Wisconsin is the home of 3,400,000 registered voters. Below is the question of greatest interest in the most recent Marquette poll:
A Q14 (ALL REGISTERED VOTERS)
In an election to recall the governor between Republican Scott Walker and Democrat Tom
Barrett, would you vote for Walker or for Barrett?
Walker 356 or 51%
Barrett 303 or 43%
How is this possible? Nearly 1 million signatures were collected on petitions to trigger the recall election. 1 million is nearly one third of all the registered voters in the state. The result of the Marquette poll begs the question: When one third of all the people who can vote in the state have already signed recall petitions, how is it possible that 54% of the people called said they will vote for Walker?
To his credit, the administrator of the poll Charles Franklin publishes the detail methodology in support of his findings. Most polls do not allow for us to look inside the methods, but only publish the results. This is not the first time Mr. Franklin has been challenged as to the accuracy of his sampling methods, which consistently demonstrate anomalies that skew the findings in favor of conservative and very conservative voters. Yet, his sampling methods have remained largely unchanged since his first poll of the Walker’s approval rating back in late January. No one is looking for a sampling that will favor any candidate, but we are looking for a sampling that rides on an even playing field. So, what is the problem? What are the anomalies that suggest skewing to the right. Even Mr. Franklin acknowledges his astonishment at the number of signatures collected to recall Walker in an interview on “Here and Now” last January.
Video of Full Interview
Franklin currently tweets as @PollsAndVotes
Here is the problem with Mr. Franklin’s poll, as described by Jud Lounsbury’s blog since his first poll back in January
Many in Wisconsin are shocked that Tom Barrett somehow went from being up by a point to being down by six points in a span of only two weeks. Have voters suddenly taken a sharp disliking to Tom Barrett? No.
The reason why the poll numbers have changed is because they significantly changed the sample of who they polled. In this poll there are more conservatives and moderates and fewer liberals:
Latest Poll: Conservatives 48% Moderates: 39% Liberals: 20%
Poll Two Weeks Ago: Conservatives 43% Moderates: 32% Liberals: 22%
Not surprisingly, when you poll more conservatives and fewer liberals, your going to have a very different result.
As discussed in previous posts, if we average the last three exit polls in Wisconsin (2006, 2008 and 2010), we find that the actual breakdown of the Wisconsin electorate is 22.7% liberal, 46.7% moderate, and 31% conservative.
The glaring anomaly has been pointed out by Jud in every Marquette poll since January, and in some polls, when the percentages are closer to the exit poll numbers, guess what? Barrett’s numbers go up.
One last interesting tidbit that suggests skewing in the most recent poll. Nearly one third (29.1%) of those called were of an age 60 or over. We know that older people tend to be more conservative and yet they represent the largest age segment of the poll.
Accurate, scientific polling is appreciated. Mr. Franklin the voters of Wisconsin deserve better.