UF-trained political scientist suggests ways to improve the system ...
Florida's pre-election voter-roll purging: Threats to voting rights
Deb Cupples
September/October 2001

Thousands of Floridians' voting rights were threatened before the 1998 and 2000 elections. Threats stemmed from the state's voter-list-maintenance (purge) program, which was supposed to remove ineligible voters from the Central Voter File:

Florida statutes required the Division of Elections (DOE) to conduct the purge with help from county election supervisors and a private firm. Contracts cost the taxpayers over $4 million.

Despite the price tag, similar errors recurred from 1998 through 2000, involving non-felons who were misidentified as felons or actual felons whose rights had been restored:

Among others, a county judge and an election supervisor were misidentified as felons. Many counties continued to use the flawed purge lists; some refused. Error totals are incomplete, because not all counties have reported the number of people erroneously purged.

Two causes of errors are poor data quality and the DOE's choice of approach to purging. Errors could have been prevented had the DOE followed certain statutory requirements or insisted that the private firm do so. Statutes aside, the DOE should have chosen a cautious approach from the start, if only because the U.S. Constitution, U.S. Supreme Court and various federal laws seek to protect voting rights.

Based on DOE data, Florida's purge program seems to be going strong in 2001. The state and federal governments should investigate this matter until a clear understanding is reached. Florida should consider the recommendations below for improving its program:

Thousands of citizens' voting rights were threatened before the1998 and 2000 elections, due to Florida's voter-list maintenance (purge) program. Chapter 98 of the statutes established Florida's Central Voter File and guides the purge program. The legislation stemmed from concerns over voter fraud in Miami's 1997 mayoral election.

Had the Department of State's Division of Elections (DOE) more closely followed certain statutory provisions, errors could have been prevented. The failure to follow said provisions has not yet been explained. Statutes aside, the sacred standing of voting rights in the US should have compelled extreme caution on the DOE's part.

The central issue arising from the flawed program is voting rights, which federal laws and the US Supreme Court have repeatedly sought to protect. Today, Florida's purge program seems to be going strong. As well as continued investigation, this article recommends specific changes that should be made to more effectively protect citizens' voting rights.

The Central Issue: Voting Rights
There is no question that voting rights are or should be regarded as sacred in the US. Beginning in 1870, the US Constitution expressly recognized citizens' right to vote (see Amendments XV, XIX, XXIV and XXVI). 1 Through a string of decisions starting in1915, the Supreme Court has repeatedly protected voting rights [e.g., Guinn v. United States (1915), United States v. Mosley (1915), Reynolds v. Sims (1964)]. 2 The Voting Rights Act of 1965 (VRA) protects the rights not only of minorities but also of all citizens. 3

One expressed purpose of the National Voter Registration Act of 1993 (NVRA) is "to establish procedures that will increase the number of eligible citizens who register to vote in elections for Federal office." 4 The NVRA also stresses non-discriminatory methods in voter-roll purging.5 The VRA and NVRA list criminal penalties for certain violations. 6

Florida's Purging Process
In 1998, the Statutes required the DOE to maintain voter rolls with help from a private firm and county election supervisors.7 The goal was to remove ineligible voters including those who had died, moved, or been convicted of felonies and had not had their civil rights restored. Florida was the only state to require private-firm involvement. It was a $4 million, multi-year contract: a sum that some data specialists consider high given the tasks involved.

In July 1998, a contract between the DOE and a Tallahassee firm was terminated. In November 1998, the DOE (under Sandra Mortham) contracted with another private firm, Database Technologies (DBT). In December 1999, (under Katherine Harris), the DOE renewed the contract. In May 2000, DBT merged with a company called ChoicePoint. That time line is based on a DBT Vice President's testimony before the US Civil Rights Commission.8

The first contract between DBT and the DOE lists the following "Disqualification Parameters" as grounds for purging: a voter who is deceased, who had registered to vote in more than one county, or who "had been convicted of a felony and has not had this (sic) voting rights restored."9 My copy of the contract mentions five phases but lists headings for only four.

Phase I required the DOE to give data to DBT to develop methods for identifying voters that should be purged. DBT was to submit a report describing methods, changes, databases used (including age of data), and an "assessment of the accuracy and effectiveness of DBT's procedures..."10

Phase II required the DOE to review the Phase-I report and notify DBT if they should proceed with the next phase, if accuracy should be improved, or if the DOE would terminate the contract. Phase III gave DBT 30 days to make DOE-specified changes and to assess the resulting accuracy.

