Poll Vault

[MattusMaximus]

One note regarding Gallup that I was mistaken about (not sure if I stated it here or not though), they don't weight their poll for party affiliation.

Ah, this is something which I've wondered about for some time. I now know that Rasmussen does weight for party affiliation (thanks Wangler - or was it Brainster - well, you know who you are )

I, too, have been cautious about some of the polls for this very reason. The question of methodology is very important, not only regarding party affiliation but also making phone calls to cellphone users. Many younger people don't have a land line and many polls don't call cellphones, so that cuts out a potentially large chunk in the polling.

What was it that someone said about damned lies & statistics?

 

[MattusMaximus]

Two days later than I predicted:

[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_977748c81d80cc62e.jpg[/qimg]

Like I said, I'm not sure what's driving this today. Psychological barrier pierced?

I saw that too. It's damned weird, because when I look at NewsFutures and the Iowa Electronic Markets I see something completely different...

NewsFutures
Obama 57%
McCain 43%

IEM - WTA
Obama 56.6%
McCain 44.3%

What the hell is going on here? Up until this week, all the prediction markets that I followed (including Rasmussen and Intrade) all pretty much had the same numbers (within error) and trends. But now we've got two major prediction markets with Obama below or even with McCain, while two other major markets are little changed from last week.

Can anyone here sort this out for me? I am terribly confused at what I'm seeing here. If I cannot figure this out, I don't know how to trust anything that I'm seeing, whether it's in the polls or the markets.

Urrrgggghhh...

 

[MattusMaximus]

Yep,

This prediction market has been very volitile (maybe they all have), and I wouldn't expect McCain to actually lead until a debate has been concluded (assuming he "bested" his opponent).

It seems the volatility that you've noted is limited only to Intrade and market outlets that get feeds from Intrade (such as Rasmussen). As I noted in my post above, the IEM and NewsFutures markets are pretty much stable as they were late last week, putting Obama ahead on WTA by 12-15 points.

I think that it would take some serious event for McCain to have a long-duration 10-20 point prediction market lead like Obama had for quite a while.

I agree, but I am beginning to wonder about Intrade. Do they have something different in the manner in which they run their market as compared to others such as IEM? I know that IEM has a cap of $500 per trader so that no one person can swing the market drastically - are there any such safeguards over at Intrade?

For that matter, are the numbers we're seeing at Intrade the WTA prediction? Or are they vote-share? I'd always thought they were WTA.

 

[Wangler]

All good questions, Maximus.

These polls and PM's can give you a headache!

 

[Pookster]

Two days later than I predicted:

[qimg]http://www.internationalskeptics.com/forums/imagehosting/thum_977748c81d80cc62e.jpg[/qimg]

Like I said, I'm not sure what's driving this today. Psychological barrier pierced?

Could some be buying McCain shares anticipating/gambling they would go up in price? Basically, are people lining up at the gas pump and filling up at the current price before the hurricane potentially disrupts the supply causing prices to rise?

 

[Pookster]

I, too, have been cautious about some of the polls for this very reason. The question of methodology is very important, not only regarding party affiliation but also making phone calls to cellphone users. Many younger people don't have a land line and many polls don't call cellphones, so that cuts out a potentially large chunk in the polling.

We all should use a critical eye when looking at any poll. Weighting for political affiliation carries it's own issues, just as not weighting. The same goes for the cell phone issue. Some account for it when they don't actually call cell phones. It's not an exact science. It's why there are margins of error and confidence levels released with each poll.

 

[Pookster]

State poll fun ...


Ohio: Tuesday, Fox/Rasmussen has McCain up by 7 (51%-44%). Today, Quinnipiac has Obama up by 5% (49%-44%)

Florida: Tuesday, Fox/Rasmussen has it at a tie (48%-48%). Today, Quinnipiac has McCain up by 7 (50%-43%)


You gotta love it.

