The Fake 60% Approval Rating

[Lucrum]

Quote from Ricter:

So what's the argument here, that AP's sample is not representative because it does not have an equal number of those who ID as democrat and those who ID as republican?

Sounds like a good argument to me.

 

[bugscoe]

Quote from seneca_roman:

If I remember my stats courses, assuming the sample was truly random and there is no evidence it wasn't then the results represent the national mood. Also it shows almost twice as many people identify themselves as Democrats, bad news both ways for the GOP.

Seneca

Oh good lord.

If the sample had randomly selected only republicans and the result was a 10% approval, would you make the same argument? Would you also conclude that there are no longer any democrat voters?

No.

Get your head in the game!

 

[seneca_roman]

Quote from bugscoe:

Oh good lord.

If the sample had randomly selected only republicans and the result was a 10% approval, would you make the same argument? Would you also conclude that there are no longer any democrat voters?

No.

Get your head in the game!

That is not what happened. But, if the sample was truly random, then I would conclude that the results were valid:

1. O. had a 10% approval.
2. All eligible voters were Republican.

Seneca

 

[bugscoe]

Quote from seneca_roman:

That is not what happened. But, if the sample was truly random, then I would conclude that the results were valid:

1. O. had a 10% approval.
2. All eligible voters were Republican.

Seneca

And truly reflective of the national mood, correct?

 

[seneca_roman]

Quote from bugscoe:

And truly reflective of the national mood, correct?

Yes.

BTW-what is it you don't like about the AP-GfK poll? I hope you are not mixing the observed results vs what you think they ought to be.

The Gallup poll sheds some more light on the current trend:

Republicans have shifted to favor Obama.

Of course, we Americans are a fickle bunch, and Obama's/Democrats new popularity can change quickly, especially if the economy gets worse.

Seneca

 

[Ricter]

Randomness is not the central issue, though it is important. The representativeness of the sample is being questioned. That's fair, and sampling methodology is well understood. This is where the author needs to work, the assertion "it aint so" is largely baseless at the moment.

 

[Tsing Tao]

here i a good write up from dick morris. i've omitted most of the opinional commentary and tried to just copy the interesting facts.

The AP sample contains twice as many Democrats as Republicans (35% Democrat vs. 18% Republican). While party identification is, itself, a variable to be determined by polling, the normal sample usually contains an approximately equal number of Democrats and Republicans. Lately, Republicans have actually been slightly in the plurality in most statistically valid samples.

This AP poll — with its 2:1 Democratic edge — is utter hogwash.


Even by the standards of AP’s previous polls, this sample doesn’t stand up. It is vastly more Democratic than any of the organization’s recent polls:

PARTY IDENTIFICATION IN AP POLLS

Democrat Republican
October, 2010 32% 25%
November, 2010 29% 24%
January, 2011 28% 25%
March, 2011 29% 20%
May, 2011 35% 18%

Of course, all of these AP polls reflect a decided pro-Democrat bias compared to most national survey findings. The Associated Press needs to search its soul — and consult with its polling company — to determine if this bias can be corrected.

Most polls, such as that by Rasmussenreports.com, suggest that Obama has, indeed, gotten a single digit bounce from the bin Laden killing. Rasmussen has his approval now at 48% after it had languished in the low 40s for some months.

 

[Ricter]

Hmm, looking at the data since October, it's not quite as "shocking". Still worth looking into.

 

[Tsing Tao]

as far as i'm concerned, rasmussen is the best in the business.

 

[seneca_roman]

Quote from Tsing Tao:

here i a good write up from dick morris. i've omitted most of the opinional commentary and tried to just copy the interesting facts.

Here is what I see when looking at the AP data re. party ID:

Oct 2010: 48% Republican
42% Democrat

May 2011: 29% Rep.
46% Dem.

So, AP is noting a down trend in the people who say they are Republican.

Other polls have it closer to an even split, but trending in the Dems. favor. EG: Reuters/IPSOS has it 44% Dem./ 40% Rep.



http://www.ap-gfkpoll.com/pdf/AP-GfK_Poll_May_Topline_051011_POLITICS.pdf

Seneca