[jack gillis]

I reject the validity of the Zogby Battleground instrument until it has been shown valid by replication and predictive confirmation.

I do so because any poll that relies to any degree on self-selection has to have a built-in bias. It's the reverse-only-worse of the participation-rate problem.

[Dales]

Exactly.

And the most likely to not be caught by such methods are the purple.

Finals go well?