Heh, Pew Research Poll... sampled 80% Moderate to Conservative voters, nearly 40% from the south,

[berni_mccoy]

Last edited Mon Oct 8, 2012, 11:55 PM - Edit history (1)

and Hispanics not factored in.

Remember the national poll today that had Romney ahead?

http://www.people-press.org/files/legacy-pdf/10-8-12%20Political%20Release.pdf

Out of 1112 Likely Voters

by RACE
White, non-Hispanic
869
Black, non-Hispanic
116
Hispanic
57 (they did not actually factor hispanic vote into percentages)

by Region
Northeast 201
Midwest 271
South 417
West 223

by Ideology
Conservative 505
Moderate 382
Liberal 204

Nuff said.

[regnaD kciN]

1. Does that mean we should start "unskewing" polls?

[DemocratSinceBirth]

3. No. But I Do Think, As We Discussed, When The Poll Was Taken, Is Of Great Consequence

Their internals are what they are but they were in the field at the worst possible time for the pres and left the field just as other pollsters suggested things were looking a little better.

[berni_mccoy]

5. This is raw data. Don't think anything was skewed.

[Bjorn Against]

22. No, it means we should reject polls with bad methodology

There is no way to unskew a poll, if it is skewed it is a bad poll. I am not a person who dismisses polls without good reason, but based on the numbers above there is good reason to dismiss this one. A good poll requires a random sample taken from a wide range of people, the demographics of this poll do not even come close to representing the demographics of America. I would say the same thing about a poll that was concentrated in the Northeast and underrepresented the South.

[movonne]

63. I say pew to pew...

[magical thyme]

2. they also apparently did the vast majority of their polling immediately after the debates

and before the Sunday Obama surge. They did only 1/10th or so of the poll on Sunday, the day that Gallup had a big surge for Obama.

[DemocratSinceBirth]

4. I Really Don't Think That Qualifies As Unskewing

.

[Yo_Mama_Been_Loggin]

6. Some real demographics eh?

Almost makes me want to use the Sarah Palin phrase "lame stream media".

[patrice]

7. K&R

[Cha]

8. What the Heck, Pew!

Screwing with my head.

Thanks berni

[Panasonic]

9. In other words, an outlier.

And should be dropped. Incorrect sampling, plus the samples are way too small. 1200 people?

Like I said on the other thread, I'd rather sample 20,000 people to get a better idea.

So therefore, the polls from here on is ignored because M$M was a horse race, and it's a Arabian vs a mule (as in Rmoney is a mule) and they're desperate to keep the ad revenue rolling.

[WCGreen]

35. The size of the sample doesn't matter after say 1,000 as long as your

sample reflects the population you are interested in polling.

Sampling is far more important than the size of the people contacted.

Usually, one voter in an area is chosen and then a cluster of voters is determined from that starting point. You then look to the surrounding cluster to find the demographic you need.

It also allows you to call others in the area if that person isn't available. That would be left to the caller.

I haven't done any polling since the demise of the family wall phone. So I am sure they have a different but similar way to build their sample.

[SheilaT]

49. Actually, if the sample is a good one, meaning properly weighted,

1200 is the right number of people to sample to get a very good picture of what's going on.

Way back when I took statistics I learned the math behind why 1200 is sufficient. I no longer recall it, alas.

[Grammy23]

66. Yep, I took statistics a long time ago and was not a wiz kid when I took it---

But in spite of that, I do remember that HOW you select your sample is really important. If you don't select a sample that fits the people you are trying to probe for their opinions, you're not going to get an accurate picture. Even well designed samples can go awry, but most of the time you get pretty close to accurate.

It would be helpful if the pollster told us how and why they designed their sample the way they did. Unless they don't want us to know.

And 1200 sounds like the number I remember as the right amount to uset if your sample is designed right. You do not need to ask 25,000 people if you've done your homework.

[patrice]

10. WHY would they do that?

[CakeGrrl]

14. The Corporate-controlled MSM? They have skin in this game.

Romney will save THEM lots of money.

Besides, this is the first thing they could grab to hint at a Romney comeback. They're desperate for one.

