PDA

View Full Version : Misc How Pollsters Affect Poll Results


DaFace
09-15-2008, 09:04 AM
I know I've seen a couple of discussions on here about how polls are useless or why the results differ from one organization to another. I ran across this blog post today and thought some of you might find it interesting.

http://politicalarithmetik.blogspot.com/2008/08/how-pollsters-affect-poll-results.html

I'll warn you that it's a bit technical, and the graphs take a bit of work to understand. For any of you interested in this topic, though, it's an enlightening read.

HonestChieffan
09-15-2008, 09:08 AM
Most people have no idea of statistics. No possible way to communicate the issues of sample, sample size, make up and variations on a discussion board.

Garcia Bronco
09-15-2008, 09:20 AM
You also have to consider the halo effect.

jAZ
09-15-2008, 09:38 AM
This is the issue.

Publicly published polling is conducted with fair intentions across the board, but with a "for profit" bias. And by that I don't mean partisan towards or against one party, but conducted to make news, attract interested viewers and thus profit to the paper and pollster.

IMO, the only meaningful polling is the polling we never see (unless it's released for partisan reasons) and that's the internal polling of the canidates. Those parties are the only ones with a vested interest in getting the results as close to perfect as possible. Everyone else has a vested interest in their poll making news (good or bad for Reps or Dems alike).

That's not to bash the pollsters for cheating or having a bias. They don't and they don't. But there are systemic factors that lead to their existance in the first place and those same factors lead to decisions that first ensure their continued existence and second to accuracy.

RINGLEADER
09-15-2008, 09:52 AM
This is the issue.

Publicly published polling is conducted with fair intentions across the board, but with a "for profit" bias. And by that I don't mean partisan towards or against one party, but conducted to make news, attract interested viewers and thus profit to the paper and pollster.

IMO, the only meaningful polling is the polling we never see (unless it's released for partisan reasons) and that's the internal polling of the canidates. Those parties are the only ones with a vested interest in getting the results as close to perfect as possible. Everyone else has a vested interest in their poll making news (good or bad for Reps or Dems alike).

That's not to bash the pollsters for cheating or having a bias. They don't and they don't. But there are systemic factors that lead to their existance in the first place and those same factors lead to decisions that first ensure their continued existence and second to accuracy.

Very true.

All you have to do is change the weighting (which no one knows to be honest) and you can make the same poll that shows McCain ahead +10 show Obama +10. Currently most are re-weighting to the Dems favor by anywhere from 5-10 points but if Obama is able to translate the investment he's made on the ground that advantage could be even bigger on election day. Or, it could implode. No one really knows because no one has really spent the money to create the kind of infrastructure he has put together.

All that said I think the polls do give you a sense of direction and not a lot more. That direction has been towards McCain but now that Obama is finally targeting the top of the GOP ticket instead of the bottom of it that will start to change I'm sure.

I still say that we'll have a great sense of who's going to win this thing after the Veep debate and there won't be a lot to move the narrative, absent some big revelation, to move the remaining squishy voters who haven't made up their mind by then. But if I was in the Obama camp I'd be prepared for the onslaught that will attack his experience in the final two weeks of the campaign -- it's where he's most vulnerable and can be framed to bleed into other topics like the economy.

jAZ
09-15-2008, 10:05 AM
... the weighting (which no one knows to be honest) ... Currently most are re-weighting to the Dems favor by anywhere from 5-10 points...
How can you reconcile those two statements?
...now that Obama is finally targeting the top of the GOP ticket instead of the bottom of it that will start to change I'm sure...
That's not a very accurate portrayal. The Obama campaign hardly touched Palin. They overwhelmingly if not entirely focused on McCain this whole time. That's not to say that supporters and bloggers or media have followed suit.

alnorth
09-15-2008, 11:03 AM
How can you reconcile those two statements?


Ignoring the accuracy of the statements, I dont see what the first statement has to do with the other. The first is a technical statement about how pollsters decide to weight their polls based on what they think the turnout will be, the second is Obama strategy and a prediction on how it will impact the race.

(He's right on the first statement by the way, every poll assumes the Dems will turnout 5-8% more than the GOP, and McCain is winning in spite of that inherent disadvantage because of the independents and Obama's failure to completely lock up the self-identified Democrat voters.)

RINGLEADER
09-15-2008, 11:27 AM
How can you reconcile those two statements?

No one knows what the final weighting on election day will be. But the pollsters make decisions on the weighting based upon what their polls tell them and historical models. Currently they weight in favor of Dems which I don't disagree with given the sentiments that are out there but they could be under-weighting or over-weighting that adjustment.

That's not a very accurate portrayal. The Obama campaign hardly touched Palin. They overwhelmingly if not entirely focused on McCain this whole time. That's not to say that supporters and bloggers or media have followed suit.

They mobilized a bunch of female Dem proxies just this last weekend! :shrug: