The Voting Integrity Project, Inc. REPORT ON INVESTIGATION POLK COUNTY, FLORIDA COUNTY COMMISSIONER'S RACE November 5, 1996 Marlene Young (D) (incumbent and successor) Bruce Parker (R) (challenger) November 10, 1997 Table of Contents Executive Summary The system of checks and balances which should define a public election system can be likened to working clay on a potter's wheel. If the system becomes tilted in any given direction without being immediately corrected, the clay will become distorted and soon wend out of shape and fly dramatically off the wheel. In Polk County, Florida, the clay flew off the wheel in November 1996 during a Commissioner's election. This may not have been the first time. And it may indicate continuing problems with Polk County's election system. After a detailed review of all available evidence, VIP has concluded that a conspiracy to commit voting fraud did not occur in this election. VIP cannot, however, report that the election outcome, as certified by the Election Supervisor, is correct. Since the losing candidate has conceded the race, VIP has concentrated its efforts on identifying weaknesses in Polk County election practices which may have contributed to the dispute. The many counts/recounts; continued uncertainty about basic facts regarding the election, including which facts were presented to a Grand Jury convened on this issue; and the continuing threat of lawsuits related to this election; have produced a lack of confidence in the outcome. We may never know whether the final result of the election is, indeed, accurate. Although the written procedures detailed in Florida election law and the Elections Office training manual may be adequate under most circumstances, even the best election protocol can prove inadequate when put under the pressure of heavy turnout or complicated ballots. There may exist opportunities for knowledgeable individuals to interfere in Polk County's process. There were notable gaps in ballot security, and even those checks and balances which do exist may not have been detailed enough, or understood properly, by those in charge of administering elections. Although VIP found no evidence to support a conspiracy to commit fraud, it is in the best interests of Polk County to continue its efforts to determine, to the extent possible, whether fraud existed in this or any other election. The Voting Integrity Project, Inc. is a non-profit, non-partisan national organization established early in 1996 to address the growing incidence of voting fraud in the United States. It enjoys the support of Republicans, Democrats, supporters of various third parties and independents. VIP's mission is to educate and equip the American public to recognize and combat election fraud in their own communities. We work directly with citizens in formalized long-term state projects, or in short-term remedial programs to address flaws in election procedure and policy. We also conduct voting fraud investigations, where there is reason to believe the election process has been manipulated to an extent that could change an election outcome. VIP depends solely on individual contributions. However, we accept no contributions from requesters of VIP investigations. VIP has been organized and is operated as a 501(c)(3) tax-exempt organization, having filed its Application for Recognition of Exemption as such a tax-exempt organization with the Internal Revenue Service. Concern about the integrity of this election cross political party lines. VIP received numerous requests for help. The first request came from an official with the Republican Party of Florida. Several individuals (political orientation unknown) followed by telephone and letter to request our involvement. An individual in the Washington, D.C. area, who identified herself as allied with the Democrats, asked for our involvement. Finally, VIP, at the request of Polk County Commissioner Nancy Rouse Hedrick (District 2) , and by vote of the VIP National Board of Governors, agreed to conduct an investigation into whether there was evidence to indicate a conspiracy to commit voting fraud in the election in question. VIP investigated all the key allegations raised by all parties in this incident and addresses each in this Report. It is important to note that Commissioner Hedrick was a Democrat while serving on the Canvassing Board, and is now aligned with the Republican Party. The Canvassing Board was named in a lawsuit brought by the Republican Party of Florida regarding this election. At the time VIP decided to investigate, a Grand Jury had just been convened. Normally, VIP would not enter into an investigation under this circumstance, but some of the allegations presented to VIP included the potential for compromise of the normal legal channels of investigation. Following the Grand Jury presentment, VIP re-evaluated whether to continue its investigation. Concerned that many questions still persisted, VIP determined to continue its investigation. VIP retained the services of a former FBI agent whose career qualified him in the area of voting fraud investigation. He was also familiar with the Polk County area. The investigation was conducted from May - November 1997. Our investigator interviewed all principals in the case, with the exception of incumbent Marlene Young who did not respond to several attempts to contact her. Other VIP personnel interviewed secondary participants in the case. This Report is the result of over 600 hours of research, interviews, investigation and analysis by VIP personnel. Allegations from both ends of the political spectrum contended that the outcome of the November 5, 1996 election was in question either by fraudulent manipulation or error on the part of those tasked with conducting the election. The allegations included: ú Deliberate manipulation of the vote count by the Canvassing Board in order to deprive reelection to the incumbent; ú Deliberate sabotage of the vote counting machine by the jerry-rigging of a reporter's computer to the vote counting machine; ú Deliberate adjustments to the election equipment so as to cause the machines to kick out thousands of resolution ballots ú Conspiracy to insert fraudulent ballots into the court-ordered manual count; ú Conspiracy by partisan workers to re-mark thousands of "resolution ballots" in favor of Marlene Young. VIP was unable to corroborate any of these allegations to the level of conspiracy or deliberate fraud, in the course of its investigation. However, VIP did identify numerous areas where the opportunity to affect the outcome of this election existed, either by design or happenstance . VIP notes in the report those instances where there seemed to be confusion as to the law in critical areas relating to how the count should be conducted. In addition, VIP notes that it was supplied two apparently conflicting vote counts by the Supervisor of Elections, Helen Gienau, who admitted to our investigator that one set of numbers had been supplied to her by Commissioner Marlene Young and mistaken for the certified vote count. However, it is that very set which seems to more closely match numbers in the transcripts of the original machine counts and Ms. Gienau's own certification. That is why VIP supports the immediate release of all evidence presented to the Grand Jury, since this may clarify