As long as there have been opinion polls, politicians have been saying something along the lines of “the only poll that matters is on Election Day.”
If you’ve ever doubted the truth of this, Tuesday’s election should have cured you.
Going into Election Day, virtually all the polls showed Hillary Clinton with a small but persistent lead that would send her to a comfortable victory in the Electoral College. Even in the hours before the polls closed, national pollsters and pundits were giving Donald Trump perhaps a 30 percent chance of succeeding.
In the end, of course, Trump nearly ran the table in the key battleground states where Clinton was presumed to be holding onto a lead, sometimes a comfortable lead.
There has been a lot of discussion in the days since about why the polls were so wrong. Perhaps it is that the rise of the cell phone has made polling harder. Maybe it was that there were lots of pro-Trump voters who were, for whatever reason, reluctant to admit their preferences publicly.
It could be too that the polls were not particularly wrong, but the analysts were reading them wrong. Perhaps the chattering class and Washington press corps had simply decided that there was no way that Trump could win, a thing they had largely assumed from the moment he started signaling his intent to run. With that attitude, it would be easy to see the polls as more decisively in favor of Clinton than they were, discounting Trump’s underdog – but by no means improbable – path to victory.
Maybe it was a combination of all of the above. In any event, it seems even Trump loyalists were astounded by what happened Tuesday night.
It wasn’t the only example of polling that was misleading this year. Locally, I hear of some internal polling by a candidate showing a very tight race but when the votes were counted, it was a blowout.
I have long been suspicious of polls for very practical reasons. I have written more horse-race kind of stories over the years – Candidate A is XX percent ahead of Candidate B in the most recent poll – than I can comfortably count. They fill space, they feed the editors’ voracious appetite for stories, and they are studied by political junkies with remarkable zeal. But the truth is, even 20 or 25 years ago, when nobody had a cell phone and polling was king – the numbers didn’t really tell you very much about what would happen on Election Day.
Here’s a good example.
In the four weeks leading up to the Feb. 1, 2000 primary in New Hampshire, there were 10 major public polls released. They generally painted a picture of Sen. John McCain gaining traction against Texas Gov. George W. Bush, but they were all over the place in saying how much. They showed McCain up by as little as 3 percent (with a 5 percent margin of error) to 11 percent (with a 4 percent margin of error).
The American Research Group even came up with a poll showing Bush with a small but solid lead – up 5 points with a 4 point margin of error.
News organizations all over the country cranked out stories about McCain’s lead and the Bush campaign’s confident moves to bring the race back under control.
In the end? The polls weren’t even close.
McCain delivered a historic 18 point shellacking that nearly derailed what most had assumed was a coronation for George W. Bush in the Republican primaries.
Even McCain appeared stunned by the margin as he delivered his giddy victory speech in New Hampshire that night.
Observing that election at close hand has made me wary of placing too much stock in polls. I’ll admit to being surprised Tuesday night as Trump began to run away with the race, but in the weeks leading up to the vote, I had been cautious about making predictions when people asked things like “He can’t win, can he?”
Because the cliché really is true – the only poll that counts is the one on Election Day.