The Presidential Prediction Poll

[NorthSideSox72]

Not fear at all. It's the same phenomenon that had people telling surveys they are watching Masterpiece Theater when they are watching Dukes of Hazard. Imagine polling people on their masturbation habits. In the words of House, "people lie". In this case, perhaps not out of fear but out of wanting the pollster's approval.

Who on earth wants the approval of a random phone pollster? If anything, people are liable to lie to piss them off, more than the other way around.

 

[GoSox05]

Nader in a close one.

[BigSqwert]

Not fear at all. It's the same phenomenon that had people telling surveys they are watching Masterpiece Theater when they are watching Dukes of Hazard. Imagine polling people on their masturbation habits. In the words of House, "people lie". In this case, perhaps not out of fear but out of wanting the pollster's approval.

 

So after hundreds, if not thousands, of polls over the past few months you're saying that time and time again people keep lying. And enough of them keep doing so that it's actually McCain who's leading in most of these polls? I find that a little hard to believe.

[Texsox]

So after hundreds, if not thousands, of polls over the past few months you're saying that time and time again people keep lying. And enough of them keep doing so that it's actually McCain who's leading in most of these polls? I find that a little hard to believe.

 

No, I am saying that there are reasons why legitimate polls post margins of error, it is factored in that people are not always truthful in their responses. Further, there is some error based on sample populations. My hunch is we will find that more McCain supporters lied and said Obama than Obama supporters lying and saying McCain. Pollsters are hoping that people will lie both ways and it will balance, and it does in most polls. Looking at the sample sizes on many of those polls, it would take dozens of people to sway those polls, not tens of thousands. I also believe there will be a late push towards McCain.

 

I do think Obama would have won if the election was a couple weeks ago.

[Texsox]

Who on earth wants the approval of a random phone pollster? If anything, people are liable to lie to piss them off, more than the other way around.

 

OK. So are they lying and saying McCain or Obama to piss them off?

 

Bottom line, there are margins of error in all polls. I believe these margins are actually in Obama's favor and we'll see McCain on election day with better than poll numbers for the reasons I mentioned.

 

Do you believe the polls will be 100% accurate? Or favor one candidate or the other?

[BigSqwert]

There could easily be a number of people in lifelong Republican families lying to pollsters saying that they're voting McCain because their spouse is in the room during the call. They can just as easily vote Obama when in the privacy of the voting booth without their die-hard Republican spouse ever knowing.

[BigSqwert]

The reason I cite an example like that is because that's how my parents are. My mom secretly told me she's voting Obama but would never let my father know. He'd be furious.

[Texsox]

There could easily be a number of people in lifelong Republican families lying to pollsters saying that they're voting McCain because their spouse is in the room during the call. They can just as easily vote Obama when in the privacy of the voting booth without their die-hard Republican spouse ever knowing.

 

Exactly. And that is a bedrock principle of polling, that those errors are equal on both sides, therefor having accurate results.

[NorthSideSox72]

OK. So are they lying and saying McCain or Obama to piss them off?

 

Bottom line, there are margins of error in all polls. I believe these margins are actually in Obama's favor and we'll see McCain on election day with better than poll numbers for the reasons I mentioned.

 

Do you believe the polls will be 100% accurate? Or favor one candidate or the other?

Margin of error is a statistically derived value about the ratio of sample size to total pool. Its got nothing to do with perceived dishonesty or anything subjective at all.

 

[Texsox]

Margin of error is a statistically derived value about the ratio of sample size to total pool. Its got nothing to do with perceived dishonesty or anything subjective at all.

 

But we know that people are not always 100% in their responses. You mentioned they could be lying to piss off the pollsters.

 

Are you thinking the polls are accurate or biased towards one candidate?..And instead of margin of error, I will say that throughout history the polls have never been 100% accurate. Why should they be now? Use a different term to describe when results are di9fferent than predicted and we'll use that.

[NorthSideSox72]

But we know that people are not always 100% in their responses. You mentioned they could be lying to piss off the pollsters.

 

Are you thinking the polls are accurate or biased towards one candidate?

I think they are probably a bit biased towards Obama, but it will vary by state. Lots of factors involved. But I think the polls are probably pretty close, just a few points off. There are so many unpolled factors that offset each other.

 

[Texsox]

I think they are probably a bit biased towards Obama, but it will vary by state. Lots of factors involved. But I think the polls are probably pretty close, just a few points off. There are so many unpolled factors that offset each other.

 

Which is what I've been trying to say. And it seems close enough in a few instances that it could tilt the election to McCain. I'm also expecting a late rally towards McCain as the undecided play it safe with an establishment candidate who has been around for decades.

[HuskyCaucasian]

Damn, they have it 375-157 Obama.

RCP has it 375-163. 306-157 if you dont throw in "toss-up" states.

