Optical Scan Machines Performed "Very Well" Says Audit Report

The Secretary of the State's office and the University of Connecticut today released a preliminary report on the recent audit performed on optical scan voting machines. The report suggests that the optical scan machines are just as reliable as the lever machines. From the press release:

Secretary of the State Susan Bysiewicz today released a report from her office and the University of Connecticut which shows that the optical scan voting machines used in several cities and towns on Election Day performed extremely well, and were proven to be a safe and secure form of voting technology.

The report was released following a randomly conducted series of audits in a number of cities and towns which used optical scan machines in place of lever machines this year. The audits were performed during the final week of November on optical scan machines used in seventeen polling precincts in nine cities and towns throughout the state.

The audits involved performing a manual recount of each voting machine used in these seventeen precincts to match the machine results with the separate hand-count. In the majority of the precincts, the counts matched up perfectly, and in those where the results did not match, there were only minor changes reported. In each instance, the change was due to a mismarked ballot, not to machine error, Secretary Bysiewicz said.

See the full report here. Some of the key findings summarized on the first page of the report:

1. The overall discrepancies between the machine counts and hand counts are not statistically significant.
2. The effect of individual machines (locations) is not significant.
3. The machine counts and hand counts are proportional with the proportionality constant 1, although the machines have a tendency to overcount, on the average by 1/2 vote.
4. The accuracy of the voting machines did not affect the election results (although no audited races were particularly close). One needs to be careful, however, in relying on machine counts when the total number of ballots is small, especially in multi-opening races.

The report itself is worth a read, especially if you're interested in statistics. It's important to note that there were some minor problems where ballots were mismarked, but the machines themselves seem to have performed well enough. The nice thing here is that these sorts of audits can be performed at all. Neither lever voting machines nor touch-screen machines have a paper trail that can be verified.

I'm looking forward to using these machines myself next November.