Phase IV required data exchanges between the DOS and DBT, including updated copies of the central voter, death, felony convictions and clemency files. DBT had 30 days to identify which voters should be purged. Reports were supposed to include voters' names, counties, birth dates, genders, races, and social security numbers if available. The contract further states: "DBT will then verify the accuracy of data contained in the output files."11

Florida Statutes required county supervisors to verify ineligibility before removing voters' names. However, Florida Statutes' Section 98.0975(1) specifies who should be included on the purge lists given to supervisors: regarding felons, lists should identify each ineligible voter who "... has been convicted of a felony and has not had his or her civil rights restored." 12 As evinced by errors discussed below, the DOE repeatedly violated that provision of the Statutes by sending lists to county supervisors that included mis-identified non-felons or actual felons whose rights had been restored.

Errors Reported Before the 1998 Election
A few weeks before the first 1998 primary election, the DOS sent purge lists to county supervisors, who then had to verify whether some 50,000 people on the possible-felon lists were eligible to vote.13 Despite section 98.0975(1) of the statutes, there were numerous errors involving mis-identified non-felons or felons whose rights had been restored. For example Brevard's election supervisor reported to his County Commission that "the 4,800 names published in the paper was reduced to 1,100 after the election office talked to the State, and has been further reduced." 14 That suggests a more-than-70% error rate.

In August 1998, Bay County's election supervisor received a list of 455 voters identified as felons and mentioned errors including mis-identified non-felons and felons whose rights had been restored.15 In August 1998, Alachua County's supervisor reported that its possible-felon list had 727 names, was 40 pages long and less than accurate.16 Note: DBT seemed to be uninvolved before the 1998 election.

According to Hillsborough County's supervisor, the state had sent out a "fatally flawed" list earlier in 2000 and a "fairly inaccurate" one in June 2000. 17 The Polk County supervisor estimated that "as many as half of the reported felons in Polk County may have been wrongfully identified or had their civil rights restored."18

As election 2000 grew closer, Palm Beach County and Treasure Coast reported a combined 657 errors, including "people erroneously identified as muggers, burglars and car thieves." 19 Leon County's supervisor said that only 34 of 690 voters targeted for removal proved to be felons. 20 Miami-Dade reported that 6000 voters were notified but that the county does not know how many voters had successfully appealed their designation as felons. 21

Errors Reported After the 2000 Election
In 2001, the US Civil Rights Commission reported that "Non-felons were removed from voter registration rolls based upon unreliable information..." 22 How did that happen? According to DBT's James Lee, of 173,000 names on a purge list, some 58,000 matched the names on DBT's possible-felon list.23 Of that 58,000, the Associated Press reported "about 12,000 people were listed as having felony convictions in Texas and other states even though many had been convicted of misdemeanors." 24 Of that 12,000, DBT said that roughly 8000 falsely labeled "felons" came from Texas, and others came from Florida.25 This suggests a more-than-70% error rate. Gregory Palast, a columnist for London's Observer, found 714 errors similar errors involving people from Illinois and 990 from Ohio. 26

According to the Lakeland Ledger, Duval County's assistant supervisor said that many voters targeted for purging were "ticked off because some of them had never had so much as a parking ticket."27 Dixie and Washington Counties continued to find mistakes even after purge lists were revised.28 Only some counties ignored later lists due to earlier errors (e.g., Palm Beach, Broward, Leon and Duval). 29

Polk County's supervisor sent 1700 letters to voters over the summer of 2000: she said that 1342 didn't respond or couldn't be reached at their address, and they were purged. 30 It is unknown how many of that 1342 were eligible and ultimately allowed to vote.

The Deputy Supervisor of Orange County said that about one-third of the letters sent to voters were returned as undeliverable.31 Orange County did not know how many voters had complained, but those who did were given a law enforcement agency's phone number to appeal directly. 32 One official stated that Volusia County basically accepted purge lists at face value, did little to confirm accuracy, and generally did not inform citizens before deleting them from the voter rolls. 33

Possible Causes
One cause of errors is what DBT's Vice President George Bruder called a false positive. A county supervisor cites the following example of a false positive: a DBT list had matched a non-felon named "Christine" with a felon named "Christopher" who had the same last name. 34 This suggests that gender may not have been used to remove unlikely matches before lists were sent to counties. Notable examples of false positives include a Hillsborough County judge 35 and the Madison County election supervisor: both were on a possible-felon list, though neither had been convicted of a felony. 36

DBT's Bruder said that purge-list compilation "was as per specifications of the division of elections... They wanted false positives on search parameters to cast as broad a net as possible." 37 The age of records is another possible cause of errors. Orange County's Deputy Supervisor attributed the high rate of incorrect addresses to the age of information sent by DBT, "some of which was close to 20 years old, she said." 38

