> All the technical problems we've been talking about seem not to have
> reduced their predictive power concerning elections.
But aren't the binomial distributions that govern election sampling far more robust than the standard distributions that are supposed to govern opinion sampling?
Also, is it true their predictive power hasn't been reduced? I thought in the last presidential election the unchallenged consensus of the polls for 3 months running up to the election was that Bush would win the popular vote. (Although some thought Gore might possibly win the electoral college.)
But perhaps that mistake fell within margin of error.
Michael