Keeping Voting Machines Honest

by Elizabeth Walter Clarkson, Ph.D, Kansas statistician/ researcher investigating possible election manipulation in the Kansas 2014 election, has been barred from having access to the “Real Time Audit Logs”, aka “Voter Verifiable Paper Audit Trails” or just  “paper trails” needed to verify the election results.  The judge allowed Clarkson’s lawsuit to move forward, but prohibited her from obtaining the documentation necessary to verify the machine counts, which is at the heart of the lawsuit.

Clarkson’s lawsuit is intended to promote an understanding of why her statistical analysis of a 2014 election reveals what has been called a “red shift,” a shift unexpectedly found in  a number of recent elections…. 

Contrary to expected results, she, and other statisticians, are finding an unexpected statistical pattern.   Generally, in small precincts, which are typically rural with few voters,  the Republican share of votes cast is high. As precincts get larger, and therefore increasingly urban, the share of votes cast for Republicans tends to decrease with Democratic votes gaining share.  This is the expected pattern.  In certain elections, statisticians have found this expected pattern until precinct size reaches 400 – 500 votes cast, at which time the pattern changes and they begin to see an increase in the Republican share as the precinct size gets larger, and a decrease in the Democratic share. It is this latter pattern that is considered a statistical anomaly.    Because the latter pattern is statistically unlikely, the question is why is this unlikely pattern occurring?  Is there manipulation (e.g. vote flipping) happening such that votes for Democrats are being flipped and converted to votes for Republicans?  Or is something else happening? Because the data suggests a direct relationship between a gain of vote share for the Republicans and the loss of share for the Democrats, manipulation seems to be the most obvious cause.  The only way to know for sure is to look at the actual ballots and conduct a hand count by a person versus a machine.  Preventing such a count prevents researchers and the public from determining if the vote counts are accurate or manipulated.

Clarkson notes that, when it comes to election results,  “optical scan computers may provide quick results” but those results are never verified because no person actually verifies the machine totals.  Thus she supports using a paper system with 100% of the ballots hand counted at the precinct on election night.   She supports hand counting 100% of the ballots rather than a sampling because too many people do not understand the statistics of audits  and they can be manipulated.  Public hand counts ensures transparency and it is a method that the public can understand and witness, thus creating trust and verifiability.