Again, roughly 12,000 errors involved records from Texas, Florida, Illinois and Ohio. According to the Texas Department of Safety, its conviction records included people convicted of misdemeanors; according to the Austin Chronicle, that Texas department had warned DBT that it seemed to make "unwarranted assumptions" about the data.39 According to DBT's Bruder, Florida law-enforcement agencies simply had flawed conviction records. 40

Errors involving Illinois and Ohio were different. Those states automatically restored felons' civil rights after sentence completion. Florida felons cannot vote without having rights restored in Florida, but this is not the case for felons who come to Florida with rights previously restored. In Schlenther v. Florida Department of State (1998), a Florida Court of Appeal stated: "Once another state restores the civil rights of one of its citizens whose rights had been lost because of a conviction in that state, they are restored and the State of Florida has no authority to suspend or restore them ..."41 In 1999, a Florida court upheld that ruling in Doyle v. Florida Department of State.42

The list of 8000 possible felons from Texas may have been partly corrected. In May 2000, it included 437 actual felons, but Texas had begun restoring felons' voting rights in 1997.43

The St. Petersburg Times reported that 4000 ex-felons were targeted for purging from states that restore rights after sentence completion: Ohio, Illinois, Texas, Connecticut, South Carolina, Wisconsin, New Jersey, Washington.44 It is unknown how many felons had been restored in those states before being purged from Florida's voter rolls.

The obvious question: why were records used from states that auto-restore felons' voting rights? According to DBT, "From the beginning, the DOE determined what information was to be used in the data matching."45 According to the Dallas Morning News, Emmett Mitchell (former DOE lawyer) said "We wanted these lists to be fairly broad and encompassing. It was never intended to be a cure-all that local supervisors were supposed to take on faith." 46 The obvious question: why did the DOE knowingly take a broad-net approach when citizens' voting rights were at stake?

According to the Nation, an August 1998 memo of the Florida State Association of Supervisors of Elections warned the DOE that it had wrongly removed eligible voters from the rolls. 47 That warning, alone, should have been grounds for the DOE to re-think its approach to purging. According to Brevard County's election supervisor, in 1998, he and other supervisors began contacting the DOE about the flawed purge list, because the DOE had failed to "compare that with the Division of Executive Clemency's rolls." 48

The alleged failure to compare purge lists with Clemency records seems strange, as the Department of State (DOS) likely had access to those records. According to its annual report, the DOS "Serves as depository for official orders and acts of the Governor, such as . . . executive orders of clemency . . . ." 49 Also, the 1998 contract between DBT and the DOE states: "The Division (of Elections) shall provide DBT with newly updated copies of central voter, death, felony convictions and clemency files on or before April 1, 1999. DBT shall have 30 calendar days to reprocess these files in order to identify records that meet the Disqualification Parameters."50 If those records were given to DBT from the outset, then purge lists should have been more accurate in 1999 and 2000.

Because of problems with Clemency records, some counties, such as Bay, planned to check alleged felons with the Office of Executive Clemency. According to the Gainesville Sun, Alachua County's supervisor had concerns, because "The people at the clemency board told us we could only check four names at a time and the first four we checked had all had their rights restored."51 That suggests a high error rate. In the two weeks before the first 1998 primary, more than 60 other counties also may have had to check Clemency records--four at a time. The supervisors also had to check voting machines, train poll workers and attend to other pre-election details. Essentially, supervisors encountered roadblocks to verifying and protecting citizens' rights.

In September 2000, Clemency wrote a letter to the Division of Elections in response to questions about restored felons. My copy of the letter states that restored ex-felons who did not show certification would be "required to make application for restoration of civil rights in the state of Florida." 52 That directive seems to conflict with the Schlenther and Doyle cases.

Purge-Related Data
I repeatedly requested various information, in writing, from the DOE about Florida's purging. I received only one response, dated February 26, 2001: it said that I could not have voter lists, because I was not a government official or candidate (per 98.095(2), Florida Statutes).53 That response failed to mention the NVRA 1993 and Florida Statutes Chapter 119. The NVRA calls for "public disclosure of voter registration activities."54 Chapter 119 states that if parts of a record are exempt from public inspection, the custodian "shall produce the remainder of such record for inspection and examination." 55
On April 13, 2001, during a personal visit, the DOE handed me a copy of Central Voter File on CD-Rom and a partial copy of its contract with DBT. In March 2001, I found the statistics listed below on the DOE's website, showing "new valid" registered voters and voters deleted from the rolls.