[Doctor_J]

16. because they're owned, run, and staffed by right-wing Repukes?

:duh:

[patrice]

17. Yeah, I guess I need to look at Pew closer; I thought it was research . . .

[reformist2]

11. How embarrassing. All polls should have this basic data prominently featured.

Don't "unskew" the polls. Skewer the pollsters when they release crappy ones like this!

[PD Turk]

12. this is good news though

If Rmoney can't even pull 50% with a crowd like that he's hurtin'

[berni_mccoy]

13. yep

[patrice]

15. Thanks! Berni! I shared it and will continue to do so daily for a while, a week maybe.

[Tigress DEM]

60. BUT if all they want to do is STEAL an election, then these PHONY polls are support of his "WIN"

[FiveGoodMen]

64. I'm pretty sure that's the plan.

Going back to 2000, they kept telling us that none of Dubya's gaffs were really hurting him...

...and then they stole it.

Looks very familiar.

[secondwind]

18. Diane Sawyer announced this poll tonight... ABC News should know better!

[demwing]

30. No, "Pew" is a name that should be trustworthy

who knew they really do stink?

[alp227]

38. Really? Pew had a good reputation for honest, nonpartisan research.

That means pew conclusions will not always be withe left our right always want to hear.

[demwing]

54. We're saying the same thing

Pew was once a trusted name. Why would anyone "know better?"

[SidDithers]

19. Here are the numbers...

check the N value column for how many polled in each category.



Sid

[maxrandb]

20. Do I read that right?

they polled twice as many people from the South than from any other region??

[Bjorn Against]

26. Yes, and they polled the fewest number of people in the Northeast and the west

The Northeast and the West of course are not only Democratic strongholds, they are the most heavily populated regions of the country so they should have the largest samples.

[James48]

34. Yes-

It appears they polled mainly white conservative men over age 50 from the south.

That's the biggest single group in their data. I would think that Romney would have scored better among white conservative men over age 50 from the south.

[maxrandb]

39. By those numbers

RMoney would have a hard time winning the Presidency of the Confederacy!

If everything goes according to the MSM plan, Obama will hit a home-run at the next debate, bringing debate #3 up as the "ultimate" rubber game.

I don't know what's worse. That the "news" today sucks so much, or that we've become so easily manipulated.

Oh well, gotta watch Honey Boo Boo.

[tritsofme]

21. 1 out of every 20 polls is complete bunk.

This is one of them.

[caseymoz]

25. Scientifically speaking, yes.

That's something pollsters don't tell you. There's a 95% chance, if they follow all the procedures, that the poll will be correct within the margin of error. This means there's a 1/20 chance that the poll is utterly wrong.

And that's if they follow all the procedures correctly. This looks like a bad sample.

[pkdu]

23. When 10% of your sample says they are liberal and voting Romney.....

Your data is a pile of shit.

[Wounded Bear]

24. So this is why they attacked the BLS......

cover for putting out bogus polls.

[SansACause]

27. Gallup switches to LV model tomorrow.

The Obama +5 will disappear literally overnight as Gallup goes to likely voter model. We need to GOTV!

[Sparkly]

28. Yet Chris Cillizza wrote a WaPo article of "takeaways" based on that poll alone.

http://www.washingtonpost.com/blogs/the-fix/wp/2012/10/08/8-takeaways-from-the-new-pew-poll/?hpid=z2

And WaPo has it on their front page.

Just to stay "fair and balanced," I suppose.

[cheriemedium59]

29. Whether we agree or disagree with the data on each poll...

This is going to be a roller coaster ride until the finish..

[OldDem2012]

31. I posted this in another thread...

...The Pew Research Center is funded by The Pew Charitable Trusts :

"The Trusts, a single entity, is the successor to, and sole beneficiary of, seven charitable funds established between 1948 and 1979 by J. Howard Pew, Mary Ethel Pew, Joseph N. Pew, Jr., and Mabel Pew Myrin—the adult sons and daughters of Sunoco founder Joseph N. Pew and his wife, Mary Anderson Pew. The Trusts is based in Philadelphia, Pennsylvania, with an office in Washington, D.C.

