IOW Rep “hierarchy” as a whole feel they should win every potus election regardless. If not they will provide fake news on why not.

        2. As I noted, there are statistical methods that would allow Nate to assign proper weights to each of the pollsters, but he doesn’t use them. He uses an arbitrary evaluation of their methodology. Because of this, his inaccurate weighting distorts his calculations, and his poll averages are therefore no more accurate than RCP’s, even though they just take a straight average and make no effort to discard the chaff!

          At some point you have to let reason enter the argument and stop saying, ” Well, I know Monmouth always produces the worst results, but gee, they’re just so scientific.”

        3. I’m skeptical that anyone even whose field it is to do the job UncleScoopy envisions would be able to develop a model that actually works. Generally, Michael Crichton’s CalTech speech 2 decades ago put his finger on something important that has been borne out. Complicated models of complex phenomena are self-defeating.

          Even in the “hard science” of particle physics, the very successful theory of the strong nuclear force, there’s still not a good understanding of the low energy details. Like, a proton is complicated & chaotic at its surface, where the proton ends rather abruptly. (Yes, this is a low energy detail.)

          It’s thought that the chaotic mess reaches a critical point & coalesces into a quantum condensate, that then permeates the rest of empty space. The point being that a particle physicist can use the theory to make a simulation that replicates what she knows about what actually happens, but there isn’t a deep understanding of why it should end where it does nor a clear picture of “what it’s like” on either side of the transition (the outer edge). The math is unsolvable, so all we can do is plug in our actual data. Experimental data is the theory’s spackle.

          Econ, the—what was it? evil? dirty? ugly? oh yeah, dismal—science has woken up to the failure of macro theory. That there is no micro foundation that anyone has discovered for it. That, as Scoopy says, before we can begin to theorize about such a beast as “the U.S. economy” (let alone the world’s), we must gather data. Tons of data. Then we look for patterns, we play back the tape until history confirms those patterns, and only then we can start to cobble together our first meager theories. Needless to say macro hasn’t gotten to that point yet.

          Yes, I think someone expert in the methods Scoopy has in mind can conjure up a weights system that will not break in the near future. But I think pollsters avoided a miss in these (2022) midterms as much by luck—that is, that confounding factors washed each other out more or less randomly—as by good & accurate methods.

          Scoopy’s expert would be concocting an ad hoc machine that will fail inexplicably & all of a sudden. That’s what I think. You’ll disagree with me, but it don’t matter. It just ain’t a-gonna happen.

          1. You’d be surprised how well some phenomena can be modeled in the social sciences.

            When I was working in international strategic planning with Shell, I was able to build a model to predict how many profitable convenience stores they could build in a given market, given the results in the first 1-3 stores, provided that I was able to choose those model stores for them. (It turns out that organizations, in some respects, behave like organisms, except that their behavior is modeled with a lower degree of certainty.)

            And when I modeled the expected sales per site in the gas stations that became c-stores in Norway (where I developed their c-store network), I was able to predict the post-conversion sales so accurately that they thought I had supernatural powers.

            And if I knew last year’s weather and this year’s, I could predict the year-on-year sales increases quite accurately. The line managers hated this, because they would come to a chart meeting proud of a 10% increase in July’s year-to-year sales, ex inflation, and I’d tell them that if weather conditions had been identical both years, their increase would have been 2%, and it would almost surely fall back to 2% in August, all things being equal.

            On the other hand, there were some problems in retail strategic planning that defied modeling, or at least they defied my attempts, and some of those lessons must apply to polling. I never was able, for example, to properly adjust data that required input from store managers or customers. It turns out that when you ask customers how often they come to your store and how much they spend, or even how close they live to the store and how old they are, that they either have no idea or bullshit you. I suspect that pollsters must face the very same problems.

            I’m guessing that a certain small percentage of customers who answer “Trump” or “Biden” don’t actually know who those people are or what they stand for. One can never underestimate the ignorance of the average person, especially the young ones.

            And I’m guessing that there is a very high percentage of fibbing about whether they are likely to vote or are even registered to vote, and I’m not sure how pollsters adjust for this, or whether they have any system to detect it.

            The applicable Scoopy rule here is “people are totally full of shit.” I remember that a reporter for the Miami Herald once interviewed a random sample of Cuban exiles on the streets back in the 60s or 70s, and asked them what they left behind in Cuba. When he extrapolated his results to the total population of the exile community, it turns out that they owned so much land that Cuba must have been the world’s largest island, and being larger than Australia, should probably be considered a continent.

      2. And then there’s the issue of who actually talks to pollsters. Most people I know immediately hang up on them even if they answer the phone.

Comments are closed.