CT: More (Human?) Errors In Reported Results?

[Wilms]

CT: More Errors In Reported Results?

By Luther Weeks

December 25, 2008

We have covered reports of differences between the results posted on the Secretary of the State’s web site and actual election results in Shelton and Stamford. We have learned of a third problem in the results posted for the Congressional race in Avon, with the possibility of similar problems in several other towns.

David Bedell (Green Party candidate for Registrar, Stamford) was researching the effect the Working Families Party may have had in the recent election. He noticed several anomalies: 9 towns with instances of zero counts for WFP and one with duplicate results for the Democratic and WFP for a candidate.

He checked with the Avon Registrar and she reported U.S. Representative Chris Murphy’s results were Democratic - 5111 and WFP - 269, making a total of 5380.

Checking this morning at the Secretary of the State’s web site we see totals of Democratic - 5577 and WKF of 0, an extra 197 votes. By our calculations that amounts to about 1.9% of the initially reported results for the race in Avon.

As we have pointed out before, the manual calculation of votes, usually with three transcriptions and several additions along the way is an error prone process. It can be a daunting task for voters, candidates, and parties to check results independently, since the State publishes summary results which must be verified against multiple hand written reports and machine tapes held in each town hall.

~snip~

http://www.ctvoterscount.org/?p=1242

[eridani]

1. Yes, it is. Handcounting is the most accurate method only when counting single races

More than one, and optical scanning is the most accurate. Of course that can only be confirmed with random auditing.

[Wilms]

2. Whoa. Why is that?

If I count a single race HC is more accurate.

But if two races are involved, the scanner beats HC??

I'll need help with that.

[eridani]

5. Because people get tired and crabby trying to do more than one thing at once n/t

[Wilms]

6. Sometimes I get crabby doing only one thing at a time.

Do you have a link?

[Bill Bored]

3. This has nothing to do with counting.

It has to do with transcribing precinct tallies. In other words, "copy and paste." I think it's fair to say that computers might be pretty good at doing that, provided the source of the data being copied is correct and that people verify the copies.

What CT seems to be doing is using lever machine procedures with computers. But without such procedures, you have to rely on the memory cards and/or data networks to transmit this data. It should always be checked against the election-night tally reported on the scanners' paper results reports. This doesn't ensure that the count is correct of course -- just that the data has been copied correctly to the EMS.

After that, you have to check to see if the EMS has added up the precinct tallies correctly too.

[Wilms]

4. One thing at a time, Bill.

But, yes. This is a different problem.

And it should remind us that ballot and ballot count integrity only BEGIN at the precinct.