A Day At The Recanvass

When initial election results are close, in Connecticut, there is an automatic recanvass.  Loosely, a recanvass is often called a recount.  Yesterday we had municipal primary elections in twenty-nine of our one-hundred and sixty-nine towns.  I woke up to a Hartford Courant article about a close primary in Cromwell that would be “recounted” at 10:00am today.

Although I have observed nineteen post-election audits, I have only attended one previous recanvass and that was a lever machine recount after the November 2006 election.  I was available and Cromwell is nearby.  From the Courant <read>

The town will hold a recount Wednesday at 10 a.m. in room 224 at town hall to determine the winners of the remaining three spots on the Nov. 3 ballot.

Emanuele — the only candidate whose vote total was more than 20 votes higher than those of his competitors — received 370 votes, according to unofficial results. David Murphy received 317 votes; Ann Marie Halibozek received 313 votes; Anthony Varricchio Sr. and Stephen Bayley each received 305 votes; and Richie Waters received 249 votes. All of the vote tallies are unofficial.

A recount is required in elections in which there is less than a 20-vote difference between candidates, there is a tie or there is a difference of half of 1 percent of votes cast. About 16 percent of the town’s 3,272 voters participated in the primary, with 542 people casting ballots, DiProto said.

If officials confirm the tie between Bayley and Varricchio Wednesday, there will be a runoff in three weeks, DiProto said.

To better follow the action, I printed a copy of the latest Recanvass Procedure Manual from the Secretary of the State’s Office.  Yet, there was no time to read it.   I will have more to say about the procedures later.  Readers familiar with CTVotersCount know that we have concerns with the basic structure of the recanvass, since it is not a manual count but primarily relies on recounting most ballots through the same type of optical scanner and presumably a duplicate memory card from our out of state vendor. <see Minnesota Recount vs. Connecticut Recanvass>  I am hardly as familiar with the recanvass procedures as I am with the post-election audit procedures.  After each post-election audit period I have submitted procedure suggestions for improvement, many of which have  been accepted.  Today my intent would be watching and absorbing information, clarifying my understanding of the current procedures.  I would find several areas which could be improved and several areas where my understanding of the actual procedures was formed by observer training  received from national party officials, as part of the 2006 2nd Congressional District recanvass (it was good training, but much of what the trainers said about the role and rights of official observers was not covered in the procedure manual and apparently not common knowledge of the candidates and parties, judging by my experience today).

I arrived at about 9:50.  Several  town officials were there with the sealed ballot bag and a tabulator sealed in a bag.  The meeting room may be the town council meeting room, as it had several rectangular tables arranged in a “U” with rows of folding chairs facing the tables.  When the proceedings started there were a total of perhaps 16 people including, I believe , 5  to 7 officials, candidates, party observers,  and other observers – all from Cromwell except for me.   These included Moderator Alice Kelly, Democratic Registrar of Voters Linda Duren, and Town Clerk Darlene DiProto.

The recanvass was conducted and supervised by Alice Kelly, the Moderator.  Most post-election audits are conducted by one or both registrars – but the procedures specify that the Moderator conduct the recanvass.  This seems to me to be a good idea for two reasons:  First, moderators are supposed to be neutral – there is no guarantee, but it sets a better tone.  Second, moderators work for the registrars, so with one or both registrars present the registrars can observe and, if necessary, correct the moderator.  My observation of post-election audits leads me to conclude that registrars are often reluctant to correct the election officials they are supervising, yet since everyone works  for the registrars, other election officials are reluctant raise concerns.

Alice did an excellent job, explaining what was happening, and offering everyone opportunities to ask  questions.  After each major step of the process, she asked “Are there any questions?” or “Are there any objections?”.

She explained that they were using the backup AccuVote-OS optical scanner which had undergone pre-election testing.  She or another official read the serial numbers from the bag while an official observed up close.  After the bag was opened she showed the public the other seals on the scanner and read those numbers aloud.  This is where I should have asked if she was comparing them to another record or writing them down etc.

Part of my recall of that training from 2006 was that two observers designated by party chairs or candidates were the only ones who could observe up close and ask questions, so at first I was reluctant to speak up.  Checking now, I note the procedures say in two places:

Each chairman may send two representatives to be present at the recanvass.  The representatives may observe the recanvass, but may not participate in the recanvass.


All recanvass procedures shall be open to public observation, however, no member of the public can actively participate or interfere with the conduct of the recanvass.

I don’t see similar language specific for candidates in primaries, such as this one, where some candidates are competing with the party slate.

