Thread by @dogemperor

@MaryBushard @SimonWDC I'm still trying to find out with the most recent ones, but with the polls people were panicking about last weekend, it's a mix of phone + "paid survey" online, and where age data was available it skewed STRONGLY 45+

https://x.com/dogemperor/status/1845845677718274335

@MaryBushard @SimonWDC Also, for the phone polls it was *entirely* registered-voters (only one out of these four sorted by "likely voters"), and at least two are based on completely opt-in "paid survey" portals (YouGov and Ipsos) which are susceptible to poll-bombing

@MaryBushard @SimonWDC So far, all I've found for more recent stuff (that does not constitute flat out junk polls meant to drive a phantom "red wave") is TIPP,

And even then, NOT to be entirely trusted as actual methodology is behind a paywall (never, ever a good sign)tippinsights.com/tipp-tracking-…

@MaryBushard @SimonWDC What we DO know is that there is (just like every other poll) a VERY small pool (1248 based on Likely Voters), and if it is like EVERY other polling to date, it probably has an *extremely* small pool of under-30 and a major skew to 45+ (which is functionally uncorrectable)

@MaryBushard @SimonWDC And honestly, at this point we should consider polls a distraction; even if the polling shows Harris up 10, we should GOTV like she's 10 behind

Remember the assignment, kids. Get out there and VOTE in the only poll that matters in the end

@MaryBushard @SimonWDC I will also warn folks that TIPP Insights leans *distinctly* right-wing (seriously, look on its front page)

Them showing a Harris lead is actually *good* for us, it's an indication Harris's *actual* lead is probably more in these statestippinsights.com

@MaryBushard @SimonWDC Re Marist Poll: Did find it, it is a university poll maristpoll.marist.edu/latest-polls/ and is one of the few good polls that show a good breakout for age and actually HOLY SHIT PUBLISHES ITS DATA maristpoll.marist.edu/wp-content/upl…

So let's dig into the data! :D

1/x

@MaryBushard @SimonWDC So like practically all nation-wide polling it's got a TINY sample size (2021 for nationwide) and--uniquely--they tried one approach to get past the "callblock & also thinks paid survey portals are sus" gauntlet in using SMS *and* paid-survey-portal methods

2/x

@MaryBushard @SimonWDC (One of the groups they used for collecting info is Cint; Cint is, similar to Ipsos, a polling firm that heavily contracts with "paid survey" portals and paid focus groups.)

This, too, seems to have covered both likely AND registered (similar to NYT/Sienna)

3/x

@MaryBushard @SimonWDC Notably, this is one of the FEW polls that seems to have even TRIED to weight for age or generation, but even *here* you STILL had almost two-thirds of respondents under 35 (and a closer balance than average of over-45s and under-45s).

4/x

@MaryBushard @SimonWDC And ALSO notably--Marist is also to date THE ONLY poll we've run into (that publishes its data) that shows an attempt to recognize GenZ and Millenials as a significant voting bloc, around 46% of respondents.

But even HERE, there's issues with the "likely voter" segment

5/x

@MaryBushard @SimonWDC When narrowed down to *likely* voters, you have the same WILD understampling of under-45s that show up in *EVERY* survey (around 35% under 45, and 24% under 35, and a massive oversample of 45+ at 65%)

6/x

@MaryBushard @SimonWDC And when narrowed down to registered voters they were able to actually contact on the survey: 44% responded who were under 45 (which is by far the closest I've seen to ACTUAL population, at 48%) and 32% under 35 (by far, best response in a survey, usually it's more like 15%)

7/x

@MaryBushard @SimonWDC So ALREADY we see one major difference in Marist versus literally everyone else--as long as they focus on *registered* voters (and comparing to population figures in the American Community Survey) they ARE actually getting expected numbers of registered voters.

8/x

@MaryBushard @SimonWDC The LIKELY voters do show the same skew as other polls, though, which indicate that people who are likely to vote who are younger may in fact not be as willing to divulge their voting prefs in a survey (which can be due to privacy concerns or the Dobbs Effect or both).

9/x

@MaryBushard @SimonWDC Marist also (in its "likely" and "registered" voter vs ACS census info of all people called) DOES tend to match significantly better with population demographics, though a heavier proportion of those surveyed do tend to be in the South which risks skewing a survey R

10/x

@MaryBushard @SimonWDC (I've actually discussed at length with the NBC phone poll how certain states can significantly skew figures, especially if a high percentage of those polled are in the "Christian Nationalism Belt": This is EXTREMELY difficult to correct for.)

11/xthreadreaderapp.com/thread/1845509…

@MaryBushard @SimonWDC So all in all, it looks like there was at least a decent attempt to correct at least SOME of the issues that tend to come up in polling, especially with younger pops

Also of note: YES they do ask the "Evangelical" question. This one shows FAR less skew than the NBC poll!

12/x

@MaryBushard @SimonWDC Persons who identify as "Evangelical Christian" in a political poll are significantly more likely to be Christian Nationalist, and even overtly New Apostolic Reformation, than the average population (as per PRRI and Pew Research Center studies).

