Democratic political consultant Dan Payne, writing in today’s Globe about Hillary Clinton’s victory in Massachusetts, says, “Once again, pollsters failed to render an accurate snapshot of the race, missing a 56-to-41-percent landslide, making prognosticators like me look bad. This has got to stop or there will be blood.”
Really? The final WBZ-TV/SurveyUSA poll of registered voters, taken on Saturday and Sunday, had Clinton over Barack Obama by a margin of 56 percent to 39 percent. Yes, the WHDH-TV/Suffolk University poll had Obama ahead by two. But SurveyUSA called it almost perfectly.
Discover more from Media Nation
Subscribe to get the latest posts sent to your email.
This is the money graf: The WHDH-TV poll said “27 percent of Democratic voters and 24 percent of Republican voters say they may change their minds before tomorrow.”That, and not the 46-44% lead their poll showed Obama as having, is the story. And the question is why their poll showed such a high percentage of undecided compared to the other polls.
Josh Marshall notes similar success for SurveyUSA in California (who had Clinton up 10 – right) as opposed to Zogby (Obama up 13 – wrong). He also mentions Missouri, which Zogby (Obama by 3) had correct, while SurveyUSA (Clinton by 11) was way off.Perhaps Clinton-Obama is more highly demographically driven than other elections? The trick would be to predict who will turn out (harder) rather than how the various demographics will vote (easier). The pollsters don’t always seem to have that down.But SurveyUSA seems to be biased in some way to Clinton, and Zogby to Obama.
I don’t understand the obsession with the polls. More often than not they are wrong or biased.
Anon 11:34, not sure why I’m bothering to respond, except that when accepted survey design practices are followed and the margin of error accounted for, polls are generally right.
Mike – “polls are generally right”. Well, so far this year I’m not sure that’s correct, hence this thread. I don’t know of any polling organization that has a flawless track record this year.There’s something unusual happening in these primaries that are making them extremely difficult to poll well. You’re having different polling organizations polling the same race at the same time and giving wildly different results. It must be driving them nuts, and it’s driving their credibility down.
Steve: I haven’t done a study, but it strikes me that the pollsters who have showed wild, last-minute swings, especially in Obama’s favor, have turned out to be the most wrong. Those whose findings show that Clinton has held on to leads that she’d had for months have generally turned out to be right. See: N.H., Mass.
Steve, as I’ve noted in past strings, the NH vote for instance, once you account for the huge amount of undecideds, the polls have been right. You simply can’t blame a poll for being wrong when 20% or so of the electorate remains undecided 24 hours before the vote.