November 20, 2018

The Potential Need for a Hand Recount of All Florida Ballots for the US Senate and Governor Races

According to Alex Harris and Martin Vassolo of the Miami Herald, there were “about 32,000 undervotes and overvotes in the U.S. Senate race between Bill Nelson and Rick Scott”. Florida law mandated a hand recount of the Senate vote (to include only those undervotes and overvotes), as a result of the fact that Scott’s lead in the initial machine vote count and the machine recount was less than 0.25% of the total Florida vote for Senate.

Before the hand recount, Scott’s lead was 12,603 votes out of more than 8 million votes cast in the race. The hand recount reduced Scott’s lead to 10,033 votes, according to CNN, a decrease of 2,570 votes. But according to Harris and Vassolo, “of the more than 32,000 ballots counted, only 410 were valid votes for Nelson and 136 were votes for Scott…” That would suggest a pickup of only 274 votes for Nelson on the hand recount. That sounds wildly inconsistent with the above noted decrease in Scott’s lead of 2,570 votes.

In order to understand the true significance of the hand recount, the following issues need to be clarified:

1) The apparent discrepancy between the hand recount gains for Nelson of 274 votes according to the Miami Herald vs. 2,570 according to CNN needs to be clarified.

2) Regardless of whether Nelson’s gain in the hand recount was 274 votes or 2,570, or something in between those two figures, it must be recognized that the 32,000 ballots involved in the hand recount constitute less than a half of a percent of the total vote for U.S. Senate in Florida. If even the lower figure (274 votes) were extrapolated to the whole state of Florida, Nelson’s total gains would be greater than 65,000 votes, far more than enough to overcome Scott’s 12,603 vote lead prior to the hand recount.

Furthermore, in either case the Nelson gains are far too great to be attributed to random statistical error. In other words, the odds against such gains occurring by random chance are astronomical – millions or billions to 1.

That means that there must have been some systematic reason for Nelson’s gains in the hand recount, compared to the machine counts. Perhaps there was a benign reason for those gains. Any benign reason would involve the presumption that Nelson voters were far worse than Scott voters in properly marking their ballots, or that machines used in Democratic Florida precincts were far worse at recording and counting the marks on the ballots.

In order to get a better understanding of this possibility, it would be necessary to have a better understanding of what characteristics of ballots contributed to the Nelson pickups. If nearly all of these ballots involved marks for the Senate race that were partial or hard to visually read in any way, that would significantly support a benign explanation for the Nelson pickups. But if a substantial number of these ballots (especially those that were changed to Nelson votes in the hand recount) involved marks that visually appeared to fully cover the circle, that would cast substantial doubt on any benign explanation and therefore support a non-benign explanation – i.e. that voting machines were programmed to discount Nelson votes or add Scott votes.

This issue could be easily resolved by inspection and analysis of a statistically valid sample of the ballots that were changed by the hand recount.

3) If an inspection the ballots changed as a result of the hand recount cast doubt on benign explanations for the initial vote count in favor of Scott (as explained above), that would bring into serious question the validity of the vote count involving the more than 99.5% of ballots that were not counted in the hand recount. Only a hand recount of the entire state of Florida would than resolve that issue.

4) If a hand recount of Florida showed a Nelson victory, that would provide a very good reason to hand recount the Florida Governor race as well.