No tightening here , folks- The Economist's pollster G. Elliott Morris

[octoberlib]

Link to tweet

G. Elliott Morris
@gelliottmorris
·
1h
Trump gets a statistically insignificant 2-point RNC "bounce" according to a new Emerson College poll (which we don't include because of pretty egregious data quality issues)


I don't care about the 538 rating. Emerson uses two of the most unreliable sampling procedures in the industry, then proclaims that they have fixed all the errors with post-processing, even though we know MTurk errors run deeper than that

There have been multiple news cycles this year driven by an influx of low-quality polling data, each time shoring up Trump's numbers and the betting markets—and each time they have reverted to the mean (Biden +9 or so) once we got good online and live-phone data

Y'all are free to include this type of data in your average if you want, but it's bad and we shouldn't treat it the same as eg an NYT or Pew survey https://journals.sagepub.com/doi/pdf/10.1177/2053168018822174

I will adjust my expectations for Trump's post-RNC bounce (currently around two points or so, and within the margin of error across surveys) once we get actual high-quality data

Link to tweet

It actually makes a whole lot of sense that Emerson has Trump's net approval above water (& better than his Nov numbers). One thing Pew found when they looked into low-qual data & survey marketplaces last year was a tendency for people to just spam "approve" on a lot of questions

Link to tweet

[vercetti2021]

1. Thank you

I'm sick of the alarmists that begin to freak out over a unreliable pollster poll. No fucking chance in hell Trump would be up 19% with AA. No way.

[octoberlib]

2. Agree.

[Thekaspervote]

3. Oh me too! That crap poll has bee posted here twice. As with this OP shows there's a wealth of

Information to the contrary

[vercetti2021]

4. 4 times actually

I've been keeping count.

[Thekaspervote]

6. Oh geez!! I swear... what about all the good polls, the reliable polls why aren't posted 4 times

[vercetti2021]

8. Well I can tell you why

Keep peeps on their feet. Or toes rather. Like I understand don't get complacent but I don't think we are because people who vote are going to vote this asshole out of here there's no doubt about that

[Thekaspervote]

5. And this is 1 more reason 538 should no longer be considered a gold standard in polling.

538 Giving Emerson an A- when it’s polling methodology is really questionable??


https://www.commondreams.org/views/2019/10/30/nate-silver-making-he-goes

[octoberlib]

7. Yeah, there were a lot of comments on Twitter on Silver changing their rating from a C

to an A-. During the primaries their polling was crap. 538 is still a pretty good aggregator despite that.

[Roland99]

9. Would like to see this model's comparison to 2016 and 2018

[octoberlib]

10. I can't access it because of the paywall. Good question.

[Drunken Irishman]

13. I am not sure they did a forecast four years ago.

[Awsi Dooger]

11. He makes a lot of good points

Regarding the spamming of the same response every time I took an interesting automated phone survey yesterday that obviously tried to address that. They varied the ordering of the approve and disapprove categories, along with varying whether the Democratic nominee was first or second.

So the industry is trying to adapt. This was a landline call

[BlueWavePsych]

12. Smarter heuristics.