They're calling it awfully quick...

February 12, 2008 4:31 PM

How are news organizations able to project winners with 0% reporting?

Many news organizations have called Virginia for Obama, minutes after polls closing. Are they relying entirely on exit polls? Have exit polls become that much more accurate since "Dewey Defeats Truman"?

posted by waraw to media & arts (10 comments total)

"Dewey Beats Truman" was SIXTY years ago. Of course, they're more reliable. And the mistakes in that projection have been long ago explained and corrected for.

International Election Observers confidently use exit polls to determine if election fraud is occurring everywhere else in the world. The Exit Pollsters working for the Media in the USofA reached a very high level of accuracy more than 20 years ago, if I recall (when Bush Sr. beat Dukakis). Yet, in more recent years, there have been more discrepancies between the exit polls and the "official" vote count, and the media make the assumption that their pollsters goofed and have never done any serious investigation on the discrepancies, motivating some of us to pull our tinfoil hats down over our ears...
posted by wendell at 4:42 PM on February 12

The media have been cautious about exit polls since 2000. Virginia's must have been so overwhelmingly for Obama that they felt comfortable using it. (IANA exit pollster, but I do work for a large newspaper.)
posted by Airhen at 4:53 PM on February 12

And they still blow it all the time. Just a few weeks ago the AP called Missouri for Clinton(the networks did not) and she subsequently lost. let's not even talk about 2000 and Florida. Really, lets not. Exit polling remains very unreliable and television news has gotten only marginally better about not relying on it.
posted by The Bellman at 4:54 PM on February 12 [1 favorite]

How does CNN make primary projections?
posted by milkrate at 5:00 PM on February 12

"Dewey Beats Truman" was SIXTY years ago. Of course, they're more reliable.

So some news organizations claiming The Bradley Effect.

Also, Dewey Vs. Truman was due to a flawed sample based on phone ownership. We're having similar problems with phone-based polls this go-round, because younger and more tech-savvy voters are more likely to have a mobile-only connectivity which is underrepresented in numbers pulled by pollsters. (Not exit polls, but phone polls) This means more inaccuracy until we change sampling methods.

Which doesn't even get to the fact that low-income voters, people for whom English may be difficult and anyone who's not old are bored are often unwilling to participate in polls - on the phone or coming out of the voting booth.

Taken together, gives me a good reason to think that our polls are just as unreliable as our grandparents polls.
posted by Gucky at 6:27 PM on February 12 [1 favorite]

Gucky, what evidence do you have that pollsters as a whole are unreliable? I've heard the "mobile phones" argument again and again, most recently with Ron Paul supporters arguing how he was going to sweep the elections and wasn't on the radar for the pollsters, etc, his supporters are too hip to own landlands. But this argument never seems to pan out. Any cites you have of this would be useful. It seems that polls do a pretty good job of predicting winners and usually do so within their margin of victory.
posted by Happydaz at 7:45 PM on February 12

Exit polls: What you should know 2006
From the blogger fka Mystery Pollster.

Also, Ruy Teixeira on why exit polls can be "off".

In any event, none of the races today was close for either Obama or McCain, except maybe Virginia for the latter. The Obama final numbers show a nearly 2:1 advantage over Clinton for MD and VA. That's way safe enough for anybody to call. It's when you get races that are 35%-34% or 47.9%-48.4% that you have to be really, really careful.
posted by dhartung at 2:02 AM on February 13

Dont forget the 1992 UK General Election. Opinion and exit polls had predicted a hung parliament, yet the Tories held onto power. The wrong predictions were blamed on Shy Tory Factor. There has been some speculation that this might also apply to Republicans and/or Hillary voters, but I'm not sure how credible that speculation is.
posted by Jakey at 2:24 AM on February 13

They do exit polling. Much like any pre-election poll, an exit poll aims to find a truly representative sampling of a total pool of voters. Those polled people offer opinions and networks base projections solely on those answers. They go on the air with them. Networks have been much more cautious with them since Florida in 2000.
posted by greenchile at 9:44 AM on February 13

Apparently I wasn't the only one to notice the early call: it was mentioned on both The Daily Show and The Colbert Report.
posted by waraw at 4:45 AM on February 14