Since we're discussing potential change in vote intention due to the debate, a reminder about polls and statistical power.
There's an old Jackman presentation that has a nice plot showing the sample size required to detect change in voter support. I think it assumes SRS.
I've written about this basic problem professionally and it still never occurred to me until this post that it would affect opinion polls as well. You need far more precision to capture change than to capture an initial level, and of course most polls are going to be powered to detect a level. Wow.
