Poll Spinning 101

So there’s a new poll out today, have you heard?

  • Mike Michaud:  38%
  • Paul LePage:  36%
  • Eliot Cutler:  15%

Of course you have.  The poll, conducted by Democratic polling outfit Public Policy Polling, was out for about twenty five seconds before the left-wing echo chamber sprung into action to spin and define the results to somehow show that Mike Michaud is in the lead and in a strong position to win.

For instance:

No public poll since Michaud said he’d likely run for governor has had LePage ahead. Yes, I know LePage told a conservative women’s group he has “a very strong lead,” but there’s still no evidence for that.

Concurrent with this argument is the constant – if entirely futile – refrain that Michaud or Cutler would win the race if it was a one on one race.  Which it is not.

And then there’s this:

This lack of movement is likely good news for Michaud. Since the last poll, both of his opponents have officially launched their campaigns. These scripted events and the attendant media coverage should have been opportunities for them to gain ground, but that doesn’t appear to have happened.

So basically, the results of the poll are another in a long list of great polls showing a strong lead for Michaud, and the lack of movement is good news for Michaud because, reasons.

Allow me to politely throw a little cold water on the gleeful cheerleading of Michaudistas.

The Michaud “lead” is in the margin of error

The survey conducted by Public Policy Polling included 964 Maine voters, and had a margin of error of 3.2%.  Michaud’s lead in this survey was 2%.

In the business, we say that this lead is in the margin of error.  It means that Michaud could be up by a little over 5%, or… well… Paul LePage could be up by more than a point.

Were the results of the survey to have been reversed, with LePage up 38% to 36% over Michaud, you can bet the same liberal commentators who are declaring a strong, steady, important lead by Michaud would be telling us that the lead wasn’t an actual lead and that the race was a “statistical tie.”

Doing the math (since PPP doesn’t include the actual numbers) this means 366 respondents chose Michaud, 347 chose LePage and 145 chose Cutler.

If ten people answered differently, the entire poll changes.  If the sample was slightly different and surveyed other areas of the state or other voters and just a few votes were different, the “lead” changes.  We are dealing with small changes that would make a big difference.

Which leads me to my next point.

Public Policy Polling has some problems

I’ve never particularly liked the way that Public Policy Polling has modeled Maine voters.

To start, they’ve never been particularly transparent about all their numbers.  They release cross-tabs with percentages but no numerical totals, and don’t include important data about number of respondents by county or town, or other information we could use to evaluate their method.

But a quick look at the numbers makes a guy curious.

To start, the PPP sample was somehow 9% non-white, which is weird for a state that is over 95% white.  This by itself isn’t necessarily troubling, but it does beg the question of what areas of the state PPP used to populate the poll.  Heavier in more diverse areas of the state (Portland, for instance) would tend to naturally provide a more liberal response.

Beyond that though, the sample they used was 37% Democrat, 31% Republican and 32% Independent.  The actual registration rate in Maine is 31.92% Democrat, 27.25% Republican and 36.95% Independent.

In other words, they oversampled Democrats by 5.08%, Republicans by 3.75% and undersampled Independents by 4.95%.

This can be particularly problematic for two reasons.  First, as an off-year election, the electorate nearly always naturally trends more Republican than in presidential years.  Second, independents typically trend toward Republicans more than Democrats, and in Maine a rather large proportion of Independent voters are former Republicans who unenrolled, but are still more closely aligned with the GOP than the Democrats.

Now, we don’t know what a more realistic Democratic number with more independents would produce for a final number, mostly because PPP didn’t include the numbers that showed where the Independents they polled broke.

But one look at the cross tabs shows that the LePage favorabilities of “moderate” voters (31-62) are underwater, but his favorabilities among “somewhat conservative” voters is extremely strong (72-23).  There is also twice as many undecideds in the “moderate” camp.  Similarly, Michaud does very well with “moderate” voters (60-22), but is very unpopular (30-52) with “somewhat conservative” voters.

