From: Rahul Dave (rahul@reno.cis.upenn.edu)
Date: Tue Nov 28 2000 - 13:49:22 PST
I mean to say N as a margin of separation, not the margin of eror
on each count. But it still does not become 4% if you have two spreads of 2%
each, its more like \sqrt(8), or less that 3% on the difference.
But yes, the certainty needs to be higher. This aint bad though. More likely
spreads though are likely to be less than 1%..in this election, 4-5 states
fl, nm, wi, io, or may qualify..fl and nm surely, i'm not sure about the others..
Rahul
>
> But I fear this would cause twice as many "close races". Let's say the
> margin of error is defined to be 2%, for each candidate. That is, for any
> candidate, their votes might be overcounted or undercounted by up to 2%.
> Any difference up to 4% between the votes for the top two candidates would
> fall in this margin. So if A got 47.9% and B got 44.1%, your formula would
> call that a "tie", but I'm sure A would argue for a recount so that A would
> get all the votes instead of just half. Or if A got 48.5% and B got 43.5%,
> B would certainly argue for a recount in order to get at least half the
> votes instead of none. The same thing would happen if A and B were
> reversed. We'd still get political wrangling for recounts, just more often
> and for lower stakes.
>
> lisa
>
This archive was generated by hypermail 2b29 : Tue Nov 28 2000 - 13:54:45 PST