HOW CAN THIS POSSIBLY BE?? MADMAN UP +3 IN GALLUP??

[PearliePoo2]

WTF??
And this would include the Comey firing, right? Pathetic...



http://www.gallup.com/poll/201617/gallup-daily-trump-job-approval.aspx

[MADem]

1. Gallup is a rightwing pollster. That crew has always been in love w/the likes of

Nixon and Reagan and W.

They always lean right, and then give the benefit of doubt within the MOE.

In terms of approval ratings, I always subtract three to five from them if they're talking about Republicans, and add three to five if they're talking about Dems, and I think that's probably closer to the mark.

[RKP5637]

5. I never knew that about Gallup. I always thought they were neutral. Thanks! n/t

[titaniumsalute]

18. You couldn't more incorrect

I work in research and statistical methodology. Gallup has always been one of the highest credible research companies for reliability and methodology. If you look at all current polls Gallup has actually had his approval the lowest on average. Being a daily poll you will see quick + and - from day to day. Always has been like that. Remember standard error is +- 3 so that is why you see it bounce around.

You can add and subtract whatever you want but that doesn't make it scientific.

[MADem]

23. George Gallup was a die hard rightie. His legacy follows suit.

His only modern nod was that he liked a "wifely" paycheck. He thought women should be "allowed" to work.

His gave Dewey v. Truman to Dewey.

He conned Nelson Rockefeller, his buddy, into running because he felt he could beat Nixon.

He always weighted his polls in favor of Republicans.


I'm not the only one who thinks this--the fact that it is some kind of surprise to you is curious. You work in the industry and this is news? Really?

Here, some old school coverage of the issue-please read it, it explains the facts at hand quite concisely:

https://www.theguardian.com/commentisfree/2012/sep/25/gallup-rasmussen-polling-outliers-lean-republican


As far as I'm concerned, nothing's changed. My method works, so you can call it alchemic as opposed to scientific if you'd like, and call me Merlin, too--I won't mind! What it does do is account for the rightward tilt of Gallup.

[titaniumsalute]

25. Thanks

I'm very familiar with Gallup's background and the thoughts that his research skews to the right. In the world of research, however, this belief isn't as dominant. Gallup tends to be one of the most transparent companies with its methodology. If a reputable company wants to purposely skew their polls why would they put the methodology out there for everyone to see? These research companies exist because companies/governments, etc. want to produce research that is reliable for decision making capabilities mostly. Just because the CEO of a company is political one way or the other doesn't mean that their data from research will be skewed. I know some very left-wing CEOs/Owners of research companies and they don't skew their data. Although I'm sure people at Free Republic and Breitbart assume they do.

The Dewy V Truman argument. 1. Yes Gallup showed Dewey winning...as did Roper Research and Crossley Research. All three showed Dewey winning the election. 2. Gallup stopped polling a few weeks before the election which today ends on election day. 3. They used, as well an many other new pollsters, a flawed methodology called quota sampling. They had used this sampling method in 1936, 1940, and 1944 and predicted fairly accurately the winning of President Roosevelt. But Roosevelt was very popular and and was basically a YES/NO if you'd vote for him. 1948 showed a different dynamic with Roosevelt dead, a strong candidate in Dewey, and two powerful 3rd party candidates Thurmond and Wallace. After the big flop of the polling data in 1948 Gallup ditched the quota sampling methods and openly admitted/took responsibility to increase their reliability.

Of course George Gallup is now gone for quite some time.

So you can believe whatever you want of course. But I wanted you to see it from the world of research and why they are probably more reliable than you might think.

[MADem]

32. I'm not the only one who notices that they "oversample" Republicans

even after they acknowledged this and claimed to correct it, but didn't.

[titaniumsalute]

33. Again this isn't correct

You are implying that they purposely try to get more Republicans into their sample than Democrats or Independents. What would be the purpose of this? Who does it serve to try to get more Republican responses then others? That doesn't make any sense.

1. You cannot over or under sample something that is subjective. A random call to a respondent requires to ask that respondent, in general, if they classify themselves as Republican, Democrat, or Independent. People can say whatever they want to say in this answer and it is subjective. There is no finite benchmark of how many Republicans and Democrats are out there. Voter rolls and self-identification vary greatly by state every year. There is no single source for the best up to date registrations for voters at any single time. Therefore you cannot say they over or under sample Republicans because you need to know the exact number of registered Republicans to know if they are indeed over or under.

2. What purpose would it serve to over sample a particular group if they could do that? Any reputable research company isn't going to knowingly put out flawed research because then people won't hire you for future projects if they know you skew results purposely. All of these are for-profit organizations unless they are University based research groups.

[MADem]

34. Look, all I know is that they're WRONG.

Don't believe me? Ask President Romney.

[titaniumsalute]

35. OK. Maybe try learning something instead...

...of being argumentative. I've done this for 15 years. I'm very liberal and am fighting the cause as we all are. But simply believing or saying something regarding research doesn't make it true. THAT is the definition of creating fake news. Good day to you.

[titaniumsalute]

36. Here maybe this will help...

http://fivethirtyeight.com/features/the-polls-arent-skewed-trump-really-is-losing-badly/

[MADem]

39. They didn't do any better! nt

[mythology]

37. By this simplistic reasoning, most polling organizations

Are left leaning considering the results of the 2016 election.

Or the polls are based on assumptions about the electorate of a future event and it's hard to get that exactly right.

