Public Policy Polling: Untangling the tangled web they weave

I finally received an e-mail reply to my question regarding the sample in a couple of polls they released last week. One had Feingold barely ahead of Republican Ron Johnson in the upcoming Wisconsin Senate race. The other showed Scott Walker with a romping lead over primary challenger Mark Neumann. (See the comments in this post.)

Something didn’t feel right. As I mentioned, Wisconsin doesn’t have a closed primary. How would one determine Republican Primary voters?

Now this conversation with Public Policy Polling is ongoing – it’s literally being updated as I type – but I think it’s worth sharing:

How were the 400 Republican primary voters surveyed determined? I notice there’s a 100% Republican response when asked.


Cindy Kilkenny

From Tom Jensen –

After all respondents were asked questions about the general election for the Senate and Governor’s races, they were asked if they planned to vote in the Democratic or Republican primary this year or no primary at all. The results released were among just those who said they planned to vote in the GOP primary.

I thought that survey was 638 Wisconsin voters. So 400 of your 638 were Republican? Really?

No we polled extra Republicans to get 400 interviews of primary voters who weren’t included in the 638 general election voters. The 638 was 33% Democrats, 32% Republicans, and 34% independents as can be seen in the full poll results on our website.

How did you poll extra Republicans? From where was that sample drawn?

I’ll update you when I get a reply.

Remember this guy? I’m pretty sure he’d think I need to keep asking questions. I was taught a survey is only valid when the sample is uncorrupted.

“Polled extra Republicans.” Sheesh.

Here’s a link to the “400 Republican primary voters” survey, and another to the “638 Wisconsin voters” work.


  1. They weren’t the results you wanted, hey Cindy?

  2. I don’t want anything one way or the other, Aaron, but the results really don’t make sense given the way the two Republican candidates run against the Democrat in other polls.

    If I can be satisfied with the sample and the method, then the results are what they are. Right now I can’t say it makes sense, but then maybe I’m missing something and it will all be clear soon.