Poll Vault

[MattusMaximus]

CNN is reporting Palin is stumping in Iowa. They've never led in Iowa. Why is Palin there?

I'm going to guess desperation with a healthy dose of cluelessness.

 

[gdnp]

I'm going to guess desperation with a healthy dose of cluelessness.

You know how those Alaskans are starved for glamor and culture. How could she resist a stop in Iowa?

 

[Syameese]

Colley rankings?

Just wondering what any of the regulars here think of the Colley ranking polls on the election. I don't have enough posts to allow me to post the link, but those familiar with College football will surely know who I am talking about.

I heard about them earlier in the year on the Science talk podcast (about June or July) and have been watching them since. Now I am no expert in statistics, but their use of the median results of polls did intrigue me when I listened to the podcast and I was interested in what the more qualified poll watchers on here think of their methodology.

 

[Undesired Walrus]

Obama up by 8 on Rasmussen.

 

[BenBurch]

Last night CBS radio was busy pushing a Zogby result that has the race narrowing to 5 points... Why do they focus on single polls, and this one with an MOE of +/- 3, as though they were so indicative?

 

[Peephole]

Last night CBS radio was busy pushing a Zogby result that has the race narrowing to 5 points... Why do they focus on single polls, and this one with an MOE of +/- 3, as though they were so indicative?

The Republicans are trying to keep up morale and the media doesn't want to call this for Obama yet, so they keep pushing worthless polls like Zogby

 

[gdnp]

Last night CBS radio was busy pushing a Zogby result that has the race narrowing to 5 points... Why do they focus on single polls, and this one with an MOE of +/- 3, as though they were so indicative?

For the same reason that the sportscasters tell you what an exciting game it is and how anything can happen when the score is 3-1 and time is running out. Because the closer it is the more people pay attention. They want people glued to their TV sets on election night, not tuning out as soon as Virginia, Florida, and North Carolina close their polls.

 

[MattusMaximus]

Last night CBS radio was busy pushing a Zogby result that has the race narrowing to 5 points... Why do they focus on single polls, and this one with an MOE of +/- 3, as though they were so indicative?

Two words: Republican desperation

+ two more words: Media hype

'Nuff said.

 

[MattusMaximus]

The latest on battleground state prediction markets from the Yahoo Political Dashboard (YPD). Note these numbers are winner-take-all predictions.

Nevada
Obama 77.0
McCain 26.0

Colorado
Obama 85.0
McCain 16.4

Missouri
Obama 64.0
McCain 38.5

Indiana
Obama 59.0
McCain 41.0

Ohio
Obama 80.0
McCain 20.5

Virginia
Obama 79.5
McCain 20.0

North Carolina
Obama 69.3
McCain 32.0

Florida
Obama 70.0
McCain 31.5

Predicted outcome of the Electoral College
Obama 375
McCain 163

 

[Malerin]

538 has moved Mccain back to 3.3%. A day or two ago, Mccain had clawed his way up to over 4%.

 

[Gnu World Order]

538 has moved Mccain back to 3.3%. A day or two ago, Mccain had clawed his way up to over 4%.

Whew. I was getting worried there.

 

[Dr Adequate]

OK, McCain's up by 2 in Arizona.

 

[BenBurch]

OK, McCain's up by 2 in Arizona.

If you will recall, I predicted he would lose AZ... (He still might not, but it will be a squeaker there.)

 

[David Wong]

Meanwhile, IBD/Tipp, the most accurate pollsters in 2004, have Obama down to a 2.8 point lead, within the margin of error:

http://www.ibdeditorials.com/series13.aspx

Expect the Drudge headline soon. Their difference seems to be they have McCain running pretty close in the young voters, either their likely voter model excludes them or they're under-sampling or who knows.

But who cares about likely voter models, when people are already voting! We can simply ask people who they voted for, and we'll have a hard number.

Thus, Zogby says Obama has a 24-point lead among early voters:

http://www.zogby.com/news/ReadNews.dbm?ID=1613

...while another pollster told 538 the lead is 5 points:

http://www.fivethirtyeight.com/2008/10/pollster-responds.html


19 point swing, on a question of fact. WHAT?!?!?!

This will drive me insane before Nov 4th gets here.

 

[davefoc]

If you will recall, I predicted he would lose AZ... (He still might not, but it will be a squeaker there.)

I want to go on record as having predicted that McCain/Palin will win in Alaska.

 

[Pookster]

Meanwhile, IBD/Tipp, the most accurate pollsters in 2004, ...

