Crunching New Hampshire
by WonkoSteve
Shortly after the New Hampshire primary, posts were floating around about possible vote fraud because of unusual patterns in machine-voting results. Yesterday, Jon Stokes posted what seems like a pretty sober analysis at the Ars Technica blog. It seems that the “patterns” people were seeing in the numbers were the result of incomplete counts and that the resulting incomplete conclusions were amplified as they flew around the Wonkosphere. As Stokes wryly puts it:
The Internet is full of people who have four things that make them dangerous, both to would-be election fraudsters and (paradoxically) to the larger cause of election integrity: computers, intermediate math skills, a mix of patriotic and entrepreneurial zeal, and the ability to publish in the blink of an eye. When you add a stream of evolving vote tallies to this mix and shake vigorously, the resulting concoction will produce lots and lots of foam.
In the end he says that when you skim the foam there is still a correlation between who won a precinct and its vote-counting method, but ir remains to be seen whether some other variable accounts for it.
The real problem, according to Stokes, is that because 34 states don’t require a verifiable paper trail and/or random audit of results, there is a significant danger of a repeat of this controversy in other primaries and especially in the fall general elections.
