Do National Polls Mean Anything?

"National polls don't mean anything at this point," as we all know. Chris Bowers, however, notes that they may not mean anything at all:

By sampling between 40-50% of all registered voters, the vast majority of national Democratic primary polls are not specifically sampling the Democratic primary electorate. As such, these polls should not only be taken with a grain of salt, but should almost be dismissed entirely as useful indicators of the current state of play in the Democratic primary / caucus season. If they are not polling the Democratic primary / caucus electorate, or even coming close to doing so, then they should not be used as indicators of opinion in the Democratic primary / caucus electorate. To use an analogous example, one does not poll the entire nation as a means of determining public opinion in California. One does not poll all Democratic self-identifiers and leaners in order to determine the opinion of Democratic primary and caucus goers.

This seems correct. We know from history that the universe of Democratic primary voters is different from the universe of Democratic self-identifiers. The problem, from the pollsters' perspective, is that with the primary calendar constantly shifting around, it's far from clear that beyond the traditional early primary states one would have anything on which to base a turnout model. Who's likely to go to the Nevada caucuses? Who knows? Of course, from the media's point of view this is all irrelevant since accurate forecasting of elections and accurate portrayal of public opinion aren't things media outlets are interested in. Shocking, yes. But they're trying to attract attention -- i.e., readers and viewers and ad dollars -- while minimizing costs. Actual candidates, however, do have an interest in accurate information and it would be fascinating to know how they handle this.

Comments

Ignore the polls. If you need a quant fix go directly to Intrade.com. You can 1) see what the market thinks about these guys and/or 2) put your money down. Check out the Fred Thompson chart on the day he announced his lymphoma; he droped faster than the news got to you I'll bet. My recommended trade is is long Obama, Edwards, Richardson and short Clinton.

Posted by Drew | April 14, 2007 5:21 PM

We should be able to study this effect. At the very least, we should be able to examine opinion polls taken in the time frame right before the start of primary season and compare that to how people actually did. Last time around, I remember that Howard Dean did indeed start to drop sharply in the polls in the days before his poor finish in Iowa.

Posted by KevinA | April 14, 2007 5:33 PM

I think this primary season will be the most interesting and unpredictable one in decades. Last time around, Howard Dean had all the big mo' going into Iowa and New Hampshire, and it really was a big surprise when Democrats turned out to favor Kerry instead. On the other hand, Dean was probably on the front of a wave, maybe too far in front, and he did end up heading the party...and as you say, the states are all changing dates so fast...

So everything's in flux! My prediction: many seemingly well-founded predictions will be proven false. This election season looks like Brownian motion in both major parties.

Posted by Bob Ellison | April 14, 2007 5:59 PM

on electronic markets - Drew, you point out the market's reaction to Thompson's lymphoma announcement as a demonstration that the market is rationally updating expected probabilities of outcomes - in short, "learning." But is it clear that the bettors made the right interpretation of the evidence?

Thompson announced that his lymphoma was in remission. He's getting all the information about his health history out there so he won't be attacked for withholding crucial details when the time comes. If anything, revelation of a past bout with cancer with a decent 5-yr survival rate is evidence that Thompson is more, not less, likely to run. For unannounced candidates, like Thompson or Gore, this kind of revelation should increase, not decrease, their likelihood of winning the nomination, since their participation in the race is as big an unknown as the effect of competition from other candidates. The online gamblers are not putting the right interpretation on the evidence - the wisdom of crowds here is likely wrong.

Still, I agree with the larger point about the uselessness of national polling.

Posted by David | April 14, 2007 6:04 PM

What do mean something are well-designed polls in New Hampshire and Iowa.

Edwards, IIRC, is running first in both states. Hillary is running third in Iowa and close to it in NH.

It may well turn out going first has even more impact than before on the remainder of the race. The winners in NH and Iowa will get a boost money can't buy.

Maybe there's money to be made betting that the crowd at Intrade isn't so wise after all.

Posted by Solomon | April 14, 2007 7:31 PM

Not only that, but the universe of self-identified Democrats doesn't necessarily match the universe of registered Democrats. And in several states, voters don't even register their party affiliation.

And turnout by district or by state is also affected by whether or not there are contested House or Senate primaries, which affects how much money is spent in those districts or states.

Posted by Michael Ditto | April 14, 2007 8:57 PM

Not sure about Bowers claim that polls are not measuring Dem primary voters.

http://www.cbsnews.com/stories/2007/04/13/opinion/polls/main2681317.shtml

Above CBS polls says Dem primary voters.

Posted by DonB | April 15, 2007 12:07 AM

"Thompson announced that his lymphoma was in remission."

Thompson's health will be an issue. Remember Tsongas? He had lymphoma. In 1992 when he was running he was "cured".

Posted by DonB | April 15, 2007 12:19 AM

http://www.latimes.com/news/nationworld/politics/la-na-thompson12apr12,1,4093512.story?coll=la-news-politics-national

This article confirms that the Tsongas and Thompson lymphoma cases are not comparable. Plus, Tsongas was diagnosed and treated 24 years ago, a pretty significant gap in cancer research time. Kerry's prostate cancer was never an issue in his campaign. In fact, his openness with his medical situation in 2003 did much to insulate him from that kind of attack. Elizabeth Edwards' cancer has already made Katie Couric look awful for questioning their campaign. Cancer survival is looking more and more like an asset to modern campaigns.

Posted by David | April 15, 2007 11:26 AM