whether these differing vote counts are tied in any way to the changed outcome in this race. VIP offers additional conclusions and recommendations in this Report, and believes Polk County would be served by continuing scrutiny of operations at the Elections Office and review of its protocols and equipment system. VIP applauds the commitment of concerned citizens of Polk County to reach a level of understanding and confidence in its elections, and endorses recent actions to convene a non-partisan citizen task force to develop recommendations for an improved election protocol in Polk County. VIP is available, if requested, to consult with this task force. Background Polk County Florida uses optical scanning equipment manufactured by American Information Systems of Omaha, Nebraska. This equipment was purchased in 1988. The scanner is programmed by its manufacturer to read paper ballots marked by carbon pencils provided at the polls. Accurate reading of the ballot by the scanner requires that the marks be precisely placed in small designated places on the ballot. Signs reading "Blacken the Oval" are placed at all tables. All poll workers must be registered voters in the county, be able to read and write the English language, and complete the required classroom training provided by the Supervisor of Elections prior to each election. Poll workers are divided into three classes: precinct clerks, affirmation clerks, and inspectors. The precinct clerk acts as the chief election official for that precinct. All precincts are issued cellular phones and special call-in numbers so that they may be in contact with the Supervisor of Elections and the Canvassing Board whenever necessary. The procedure for opening the polls is standard. A clerk and pollworker of opposite political parties unseal the ballot boxes, count the blank ballots issued and certify they match those allocated. This is memorialized on a "Ballot Accounting Form." The ballots are printed in sequential numbers and are issued sequentially by the lowest number first. In addition, there is a "Spoiled Ballot Log" to document ballots which were unacceptable. This is completed and signed before the polls open. The Poll List is updated against the Absentee Ballot List to prevent duplicate ballots being cast. All unused ballots are to be properly secured during voting. The metal ballot boxes are opened, witnessed and re-sealed with seals signed by all poll workers. After the polls are opened, each voter is verified as to identity and address, and then asked to sign the poll list. Signatures from furnished identification and the poll list are compared. A ballot is issued and its number is recorded next to the voter's name. (This number can be removed at voter's request and is removed at the ballot box to protect secrecy of the vote.) Then the poll worker initials that line in the poll list. The worker demonstrates for each voter how to "blacken the oval" and advises the voter to use only those pencils supplied for that purpose. The voter is shown how to use a "sleeve" to protect the secrecy of his ballot choices. Procedures provide for numerous checks on voters that would halt the processing and require further deliberation by an elections official. Such checks include voters who have changed their circumstances (name, address); voters not on the list; voters mis-assigned to precinct; and voters who do not have appropriate signature identification. These voters are processed at an "affirmation" table where their status is verified by calling the Supervisor of Elections Office. All completed ballots are placed in the ballot box, which has its opening sealed at the close of polls and prior to transportation to the central counting location. Supervisor of Elections, Helen Gienau, was herself running for re-election and thus was unable to directly monitor the election counts. Ms. Gienau's staff, however, substituted. In addition, a three-person Canvassing Board was appointed to oversee the elections. The Canvassing Board consisted of Administrative County Judge Michael Raiden, designated Chairman; and two County Commissioners, Jerry Carter and Nancy Hedrick. It is the role of the Canvassing Board to certify the results of the election. The election was supervised by Judy Walker, an employee of the Election Board. Also present was Mr. Jeff Hintz, a former employee of the equipment manufacturer, AIS, contracted as computer technician for this election. The Elections Office provided manuals on how to properly conduct counts to the members of the Canvassing Board, but this was accompanied by little, if any, training on correct procedures on how to handle resolution ballots. The count was actually performed by individuals recruited for this purpose by the Supervisor of Elections. One of the new issue areas raised in the course of our investigation involved these individuals - the amount of training they had for conducting count operations; whether they were indeed comprised of bi-partisan pairs as claimed by Supervisor of Elections Gienau; and whether their status as patronage employees in any way compromised the integrity of the vote count. The ballot that day was a complicated one, including presidential, congressional, state legislative, county commission, school board, local constitutional officers, court officials and city officials, and several constitutional amendments. Depending upon the jurisdiction, the ballot ran 2-3 pages and allowed, in some areas, that voters write in names of their candidates. This complicated ballot preparation. Ballots were distributed to the election clerks for each precinct a week prior to the election. Following the close of the polls, ballots were transported to a central counting location where, supervised by the Canvassing Board and Elections Office personnel and witnessed by a group of candidates, party officials, press and other interested parties who were physically separated by an observation window, the count proceeded. Chronology of the Election of November 5, 1996 October 30, 1996 - Canvassing Board Meeting The Canvassing Board met to perform the Logic and Accuracy Tests on the ballot readers and tabulators. Absentee ballot envelopes were also reviewed at this meeting, with questionable ones set aside for later disposition. A meeting notice was published, and candidates were notified by certified letter as to time and place. November 1, 1996 - Canvassing Board Meeting One of the items on the agenda was review of questionable absentee ballot envelopes. November 2-3, 1996 - Preparation of Absentee Ballots The Elections Office staff worked through the weekend to prepare absentee ballots for counting. Tuesday, November 5, 1996 - Election Day Polls were opened from 7 a.m. to 7 p.m. Approximately 154,000 voters participated - After the polls closed, precinct clerks were required to complete all documentation of the ballots and sealed the ballot boxes for transportation to the central count. Allegations surfaced after Election Day that in at least 20 precincts the number of ballots listed on the ballot accounting form did not match the number of ballots counted by election workers in Bartow; that some forms were illegible, or did not add up to the number of ballots issued to the precinct. Discrepancies such as this are rampant throughout this case. The Canvassing Board