Centerfield

Golly, Tulley, this noncommie, nonpinko, nonlibrul Wally Street people even mention our friend Ruy! How thoughtful of them. Must send cookies.

Yeah, ain't it sweet, Erasmus? But flowers or a card would be better--the mailroom would either eat the cookies themselves or send them to the lab for analysis.

My two favorite slice-quotes from the WSJ story are Dem pollster Peter Hart:

"This is art. This isn't science. Nobody knows."

And this one:

"We're open to any scientific evidence that would point to our modifying our likely-voter model," responds Frank Newport, editor-in-chief of the Gallup Poll. Mr. Newport says so far he hasn't seen any.

More interesting to me will be matching the weighted and non-weighted polls from the last week of October to the actual election results. The pollsters can use that data to be wrong next time. ;-)

I wonder what the correlation is between "likely voters" and "actual voters."

I also wonder why none of these polls attempts to balance out results when one party or the other is overrepresented. They should make some sort of presumption of what the expected ratio of dems to gopers will be, and then balance out the results by randomly discounting some of the votes of the overrepresented group. For example, if the survey ends up asking 520 gopers and 480 democrats, then if they presume a 50-50 split come election day, they should randomly selected 40 gop opinions and not count these. It seems foolish to me to randomly discount such week to week discrepancies as "drift."