A news report says that trained personnel entering test votes on the eSlate made errors on 40% of the test ballots in the first day of testing; 25% of the ballots the second day; and 14% the third day. Voting integrity advocates worry that voters using the machines for the first time will have comparable error rates.*

VotersUnite has learned that this is a misrepresentation of the facts. We received this information from Dennis Paull, a county poll Inspector and election observer for the Democratic Central Committee, who attended the testing (our highlighting):

The 40% error rate refers to the precincts tested, not to the ballots entered. 40% of the precincts were found to have one or more of the votes entered on one or more of the 25 ballots be in error. So the error rate per ballot was closer to1.6%. The number of races per ballot varied but the max was 43. Thus the error rate per vote was closer to .04%.

The elections staff attributed the errors to tester errors. They repeated the entry of all ballots for the precincts that showed an error. Again, the error rate was about the same but the number of precincts was reduced. This was repeated twice more and finally all votes came out as expected.

I do not consider this test procedure to be appropriate as it is not clear whether the errors were people errors or random machine errors. So it is not at all clear whether the machines will function correctly on election day. But the testing at least showed the entry of precinct-specific election data was most likely correct.