Florida Data/Machines

[CHIMO]

I entered this on another board and have not received any answers, so I am posting here. Maybe someone can justify the divergence?

Did some simple calcs based on the numbers in the table below.

http://ustogether.org/Florida_Election.htm

I have not checked the validity of the table but a post below for one county matches.

Assumptions:

Expected Votes in each county for Dems are proportional to the percentage of registered voters for the Dems.
ie Dem/(Dem + Rep) x Total Vote

Population using touch screen is the same as that as Op Scanners.

Using the above assumptions:
The ratio of Expected votes to actual votes for touch screen is 1.02
The ratio of Expected votes to actual votes for op scanners is 1.22

Assumptions are wrong.

If one plots expected vs actual for the touch screen and also for the op scanners the plots looks fairly linear for both the touch screen and the op scanner(The op scanner appears to have outliers for the two largest vales but eyeballing it looks OK?)

From this I would say that the assumption that the populations are the same is incorrect. I do not know the way the machines were distributed. Is this an expected item "That the populations are not the same?"

Additional: If one further assumes that the touch screen and op scanners have the same population and that the op scanner is biased, then moving those votes to get a ratio of 1.02, then that makes the % vote for Dem 51.9%. Dam close to the exit polls.

Hopefully some of the optical scanned votes could be checked against the input cards.

Would appreciate if someone could provide a good reason for the difference between the two methods of recording the votes.

Thanks