LikeLike

]]>LikeLike

]]>I still wish they’d do it anyway!

LikeLike

]]>> of less than 1%. Three and four percent is clearly no longer acceptable.

Unfortunately, the way statistics works with random sampling, that becomes prohibitively expensive. For example, let’s take Ohio, with 5 million voters, and assume that both candidates are at around 50%. A random sample of 600 gets you a margin of error of 4%. To get the MOE down to 3%, you would need a random sample of 1067 — about 1.8 times as many. But to get the MOE down to 1%, you would need a random sample of 9590 voters — almost 16 times as many as a 4% MOE and 9 times as many as a 3% MOE.

(Other than that, I think you make some very insightful points.)

]]>LikeLike

]]>LikeLike

]]>LikeLike

]]>LikeLike

]]>LikeLike

]]>