Bush leads Kerry 50-45 percent of nation-wide RV’s, with 2 percent for Nader, according to an ABC News/Washington Post Poll, conducted 10/1-3.
Thanks for signing up!
TDS Strategy Memos
Latest Research from:
tds-wwc-roundtable
Editor’s Corner
By Ed Kilgore
-
October 2: The Problem With Optimistic Hype and Spin
Lately I’ve noticed a tendency in some parts of the pro-Democratic commentariat to scorn anyone who isn’t confidently predicting a Kamala Harris landslide in November. So I wrote a warning about this self-delusion at New York:
It’s the nature of the political game that some fans perpetually insist their team is on the brink of a landslide victory. Maybe it’s just a device for maintaining their own enthusiasm, or maybe they truly believe that hype and spin add votes to their candidate’s columns. I usually associate this tendency with conservatives, but in this tense 2024 presidential contest, the landslide-predictors aren’t all in the MAGA camp. There are assorted voices pouring buckets of cold water on the data suggesting the contest between Donald Trump and Kamala Harris is a dead heat, insisting Harris is on the brink of a big victory.
Personally, I try to keep it all in the road by looking at the broadest array of public polling data available, by sticking with averages rather than enthusing or panicking over isolated poll findings, and by using history and common sense to discern trends and avoid overreaction to every survey that comes down the pike. But there’s just so much science you can muster on these matters given the relatively small sample size we have for presidential elections, which occur only every four years. So there’s plenty of grounds for legitimate nitpicking about polls and the experts who interpret them. It can get pretty wonky, as with Nate Silver’s criticism of the presidential-election model being used by his successors at FiveThirtyEight. But it probably affects only the numbers — and the expectations — at the margins.
Another entirely legitimate form of carping about polls and predictions emanates from those who dispute interpretations based on poor analysis of the timing, mix, and provenance of public polls. My old colleague Simon Rosenberg became somewhat famous doing that ahead of the 2022 midterms, when he denied he saw any “red wave” in the data (there ultimately wasn’t one despite an insistent media narrative) and pointed to the skewing of polling averages by strategically released state polls from conservative outlets and firms. He was largely right, though all in all, 2022 polling was very accurate, even if the predictions weren’t. This cycle, Rosenberg is again more optimistic about Democratic prospects in 2024 than your average pundit, but he remains firmly anchored in the world of data and objective reality and admits it’s a close race.
Then you have pro-Harris critics of the polls and pundits who rely on quite different talking points. The distinguished political scholar Michael A. Cohen uses data to support his supposition that Harris is actually in command of the race but ultimately relies on more fundamental arguments that are impervious to numbers:
“I have a few well-argued beliefs about this campaign (and they predate Biden’s departure from the race). Donald Trump is a high-floor, low-ceiling candidate who voters, by and large, don’t like — and they have made up their minds about him. Trump is incapable of and unwilling to change his political messaging to appeal to voters outside of his MAGA base. Democrats have a growing edge with women and college-educated voters, as evident in multiple elections since 2018, which gives them a significant advantage in crucial swing states like PA, WI, MI, GA, and AZ. Democrats have the easier path to 270 electoral votes.
“In key regards, this election is playing out in much the way I expected — even when Biden was on the ticket. With Biden out and his age and mental acuity no longer an issue, it’s that much harder for Republicans to win.”
Maybe Cohen is right. I’ll believe it when I see it empirically. But it turns out my attitude is exactly the kind of thing that veteran progressive media critic Dan Froomkin considers a sign of terrible bias:
“What if Kamala Harris — after a spectacular entry into the race, a stunningly unified convention, and a devastating debate — is basically running away with it, leaving Trump in the dust, while the national media — still mortified by its failure in 2016 to see the extent of Trump’s support — stubbornly sticks to the safer narrative that it’s a horserace going down to the wire?
“But wait, the polls aren’t showing Harris way ahead, you say. At most, they’re showing her with a narrow lead. Well, polls are garbage these days. And the pollsters, whose arbitrary weightings make a mockery of science, travel in packs. They, more than anyone, are terrified of underestimating Trump support again. So maybe this time they’re overestimating it? (Which they sure did in 2022.)
“You could, by contrast, make a solid vibes-and-momentum argument that Harris is winning handily.”
There’s a lot to unpack here. The argument that “polls are garbage these days” isn’t supported by the 2022 record, and to the extent that polls have been garbage lately, they have indeed underestimated Trump’s support. The “arbitrary weightings” of pollsters may or may not be precise, but they are not “arbitrary”; they reflect what we know about the shape of the electorate. And just tossing out all data in favor of a “vibes-and-momentum” take that is 100 percent subjective hardly empties analysis of bias.
