Two weeks ago, a badly flawed Rasmussen Poll indicated that Sen. Amy Klobuchar and Gov. Tim Pawlenty enjoyed equally high and very impressive 56 percent job approval ratings, while only 41 percent rated Al Franken as "excellent" or "good" in the way he is performing as a senator.

The Strib relayed the results, without noting the serious flaw, which is that Rasmussen's questionaire did not use the same language when asking about Franken as when it asked about Pawlenty. My friend and former colleague Rob Daves (former director of the Strib's Minnesota Poll) said Rasmussen's questionaire was "fundamentally flawed."

I'll go into the polling issues a bit below and link to everything. But the main takeaway is that if you read about that previous poll and believed a solid majority of Minnesotans approve of Pawlenty and disapprove of Franken, you should know that a more recent poll, by Survey USA, which used the same language in asking about all three of the state's top elected officials, finds the approval/disapproval ratings as:

• Klobuchar: 64/31
• Franken: 49/44
• Pawlenty: 45/52.

I wouldn't want to make too much of the new SUSA poll, which is a robodialed poll (so is Rasmussen) of just 600 adults (not registered voters) with a margin for error of 3.9% (which means that there is no statistical significance between Pawlenty and Franken on approval/disapproval). But it is a big improvement, on basic methodological grounds, over the Rasmussen and has received less attention (mainly because the Strib wrote about the Rasmussen poll).

The main hero bringing these problems to light has been Jonathan Singer writing for the pro-Dem website MyDD. He seems to the one who noticed the bizarre questionaire error by Rasmussen. Here's what it is:

Rasumussen's questionaire used this question when asking about Franken and Klobuchar:

How would you rate the way that Al Franken/Amy Klobuchar is performing his/her role as Senator....excellent good fair or poor?

But this question when asking about Pawlenty:

How would you rate the job Tim Pawlenty has been doing as Governor... do you strongly approve, somewhat approve, somewhat disapprove, or strongly disapprove of the job she's (sic) been doing?

Singer seems to think Rasmussen did it on purpose, but Rasmussen told Mark Blumenthal of Pollster.com that it was a "mistake that fell through the cracks." (Click through to Blumenthal for the full Rasmussen explanation.)

There is no legitimate reason not to use the same wording when asking about the same topic (job approval) on two different public officials in the same poll. This is the mistake that Daves called "fundamental." It is especially signficant in this case because, as Blumenthal also explained, offering respondents the word "fair" and then counting "fair"  as a response of disapproval runs a serious and proven risk of attracting respondents who think "fair" means something like "not bad" or even "pretty good." But whatever the advantages of using "fair," there is really no excuse for using it for one politician and not another.

Survey USA, whose main local client is KSTP-TV and who does a regular monthly round of approval rating polls, uses a very straightforward questionaire:

Do you approve or disapprove of the job Tim Pawlenty (Amy Klobuchar/Al Franken) is doing as Governor (Senator)?

The answer can be only "approve" or "disapprove," and there's not much room for confusion.

If you'd like to peruse the internals of that SUSA poll, the two senators are here and TPaw is here.

On the same poll, SUSA asked Minnesotans about Pres. Obama with this result: Approve: 55. Disapprove: 40.

That's a small improvement from SUSA's August sample in Minnesota, which gave Obama a 53/44 approval/disapproval.

In the cases of TPaw and the senators, the late September SUSA ratings showed a small (statistically insignificant) improvement for Franken and Klobuchar over the late August numbers, and a small decline for Pawlenty.