« Housekeeping | Main | Likely Voters VII: CBS/NYT »

October 28, 2004

Answering Erikson

The hardest part about blogging the last few weeks is that I get less time to simply read and surf than I used to. So you may already know about the following. .

First, I discovered that Philip Meyer, a Knight Professor of Journalism at the University of North Carolina at Chapel Hill, wrote a column in USAToday on Erickson's POQ article and the Gallup likely voter model. Money quote:

A likely-voter poll is the right thing to do if all you want is to predict the outcome of the election - but that's a nonsensical goal weeks before the event. Campaigns change things. That's why we have them.

It would be far more useful to democracy if polls were used to see how the candidates' messages were playing to a constant group, such as registered voters or the voting-age population. Whoever is elected will, after all, represent all of us.
Read it all.

Second, another must-read for the polling obsessed that those outside the Washington Beltway may have missed: Washington Post Polling Director Richard Morin's lengthy exposition on how tough it is to be a pollster these days. It covers all the topics we've been chewing over and some we haven't gotten to yet: cell phones, reponse rates, Internet polling, etc. If you enjoy this blog, you'll find much of interest.

Related Entries - Likely Voters

Posted by Mark Blumenthal on October 28, 2004 at 06:17 PM in Likely Voters | Permalink


Is anyone here even slightly concerned about electronic voting? I mean the way its looking, a few extra votes couldn't hurt some people...

Posted by: Mapreader | Oct 28, 2004 8:58:14 PM

"It would be far more useful to democracy if polls were used to see how the candidates' messages were playing to a constant group, such as registered voters or the voting-age population."

If "registered voters" are "a constant group", what are registration drives for? And what are laws against felons voting for? And so on.

Is "a constant group" some strange term of art? Or what?

Posted by: David Blue | Oct 29, 2004 6:47:18 AM

FYI, it looks like MysteryPollster accidentally duplicated the second URL into his third link in this post; I suspect the real article he was trying to link to was http://www.washingtonpost.com/wp-dyn/articles/A3735-2004Oct27.html

Posted by: Sweth Chandramouli | Oct 29, 2004 9:27:01 AM

Mapreader: As it stands today, I worry much more about voting supression than crooked electronic voting machines. If you can't get people to vote, It won't matter very much if the electronic voting machines are accurate or not.

Posted by: ClaudeB | Oct 29, 2004 9:33:57 AM

Perhaps I am running ahead of things, but talking about how the changing political landscape is making polling more an more prone to pitfalls, has anyone put some thought on how the early-vote rush may alter exit-polling on election day? Could we see a series of consistently erroneous exit polls early on on Tuesday night, with States (and possibly the election), being called early by the media for one or the other candidate, and retracted later based on more complete projection data? Is anyone keeping track on who these early birds are, and which candidate they support?

Posted by: slo-mo-joe | Oct 29, 2004 11:00:15 AM

Slo-mo-joe hits a HR!

With all the concern about the possibility that cell phone only voters are fundamentally different that those landline owers of the same demographic age who are polled, I am refreshed to see a very legitimate issue aroused.

Does anyone have any early indications for these voters?

Posted by: Eric | Oct 29, 2004 1:15:11 PM

Aren't the early voter results tabulated as they're cast? (That at least is the implication on the Volusia county website -- google volusia and "early voting") The delay on overseas/absentee reporting is b/c they only have to be postmarked before election day, and thus some can take a while to arrive and be counted.

The early-vote data could conceivably be released as soon as the polling places close, so we start from a baseline of, say, 20% of the vote rather than 1% of precincts. If the precinct location is included in the count data then you'd just throw the early counts into the predictive extrapolation model.

Hmm. Assuming the computer models will accept partial data. Slo-mo-joe may be right after all.

Posted by: stanley | Oct 29, 2004 3:33:30 PM

Erikson's comments made sense to me, for about five seconds.

Then I thought "you know, if you don't vote, you don't matter. There's no reason why a democracy should care about the opinions of people too lazy or uninterested to vote."

So I think the focus on likely voters is a feature, not a bug.

Posted by: Greg D | Oct 29, 2004 6:42:09 PM

OK, let me rephrase this. I understand that early votes are tabulated as they are cast, but we don't know what population subset they represent. For instance, I read that 30% of Floridians have already voted, and Kerry is leading in that group by >10%. Clearly, early voters seem to be more pro-Kerry than the population of Florida as a whole (based on opinion polls). But by how much?
If in tomorrow night's poll projections these votes are dumped whole in the projection model, they will skew the results pro-Kerry. If they are excluded, and only exit polls (or early projections from tomorrow's votes) are relied on, the result will be to overestimate the Bush vote. Both effects could be really big, and the more skewed the early vote, the more the error in the estimate will be.

Basically, I don't think we should trust exit polls or early projections in any state with more than 10% early votes (possibly even less, if the race is tight). It's gonna be a long night.

Posted by: slo-mo-joe | Nov 1, 2004 1:27:47 PM

The comments to this entry are closed.