If political polling has egg on its face after this month’s election, the Hamden-based Quinnipiac Poll has an omelet dripping from its visage.
The normally well-regarded Q Poll was so far off in so many places that President Trump singled it out for criticism. On Election eve, Quinnipiac had Biden up by 5 points in Florida. He lost by 3. In Ohio, Q put the race at 47-43 Biden. Trump won by 8 points. Its last national poll had Biden ahead by 11. As of this writing, he’s winning by slightly more than 3.
What happened? CTNewsJunkie asked longtime Quinnipiac Poll Director Douglas Schwartz. He declined to get into details, saying, “We learn with each election cycle and our experts will examine our polling methods and make any necessary adjustments in future years.” It could take as long as six months to figure out what went wrong, Schwartz said.
Of course, the Q Poll wasn’t the only one to fall short this year. Virtually every polling outfit had a bad night, leading West Hartford native Frank Luntz, perhaps the greatest pollster of the last 25 years, to say his industry is “done.” This comes after a major, but ironically smaller failure four years ago that sent chastened pollsters scurrying back to their drawing boards. Whatever they did clearly didn’t work.
I’m not not going to try to get into the weeds of polling since I know as much about statistics as I do about operating a nuclear reactor. I’ll leave it to Schwartz to give us an explanation at some point as to why the Q Poll missed the mark.
But it seems to me that this polling failure is yet another example of how our blind faith in technology is leading us astray. It’s the political equivalent of refusing to look out the car window even as the GPS keeps sending us in circles.
I’m not saying polls should be discarded. They are an important, if imperfect, tool in the politician’s and journalist’s toolbox. But modern polling, through its use of ever more powerful computers, ever bigger data sets and ever more sophisticated algorithms and statistical devices, has promised to make polls better. Instead they are getting worse. And people wonder why experts are increasingly held in disrepute by much of the American public.
Poll aggregating sites like Nate Silver’s FiveThirtyEight, it seems to me, add to the problem. Silver takes these already complicated and often problematic polls and subjects them to his own methodology — still more arcane statistical tools and algorithms. The result is polling averages derived from numerous surveys predicting the probability of an outcome. There’s a heads-I-win-tails-you-lose quality to this. Silver is never wrong because his predictions are only probabilities and if you criticize his numbers, he responds that you don’t understand probability: I told you there was a 5 or 10 percent chance that would happen, so why are you complaining?
Which raises the question, what good is any of this, especially if the underlying data may be flawed?
This approach also divorces elections from their core mission — voters choosing elected representatives based on what they say they will do — and turns them into a crap game. In the real world, candidates don’t win or lose elections because they rolled a 7 per the immutable laws of probability. They win or lose because of an often amorphous combination of policy proposals, partisan preference, image, likability, emotion, circumstance, name recognition, voter mood, instinct and just plain luck that no algorithm or statistical tool can ever fully capture.
I recall watching a documentary a number of years ago in which a camera crew followed then-Speaker of the House Newt Gingrich during the 1998 midterm elections. It was right after the Clinton impeachment, and Republicans were clearly in trouble. Gingrich is constantly meeting with local GOP officials. His first question to them wasn’t, “What do the polls say?” It was, “What are you hearing?”
You may or may not like Gingrich, but there is no denying that he is a masterful politician. He was doing what all gifted politicians do, which is talk to people and, most importantly, listen to them. It’s what Connecticut Democratic Chairman John M. Bailey, who dominated state politics from the early 1950s until his death in 1975, did. When Dick Lee, New Haven’s legendary mayor in the 1950s and 1960s, had free time, he’d go to wakes and funerals as a way to keep his finger on the pulse of the community. Bill Clinton, another superb politician, would talk to anyone anytime anywhere. And yes, Donald Trump has that quality as well. In 2016, he rightly sensed, polls and political experts be damned, that deep anger at elites, immigration, outsourcing and endless wars would propel him to victory.
Yes, polling is important. But it can never substitute for simple human contact and conversation. I’d advise everyone involved in politics, especially at the national level, to engage in more of both.
Christopher Hoffman lives in North Haven. He has written for the Hartford Courant and the New Haven Register, where he covered the statehouse, and he worked as a policy and communications advisor at the state Attorney General’s Office.
The views, opinions, positions, or strategies expressed by the author are theirs alone, and do not necessarily reflect the views, opinions, or positions of CTNewsJunkie.com.