Outlier? 56-26-3 - Updated x2
Sunday, Sep 17, 2006
About one in twenty polls are “wrong.” It’s the nature of the business. That’s why you can’t just consider one poll when looking at a race (it’s also why I set up the statewide polling page, so you could look at all recent polling).
Also, the smaller the sample size, the more likely it is that the poll is wrong. The latest Sun-Times/NBC5 poll of the governor’s race surveyed just 400 people. That’s embarrassingly small. It’s the usual sample size for a quickie state legislative poll (300 is the absolute bare minimum in those races), not a statewide survey.
But I wonder whether the new Sun-Times poll - which shows a gigantic 30-point lead by Gov. Blagojevich - will be put into perspective by the mainstream media, where things are looked at with a more simplistic bent (although Paul Green, to his credit, reportedly said on City Desk this morning that he didn’t believe the results). The resulting headlines are not gonna be good for Topinka and will only strengthen the “inevitability factor” for Gov. Blagojevich.
The CS-T will publish these numbers online soon, but they were released this morning on City Desk and sent to me by an alert reader.
Because nothing major has happened in the governor’s favor in the few days since either the Tribune poll or the Rasmussen poll were taken, which both had Blagojevich below 50, this looks a lot like an outlier. Also, compare this result to the last five polls and you can obviously see a difference. Plus, if the governor’s campaign had poll results even close to this, I imagine they’d release them.
Comments will be opened on Monday.
[Hat tip: wndycty}
*** UPDATE *** The Sun-Times story is now up. They make some defenses of the poll in the piece, including the fact that they’ve been using that dinky sample size in polls for eight years. Being close to right about Glenn Poshard in 1998, however, is still no excuse for cheaping out. As I recall, almost nobody believed that poll, either - for the same reason.
Also, the pollster apparently overweighted Democrats in the original sample and went back and weighted the results to achieve a 47-30 Dem-GOP partisan split. The result after the weighting was 51-30, Blagojevich over Topinka.
Comments are now open.
*** UPDATE 2 *** Carol Marin’s column yesterday warns readers not to get too excited about Topinka’s current predicament.
Take 1980 when, according to the polls, Cook County State’s Attorney Bernard Carey was going to clobber challenger Richard M. Daley by 20 points.