Not quite. They were one of the closest on the margin of victory, but not close at all on the final percentages.


http://poq.oxfordjournals.org/cgi/content/full/69/5/642/T1

					Bush	Kerry	Nader	Other	Bush 
									Lead

Election Result 			50.7 	48.3 	0.3 	0.7 	2.4 
			
Harris (Online) 	10/29–11/1 	47 	50 	1 	2 	–3.0 	
Battleground: Lake 	10/31–11/1 	48.6 	50.7 			–2.1 	
Fox News 		10/30–31 	46 	48 	1 	5 	–2.0 	
Marist College 		11/1 		49 	50 	0 	1 	–1.0 	
Democracy Corps 	10/29–31 	47 	48 	1 	3 	-1.0
Gallup 			10/31–11/1 	49 	49 	1 	1 	0.0 	
American Research Group	10/28–30 	48 	48 	1 	3 	0.0 	
Zogby 			11/2 		49.4 	49.1 	1 	0.5 	0.3 	
Harris (Telephone) 	10/29–11/1 	49 	48 	2 	1 	1.0 	
NBC News/WSJ	 	10/29–31 	48 	47 	1 	4 	1.0 	
ABC News 		10/28–31 	49 	48 	1 	3 	1.0 	
Washington Post 	10/28–31 	49 	48 	1 	3 	1.0 	
CBS News/N.Y. Times 	10/29–11/1 	49 	47 	1 	3 	2.0 	
CNN/USA Today/ Gallup 	10/29–31 	49 	47 		4 	2.0 	
Pew Research Center 	10/27–30 	51 	48 	1 		3.0 	
TIPP 			10/30–11/1 	48.6 	45.3 	0.9 	5.2 	3.3 	
Battleground: Tarrance 	10/31–11/1 	51.2 	47.8 	0.5 	0.5 	3.4 	
GW Battleground 	10/31–11/1 	50 	46 	1 	4 	4.0 	
Newsweek 		10/27–29 	50 	44 	1 	5 	6.0

 

[chipmunk stew]

After three straight days of maintaining a record 50.4% RCP average, Obama ends at a new record, 50.5%. It's his fifth straight day above 50%.

 

[BenBurch]

After three straight days of maintaining a record 50.4% RCP average, Obama ends at a new record, 50.5%. It's his fifth straight day above 50%.

Need 8 more...

 

[R.Mackey]

Meanwhile, IBD/Tipp, the most accurate pollsters in 2004, have Obama down to a 2.8 point lead, within the margin of error:

http://www.ibdeditorials.com/series13.aspx

I don't know if this will make you more or less anxious, but this poll looks a bit funny to me.

Consider: Obama polls with 47.0% percent of respondents, McCain with 44.2%. But how they each get to that total is different.

Let's start by looking at party split. Obama collected 87% of Democrats, 44% of Independents, and just 6% of Republicans. McCain collected 88% of Republicans, 39% of Independents, and 9% of Democrats.

If we work from these data, we cannot compute how many of the total sample are actually Democrats etc., but we can show that Republicans are slightly outnumbered by Democrats unless both populations are very small (specifically, R = 3% + 0.82 * D, where R + D + I = 1.00). One reasonable solution is for exactly half to be Democrats, then 44% are Republicans, and 6% are Independent. This is consistent with our expectations.

However, this poll shows McCain gathering more Democrats than Obama gaining from the Republican party. This is the only poll I can recall seeing that shows this result. If anything, trends seem to be going the other way.

Next, let us look at ideology, which shows a very different picture. Now Obama's total is made up of 94% of Liberals, 57% of Moderates, and 20% of Conservatives; this adds up to 47% of the total sample. McCain, a mere 2.8% behind, received 71% of Conservative votes, 33% of Moderates, and a paltry 2% of Liberals -- an apparent trouncing! How can this be?

In order for these numbers to make sense, the Conservatives must outweigh Liberals by a comfortable margin (in this case, C = 29% + 0.88 * L, C + L + M = 1.00). The solution where Liberals are closest to Conservatives in number has a sample made up of 20% Liberals, 33.4% Moderates, and a whopping 46.6% Conservatives.

This makeup does not appear to be consistent with either the party affiliation results, or with the expected population of voters anywhere north of the Mason-Dixon Line (except Utah). Either a large majority identifies as Moderate, and Liberals are in extremely short supply, or else the sample is highly biased towards Conservatives.

---
ETA: I should have kept reading. The starkest contrast is in household earnings. There, Obama received 47% of Upper/Upper Middle Class, 45% of Middle Class, 53% of Working Class, and 60% of Lower Class. McCain, 49% of Upper/Upper Middle, 46% of Middle, 39% of Working, and 34% of Lower Class.

This poll therefore only makes sense if nearly all replies are biased heavily against poorer households. The result follows naturally.
---

This does not mean that the data are wrong. The data are data, no more and no less. However, taken in the light of expected results and compared to other polls, it seems there is a structural anomaly present in the sample, one that if truly present would handily explain the seemingly anomalous result.

Only time will tell. I will note in passing that even this poll consistently predicts an Obama victory -- we are merely fussing over the margin, and this particular methodology does not seem well suited to estimate margin. But perhaps they know something I don't. Very difficult to judge from such few results.

Cliff's Notes: After some analysis, this poll looks to be a simple outlier, nothing more. We'll find out in a week or so.

 

[David Wong]

Gallup has it closer than that. Two points in their likely voter poll:

http://www.gallup.com/poll/111568/Gallup-Daily-Presidential-Race-Narrows-Slightly.aspx

49-47.