Except that is 59 tests out of 1600. Let's at least assume all 400 people in the division wanted to vote D. Let's say there are 1000 tests of these 400 people (400k out of the 675K blacks). That they got it right 59 out of 1000 times is not impossible. The machines are dumbed down to prevent mistakes.
Here's how I look at it:
There are at least 59 precincts out of 1600. Each precinct reported 100%, and that the vote was 100% 0bama. If there are roughly 6000 registered voters per precinct, and 6000 (or more) of them chose 0bama, then that is a big enough sample to make some assumptions...
If 90% of black voters voted for 0bama, then 10% voted for other candidates. So, not including any errors (e.g. Ethiopians, Somalis, Mexicans were coached through the process with no errors), here's the population of the sample: 354,000.
Now, you and I know that in Chester, S. Philly, etc., Romney was not going to get 10%. But let's say that there were 1000 Republicans in those 59 precincts. That equals .002.
I should have said, statistically improbable, not impossible.
5.56mm