Let the record show that the Zogby Poll missed on its prediction of the California primary race by — and this could be a new world's record for polls that are generally well-respected — about six miles. Their last, pre-election projection was Obama with 49% and Clinton with 36%. The final total was Clinton with 52% and Obama with 42%.
How bad is that? My guess is that if I'd asked every reader of this site to just take a wild guess without doing any polling or research gathering or anything…if you'd just plucked numbers out of the "hunch" part of your brain…almost all of you would have been closer than Zogby. I would have guessed Obama to beat Clinton by a point or two.
Mr. Zogby has occasionally been accused of slanting his polls towards one candidate over another, either because he personally favors that candidate or because he's being paid by a news organization that does. Leaving aside fake polls that a campaign may commission just for its own p.r. value, I don't think any of the major pollsters ever intentionally give us anything but their best possible estimate. It's too embarrassing when they're wrong and, besides, one poll skewing towards a preferred candidate wouldn't have any impact. Zogby's last poll in California couldn't have looked much better for Obama, and Obama still finished below all expectations.
I think the polls still have a value, especially when viewed in consensus. The consensus this time was probably, generally right…although it's interesting that so much of the exit polling was way off, causing several TV networks to say X was likely to win when Y won, and to even actually call one state for the wrong candidate. I don't know if it's this election in particular or if something about our national approach to voting is changing…but something's amiss.