We’ve had a couple rather interesting poll results lately – results that bring up a lot more questions than they answer.
After Paul LePage lead the race for Maine governor for the entirety of the summer by anywhere from 8 to 18 points, we have seen two successive polls show the race tightening. But both provide some rather interesting head scratching moments.
A week after LePage’s “angry press conference” as it is now known as, Rasmussen conducted a poll of Maine voters which showed a nearly 20 point advantage for LePage, reassuring his supporters that the flash of anger had perhaps helped rather than hurt the nominee. This was especially significant, because up to that point Rasmussen had been showing the smallest leads for LePage out of all the polls done.
Then came Libby Mitchell‘s internal poll which showed her only four points behind LePage, 38% to 34%. Quite a different result than what Rasmussen came up with. Many people – including me – dismissed the results of the poll, mostly because it was a campaign’s own internal poll, and the campaign refused to provide the cross tabs to show us who was polled, what questions were asked, and what demographics were hit. They simply claimed that it was 500 likely voters.
While I actually do believe that campaign internal polls are generally just as (sometimes more) reliable than public polls – candidates want the best information possible to make campaign decisions with – I simply didn’t buy it in this case. The trouble with campaign internals is that the pollster will simply ask whatever the campaign asks them to ask. Asking a voter, “do you have a positive or negative impression of Paul LePage as of right now?” is going to produce drastically different results than if you were to ask, “after seeing the video footage of Paul LePage shouting at a reporter when pressed on his wife’s tax dodging, do you have a positive or negative opinion of Paul LePage?”
Since the Mitchell campaign refused repeated requests from myself and other members of both the blogosphere and the media to release the full data, I viewed these results as likely very suspect, especially in light of the most recent Rasmussen survey of voters. Still, the poll had some merit, and the truth of where the race was probably was in between the two polls.
But then came the most recent Portland Press Herald / Critical Insights poll results, which literally shocked just about everyone. It showed Mitchell leading LePage 30% to 29%.
Critical Insight’s first poll results from a little more than a week prior showed a very different story. That poll had LePage up by 13% over Mitchell. At the time many insiders scoffed at the results. Dennis Bailey, when asked about Critical Insights track record on Twitter, said the following:
Not so great or experienced with political polling, better at consumer research. Suspicious of a 1-day poll. Never seen that before
Additionally, Mike Tipping published a piece in DownEast, asking questions about the somewhat cozy relationship between the polling firm and Eliot Cutler’s campaign:
One gets a different sense of Cutler’s chances in the coverage of the poll in the MaineToday papers, however. Their piece on the results has the pollster, Mary Ellen Fitzgerald, commenting on the high number of undecided voters, calling it a “stinging indictment of the traditional parties.” She is also quoted as saying that “I think it sort of augurs better for the independents, even though the independents didn’t fare well in this poll,” and “I expect this will change drastically over the next few weeks.”
Fitzgerald offers no evidence to support her conclusion that the undecided numbers are a reflection of dislike for major party candidates rather than a by-product of candidates that are relatively new to the electorate contesting an open-seat race. The toplines released show no questions about name recognition or candidate favorability.
Critical Insights has a history of having a higher number of undecideds than other polls with different methodologies. In 2006, during the last gubernatorial race, CI released a poll conducted on October 23rd (their only poll of the race) showing 11 percent of the electorate still undecided, even that close to the election. A poll conducted the day before by SurveyUSA found only 2 percent undecided.
It would have been useful for the papers to disclose that Critical Insights, Fitzgerald’s company, has done work for the Cutler campaign. Public expenditure reports show a $2,000 outlay on April 16th from the Cutler campaign to Critical Insights for “polling and survey research.”
I myself had heard that Critical Insights had a good reputation, but after doing more digging I found that their reputation was based on the polling they do in other industries, not politics. As Bailey said, their history in politics isn’t exactly extensive, and their results in the past haven’t been particularly helpful. Remember that a couple weeks before the election, Critical Insights published a poll that showed name ID for the GOP candidates thusly:
- Les Otten – 30%
- Peter Mills – 16%
- Steve Abbott – 8%
- Bruce Poliquin – 8%
- Paul LePage – 7%
- Matt Jacobson – 3%
- Bill Beardsley – 2%
Given LePage’s results, it is very obvious that his name ID was much higher than 7%, given that only two weeks later he crushed Otten, and nothing he did in the last two weeks would have driven his ID that much higher – he bought one commercial on TV, which was joined by one from just about everyone else. Much like Tipping said, they had a massive undecided number when in reality the number was a great deal smaller.
