Ignore Exit Polls

[Rohirrim]

Good article on why exit polls are worthless.

But exit polls are really more trouble than they're worth, at least as a predictive tool. An independent panel created by CNN in the wake of the Florida disaster in 2000 recommended that the network completely ignore exit polls when calling particular states. I suggest that you do the same.

http://www.fivethirtyeight.com/

[Garcia Bronco]

Of course exit polls are worthless. You don't even have to research it, it's full of so many confounding variables.

[Kaylore]

Good article on why exit polls are worthless.

But exit polls are really more trouble than they're worth, at least as a predictive tool. An independent panel created by CNN in the wake of the Florida disaster in 2000 recommended that the network completely ignore exit polls when calling particular states. I suggest that you do the same.

http://www.fivethirtyeight.com/

But I thought the "proof" Bush stole the election was that the exit polls favored Kerry. So are we to ignore exit polls if they favor McCain but if they favor Obama and McCain gets more, then they're right?

[Old Dude]

But I thought the "proof" Bush stole the election was that the exit polls favored Kerry. So are we to ignore exit polls if they favor McCain but if they favor Obama and McCain gets more, then they're right?


I think the studies concluded that, for whatever reason, democrats consistently respond to exit polls more often than republicans. So exit polls are almost always skewed in favor of the dems.

If the exit polls tilt toward Obama, then it means nothing, because it's expected that they will, even if the race is much closer. But if the exit polls show a big trend toward McCain, that would be a very bad sign indeed for the dems.

Chances are that none of the major networks will release anything very specific about their exit polls unless and until there is some very significant development. They've been burned too many times. However, they probably will release "secondary" info such as the degree of turnout, general demographics, priorities of issues, etc. Except for the turnout numbers, that should also be taken with a grain of salt, for the same reasons as above.

The turnout data could also be misleading in some states. For example, in Colorado, some people may have mailed in ballots that are defective for some reason (most likely due to inadequate identification or receiving the ballots too late.) In cases like that, the voters may go down in person and cast "provisional" ballots - which won't be counted unless their mail in attempt is rejected. So we may have a situation here where, on the surface of things, it looks like there are x number of mail ins (connected with either party) plus people showing up to cast provisional ballots - - but only one or the other will count.

As of early last week, something like 18,000 voters in Denver had not yet received their mail-in ballots. Given Denver's demographics, those were probably mostly democrats. Bottom line is that, at least so far as Colorado is concerned, we may not know the outcome until very late tonight or even Wednesday no matter what the turnout or exit polling looks like.

[Old Dude]

That said, my own super-secret exit poll sources tell me that turnout so far ...

1. Is surprisingly light in Florida.
2. Slightly more than expected in Georgia.
3. Pretty heavy in Missouri.
4. Utterly humongous in Virginia.

[Kaylore]

They should just stick to the real votes. Who cares what people say as they're leaving. What matters is what you wrote on the ballot.

[Spider]

In Wyoming , they had the kiddies vote as if they were part of the election ... vote for president of the USA , even 1 st graders ......Obama won in all of my Kids classes .by a landslide ........ Just goes to show kids do not listen to their parents ROFL! ....... I know alot of those kids come from republican households ......

[Kaylore]

Here's two polls you can bank on.

1. The Washington Redskins game before the election. An overwhelming majority of the time if Washington wins their game before the election, the incumbent party keeps the white house. If they lose, a new party takes power. The Steelers killed the Redskins. Advantage Obama.

2. The presidential candidate whose costume sells the most during Halloween always wins the election going back thirty years. Obama outsold the other costumes by a margin not seen since the Reagan years.

All signs point to an Obama landslide. You can thank me later for tipping you all off. :approve:

[Dudeskey]

Just ask John Zogby about exit polling!

[Garcia Bronco]

Here's two polls you can bank on.

1. The Washington Redskins game before the election. An overwhelming majority of the time if Washington wins their game before the election, the incumbent party keeps the white house. If they lose, a new party takes power. The Steelers killed the Redskins. Advantage Obama.

2. The presidential candidate whose costume sells the most during Halloween always wins the election going back thirty years. Obama outsold the other costumes by a margin not seen since the Reagan years.

All signs point to an Obama landslide. You can thank me later for tipping you all off. :approve:


the last Washington home game before the eclection is 17-0, soon to be 18-0, all time.

[Beantown Bronco]

I never knew this:

Among the 50 states, Missouri has a record of picking presidents that’s hard to match — the Show Me State has voted for the eventual winner in every election since 1904, with the exception of 1956, when it voted for Adlai Stevenson. Ohio’s not a bad predictor, either: It is almost always close to the national average, and no Republican has ever been elected president without carrying the Buckeye State. In fact, in the 14 presidential elections since 1952, Ohio has gone with the winner 13 times. Just three other states can boast that record of accuracy: Missouri, Nevada and Tennessee.

[Crushaholic]

That said, my own super-secret exit poll sources tell me that turnout so far ...

1. Is surprisingly light in Florida.
2. Slightly more than expected in Georgia.
3. Pretty heavy in Missouri.
4. Utterly humongous in Virginia.

Your poll is utterly humongous in Virginia? I would consider moving there, then...

[Bob]

Good article on why exit polls are worthless.

But exit polls are really more trouble than they're worth, at least as a predictive tool. An independent panel created by CNN in the wake of the Florida disaster in 2000 recommended that the network completely ignore exit polls when calling particular states. I suggest that you do the same.

http://www.fivethirtyeight.com/

We are agreed ... folks voting decide -- think 2004.

[Old Dude]

Your poll is utterly humongous in Virginia? I would consider moving there, then...

They don't call me "kickstand" for nothing ...

[alkemical]

Good article on why exit polls are worthless.

But exit polls are really more trouble than they're worth, at least as a predictive tool. An independent panel created by CNN in the wake of the Florida disaster in 2000 recommended that the network completely ignore exit polls when calling particular states. I suggest that you do the same.

http://www.fivethirtyeight.com/

maybe the exit polls are right, and that's our "receipt" for voter fraud :)

[Old Dude]

Of course exit polls are worthless. You don't even have to research it, it's full of so many confounding variables.

And they'll probably be even more worthless than usual this year. There's been a big surge in early voting across many states - and by both parties. So any kind of on-site exit polling will be blind to what, in some cases, could constitute anywhere from a third to a half of all the votes cast.

For example, let's say, hypothetically, that exit poling in Indiana shows this year that 22% of the electorate is under 30 - compared to 2004 when only 14% was under 30. On the surface, that might look like an indication that the youth movement is having a major impact. But it might be that under-30 voters, for whatever reason, were just more likely to vote in person on election day than to vote in advance.

Like you said, there are just way too many variables to sort out.