Polls come in all sizes and intents. It's one of the reasons that public opinion surveys conducted at nearly the exact time come up with very different results.
The wide-ranging nature of polls was on display in the Nebraska U.S. Senate race between Republican Deb Fischer and Democrat Bob Kerrey. One of the most widely discussed polls in the race was sponsored by The World-Herald.
That poll found that Kerrey had narrowed the gap from 16 percentage points in a mid-September survey among likely voters to 3 points two weeks before the election.
In the end, though, Fischer cruised to victory by 16 percentage points.
Why the disparity?
Aaron Trost, Fischer campaign manager, said the campaign's own tracking polls found volatility in the electorate, particularly among independents, two weeks out.
The campaign released its own poll claiming a 16-point advantage for Fischer in the same time period as the newspaper poll. But Trost said his educated guess is that Fischer probably was no more than 10 points ahead at that time.
A Kerrey-sponsored poll released about the same time put the race at a 5-point Fischer advantage. There were reports of other third-party polls that also had the race in single digits.
What helped boost Fischer's confidence, Trost said, was that her campaign survey showed much more intensity among Republicans and Republican-leaning independents than among Democrats or other independents.
Also, the Fischer campaign and Republican groups knew they would focus their advertising in the final week on tying Kerrey to President Barack Obama. They were confident that would hurt the Democrat, given the high negative ratings for Obama in Nebraska.
A key difference between The World-Herald Poll and the Fischer tracking poll was the assumption on the makeup of the electorate.
The World-Herald used voter registration lists, with the sample weighted to match the makeup of registered voters: 48 percent Republican, 32 percent Democrat and 20 percent independent.
The Fischer camp assumed that a higher proportion of Republicans would vote, as they had in the previous two presidential elections, and that no more than 15 percent of the electorate would be independents.
That key difference illustrates how campaign polls attempt to create a model that predicts voter turnout, while most newspaper and other public polls simply attempt to measure public opinion at a moment in time.
“There is a whole different level of sophistication involved if you do that, a voter turnout model,” said Tom Wiese, president of Wiese Research Associates of Omaha, which conducted the survey for The World-Herald.
Wiese said the margins of error also have to be considered. In the case of The World-Herald Poll in October — with a random sample of 800 registered voters — the margin of error was plus or minus 3.5 percentage points. That could mean that the results could move 7 points in either direction.
Also, as with all scientific polls, there is a 5 percent chance that the results of any survey are simply wrong.
What is unusual in the case of the second World-Herald Poll was that its other results, including an expanding victory margin for Mitt Romney in Nebraska and the tightening of the 2nd District U.S. House race, signaled trends that were borne out by the election results.
Contact the writer: