If you can read this, either the style sheet didn't load or you have an older browser that doesn't support style sheets. Try clearing your browser cache and refreshing the page.
Fark SearchWeb Fark

         more options... Create account

(The Morning Call) Obvious Ever wonder why the poll numbers vary so much? Check out the 'changes and adjustments' they make   (mcall.com) divider line 42
More: Obvious  
•       •       •

3369 clicks; posted to Politics » on 25 Oct 2008 at 1:09 PM   |  Make this a Fark FavoriteFavorite    |   share: Share on OMGTWITTER WEB2.0share on StumbleUponshare on Facebook  more»   |    Get this fabulous T-Shirt and impress the methane out of your friends! shirt it!

42 Comments   (+0 »)


Fark.com's  Political Inclination Thermometric Analyzer:
Neutral 3.04% Fascist
Archived thread
 
downstairs [TotalFark] 2008-10-25 11:53:49 AM  
The nice thing about polling is there is no way to measure if a poll is right or wrong, seeing that opinion sways from day to day and the only 100% accurate poll is the actual election.

So Pew Pew Pew is off by 5% on October 25? Well, they can say McCain simply got more popular over the next week and a half.

 
UNC_Samurai [TotalFark] 2008-10-25 11:59:21 AM  
Someone's already figured out the polls:

farm4.static.flickr.com

 
dahmers love zombie [TotalFark] 2008-10-25 12:24:19 PM  
Done correctly, proportionate sampling is fine. It's (as the article notes) the definition of "likely voter" that is the sticking point. Asking people "are you likely to vote" may not be the best indicator, nor might be a measure of how many times they've voted in the last N elections (this one seems to be engendering more voter interest, but honestly they say that every time).

If they wanted to be a bit naughtier, but perhaps more accurate, they could shoot for a pre-election "exit" poll of early voters. They'd probably get in trouble, however.

 
sloppy shoes 2008-10-25 12:47:29 PM  
dahmers love zombie: Done correctly, proportionate sampling is fine. It's (as the article notes) the definition of "likely voter" that is the sticking point. Asking people "are you likely to vote" may not be the best indicator, nor might be a measure of how many times they've voted in the last N elections (this one seems to be engendering more voter interest, but honestly they say that every time).

If they wanted to be a bit naughtier, but perhaps more accurate, they could shoot for a pre-election "exit" poll of early voters. They'd probably get in trouble, however.


I don't think they'd get in trouble, but they would have 2 main problems- 1. identifying who has already voted and 2. early voters are doubtfully a representative sample of total voters.

 
sloppy shoes 2008-10-25 12:53:50 PM  
I should add,

The problems with polling are the same with economics- there are many different ways to see the issue and there are many different data sets to analyze- that's why it is so important to talk to multiple economists and look at multiple things.

Sorry for the rant, but I'm just furious with how much of the media is analyzing the financial crisis. They keep talking about swings in the Dow Jones as if that's God's gift to finance. Yes it's important for your 401k or whatever else you have, but the heart of this financial crisis is not the stock market.

If people are worried about jobs or unemployment or many other things to come, which is barely being talked about compared to all the non-news about the election, they need to start looking into some risk indexes, consumption (which they are starting to do), and other variables.

Sorry. Rawr.

 
dahmers love zombie [TotalFark] 2008-10-25 12:54:36 PM  
sloppy shoes: I don't think they'd get in trouble, but they would have 2 main problems- 1. identifying who has already voted and 2. early voters are doubtfully a representative sample of total voters.

Both good points (I guess you could ask them "have you voted", but...). I do think they'd get in trouble for the same reason that networks have gotten hand-slapped before -- discussing a state's actual results (through early returns or through exit polling) before the polls close.

 
sloppy shoes 2008-10-25 01:15:04 PM  
dahmers love zombie: sloppy shoes: I don't think they'd get in trouble, but they would have 2 main problems- 1. identifying who has already voted and 2. early voters are doubtfully a representative sample of total voters.

Both good points (I guess you could ask them "have you voted", but...). I do think they'd get in trouble for the same reason that networks have gotten hand-slapped before -- discussing a state's actual results (through early returns or through exit polling) before the polls close.


