I don't understand: Rasmussen says the MoE for these small states is 5%, which would mean a sample size of about 500.
But if you take 500 people a day, you'll only get about 4 or 5 each day from Arkansas, Alabama, Oregon, and South Carolina and only two/day from Maine. This would be samples of 120-150 for the medium-sized states, a MoE of around 9%, and an insanely small 60 person sample for Maine.
Is Rasmsussen oversampling the small states, or it this whole project BS?
Actually, +/- 5% is only a sample of about 400 (384 if we want to be picky)
Yes he is oversampling, and/or doing supplemental interviews in ceratin states to get up to a minimum sample size.
Three states which (I think) might blow up real bad on Rasmussen are Colorado, Arizona, and Washington. I am curious about Nevada also.
These states are big enough that most of their interviews will be "subset" interviews from the daily tracks all put together.
Since the really little states will be mostly suplemental interviews they "might" be ok.
The statistical validity of this thing is getting pretty thin... IMHO.
And again, at +/- 5% if we get 40 of these state polls, be can expect 10-15 of them to be wrong by more than 5%.
Ah, I see that now at the bottom of each article. This could actually make the small state polls more reliable than the larger state polls. And I meant 400 instead of 500...sorry.