Tuesday, September 9, 2008

More on the SurveyUSA poll

The crosstabs are up now on the SurveyUSA poll and there's already one pretty clear explanation for the bizarre numbers. The party identification breakdown on the poll is 41% Republicans, 40% Democrats, and 16% independents.

A month ago SUSA found a 46-33 Democratic party identification advantage. Although it is possible there could be some shift after the Republican convention a 14 point shift strains credulity. PPP has found a Democratic id advantage of anywhere from 8-15 points in its general election polls, and I think Civitas has found an id advantage of 11-13 points pretty consistently.

I am not one to trash Survey USA- I think they are a top notch company, but we all have our bad polls and I suspect for them this is one of them. We'll have new numbers tomorrow.


Anonymous said...

Whoa. You're going to try to debunk a SurveyUSA poll by conducting your own survey over ONLY ONE NIGHT and with NO CALLBACKS?

And are you going to weight by party? If so, what will you decide the spread should be?

Chris said...


Do you know if they do party affiliation by observation or do they ask the respondent.

We do ours by observation from the voter file.

If you're asking people how they are registered, there may be a demographic shift as people re-identify with the good vibes from the GOP convention.

Tom Jensen said...

To your first point: you mean like every single Rasmussen and Insider Advantage state poll? Hope you get on their case.

It is usually not my preference to do a one night poll but this is an unusual circumstance where I think some more realistic data needs to come immediately.

We do not weight for party id and will not for this poll.

Tom Jensen said...


I think they do it by self reporting since they do RDD polling if I am correct.

I have no doubt there's been some party id shift toward the GOP- we found it in the Florida poll we released today- but nothing of the magnitude conveyed in their poll.

Anonymous said...

Anonymous and Tom,

I think, it is the regan democrats that are supporting the GOP. Sarah Palin seems to be the lady Regan. When the Republicans had their primary, people were praying for a candidate like Sarah. She is becoming popular because of her stand on issues and her record in Alaska. Added to it, she is good looking and has good delivery of speech.

Let us wait and see, whether she is going to stumble in the debates or interviews.

Chris said...


I figured if a firm is doing RDD you have to use self-id party affiliation.

But remember the poll last week from Democracy Corps had R and D percentages much closer.

One thing from the SUSA polls that we know isn't right is the % of black vote going Republican. I don't care what any poll says, if the number of blacks voting Republican is over 4%, there's a problem.

Anonymous said...


I don't know about the party ID thing but you can look at the results within Party ID and see that what has happened is an enormous move to the Republican candidates by the Unaffiliateds.


Anonymous said...

Even if you keep the core cross tabs in terms of levels of support between Reps, Dems, and Inds for McCain and Obama, the result would still be much different if one used SUSA's last party id breakdown. It would turn their 20 pt lead for McCain into an 8 pt lead (52 M to 44 O). As to the sudden switch in Inds support, that may be true but come on McCain getting close to 60% support among Inds (he does not even get that level of support in some of the redder of red states). I'm wondering (given the relative small sample size for the Inds whether the Indie number is on the high end of the margin of error?).

Web Statistics