I don't think the state can release the results early. So a network shouldn't be publishing official results until the county clerk or whoever certifies them.

The problem with early voting and asking "Have you voted?" is you only call a limited sample. A good pollster will try work out his conditions before he calls people. Given how little early voting there actually is, if you call 1000 or 1500 people needed to get a decent sample, you're only gonna get 300 or less who have actually voted on average. (I doubt more than 30% of votes are early, but I could be wrong).

Now, what I would say- I don't think they should be publishing the results of early voters. Even though you don't actually find out the individual names of voters, it still eliminates the concept of secret ballot partially. For primaries I think having staggered voting is fine because it allows the party to coordinate with the rest of the country and not split itself up; however, for the general election, I think it would lead to problems.

However, if we ever get to the point where we have, say, a 2 week voting period, I suspect that you would actually be able to use early voting as a predictor for the elections because 1. enough people would early vote and 2. if it were everywhere in the country you could do state by state models that probably got pretty accurate.

But then you have the problem of last minute voters viewing the election as already over if past early voting models were correct. In which case you just lead the choice to vote down this cyclical argument over whether or not to vote based on what's happening with the polls. That's already a problem.

 
Mortician 2008-10-25 01:19:53 PM  
The problem with sampling early votes is that they're the decisive ones :) And people wonder why early voters seem to be split 50/50...

 
roncofooddehydrator 2008-10-25 01:27:21 PM  
Come on now people, all farkers should know by now that fivethirtyeight.com is the definitive place for polling information.

 
organizm 2008-10-25 01:37:52 PM  
roncofooddehydrator: Come on now people, all farkers should know by now that fivethirtyeight.com is the definitive place for polling information.

I came in here to say this. And hopefully, by saying that, I'm letting you all know that this is what I would have wanted to say but instead I had to say that.

 
kevinatilusa 2008-10-25 01:39:46 PM  
538 has exact looking numbers, but those numbers are also based on quite a few assumptions.

Most notable in the "pollster ratings" above is the assumption that a pollster has the same amount of methodological error in the general election as in the primaries, even though the demographics in those two elections are very different.

 
Atypical Person Reading Fark [TotalFark] 2008-10-25 01:45:26 PM  
Actually, I like electoral-vote.com very much as well.

Trends matter - and when one sees, for example, Montana and ND getting increasingly less Republican over a period of weeks near the election, until finally, the two parties are tied in nearly all polls, it's a pretty sure bet neither state will suddenly surge back all the way to 60% McCain. Once ND got less red, then Montana did as well - because there are regional forces at play.

So it didn't surprise anyone when SD got a little less red as well. The entire northern border of the U.S. (with the exception of Idaho, if you want to count that tiny bit of border) is either blue or close to neutral - and all trends show an increasing number of blue voters. Heating oil concerns? Pollsters don't know why their numbers behave as they do - but looking at the numbers over time is very predictive of voter behavior on election day.

This time, we have more polls - and more frequent polls, all trending the same way.

Only 16 states remain strongly red in the view of most pollsters.

 
dofus 2008-10-25 01:48:47 PM  
Florida releases this information every night. Includes names and party affiliation of those who have already voted.

Took about 15 minutes with a spreadsheet to determine that as of this morning, 10,833 Democrats, 8,001 Republicans, 3,060 'no party affiliation' (NPA) and 1,238 'other' voters have done their thing in Pasco County.

Pasco is about 6:5:3 Republican, Democrat and NPA. I _think_ the total is in the 150,000 range. (It's available, I'm just too lazy to look it up.)

 
Dalisca [TotalFark] 2008-10-25 01:51:09 PM  
A large portion of science is just educated guessing. More evidence makes the guesses more educated, but even science will tell you that theories are not 100%. There's a difference between a hypothesis, a theory, and a proof. Poll results and election predictions are theories, which leaves wiggle room for some variation. Probabilities trading will say that Obama has an 86% chance of victory where McCain has a 14% chance of victory. Though the Obama victory is more probable, no pollsters or statisticists will guarantee that victory. McCain still has a chance, though it is pretty marginal at this point.

Polling is just a method to try and predict the future. As time passes and more variables are added to each equation and those guesses become more and more educated, the theories will become more and more accurate.