reconvened at noon to complete the review of questionable absentee ballot envelopes. Ballot boxes were checked in and opened. At this point ballot accounting forms were removed and ballots were transported to the count room. The scanning machines were employed to perform an initial count. All ballots were run through the machine, with approximately 6,000 "resolution" ballots - where the voter had in some way rendered the ballot unreadable by the machine -- kicked out by the machine readers. At this point, the Canvassing Board was advised by the Elections Office Counsel on the procedure to be employed with resolution ballots. According to an interview with Commissioner Hedrick, they were advised that, wherever possible, a determination as to the voter's intent was to be made by a Resolutions Committee. This seven-member committee began working in bi-partisan teams. Where a determination could not be made as to intent, the ballot was passed to the Canvassing Board. Wednesday, November 6, 1996 By 2:00 a.m., 100% of the precincts had reported in. The machine count was completed at some point in the early morning hours. However, the resolution ballots were still largely unresolved. At this point, officials from the Supervisor of Elections Office recruited several county employees still on hand. According to witnesses, these additional individuals were given little training and worked singly, rather than in bi-partisan teams. By 4:00 a.m. the results of the first machine/resolution count (Machine Count #1) were: Parker - 71,985; Young - 71,955; Votes Not Cast ("No Votes") - 10,014; for a total of 153,954. This first count resulted in a margin of victory for Mr. Parker of only 30 votes, triggering an automatic recount. At this point, no ballot audit had been completed. Thursday, November 7, 1996 A review of the election returns that afternoon revealed that one precinct had a reported turnout in excess of 100%. At this point a ballot reconciliation report had still not been performed. Commissioner Hedrick left before the count was concluded and Commissioner Richardson was substituted. An automatic recount was conducted which reduced that margin to 19 votes (Machine Count #2): Parker - 71,984; Young - 71,965; 10,020 "No" votes for a total of 153,969 ballots cast. This reconstituted Canvassing Board voted unanimously to certify Parker as the winner by 19 votes. November 14, 1996 - Manual Counts A partial manual count was pressed for by Commissioner Young. Under law, the partial manual count would be comprised of six precincts, three selected by each candidate. However, Bruce Parker permitted Young to select all six precincts. The Canvassing Board conducted a partial manual count of six precincts. Results of this count indicated the margin had been reduced even further to 16 points. The Canvassing Board decided to request the State to come in and validate the machine tabulation software and the Board ordered a full manual recount. Workers were called back to conduct the count. Ballots were sorted into Parker, Young and No Vote batches and then resorted by precinct and category (machine ballots, resolution ballots, absentee ballots). In order to be tabulated, two teams had to agree on the count total for a given category/precinct. If no agreement could be reached, the mission passed to subsequent teams until agreement was reached. Each teams's results were recorded on tally sheets. November 16 - 18, 1996 Commissioner Hedrick provided a statement that on this day, two days following the partial manual count, "I stopped in to see how everyone was doing and saw that boxes were not taped, by [sic] were "dog-eared" (one flag up, one flap down). I noticed a 2-inch stack of ballots sitting alone on a table. When I asked about them, I was told that they simply must not have been put back in the right box and an elections worker put them back." A total of 43 resolution ballots are decided by the Canvassing Board. The count was completed the afternoon of the 18th. November 19, 1996 - Final Count Certification According to the Grand Jury presentment, the Supervisor's Office requested additional time to review the final manual count results, but was denied this request, and a review of the results by the Canvassing Board was commenced. At around 5:30, the Canvassing Board conferred with counsel to discuss procedures and then privately reviewed numbers for several hours. The presentment reported that before this review was completed, one of the Canvassing Board members announced that due to questions regarding the tally sheets, that member would be unable to certify the election. At that point, the Board proceeded by a 2-1 vote to let the certification of the results of Machine Count#2, giving the seat to Mr. Parker, stand. November 20, 1996 - Young requests Tally Sheets Marlene Young requested copies of the tally sheets used from the manual recount. November 21, 1996 - Young Protest Filed Mrs. Young filed a protest in Circuit Court for Polk County, alleging fraudulent counts. November 27, 1996 - Tally Sheets Received Young is issued over 1,000 pages of tally sheets for the manual by the Supervisor of Elections office. November 28-30, 1996 - Young Performs Audit of Tally Sheets Young compiles the results from the copies of audit sheets give her by the Supervisor of Elections. Her results show she wins by 16 votes. December 5, 1996 - Canvassing Board Meeting Called There is no quorum for this meeting. However, Judge Raiden instructs Counsel David Cardwell not to take a position on an upcoming conference on the status of the case, since he represents both the Supervisor of Elections and the Canvassing Board. A few days later, Paul Craft from the Secretary of State's office visits the Supervisor of Elections to inspect the voting system and software. December 10, 1996 - Status Conference on Case A status conference on the case is held with Judge Owens in Sarasota. A hearing to consider all motions is set for December 19, 1996. December 12, 1996 - Canvassing Board Meeting The Canvassing Board hires attorney Wayne Malaney to represent them. December 17, 1996 - Department of State Report Received A letter from Paul Craft of the Department of State is received reporting that the voting system had not been certified. December 19, 1996 - Hearing before Judge Owens Judge Owens dismisses the Canvassing Board from the protest. Special Master Samuel Goren is appointed to assist the Court. January 31, 1997 - Court-ordered Recount The procedures for the recount included a complete ballot reconciliation, with Young and Parker each agreeing to the results on a precinct by precinct basis. The court-ordered manual recount proceeded with drastically different results from all previous counts. This time Young emerged as victor with a margin of 18 votes: Young - 72,108; Parker - 72,090; No Votes - 9830, for a total of 154,028 ballots cast. These results reflect an additional 59 votes for which there is no explanation -- a number which is clearly sufficient to address the vote margin several times over. The Grand Jury presentment does not address this discrepancy. This count was witnessed by Samuel Goren, a Ft. Lauderdale attorney appointed as a mediator, and Helen Gienau, Supervisor of Elections. Judge Owens certified this final count. Mr. Parker conceded the race to Mrs. Young, and she was seated