Thread by @dogemperor

@RobertVenditti4 @schlagteslinks Fox, HarrisX are known to be right-wing; NBC is known to have issues (oversampling of older *and* particularly "white evangelical" pops); Quinnipiac is known right-aligned; Morning Consult is known to have issues in methodology; Suffolk & Emerson have similar issues w/ pools

1/x

@RobertVenditti4 @schlagteslinks Re the Suffolk poll (yes I am about to do one of my Deconstructions), it's another poll with a *massive* skew towards South (33.6%), which *INHERENTLY fucks a poll because that's right in the Christian Nationalist Belt

2/xsuffolk.edu/academics/rese…

@RobertVenditti4 @schlagteslinks Two massive risks of phone polling involve overpolling of the South, and overpolling of self-identified "white evangelical Protestants"; most of the South is within an identifiable belt of high numbers of Christian Nationalists up to 2x population average

3/x

@RobertVenditti4 @schlagteslinks Something like 66% of people who self-identify as "white evangelical Protestant" are found to have Christian Nationalist sympathetic viewpoints, with around 30% as "hard" Christian Nationalists; 40.1% of white evangelical Protestant at least partially believe in NAR theology

4/x

@RobertVenditti4 @schlagteslinks Even with this, *on average* only 30% of Americans share at least some beliefs associated with Christian Nationalism, and only 10% or so are "hard" Chrstian Nationalists (which would include New Apostolic Reformation Christian Nationalism)

5/x

@RobertVenditti4 @schlagteslinks And any poll that shows more than around a 15% total of *self identified* white evangelicals--especially with them trending extremely Trumpy--is a sign you oversampled a highly Christian Nationalist area

I discuss this re the NBC polls:

6/xthreadreaderapp.com/thread/1845509…

@RobertVenditti4 @schlagteslinks I have also extensively discussed the risks of oversampling the South specifically with the Emerson poll (which is itself fairly skewed towards the right in its questioning bias)

7/xthreadreaderapp.com/thread/1848419…

@RobertVenditti4 @schlagteslinks And we LITERALLY see the same pattern of oversampling of the South with the Emerson poll, and it also entirely concentrates on voice calls which ALSO introduces an additional source of self-selection that is frankly something polling *has not* caught up with

8/x

@RobertVenditti4 @schlagteslinks To whit: People who answer a *voice* poll are generally older, do *NOT* have callblocking tools, do NOT whitelist, are MUCH more likely to pick up a telemarketing or phishing call, and are part of a psychological population of "easy marks"



9/xthreadreaderapp.com/thread/1842984…

@RobertVenditti4 @schlagteslinks "Easy marks" are worse at telling truth from "truthiness", *generally* trend conservative, AND notably are not only more likely to fall for conspiracy theories AND affinity frauds, but are more likely to propagate disinfo.

This is *very* difficult to correct for

10/x

@RobertVenditti4 @schlagteslinks Another factor that I see over and over in R-trending polls--an EXTREME oversample of people 50-70, with 53.9% of the pool being 50+ and 27.3% being 50-64 (aka Fox News target audience)

In comparison the SMALLEST pool was 18-24, 7.4%; under-35 only 22.4%

11/x

@RobertVenditti4 @schlagteslinks Even with this fairly extreme skew, Harris STILL is a point ahead, and (in a sign we are dealing with a heavily R-skewed pool) Cornel West got a full percent

This persisted even with a "two options only" with 9.71% "undecided" and a fairly high refusal rate

12/x

@RobertVenditti4 @schlagteslinks Another sign that this pool had almost NO members under 25 (one of the fastest growing demographics of voters) is that only 8% did not vote, and 1.7% voted third party, in 2020. (You'd expect more "did not vote" with more younger people in the pool.)

13/x

@RobertVenditti4 @schlagteslinks In a sign that is *incredibly* sus, a lot of questions specifically regarding willingness to accept electoral results, economy, most important issues, etc. are completely embargoed, which is a sign they probably don't want to release details that can show level of skew

14/x

@RobertVenditti4 @schlagteslinks And yes, those kinds of crosstabs can show an unusually conservative population--for example, pools of young folks listing "migrants" as their top issue is a MAJOR warning you're dealing with skewed polls or possible poll-poisoning (esp. with online polls).

