Thread by @ThyConsigliori

1) This one's for you, @TamaraLeigh_llc. I don't know if I can refute this poll, but I will be able to give it some very real question.

breitbart.com/politics/2020/…

2) Methodology:

"This poll was conducted ... among a national sample of 1991 Registered Voters. The interviews were conducted online and the data were weighted to approximate a target sample of Registered Voters based on age, gender, educational attainment, race, and region."

3) I omitted the data, July 17 - 20, simple due to Tweet sized constraints.

As far as size goes, it is a good deal larger than the 1,500 line, so that's good. And Registered Voters are better than adults, while not as good as Likely Voters.

4) The first thing I don't like is this phrase: "target sample of Registered Voters based on age, gender, educational attainment, race, and region." Note the word "sample." How do we build such a thing? Based on what data, and sample, or target model?

5) They actually give us a tremendous amount of information about this, and all the answers to all their questions, over 300 pages worth, here:

assets.morningconsult.com/wp-uploads/202…

6) We'll discuss what all that massive volume of data means, in a minute, but they offer more data under this heading: Respondent Demographics Summary, from pages 308 - 311. I've gone over it twice, but have to check a third time. I'll be right back...

7) I missed it the first two times, but here it is, I found it!

Dem (no lean): 768 - 39%
Ind (no lean): 548 - 28%
Rep (no lean): 675 - 34%

We could dive far more deeply into the weeds of their mountains of data, but the key you always look for is just that, how many of each?

8) I'm not sure I know what "no lean" means. I was tempted to erase that, but just left it in. To judge any poll, we have to build our own model of which voters will show. That's why Registered Voters is weaker than Likely Voters. The idea is to predict the outcome, accurately.

9) Now, I do NOT find this model to be egregious. I've certainly seen far worse, and with far fewer polled, and at that, Adults as opposed to Registered Voters. So, by the number polled and the credibility of the split, not near as bad as I've seen. However...

10) I'm not buying this poll, anyway. The first tell is in the headline itself: "A majority of voters support a mask mandate in their respective states punishable by fine or jail time, a Morning Consult/Politico poll released this week showed."

11) Can you count the number of questions? I can't. It's over 300 pages of polling data in the raw. They asked such things as Favorability of Nancy Pelosi; Favorability of Congress. On and infinitely on. Imagine yourself giving answers to, let's guess, maybe 50 - 100 questions.

12) At what point do your eyes glaze over and you just click, click, click in order to get done? Nah, I don't buy the method, at all. Asking countless questions and tallying the countless answers, what's it all mean? Who's got the focus? I sure don't. Would you?

13) But let's get back to 39% Democrats vs 34% Republicans, and 28% Independents with no apparent lean. I smell a rat here. Why should we believe that there are 5% more Democrats than Republicans? And these aren't even Likely Voters, just Registered?

14) If I were to guess, I'd say Independents are growing, where Democrats and Republicans are shrinking. Why doesn't their model have a larger number of Independents? No answer is given. And, who knows which way Independents are breaking? I sure don't.

15) Friends, Tamara, all of this is simply too complex. Maybe there's science here. Maybe. If so, I sure can't ferret it out. Rather, what I see are answers looking for people to give them. Who knows? What they assess may be true. Or, it could be 100% false, hidden in complexity.

16) If you wished to show how people feel about masks, your poll would focus on that alone. You'd ask a short sequence of logical questions designed to create clarity, not confusion. And you wouldn't be asking about Nancy Pelosi's or Congress' favorability. That's a distraction.

17) So, for those tempted to take this seriously, and believe that its results are very, very bad; and for those who trust that Breitbart would vett the poll methodology, I say...balderdash. The real bottomline of this poll is nothing more than we don't really know.

18) We've got a way to go, before our site hits minimum adequate functionality, but I still proudly offer it as an honest mind's alternate to all this fake science presented as if it were the search for truth.

BetweenTheLines.vote

19) And @TamaraLeigh_llc, be confident in telling any of your friends that the quasi- or worse, the fake science of polling is just about the easiest thing in the world to manipulate. Exciting headlines predicting doom are just about always a tell.

20) What we need in honest polling is humility, not hubris. We need curiosity, not agenda. We need to seek for the truth, not for justification for the headline we chose prior to asking any questions. Fake Science. Polling is its favored child this season. Be warned.

Thread ends at #20.