General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region Forums538's Walk of Shame continues.
One of these polls is not like the others
One of these pollsters does not belong
https://projects.fivethirtyeight.com/biden-approval-rating/
538's eccentric decision to include, and continue to include a rogue pollster, Rasmussen Reports in its aggregates, is beyond a joke, it is a locus of shame.
Rasmussen has a infamous reputation for manipulating numbers, presumably within the Margin of Error, for partisan purposes. As 538 themselves have said, after being extreme outliers in favor of the GOP during general elections, their polls 'magically' fall into line with reputable pollsters on election eve.
The most charitable thing one can say about Rasmussen is that they have a heavy methodological and sampling bias in favor of the GOP. A less charitable conclusion would be, they gleefully cook the books for prominence on RW media, to butress funding.
538's pretensions to adjusting Rasmussen's numbers are a fig leaf in a storm. No amount of stat-'splaining can lend a decent sheen to preposterous results. Rasmussen have reported Biden +0 (49/49) twice now. In Bill Clinton's words - one way or another, 'that takes a lot of brass'.
But the problem is not whether particular results are subject to ridicule, it is the ongoing corruption. 538 presents its adjusted weighted average of polls as an indication of how the president's approval is tracking over time. But arbitrarily playing around with the numbers, within the MoE, fatally compromises a times series - And that is what Rasmussen Reports does. They do it during General Elections, and they are doing it now.
Today, Rasmussen's numbers suggest that the American Rescue Plan was unpopular with voters, damaging the president's approval.
Biden - Job approval
March 10-14 +6 (52/46)
March 11-15 +4 (51/47)
March 15-17 +4 (51/47)
March 17-21 +0 (49/49)
Horseshit.
It's time for 538 to bite the bullet. Rasmussen Reports do not belong in any survey of respectable pollsters, let alone in an aggregate featured on the front page, every day.
brooklynite
(94,502 posts)538 provides data. I'm intelligent enough to assess its value.
speak easy
(9,238 posts)at any opportunity to defend Rasmussen Reports. Your'e welcome to them.
brooklynite
(94,502 posts)I'm better off hearing opinions that disagree with my own. So are you.
speak easy
(9,238 posts)You mean, 'both sides', right?
brooklynite
(94,502 posts)The point isn't always to validate the opposing viewpoint. Its to understand what people who disgree with you are hearing and thinking.
speak easy
(9,238 posts)who believe the election was stolen?.
If you want to see Rasmussen's numbers - go FNC - plenty of coverage there. Not to put too fine a point on it Rasmussen's results are tailored to RW propaganda. Trump had a higher approval rating that Biden? Rasmussen - check; FNC - check. What is left to understand?
Enough of this shit.
ProfessorGAC
(64,995 posts)When data needs to be adjusted as much as 538 does (3% up in approval, SEVEN down in disapproval) they are acknowledging flawed information.
In fact, it's flawed to the point that even after adjustment Ras has the lowest approval & highest disapproval.
I'm pretty damned sharp too, and one of my 3 advanced degrees is in statistics.
There's no valid mathematical reason to include subjective, clearly polluted data in the aggregation.
They want to report them and describe them as an outlier, fine.
Including them in the aggregate is bad statistics.
And, despite their explanation of their methodology, it strains credulity to grade them C+ when the bias is known and suspected willful.
USALiberal
(10,877 posts)ProfessorGAC
(64,995 posts)I'll presume you've never read Silver's methodology page.
Or, if you did you didn't see the subjective assumptions used in his waiting.
Either way, it's pretty clear you don't grasp this stuff the way you think you do.
USALiberal
(10,877 posts)Hestia
(3,818 posts)when we all know he most likely wasn't above 17-20% of the US ever at any given time. It seems they were included to assuage * so people weren't subjected to his outright evil rages.
I think it does make a difference on what people see, especially if they are like me and don't drill down into the "meaty parts" due to time, patience and personal dedication.
uponit7771
(90,335 posts)WarGamer
(12,436 posts)Stallion
(6,474 posts)DanieRains
(4,619 posts)Only thing that makes sense.
Tree Lady
(11,451 posts)I started doing surveys for fun, mostly shopping stuff, computer etc back then it was called polling point before they changed name to yougov
I have been doing them for about 15 years now, the last 5 years they ask more political questions.
They know your full name, address and email, don't sell it anywhere (so they say)
In the political ones they ask what party you are, when you say democrat they ask if strong one.
You can tell if someone pays for a survey and the questions try to trick you into saying something that would slant answers to look like you were more on Trumps side. I would make sure end of survey where you can tell them what you think what I thought about that.
You get points for each survey around 300-500 and after 30,000 points you can get a $25 GF I usually get Michael's and buy candles. Always takes me a whole year to get it so I stock up on tea lights.
I have thought about quitting because most surveys are boring but after they got political I wanted to make sure some progressive voice gets counted.
I always wonder how they can get a honest result though when they know ahead of time if someone says for years they are strong in their party.
Forgot to say all the surveys are online no phone calls. But its through your yougov account and they verify your address.
uponit7771
(90,335 posts)speak easy
(9,238 posts)Time to call in the fumigators.
regnaD kciN
(26,044 posts)speak easy
(9,238 posts)Hortensis
(58,785 posts)this entire century. Ask the wrong questions, get stacked, misleading answers offered as "proof" of lies.
There was a wakeup when the widespread popular response to Elizabeth Warren's big ideas belied what they'd been reporting. But they've been repairing that exposure ever since, and have staked a new offensive position that Democrats are split between conservative-leaning "moderates" and "progressives."
Horseshit. From our left to our right, we're overwhelmingly progressive liberals. There is almost literally no such thing as a Democrat who is not at least somewhat progressive, but usually very. Even our relative few conservatives.
Why would pollsters, whose product is enormously influential, not be infiltrated by the same RW agents who shape the product of most of the MSM, and of the Republican Party? Of course they are.
Because techniques to deceive, divide, demoralize, disillusion, divide, self-disenfranchisement, and divide by providing disinformation work.
Locutusofborg
(525 posts)Was fairly accurate. They predicted Biden 48% and Biden got 51.4% and they predicted Trump 47% and Trump actually ended up with 46.9%.
https://www.realclearpolitics.com/epolls/2020/president/us/general_election_trump_vs_biden-6247.html
speak easy
(9,238 posts)Yavin4
(35,437 posts)Data, no matter how bad it is, is never the problem. The problem(s) come(s) with how you use it.
538 should report on the polls, offer their analysis and adjustments, and don't aggregate them. Let the public look at each poll and come their own conclusion.
When they aggregate polling data, that's when they get into trouble because most people will see that top line number and run with it.