13/x

@MaryBushard @SimonWDC And while estimates of NAR adherents swing WILDLY (anywhere from 3 to 33 million adherents in the US), *in general* the number of "hard" Christian Nationalists nationwide tends to be around 10-15%, as well as number of self-identified "evangelicals" GENERALLY in the US.

14/x

@MaryBushard @SimonWDC The one exception to this is (again) in a "Christian Nationalism Belt" identified in PRRI research. Image 1 is support for Christian Nationalism (both by "hard" Christian Nationalists and "sympathisers") by state, and image 2 is just for white Americans

15/x

@MaryBushard @SimonWDC This likely has severely skewed NBC polling, which has not only not attempted to correct for age of voters, but has hit anywhere from 25-30% of callers who "identify as MAGA" and a constant 30% or more who identify as Evangelicals.

Methodology is important!

16/x

@MaryBushard @SimonWDC Anyway, back to Marist

Some good insights--most people who approve of the job Joe Biden is doing are college graduates, and there's a definite 11% "unsure" in "independent women" (which can even be "women who didn't register to a political party").

17/x

@MaryBushard @SimonWDC To date, this is the only national survey that I've seen that shows breakouts by region, and shows a lot of the conservative swing *is* specifically in the Christian Nationalism Belt (Midwest and South), and also non-college-graduates (esp. white non-college-graduates)

18/x

@MaryBushard @SimonWDC There is a SIGNIFICANT group of folks under 45, and especially under 30 (close to 20%) who "aren't sure" about whether they approve or disapprove of Biden's presidency, so it may be helpful to note what areas Kamala Harris MIGHT differentiate herself

19/x

@MaryBushard @SimonWDC Again, most disapproval of Harris seems to be concentrated in the Christian Nationalism belt (ahead in both West and Northeast for "Favorable")--and a solid 10% who "aren't sure" in the Midwest, which are Approachable and a good opportunity to work towards

20/x

@MaryBushard @SimonWDC Unsurprisingly, more minorities approve of Harris, but with a 12% "not sure" quotient--and similarly when broken out by African-American and Latino (with a strong pull to Harris re African-American voters, and even a decent spread among Latinos pro-Harris).

21/x

@MaryBushard @SimonWDC And out of these--50% favorable overall for likely voters for Harris, vs 47% unfavorable; likewise, 48% favorable for Harris for registered voters vs 47% unfavorable

This survey also has another interesting insight or three as well

22/x

@MaryBushard @SimonWDC Surprising is that there's a 22% *FAVORABLE* rating for Harris among *self-identified evangelicals*, and only a *74%* disapproval rating.

Again, remember: most people who self-identify as "Evangelical" on a survey are Christian Nationalist and are often full-on NAR/7 Mtns.

23/x

@MaryBushard @SimonWDC THIS is in and of itself a sign of a phenomena I've tried to point out is *VERY* difficult to capture--this is functionally a "wake" of a hidden Dobbs Effect, particularly by evangelical women voting for Harris (who don't want the husband to know) AND by #exvangelical kids

24/x

@MaryBushard @SimonWDC This is also (to date) the only survey I've seen that breaks down by large city, small city, suburb, small town, and rural. Unsurprisingly, small towns/rurals and smaller cities do trend "disapprove", but...suburbs actually ARE trending "approval", 48-44 with 7% "unsure"

25/x

@MaryBushard @SimonWDC And when you focus it by small cities and *men* vs *women*--54% of women approve of Harris (vs 37% disapprove and 9% unsure), while men tend to disapprove (53%, vs 40% approve and 7% unsure)

Also, white male college grads 52% approval vs 46% disapproval

26/x

@MaryBushard @SimonWDC Slightly more people under 45 approve vs disapprove (44% vs 43%) with a WIDE "unsure" of 14%--and 18-29s are functionally tied at approval/disapproval with 15% unsure

Most disapproval seems to be from 45-59 year olds, and again, significant skews re educational level

27/x

@MaryBushard @SimonWDC Now, to compare with TFG...49% disapproval rating and only a 44% approval rating nationwide, with only significant favorable bases being with R

Even the "favorable" rating for TFG in the Christian Nationalist belt is low (South, 49% fav/46% unfav; MW 48% fav/45% unfav)

28/x

@MaryBushard @SimonWDC Of note: TFG also has a 51% unfavorable rating among Latinos (vs 41% fav), and...most of the support isn't among the poor, but people making 50,000% or more (and very high unfavorability rating of 64% for college grads in particular)

29/x

@MaryBushard @SimonWDC There are some indications, even here, of an effect I keep pointing out--the basic nature of polling inherently will skew polling rightward (due to "easy marks" also trending more conservative, and "easy marks" being more willing to answer a SMS message or email survey)

30/x

@MaryBushard @SimonWDC And there is a SLIGHT signal of this--*every* demographic measured has a higher rate of disapproval vs approval for TFG, save for GenZ/Millenials--and even then it's 45% vs 44% with a WIDE 11% "not sure" (and broken out by age, *all* demos have higher disapproval)

31/x

@MaryBushard @SimonWDC Also impressive, again--that same 74/22 split in favorability by white evangelicals (74% fav towards TFG, 22% unfavorable).

