Is Rasmussen using D+3, or D+6?

Self | Oct 29, 2012 | Self

[HannibalHamlinJr]

Got it - so the D+6 was probably a reference from a state poll?

Assuming so, thanks PJ

[gswilder]

nhwingnut....

why does RAS weight like that then? If his own party ID split is different? Not challenging you, but you have provided some pretty insightful information. Thanks in advance....

[CatOwner]

You’ve provided specific information, duplicated by “Political Junkie Too”, and yet some don’t seem to want to hear the answers.

[ROCKLOBSTER]

Ras isn’t weighting the polls—the reality is that there are just more Ds.

Gallup doesn't ask their party affiliation.

[InterceptPoint]

Thanks for posting. This needed to be flushed out.

[nhwingut]

Once it gets out on Twitter it’s hard to bring it back.

But I am positive that Rasmussen is using a D+3.

[Utmost Certainty]

I’ve been under the impression that most of the legit pollsters simply report whatever the party representation of the sample happened to be.

[HannibalHamlinJr]

Thanks NH ...

Certainly makes sense that it’s still D+3

And probably it was a serious discussion about D+6, but referring to an individual state, or just quoting a bad source. Not going to investigate the entire thread.

[nhwingut]

why does RAS weight like that then? If his own party ID split is different?

It's the million dollar question. It's frustrating that his own surveys show a R+2.6 and yet he re-weights it to a D+3 - almost a 6 point swing.

In the past (i.e. 2008) he always used his party id survey for his daily polling. I mean isn't that the purpose of the party id survey? That's what Gallup does. They do not weight. They use what they get. And they are getting R+1. That's why it is unlikely that Romney will ever drop in the Gallup poll with that sample, and a week to go.

My opinion is that Rasmussen (who takes heat from the left) is hedging his bets, and being conservative.

If Romney ends up being up 3-4 in the final D+3 Rasmussen poll, add an extra 3 points. That's why I think it'll be 52-47 for Romney.

[Political Junkie Too]

Probably.

I just found the North Carolina poll from Saturday was 500 likely voters 34R/41D/24O.

The Pennsylvania poll from Thursday was 500 likely voters 37R/43D/20O.

The Michigan poll from Friday was 500 likely voters 34R/37D/29O.

The Massachusetts poll from Tuesday was 500 likely voters 17R/43D/40O.

I could go on.

-PJ

[Ravi]

I was wrong. Retweeted @numberscruncher who was incorrect. Nhwingnut’s explanation is exactly correct. Romney was winning only 85-88% of republicans up until today so that’s why he wasn’t cleaning pbama’s clock even though he was cleaning up wit indies.

[CatOwner]

Would love to see the CA numbers:

95D/-10R/15O (or at least it seems that way having lived here for over 50 years)

[nhwingut]

Some do, some don’t.

It depends on the sample size. Gallup does a massive sample (2500 plus) so they will get a good read of the electorate and will not weigh.

But smaller samples have to be weighted or else they would be useless.

[CatOwner]

"Romney was winning only 85-88% of republicans up until today so that’s why he wasn’t cleaning pbama’s clock even though he was cleaning up wit indies."

Where in the heck is he polling these Republicans? I do not know of one Republican in CA where I live who is voting for Obama. Squirrelly, if you ask me.

[Utmost Certainty]

Cool, thanks. Gallup seems the better barometer to me in that case.

[X-spurt]

Not according to the guy who runs Gallup. They do not ask party at beginning of poll, they do ask party at end of poll.

[HannibalHamlinJr]

That’s been my explanation to friends.

Rasmussen has a reputation to uphold in terms of accuracy, but he also has a business presence & narrative.

No one can ever prove you were wrong 2 weeks before the election.

If I’m Rasmussen, I’m thinking ‘stay close enough to the truth that I can veer towards it a few days prior to election, but stay within the pack, giving me the most options.’

‘Hedge’ and ‘conservative’ are the exact two words I’ve probably used 10 times in emails to explain to friends in the past week ...

Thanks!

[stickywillie]

if ras is not sampling properly....+3 or +6D, when really it’s + something for R, why is he any better than the rest?....because he’s not +13D? why bother to sample @ all if your not being honest, like the lib polls?

[Political Junkie Too]

The last Rasmussen California poll was in February. 31R/38D/31O.

-PJ

[ROCKLOBSTER]

Yes, but I also thought I heard him say they don’t use it for weighting.