General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region Forums*Emerson is a joke*
I welcome all Stats and Poli Sci majors to tell me if these findings are remotely tethered to reality:
Link to tweet
The suggestion that Trump is winning by 1 in urban areas and losing by 20 in suburban areas is risible. That's just not how people vote, ever. The Democrats do much better in urban areas and less well in suburban areas, not vice versa. That's a finding that's been consistent for decades.
More on Emerson:
Link to tweet
Beware of Emerson with it's IVR/MTurk frame despite it's A- rating. It's garbage methodology.
Focus on the average. If a poll looks too good or bad to be true it probably is.
beachbumbob
(9,263 posts)Demsrule86
(68,556 posts)I believe it is seriously flawed and ignored.
Johnny2X2X
(19,060 posts)Emerson is a well regarded poll, A-, but it doesn't mean they can't get it wrong from time to time. This one seems really odd. But just take it as a data point, same as a single poll showing Biden up 10.
bearsfootball516
(6,377 posts)They changed to MTurk, a notoriously bad polling methodology. But 538 only updates it's ratings every couple years, so it's stuck with the A- rating despite being a significantly worse polling firm since changing it's methodology.
Thekaspervote
(32,762 posts)Flawed polling methodology
The Economist pollster will not even put them in their averaging mix
Silver is still trying to recover from the ridiculous polling he did for the dem primaries. After 2 caucus nights silver declares that sanders will win 49 states despite the fact that the statistics showed sanders vote totals were down 25% from 16 and there was no youth voter turn out. Declared Bidens campaign to be dead.
Amishman
(5,557 posts)This looks more like a total fluke of a sample. It happens even to good pollsters.
Random sampling is random, and will sometimes spit out a complete head scratcher. An honest pollster can't just toss it because they doubt it or dislike it. They report it and move on.
Codeine
(25,586 posts)you can get an occasional wonky result. Its one poll, with the rest of the polling industry coming up with differing results. Not a big deal, and not worth the gloom and doom.
uponit7771
(90,335 posts)... seriously even though their 16 swing state polling a week and 3 days out were horrible.
BGBD
(3,282 posts)Shouldn't be used.
It's a lot cheaper to just pay a bunch of random people a few cents to take a poll than try to call the number of people it takes to get a good sample.....however, they are also unreliable.
Many of them just click through these polls so that they can complete as many as possible in the lease amount of time. They are doing it for money, and they want to maximize their earnings. It's a quantity game, not a quality game on MTURK.
DemocratSinceBirth
(99,710 posts)BGBD
(3,282 posts)what you pay for.