So really, it would depend on what kind of independent voter is being polled.  In an election like we are about to go through, I would broadly assume it would be a slightly more conservative independent voter than usual, which would seem to favor LePage were the Independent numbers higher.

All of that said, Public Policy Polling has a questionable history, anyway.

In Virginia, the day before the electionPublic Policy Polling showed Democrat Terry McAuliffe up on Republican Ken Cuccinelli by 7%.

The actual result?  McAuliffe won by just over 2%.  The day before an election in a large state they have extensive experience in with easier to model voters and areas, they got it wrong by 5%.  They get away with it, of course, because they were still right about the result, which is how they have avoided disaster a number of times in the past.

They certainly get things right too, but as we all know a stopped watch is right twice a day.  They had a generally good year in 2012, but they messed up a lot of stuff in 2010, and have not had a good record in 2013 at all.

Indeed, their checkered history makes sense, given their questionable methodology.  It is so questionable that the liberal publication, The New Republic, engaged in a rather remarkable takedown of how they do what they do.  I’ll let them take this one:

After examining PPP’s polls from 2012 and conducting a lengthy exchange with PPP’s director, I’ve found that PPP withheld controversial elements of its methodology, to the extent it even has one, and treated its data inconsistently. The racial composition of PPP’s surveys was informed by whether respondents voted for Obama or John McCain in 2008, even though it wasn’t stated in its methodology. PPP then deleted the question from detailed releases to avoid criticism. Throughout its seemingly successful run, PPP used amateurish weighting techniques that distorted its samples—embracing a unique, ad hoc philosophy that, time and time again, seemed to save PPP from producing outlying results. The end result is unscientific and unsettling.

Then there is the fact that the firm is owned and operated by Democrats.  I don’t typically use this against pollsters, but in this case I will for a very good reason.

PPP has actively withheld poll results that it, though its partisan Democratic lens, didn’t believe or agree with.  Earlier this year, PPP conducted a poll in Colorado that showed Democratic state Senator Angela Giron trailing by twelve points in a recall election.  It then withheld that poll, and claimed it did so because it just couldn’t understand how she could have been losing that race, given that Barack Obama did so well in her district.

It is no wonder that The New Republic isn’t the only critic. The polling whisperer, Nate Silver attacked PPP for withholding the Colorado poll, saying on Twitter that it was a “VERY bad and unscientific practice for @ppppolls to suppress a polling result they didn’t believe/didn’t like.”

But more damning than that is his statement that “…what they’re doing barely qualifies as POLLING, if at all.”

Hell, even Mike Tipping took a swipe at them.

Which is all to say that they are a partisan outfit with very questionable polling ethics and methodology, who industry experts struggle to even call polling (Silver, for instance likes to call it forecasting).  They used questionable samples of Maine voters, didn’t transparently report results or methodology, and got a pretty large, high profile race wrong last week.

So trumpeting this result as very credible is a bit much, for me.  Oh, but that’s not all.

Michaud should be up by a lot more right now

Mike Michaud has had more than a decade as Maine’s affable, likeable, uncontroversial congressman from the second district.  He has had one really competitive race for his seat – the one that elected him there in the first place – and since then has not really been dusted up.

None of his opponents were able to lay a real haymaker on him, try as they might have.  They either didn’t have the money, didn’t have the attention, didn’t have the skills, or didn’t have the environment to really expose Michaud’s record to light and bring into question his effectiveness and worth as a congressman.

He then had a high profile and successful rollout, full of glowing (I’d even say fawning) press coverage for his run.  Even his revelation of being gay turned into a net positive for him, as it didn’t lose him any real voters and probably helped solidify his standing with the liberal base, keeping them from Cutler.

At the same time, LePage and his allies have not really said much about Michaud yet, no real ads have been purchased, no debates commissioned, and no contrasts drawn.  LePage had his own rollout, sure, but unlike Mike Tipping’s comments from earlier in this piece, he was never going to get much of a bounce from that, as we all know Paul LePage and have for four years.  Making an announcement with waving signs wasn’t going to do anything there.

