PPP Poll for January 1 (reason for optimism?)

[Heath]

Contrary to the media spin, Ron Paul actually improved during the course of the most recent PPP poll.
This info is from the NY Times 538 blog:

Jan. 1
PPP Survey of 670 likely voters

Ron Paul 21 %
Mitt Romney 18
Rick Santorum 18
Newt Gingrich 14
Rick Perry 10
Michele Bachmann 8

 

[Havax]

I suppose you fragment this poll like they did with The DMR to show Santorum in 2nd to balance out media fairness of stupid ways to look at polls.

 

[kylejack]

Yeah, but this poll includes Democrats and Independents. I'd like to see how he did with Republicans on each day. Anyone have the crosstab?

 

[Tunink]

What's the possibility that a good bit of those called/polled were simply not in that would have been Paul supporters due to events with the campaign?

 

[sailingaway]

Yeah, I've been saying that.

 

[kylejack]

Don't count too hard on the polls being wrong. Some of us tried that last cycle.

The only real game-changer I see that could alter the numbers is Phone-From-Home. The other campaigns don't know who their supporters are, but we do. That's got to be a huge advantage for turnout.

 

[FA.Hayek]

Don't count too hard on the polls being wrong. Some of us tried that last cycle.

The only real game-changer I see that could alter the numbers is Phone-From-Home. The other campaigns don't know who their supporters are, but we do. That's got to be a huge advantage for turnout.

this

 

[Diurdi]

Don't count too hard on the polls being wrong. Some of us tried that last cycle.

The only real game-changer I see that could alter the numbers is Phone-From-Home. The other campaigns don't know who their supporters are, but we do. That's got to be a huge advantage for turnout.

You have to take into account the margin of error for these things though. When everyone is within a couple points of eachother, you really can't say one way or the other.

 

[Romulus]

We have to remember too... look at what the polls were saying before Rand won by 10%. Not saying that could happen... but its very possible we could pull this off if we work like we're down 10.

 

[Liberty74]

Did anyone see the question asking the voters who they thought was going to win the caucus?

Romney 24
Paul 20
Santorum 9

 

[Liberty74]

What's the possibility that a good bit of those called/polled were simply not in that would have been Paul supporters due to events with the campaign?

That is what the margin of error is for.

 

[SaulPaulinsky]

Yeah, but this poll includes Democrats and Independents. I'd like to see how he did with Republicans on each day. Anyone have the crosstab?

They're on PPP's twitter feed. I spent a lot of time with them last night.

I think PPP *could* be underestimating the share of 18-29 demo voters. They're basically matching the last caucus, but the Dems had a caucus then and their share was much higher, according to a post on this board on another thread around 20%. Obviously, if that share is higher that favors Paul.

And we have the strongest support, least likely to change. Over 1/3 of Santorum's support came in the last week, and 48% of all people that said they decided in the last week said they could still end up supporting someone else. That is to say, they haven't really decided.

72% of Gingrich's supporters watch Fox News regularly.

This is really going to be about GOTV, because I rather doubt Santorum's supporters will end up supporting Paul in numbers.

I have it Paul 24, Santorum 20, Romney 19. I hope I'm right, because that would be a great result.

Apparently 20% of GOP caucusgoers favor WWIII.

 

[Exponent]

That is what the margin of error is for.

Technically, the margin of error only represents the likely error due to purely random effects. Non-random effects, like a certain candidate's supporters being less available for taking calls, is not covered by the margin of error. Pollsters attempt to compensate for some of these effects (especially ones like demographics, or who is expected to show up to caucus), but it is a guessing game with no calculable margin of error available.