Although today the Pew Charitable Trusts is non-partisan and non-ideological, Joseph Pew and his heirs were themselves politically conservative. The mission of the J. Howard Pew Freedom Trust was to "acquaint the American people with the evils of bureaucracy and the values of a free market and to inform our people of the struggle, persecution, hardship, sacrifice and death by which freedom of the individual was won." Joseph N. Pew, Jr. called Franklin Roosevelt's New Deal "a gigantic scheme to raze U.S. businesses to a dead level and debase the citizenry into a mass of ballot-casting serfs."

Most of the early beneficiaries were such conservative organizations as the John Birch Society, the American Liberty League, and the American Enterprise Institute, although other beneficiaries included a cancer research institute, a museum, higher education, the American Red Cross, and historically black colleges. For many years, the Trusts tended to fund charities and conservative causes in Philadelphia.

In 2004, the Pew Trusts applied to the Internal Revenue Service (IRS) to change its status from private foundation to non-profit organization. Since the Pew's change to a charitable foundation, it can now raise funds freely and devote up to 5% of its budget to lobbying the public sector."

Always follow the money...it usually tells you quite a bit about the philosophy of those handing out the checks.

[ejbr]

32. The female voter switch seems absurd

56 - Obama, 38- Romney to 47 Obama, 47 Romney?

White women

49 Obama, 46 Romney to 38 Obama to 57 Romney?

I just can't imagine women being so naive as to change their views based on a debate when you still have Rmoney supporting the personhood amendment, among other outrages.

[JoeyT]

51. Same reason everything else changed.

Asking white women spread around the country is going to get much different answers than asking white women that live in the deep south and identify as conservative, which is apparently what they did to get these results.

It's likely that women's support of the president hasn't actually changed. Bad polls with bad methodologies give insane results. GIGO.

[DavidL]

33. Who the heck does a poll like that and say it's a "random" sample

of the USA?

almost 40% from the South? less than 5% Hispanic? almost 48% Conservative?

This represents a "random representative, statistically significant sampling" of the current US likely voter population?

Who bothers to publish such an outlier polling?

[Sirveri]

56. If you poll primarily in the South, you should not be surprised with conservative turnout.

We need to stop freaking out about polling.

First off, national polls are worthless, the only numbers that matter are state polls since the vote is by electoral college.

Second, the Republicans like to freak out about polling methodology. Don't be like Republicans.

Third, who cares? The only poll that matters is the one in November, while poll tracking is fun, it's ultimately useless for most everyone here since we're not running or advising the campaign on what strategy to employ and which state to target.

So they polled the South, a lot of Republicans were home, and they skewed it, so what? It's a regional over sample, adjust down to account for population and run the numbers. That grants you a national poll number +/- 5%, which is worthless since the election is decided state by state.

[Dawson Leery]

36. Pew just proved themselves to be garbage.

Half of the electorate is "conservative"?

Why did they poll so many white conservative men over age 50 from the south?
The Northeast and West were under represented in this poll.
Latinos were not part of the sample.

Shame of Pew. Fitting name.

[Ilsa]

37. The sample also has ALOT of people over 50.

I realize older people are more likely to vote, but this is very skewed (2/3) towards older voters.

[OldDem2012]

40. Holy crap....Rachel's using the OLD poll numbers!!!....

Who the heck is responsible for getting her the CORRECT numbers??

Incredible.

[Gabi Hayes]

41. since people are going totally ape over all the polling swings, this thread is one of the most

valuable I've seen around here in quite some time

I'd like to see some non-partisan statistician analyze what these numbers mean, but to, a know nothing layman, they are definitely redolent of the eponymous 'foundation' whence they emanate

thanks for the info on Pew. I'd somehow developed a notion that they were, somehow above the fray, certainly unaligned.

I remember talking with one of their spokespeople on a radio show a few years ago, and was taken aback by his attitude on the question of the day, which, IIRC, had something to do with the myth of the liberal media, to which myth he heartily subscribed

the Liberty League, for lord's sake? they were behind the Business Plot, weren't they?

chekachekachekacheka......

yup:

http://www.rooseveltinstitute.org/new-roosevelt/how-fdr-took-forces-wealth-and-power

[ClusterFreak]

42. Push poll. Throw it out. n/t

[City Lights]

43. A concerted effort to mislead the public.

Screw the pollsters and the corporate media that promote their lies.