I find no other reference to observers in the manual.  It could benefit by language similar to that in the latest Post-Election  Audit Procedure Manual:


The State of Connecticut is committed to an open, public, and transparent process. Public Act 07-194 specifically provides that the audit “shall be open to public inspection.” This means that observers should be allowed to view every aspect of the proceedings, including being close enough: (1) to actually see ballots as they are being counted; (2) to see tally sheets as they are being marked and when they are complete; (3) to see report forms to be sent to the Office of the Secretary of the State; (4) to be able to verify that the counts from tally sheets are properly added and conform to the totals on the report forms; (5) to be able to verify the seal number(s) and observe the condition of the seals before they are broken; (6) to observe the moderator’s report, machine tape, and other documents created on election day; and (7) to view any other documents created in the process and to ask questions of Registrars regarding the process.

Observers should be allowed freedom of movement sufficient to enable them to view the items described above. However, it lies within the discretion of the Registrars of Voters to ensure that no observer disrupts the integrity or the orderliness of the process.

The recanvass should provide this same access to the official party and candidate observers.  There also may be a need for a similar, but perhaps restricted role articulated for other observers.  In addition, I note that only two party/candidate observers are specified.  I would require the moderator to add to the notice of the recanvass a maximum number of teams that will simultaneously be performing the recanvass and allow each party/candiate at least two observers per team, in order that representatives can watch every aspect of the process.

I also note the value of having some type of “Observer Guide”  to be made available at recanvasses or audits such that the public, party observers, and candidate observers would know what to expect and to what extent they can observe and question the proceedings.  We are aware of such an  Election Observer  Handbook, in San Mateo County California.

The ballots were unsealed and five election officials began to organize the ballots. And as required by procedures, separate out the ballots with questionable ballots.  When asked for questions or objections, I stated that I thought that the procedures required that two opposing officials check each ballot and agree on which had questionable votes.  Alice said I was reading the procedures incorrectly and that she had checked with the Secretary of the State’s office and only one person checking each ballot was OK. I quickly scanned the procedures and found this:

The recanvass officials of opposing political parties shall remove all other ballots in the ballot transfer case (except any ballots marked “spoiled ballots” from a polling place in which the marksense machine was used for polling place voting).  They shall  examine all these ballots which were machine counted on election day to determine whether the markings for the office being recanvassed are sufficiently clear to be read by the machine.

Reading this I saw it was ambiguous, yet I am still convinced that the intent would seem to be that two officials review each ballot.  In any case, I did not further object.  However, reviewing the procedures later at home, I reviewed the complete paragraph:

The recanvass officials of opposing political parties shall remove all other ballots in the ballot transfer case (except any ballots marked “spoiled ballots” from a polling place in which the marksense machine was used for polling place voting). They shall examine all these ballots which were machine counted on election day to determine whether the markings for the office being recanvassed are sufficiently clear to be read by the machine. (See examples of properly and improperly marked ballots in this handbook as a guide) Also, if a stickered race is being recanvassed, make sure that early absentee ballots issued without the corrected name are not machine counted. If any such error or defect is found, the ballot should be set aside for hand counting of the races involved in the recanvass. If two recanvass officials of opposing political parties agree that such ballots are sufficiently clear to be read by the machine, such ballots shall be processed through the machine. When finished, open the write-in bin, handcount any write-in race that is subject to the recount, and enter it on a “recount write-in bin” line.

At first reading it would seem to me that the underlined sentence would imply two officials concurrence on the machine readability of each ballot.  Yet, there is ambiguity, as this sentence could be interpreted only to apply to the stickered ballots, if any.

One good result was that in asking my question, the official observers realized the value of having two individuals check the marks on each ballot for machine readability.  So, before Alice fed each ballot into the scanner, to observers looked closely at each ballot – in the end they added thirty-two more to the originally pulled 59 ballots.  (altogether there were about 650 ballots.  Having a single polling place, Cromwell scans absentee ballots into the same scanner on election day, so they were not segregated in the recanvass.

Also upon that closer reading of the procedures at home, I discovered this step:

The recanvass officials must, in the presence of the moderator and town clerk, open the depository envelopes containing the discarded outer and inner envelopes and rejected absentee ballots. The recanvass officials must check all absentee ballot outer envelopes against absentee ballot inner envelopes and against the check list to verify postmarks, addresses and check list markings and to verify that the number of outer and inner absentee ballot envelopes is the same as the number of persons checked as having voted by absentee ballot.

I saw the moderator, Alice, display the sealed envelopes, yet I did not observe, but possibly could have missed the processing of the absentee ballot envelopes while other activites were underway. I resolve to do my homework in the future and read the entire Recanvass Procedures closely and become much more familiar with the process.  Just as I am a much better observer after nineteen post-election audits, I can  grow in my ability to observe recanvasses.  It would help to have my own checklist and another observer to make sure everything is done and double checked.