Note: The HIGHEST predictive factor for Trump voters is "Are they NAR". (Please see my previous notes in that discussion)

32/x

@MaryBushard @SimonWDC And again, notably--a *slight* favorability lead for TFG among small city/suburb men (49% to 46%) but a STRONG unfavorability among small city/suburban women (57% disapprove vs 37% approve).

This, kids, is the Dobbs Effect in action

33/x

@MaryBushard @SimonWDC While there seem to be more "independent men" fans of TFG (59% TFG vs 39% Harris) independent women favor Harris (51% Harris vs 46% TFG). Independents only had 1-2% who stated they'd vote for a third party

34/x

@MaryBushard @SimonWDC This is a rare poll that also notes voting preferences--and there's indications that MAGAts still think mail-in voting is sus as heck.

For in-person voting, 43% noted they'd vote for Harris (vs 56% for TFG), for early voting, 46% for Harris (vs 53% for TFG).

35/x

@MaryBushard @SimonWDC (As a minor aside: One thing I DO wish this particular poll had taken into methodology is the *availability* of voting options available. 3 states (Alabama, Mississippi, and New Hampshire) don't provide in-person early voting OR no-excuse vote-by-mail options.)

36/x

@MaryBushard @SimonWDC Certain other states require preregistration for excused in-person absentee voting, or have very limited windows for early voting (so the options may not be there). Yes, this is more common in the Christian Nationalist Belt



37/xncsl.org/elections-and-…

@MaryBushard @SimonWDC Anyways, back to survey, and the answer WHY expanding voting access terrifies R legislatures--fully 75% of people who reported they were likely to vote mail or absentee were explicitly voting Harris (this presumably includes in-person absentee voting).

38/x

@MaryBushard @SimonWDC ALSO significant: The ONLY place where a clear "I'll vote for TFG vs Harris" showed up was in teh South, and not by much (47% Harris vs 52% TFG). The Midwest is a dead heat (49%/49% with 2% other parties). Everywhere else is strongly Harris by 57%/42% or 57%/43%

39/x

@MaryBushard @SimonWDC EVERY age demographic polled is trending for Harris (including 54% of people under 45, 53% of GenZ/Millenials, and 55% of Baby Boomers)

Even GenX is trending 51% for Harris

40/x

@MaryBushard @SimonWDC Another strong ping hinting at a Hidden Dobbs Effect (women who are "hidden Harris voters" in evangelical households)--18% stating a vote for Harris, 81% for TFG (compare with 22% approval for Harris and 22% disapproval for TFG)

41/x

@MaryBushard @SimonWDC In regards to cities vs the country, only small towns and rural populations are showing majorities for TFG (55% and 65% respectively)--and small city/suburb men also slightly towards TFG (52% vs 48% for Harris)--and a *60% for Harris among small city/suburban women*

42/x

@MaryBushard @SimonWDC You know why I keep harping about the Dobbs Effect?

THIS is why I keep harping about the Dobbs Effect.

Media and GOTV folks need to keep up showing JD Vance's horrendously misogynist commentary, TFG's creepiness on main, and so on...and remind women their vote is private!

43/x

@MaryBushard @SimonWDC This poll also had a specific breakout for registered voters only on "how do you think you'll vote" based on candidate and party preference

Again, indications are that MAGAts still think mail/absentee voting is VERY sus (over 80% voting in person, either early or on Nov 5)

44/x

@MaryBushard @SimonWDC Also an interesting tidbit--something like 3% of likely Trump voters, and 2% of Republicans and 5% of Independents (including Libertarians, Green Party, etc.) indicate they don't intend to vote at *all* on Election Day despite being registered (vs 2% leaning Harris, 1% Dem)

45/x

@MaryBushard @SimonWDC And something that shows impact of voter suppression efforts in the South in particular (lack of vote-by-mail and often early voting or no-excuse absentee voting, early polling location closing hours, etc.) shows 4% of Southern registered voters being unlikely to vote

46/x

@MaryBushard @SimonWDC This also shows up when looking at percentages with people who are not college graduates (4%), and people earning under $50k who often aren't able to get time off to vote (6%)

6% are also under 35, so definitely want to encourage younger folks to vote

47/x

@MaryBushard @SimonWDC But all in all--even though there are *uncorrectable* problems (mostly related to "the person who'll do a survey based on an unsolicited email or an unsolicited text message is LIKELY to trend more conservative than average"), Marist...is a *decent* poll, all in all

48/x

@MaryBushard @SimonWDC Now, one thing I also would like is some more info on question phrasing, but at least THIS poll provides the actual info on what you can use to evaluate if it is, in fact, a quality poll or not and what they may be doing to attempt to correct for issues in polling

49/end

@MaryBushard @SimonWDC @threadreaderapp unroll