(C) Daily Kos
This story was originally published by Daily Kos and is unaltered.
. . . . . . . . . .



Holy Moley, Can Everyone Just Take a Deep Breath? [1]

['This Content Is Not Subject To Review Daily Kos Staff Prior To Publication.', 'Backgroundurl Avatar_Large', 'Nickname', 'Joined', 'Created_At', 'Story Count', 'N_Stories', 'Comment Count', 'N_Comments', 'Popular Tags']

Date: 2022-10-22

The recent change in our fortunes utilizing a very simple poll composite model. No secret sauce.

The sky, as you know, is falling. The most recent twitter conversations have been about what the lame duck Congress can do to safeguard the economy and democracy because, as we all know, we are absolutely 100% going to lose. At least, that’s the conventional wisdom.

Above, I posted a simple election modeler for the Senate. And yes, the Senate isn’t the only game in town, but it definitely the cause of the recent freak out. My model is much simpler (but honestly, more intellectually honest) than the 538 modeler which had us going from a 70% of keeping the senate a month ago, down to 55% today. My model simply takes the aggregate polls of the last 3 weeks. I take the mean if there are at least 5 polls, and the median otherwise.

Which polls? All of them. Any pollster in the 538 database (which is already more honest than the cherry-picked RCP aggregator). I did get rid of Echelon and Center Street, but only because they were predicting Dem+20 races and I simply didn’t want to justify their inclusion. I then assume normal distributions of errors (about 4% per state is typical) and look at what happens if the polls were unbiased, or 2% too Dem (red) or 2% too Republican (green).

First, the good news. If the polls are unbiased, then we have a very good shot of holding the senate. Everyone seems to have internalized the idea that this is an impossibility — that of course — Republicans will always outperform their polls, that no-one is even considering this possibility. Never mind that we’ve clearly outperformed polls all year: in the CA recall, in NY Specials, in the KS abortion vote, and so on. Outperformed them by a lot more than 2%.

However, it looks like we’ve fallen off a cliff in the last two weeks. And, to be blunt, things have changed. But by how much?

Rather than look at poll averages, let’s look at how much a given pollster has changed in October compared to previous polling (generally August or September). Here’s the generic congressional ballot:

Pollster Recent Date Recent Margin Initial Date Initial Margin Change Rasmussen (Pulse Opinion Research) 2022-10-20 -4 2022-09-15 -1 -3 Emerson College Polling Society 2022-10-19 -4.9 2022-08-24 -1 -3.9 YouGov 2022-10-19 3 2022-09-13 3.9 -0.9 Ipsos 2022-10-18 1 2022-09-13 4 -3 Data for Progress 2022-10-18 -3 2022-09-15 -2 -1 Monmouth U. 2022-10-17 -6 2022-08-01 3 -9 McLaughlin 2022-10-17 -6 2022-08-24 0 -6 Morning Consult 2022-10-16 3 2022-09-15 3 0 Hart Research Associates/Public Opinion Strategies 2022-10-16 -2 2022-09-13 0 -2 Pew 2022-10-16 1 2022-08-14 2 -1 Harris Poll 2022-10-13 -6 2022-09-08 2 -8 Trafalgar Group 2022-10-12 -5.3 2022-09-09 -5.7 0.4 Beacon Research/Shaw & Company 2022-10-12 3 2022-09-12 3 0 Siena College/NYT Upshot 2022-10-12 -3 2022-09-14 1 -4 Susquehanna 2022-10-11 0 2022-09-27 -1 1 Big Village 2022-10-07 5.4 2022-09-09 5.3 0.1 InsiderAdvantage 2022-10-03 -2.1 2022-09-01 1.2 -3.3

There are some whoppers. Monmouth and Siena got big news, but most of these pollsters have much smaller changes. Assuming that shifts within a poll are what matter, the average shift is about 2.6 points toward the Republicans. To be clear, this is significant. It means the difference between a nominal Dem advantage and a nominal Republican advantage, and with it, probably significantly decreases the change of keeping the House.

But it is a far cry from the claims that somehow PA and GA and AZ have gone from Dem+10 to even. The numbers simply don’t add up. And, you might suppose, as these are for the nation as a whole, maybe there are specific moves within states. So let’s look at them (and yes, I wrote a variant of this a few days ago, but these are updated numbers, hopefully a few more insights, and in slightly more accessible format).

State Pollster Recent Date Recent Margin Initial Date Initial Margin Change Arizona Trafalgar Group 2022-10-17 1 2022-08-27 3.3 -2.3 Arizona InsiderAdvantage 2022-10-11 4.4 2022-09-07 6 -1.6 Arizona OnMessage Inc. 2022-10-10 3 2022-09-11 10 -7 Arizona OH Predictive Insights / MBQF 2022-10-06 13 2022-09-09 12 1

Arizona, where Kelly is leading by about 5 points per both a simple average, and according to 538 has a similar margin to Florida — and nobody is treating that as a tossup. This 5 point Dem lead is after shifting by about 2 points. Nothing is guaranteed in this world, but this is most definitely not a tossup.

Office State Pollster Recent Date Recent Margin Initial Date Initial Margin Change senate Georgia InsiderAdvantage 2022-10-16 2.9 2022-09-07 -3 5.9 senate Georgia Trafalgar Group 2022-10-11 1.5 2022-08-27 -0.8 2.3 senate Georgia Quinnipiac 2022-10-10 7 2022-09-12 6 1 senate Georgia Emerson College Polling Society 2022-10-07 2.2 2022-08-29 -1.7 3.9 senate Georgia U. Georgia SPIA 2022-10-04 3 2022-09-16 -1.6 4.6

As many people have noted, Georgia has moved about 3.5 points toward Warnock since his awfulness has been further revealed. This is not a particularly close race. Both Nevada and Wisconsin have been barely polled. Marquette showed a big move in their poll — about 5 points, but consistent with simple sampling error. Nevada has two polls moving toward the R by similar margins, but it’s worth knowing that the better of them, Suffolk, still has Cortez Masto ahead. Ohio has been almost completely unchanged. And then we get to my home state, Pennsylvania. In all of October, there have only been 5 polls at all, all month. 3 are from hard right partisan pollsters: InsiderAdvantage, Fabrizio Lee, and Trafalgar (only 1 of which 538 identifies as partisan). 1 from Echelon. And 1 from Wick — a pollster I’d never heard of — and the only pollster to show an Oz lead over the entire election cycle (they also show Shapiro up by only 2. Likewise, they show Kelly only up by 3, Warnock tied in Georgia. But on the bright side, their random number produced a result where Hobbs is up by 1 for AZ Gov). And that just showed up on 538 even though they polled it a week and a half ago. And only 2 pollsters polled it both before and after the shift: State Pollster Recent Date Recent Margin Initial Date Initial Margin Change Pennsylvania InsiderAdvantage 2022-10-19 0.8 2022-09-24 3.1 -2.3 Pennsylvania Trafalgar Group 2022-10-11 2.4 2022-09-15 1.8 0.6

According to this, the average has barely moved.

Folks, the media is being flooded with polling misinformation, or perhaps disinformation. It will be interesting to see what happens when the usual suspects (Quinnipiac, Marist, Siena, etc.) poll PA again. My bet is that we start seeing a lot more Fetterman +5 to +8 results.

[END]
---
[1] Url: https://www.dailykos.com/stories/2022/10/22/2130605/-Holy-Moley-Can-Everyone-Just-Take-a-Deep-Breath

Published and (C) by Daily Kos
Content appears here under this condition or license: Site content may be used for any purpose without permission unless otherwise specified.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/dailykos/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.