Mike Farb's site links to VoteSleuth.org, which is very interesting and well written. I'd love to know what the statisticians among us thought of this?
www.votesleuth.org/about/
About
We had been following @mikefarb1 with great interest on Twitter. He and a crowd-sourced team of crack data analysts have been finding some very interesting peculiarities in precinct-level election statistics from recent elections. His website is here.
At some point @mikefarb1 posted a link to an article about a statistician’s denied request (www.kansas.com/news/politics-government/article61222187.html) to audit Kansas voting machines based on some statistically anomalous results she had found when looking at precinct-level election returns vs. precinct size.
This led us to some extremely alarming papers by statisticians who have noticed odd-looking data when they tally precinct-level votes starting with the smallest precincts and moving to the largest.
We found this quite shocking. Their results looked bizarre:
How Trustworthy are Electronic Voting Systems in the US? (Beth Clarkson) ^counterinformation.wordpress.com/2015/06/26/how-trustworthy-are-electronic-voting-systems-in-the-us/^
Republican Primary Election 2012 Results – Amazing Statistical Anomalies ^electiondefensealliance.org/files/PrimaryElectionResultsAmazingStatisticalAnomalies_V2.1.pdf^
So we decided to investigate this ourselves.
This website is a record of our ongoing work.
Specific discussion of the 2016 US election stats at: www.votesleuth.org
And they have a very readable page on The Law of Large Numbers - what you would expect a graph of precinct-by-precinct results to look like, and why it's suspicious for large precincts to be inconsistent with small precincts.
www.votesleuth.org/the-law-of-large-numbers/