Froomkin tips his hand by quoting SFGate columnist Drew Magary, who angrily argues that Harris is “destroying Trump, because Trump is a deranged old s—tbag” and that anyone who doesn’t see that is an Establishment hack.
Maybe all these critics are right and, for various reasons, pollsters and pundits are afraid to admit Harris is romping to victory. I don’t think there’s much question that a lot of the people being accused of overestimating Trump would be ecstatic if Harris wins. But being careful about it, given what we actually know about this election and the electorate, is not just prudent but mandatory.
There are significant methodology differences between the Washington Post and New York Times polls. The Times poll weights for Hispanic status as well as race; the Post poll weights only for race.
I would expect low-income Hispanics to vote strongly Democratic and be particularly underrepresented among poll respondents. Failure to adjust the sample for income, Hispanic status, or any proxy for these variables will impart a Republican bias to the results. (Education is not a proxy for income this year — the Post shows no trend of voting with education, and the latest Pew poll even shows Kerry support rising with education level.)
The Times poll also weights for whether the respondent lives in a Democratic, Republican, or swing county; the Post poll does not. This variable may function to some degree as a proxy for income — although as I noted out in my posting on the Times poll, the weighting is based on 2000 turnout and thus partially cancels out increases in Democratic registration. This weighting functions in some ways as a weaker version of party-ID weighting.
It seems to me that these methodological differences could easily explain much or all of the Times poll’s tendency to show Kerry doing better. It’s really unfortunate that none of the polls seem to publish any of their demographic weighting coefficients (except that we sometimes get numbers from which the party ID coefficients can be calculated) — it would be very helpful to know how significant and how stable the weighting coefficients are.
Ok,
Buhrabbit, I’m not referring to ANY conspiracy theory. They don’t hide anything in any manner whatsoever. Fox does their propaganda in the open. CNN has tried to be fair lately, but they are under enormous pressure from Fox to be just as un-journalistic. CBS is numbed (due to their own failure, largely). NBC has the same problems as CNN. And everyone who claims Washingtonpost and NYTimes are liberally biased is just a freeper. There’s just the LATimes and the magazines.
The rest what’s going on is reported on Media Matters. It’s bad enough.
RKC,
1. Gore’s reputation of being a nerd is part of the spin put on him. I have heard people talking of genuine affection to him. He’s known to be a interesting and interested fellow when the lights are off. He had problems to show this. I agree that he failed miserably in that first debate. But hey, that guy on the other side was even harder to bear..
2. The media hasn’t been any fairer to Kerry! That’s just bogus, with all due respect. What’s happened is that the left has learned a few lessons. They see what’s coming, they anticipate. There is a blogosphere, there is moveon.org. There is an increasingly unpopular war and a higher misery index. But that’s about it. The media went for more than a month with a campaign built on PLAIN LIES, for Christ’s sake!
3. Scepticism of polls? The bad September polls had a devasting effect on the Democratic electorate, much worse than four months of bad polls on the Republicans.
4. The registration efforts are good signs, and I’m glad for them. But the Reps have learned one or two things from the late surge to Gore in 2000. We’ll see if they can match us.
Listen, I don’t want to be gloomy. On the contrary, I’m all giddy since the debate. But this is an uphill struggle, and the opponent is not just the administration. I’m absolutely devasted of the things going on in the media, these days. We’ll need 20 years and more to reestablish some sort of decent journalism.
That’s all. But I hope that my pessimism is nuts.
Solving for party affiliation (minimizing RMS error against WP data), I get the following breakdown in party affiliation in the WP poll: (36%D/38.2%R/25.8%I).
If renormalized to the 2000 results of (39%D/35%R/26%I), the WP/ABC poll yields the following prediction:
Bush: 49.1
Kerry: 48.2%
Nader: 0.5%
The mean RMS error for my fit was 0.46%, within the 0.5% rounding error but just. Why doesn’t their poll supply party affiliation sample percentage, anyhow?
Solving for party affiliation (minimizing RMS error against WP data), I get the following breakdown in party affiliation in the WP poll: (36%D/38.2%R/25.8%I).
If renormalized to the 2000 results of (39%D/35%R/26%I), the WP/ABC poll yields the following prediction:
Bush: 49.1
Kerry: 48.2%
Nader: 0.5%
The mean RMS error for my fit was 0.46%, within the 0.5% rounding error but just. Why doesn’t their poll supply party affiliation sample percentage, anyhow?