[lostfan]

Which is what I've been trying to say. And it seems close enough in a few instances that it could tilt the election to McCain. I'm also expecting a late rally towards McCain as the undecided play it safe with an establishment candidate who has been around for decades.

All Obama has to really do is win Ohio or FL and right now he is leading in both. The rest of those states are mostly just nails in the coffin that increase the odds of Obama winning. The late rally would have to be HUGE and span all across the country.

 

A McCain victory isn't impossible by any means but at this point it appears really, really unlikely. All of the polling trends are breaking Obama right now and McCain is running out of time.

[Disco72]

Margin of error does not always factor in all possible biases. The bias that TexSox is talking about is called Social Desirability. You'd be surprised what people "lie" about even in confidential surveys. I do research on businesses, particularly internationalization. People will lie on confidential surveys (not tied to their name or company) about their international experience, sales numbers, or any number of things that do not really matter that much. If people will "lie" about those issues, it is perfectly reasonable that there is a social desirability bias in the polls based on a much stronger issue.

[BigSqwert]

Margin of error does not always factor in all possible biases. The bias that TexSox is talking about is called Social Desirability. You'd be surprised what people "lie" about even in confidential surveys. I do research on businesses, particularly internationalization. People will lie on confidential surveys (not tied to their name or company) about their international experience, sales numbers, or any number of things that do not really matter that much. If people will "lie" about those issues, it is perfectly reasonable that there is a social desirability bias in the polls based on a much stronger issue.

 

There really wasn't much evidence of this going in the primaries so I'm not sure it will happen on the 4th.

[Rex Kickass]

All Obama has to really do is win Ohio or FL and right now he is leading in both. The rest of those states are mostly just nails in the coffin that increase the odds of Obama winning. The late rally would have to be HUGE and span all across the country.

 

A McCain victory isn't impossible by any means but at this point it appears really, really unlikely. All of the polling trends are breaking Obama right now and McCain is running out of time.

 

Actually all Obama has to do is win in PA and NH (the only two states in any perceived danger that were Kerry states, and is currently comfortably ahead of McCain), adding in Iowa and New Mexico (which appear to be comfortably for Obama) and pick up Virginia (which also seems more likely) and Obama has 278 electoral votes.

[NorthSideSox72]

There really wasn't much evidence of this going in the primaries so I'm not sure it will happen on the 4th.

Different candidates, different deal. Obama's main alternative in the primaries was a woman, who might experiences many of the same biases that Obama does. Not the case here.

 

[BigSqwert]

Different candidates, different deal. Obama's main alternative in the primaries was a woman, who might experiences many of the same biases that Obama does. Not the case here.

 

I would think race is a bigger issue then sex. How many people do you run across in your lifetime that despise women?

[Disco72]

There really wasn't much evidence of this going in the primaries so I'm not sure it will happen on the 4th.

 

My own personal opinion is that Obama's poll leads are big enough to account for any social desirability effect, but it will still be interesting to see what happens on election day. My prediction is a decent lead in the popular vote (over 50 but under 55%) but a pretty safe win in the electoral college.

 

However, regarding the primaries vs. the general election on social desirability, there are some important reasons why it might not have been evident in the primaries but possibly evident in the general. The Democratic primaries had several important candidates that were minorities or women, whereas the general is obviously a much starker contrast between candidates and a much broader set of the population.

 

(Edit: NSS beat me to it and said it in fewer words too!)

Edited October 24, 2008 by Disco72

[lostfan]

Actually all Obama has to do is win in PA and NH (the only two states in any perceived danger that were Kerry states, and is currently comfortably ahead of McCain), adding in Iowa and New Mexico (which appear to be comfortably for Obama) and pick up Virginia (which also seems more likely) and Obama has 278 electoral votes.

Either way, he has more than one path to victory. Among his "safe" or at least "almost safe" states, he has a built-in electoral advantage. Meaning he could lose all of the toss-up or slightly Dem-leaning states (unlikely) and still win.

[knightni]

non-internet using? Most polls are conducted by phone.

I mean that, the voters who come out aren't the left-leaning people who, frequent message boards, create anti-Bush/McCain sites, etc.

[BigSqwert]

I mean that, the voters who come out aren't the left-leaning people who, frequent message boards, create anti-Bush/McCain sites, etc.

 

Early voting results suggest that democrats are coming out in larger numbers.

[knightni]

Early voting results suggest that democrats are coming out in larger numbers.

You're missing my original comment.

 

People who come out election day and right-lean, are people who are less likely to use the internet or answer polls (conservative church-goers, retirees, military personnel, etc.)

[BigSqwert]

You're missing my original comment.

 

People who come out election day and right-lean, are people who are less likely to use the internet or answer polls (conservative church-goers, retirees, military personnel, etc.)

 

Why would conservatives not answer polls? What evidence do you have that this is true?