After the initial separation of ballots  five officials counted the number of ballots.  From my position in the peanut gallery (Archaic:  see “B&W TV” or “Howdy Doody”), it looked like there was little organized cross checking while counting ballots – I could tell that they were off – based on their worried expressions – they had missed a pile of perhaps 30 ballots they had set on another table!  Its not in the recanvass procedures, but I would rather see a process where two officials verify every step of the process and where ballots are counted in piles of 25, and piled in a standard way by separate teams of counters, not five people around one table with at least five piles of ballots.  Here is another instance where the post-election audit procedures go farther, yet even they do no articulate a detailed process for counting ballots or counting votes.

Here my opinion comes in.  I would like every part of the process observable; every part  double checked by two opposing officials; and blind, that is, so if there are differences no counting official knows by how much or the election night total until all the counting is over.  Most of that is covered in the post-eleciton audit procedures, yet without detailed procedures not well understood by many election officials.

After the votes were read through the scanner, Alice fed the election ender card and then produced the report tapes with the counts, just like election night procedure.

The next step was counting the remaining ballots by hand.  Alice read each vote as two officials made hash marks on separate sheets.  Two observers looked on.  Once again, there are no standards and eventually the process yielded a credible result.  I would have a 2nd official watch that the person calling the votes to  make sure they made no errors.  For the 1st half of the votes, the observers could not see the ballots as they were being read.  They  were able to watch the hashing closely.  In the 2nd half of reading the ballots, Alice read them off the pile, so  that the observers had the opportunity  to see the votes – in fact, on one ballot an  observer corrected one vote she had missed.  After the ballots were counted the hash marks were off by one for each of two candidates (here it is fine to share that information with the team – it has nothing to do with the comparison to the election day count).

So, the votes were read and hashed another time for the two candidates with unequal totals.  I would have counted in batches of twenty-five, rather than doing all 91 in one batch.  It would have taken more time, but it would have reduced the recounting time, since presumably only one batch would have had an error.  A small difference here but when counting several hundred ballots it would  have saved a lot of time, especially if there were errors in each full count.

Its even possible that I contributed to the difference in the 1st count.  I had never seen the type of ballot bag used before.  While the counting was occurring, I asked the registrar, Linda Duren, where it was obtained – she said it was purchased from Diebold – back at home, I found it for sale from our distributor, LHS Associates.  Linda was across the room from the counting and I thought we were whispering, but Alice asked me to be quiet as it was disturbing the concentration of  the count.  Obviously if she said it did, it must have interferred.  On the other hand, I recalled many audits with three or four teams around a conference table or in very close proximity, up to nine teams in about the same space.  I always wondered how they managed to focus over the din of every team calling the same names over and over.

The counts were added and the differences announced.  The tie vote for two candidates remained and there will be a run-off on October 6th.  Until the last municipal election cycle in 2007, ties were decided by a random drawing – but, one occurred in 2007 and subsequently the law was changed.  All the totals matched election night, except one that increased one candidate by one vote.  Other observers were quite surprised.  I assured them that it was normal to have small differences, and that increases were more typical than decreases, because often a few votes are not completed well enough to be read by the machine.  Earlier many had expressed the opinion that if the machine read it on election night it must be OK, as the machine rejects errors – I pointed out that, in Connecticut, it only rejects overvotes and completely blank ballots, undervotes are not rejected and there are reasons that incorrectly filled in votes are  not read correctly.  I found that the other observers quickly understood my explinations and the reasons why I thought two people should check each ballot.

One ofter aspect of the process, I would change.  The adding of the vote subtotals from the machine and the hand count was not transparent to me.  I saw no indication that two people checked the totaling , transcription involved, or that an observer checked that aspect of the process.  It is likely some checking did occur.  Without a more defined role in the procedures, and my incomplete familiarity with it, I was reluctant to do much more than observe from a distance.

Once more, Alice did a very good job of articulating the process and soliciting questions.  The officials worked effectively and deliberately together. As everyone was leaving Alice and Linda asked for my feedback. I acknowledged that I would change the procedures in several ways but did go into that.  I suggested they consider two people checking each ballot, batching ballots  for counting, having a second official check  the reading of the votes, and holding the ballots such that observers can see them as they are read.

Finally, this whole process took about one hour and fourty-five minutes.  For integrity purposes I would add one official to the hand counting team.  Added time in counting more carefully would likely be saved in reduced rework.  Judging from the time it took to feed ballots into the scanner, print scanner tapes, and hand count 91 ballots, doing a full manual recount instead would have added about fourty-five minutes to an hour to the day.  Well worth the extra effort in my opinion.


Leave a Reply

You must be logged in to post a comment.