Florida's Registered vs. Deleted Voters 56

Date New Valid Deleted
1995 991,596 310,940
1996 1,180,050 298,052
1997 601,656 285,708
1998 618,681 350,738
2000 1,042,225 365,318
Jan-01 62,356 104,669

From 1995 through 2000, the number of people deleted from the voter rolls was between 25% and 56% of the number of new-valid registrants. In the first month of 2001, the number of people deleted was 167% of the number of new-valid registrants. The state's purging efforts seem to be going strong, but I do not know who is handling the tasks or how.

From the start, errors could have been prevented had the DOE followed certain statutory guidelines regarding the purge program. Instead, the DOE chose to take a broad-net approach. Regardless of the Florida Statutes, such an approach should never have been considered, given the aggressively protective stance that the US Constitution, the US Supreme Court and federal laws have taken regarding voting rights.

The initial choice seems understandable at the time, given the public's outrage over the 1997 Miami election. Moreover, in early 1998, the DOS may not have grasped the enormity of the lists or mistakes that a broad-net approach would produce. However, the choice to continue using a broad-net approach is not understandable. From August 1998 through June 2000, the DOE repeatedly received error involving mis-identified non-felons or felons whose rights had been restored. Because errors were similar and recurring, the DOE had reason to suspect that its own, broad-net approach was the cause.

At various points before Elections 1998 or 2000, the DOE should have tightened data-matching parameters and excluded criminal records from states that auto-restored felons' rights. As well as coinciding with various laws and Florida case law, such error prevention would have reduced county supervisors' workloads, better enabling them to verify voters' eligibility and protect their rights.

Whether errors resulted from contract specifications or the failure to follow them is irrelevant. The bottom line is that the DOE could have either restructured the purge program or terminated the DBT contract. In 2001, South Florida's Sun-Sentinel quoted Secretary Harris as saying "I inherited that contract. I had concerns when I first came in." 57 Why the contract was renewed is a mystery.

If voting rights are not fiercely protected, then people from all political parties could one day lose them. At this point, partisanship, election results and personal feelings must be put aside so that we can re-focus on protecting voting rights and the integrity of the electoral process. Below are some recommendations that Florida should consider.

Private Firm Involvement
The state should prohibit private firms from playing crucial roles in voter-roll purging. Because purging is required annually, contracts with profit-seeking firms may cost more than in-house servicing. Additional costs could result from errors, problems or legal battles. Comparative cost-analyses should be done. Also, involving a private firm hindered the ensuring of both public accountability and the public's access to records.

Any firm that profits from selling records-access and has access to state faces an inherent conflict of interest. Both the FBI and DEA recognized that conflict when terminating their contracts with DBT in 1999 over concerns that the firm could monitor ongoing investigations. 58 In January 2000, the Pennsylvania Department of Transportation halted its contract, because DBT had allegedly violated the confidentiality of that state's driving records.59 Another potential conflict of interest could arise from dealings with DBT specifically, as Florida's retirement system holds or held stock in ChoicePoint (which had merged with DBT).60

Inter-Governmental Cooperation
State, local and federal agencies could agree to share records relevant to voter-list maintenance. Data could be free of charge, since all participants would benefit. Those agencies could factor voter-list-maintenance requirements into their record-keeping methods, which could make all states' voter lists more accurate.

Funding of List-Maintenance Programs
Before Elections 1998 and 2000, county supervisors did not have adequate resources to check all of the possible matches on the purge lists. That was unfair to them and to citizens. The state should adequately fund its purge programs.

Barriers to the Public's Access to Information
The state and counties should give the public easier access to registration and purge lists. The records could be available as electronic files and at websites. Purging efforts and schedules should be better publicized, notifying citizens when new purge lists have arrived. Counties could make public-service announcements and write newspaper press releases or editorials. The state and counties could also use websites as bulletin boards: alphabetized purge lists, instructions and disclaimers could easily be pasted into HTML documents.

The recommendations discussed above would require resources, but the likely results are improved services and reduced long-term costs. Costs aside, voter-list maintenance involves constitutional rights, which should be handled with extreme care. As evinced by the NAACP's 2001 class-action complaint against Florida election officials and DBT, even unintended infringement on voting rights could bring severe consequences to the state and its citizens.61

