Exactly, this is a tempest in a teapot unless there is more information. One needs to show that the effect of electronic versus hand-counted votes on the election is statistically independent of the composition of the population (race, median income, M/F ratio, etc.) of each district. Probably districts where hand-counting is still used are on average poorer than those with e-voting machines, for instance.
The linked site makes a first stab at this by breaking down the data into "small towns", "medium towns" and "large towns", but that's only a start at making a real convincing argument. As the parent states, a comparison of the difference %(Obama) - %(Clinton) to the same quantity in the exit poll results, in hand-counted vs. e-voting districts, would be much more convincing. (Note, I say this as someone who much prefers Obama over Clinton.)