 

[Pookster]

Rasmussen has the race tied again (48%-48%) ...

http://www.rasmussenreports.com/pub...ial_election/daily_presidential_tracking_poll

McCain is now viewed favorably by 56% of the nation’s voters while Obama earns positive reviews from 55% (see trends). Sarah Palin is viewed favorably by 56%, Joe Biden by 53%.

Yesterday, for the first time ever, Rasmussen Markets data gave McCain a better than 50% chance of winning the White House in November. These figures are updated on a 24/7 basis by market participants and currently give Obama a 49.0 % chance of victory. Prior to this past weekend, expectations for a Democratic victory had generally been in the 60% range.

 

[MattusMaximus]

State poll fun ...

Ohio: Tuesday, Fox/Rasmussen has McCain up by 7 (51%-44%). Today, Quinnipiac has Obama up by 5% (49%-44%)

Florida: Tuesday, Fox/Rasmussen has it at a tie (48%-48%). Today, Quinnipiac has McCain up by 7 (50%-43%)

You gotta love it.

Good gawd, we might as well just consult a damned Magic-Eight Ball

Well, until anyone can tell me what the hell is going on with Intrade, I'm going to follow the Iowa Prediction Market. They've been at the prediction market business longer than anyone & seem to be the gold standard. And they've also got safeguards in place to prevent the kind of overly-slanted trading to which Pookster is referring - i.e. by limiting each trader to $500 total they make sure that single individuals cannot dominate the market and thus cause crazy swings.

Today's IEM-WTA (as of now):

Obama 55.3
McCain 45.2

 

[Pookster]

Gallup's take on party affiliation changes since the conventions ...

http://www.gallup.com/poll/110215/GOP-Increase-Party-After-Convention-Unusual.aspx

The percentage of Americans identifying themselves as Republicans has increased from 26% immediately before last week's Republican National Convention began to 30% immediately after it. That increase, combined with a slight 2-point drop in Democratic identification from 37% to 35%, has reduced the Democrats' formidable advantage in national party identification from 11 points to 5.

Such short-term shifts in party affiliation are a regular occurrence after a party's convention -- Gallup has measured an increase in the percentage of Americans identifying with a party after its convention since 1992, averaging 4 percentage points.

Each party has enjoyed an increase in leaned party identification after its convention, with the exception of this year's Democratic convention. That may reflect the already-high percentage of Americans identifying with or leaning to the Democratic Party this year before its convention began.

But are those increases temporary, or do they have some staying power? In general, it appears that the increases are short-lived and fade as the enthusiasm from the convention subsides.

Since 1996, the percentage of Americans who identify with or lean to a party has dropped in the poll after that year's post-convention poll for each party. So to the extent history is a guide, Gallup's next survey would likely show a slight drop in the percentage of Americans identifying with or leaning to the Republican Party.

 

[Pookster]

Gallup has McCain up by 4 today (48%-44%) ...

http://www.gallup.com/poll/110227/Gallup-Daily-McCain-48-Obama-44.aspx

In addition, since Sept. 5 -- the first night after the Republican National Convention -- he has outpolled Obama in each of the last six individual night's polling. That consistent pattern in the night-to-night data suggests that McCain has a stable lead for now.

 

[Brainster]

Rasmussen finally shows the bounce. McCain up three percentage points both with and without leaners. Interesting comment from them about the moving average:

It is unusual to find a three-point jump in one day on the tracking poll. Daily tracking results are collected via nightly telephone surveys and reported on a three-day rolling average basis. Today’s gain for McCain comes partly from a good night of polling last night and partly from the fact that a good night for Obama on Monday is no longer part of the sample.

 

[Foolmewunz]

The states that are turning into batlegrounds this week should really worry the Democrats. Michigan? Pennsylvania?

This is not good.

The RCP average this morning (evening over here) is awash with red ink.

 

[David Wong]

Has there ever been an election where the same guy just lead pole to pole, from nomination to election day? I can't remember one. Kerry had leads, Gore had leads, Dukakis had leads, even Mondale.