But again, that doesn’t necessarily invalidate this poll. Just because they got things wrong in the past doesn’t mean they aren’t improving and it doesn’t mean this poll is suspect.
However, the actual poll itself provides plenty of reasons to question it.
While the problems with the previous polling from Rasmussen and PPP were mostly demographic skews slightly to the older part of the electorate (which, as I’ve argued before, is irrelevent to the totals since both the very young and very old demographic showed LePage and Mitchell tied, so any skew the other direction wouldn’t change things), the problems with this poll are more structural.
First, let’s start with the sample size. Typically for a statewide poll like this, a sample size of 500 is about the smallest you want to get. That’s what Rasmussen has been using as they poll states, while other outfits use a much higher sample. Public Policy Polling, for instance, polled 1,468 people in their survey in early September. Indeed, Critical Insights got responses from 603 voters in their first poll.
But in this poll, they surveyed only 405 respondents. Their margin for error was almost 5%, which for a poll of this type is really outside of the norm. Rasmussen’s polling in the state was about as high of a margin of error and as low of a sample as I think anyone could find useful. This is just very small.
Smaller still is the sampling on the congressional races. Anyone who wants to put any stock in those numbers might want to note that only 200 people were surveyed per race. However small a sample of 400 respondents was, 200 is simply worthless.
Then there is the oversampling of Democrats in the poll. In this survey, the ratio of Democrats to Republicans was 39% to 33%, respectively. While this would probably make sense in a presidential election or a “standard” midterm election, given everything we are seeing in this cycle it represents a gross misinterpretation of what is happening on the ground.
Even somewhat positive national polling for Democrats is showing a massive enthusiasm gap:
But the Gallup poll also found that Republicans “remain significantly more likely than Democrats to say they are ‘very’ enthusiastic about voting in this year’s midterm elections,” with 47 percent of Republicans very enthusiastic about voting compared with 28 percent of Democrats.
Gallup notes in its analysis: “Given this continuing enthusiasm gap between Republicans and Democrats, even a tie in registered voters’ preferences will almost certainly mean the Republicans will garner the most votes on Election Day.”
In 2008, the partisan makeup of Maine was 33% for Democrats to 27.4% for Republicans with 39.6% unenrolled. That is a 6 point split, which is what Critical Insights used in their survey. However, the registration data as of the primary election this year shows a slight tightening in registration (about a half a percent, but still significant)
But more importantly, that enthusiasm gap described above means that the likely voter universe should be quite a bit narrower than what Critical Insights is using. The other polls are putting the two parties within a couple percentage points of each other one way or another.
However, Republicans should be careful to not completely dismiss what is going on with this poll. It is critically flawed, but it likely does tell us that the campaign is tightening a bit, which I think we all expected.
This should be expected. LePage has been in the news a great deal lately, and most of it has been unhelpful to him. The press conference issue may have helped him with some voters, but likely made a few independents raise their eyebrows. In the last week, Maine Democrats have continued to hammer away at LePage, and have been attacking him for what was obviously a joke about punching A.J. Higgins (kudos to Gerald at Dirigo Blue for being fair on that issue) and for saying that he would be on the front page of the papers a lot for telling the president to “go to hell“.
I’ve been pretty hard on LePage, but I think both attacks on him are ridiculous. One was a joke, and the other was very clearly hyperbole. None the less, Democrats are sinking their teeth into LePage, and it would be foolish to believe that it wasn’t having at least some effect on his favorability ratings, which in turn likely drag down his numbers a bit.
So after all of this, what do I believe?
I believe that the race is probably somewhere between a 5 and 7 point lead for LePage at this moment in time, but that if he doesn’t regain his footing and recapture the message war a bit, those numbers could further erode.
Luckily for you the reader, Pine Tree Politics is commissioning a poll to make sense of all of this. We will be partnering with the Maine Center for Public Opinion to get a more detailed, accurate snapshot of the race. Our sample size will be significantly higher than both Rasmussen and Critical Insights, we will be polling likely voters, and overall we should get a very accurate picture of the race.
Our questions will be straight forward and non-partisan, because we want to know what the hell is really going on in this race – something that Mitchell’s poll and Critical Insights have not told us. I’d like to thank both MECPO for providing an affordable solution that I trust, and my partners here at Pine Tree Politics for helping me split the cost.
So stay tuned. Is the race tightening? You’ll find out for real soon.
Account
Start a blog
Send us your news
Post an event
See it. Fix it.