We all know, by now, that we do a horrible job at counting votes. Who even knows if any of the presidents we have ever elected have ever actually won?
So, up at my grandparents home this weekend, I asked my grandfather statistically how could we best count.
Here were his thoughts:
Take a sample of 3,000 to 4,000 Americans, randomly selected on November 4th. We make sure every single one of them vote.
We will come out with an answer that has a margin of error less than 1%.
I'll get the exact math on this if anyone really wants to know.
Migrated to Tumblr
13 years ago
1 comment:
To get a national number that might be useful (I'm dubious); but wouldn't we need a sample in each state? And unless the sample size were to get unmanageably large, in close contests the results would be within the MoE, and tell us little. Besides, isn't this what Exit Polls seek to do -- and, as we learned from 2004, they have their own problems. . . .
Post a Comment