-Fe Wm.
A least-squares fit on the WP/ABC data for party affiliation (why don’t they just provide this?) is as follows:
D= 35.99%
R= 38.21%
I= 25.8%
If the data is rescaled to Gore’s D39% R35% I26% model, we have, for the presidential race:
Bush: 49.1%
Kerry: 48.21%
Other: 0%
Neither: 0.65%
Would Not Vote: 0.26%
DK/No Opinion: 0.65%
Not a perfect fit on the thing – mean RMS error of 0.46% (still within the 0.5% rounding error, but just) – maybe I miskeyed some data, or something. Still, close race.
-Fe Wm.
This poll has weird sampling.
In the internals, for 1169 likely voters you get:
Bush 12%D 92%R 47% Ind = 51% of the total
Kerry 86%D 7%R 47% Ind = 46% of the total
Nader 0%D 0%R 2% Ind = 1% of the total
Using this information, you can solve backward to find out how many Ds, Rs and Is were likely voters. Here’s what I came up with:
A party ID of 24% Democrat, 27% Republican and 50% Independant out of likely voters. That seems to be an absurdly high number of independants as well as oversampling Republicans. I guess the most important thing to notice from this poll is they are running even among independants. Also, Kerry improved his status with independants by 4% from the last poll.
In response to Frenchfries’ comments, consider the following:
1. In 2000, even those who supported Gore generally considered him an unlikeable nerd. In 2004, while the flip-flop label has stuck amongst Republican partisans, most democrats know that it’s spin and consider Kerry a man of integrity.
2. In 2000, the media was totally unfair in its treatment of Gore. In 2004, the media has been largely unfair to Kerry, but he’s been treated a little better than Gore in 2000.
3. In 2000, all the polls right before the election showed Bush ahead, which should have discouraged Gore supporters from turning out. We don’t know what the polls will say right before the election this time, but we have learned to be more skeptical of polls and therefore are less likely to be discouraged by them.
4. In 2000 Gore did not have a massive voter registration drive in his favor. I’d guess that with Gore in 2000 and Bush in 2004, there was little participation in the primaries because the nominee was never in doubt. In 2004, we’ve had two waves of voter registrations: First, for the democratic primary process, in which there was a record level of participation, and second for the general election, and signs are that the democrats are way ahead in registrations.
My point is this: Even with everything against him, Gore still won the popular vote in 2000. This time, Kerry has the same things going against him that Gore did, but he has several advantages in his favor that Gore did not have. That suggests to me that Kerry still has a good chance of winning, even if he is behind in the polls on election day. I think the opposite of Frenchfries’ point is true: Bush can only win if there is a surge to HIM.
When the Newsweek poll came out showing Kerry ahead, the Kerry campaign downplayed it, saying they’ve had questions about that poll in the past. I think that’s the right approach: Whether we’re up or we’re down, we should downplay the polls. Remember, if Kerry is ahead in the polls on election day, there is a risk that Kerry voters won’t show up, assuming that they don’t need to. IMO, the message from the Democrats should be: Never mind the polls, whether we’re up or we’re down, just GET OUT AND VOTE.
I hope its clear from my post re: Zogby that I am not referring to a conspiracy theory. My beef is that at some point the media’s definitions of “fair” has ceased to be about truth finding, but instead is about speed. Also, there is their need to convince the right that they are balanced so they often will entertain ideas no matter how factually incorrect. I have a question for example, with no only the polling issue, where a 2 pt Bush lead is a lead, and a 2 pt Kerry lead is a tie (as I said do a google search), but also with a media that can’t say that Kerry’s Global Test is not the same as what Bush is saying on the stump. They have a very clear point of reference to make this judgement- namely what Kerry said in the debate- yet no mention is made of this. How much sense does this make if you are going to go with your lead story being about the global test. This goes beyond theories of laziness, it goes to theories of fear of not appearing balance, and as a result, not actually being balanced. Here, the solution should have been not just repeating Bush’s lie, and then saying Kerry says my opponent is lying about what I said (which confuses the point) but to add this is Kerry’s statement in the context of the debate. This way neither the right or left could claim unfair treatment- yet the way they choose to do it- clearly is unfair treatment. I think its just poor reporting for ourside that’s hurting us a lot. I am not certain what can be done about this in this cycle but we need to address this.
I highly highly doubt Bush is up 5 after the first debate.