[Gabi Hayes]

44. need to know anything more about Pew? read this:

http://www.usnews.com/opinion/blogs/mary-kate-cary/2009/10/14/fox-news-less-biased-than-cnn-msnbc-in-white-house-coverage-

Fox is less biased than other news outlets?

love to see their working numbers!

[ProudToBeBlueInRhody]

45. There's a bigger question here

Has the term "liberal" been vilified to the point you can't find enough people willing to admit they are to poll them?

[JackN415]

46. If I were Pew and need to be in the news limelight

I would deliberately choose certain sample that would produce "interesting" or "breaking news" results. For example, if I sample members of DU, I wonder who might be in the lead?

[TomCADem]

47. Never New that The South Was As Populous As West and Northeast Combined!

I guess you learn something new everyday.

[svip]

50. About regional specifics

There are alot of things to question this poll about, but regional specifics are not one of them for the most part.

The south, as most consider it, makes up about 16 states. The population of those 16 states is about 114,500,000 (rounded). It makes up a little over 37% of our population. In the poll, southerners are represented by about 37.5% of LV.

They do need to flip the midwest and west to get a more accurate representation solely of population.

But that's before going into the number of eligible voters from each region, which could change the numbers completely. And I'm not taking the time to look into that.

As far as voter turnout of those registered in each region, it was nearly identical in every region in '08.

I'd focus more heavily on the age of voters. +5 Repub crosstab. Voter ID for Repubs level beyond even that of 2010. Women swinging nearly 20pts in favor of Romney. Young voters going to Romney. And more importantly, Romney gaining 10pts on who would help the middle class and poor.

Remember, Pew's last poll was out BEFORE the 47% comments. But even with those, Pew still has Romney gaining pretty heavily in this regard.

[TomCADem]

52. But from an electoral college point of view, heavy pro-Romney Southern support...

... does not translate into more electoral votes, since Romney was winning in the South anyways.

[svip]

53. Very true

But this poll isn't a state poll, but a national one. Only meant to reflect the feelings of the nation currently.

But like I said, there are alot of things to question in this poll. It's just regional specifics aren't one of them.

[berni_mccoy]

55. That population you cite includes Maryland, Delaware, VA and DC

None of which are likely to go for Rmoney, and most have heavy African American populations. The poll does not sample likely voters indicative of that region of "the south" as you are defining it (and yes, I read the same wiki article you are citing). Take away those heavy Democratic regions of the South and you've got a far reduced population from the 37% you are referring to (more like 25%).

[trailmonkee]

48. that's why they call it pew!

[Tsiyu]

57. Of course

Pew = FOS.

They sure got all the bully-adoring pundits drooling, though.

[Tyson73]

58. 2 to 1

I"ve heard it said that there are more conservatives than liberals.

But not by more than 2 to 1. That's rediculous.

505 COnservatives
204 liberals ?????????

Give me a break.

I think we've gone into "unskewing" overload.

Repukes are desparate.

[Vattel]

59. Based on census numbers the south is not over-represented.

The Midwest seems a bit over-represented though. Depends on how you divide the regions.

Hispanics do seem to be under-represented. Maybe there is an explanation though.

[Wednesdays]

61. K&R

[Roland99]

62. So, 5-2 Repuke over Dem (and moderates in between) and the poll is still that close?

I'll take that!

[Blue Idaho]

65. This poll supports the required M$M narrative.

The M$M needs a "horserace" to sell the final blocks of political ads at the highest possible rates. They also need to drive viewership of the countless talking heads political opinion shows on every channel from Fox to MSNBC. I'm not saying opinion was not swayed by the first debate but I am asking who does the this polls demographic really benefit?

I say it benefits the crooked media even more than it benefits the republican party.

[lindysalsagal]

67. This is perfect. It will just make the repugs that much more pissed off when it drops them down hard

on the floor again.

I love it!