And Eliot Cutler sits, lurking.  He hasn’t touched Michaud either.

In other words, Mike Michaud is about as clean and popular as he is likely to ever be in this race.  He has a 51-33 favorability rating and no haymakers have been landed on his jaw.

And he can only muster a statistical tie in a Democrat poll that oversampled Democrats?

Earlier this week, Ethan Strimling sent me a text message and asked what I thought the PPP results would be.  I told him that I thought Michaud would be up by about 8 points, because Michaud has had more of a coronation than a rollout, and LePage hasn’t engaged in the race yet.  I told Ethan that as the race truly began, I expected it to close and become a real fight, but that Michaud would probably be up at this moment in time.

How is Mike Michaud not winning right now?  The Democrat whining about Cutler being in the race ignores the fact that the Democrats are more or less unified behind Michaud right now.  This should (but won’t) prove to them, finally, that Cutler’s voters are not “theirs” but are in fact just as annoyed with the Democrats as they are with the Republicans and are voting for Cutler on his own appeal.

But if Michaud has a 51% approval rating, a unified party, all the positive press he could ever dream for, and a great rollout and can only muster 38% of the vote, what does that tell us about both him, and how this race is going to be going forward?

Put simply, if he really wants to win, he should be up by high single digits or more right now, because…

The dynamics of the race do not favor Mike Michaud going forward, and that makes this poll a big problem for him

Michaud is going to face something he has never faced in his political life: a challenge to his image.

LePage and the Republicans will have enough resources to actually define Michaud, which is not something that has ever happened to him before.

They’ll be able to actually tell the Maine voter that Michaud has gotten almost nothing done in Congress in a decade, with both Republicans and Democrats in charge.  They’ll be able to point out his craven flip flopping, particularly on issues like abortion.  They’ll be able to point him (rightly) as a puppet for liberal interests, and a willing accomplice to bloated spending boondoggles like the stimulus, and the increasing albatross of Obamacare.

And Eliot Cutler has one path to victory, and that is to savage Michaud and prove that he is the more electable alternative to LePage.  At this point, virtually none of the 36% that is with LePage will defect to Cutler, and Cutler will need to overtake Michaud, with Michaud voters, to win.

His favorability numbers will not remain 51-33, for once.

At the same time, Paul LePage has a story to tell that he has yet to really be able to tell to the public.  His four years in office will have been wildly successful, and he will have a lot of truly remarkable accomplishments to point to.

Pension reform.  Welfare reform.  Tax reform.  Unemployment dropping.  Jobs created.  Spending curtailed.  Hospital debt repaid.  The list goes on.

And his positions on the issues are ones which Maine people agree with.  LePage’s problem has always been that the media hoopla and distraction that can sometimes come from what he says overwhelm that fact.  But don’t think Maine voters will not have a chance to be reminded of that fact.  They will.

The point here is that Mike Michaud has been perched on an ivory tower, untouched and unsullied for his entire career, and he is about to be in a street fight.  I suspect he has a glass jaw, but even if he doesn’t, once the race begins and ads are run and he begins to be defined for the voters, he will drop, not rise.

LePage on the other hand has been through the fire for four years, and if you don’t already dislike the governor, you certainly won’t get to dislike him because of anything Michaud says.  To the contrary, there are a number of people who abandoned the governor but are able to be regained once he makes his case.

So in the future, Michaud is likely to go down, and LePage and Cutler are likely to go up.

That makes the results of this poll truly frightening for Michaud supporters.  They can spin it away all they want, but even if the poll results were legitimate (which I think they really aren’t) they would present a significant problem to camp Michaud.

LePage-land is no doubt very happy right now.

Matthew Gagnon

About Matthew Gagnon

Matthew Gagnon, a Hampden native, is a Republican political operative. He serves as the Director of Digital Strategy for the Republican Governors Association, and has previously worked for Senator Susan Collins and the National Republican Senatorial Committee.