My NationalJournal.com column, which looks at how the leaked and at-poll-closing exit polls compare to the actual results, is now online.
Excellent work. Two thoughts come to mind:
1) Do you have crosstabs for the early and final exit polls? Because I'd like to know how different they are. My hunch is that people who are more "into it" vote right away, which would bias Obama, but you'd have to see the skewed demos for that to even seem plausible.
2) Anytime a poll overestimates Obama, as it did in Ohio, you know we're all gonna shout "Bradley Effect!" rather loudly. The beauty of the Bradley Effect is that it's impossible to prove or disprove, but those of us who are into it will insist that it's very real. Any thoughts?
Thanks again. Good stuff, especially about being patient for good data. And doing volunteer work while you wait. :-)
Posted on March 7, 2008 4:31 PM
Back in a former life I used to be a senior economist with BEA. And we had a saying that pollsters should also heed: Accuracy must be balanced by timeliness.
The key word there is balanced. Right now, there is no doubt that in the rush to be first, not just on the TV news but in the blogosphere, accuracy takes a back seat. But this is not inherently a negative thing. In an objective sense, I agree Crowley's comment that a "few hours wait" is worth it for better data. But I also recognize the fun on being caught up in the moment as the data is coming out. There is an "entertainment value" that should not be casually dismissed.
As I have harped on before, the real issue is one of education. The difficulty is that most people don't understand what is good data or not and they want to rely on an "expert" to tell them. Rather than rushing to be accurate and first, it might be better to take a step back and educate people to be better "data consumers". This way, whoever is first, consumer have the ability to evaluate the data and respond as they see fit.
Posted on March 7, 2008 10:35 PM
I still think that the exit polls oversample the city voters, because it�s easier to reach the people there.
But in the cities Obama�s support is stronger than in rural areas, so the exit polls are wrong, because their sampling is wrong- not the race, or gender part, but the location of the respondants.
Posted on March 8, 2008 6:46 AM
Rasmus, in principle that is just about untenable, because the exit pollsters (1) have a lot of time to set up their sample so that they cover all sorts of places proportionally, and (2) can adjust their estimates to correspond with turnout estimates, beginning even before the polls close.
In the 2004 general election, the national tab seems to have ended up with too many big-city respondents. Based on the archived data, I don't think this had to do with the number of people being interviewed; I think some garbage info sneaked into their turnout estimates for some big-city precincts. (Or it may have happened as a side effect of some other weighting.)
Posted on March 8, 2008 8:46 AM
I agree that there might be oversampling in cities, Obama has consistantly won the cities. But lack of cooperation among older voters with younger pollsters? Come on.
I don't think you can just dismiss the problems our country has with voting fraud. The New Hampshire recount investigation found some major problems with the chain of custody of the ballots. After all, Yushchenko of the Uraine would not be in office, if it weren't for exit polls! Go Orange Revolution!
Posted on March 8, 2008 5:20 PM
Rather than mindlessly speculating, I actually was an exit pollster for a precinct in California for the February 5th election. The morning samples are horribly biased because everyone is running to work or school and so you are unlikely to get more than every third or fourth person to stop. I was told to exit poll EVERY SINGLE PERSON COMING OUT of the polling place, which can be a problem when you have groups of people. At least I was able to set up shop right outside the door of the fire station where the polling place was, as technically you are supposed to stand 25 feet from the entrance, and in other states they don't even do that.
Posted on March 16, 2008 5:01 AM
Comments: (you may use HTML tags for style)
Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.
Please email us to report offensive comments.
See our comment policy here. Note that we require commenters to share their email address via Typekey. We will never share your email address with anyone without your explicit permission.
MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR