Down East 2013 ©
Three polls have been released in the last two weeks in which Maine voters were asked to express their opinion of Paul LePage and his performance as governor. Many are pointing to the most recent, a Critical Insights survey showing LePage’s approval rating at 31%, as evidence that support for the governor has quickly plummeted.
The real answer to how the public views LePage and his time in office is much more complicated.
First, let’s look at the results.
Pan Atlantic SMS published a poll of 401 registered voters who voted in the last election  commissioned by Maine Today Media, conducted from April 25th to May 2nd, that showed 38.9% of Mainers with a somewhat or very favorable opinion of governor LePage, 56.1% with a somewhat or very unfavorable opinion, and 5% unsure or undecided.
The Maine People’s Resource Center published a poll of 970 registered voters  from May 1st through May 8th showing LePage with a job approval rating of 41.1%, with 52.7% of respondents disapproving of his performance and 6.2% unsure or undecided. (Important disclosure: I work for MPRC and had a large role in developing this poll.)
Critical Insights published a poll of 600 registered voters , conducted May 5th through 9th. This survey showed LePage with the approval of 31% of respondents, with 54% expressing disapproval and 15% unsure or undecided.
So, did LePage’s approval rating go up two points and then down ten in just a few days, or is something else going on?
First, we should note that these polls have margins of error of 4.9%, 3.14% and 4%, respectively, at the 95% confidence level. Each result should be considered an approximate range, rather than a single data point. It’s interesting to note that the disapproval numbers for the three polls were almost all within each other’s margin of error. It’s the approval numbers that vary more widely.
Second, we should look at how the questions were worded.
PA SMS: “I would like you to tell me if you have a very favorable, somewhat favorable, somewhat unfavorable, or very unfavorable opinion of Governor Paul LePage.”
MPRC: “Do you approve or disapprove of how Paul LePage is handling his job as Governor?”
CI: “Do you approve or disapprove of the way Governor LePage has handled/is handling his job as governor of Maine?”
In addition to asking about LePage as a person, rather than his job performance (they asked a separate job performance question that isn’t easily comparable) PA SMS offered respondents a scale of possible responses. Offering people more options may have increased their ability to capture the opinions of people who leaned in one direction or another, which might account for the 10-point difference in the rates of undecided respondents between the PA SMS and CI polls.
The undecided rate for the MPRC poll, which is also much lower than the CI survey, despite the similarity of their questions, may have been influenced by the fact that the poll was conducted using an Interactive Voice Response (“press one for yes”) system rather than a live interviewer. Studies indicate that people may be more willing to express their opinions (especially if they perceive them as unpopular) to a computer rather than to a real person.
There are a vast number of other differences between the polls that could potentially influence the results, everything from how those surveyed were chosen to the method and factors used to weight the results to the times of day the poll was conducted and the number of attempts made to reach each respondent. All of these were almost certainly different between the three polls.
PA SMS, for instance, attempted to weed out respondents who hadn’t voted in the last election, while CI and MPRC just polled all registered voters. (Although it’s not clear exactly how much of a difference this would make – respondents grow more likely to report that they voted, even if they didn’t, as time passes after an election.)
One factor that could have had a significant impact is what questions were asked prior to the questions about LePage. If people were asked about abortion immediately beforehand, for instance, they might be more likely to judge LePage based on their perception of his performance on social issues, rather than on economic or other factors.
My favorite exaggerated illustration of this effect is this scene from Yes, Minister .
While the MPRC poll asked about LePage’s approval right away, after confirming that the respondent was a registered voter, both CI and PA SMS asked other questions first. CI asked about a series of national issues and then moved on to state issues, including gubernatorial approval, rotating the order of the questions. This means that respondents first heard a series of questions on things like the national economy and nuclear power before they were asked about LePage.
PA SMS asked four questions on the economy and several on online banking before asking their questions about the governor.
So, how can we interpret these results?
The most useful thing we can do with these numbers is probably to consider them as signposts and compare them with previous and future results collected by the same organizations using the same methods. This should give us a good sense of how opinion is changing over time.
If you’re looking for an idea of where exactly things are now, I’d recommend looking at all these results together. There are various methods of comparing and averaging polling numbers (including looking at the margins between the approval/disapproval numbers, rather than considering each individually), but my best offhand guess for where LePage sits with the public, based on these polls, is probably an approval rating somewhere in the mid to high 30s, but with softer support than opposition.