It appears just about every democraphic factor that has been tried, including the size of the precinct both as a continuous variable and as a large vs. small categorical variable, has failed to remove the Diebold effect. I am not sure gender has been tried, though. I am also not sure that the regressions involving both size and method have been carried out properly.

However, for the correlation between Clinton's lead and count size to be so "smooth", the amount of stuffing would also have to be proportionate to ward size.  Do we really think that a fraudster would be that resourceful and clever?

No, in this case the suggestion is a simple reversal of the vote counts between Clinton and Obama.

Flipping count method virtually eliminates any correlation between Clinton's lead and ward size.

Ergo - Clinton's lead is actually correlated with ward size, but because that also correlates with count method, it has been confused with the correlation with count method.

No, this is not "flipping count method". It is flipping the sign of the lead between Clinton and Obama for one of the two methods.

We have met the enemy, and he is us — Pogo