Still just theories, though.

 
HitInTheJunk 2008-10-25 01:52:14 PM  
I've been saying for years that all these pollers are in huge liberal citys. Of COURSE they're going to make 'changes' to their numbers. I bet if you look at the raw numbers without the filter of the lib MSM media, McCain/Palin would be up by 10 points, which is much more accerate.

 
deltabourne 2008-10-25 01:52:24 PM  
roncofooddehydrator: Come on now people, all farkers should know by now that fivethirtyeight.com is the definitive place for polling information. place that assauges my confirmation bias the most

FTFY

 
Dalisca [TotalFark] 2008-10-25 01:53:10 PM  
HitInTheJunk: I've been saying for years that all these pollers are in huge liberal citys. Of COURSE they're going to make 'changes' to their numbers. I bet if you look at the raw numbers without the filter of the lib MSM media, McCain/Palin would be up by 10 points, which is much more accerate.

Wow, that must be some good crack you're smoking.

 
dofus 2008-10-25 01:53:20 PM  
re: my own post

Looks like Democrats are twice as like to vote early than the Republicans and NPAs.

(I think there is a party named 'Independent Party' in Florida so they use the NPA thing to designate people who don't specify a party aka 'independents' pretty much everywhere else.)

 
deltabourne 2008-10-25 01:54:06 PM  
Dalisca: A large portion of science is just educated guessing. More evidence makes the guesses more educated, but even science will tell you that theories are not 100%. There's a difference between a hypothesis, a theory, and a proof. Poll results and election predictions are theories, which leaves wiggle room for some variation. Probabilities trading will say that Obama has an 86% chance of victory where McCain has a 14% chance of victory. Though the Obama victory is more probable, no pollsters or statisticists will guarantee that victory. McCain still has a chance, though it is pretty marginal at this point.

Polling is just a method to try and predict the future. As time passes and more variables are added to each equation and those guesses become more and more educated, the theories will become more and more accurate.

Still just theories, though.


Stop using the word theory when you aren't sure of its meaning.

 
Dalisca [TotalFark] 2008-10-25 01:57:39 PM  
deltabourne: Dalisca: A large portion of science is just educated guessing. More evidence makes the guesses more educated, but even science will tell you that theories are not 100%. There's a difference between a hypothesis, a theory, and a proof. Poll results and election predictions are theories, which leaves wiggle room for some variation. Probabilities trading will say that Obama has an 86% chance of victory where McCain has a 14% chance of victory. Though the Obama victory is more probable, no pollsters or statisticists will guarantee that victory. McCain still has a chance, though it is pretty marginal at this point.

Polling is just a method to try and predict the future. As time passes and more variables are added to each equation and those guesses become more and more educated, the theories will become more and more accurate.

Still just theories, though.

Stop using the word theory when you aren't sure of its meaning.


the⋅o⋅ry

-noun, plural -ries. 1. a coherent group of general propositions used as principles of explanation for a class of phenomena: Einstein's theory of relativity.
2. a proposed explanation whose status is still conjectural, in contrast to well-established propositions that are regarded as reporting matters of actual fact.
3. Mathematics. a body of principles, theorems, or the like, belonging to one subject: number theory.
4. the branch of a science or art that deals with its principles or methods, as distinguished from its practice: music theory.
5. a particular conception or view of something to be done or of the method of doing it; a system of rules or principles.
6. contemplation or speculation.
7. guess or conjecture.

_________________________________________

Maybe you should check your dictionary before questioning definitions?

 
Biological Ali 2008-10-25 01:58:18 PM  
deltabourne: roncofooddehydrator: Come on now people, all farkers should know by now that fivethirtyeight.com is the definitive place for polling information. place that assauges my confirmation bias the most

FTFY


Does any electoral vote projector show McCain ahead right now?

 
Dalisca [TotalFark] 2008-10-25 01:59:43 PM  
Biological Ali: deltabourne: roncofooddehydrator: Come on now people, all farkers should know by now that fivethirtyeight.com is the definitive place for polling information. place that assauges my confirmation bias the most

FTFY

Does any electoral vote projector show McCain ahead right now?


Not a single one. Even FOX News has McCain down by nine.