as Commissioner. March, 11, 1997- Grand Jury Convened On March 11, 1997 a Grand Jury was selected and convened to investigate allegations surrounding the election. First testimony was not taken by this Grand Jury until May 1, 1997. June 16, 1997 - Grand Jury Presentment Released to Public The Grand Jury concluded that "...the election process in Polk County is not broken." Although citing some deficiencies in process and making some minor recommendations, including the hiring of additional personnel at the Elections Office, the presentment criticized the Canvassing Board and the Supervisor of Elections, but found no evidence to support allegations of fraud. To this date, however, Polk County citizen calls for release of supporting documentation reviewed by the Grand Jury have not been granted. Analysis of Vote Counts VIP does not question the integrity of anyone associated with the various counts. However, one disturbing aspect of this investigation has been the difficulty in reconciling the different vote counts. Much of this discrepancy remains unexplained, and may be attributable to Polk County election procedures and to the specifics of how this count was conducted. Following are the results furnished to us by Helen Gienau, Supervisor of Elections on separate occasions. These were presented in two different formats and, except for the signed 3/21/97 certification, were not readily identifiable. Ms. Gienau distributed the results in Tables 1 and 2 to our investigator in July, followed by different results in Table 3 in August. The explanation offered was that she had confused the certified numbers with Marlene Young's numbers, also in the Elections Office computer. VIP wonders how such confusion could possibly occur, and whether it is appropriate for an Elections Office to be the repository of one candidate's data. This incident alone has continued to be a source of speculation among VIP officials and its investigator, and begs further explanation by Ms. Gienau. It is especially troubling because of the difference in how the ballots were attributed, the difference between the Canvassing Board supervised counts and the subsequent court-ordered count as well as the lack of explanation for such difference in the Grand Jury presentment. What is most troubling is that the first set of counts supplied to VIP, which Ms. Gienau later identified as belonging to Ms. Young's campaign, are identical to the numbers cited by the Grand Jury presentment, while the numbers supplied by Ms. Gienau and identified as the actual vote outcome vary significantly from the earlier numbers and from the signed certification by Ms. Gienau. We have yet to obtain an adequate explanation for the discrepancies, though such explanation may, in fact, exist. This does, however, raise the question as to which set of numbers were supplied to, and relied upon by, the Grand Jury. First Counts Furnished by Ms. Gienau Table 1 Machine Count #1 Machine Count #2 Complete Manual Count Gienau 3/21/97 Certification Differential Range Young 71,955 71,965 72,114 72,108 159 Parker 71,985 71,984 72,098 72,090 114 Margin 30 19 16 18 48 Total Votes Cast 143,940 143,949 144,212 144,198 272 No Votes 10,014 10,020 9,885 -- 135 Total Ballots Accounted 153,954 153,969 154,097 -- 143 Ballot Reconciliation Table 2 Ballots Issued 231,039 Ballots Used 133,040 Ballots Unused 85,961 Ballots Spoiled 993 Write-ins and Resolutions 11,045 Total Votes Cast 144,085 Second Counts Furnished by Ms. Gienau Table 3 Machine Count #1 Machine Count #2 Complete Manual Count Gienau 3/21/97 Certification Differential Range Young 64,913 64,922 65,066 72,108 7,195 Parker 63,814 63,807 63,919 72,090 8,283 Resolution Ballots 8,883 8,894 8,699 -- 195 Total Votes Cast 137,610 137,623 137,684 144,198 6,588 The discrepancies between Tables 1 and 3, and within Table 2 raises a number of questions which remain unresolved. The numbers in Table 1 are the same as those delineated in the Grand Jury presentment and identified as unofficial hand tallies. They also match those numbers in Ms. Gienau's 3/21/97 certification. These numbers were later identified by Ms. Gienau as "Marlene Young's numbers." At that point, the numbers in Table 3 were supplied to us as the actual final certified vote counts. These numbers do not fit any reported in the press, the Grand Jury presentment, nor the Canvassing Board transcripts. VIP remains questioning of the final certification numbers. We have heard no plausible explanation to date for the sudden increase in the final numbers by 59 votes. We are struck by the careful recount performed by the Canvassing Board wherein the total number of votes was within two of the original count, and discrepancies were identified by precinct, with only two precincts identified with a discrepancy over 10. That recount confirmed the election outcome of the original count. An additional point about discrepancies. Military ballots were apparently lumped in with all absentee votes. The transcript suggests that these military ballots may initially have been counted as to state and local races. Page 129- 130 of the November 7 transcript implies that only the 17 late arriving military ballots were correctly tabulated for federal races only. Allegations of Deliberate Manipulation by the Canvassing Board This was the easiest allegation to dismiss, since all Canvassing Board meetings, including the counting sessions, were conducted with a court reporter present and in front of numerous witnesses of every political persuasion, as well as reporters. VIP found no suggestion of any motivations, other than civic duty, present during these sessions. If fact, we found the perseverence of the Canvassing Board, and non-partisan attitudes evident, truly striking in light of the deficiencies of the system which they were tasked with addressing. Finally, the Canvassing Board members had neither the method, means or motive to swing this election in favor of one candidate or the other. Allegations of Tampering with Ballot Tabulators The ballot readers transfer their counts to a tape cartridge which is itself transferred to an accumulator which tabulates the results. In an effort to assist rapid reporting of the results to the media, an additional computer was attached to the accumulator. It was this non-certified change in election equipment which triggered an allegation that such "jerry-rigging" had affected voting tabulations. The grand jury, however, investigated this and declared in its presentment that the ballot counts were not interfered with by this change. Elections office personnel were apparently unaware that such a change de-certified their equipment according to state election law. In a letter dated December 17, 1996, responding to an inquiry by Canvassing Board Chairman Raiden, the Florida Department of State Computer Audit Analyst, Paul W. Craft, delineates and explains the nature of the decertification of the Polk County machines. Mr Craft writes, "A voting system is a configuration of components. When we certify a system, it is certified as a particular configuration of components which we have tested and found capable of meeting the Florida requirements for voting systems. If the configuration or any of the components of the system are altered, the system ceases to be a certified system." One of the criteria of the State