Bruce wrote:

>>A law passed by the Texas legislature does not constitute evidence.
Somebody would need to prepare several collections of ballots in various
states of almost-votedness, and give them to a bunch of groups of people
to count, and compare the results among the groups. Does *anyone* know of
any statistical tests of hand-counting?

Quite true, except the Texas law was written in response to earlier
concerns AND tests they had done which confirmed there was a potential
problem.

As for the accuracy of a hand-count.  A hand count is used to calibrated
and verify mechanical voting machines, in its simplest form it is the gold
standard.  But to be fair the only way to be assured that a hand count is
absolutely correct is to keep re-counting until you get the same result
several times.  :-)

Punch tab voting and the counting machines can be extremely accurate IF the
voter checks and makes sure that the chad is completely removed from the
ballot.  Unfortunately that is difficult to explain to many in the general
population.  It can be argued that perhaps they shouldn't vote then,
however that is quite contrary to the goals of our founding fathers.

The error analysis for the punch tab voting process was that the voter had
successfully voted (only one whole punched) but that the chad did not
separate completely and when fed through the tabulating machines, the chad
is flipped back into the hole causing the ballot to appear as if a vote
wasn't cast.

Because of problems like these with the punch tab machines, many counties
are phasing them out for either AccuVote machines OR in the case of Erie
NY, they've gone back to the old mechanical pull tab machines.  The auditor
there said they were more accurate!