<$BlogRSDUrl$>

Tuesday, August 03, 2004

Polling

There's a reason you don't see much about polling in these semi-hallowed pages. There is something inexact about the whole thing that bothers me. As soon as a poll is released, whatever the parameters, frequently the candidates jump to spin the results based on how they fared. I think it's fair to say that national polls influence elections as much as measure their potential outcomes.

Finally I've found someone who shares some of my feelings:

”NO ONE is "ahead in electoral votes" because the current electoral vote stands at ZERO to ZERO. The worst part of this story is not people overinterpreting polls, or getting it wrong, or assigning states incorrectly, but in confusing polls with reality. Even if the supposed "margin of error" were zero, those margins are comparable to the scientific notion of "precision." They don't necessarily have ANYTHING to do with ACCURACY, and more than "precision" weapons are accurate when they precisely target the wrong house and kill innocent civilians. There are a million ways polls can be extremely precise, yet be completely inaccurate. For example, you could phone people only from 9-5, and exclude the entire population of people who work during the day from your poll. And that doesn't even count the usual attempt to differentiate between "likely voters" and "all registered voters" or even "all eligible voters," since there is still time to register between now and the election.”

| Permalink Here


This page is powered by Blogger. Isn't yours?

http://www.top-blogs.com/cgi-bin/rankem.cgi?id=ebradlee
Site Feed



















Site Meter

+ : nothing blogs : +

PhillyBlog
<< <5 | < | list | random | > | 5> >>

Listed on BlogShares

Performancing

Technorati Profile

Who Links Here?