McCain not as forked as the McCain Forked thread, but still forked

[Defensive_Indifference]

@Marley23

This is nobody’s fault, but it’s one of those things that shows you how snakebit the McCain campaign is these days:

On the ESPN ticker a few minutes ago, I saw that Sarah Palin dropped the puck at the St. Louis Blues-L.A. Kings game. She walked onto the ice on a red carpet, and Blues goalie Manny Legace stepped on the carpet and fell. Strained hip flexor, gave up two goals in the first period and then left the game. The Blues lost 4-0.

Ah, shit.

And she didn’t even get booed as lustily as she did in Philadelphia. She got a mix of boos and cheers, apparently.

[Frank]

@Bayard

She got a mix of boos and cheers, apparently.

Yeah, I couldn’t tell any preponderance on TV, and neither was really noticeable. Given that this is the Midwest, there was polite applause as well.

[DSeid]

A GQish question: I’ve been noticing that Rasmussen is lately going higher +Obama than even the “expanded LV” Gallup tracker…

Does Rasmussen modify their LV screen methodology on the fly? Are their results reflective of a real change in the numbers, a statistical random walk, or a change in their sampling methodolgy either as the election gets closer or as they consider the LV question more throughly as they go along?

[Lakai]

Bad news for McCain in Ohio as Obama shot up to +6 at RCP.

FiveThirtyEight has him at 2.9.

[Marley23]

That’s not what I’m seeing over there (5.1%). Does the site just not like me anymore?

[RTFirefly]

@DSeid

A GQish question: I’ve been noticing that Rasmussen is lately going higher +Obama than even the “expanded LV” Gallup tracker…

Are you sure you’ve been comparing properly? Starting today and working backward, here are the Obama leads I see on the respective websites:

Gallup Exp: 8,7,6,8,10,9,7,4,6,6,8,10,10,6,9,9,9
Rasmussen:  8,7,7,6, 4,4,6,5,4,4,5, 5, 5,6,7,5,5 

For Gallup, I’m going by the middle graph here, and Rasmussen numbers are here.

[RTFirefly]

@Marley23

That’s not what I’m seeing over there (5.1%). Does the site just not like me anymore?

It must not. RCP’s Ohio page is showing a 6.1% Obama lead, at least when I just opened it right now.

[Marley23]

That’s what I see over there too. But I thought Lakai was saying FiveThirtyEight is giving McCain only a 2.9 percent chance of winning the whole enchilada, which isn’t what I see.

[tnetennba]

It was lower a couple of days ago.

[Merkwurdigliebe]

Remember, as early voting progresses, it could be that voters who were not considered likely to vote, are actually reporting having already voted, meaning more support for Obama than previously indicated.

[Knorf]

@cricetus

It was lower a couple of days ago.

Nate Silver wrote this yesterday:

“McCain’s position has improved very slightly: we now give him a 5.1% chance of winning the election, up from 3.7% yesterday.”

[wonky]

@DSeid

Does Rasmussen modify their LV screen methodology on the fly? Are their results reflective of a real change in the numbers, a statistical random walk, or a change in their sampling methodolgy either as the election gets closer or as they consider the LV question more throughly as they go along?

I think Rasmussen modifies their party weighting on the fly.

[DSeid]

RTF,

Yeah, I think I’m reading it right … the last 4 days they average about the same and Rasmussen was even putatively higher for one day. Before that the Gallup LV-e was virtually always higher than Rasmussen’s results, often between 4 or 5 points higher. About the same that Gallup’s LV-e has been running higher than their LV-t. Gallup began your sample run in a stretch of 9’s while Rasmussen began it with a couple of 5’s. Now Gallup is pretty much the same if not a hair’s breadth lower while Rasmussen has come up Gallup’s numbers.

Now again that could be just statistical noise and the race can be rock solid. Or the race could be moving more Obama-ward slightly as it seems by Rasmussen’s numbers … or Rasmussen could be modifying their LV screen on the fly.

Does Rasmussen share their methodology in real-time? Do they modify their LV screen as the election looms?

On preview, jsgodess, is party weighting modification all of the explanation?

[Lakai]

@Marley23

That’s what I see over there too. But I thought Lakai was saying FiveThirtyEight is giving McCain only a 2.9 percent chance of winning the whole enchilada, which isn’t what I see.

The 2.9 is for Ohio.

[wonky]

@DSeid

On preview, jsgodess, is party weighting modification all of the explanation?

It probably could be.

I know Nate Silver talked about one of the big pollsters doing weekly party re-weighting and I think it was Rasmussen.

ETA: Not weekly:

To the extent that any pollster should weight by party ID, I think Rasmussen is going about it the right way, setting targets based on a six-week rolling average of all interviews conducted.

Quirks/Concerns: The poll sometimes crosses the line from being stable into lower-case-c conservative, as the strict party ID weighting scheme means that it often won’t react as strongly as others to changes in the mood of the electorate.

[DSeid]

Adjusting on a six week rolling average seems especially unlikely to move them up like this.

[wonky]

@DSeid

Adjusting on a six week rolling average seems especially unlikely to move them up like this.

I don’t know. Once the move starts, it should continue, so if they are seeing a shift in the party ID that started a while back, they might only now be catching up to the party IDs the other polls were using in the first place?

[Lantern]

An articlewhich looks at McCain’s oddly frequent visits to Iowa. No real answer except some vague talk about numbers which supposedly show the race is closer than public polls suggest.

[DSeid]

This Mark Blumenthal article (discussing the 2004 election) may have my answer!

Via email, Scott Rasmussen said he uses screening questions to select the sample, and questions about likelihood of voting, past voting and interest in the election (closer to election day) to “further refine” their likely voters (“i.e.-you might get through the screening questions but still be considered a low probability voter”). Rasmussen does not aim for a specific cut off percentage.

New Registrants - Via email, Rasmussen said it was “possible” for new registrants to be classified as likely voters, adding only: “As a practical matter, it is easier for younger voters who are first time registrants than older voters.”

(Bolding mine.)

So as election day approaches Rasmussen weans off of past voting participation and more into intent to vote as his model, ie from the Gallup LV-t to the Gallup LV-e style model. He also apparently varies how many get to be counted based on expected turn-out as election day approaches. Or at least that was how he did it then …

[wonky]

DSeid, interesting investigative work! I wonder if any pollsters start pushing undecideds harder as the day approaches, too.