 
Atypical Person Reading Fark [TotalFark] 2008-10-25 02:04:15 PM  
Dalisca: A large portion of science is just educated guessing. More evidence makes the guesses more educated, but even science will tell you that theories are not 100%. There's a difference between a hypothesis, a theory, and a proof. Poll results and election predictions are theories, which leaves wiggle room for some variation. Probabilities trading will say that Obama has an 86% chance of victory where McCain has a 14% chance of victory. Though the Obama victory is more probable, no pollsters or statisticists will guarantee that victory. McCain still has a chance, though it is pretty marginal at this point.

Polling is just a method to try and predict the future. As time passes and more variables are added to each equation and those guesses become more and more educated, the theories will become more and more accurate.

Still just theories, though.


"Just" theories? Well then, so is the balloting process itself, which is also subject to the whims of human error. Lost ballots, hanging chads, poorly regulated computerized voting - etc. There is no human process that is 100% pure.

In most of science, one can control for variables (polling isn't a science, btw, it's a statistical method). Therefore, what is "just a theory" in science is usually the best guess we humans can come up with.

Example: theory of microorganisms causing human illnesses. Got a better one? "Just" a theory - that's saved millions of lives.

Whether it's a good thing to save so many human lives - well, there's where you'd need a poll (if you think that's how we decide such things) or a system of ethics (too hard for many humans to operate those, though).

Those are just values though - no big deal.

 
Dalisca [TotalFark] 2008-10-25 02:33:44 PM  
Atypical Person Reading Fark:
"Just" theories? Well then, so is the balloting process itself, which is also subject to the whims of human error. Lost ballots, hanging chads, poorly regulated computerized voting - etc. There is no human process that is 100% pure.

In most of science, one can control for variables (polling isn't a science, btw, it's a statistical method). Therefore, what is "just a theory" in science is usually the best guess we humans can come up with.

Example: theory of microorganisms causing human illnesses. Got a better one? "Just" a theory - that's saved millions of lives.

Whether it's a good thing to save so many human lives - well, there's where you'd need a poll (if you think that's how we decide such things) or a system of ethics (too hard for many humans to operate those, though).

Those are just values though - no big deal.


I think we're kinda' trying to say the same thing here.

I'm not saying that theories aren't good enough to rely on pretty dependably, or be used for the greater good. But, they're still not proofs. The great thing about theories is that they can be modified over time to increase their own accuracy. This is why the details in the theory of evolution have been changed so many times, each time a new discovery is made. Yay progress.

Theories themselves are open to their own evolutionary process. Even Einstein's Relativity exists now in a different form than when it was originally concieved.

That's why we have different and better medications now to treat those illnesses caused by microorganisms than we did a hundred years ago.

As each election comes and goes, more accurate comparisons will be manufactured to increase the accuracy of polling data vs. electoral results in order to better predict those electoral results. These polls are probably more accurate than those from four years ago, and polls will probably be more accurate in the next electoral cycle.

We may still see some surprises on election night, but I don't think any of them are going to make enough of an impact to change the predicted outcome of this election.

 
alostpacket [TotalFark] 2008-10-25 02:36:17 PM  
kevinatilusa: Most notable in the "pollster ratings" above is the assumption that a pollster has the same amount of methodological error in the general election as in the primaries, even though the demographics in those two elections are very different.

I don't think that's the only criteria he uses on pollster ratings. I'm pretty sure he has some type of evaluation based on their deviance from the average of all the polls. As well as some subjective stuff but I'm not 102.5% sure.

 
Dalisca [TotalFark] 2008-10-25 02:52:08 PM  
Atypical Person Reading Fark

And on a side note, after the past two elections, I won't be ready to put on my party hat and do my victory dance until the votes are counted and there is an officially declared winner. I really want to, especially from the polling data, but odder things have happened than polling data being an inaccurate indicator of outcome.

Here's hoping the champagne tastes extra sweet on the fourth.

/Cheers

 
unyon [TotalFark] 2008-10-25 02:54:46 PM  
Do the polls factor in the votes that will be lost to disenfranchisement, diebold, supression, and election fraud?