for the system is the inclusion of a software component called "Election Administrative Database (EAD)," which maintains the definitions of the political subdivisions and jurisdictions for ballot configurations. The State found that the EAD software had not been used to program the November 5, 1996 election. This was confirmed by Mr. Hintz, the independent contractor consulting on election equipment to the Supervisor of Elections Office, who told Mr. Craft he had programmed the election directly in the Election Programming System (EPS) software. This was an unauthorized modification. The programmable read-only memory chips in the Model 350 Accumulator were programmed using files which had not been components of the system as certified. This modification permitted the Accumulator to output results to an attached personal computer linked by serial port. This was an unauthorized modification. The modified file disabled the clock card and time stamp capability in the Model 350 Accumulator. This was an unauthorized modification. Mr. Hintz' status as an independent contractor may have necessitated these changed. He explained these modifications to Mr. Craft as necessary to produce a screen and menu item for communication on the Model 350 Accumulators, and to permit model distribution of results. Mr. Craft concludes in his letter that "Although Polk County purchased a certified voting system and possesses all the components of a certified voting system, the voting system used by Polk County to tabulate the November 5, 1996 General Election was not a voting system certified for use in Florida." With interest we note that Mr. Craft concludes that these modifications "...introduce(d) the possibility that the election results...could be modified on the attached personal computer," but that these departures "would not affect the logic and accuracy of vote counting or the results written to tape by the scanners." VIP has no evidence to indicate that deliberate sabotage was attempted or occurred because of these unauthorized changes in the system. However, it is troubling that such departures occurred and with such tremendous potential impact. Allegations of Deliberate Adjustments to the Ballot Scanners This allegation was one which had the most credence, since it offered a plausible scheme by which thousands of ballots could have been manipulated. The allegation was that the ballot scanners had been adjusted to cause them to be "less sensitive"in reading the ballots, thereby kicking out thousands of resolution ballots which required subsequent handling, interpretation, and marking by election workers or even Canvassing Board members. VIP notes that the AIS equipment in question has three sort options available as hardware switches -- blank ballots (under-votes), over-voted ballots (where more than one candidate was selected per office), and ballots that contain write-in votes. AIS provides these options in its equipment in order to meet a broad range of need for its customers, which have varying requirements based upon their state laws. It is up to the customer to determine which switch settings meet their legal requirements and need. When activated, the machine halts and indicates to the operator the reason for the ballot "kick-out." Depending on the ballot, and providing the appropriate switch is "on," the machine will read "SORT BLANK" "SORT OVERVOTE" OR "SORT WRITE-IN." The operator than removes the ballot and manually sorts for review by the Resolution Board. In its Election Procedures Manual for Administrators, AIS instructs that after all ballots are scanned, resolution ballots are run through the scanner a second time. This is called a "Resolution Pass" and all switches should be turned off during the Resolution Pass, except the SORT WRITE-IN switch. This is corroborated in the transcript of the November 7 machine recount. VIP cannot determine whether the allegations about machine settings affecting resolution ballots stems from this off/on procedure, but notes that the consultant, Mr. Hintz, was available throughout the election count, and would have presumably monitored the equipment for correct switch settings. Thus, VIP can find no corroboration for the allegation that machine settings were deliberately tampered with so as to force high numbers of resolution ballots. It remains a mystery, however, why there were such high numbers of Resolution Ballots in this election, as opposed to any prior election. We note that in the last presidential election, where the ballot was as complex, there were few if any resolution ballots reported. Allegations of Ballot Stuffing After the partial hand count, Commissioner Hedrick observed boxes of ballots sitting in the counting area open. The box tops were folded to close the top of the box, but they were not sealed. She advised she also saw a stack of ballots sitting near these boxes. Blank ballot tablets which she said were about 80% "empty" were also there. Commissioner Hedrick advised that this situation would be conducive to the fraudulent preparation of additional ballots, or "ballot stuffing." Additionally, allegations were made that the storage of blank ballots with executed ballots during the period of time prior to the mandated court recount resulted in a different outcome from the Canvassing Board-supervised count and recounts. The supposition was that security of the stored ballots had been violated by someone and that additional blank ballots had been marked for Marlene Young. The proof of this was, allegedly, that the vote totals mysteriously swelled. VIP notes the emergence of 59 additional ballots in the court-ordered hand count, following an extended period of ballot storage. Without questioning the motives of any involved in that final count, there has been no official explanation offered for these additional ballots. Though a reconciliation was performed in the Court-ordered recount, we are unsure whether it was a reconciliation of the number of votes cast per precinct registers against those counted per precinct -- the only type of reconciliation which would address this issue. Interviews with the Supervisor of Elections disclosed that executed ballots and the remaining blank ballots are still in existence and remain in storage as required by law. However, VIP has not had access to those materials. Allegations of Marking Resolution Ballots During the initial count on Election night, conducted by machine and supervised by the Canvassing Board, the machine would periodically reject a ballot because of inconclusive markings on the ballot. In other words, the machine failed to "read" the ballot. These ballots are known as "resolution" ballots, and are placed in a separate pile for further disposition. In this election, there were thousands of such resolution ballots. As VIP investigated this matter it found that a great deal of discretion was given to elections personnel in the treatment of these ballots, and that such discretion may fall outside the statutory requirements for treatment of resolution ballots. Each resolution ballot was reviewed by a pair of elections workers who constituted the Resolution Board. These individuals are supposed to review each resolution ballot to determine whether the intent of the voter is clear on the ballot. If it is not clear, that ballot is supposed to pass to the Canvassing Board for determination. For example, occasionally a voter will not sufficiently darken a circle as to enable reading by the machine. In those cases, the voter's intent is fairly obvious. In other instances, however, the mark may not be exactly on the circle, or is just a scribble and its meaning cannot be determined. Election protocol requires that in instances where the intent of the voter is clear, but where extraneous markings may interfere with the machine reader, white stickers are used to cover any extraneous marks and the ballot is to be re-fed into the counting machine. Where the intent of the voter is not clear, the Canvassing Board reviews the ballot. It is the Canvassing Board's prerogative to not count a ballot where the intent is not clear. This may not have been the actual process used in this election. Commission Hedrick described the process in a letter to Tom Lee, a Florida legislator, she stated: "At 7:00 p.m. we started counting the election returns and a table of four poll workers (most were experienced) was set up as the Resolution Board who corrected errors on the paper ballots if they were simple (such as a stray mark.) If they were unsure or if the voter's intent was difficult to ascertain, they asked us, the Canvassing Board, to render a determination (as mandated by Statute.) All of our determinations and criteria for their work were given verbally and recorded by a court reporter. Transcripts relate several factors that may have contributed to the unusually large number of resolution ballots in this election. Elections Office workers seem to have been engaged in making initial interpretations of voter intent and the re-marking of ballots in order to facilitate their being read by the scanners. Untold numbers of "over votes," where the voter marked more than one candidate for an office, were "corrected" with white adhesive and passed through the machine. This is another opportunity for "interpretation," which has not been addressed. The Canvassing Board members remark in the transcript of the November 5-7 machine counts that relatively few of these resolution ballots were brought to them for a decision. VIP notes that one possible source for misunderstanding proper procedure, may be the Election Procedures Handbook supplied by the equipment manufacturer. Florida Election Statute 101.5614 Canvass of returns mandates a different procedure than that utilized in the Polk County election: "...(5) ...If any paper ballot is damaged or defective so that it cannot be counted properly by the automatic tabulating equipment, the ballot shall be counted manually at the counting center by the canvassing board. The totals for all such ballots or ballot cards counted manually shall be added to the totals for the several precincts or election districts. No vote shall be declared invalid or void if there is a clear indication of the intent of the voter as determined by the canvassing board. After duplicating a ballot, the defective ballot shall be placed in an envelope provided for that purpose, and the duplicate ballot shall be tallied with the other ballots for that precinct. "(6) If an elector marks more names than there are persons to be elected to an office or if it is impossible to determine the elector's choice, the elector's ballot shall not be counted for that office, but the ballot shall not be invalidated as to those names which are properly marked." By all accounts, there may have been many times the margin in this race of resolution ballots where the intent of the voter was not clear, but the ballot was marked and re-fed through the machine and counted for one candidate or the other. The confusion about proper treatment of such ballots was clearly and repeatedly reflected in the transcripts of Canvassing Board sessions. Despite protestations by Ms. Gienau and members of her staff that no ballots were marked, there are several instances in the transcript where Ms. Gienau and others speak of marking these resolution ballots. VIP is aware that a libel suit has been filed against a citizen who raised this point in a Letter to the Editor. Though no individuals were named in the letter, two elections office workers have pressed their legal case. At the risk of similar actions, VIP considers this a legitimate area of inquiry. Several parties to these events have also contended that there was no attempt to have bi-partisan teams performing work on the ballot. Although Ms. Gienau now maintains that such bi-partisan teams are not a requirement of the law, in the transcript for the November 7 Canvassing Board meeting Counsel Cardwell specifies the importance of having bipartisan teams review the resolution ballots. Commissioner Carter points out that as the evening count wore on, additional people were added to these teams, and that some were not members of the Supervisor of Elections staff. Our attempts to obtain lists of the workers and their party affiliation were repeatedly frustrated by Ms. Gienau, who finally told our investigator that such lists did not exist. This is contrary to indications in transcripts of Canvassing Board meetings. Finally, we are impressed that the Polk County election system itself requires extraordinary input by scores of relatively untrained, but well-meaning, individuals. Each time a ballot in this system is handled, it degrades for future interpretation. As Canvassing Board Chairman Raiden expressed it, "You know, it's comical sounding; obviously not for the people that have a stake in the outcome of this, but we saw a lot of very peculiar things. And it's - nothing really prepares you, either computer training or legal training, to make these kind of decisions sometimes." Thus, thousands of ballots were counted where the intent of voter may not have been clear, or where such intention was determined inappropriately by election workers, rather than the Canvassing Board. With so much handling of the ballots, and differing interpretations of questionable ballots, no one will ever really know the correct outcome of this election. Conclusions On two significant occasions during this investigation, conflicting information was supplied to us by Ms. Helen Gienau. VIP was able to corroborate through unofficial sources that the first set of numbers supplied by Ms. Gienau, which match those given and cited by the Grand Jury, match the tabulations generated by the computer and supplied to media and others. However, without the AIS machine audit tapes, no one can substantiate that these were, in fact, the machine totals. The substitution by Ms. Gienau of the originally supplied numbers with a different set, and the characterization of the original counts as Ms. Young's, begs clarification. We question the reason and propriety of storing any numbers, other than official counts, in the Supervisor of Elections computers. There is some evidence that would suggest these numbers were generated by and used to support Marlene Young's protest of certification. Another explanation could be that the second set of numbers is the correct one, and the grand jury and others have been supplied the wrong set, either inadvertently or by design. We are left with the disturbing question of whether the final numbers relied upon to overturn this election were correct. VIP urges Polk County citizens to press for further investigation on this point by the appropriate officials. Paper ballots must be easily read, easily marked and their security tightly managed. This is especially important with the post-forty generation. Polk County is a retirement community with many senior citizens who comprise one of the most active voter classes. "Decentralization," where initial counts are performed in the precinct by the same election workers who have monitored the voting, result in fewer allegations of "ballot stuffing." Such a safeguard would require a change in Polk County election procedures, but would also provide a layer of audit trail currently lacking. In addition, separation of the ballots from the ballot reconciliation forms before an independent reconciliation is performed provides an opportunity for precinct ballots and fraudulent ballot reconciliation forms to be incorporated at the central location. Polk County's counting machine system should be reassessed. A paper ballot system is not in itself a problem. In fact, electronic systems may pose different types of security issues. VIP and many others critical of the move across the Nation toward electronic tabulation note that despite certification of machines, questions abound relating to portions of programs that are held as proprietary, and therefore not subject to scrutiny. VIP questions the necessity for release of ballots to precinct workers a full week in advance of the election. With 152 precincts, how can Polk County ensure ballot security in this process? Although the training required of poll workers may be sufficient, an additional emphasis needs to be placed on training of all elections personnel, including the Canvassing Board members, who are involved in the counting process. This should be emphasized even when the Supervisor of Elections is available to oversee the count. VIP notes that ballots remain in storage for this election. Given the circumstances, that is wise. However, under normal procedures, witnessed destruction of unused ballots as early as possible in the process, or segregation of unused ballots from executed is recommended. Destruction of any remaining, unused ballots is authorized under Florida election law. The benefit in so doing is to obviate allegations such as those generated by the storage of executed and blank ballots in this election. In summary, we may never know the truth about the actual vote counts in this election. However, there are remedies available to the citizens of Polk County to ensure integrity in future elections. Recommendations 1. Re-evaluate the machines and system used by Polk County, including requesting the Florida Legislature to provide for an independent audit of the Polk County voting system. If it is determined that the system is inappropriate for the size of Polk County, or the machines are no longer functioning properly and cannot be repaired to do so with some certainty, replace the election equipment used by Polk County with a more appropriate system. 2. If current equipment system is retained, revise Polk County's ballots to make them easier to fill out. 3. Decentralize the ballot counting process or amend to include a precinct count by the poll workers before ballots are transferred to a central counting location. Deferring the count to a central location invites fraud or allegations of fraud. 4. Training protocols for all election personnel should be reviewed and revised to ensure adequate training of all persons involved through the certification process. 5. Stop practice of using patronage county employees to perform election duties. This is a clear conflict of interest and invites fraud or allegation of fraud. Utilize paid citizen volunteers who have been certified as to training. 6. Change designation of elections workers to require they be residents of the precinct in which they live. This one change alone will raise confidence in election outcomes. 7. Ballot security is the primary area where vote fraud can and does occur. There were several breaches of ballot security in this election related to existing security protocol that was not followed. Stricter compliance to existing security needs to be assured. In addition, the practice of early release of ballots to precinct workers should be stopped. It is virtually impossible to ensure ballot security with thousands of ballots sitting in car trunks, on kitchen tables, etc. for up to a week prior to the election. 8. There needs to be established a detailed written and published protocol on how resolution ballots will be determined in the future. Voters should understand in advance of their marking a ballot how their ballot will be interpreted. There should include a clear prohibition on any marking of any ballot by anyone other than the voter of that ballot. 9. Investigate the release of separate vote counts by Supervisor of Elections, and cease the practice of storing counts, other than official counts, in the computer. 10. Release all documents supplied to the Grand Jury. nee, Caldwell. Letter from Nancy Rouse Hedrick, Commissioner, District 2, Polk County Board of Commissioners, April 28, 1997. Each poll worker is required to undergo 1-1/2 to 2-1/2 hours of training. Normal procedure provides for all ballots to be physically checked by the Supervisor of Elections Office prior to distribution to the poll workers. In this election, however, Ms. Gienau admitted that with 200,000 ballots, such a check was not performed on all ballots. This resulted in overlooking a printing error that resulted in mis- allocation of ballots between two precincts. This was ultimately identified and rectified by the time of the recount. Polk County Poll Worker Training Manual, Elections `96, Helen B. Gienau, Polk County Supervisor of Elections. This election had unprecedented high numbers of resolution ballots - thousands versus a mere handful in previous elections. Absentee ballots are date and time stamped upon receipt and are accepted up to 7:00 pm on the day of the Election. Any absentee ballots received after that point are logged in as to date and time of receipt but are not counted, with the exception of military absentee ballots from overseas which may be counted if postmarked by Election Day received within ten days following the election. There were 21 such ballots received between Election Day and the mandatory recount. In addition, it is important to note that military "advance" ballots - those that are mailed in advance of the runoff primary, need to have their selections transferred to a regular ballot for feeding into the ballot readers. This is a labor intensive effort, estimated at 10 hours for this election, and which could result in error. The Canvassing Board transcripts reflect an initial misunderstanding as to whether military ballots were counted for state offices. In the first count the transcript reflects that non-federal offices may have been counted. By the time of the recount, that issue was clarified to reflect that federal and Florida law permit only federal offices to be counted for military absentees, including advance ballots. This raises the question as to what, if any, effect this may have had on vote count discrepancies. "Polk Ballot Counts Disagree," Rick Rousos, The Ledger, 2/14/97. The Canvassing Board operated with a court reporter present at all times. VIP reviewed