Citizen Audit of Allegan Elections Shows Weaknesses
in Ballot Security, Count Accuracy

(Lansing, April 2, 2014) -- An unprecedented citizens’ audit of election returns from Allegan County in 2008 and 2012 shows “egregious and numerous failures of ballot security and ballot accounting along with other election integrity weaknesses.” The audit, released today, raises concerns of election integrity and tabulator accuracy. However, the audit vote totals did not show large enough discrepancies to suggest intentional tampering.

The study of election returns in two Allegan County elections – the 2008 general election and 2012 Republican primary – was initiated by the West Michigan News Company and brought to completion by the Michigan Election Reform Alliance (MERA). It was motivated initially by a failed recount in a 2008 judicial election. In the attempted recount, MERA notes that due to broken seals on the ballot storage containers in over half the precincts, the recount board could not be certain that the ballot sets were intact and unaltered, and the recount had to be dropped. “The State Board of Canvassers refused to investigate further,” said Phil Shepard, who lead MERA’s work on the audit.

MERA completed the project in part to determine how accurately the optical scan tabulators used in Michigan count votes. Audit data on 17 precincts from the ’08 election and 11 from the ’12 primary show average tabulator error rates of 0.26% and 0.42%. Single race error rates from the ’08 election ranged from 0.0% to 1.08%. In the ’12 primary, the single race range was from 0.0% to 1.78%.

These data indicate a critical need for manual recounts in close local races with small vote totals. The recount practice of running the ballots through the same tabulators again clearly does not provide a meaningful check on the correctness of close outcomes. An earlier report -- Facing Michigan’s Election Cliff (see below) -- demonstrates the growing unreliability of the optical scan counting systems used throughout Michigan.

“The high tabulator error rate, coupled with the various anomalies and election integrity failings detailed in the report, raise a very serious question: Should the current system of vote counting be replaced, perhaps by the time-tested practice of hand counting the vote on election night?” said Shepard.

The study was complicated by a reluctance of Allegan County elections officials to cooperate in the effort. With only one exception, Allegan’s local clerks initially refused a Freedom of Information Act (FOIA) request to view and photograph the ballots. The Attorney General then ruled that ballots were subject to FOIA once an election had been certified (AG Opinion #7247). Some local clerks continued to resist the audit of the 2012 Republican primary, resulting in a smaller number of precincts being audited.

The report details a variety of ballot security issues discovered during the audits. In one case, a precinct’s sealed set of ballots were found to be about 200 ballots short of the official total. Then the clerk went in a back room and retrieved about 200 ballots that were not from the sealed set and presented them as voted ballots. “Evidently ballot security in this precinct in ’08 was less than perfect,” said Shepard.

The audit also found small errors (1-3) in the total number of ballots that were presented to the audit by local clerks. When the slightly incomplete precinct ballot sets were excluded from the accuracy calculation, the average tabulator error rates were 0.21% (’08 general) and 0.35% (’12 primary).


Mark Grebner published a deeper analysis of MERA's data: "Q: How many votes change do recounts change? A: About 5 per 1000" and a comparison to findings from a different recount: "More fun with recount statistical analysis"
In addition, some of the ballot containers photographed showed certificates with three or four seal numbers, indicating that they had been opened and resealed two or three times before being seen by the audit. The attempted recount of the contested judicial race in ’08 may account for some containers being resealed at least once. Other re-sealings could have resulted from the same recount attempt if ballot sets were opened for the recount but not dealt with timely. Otherwise, the audit has not been able to account for multiple re-sealings, including cases from the ’12 primary as well as the ’08 general.

The full Allegan audit report is available at:
http://michiganelectionreformalliance.org/AlleganAudit.pdf.