This is just flatly wrong.
Wow. “This is just flatly wrong.”
Sounds like Argument Clinic time. “No it isn’t.”
These are no more than wild guesses. It’s not reasonable to expect strangers on the internet to assume the political predictions of an internet stranger, barring some sort of model with a record like Nate Silver, are anything more than wild guesses.
Same goes for anyone else proclaiming that certain candidates are or are not “electable”. Just wild guesses. It’s way too early to have any idea about this.
It’s not too early, when you have evidence from a recent general election like we do with Warren.
Are you claiming there was logic or evidence in the post of yours I was responding to? :dubious:
Yes it is. Obama lost his first congressional election (the primary, anyway)… that didn’t mean he was bad at politics. Maybe he needed to be in the right situation. Maybe the same for Warren.
I’m not saying Warren is an A+ politician – I doubt she is. Those only seem to come around every 20 years or so (or maybe every 16 years, if we look between '92 and '08). But I think she’s got skills, and they’re only growing. Her last election under-performance might have just been due to inexperience in running, and/or a sub-par team. She’s doing much, much better this time. Watch her on camera – she’s very sharp, and even funny sometimes. Not Obama or Bill Clinton-level, but a lot better than Hillary, IMO. So far I’m very impressed with her political skills. One sub-par election in MA doesn’t mean she’s a bad politician. A steady rise in the primary shows real political skills.
George McGovern had “real political skills” too. Being catnip to primary voters is a real skill, but it can be the opposite of the kind of skill needed for general elections.
She underperformed in her previous campaign as well, and at the beginning of the primary season she was Massachusetts voters’ third choice for presidential nominee.
I normally try to keep my subjective impression out of this discussion, but since you specifically mentioned watching her: I do, and I find her extremely grating every time. It’s not about her gender, because I did not feel that way about Hillary and do not feel that way about Kamala. I don’t like Warren’s vocal intonation, her preachy speaking style, or the cross look she gets on her face that her staff has clearly essayed mightily to train out of her, to inconsistent effect.
We’ll see.
Fortunately Trump looks like a walkover at this point. 2022 and 2024 are what I worry about.
Well, I said a lot of stuff. If it’s “just flatly wrong,” I’m sure you can dismantle it, thereby demonstrating the absence of logic or evidence.
If you’re going to just assert I’m wrong, without attempting to demonstrate it, I can only conclude that your assertion is all you’ve got.
Yeah, I struggled to find a word there. You’ve noted your interest in Warren often, but the assumption that certain people should be shown with disregard for the written perameters also shows an interest or an assumption.
Huh. I missed that. Was there a reason for that?
I saw that. I wondered why there was an assumption that anyone would be asking about Beto when his average was so close to the 2% perameter you set. It seemed reasonable to think he’d fall below it on any given week.
Yup, you were right. I wonder how many of your predictions show up in your display of your numbers.
Exactly right. When people expect to see patterns in the data, they often magically show up. It’s especially pronounced when you’re the one creating the data, reporting the data and interpreting the data. Your predictions often magically show up in the data.
Ah, OK, this is a rule of displaying the data that was not transparent. Makes me wonder how many of these rules there are like this with assumptions that are not communicated.
Thanks. Your sarcasm here belies your objectivity.
When people look at reports or lists of numbers, they have a tendency to think that those numbers are somehow evidence of something. They especially have that tendency when there’s a summary of those numbers that looks like a conclusion.
I was just listening to Eric Weinstein about a trick called magician’s choice. It goes like this:
What people decide to show and not show affects the tendency of people to make conclusions that they think they’ve drawn themselves. Sometimes this can happen without any intention on the person showing the numbers. Mostly and in this case, it’s no big deal.
But when it’s done by people (intentionally or not) on a national or international platform, it can have effects that people aren’t aware of.
The number either declined or it didn’t. Anything else is spin. It takes more than one person for a conspiracy. No one is claiming any kind of conspiracy here, especially not me.
Yer funny.
You could argue that, by showing Beto the week he dropped under 2%, I was disregarding the written parameters. But from that week on, it was a written parameter.
I told you already. (Post #110.)
Well, good. Then you know that I set a precedent then, and have followed it consistently ever since.
