New Hampshire election officials re-counted the paper ballots from nine of the electronic machines that counted votes in the Nov. 3 general election and found small differences from the machines' own totals, which the audit team called within expected margins. One sampled ballot in Manchester Ward 7 was read differently by the machine and by the auditors, and the audit team's explanation for it is a theory.
An audit team appointed for the Secretary of State called the audit "successful" in a memo to Secretary of State David Scanlan, the Ballot Law Commission and the chairs of the House and Senate election committees. The report is dated Nov. 6, the day the office had said it would publish it.
How the check worked
State law requires the Secretary of State to randomly select at least eight ballot counters for audit. Eight of the nine were chosen under RSA 660:35, the post-election audit law, and a ninth, in Nashua Ward 9, was chosen by the Secretary of State under RSA 660:32 with its serial number identified in advance. The Secretary of State's office announced the locations on Nov. 3, after polls opened.
The audited machines were Liberty Votes units in Hudson, Manchester Ward 7, Concord Ward 4, Laconia Ward 5 and Nashua Ward 9, and VotingWorks units in Moultonborough, Portsmouth Ward 2, Rochester Ward 3 and Woodstock.
Auditors at the New Hampshire State Archives in Concord ran the paper ballots each machine had counted through separate audit equipment, on Nov. 4 and Nov. 5, and compared that count of ballots cast and race-by-race results with the machine's. The audits were open to the public. They then pulled a random sample of 50 ballots, or 2 percent of all ballots cast, whichever was larger, and checked the paper against the voters' choices recorded by the audit equipment.
What they found
Most machines showed differences of a few votes in individual lines of the report's tables, and the audit team called all of them within expected margins. One of the larger gaps was in Manchester Ward 7, in the Hillsborough District 41 House race. An undervote is a ballot with no choice marked in a race. The audit equipment counted 1,258 undervotes there; the machine's Election Day results showed 1,280, a gap of 22.
Eight of the nine machines showed no discrepancies in the hand check. In Manchester Ward 7, one sampled ballot differed. A voter apparently scribbled through "No" and marked "Yes" on Statutory Question 2, but the mark was only slightly inside the oval. The machine is programmed to reject overvoted ballots (those with more marks than allowed) and registered no "Yes" vote. The audit equipment read both marks.
The audit team theorized that the counting machine was less sensitive than the audit equipment. Only a sample of ballots was examined, and the report does not say how many other Manchester Ward 7 ballots carried a marginal mark like this one.
Separately, the Attorney General's office reported on Nov. 5 that its Election Day hotline took about 277 calls and that it received no credible allegations of election interference. It said Election Day operations ran without significant disruption to administration of the election.