 
Dalisca [TotalFark] 2008-10-25 02:58:07 PM  
unyon: Do the polls factor in the votes that will be lost to disenfranchisement, diebold, supression, and election fraud?

No, but I don't even want to think about those things yet. Scary stuff!

 
Velkrin 2008-10-25 03:05:27 PM  
[Sir Humphrey demonstrates how public surveys can reach opposite conclusions]
Sir Humphrey Appleby: Mr. Woolley, are you worried about the rise in crime among teenagers?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Do you think there is lack of discipline and vigorous training in our Comprehensive Schools?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Do you think young people welcome some structure and leadership in their lives?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Do they respond to a challenge?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Might you be in favour of reintroducing National Service?
Bernard Woolley: Er, I might be.
Sir Humphrey Appleby: Yes or no?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Of course, after all you've said you can't say no to that. On the other hand, the surveys can reach opposite conclusions.
[survey two]
Sir Humphrey Appleby: Mr. Woolley, are you worried about the danger of war?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Are you unhappy about the growth of armaments?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Do you think there's a danger in giving young people guns and teaching them how to kill?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Do you think it's wrong to force people to take arms against their will?
Bernard Woolley: Yes.
Sir Humphrey Appleby: Would you oppose the reintroduction of conscription?
Bernard Woolley: Yes.
[does a double-take]
Sir Humphrey Appleby: There you are, Bernard. The perfectly balanced sample.

 
Atypical Person Reading Fark [TotalFark] 2008-10-25 03:10:49 PM  
Dalisca: Atypical Person Reading Fark:
"Just" theories? Well then, so is the balloting process itself, which is also subject to the whims of human error. Lost ballots, hanging chads, poorly regulated computerized voting - etc. There is no human process that is 100% pure.

In most of science, one can control for variables (polling isn't a science, btw, it's a statistical method). Therefore, what is "just a theory" in science is usually the best guess we humans can come up with.

Example: theory of microorganisms causing human illnesses. Got a better one? "Just" a theory - that's saved millions of lives.

Whether it's a good thing to save so many human lives - well, there's where you'd need a poll (if you think that's how we decide such things) or a system of ethics (too hard for many humans to operate those, though).

Those are just values though - no big deal.

I think we're kinda' trying to say the same thing here.

I'm not saying that theories aren't good enough to rely on pretty dependably, or be used for the greater good. But, they're still not proofs. The great thing about theories is that they can be modified over time to increase their own accuracy. This is why the details in the theory of evolution have been changed so many times, each time a new discovery is made. Yay progress.

Theories themselves are open to their own evolutionary process. Even Einstein's Relativity exists now in a different form than when it was originally concieved.

That's why we have different and better medications now to treat those illnesses caused by microorganisms than we did a hundred years ago.

As each election comes and goes, more accurate comparisons will be manufactured to increase the accuracy of polling data vs. electoral results in order to better predict those electoral results. These polls are probably more accurate than those from four years ago, and polls will probably be more accurate in the next electoral cycle.

We may still see some surprises on election night, but I don't think any of them are going to make enough of an impact to change the predicted outcome of this election.


Good points. I too would be very surprised (more than surprised really) to see Obama get less than 330 or so electoral votes.

In other words - a landslide.

 
Casanova.Frankensteir 2008-10-25 03:16:53 PM  
HitInTheJunk: I've been saying for years that all these pollers are in huge liberal citys. Of COURSE they're going to make 'changes' to their numbers. I bet if you look at the raw numbers without the filter of the lib MSM media, McCain/Palin would be up by 10 points, which is much more accerate.

Um, a large part of the American population lives in those big liberal cities as well. Don't be fooled by the size of the states themselves - there's a lot of non-voting farmland in the Midwest, for example.

Another example - there are more people living in Fort Worth, Texas than in the entire state of Alaska. It only took 115,000 Alaskans to vote for Sarah Palin as governor, and that was a landslide.

In most states, the big cities pretty much determine the direction of the overall vote. If the Obama trends continue the way they have been, there are some who say that Houston and Dallas and the other big Texas cities might even be able to turn Texas blue this year. A long shot, but not out of the question.

 
King Something [TotalFark] 2008-10-25 03:19:45 PM  
I've got a few links to polling websites in my profile- E-V, 270towin.com, 538 and even Karl Rove have Obama with a huge lead.