15/x

@RobertVenditti4 @schlagteslinks And even here we're seeing signs with the non-embargoed stuff that this is a fairly R-skewed pool (around half a point more think TFG will win vs Harris, with around 12.4% undecided).

The high percentage of people supporting mass deportation (45.2%) also hints at pool skew

16/x

@RobertVenditti4 @schlagteslinks There are also an unusually high number of people who would not vote for either a Dem or a Republican for Congress (2.1%) and around 4.8% undecided, which makes me suspect they hit a pool of Libertarians

And there's more evidence of a poll skew lower down

17/x

@RobertVenditti4 @schlagteslinks 35.8% self-describe as "Moderate", but only 24.7% as "liberal" of any sort (and 16.3% as "liberal but not very liberal"). 35.2% describe themselves as "conservative", with 24.6% as "conservative but not very conservative"

18/x

@RobertVenditti4 @schlagteslinks Crosstabs also STRONGLY indicate this hit a STRONGLY conservative polling group, with 18-34s going 48% for Harris and 42% for Trump AND a pool of Hispanics going 49% for Trump vs 38% for Harris

Which this does not match up with really ANY polling for these groups

19/x

@RobertVenditti4 @schlagteslinks Among other things, polling for 18-34s are usually showing something closer to a 60+ or more swing for Harris, and Hispanics also a general 55+ swing for Harris

This poll also shows unusually low nums for Harris among African-Americans (usually around 82%, this is 72%)

20/x

@RobertVenditti4 @schlagteslinks When narrowed down to two candidates (in probing of undecideds) it's *closer* to what is typically seen in polling in *general* of these pops but still rather low/conservative

And looking at breakouts by REGION shows the reason for the skewage.

21/x

@RobertVenditti4 @schlagteslinks Based on the area breakouts, the polling in the South and Midwest apparently almost perfectly tracked along the "Christian Nationalism belt", with skews showing 57% for The Former Guy in THOSE regions only

Based on Hispanic skew, I also suspect FL Cubano communities hit

22/x

@RobertVenditti4 @schlagteslinks And that's how even being less than careful about making sure you have a good age distribution and regional distribution can skew a poll significantly rightward! (Important, as outside of GA and NC not a lot of the swing states are heavily Christian Nationalist.)

23/x

@RobertVenditti4 @schlagteslinks Atlas Intel, I am *incredibly* sus about for Reasons--those Reasons, of note, being "Atlas Intel is based in a BRICS country, which is to say, it's in a country at least moderately within Putin's orbit". Its associations with OTHERS in Putin's orbit are not reassuring.

24/x

@RobertVenditti4 @schlagteslinks The fact that their official Xitter account's pinned post shouts out Nate Silver (now working for cr/y/pto betting site Polymarket owned by brogliarch Peter Thiel, and which explicitly bans US IPs) is not reassuring, nor is its promotion by a pro-Orban site

25/x

@RobertVenditti4 @schlagteslinks So Atlas Intel is of course going to drive a narrative, because Putin *very much* wants a Trump presidency so he can have his Manchurian Candidate in office (and to recruit Vance to the Genocidal Oligarch-Dictator's Club). Effectively borderline "active measures". Avoid.

26/x

@RobertVenditti4 @schlagteslinks Out of polls conducted on a national level, I have seen *ONE* that is not hot garbage on some level, that one is Marist (good aggregation of pop, good age spread, actually provides some INTERESTING info, uses MMS/email polling at random to avoid callblockers/poll poisoning)

27/x

@RobertVenditti4 @schlagteslinks But as I've said before: All polling has functional limitations, including the fact that aggressive advertising (over 40 years) has so thoroughly shat the bed for people being contactable in a survey that we are *well* in a post-polling world save for some specific insights

28/x

@RobertVenditti4 @schlagteslinks And so my advice is the same as it has been: Polling is functionally a *distraction*, at most it gives insight into reachable populations for GOTV, and even if polling shows ahead we should be ALL GAS NO BRAKES as if 20 points behind.

29/end

@RobertVenditti4 @schlagteslinks @threadreaderapp unroll