The only one I can think of where one guy never had a sustained lead was Dole in 1996, you had weird outlier polls putting him ahead but otherwise it was all Clinton, and that's the only election of my life that went that way.

Even those sure McCain would lose, should have expected him to lead at some stretch of the race. The wake of the RNC and VP pick is right where you'd expect it.

This will be a close race; there's a really good chance that on election day, the leader won't even be outside the margin of error. We could easily go in with McCain up by 2 or Obama up by 2 and basically watching to see which party gets the best turnout.

 

[Jimbo07]

This will be a close race; there's a really good chance that on election day, the leader won't even be outside the margin of error. We could easily go in with McCain up by 2 or Obama up by 2 and basically watching to see which party gets the best turnout.

Dateline, Nov. 2008 - In an entirely unprecedented move, the election is decided by the courts...

 

[David Wong]

McCain's bounce continues to erode on Gallup, down to 3 now:

http://www.gallup.com/poll/110266/Gallup-Daily-McCain-48-Obama-45.aspx

Obama was up by 2 a week after his convention, McCain finds himself up by 3 a week after his. Seems to mirror pretty closely then.

 

[chipmunk stew]

Has there ever been an election where the same guy just lead pole to pole, from nomination to election day? I can't remember one. Kerry had leads, Gore had leads, Dukakis had leads, even Mondale.

The only one I can think of where one guy never had a sustained lead was Dole in 1996, you had weird outlier polls putting him ahead but otherwise it was all Clinton, and that's the only election of my life that went that way.

Even those sure McCain would lose, should have expected him to lead at some stretch of the race. The wake of the RNC and VP pick is right where you'd expect it.

This will be a close race; there's a really good chance that on election day, the leader won't even be outside the margin of error. We could easily go in with McCain up by 2 or Obama up by 2 and basically watching to see which party gets the best turnout.

While Kerry may have been up on some polls, they were clearly outliers. Check out this graph:
http://www.realclearpolitics.com/Presidential_04/chart3way.html

Kerry never had a chance.

Compare that to this year, and Obama's looking pretty damn good.

 

[David Wong]

While Kerry may have been up on some polls, they were clearly outliers. Check out this graph:
http://www.realclearpolitics.com/Presidential_04/chart3way.html

Kerry never had a chance.

Compare that to this year, and Obama's looking pretty damn good.

Look at 2004 again. Kerry leads routinely, including through July and August... until the calendar turns September. Then Bush takes the lead and never lets it go.

Now look at this year. Obama lead through summer, but ever since the calendar has flipped to September...

 

[gdnp]

Compare that to this year, and Obama's looking pretty damn good.

I agree. The bounce hasn't finished bouncing, and most of the spreads are within the margins of error. The economy is unlikely to improve before the election. Iraq is going well enough it would be hard for it to improve more, thus the surge is unlikely to give McCain more bounce, and legitimizing Obama's call for withdrawing troops. If things turn sour, it is hard to imagine the population would support another escalation. I expect Obama to at least hold his own in the debates. So I see a lot of ways the scales could tip back to Obama and not so many that would tip them towards McCain.

Palin remains a wildcard. I don't expect her to win over too many more converts, the question is whether her celebrity will fade before or after the election. The potential for a major gaffe or scandal remains real. I still think that Palin is poorly qualified to be vice president. Then again, I also think reality TV is incredibly stupid, and SOMEONE has been watching those shows.

All in all, I'd be happier if Obama was comfortably ahead, but I'm not panicking yet.

 

[gdnp]

Look at 2004 again. Kerry leads routinely, including through July and August... until the calendar turns September. Then Bush takes the lead and never lets it go.

Now look at this year. Obama lead through summer, but ever since the calendar has flipped to September...

The difference is that Bush opened up a 6-7 point lead through September that faded in October, and ended up winning by 1.5%. McCain's lead now seems to be closer to 3%, so a similar 4-5 point swing between now and election day would put Obama back ahead.