Skip to main content

#### 171comments | Permalink

• ##### They Also oversampled(13+ / 0-)

southerners over the age of 50.

http://www.people-press.org/...

Age
18-49 = 338
over 50 = 761

note:  Likely voters are probably skewed toward this age group.

Region
Northeast = 201
Midwest    = 271
South        = 417
West         = 223

Note:  if regions were sampled according to population (west and northeast dominant) I'd figure a big skew toward Obama)

"They Own You." --George Carlin

[ Parent ]

• ##### That's what this out-of-sorts feeling is today(4+ / 0-)

I've been oversampled.  Shoulda recognized it myself.

------
Ideology is when you give the answers before you know the questions.
It is what grows into empty spaces where intelligence has died.

[ Parent ]

• ##### But they weight it for region and age(1+ / 0-)
Recommended by:
grubber

so the raw sample doesn't mean a whole lot

• ##### It is not weighted to region(1+ / 0-)
Recommended by:
katesmom

for sure. Per the 2010 census (rounded down to the nearest million) the south was very much oversampled.

Regional Population [2010]
Northeast : 55,000,000
West : 59,000,000
Midwest : 65,000,000
South : 78,000,000

Pew sampling
Northeast : 201
West : 223
Midwest : 271
South : 417

Obviously, 78 is not more than twice 55, yet the poll surveyed more than twice as many southerners as northeasterners. So the poll could not have been weighted to population per number of respondents.

If you conducted the same poll but doubled your sampling in the northeast (strong Dem preference), no doubt the result would have been strongly favorable to Obama. Wouldn't it? Am I missing something?

• ##### You are confusing sampling and weighting(1+ / 0-)
Recommended by:
Joieau

Based on the sample obtained through the random digit dialing process, a disproportionate number of respondents lived in the south.  So when they calculate the overall percentages, they weight the responses from the other parts of the country more to balance it.  Thus, the reported percentages have been adjusted, to the extent possible,  based on real population counts, not on the makeup of the survey respondents.

• ##### also looks like they undersampled(0+ / 0-)

the 18-29 and hispanic demographics

"a lie that can no longer be challenged becomes a form of madness" -Debord

[ Parent ]

• ##### This is the most(0+ / 0-)

important comment!! Southerners over 50 were over sampled. This poll is crap. Even Rachel was fooled and reported it like a real poll. I thought she was smarter.

The only foes that threaten America are the enemies at home, and those are ignorance, superstition, and incompetence. - Elbert Hubbard -9.62/-8.15

[ Parent ]

Subscribe or Donate to support Daily Kos.

Click here for the mobile view of the site