Our class visited and tweeted about the law school’s latest poll Thursday. Of course, the poll got plenty of attention, due to how close it is to the election. There were several media outlets there, and the room was quite full.
The biggest headline was, naturally, the presidential results. The poll showed Obama leading Romney 51–43 among likely voters, which beats the margin of error. But the senate race is a statistical tie. Baldwin leads Thompson by only 4 points, 47–43.
A recent visitor to our class, blogger Craig Gilbert, warned of overanlyzing poll results. Gilbert, who works for the Journal-Sentinel and blogs at the Wisconsin Voter blog, noted that there’s still a one in 20 chance that the data is off by more than the margin of error. This is even ignoring the possible mistakes the pollster could have made.
That meant that a single poll is most valuable in the context of the most recent polls. A poll showing a bigger lead than previous polls could either be the result of statistical noise or a pollster’s mistake. It could also indicate an actual change.
So what did the flurry of polls after Marquette’s show?
We can see that Obama’s edge in the Marquette Law poll has continued. However, the average lead is smaller than what Marquette gave him. Whether this is due to a minuscule gain by Romney or statistical noise is hard to tell. Some national measures of the race — including Rasmussen’s tracking poll — show a slight increase in Obama’s lead over the past few days.
Polls agree that Obama is more popular among Wisconsin’s likely voters, whether the lead has grown remains ambiguous. Unless you have reason to think polls are systematically wrong in the president’s favor, that translates to an actual lead. As fun as it is to muse, we’ll only know in about 72 hours — when the results come in after Wisconsin’s polling places close at 8 p.m. on election night.
More to read
—Gilbert didn’t post about the Marquette Law School poll itself, but looked at the past 12 months of Marquette’s polling.