MyDD :: The Numbers Rasmussen Was Hiding

The Numbers Rasmussen Was Hiding

by Jonathan Singer, Tue Oct 06, 2009 at 12:44:25 PM EST

Last month I called out Rasmussen Reports for skewing questions in its Minnesota polling to make it seem as though the state's Republican Governor -- and seeming Presidential aspirant -- Tim Pawlenty was more popular than he actually was relative to the state's well-known Democratic Senator, Al Franken. After Mark Blumenthal of Pollster.com got on Rasmussen's case, the pollster fessed to have run a bad poll. Now, via pollster SurveyUSA, we have an indication of what Pawlenty's numbers look like relative to those of Franken -- and not too surprisingly, when the same question is asked about both elected officials, they show Pawlenty in a significantly weaker position.

Do you approve or disapprove of the job Tim Pawlenty is doing as Governor?

Approve: 45 percent
Disapprove: 52 percent

Do you approve or disapprove of the job Al Franken is doing as United States Senator?

Approve: 49 percent
Disapprove: 44 percent

Rasmussen claims that it was all a big mistake, that the different questions for the two candidates was unintentional. Fair enough. Let's take them at their word. Nevertheless, what we do know is this: When Rasmussen asked two separate questions gauging approval -- one that tends to show higher approval numbers for Pawlenty, one that tends to show lower approval numbers for Franken -- the numbers came out as expected, with Pawlenty scoring 15 points higher than Franken. When SurveyUSA asked the same question about both Pawlenty and Franken, Franken came out ahead. So much for the notion of Pawlenty being popular in his home state, let alone more popular than other elected officials there.

Tags: Rasmussen Reports, Tim Pawlenty (all tags)

Permalink :: 6 Comments :: Post a Comment

Re: The Numbers Rasmussen Was Hiding (none / 0)

Good stuff! Thanks for continuing to follow-up this story.

by dvkenned on Tue Oct 06, 2009 at 03:13:58 PM EST

Re: The Numbers Rasmussen Was Hiding (none / 0)

Great post - it is fantastic to see people are watching our pollsters for accuracy. Unfortunately, you have your own inaccuracy in this post:

You say "when the same question is asked about both elected officials, they show Pawlenty in a significantly weaker position." The difference in approval ratings is 4%, and the margin of error on both polls is 4%. The word "significantly" in a statistical context implies that the difference is greater than the margin of error. This is not the case.

If you meant that the new numbers differ significantly from the Rasmussen numbers, you should revise the quoted sentence.

-Xander

by Reductive on Tue Oct 06, 2009 at 03:18:27 PM EST

Re: The Numbers Rasmussen Was Hiding (none / 0)

In the earlier poll, Pawlenty came in at 56 percent; here he came in at 45 percent -- a difference outside the margin of error.

Blogging here @ MyDD.com. Twittering @jonathanhsinger.

by Jonathan Singer on Tue Oct 06, 2009 at 03:52:31 PM EST
[ Parent ]

Re: The Numbers Rasmussen Was Hiding (none / 0)

It is significantly weaker relative to the prvious poll

Howard Dean is my go-to guy

by lojasmo on Tue Oct 06, 2009 at 04:23:29 PM EST
[ Parent ]

Why take them at their word? (none / 0)

They clearly have their thumb on the scale. A company as experienced as Rasmussen does not make this kind of mistake, and they have a well-known GOP bias. I'm waiting for Nate Silver or SOMEONE to just come out and say they're not going to take their polling seriously anymore, or at least treat every Ras poll as if the GOP itself (whether a committee or a candidate) commissioned it.

by bpfish on Tue Oct 06, 2009 at 05:30:09 PM EST

Re: The Numbers Rasmussen Was Hiding (none / 0)

Trust, but verify.

Let's give SOME props to Ras for actually responding substantively rather than calling for a blogger ethics conference.

I remember hating the beejezus out of Rasmussen because they consistently seemed to be skewing the election to the right in their poll results, but IIRC they actually ended up being one of the most accurate (non-aggregated) pollsters in the presidentials and mid-terms from 2000 on.

Please correct my memory if that recollection is in error.

I guess I'm sounding negative, what I'm really trying to say is don't be too quick to blow them off, even after this "mistake." Maybe they called a lot of elections correctly BECAUSE they skew right and looked good when elections skewed right themselves (e.g. Ras got lucky). But maybe they generally know what they're doing-- moreso than many of their competitors. Don't rule it out.

Let me conclude with, major, major props to Singer/MyDD for busting Ras on the Franken thing AND following up. Awesome work.

by alteran on Wed Oct 07, 2009 at 09:13:36 AM EST

You are not logged in.

In order to post a comment, you must be logged in. If you have a member account, please log in to comment.

If not, you can make an account right here. It's quick and free.