How polling poisoned the political well
posted at 11:01 am on November 8, 2015 by Jazz Shaw
Has political polling reached the end of the line? While some won’t find it as terribly shocking anymore, the polls in Kentucky which had previously been considered among the most reliable missed the Bevin win in the governor’s race by a wide margin. This wasn’t the first time either. The last few cycles down there saw the Bluegrass Poll taking one hit after another and they responded this week with something of an apology.
The wide gulf between the results of Tuesday’s election races and the most recent Bluegrass Poll is a source of frustration for the Herald-Leader and its media partners as well as for our pollster — Survey USA — especially after a similarly huge difference in last fall’s U.S. Senate race.
As a result, we are rethinking our approach to election research and plan to make changes for future campaigns. Depending on how we proceed, we will look for a new research firm and no longer will use Survey USA.
The Herald-Leader and Survey USA aren’t likely the problem here. Nobody else who was regularly polling the race did very well either, and that’s not some sort of rare exception. The polling system seems to be getting more and more erratic. And yet it remains the gospel for political reporting in both the old and new media. Jonah Goldberg has an op-ed this weekend which is definitely worth a look and he’s not just questioning the methodology of a couple of outlets. Ending reliance on polls could give politics back to people.
But the thing I find most intriguing about Bevin’s victory is that his opponent, Jack Conway, had led in all the polls for pretty much the entire race.
“What’s ironic,” writes National Journal’s Josh Kraushaar, “is that coverage of political campaigns is increasingly dependent on the polls, even as the polls themselves are increasingly flawed. The tenor of the presidential primary is being dictated by the daily stream of horse race polling. Credible candidates are being left off the main debate stage because of miniscule differences with some opponents in national surveys.”
It’s a national, and international, trend. The polls underestimated the scope of the 2014 midterm elections. Elections and referenda in Greece, Poland, Britain, Israel and Scotland embarrassed the pollsters who thought they knew what was going on.
Jonah goes on to quote Karlyn Bowman of AEI who speculates that this could be “the end of polling as we know it.”
That may prove to be a bit hyperbolic, but would it really be so bad? Jonah makes a valuable point in noting the effect that national polling (which has fluctuated wildly from one survey to the next over the fall thus far) has had on the debate process and, secondarily, on the election as a whole. Putting Trump and Carson on the main stage is a no brainer, sure, but after you get through three or four more ambiguously defined second tier candidates you find yourself in a swamp of people who are all polling within the margin of error of each other. The difference between who is put on the “main stage” and who goes to the warm-up game (or who is left out entirely as will happen this week) is minuscule in terms of numbers of respondents, yet could have a major impact on their chances.
But as I’ve fretted over before, the effects of such massive reliance upon and distribution of polling probably run a lot deeper than that. Candidates who are getting consistently low poll numbers tend to take those data samples and turn them into a self fulfilling prophecy. Voters are less inclined to support (or donate to) somebody they perceive as a loser while the bandwagon effect no doubt helps those who are riding high in the surveys. If there was still some element of mystery as to where the electorate was heading on any particular day, a particularly good speech or breakout moment in a debate could take somebody from worst to first, or at least give them a better fighting chance at it.
Voter turnout is also historically affected by polls. When the media is telling you day after day that your candidate has already lost the election, why bother to get up and go vote when you could stay home and mow the lawn? Polling is a great tool for those of us who talk or write about politics for a living because it gives you something to analyze and virtually limitless material for speculation and interpretation. But are we feeding a democratic death spiral by relying on them so heavily?
I’ll close with one more comment from Jonah.
You can argue that following the polls is democratic, but it’s a cheap and shallow form of democracy. We are also a republic, and in republics leaders are expected to do what they think is right, not just popular.
Toppling the tyranny of polls would put arguments back at the center of politics. And that’s as it should be.
It’s that and much, much more. This ties into what is referred to in science as The Observer Effect, which states that it’s nearly impossible to measure something without changing the thing being measured. The advent of modern polling seems to be crippling the process it was intended to analyze.