How many predictions have I made? Not that it matters: one of the advantages of having stuck to my rules is that I’ve disproved the only predictions I can recall making: (1) the Warren/Sanders inversion, where once Warren got ahead of Sanders in the polls, she was going to start sucking support away from him - and instead, he held steady or gained, until this past week; and (2) my observation a few weeks back that, despite my hopes, Warren wasn’t going to break 20% anytime soon.
And which predictions were these?
I’m serious. Produce my predictions.
But let me repeat this:
I think I’ve spelled out my rules clearly enough that anyone can follow them and replicate my results. Because this is what scientists do. Shall we go over them again?
-
The eligible polls for a given week are those that satisfy the following conditions:
a) They are national polls, not state polls.
b) The midpoint of the time they were in the field is within the past 2.5 weeks. (Started off at 2 weeks, but a few weeks ago i said I was going to extend it to 2.5 weeks.)
c) The pollsters have ratings of at least C- from 538.
d) There is no more recent poll by that pollster that is eligible per the other rules.
e) If several versions of a poll are reported at 538, LV > RV > A: that is, I’ll take the poll of likely voters over registered voters, and registered voters over adults. -
I take all the eligible polls. I multiply the results of the A-rated polls (A+, A, A-) by 5, I multiply the B-rated polls (B+, B, B-) by 3, and the C-rated polls (C+, C, C-) by 1. I sum up all those numbers, and divide by the sum of the weights. So if I’ve got two A-rated polls, three B-rated polls, and two C-rated polls, I divide the sum by 21 = 25 + 33 + 2*1.
-
And then I show the numbers on everyone who is above 2%, or was above 2% the previous week.
There. You have all the information you need to check my work. Knock yourself out. If you think I’ve been screwing with the numbers, now you can prove it.
Put up or shut up.
RTF, when we get into actual discussions of opinions, I often find myself disagreeing with yours.* But I don’t think you are guilty of cooking the books on these numbers and I appreciate your posting them.
*To wit:
I quoted a single sentence of yours which is also just an assertion: “Sorry, but none of those candidates whose polling numbers are scraping up against zero - none of them are ‘electable.’” And I do think your claim is flatly wrong, but there is no real way to prove it, any more than you can prove what you said.
Mark Shields on PBS Newshour earlier tonight sees it just the way I do:
Just to be clear, I’m not accusing you of intentionally falsifying the numbers. I also don’t know how I would prove you wrong. There are way too many moving parts, some of them requiring judgment and others demanding error-free inputting and fomulas with no cross-checks.
Some of the polls are not super clear about which numbers I was supposed to pick up. I tried to find the list of Democratic voters. I will say that the methodology to use likely voters over all adults has the potential to move the numbers toward established candidates.
That said, just for grins and giggles, I tried to replicate your report. Here are the polls I used with the weighting I gave them.
Emerson (3)
Zogby
Harris poll
Ipsos (3)
Monmouth (5)
IBD (5)
Morning Consult (3)
Quinnipiac (5)
YouGov (3)
HarrisX
Here are the results I got for Oct. 9, 2019.
Biden 27.1
Warren 25.1
Sanders 15
Buttigieg 5.3
Harris 4.2
Yang 3.4
Here are your numbers for the same time.
Biden 26.2
Warren 25
Sanders 15.3
Buttigieg 5.2
Harris 4.2
Yang 3.1
I’m not sure why my Biden number is so high. It could be input error on my part. I only spent a couple hours doing this. But the raw numbers seem to roughly match the numbers you gave in an earlier post. It’s also possible that a calculation or weighting could be faulty. But the other numbers are pretty close.
I’m also going to guess that the Zogby poll made some difference because you didn’t mention it and I didn’t see a reason to exclude it. Short of going over every number and calculation, I can only guess where the differences lie.
Shields and Brooks agree on one point. With news now focused on impeachment rather than the Democratic candidates, the race “is frozen.” This will favor the current front-runner Biden (and perhaps Warren).
Also available to Biden is the argument that Trump is so afraid of Biden he commits felonies to neutralize him.
The most recent national Zogby poll I could find at 538 was in the field September 16-17. Median field date, 9/16.5. 2.5 weeks (17.5 days) later was October 4. So it aged out between 10/4 and 10/5. It was included in the prior week’s average, but not this week’s.
Harris Poll, HarrisX, and Harvard-Harris are all the same Mark Penn-owned outfit; they only get represented once. HarrisX is updated daily, so that’s the one I use.