It's gonna be a landslide.

 
Dalisca [TotalFark] 2008-10-25 03:29:57 PM  
Atypical Person Reading Fark:

Good points. I too would be very surprised (more than surprised really) to see Obama get less than 330 or so electoral votes.

In other words - a landslide.


Thanks! A landslide would be niiiiice. As long as it's over the magical 270 number, I'm still going to be the person standing on the couch with a lampshade for a hat.

 
guilt by association 2008-10-25 03:31:20 PM  
Anything to drive a horserace narrative.

 
Fast Thick Pants 2008-10-25 03:32:30 PM  
Biological Ali: Does any electoral vote projector show McCain ahead right now?

One out of 87 listed here: 3bluedudes.com/ProjectDatabase.htm

But it looks like a bit of joke.

 
FootInMouthDisease 2008-10-25 04:09:22 PM  
Dalisca: Biological Ali: deltabourne: roncofooddehydrator: Come on now people, all farkers should know by now that fivethirtyeight.com is the definitive place for polling information. place that assauges my confirmation bias the most

FTFY

Does any electoral vote projector show McCain ahead right now?

Not a single one. Even FOX News has McCain down by nine.


Where'd you go deltabourne??

guys i think we hurt his feelings with these horrid "facts" and the terrifying prospects of "reality"

 
bronyaur1 [TotalFark] 2008-10-25 04:27:32 PM  
HitInTheJunk: I've been saying for years that all these pollers are in huge liberal citys. Of COURSE they're going to make 'changes' to their numbers. I bet if you look at the raw numbers without the filter of the lib MSM media, McCain/Palin would be up by 10 points, which is much more accerate.

I wonder if those liberal polsters in huge citys would be more accerate if they learned how to spell.

Ladies and Gentlemen, I give you the 2008 GOP's base.

 
Dalisca [TotalFark] 2008-10-25 04:44:22 PM  
bronyaur1: HitInTheJunk: I've been saying for years that all these pollers are in huge liberal citys. Of COURSE they're going to make 'changes' to their numbers. I bet if you look at the raw numbers without the filter of the lib MSM media, McCain/Palin would be up by 10 points, which is much more accerate.

I wonder if those liberal polsters in huge citys would be more accerate if they learned how to spell.

Ladies and Gentlemen, I give you the 2008 GOP's base.


Maybe he means "acerate", as in, needle-shaped, or is talking about aceric acid?

 
bronyaur1 [TotalFark] 2008-10-25 05:34:13 PM  
Dalisca: bronyaur1: HitInTheJunk: I've been saying for years that all these pollers are in huge liberal citys. Of COURSE they're going to make 'changes' to their numbers. I bet if you look at the raw numbers without the filter of the lib MSM media, McCain/Palin would be up by 10 points, which is much more accerate.

I wonder if those liberal polsters in huge citys would be more accerate if they learned how to spell.

Ladies and Gentlemen, I give you the 2008 GOP's base.

Maybe he means "acerate", as in, needle-shaped, or is talking about aceric acid?


That must be it :)

 
Gyrfalcon [TotalFark] 2008-10-25 10:56:07 PM  
Easy answer: Lies, damn lies, and statistics.

 
xria 2008-10-26 12:05:30 AM  
Of course the trends are usually more important/meaningful than exact numbers anyway - so as long as they adjustments they make are consistent each time they take the poll, then the values they return can help give an understanding of what is going on. There is always going to be variation between polls and actual votes due to a number of continuously changing systematic errors, and the weighting or adjustment made that would have corrected the polls 4 years ago will be an incorrect bias this time around as demographics, culture, economics and other changes move the goalposts.

 
giarcgood [TotalFark] 2008-10-26 07:15:12 AM  
Dalisca: Polling is just a method to try and predict the future. As time passes and more variables are added to each equation and those guesses become more and more educated, the theories will become more and more accurate.

I put a PhD scholarship proposal in about 17 years ago to try and estimate this stuff. Had all sorts of good thoughts based on above and more.

Knocked back and work in a bank now on very exciting things.

 
Displayed 42 of 42 comments


[Continue Farking]