During the height of the 2012 presidential election, a whole closet industry arose on the political right, based on outright denial of the fact that polls consistently showed Barack Obama winning the presidency.
Conservative operatives claimed the polls were unfairly and systemically biased against Mitt Romney, and a website emerged to “unskew” the results from national pollsters (which they claimed were surveying too many Democrats) to make them more favorable to Romney and the Republican Party.
Some members of the media bought into and repeated these claims, perhaps because they made the election seem closer and more dramatic than it actually was.
GAUGING PARTY IDENTIFICATION
In the end, the 2012 polling was accurate. Polling aggregators like Nate Silver ended up pegging the results almost exactly, and conservative pundits who bought into the claims of systemic polling bias were deeply embarrassed.
You might think that this would have convinced those on the right to change their ways, but you’d be wrong. This year, the exact same conspiracy theory is back in full force, even here in Maine.
The major false claim underlying these assertions of polling bias is that party identification numbers in polls are wrong. For instance, those on the right might claim that a survey of Maine voters has too many Democrats or too few Republicans when compared to how people are actually registered by party in the state. This may seem on the surface like a legitimate critique, but there are two big problems with it.
• First, the party someone tells an interviewer that they identify with isn’t necessarily the one that’s on their registration card. A series of studies has shown that answers to these questions are much more fluid and are mostly a measure of how voters are feeling about parties and candidates at a given moment.
As the editor-in-chief of Gallup put it, “There is no reliable measure of the distribution of party identification within the population … . The distribution of party identification may vary in the total population from day to day, week to week.”
• The second problem is that even if party registration could be measured completely accurately, the population that votes in November will not have the same registration profile as all registered voters in the state.
Quite simply, not everyone votes. Who actually shows up is highly dependent on things like enthusiasm for candidates and knowledge of and interest in issues.
JOURNALISTS PARROT BIAS CLAIMS
Weighting a sample to party registration numbers or to previous election results means putting a finger on the scale and making a big assumption about who will turn out to vote. For this reason, no reputable national pollster weights their results by simple party registration.
“In effect, standardizing, smoothing, or otherwise tinkering with the balance of party identification in a survey is tantamount to saying we know how well each candidate is doing before the survey is conducted,” the Pew Research Center reports.
Unfortunately, certain conservative operatives in Maine either don’t understand or don’t care about these public opinion basics, and some reporters have internalized their malarkey.
When supporters of Gov. LePage attacked a recent Public Policy Polling survey showing their candidate behind, for instance, both Portland Press Herald reporter Steve Mistler and Bangor Daily News reporter Christopher Cousins uncritically repeated their false claim that the poll “oversampled Democrats” when compared to party registrations.
Mistler and Cousins are both veteran reporters and should know better. Even if they weren’t completely familiar with the methodological nuances, they should have remembered that party identification numbers from past Public Policy Polling surveys have been similar and that their results have, regardless, been among the most accurate surveys of Maine elections.
Or they could just think back to when the same claims were made last year about polling in the 2nd District congressional race.
HISTORY REPEATS ITSELF
Then, it was the campaign for Republican Kevin Raye that took issue with party identification numbers in two surveys from the Maine People’s Resource Center (for which I work) showing their candidate down 16 and 19 points against U.S. Rep. Mike Michaud, a Democrat.
Raye campaign consultant Kathie Summers-Grice called the results “preposterous.” On Election Day, Raye lost by a 17-point margin.
What reporters should be doing is ignoring the spin and downplaying individual poll results in favor of providing a broader context.
Eight public polls have been released in the last nine months in the race for Maine governor, and together they show a relatively stable contest, with Michaud consistently about 3 or 4 points ahead of LePage. Those numbers may not be as exciting as polls showing big swings in one direction or another, but they’re more likely to be true.
Mike Tipping blogs at MainePolitics.net and works for the Maine People’s Resource Center. He can be contacted at: