2004 Ballot Recount: Observer Report

• December 20, 2004:
  Report by Green Party Observer

December 20, 2004:
Report by Green Party Observer

The Director, Chris Heizer, and Deputy Director, Steve Harsman of the Board of Elections seemed to be fairly evenly matched and balanced in the power structure. All secure areas were under double lock and key, requiring both the Democrat and Republican counterpart to be present to access sensitive materials. Sensitive computer files are double password-protected in a comparable manner.

The counting room itself was restricted to staff, authorized witnesses, and credentialed media. Other observers were permitted to view the proceedings from an adjacent room through large glass windows. Teams of four Board of Elections counters staffed each of five counting tables. One Democrat and Republican pair counted while another Democrat and Republican pair kept the tally. In addition, the entire staff was helpful and well-trained.

However, there were certain aspects of the process that were somewhat disturbing. The foremost of these involved the handling of ballots. It seemed that the process was less an attempt to see if the machines could accurately read the punchcards and more of an exercise in ensuring artificially that the hand recount and machine tabulation of the same three percent sample were a match.

Hanging chads, which caused so many problems in Florida in 2000 and which still account for varying totals from election night to the recount, were physically and forcefully removed from ballots once the Board of Elections directors had examined and discussed a questioned ballot to determine the voter's chosen candidate. Of course the hand recount will match the computer tabulation of the same three percent sample if the ballots are prepped to ensure that the machine cannot possibly misread the chosen sample.

This resulted in a loss of one vote for Bush (and the consequent addition of one "overvote" — too many presidential candidates designated), and the pick up of seven votes for Kerry (due to the correct reading of punchcards that had previously registered as "undervotes" — i.e. no vote registered). On the other hand, when the remaining 97% of the ballots were run through the machine, Bush picked up 10 votes and Kerry picked up only an additional 12 for a net increase to the total margin of 10 votes.

If the change in the three percent sample is considered statistically relevant and is extended to the full 100% of the vote, it would result in a pickup of greater than 250 votes for Kerry. While this is miniscule when viewed in relation to the total of 287,646 votes cast in the county, it is still quite different from the projected change based on the three percent sample. This proves that the final count will indeed change with a full hand recount and that a full 100% hand recount is necessary if a higher level of accuracy is desired.

Finally, because of the fragile nature of the chads in the punchcards, the more times they are run through the counting machines, the more chads will come lose and more votes will be counted (either for a candidate or as overvotes), regardless of the desire to maintain voters' initial intentions.

I was quite happy to see the level of transparency in the Montgomery County Board of Elections office. The staff was willing and able to show us anything we requested and provide us with printouts of all critical data. We even were able to receive the list of rejected provisional ballots with each individual's personal information.

It was somewhat worrisome, however, that the spreadsheet containing this data (which was sent to me by email) included not only the person's name, address, and reason for acceptance or rejection of the ballot, but also the person's social security number and birth date. This is a serious security breach, in my estimation, since this data should not be distributed freely.

Despite the slip up involving the distribution of personal identifying information, it was apparent that the Board of Elections staff at least felt entirely confident in their intent to ensure a fair and smooth-running election process. Any problems were defects of the system or of the software and hardware, not of the individual citizens involved with the Montgomery County Board of Elections.

Additional notes: Despite clearly distinguishable voter intent, pursuant to Ohio law, "pregnant" or "dimpled" chads were not counted as votes for a candidate. These "undervotes" for which a candidate was designated and voter intent was clear heavily favored Kerry. In addition, one ballot was discovered which had been placed in the cardholder backwards. As a result, the punches (which did go entirely through the card) were from back to front and didn't line up with the chads. Although voter intent could have been determined by overlaying the reversed ballot with a new ballot and punching through the pre-existing holes, this was not done because Ohio law does not require such an action.

One final note is that the machines in the counting room are only networked to the computers in the same room. Neither the tabulating machines nor the computers are connected to any network outside that room or to the internet in any manner, according to the Board of Elections officials.

Additional Stats:

• Total Ballots Cast & Counted (including provisional and absentee): 287,646
• Out of 9257 total provisional ballots cast, 7376 were determined to be valid.
• The provisionals that were deemed not valid were found as such for the following reasons: 1068 not registered, 4 unable to determine identity of voter; 31 no signature; 11 double voters; 767 wrong precinct. While these numbers (particularly the "no signature" and "wrong precinct" category were acknowledged by the Board of Elections officials to be too high ("one is too many" according to Steve Harsman), the percentages were significantly lower than in previous elections.

While invalid provisional ballots are generally left unopened, those determined to be in the wrong precinct were opened because of discrepancies between the envelope designation of precinct and the ballot designation of precinct. In some precincts that ran out of envelopes (which bear a code distinguishing the correct precinct), envelopes were borrowed from neighboring precincts. While the envelope may not have been correct for a given voter, in certain cases the ballot itself was for the correct precinct and as such the vote was counted.