I received the following question from a reader:
I have poked around in the Kansas Senate and Wisconsin Governor election results, which to my eye look the most dramatic, and I see that there does appear to be a Republican bias in the larger precincts. Not much, just a few percentage points, but the effect is real. What I don’t understand is what suggests that this might be sinister, as opposed to being just one of those oddities that crop up from time to time.
And thank you for looking into this. I have family in the Wichita area, and I want their votes to count.
Bob, thanks for caring enough to poke around the statistics. I appreciate the concurrence regarding a Republican bias in the larger precincts. You ask good questions.
Why should this small trend disturb us? Why isn’t it just considered a statistical oddity?
It’s not a statistical oddity because it crops up all over the U.S. Therefore, it’s a pattern, not an oddity or ‘one-off’ situation that isn’t expected to be repeated. It cannot be dismissed as an oddity; it is a pattern that has some explanation. Vote rigging is one such explanation, but this analysis cannot establish cause. An audit could establish cause and that is why I am trying to get access to the paper records.
Why consider it ‘sinister’ or evidence of voting machine manipulation? The evidence for this is more complex and not necessarily statistical, such as the documented vulnerability of the machines to being hacked. The most convincing piece of statistical evidence against the various voting machines is that if you separate out the different types, only some types of machines demonstrate this trend. When I separated out the Wisconsin Gov. Election results for paper ballots only, no voting machines involved, the results not only do not show this trend, but form a beautiful example of what is expected under the null hypothesis of no trend.