Rich R: Yes, the poll I described above COULD BE administered in a way that would tend to produce biased results, but I don't see any evidence that it actually WAS being administered that way. To repeat my previous observation, I don't see how you can get any more unbiased.

The only thing I can think of that might wipe out the (small) advantage in being listed first (considering that SOMEBODY has to be first) would be to make sure that Republicans were listed first in half of the polling calls, and Democrats were listed first in the other half. For all I know, maybe they were doing that. Maybe I just happened to hear the Republican-first version. (Honestly, my memory isn't that good. Maybe the Democrats WERE listed first. Don't take my transcription THAT literally.)

On the other hand, doesn't the same advantage exist in the election itself? Does the election board in your state print half of the ballots listing Republicans first, and the other half listing the Democrats first, and then distribute the ballots randomly? I doubt it (although it is technically feasible).

And I'm not even beginning to consider the complications involved in having 3 or more parties on the ballot!

At any rate, the purpose of a poll is (usually) NOT to find out what the people really want, but to predict the results of the election. Therefore, arguably, if there is any bias in the election, then there ought to be EXACTLY THE SAME bias in the polls.