It's faulty logic to assume from a single data point like you're doing.

[Ilsa]

2. His people like the drama, the "crazy".

They eat this shit up as if it was cherries jubilee. He's their hero, telling us what he really thinks, no matter how fucking batshit insane it sounds.

[underpants]

3. It's Gallup

No we shouldn't pick and choose polls (as long as they are within +/- 3 MOE) but we can recognize how certain polls trend. Gallup and Rasmussen are very much right wing and often appear to serve mostly for their world's headlines.

[Lord_at_War]

10. Polls are essentially BS on BOTH sides...

1) You question 1000-1500 people for their opinions.

2) You have to "normalize" your "sample"- i.e. make a prediction of how well the people you picked actually represent the country at large. Do you decide on a 37% Dem, 31% Repub, 32% "Independent/"non-affiliated" split? Or do you go with a 35% Dem, 30% Repub and 35% "Independent" breakdown?

What is the % of Americans that support "Democratic Party" or "Republican Party"? The pollster doesn't really know, so he/she puts their own number on it- and uses that number to support whatever conclusion they wanted in the first place...

[titaniumsalute]

19. An overall poll asking approve/disapprove must be done at random

You poll 1000 registered or likely voters and you get what ever answers you get. They must weight for age, sex, geo, etc. But you will see day to day bounces from Gallup since it is a daily poll with standard error at +- 3.

[Solly Mack]

4. Reality shows do well. The more drama the better.

[lpbk2713]

6. The Яussians have found a back door.

Look for the numbers to go up a point or two every day until it is around 60.


(Please forgive. Some times hyperbole gets the best of me)

[haele]

7. There's a certain percentage of "citizens" who view life as a Reality TV show/Cage Match.

They're the ones who go "whoot, whoot, whoa - didja see that twist coming?" whenever Drumpf seems to get away with something outrageously thoughtless.
They're disengaged from a larger real world, and don't care what happens to anything outside of their echo-chamber fortress of comfort. Nothing is real to them other than the day to day experience. So yeah, they're going to be happy Drumpf is pulling one over on those shrill liberals who are apparently threatening to drag them out of their snow globes and participate in reality.

'Cause they're Amurican, damn it, and while their Boss-Man can tell them what to do and force them to kiss his Ass for a paycheck, "You" sure as hell aren't going to tell them what's best for them; they'll put a boot up anyone other than the Boss's ass for apparently trying.

Haele

[PearliePoo2]

11. Yeah, really, really sad.

Good post.

[JI7]

28. this needs to be an OP. with all the over analysis it really does come down to this

And we already know reality shows are most popular in right wing leaning areas.

[Hawaii Hiker]

8. He's still underwater at 41-54% @ Gallup

It's just that he went up 3 pts today....

http://www.realclearpolitics.com/epolls/other/president_trump_job_approval-6179.html

[geek tragedy]

9. tracking polls are prone to big swings and spikes.

they imply greater volatility than there actually exists

[Bucky]

12. Please no BS comments about Gallup slanting the data.

Numbers go up and down all the time in popularity polls. Trump has the full leadership the Republican party out flogging the media for him. This is both believable and nothing for us to worry about as Democrats. The main story continues that Trump is unpopular, and incompetent it running government.

[titaniumsalute]

20. Absolutely

I see people say it is a right wing poll. That is 100% BS. That is a simplistic explanation for something people might not understand.

[MADem]

24. You mean like they didn't slant it in 2012?

Look, Gallup DOES "slant the data."

Gallup and Rasmussen: the polling outliers that lean Republican
https://www.theguardian.com/commentisfree/2012/sep/25/gallup-rasmussen-polling-outliers-lean-republican


But given that Hair Furor has lost his ever-loving mind, that's the least of our worries. We can hold more than one idea in our heads, we're Democrats.

[BSdetect]

13. Cambridge Analytica / Mercer manipulating people?

[Tiggeroshii]

14. From 38 to 41 within margin of error?

Not a significant gain....

[smirkymonkey]

15. Not buying it.

sorry.

[Feathery Scout]

16. He fired the guy who 'let Hillary Clinton off'. He'll get a bump for that.

[Dr. Jack]

17. Oh I thought you meant he was at 53%

What a relief. Just statistical noise from a pollster that is way past its prime. Gallup has barely done polling in the past 5 years after botching the 2012 election in an epic fashion

[Goodheart]

21. They were given a false narrative by Fox News

My brothers are relentless Cheeto supporters who think Comey was fired for not indicting Hillary.

My brothers are not the brightest bulbs in the box.

[SoCalDem]

22. Gallup.. nt

[DemocratSinceBirth]

26. The link belies the headline- 39/56. It changed again.

[Doodley]

27. Gallup has been one of the least Trump-friendly polls. Margin of error means it fluctuates.

[Orsino]

29. 38% may be where he bottoms out.

The rest is noise, and 3% is the usual margin of error.

Besides, he had the weekend off. He's ready to start fucking up again this morning.

[wryter2000]

30. What's the margin of error?

Could simply be noise in the system.

[WePurrsevere]

31. A) Gallup leans right and B) "Margin of error is 3 +/- percentage points"...

I tend to take most polls with a large cow lick of salt. It can be too much like weighing yourself every day when you're on a diet.

When you do look at polls... read the the demographics, margin of error, etc. It might help.

[mythology]

38. That's the same as the margin of error

Any one data point isn't particularly useful. It's the trend over time.