Polling Still Isn't An Exact Science

I've been reading some things on the Wall Street Journal's web site, and an item by Daniel Henninger caught my eye.

The headline reads "Thomas E. Obama" (as in "Thomas E. Dewey," the man who beat Harry Truman in the 1948 election -- according to the headline in the Chicago Tribune).

Of the outcome of the New Hampshire primary, compared to the polls completed just before the voting started, Henninger writes, "Polls before the vote had Barack Obama leaving a tear-stained Hillary Clinton in the dust, some showing a lead of 12 points. She won by three, so on paper a 15-point error, a statistical fluke of startling proportions."

Frankly, it's bigger than "startling."

In 1948, they had a reasonable excuse. Polling wasn't as scientific then as it is today, and pollsters more or less stopped polling in the presidential race several days before the election.

Pollsters were getting the impression that the election was going to be a landslide for Dewey. They wanted to focus on races that weren't regarded as foregone conclusions, and they weren't aware of the concept of "tracking polls;" if they had been, pollsters would have seen the movement to Truman occurring in the final days of the campaign.

Henninger says the internet is at least partly responsible for what happened in New Hampshire -- with its 24/7 pressure.

"Voter behavior in the new age remains a mystery yet to be explained," he writes. "A new conventional folly is forming that Hillary achieved this entire reversal because, for about 2.7 seconds, her tear ducts opened. Therefore women voted for her. Who knew politics was so easy?"

I saw the same polls Henninger did. They were conducted right up to the last day, and they all indicated that Obama was going to record a big victory. And I saw the video of Hillary's teary-eyed moment. And I saw the actual primary returns from New Hampshire Tuesday night.

I can't explain what happened any more satisfactorily than he did.

But I will say this.

As tempting as it is to believe that polls possess some great knowledge and insight that ordinary mortals don't, the only poll that still matters is the one on Election Day.

Sometimes, voters make their decision at the last minute.

And it's a cliche to say this, but it's true. Timing is everything.