[ExI] unanimous decision, was: RE: An old skeleton tumbles out of the list closet

[spike]

>...From: Kevin Cadmus [mailto:kcadmus at gmail.com] 
Sent: Wednesday, November 14, 2012 3:34 PM
To: ExI chat list
Subject: Re: [ExI] unanimous decision, was: RE: An old skeleton tumbles out
of the list closet

>...A bit of math leads me to think the result is not so unbelievable at
all:

>...67000 votes in 1687 divisions gives approximate 40 voters per average
division

>...Let's say that 500 of the 1687 divisions are essentially racially
homogeneous, namely 100% African American.

>...Let's say that we might expect 95% of voters in these 500 divisions to
be Obama voters.

>...Taking 0.95 to the 40th power should give the probability that all 40
voters in one of these 500 districts would vote for for Obama.

>...But 0.95 to the 40th power is 0.1285.

>...So wouldn't you expect approximately 500 * 0.1285 districts to be
unanimous?

-Kevin


There were 19,605 votes in 59 unanimous divisions, so best case for the null
hypothesis (no funny business with the voting machines) is to average: about
330 voters per division.  Kevin, let us assume your numbers, 95% of the
voters in those divisions to be Obama voters.  (0.95)^330 = 4e-8.  So we
would expect about 0.04 parts per million probability that there would be a
unanimous vote in any given division in such a case.  We are told that
5,596,499 votes were cast in Pennsylvania, so using the above numbers,
that's 17,000 voting divisions, so we would expect about 0.00067 unanimous
divisions, rather than the 59 seen, or about 88,000 times higher than we
would expect.  That's the best case.  If we change the distribution of those
19,605 votes over the 59 unanimous divisions, the probability of their all
coming out unanimous goes down from there.

But the critical question is if the number of Johnson and Stein (et.al)
votes in those zero-Romney divisions also went to zero.  If so, then the
error rate in those divisions is also zero, shaming those Florida voters
from 2000 who claimed errors when they accidentally voted for the wrong guy.
Pennsylvania voters don't make mistakes apparently, but Florida voters do,
enough to actually tip an entire election.  

Or alternatively, can we find any divisions in which the Romney vote was
zero but the other candidates were non-zero?  Is there anywhere we can find
that info?  Or is that info intentionally held from the public?

spike