UPDATED: What is “statistically insignificant”?

Update:  One of the Greenwich Registrars has commented on the first story below and corrected some misinformation.

by Fred DeCaro III
12:46 pm on Sunday, December 4, 2011

It is a shame that poor “reporting” and the absence of fact-checking led Mr. Weeks to comment on this story.

The “reporter” of this story simply copied from a Greenwich Time story, including mistakes in that story.

First, he places the term “statistically insignificant” in quotes. This is not a quote from me, but a quote from the Greenwich Time news story the “reporter” copied from.

Second, because the “reporter” did not verify anything, and merely copied from Greenwich Time, he copied their erroneous statement that 1,250 ballots were hand counted. There were 2,500 ballots counted by hand, and the discrepancy was .36%, which is under the threshold mentioned by Mr. Weeks, who is a well-respected CT election watchdog.

Most disturbing to me personally is when the “reporter” uses inflammatory language and says “the registrars reportedly decided that since the discrepancy was “statistically insignificant” that they wouldn’t bother to take the time to investigate the error.”

Not only is the phony quote used again, but the word choice of “wouldn’t bother to take the time” makes it seem like we didn’t care about the discrepancy.

Together as Republican and Democrat Registrars we made the decision that spending an additional $240 an hour of taxpayer money (with a minimum of 2-3 more hours necessary) to research this matter would not enhance public confidence in the process. I doubt very many Greenwich taxpayers would disagree with that decision.

Greenwich Patch: Voting Machine Audit Takes Nine Votes From Marzullo <read>

A state-mandated hand recount of votes cast in the Nov. 8 election in two of Greenwich’s polling places to determine the accuracy of the new electronic voting machines on Thursday reportedly revealed that Democratic Selectman Drew Marzullo received a “statistically insignificant” nine fewer votes than were originally tabulated by the machines…

The registrars reportedly decided that since the discrepancy was “statistically insignificant” that they wouldn’t bother to take the time to investigate the error

Revised comments: There is no agreed upon level of difference that would be considered “statistically insignificant” in Connecticut. The nine votes out of 2,500 in this case represents just 0.36% of the votes which is approaching the 0.5% threshold for a recanvass. To me anything over that threshold or anywhere near it is significant. Especially since we do not even consider differences in the audit attributable to scanners’ inability to read voters’ intent. We could benefit by established standards for accepting counts and for triggering further investigations.

In another Fairfield County town, officials are concerned about the cost of audits. The Daily Weston: Weston Officials Blast State’s Election Audit <read>

“Am I happy about this? Of course I am not happy about it. This is ridiculous, it’s an unfunded mandate,” said First Selectman Gayle Weinstein…

The purpose of the audit is to make sure voting machines are working correctly, Merrill said. “With this audit, we now must take the step of checking the machine totals from Nov. 8 to ensure the accuracy of our optical scanners. We are committed to making sure Connecticut voters have continued confidence that their votes were recorded accurately, and that’s why these independent audits are so vital.”

The actual hand counting of the three races, Weinstein said, will happen Saturday at Town Hall. But she said the audit is unfair because the town must pay poll workers to spend the day counting votes. Weinstein estimates the audit could cost taxpayers $2,500.

“We have to sit here and count each ballot by hand. I can’t believe it,” said Laura Smits, the Democratic registrar of voter. “I am hoping we get this finished in one day, but who knows. This is costing us a lot of money.”

We were in Weston to observe the audit. It was very well conducted and efficient. If we had not read the article beforehand, we would have never guessed there was any official in Weston concerned with cost. All seemed concerned with performing the audit as intended.

The audit likely cost a bit less than $2500, perhaps $1,500, which would have been about $0.60 per ballot, or about $0.06 per ballot cast in the election statewide given the 10% audit. Perhaps also around 7% of the costs of the ballots printed for the election statewide or less than 1% of the cost of running the election statewide. See <Election Costs $ – Democracy? Priceless!>

Facebooktwittergoogle_plusredditpinterestlinkedintumblrmailFacebooktwittergoogle_plusredditpinterestlinkedintumblrmail

Leave a Reply

You must be logged in to post a comment.