2008 record      


Polling: Among lessons drawn from this election, add one you'd think had been learned by now: Consider a poll or pundit's record before starting to quote it or him ad nauseam.

'Shellshocked." That's how one Romney adviser reportedly described his boss as the bad news started sink in on election night. Why, the crowds had been so large and enthusiastic in swing states! And besides, "poll after poll" showed Republicans more motivated to vote than Democrats!

And indeed, many a poll had Romney ahead, with some — including the most famous of all — showing him running away with it. With just week to go, Gallup had the governor up by a whopping 6 points.

We won't have the official results until next month, after the states are finished with their final tallies. As of Monday morning, the popular vote was Obama 50.6%, Romney 47.9% — a margin of 2.7 points. But that could change significantly.

Four years ago, the popular vote on the morning after the election had Obama beating John McCain by 6.0 points. That put the Rasmussen poll closest at that point, and Rasmussen has been claiming ever since it was most accurate pollster in that election.

The margin widened steadily over the next six weeks, however, and the final results had Obama up by 7.2 points — 52.9% to 45.7%. That was exactly what the IBD/TIPP Poll projected.

So for the second presidential election in a row (we had Bush beating Kerry by 2.1 points in 2004 vs. the 2.5 actual), IBD/TIPP was No. 1 among national polls.

This year, IBD/TIPP called it 50.3% for Obama (vs. Monday's unofficial 50.6%) to 48.7% for Romney (vs. 47.9%). In other words, our spread was 1.6 points vs. a real spread as of Monday of 2.7 — thereby an error of 1.1.

As the table above indicates, 1.1 will probably put us third this year instead of first. But among the 11 polls active in the last three elections, IBD/TIPP has the best record.

For a little different perspective, widely followed New York Times blogger Nate Silver rates pollsters' results based on performance over the last three weeks of the campaign, rather than the last poll alone. His ranking of 23 polling organizations, published Monday, had IBD/TIPP at the top.

Our record didn't cut much ice, however, with the many poll critics who came out of the woodwork this cycle. They couldn't get over the fact that our registered voter sample assumed Democrats outnumbered Republicans by 7 percentage points, or 38% to 31%, before applying our "likely voter" screen. (We had independents at 32%.)

They claimed our party breakdowns by old (2008) data, and that Republicans had narrowed the margin considerably since then.

Conservatives were especially critical of our supposedly "skewed" data, with some sure we were, as Breitbart.com concluded, "in the bag for Obama." But we stuck to our breakdown because that's what the numbers said. Our pollster, TechnoMetrica Market Intelligence, wasn't about to tailor its assumptions to anyone's political preference.

As it turned out, exit polls showed the party affiliation of 2012 voters is 38% Democrats, 32% Republicans and 29% independents — a 6-point Democrat advantage.

Rasmussen, who had Romney winning by a point, believed the electorate has more Republicans than Democrats. The hallmark of any scientific endeavor is repeatability. We're not sure if any organization has been able to repeat that finding. But that didn't keep conservative commentators from quoting Rasmussen every chance they got.

As for Gallup, which also had Romney winning by a point after having him up as much as 6, the problem was grossly overestimating the non-Hispanic white vote at 78% and underestimating the non-white at 22%. The exit poll had the ratio 72%-28%. IBD/TIPP believed the ratio was 74% to 26%.

Still, it's the Gallups and Rasmussens that pundits keep quoting. And it's why the pundits, from Karl Rove to Dick Morris to George Will to Michael Barone, most of whom had Romney winning handily, were so wrong.

Fact is, no other poll has done better than IBD/TIPP in the last three presidential elections. We want to congratulate our partner, Raghavan Mayur and the rest of his staff at Ramsey, N.J.-based TechnoMetrica, for retaining their title as "America's Most Accurate Pollster."

Submit to DeliciousSubmit to DiggSubmit to FacebookSubmit to Google PlusSubmit to StumbleuponSubmit to TechnoratiSubmit to TwitterSubmit to LinkedIn