Fun with exit polls part 1 » Forum Post by Frogboy

I can't capture the photo but right now cnn has ohio with 67% for obama (of votes counted).

http://www.cnn.com/ELECTION/2008/results/president/

Successfully updated karma reason!

Ah, Ohio...people only know we exist during election years.

~Zoo

Successfully updated karma reason!

Obviously if Obama wins Ohio McCain is toast.

Ohio just got called for Obama.

Successfully updated karma reason!

Draginol

Join Date 03/2001

+137

McCain is toast.

Successfully updated karma reason!

Mumblefratz

Join Date 07/2006

+139

Last time (2004) they buried their exit polls after it became clear they were ridiculous biased for Democrats (no bias in the media eh?).

They actually did a reasonable job of explaining this on Fox.

Exit polls are based on a limited number of people that are willing to spend about 10 minutes filling out a survey after having voted. The idea is that how "exuberant" a particular set of voters feel about the potential results of the election will affect their willingness to spend time and participate in such a survey. Given that Obama supporters had reason to be optimistic about the eventual outcome of the election it's not unreasonable to assume they would be more statistically willing to participate in the exit surveys and thus be over-represented.

This is a well known effect which is easily compensated for by waiting for some real results to be reported from precincts where exit poll information has been gathered. The difference between real election results and exit poll data (referred to as Within Precinct Error or WPE) in sample precincts can then be used to correct exit poll results for the entire state.

Sorry to disappoint you, but this is *not* evidence of any kind of media bias.

+1

Successfully updated karma reason!

Sorry to disappoint you, but this is *not* evidence of any kind of media bias.

Owned!

Successfully updated karma reason!

Mumblefratz

Join Date 07/2006

+139

Owned!

I wouldn't say that. This is simply one of many possible examples of where something that makes sense intuitively leads you to an incorrect conclusion. Even with some experience in statistical polling methods this is an easy mistake to make. That's why drawing conclusions from polling data is a dangerous thing to do even with extensive experience in the field. There are always implicit assumptions made that may be incorrect and there's always the potential of some unseen correlation between what you intuitively think are independent variables.

Always remember, "There are lies, damned lies, and statistics". In the end the only thing that's reasonably sure is that more people in Ohio voted for Obama than voted for McCain. When you start to try and postulate reasons why this was so there's always a significant element of guesswork. The intent of my reply was to point this out and try to discourage people from making conclusions about people's motives that are truly impossible to discern from the data. The data just *is*, the *why* of it is usually unknowable.

Successfully updated karma reason!

Always remember, "There are lies, damned lies, and statistics". In the end the only thing that's reasonably sure is that more people in Ohio voted for Obama than voted for McCain. When you start to try and postulate reasons why this was so there's always a significant element of guesswork. The intent of my reply was to point this out and try to discourage people from making conclusions about people's motives that are truly impossible to discern from the data. The data just *is*, the *why* of it is usually unknowable.

On the other hand, statistics are often the only way to study some phenomenas. It's hard not to rely on those for things like politics when you are using them for so many things that are core to our society.

I should know, I studied in Statistics

Successfully updated karma reason!

Mumblefratz

Join Date 07/2006

+139

statistics are often the only way to study some phenomena. It's hard not to rely on those for things like politics when you are using them for so many things that are core to our society.

Of course. Just don't try and use them to attribute specific motives to individual action and they are fine predictors of future behavior. It's just that you can predict *what* people will do but to attribute a specific reason as to *why* they do it is beyond the field of statistics and into the field of behavioral psychology.

Successfully updated karma reason!

Draginol

Join Date 03/2001

+137

Quoting Cikomyr,

comment 6

Owned!

You know what's amazingly idiotic? Antagonizing the owner of the site.

Successfully updated karma reason!

Draginol

Join Date 03/2001

+137

So it actually ended up 51% to 47%. Quite a bit different than the 53% to 45% exit poll.

Now, being quite familiar with statistics as well as building statistical models, my point is that these exit polls consistently favor Democrats and yet they never alter their models to reflect this.

This wouldn't be such a big deal except that they often will call even somewhat close races based almost purely on this data.

Successfully updated karma reason!

You know what's amazingly idiotic? Antagonizing the owner of the site.

What are you gonna do? Ban me? Destroy my Stardock account? What will it really change?

I am quite aware of your positions, and of your capabilities. But I think you are a very moral person, based on your views toward a lot of things in the Gaming and Software industries - and the reason behind many of your political views.

I just don't think you are the kind of person that will ban a poster for something like up-there. If I started to really be a disruptive element on this board, you'd not even need to do the work yourself; you have competent moderators. But for a crack?

Now, I wouldn't do a crack against one of the Sheiks in the country I am - UAE. They just don't have the same kind of morality than we do, and I know they would ban me out of the COUNTRY. But lemme tell you I feel a lot less at ease here than I do on this very board. Is that what you really want, a board of peoples who phonnily smile at you?

I haven't kept artificial politeness toward you, and I have always genuinely been honest with you. I think knowing someone is honest with you is worth bearing a crack from time to time.

But then, maybe I am wrong, and then, Good Bye JU [e digicons]:thumbsup:[/e]

Successfully updated karma reason!