Other than that, you used the right polls, and assigned the right weights.
Including Harris Poll and Zogby would have changed things only slightly due to their low weights. They had Biden at 31 and 28, respectively; both had Warren at 17. Including them should have raised Biden’s average by about 0.2%, and lowered Warren’s by 0.5%, rather than lifting Biden by nearly 1% and leaving Warren nearly unchanged. So I’m thinking arithmetic errors crept in somewhere.
The point is, I have a pretty rigorous and transparent process here, one that I went out of my way to ‘me-proof,’ so to speak. So yeah, I resent innuendos like
and
OTOH, I’m more amused than anything else by your suggestion that I had a thumb on the scales for Cory Booker, a politician that I quite honestly have a hard time working up an iota of interest in.
FWIW, I think my decision to include Beto (and then anyone else) the first week after they drop below 2% is looking pretty good: that first week might be a fluke - and was, the first time Beto and then Booker dropped below that mark; they were both at or above 2% the following week. So the question is, which makes more sense - a blank spot for that candidate for that one week, or the actual number?
No, it’s understanding that minor fluctuations shouldn’t be considered real movement.
Thanks!
Excuse me, but WTF?!
It’s true that you quoted just a single sentence, which, standing alone, is no more than an assertion. But the assertion in question was* the conclusion of a three-paragraph argument*. Your assertion* was your entire post*.
So what you’re saying is that, by quoting just my conclusion, you don’t have to deal with the argument that led up to it - and that your one-sentence post, consisting of nothing but an assertion, is just as good as mine.
That’s total bullcrap, and you know it.
This is an important point, so thanks for making it. Polls have sampling error, and while taking the average of 6-8 polls causes a certain amount of that to cancel out, it doesn’t make it entirely go away.
If we look at the numbers to date, there’s plenty of evidence of that, too. Biden dropped from 29.8 to 26.5 one week, then was back up to 28.5 the week after. Warren went from 16.8 to 19.0 - woohoo, breakthrough!, from a Warren fan’s perspective, right? - but down to 17.6 the week after. Between 8/14 and 8/21, Sanders went from 17.1 to 15.2; his fans are finally moving to Warren, right? Oops, next week he was back to 16.9, and until this week, he’d never been below 16.0 since. Maybe his drop to 15.3 this week is real, maybe it’s just statistical noise; we don’t know until we see more polls.
So yeah, reading anything in to Yang’s increase from 2.9 to 3.1 is a bad idea.
I took out the Zogby and Harris polls. Then I went back to your numbers about Warren and Biden to back into your numbers. I used the Democratic voters numbers for the IBD and Ipsos polls when I was supposed to use the total voters numbers, it seems.
The numbers match up pretty closely to your numbers except Yang’s. But meh, once the Emerson 8 ages out, the point is pretty moot, so I can’t care enough to chase it down.
When a question about the numbers being displayed out of order gets a response of post #110:
I think it’s fair to question what else in the process is affected by personal prejudice. Just because anyone can replicate your numbers doesn’t mean that anyone will or should. Duplicating effort is a waste of tme. I did it because I thought it might be interesting, which it was.
Among other things, it was interesting to me how the poll questions or the format of the poll or the methodology of the poll has the possibility of changing the direction of the results. In some of the polls, there are a LOT of questions about everything. The featured question of the candidates is buried in the middle, and sometimes not all of the candidates are listed. That gives the people with name recognition even more likelihood of showing up in a poll than someone without it. If there’s that much time to ask so many questions, it seems like there should be enough time to ask about all the candidates that RCP features. I didn’t study it closely, but I think I ran across some polls that didn’t.
It’s not about Cory Booker. It’s about the assumption that anyone cares if he shows up in the numbers enough to make special rules about it. Would special rules be made up equally for other candidates that you have personal prejudice against if there wasn’t one already?
By that logic, none of the polls’ movements are worth examining from week to week. All of the polls’ changes are within the margin of error in any given week and if even the aggregate creates some sampling error and directionality can change week to week until the race is over, then there’s not much to say about any changes week to week.
I was amused that when Warren overtook Biden for a week in a couple polls, you made a point about it, but when Yang overtook Harris in the Emerson poll, you made it a special point to note that it was within the margin of error.
But since all of this is mostly just speculation and entertainment, it’s all good.