Reply #94: Yes

[Febble]

lever precincts had a larger mean exit poll discrepancy than precincts using any other voting method. To do the comparison properly, though, I separated rural, suburban and smaller urban (pop<50,000) from larger urban precincts, as paper ballots were only used in smaller places (mostly rural) in the NEP sample. When I looked at the discrepancies in these smaller places there was no significant difference between precincts using the different technologies. However, for urban places, the discrepancy was greatest in lever precincts, followed by punchcard precincts, followed by optical scanners, then DREs. Lever precincts had significantly greater discrepancies than either DREs or optical scanners.

I agree with your last sentence. Corruption is as old as democracy, and it is perfectly easy to corrupt an election conducted with hand-counted paper ballots (an attempt was made in the UK in 2004). No voting method is immune from corruption, though some counting methods may be more accurate than others, and some are more reliable than others.

IMO, the two key issues are chain-of-custody and random audit, whatever method is used, although to be able to do the second you have to have some kind of ballot that can be independently counted.

And none of that will deal with voter suppression, or indeed, with candidates that tell lies.