transcripts of these sessions of the Board. Transcript of the Canvassing Board, November 7, 1996, pg. 11-14. This section contains a discussion of whether the resolution ballots had been properly treated by the Resolutions Committee and Canvassing Board. This discussion reveals substantial confusion on the part of the Canvassing Board and other officials present as to what the proper standard is regarding resolution ballots and interpreting the intent of the voter. Counsel Cardwell offers that there is a base standard in the statute that "...no vote shall be declared invalid or void if there is a clear indication of the intent of the voter as determined by the Canvassing Board." However, Mr. Cardwell maintained in this discussion that the bulk of these determinations could, and should, be made by the elections office personnel. Vote margins of less than one half of one percent result in automatic recounts. There were actually two races close enough to trigger mandatory recounts. The second was the school board race between Jim Miles and Edith Yates. Several other races were close calls, though outside the threshold for automatic recount. Two of those candidates wrote the Canvassing Board regarding their concern about the resolutions ballots. Both letters echoed each other, as articulated by the Republican candidate in the Tax Assessor's race, "My concern is for the ballots referred to as resolution ballots that were interpreted after the count of all precincts and absentee ballots. These resolution ballots have created a great deal of skepticism throughout the voting community. In my opinion these ballots were open to interpretation and potential contamination and should not be considered in the final tabulations." (Emphasis added) "Tally turns up unregistered voting," Jennifer Ellis, Tampa Tribune, February 14, 1997. It is contended by one Republican official that Young picked six precincts where she was favored, an acceptable strategy. This fact is worth noting since, despite that, Parker maintained a lead over Young. If accurate, this is hardly the same impression of ballot treatment given to our Investigator by Elections Office personnel. In an interview with Judy Walker on May 14, 1997, our investigator was shown a room containing the boxes of ballots, with all of the doors, windows, doorknobs and locks sealed with evidence tape. In a later interview with Supervisor Gienau, she stated that initially the boxes were sealed along the opening of the flaps. However, after that initial sealing, the boxes were sealed at a right angle to that with one piece being placed over both flaps to the right and the left. These tapes were then initialed and dated by the individual who taped them. Our investigator, however, was unable to verify these accounts, due to the lack of access to the ballot storage area at that time. In Re: An Investigation Into the 1996 Election for Polk County Commissioner, District Three, In the Circuit Court of the Tenth Judicial Circuit of Florida In and For Polk County. These counts were furnished by Ms. Gienau in response to a request for the final certified vote counts for this election. They actually conform to the numbers cited in the Grand Jury presentment as having been tabulated from the manual count tallies. The transcript of November 7 recount reflects slightly different numbers - Young, 71,958; Parker 71,974. The breakdown as to total type of vote for this race is: Precinct, 117,218; Absentees, 5,220; Write-ins, 6,020; and Resolutions, 5,474. (pg. 133) Certification of the Court-ordered manual recount. These numbers were part of the same print-out supplied by Ms. Gienau as reflect in Table 1. We have broken them out separately in order to reflect that these numbers, as totaled, do not conform to total votes cast. This is a good example of the inconsistency of accounting throughout this election. November 7 transcript, pg. 153. Letter from Mr. Todd V. Urosevich, Vice President, Election Services, American Information Systems, Inc. to Commissioner Nancy Hedrick, October 29, 1997 Pg. 88-89. Chairman Raiden offered this wisdom during the November 7 recount: "And incidentally, you know, handling these ballots every time [at] some point there is a law of diminishing return to handling these things. The more times they're run through the machine the more likely they are to wear out or tear. Over votes - erroneous votes where the board determines they scratched out, they voted for this one, changed their mind, voted for this one - at any rate, they use little adhesive dots or ovals paste over what they don't want the machine to read. And my thinking is it could be that putting these in the box with the ovals, maybe some of the ovals came off and that might explain some of the discrepancies. Recognizing that the machine tonight is not programmed to pick up over votes, so it's not going to kick out an over vote. So that could explain some of the discrepancies. The little ovals -- that's my personal guess. That is not anything official." (p. 149) "The AIS Election Procedures Handbook for Administrators," pg. 16, American Information Systems, 1997, states: "If State Law requires you to make voter intent decisions, all ballots that are rejected by the scanner for containing overvotes must be inspected by the Resolution Board. The Resolution Board must use its best effort to determine the voter's intent." Later in the same section: "If the Resolution Board cannot agree on the voter's intent, the ballot must be deposited in an envelope provided for that purpose and returned to the election supervisor. (Emphasis added) It is clear AIS intended this book as a manual on its equipment, not election procedure. However, it is plausible that this book was relied upon by election workers in Polk County as protocol for the treatment of Resolution Ballots. General understandings of those present at the November 7 meeting were reflected in the statement by Chairman Raiden that, "They're open to interpretation. But the ability to make the interpretation and the legal responsibility for making it is vested in the Supervisor's personnel, the resolution board and the Canvassing Board. That is unavoidable in my judgment." (pg. 75) This position was reiterated by Counsel Cardwell, "Your [The Canvassing Board] authority when it came to the resolution ballots was when the ballots were brought to you, when the two people working on the resolution could not come to an agreement on the voter's intent." (Pg. 79) This reflects the general acceptance that the Resolution Board (i.e., the two individuals reviewing the resolution ballot) had the first level of discretion in interpreting the voter intent, when actually the statute is clear -- they have no discretion to interpret. If the voter intent is not clear, the Canvassing Board decides. What is lacking is detailed written protocol available to all as to how resolution ballots will be treated. See Transcript, November 7, 1997, pg.26, 35 Ibid, at pg. 20. Ch.97-13, 38, 101.591. Voting system audit. The Voting Integrity Project, Inc. 1 Polk County, Florida PO Box 222034 Investigation of County Commission Election Chantilly, VA 20153-2034 November 10, 1997 The Voting Integrity Project, Inc. 19 Polk County, Florida PO Box 222034 Investigation of County Commission Election Chantilly, VA 20153-2034 November 10, 1997