Walker’s Polling History

A summary of comments from a previous post, but worthy of a few minutes I think.

Scott Walker’s polling has been all over the place prior to his other two state-wide elections. Real Clear Politics does a lovely job of keeping an archive.

Here is 2010 – from prior to Barrett right before the vote.


The 2010 polls averaged 3 points higher than Walker’s actual return.

Here is 2012 – from before it was clear there would be a recall up to the vote.


The 2012 polls averaged within one tenth of a point of the return. Kind of cool, huh? But, there was also a 15 point range in those polls, from Walker +12 to Barrett +3. I have no idea who We Ask America is as a pollster, and since all three of those polls were very high returns for Walker, I’m instinctively inclined to ignore them. I also notice the Marquette polls are darn close.

Here’s the whole list. I don’t think I’m violating use here, but if someone nudges me otherwise, I’ll take it off and let you click back and forth.


It’s early yet, but here’s the polling for 2014.

Here’s the list:


See that Marquette poll from May that has Walker +3? That’s the 46 to 46 poll from the earlier post. (The one that lauded such oddball comments from liberals who won’t even call themselves Democrats but are more than happy to chest bump over a tied poll.) The 48 to 45 read is “likely voters” while the 46 to 46 tie is from “registered voters.” No, I’m not inclined to speculate the cause of the difference. Yes, I wanted you to know it is in the RCP list above, even if it doesn’t look right to you. And I will assume RCP uses the “likely voter” numbers all the time, but I can’t guarantee it. There’s an RV or LV next to the sample number to let you know which numbers RCP included.

I did find it amusing that at least one right-sided website was quick to call Marquette out for oversampling Democrats. Geez. No one over samples on purpose. Perhaps the sample drew more Dems, but it can be adjusted for in the end if necessary. And, those were the registered voter numbers that created such a tizzy. Collin Roth, writing for Right Wisconsin, never mentions the “likely voter” spread that puts Walker ahead.

(BTW, you can open that Right Wisconsin link in an incognito or private browser window if you are past their limit for a month and not a subscriber.)

Oh, and kindly remember the tied numbers grabbed all the headlines – even the Marquette Law statement title, because THEY MAKE MORE MONEY WHEN THEY KEEP YOU EXCITED. No one is giving you poll numbers out of the kindness of their hearts. Even I wouldn’t mind your buying that book I wrote. 🙂

I still say Walker comes in around 53% next November. But heck. What do I know…


  1. “No one over samples on purpose.”

    I think this is true once the election is close. I’m so sure it is always true well before the election.

    Think about how steady Wisconsin has been over the last 4 years. Then look at your poll history for the recall election. From August 2011 to April 2012, the polls were actually pretty close with a couple showing Barrett ahead. Then we get to May and the results align well with the final outcome (which also happened to be almost the exact same outcome as 2 years prior).

    So did Wisconsin shift to Barrett and then shift back to Walker? Or did we stay the same and a couple goofy pollsters tried to create a narrative to manufacture headlines to give their candidate a shot and skewed the results in the process? I think it is a fair question.