1. US Constitution Amendments XV, XIX, XXIV, and XXVI.
2. US Supreme Court, Reynolds v. Sims, 377 U.S. 533, http://www2.law.cornell.edu, 1964.
3. Voting Rights Act of 1965, Section 11(a), Prentice Hall Documents Library, http://hcl.chass.ncsu.edu/garson/dye/docs/votrit65.htm, accessed March 2001.
4. 1973gg-6(b)(1), National Voter Registration Act 1993.
5. 1973gg-6(b)(1), National Voter Registration Act 1993.
6. 1973gg-10, National Voter Registration Act 1993.
7. 98.0975, Florida Statutes.
8. Bruder, George, Statement Before US Civil Rights Commission, Miami, Florida, February 16, 2001.
9. State of Florida Department of State Division of Elections and Database Techonolgies, Inc., Data Processing Services Agreement, signed November 24, 1998.
10. See State of Florida . . . Data Processing Services Agreement.
11. See State of Florida . . . Data Processing Services Agreement
12. 98.0975(1)(b),Florida Statutes.
13. Staff and Wire Report, State Voter Rolls, Newsherald, http://www.newsherald.com/archive/local/ld081998.htm, August 19, 1998.
14. Minutes of Brevard County Commission Meeting, http://crystalweb1.clerk.co.brevard.fl.us/minutes/m091598.htm, September 15, 1998.
15. See Staff and Wire Report, Newsherald.
16. Voyles, Karen, Area election officials check felon figures, Gainesville Sun, August 18, 1998.
17. Brune, Tom, Florida Felon Policy Tilted Outcome, Some Critics Say, Newsday, http://newsday.com, December 14, 2000.
18. Townsend, Billy, Almost 1500 Voters Tossed from Polk Rolls, Lakeland Ledger, December 15, 2000.
19. Gelbart, Marcia, Felon Glitch, Palm Beach Post, http://www.gopbi.com/news/2000/06/22/felonglitch.html, June 22, 2000.
20. St. John, Paige, Big Lie: Every Vote Counts, Tallahassee Democrat, December 17, 2000.
21. Palast, Gregory, Florida's Flawed "Voter-Cleansing" Program, Salon, http://Salon.com, December 4, 2000.
22. US Commission on Civil Rights, Status Report On Probe Of Election Practices In Florida During The 2000 Presidential Election, http://www.usccr.gov/vote2000/flstrpt1.htm, March 9, 2001.
23. Kidwell, David and Dougherty, Geoff, Florida Lists Didn't Stop Felons from Voting..., Dallas Morning News, December 10, 2000.
24. Associated Press, Florida Pays for Inaccurate Felon Lists, http://quest.searchcolorado.com, December 5, 2000.
25. See Bruder, George.
26. Palast, Gregory, A Blacklist Burning for Bush, Observer, http://www.observer.co.uk/Print/0,3858,4103063,00.html December 10, 2000.
27. See Townsend, Billy.
28. See Palast, Gregory, December 4, 2000.
29. See Kidwell, David, et. al.
30. See Townsend, Billy.
31. See Palast, Gregory, December 4, 2000.
32. See Palast, Gregory, December 4, 2000.
33. See Palast, Gregory, December 4, 2000.
34. See Palast, Gregory, December 4, 2000
35. See Brune, Tom.
36. Parker, Laura, Panel Scrutinizes Mistake-riddled "cleansing" of Voter Rolls, USA Today, January 15, 2001.
37. See Robinson, Andrea.
38. See Palast, Gregory, December 4, 2000.
39. King, Michael, Margin of Error, Austin Chronicle, http://www.Auschron.com, December 8, 2000.
40. See Bruder, George.
41. Florida Court of Appeal (Second District), Harry P. Schlenther v. Department of State, Division of Licensing, Case No. 97-04363, June 24, 1998.
42. Florida Court of Appeal (First District), Michael W. Doyle v. Florida Department of state, Division of Licensing, Case No. 98-4363, December 22, 1999.
43. Palast, Gregroy, Florida's ‘Disapeared Voters' . . . , The Nation, February 5, 2001.
44. Nickens, Tim and Goffard, Christopher, State Makes it Easier for Some Felons to Vote, St. Petersburg Times, March 31, 2001.
45. See Bruder, George.
46. See Kidwell, David, et. al.
47. See Palast, Gregory, February 5, 2001.
48. See Minutes of Brevard County Commission.
49. Florida Department of State, Division of Elections 1998 Annual Report, http://www.election.dos.state.fl.us/publications/pdf/98ANNRPT.PDF
50. State of Florida Department of State Division of Elections and Database Techonolgies, Inc., Data Processing Services Agreement, signed November 24, 1998.
51. See Voyles, Karen.
52. Governor's Office of Executive Clemency, Letter to Florida Division of Elections, September 18, 2000 (hard copy).
53 Kast, Ed (Asst. Director, Florida Division of Elections), Email to Deborah Cupples, February 26, 2001.
54. 1973gg-6(i), National Voter Registration Act of 1993.
55. 119.07(2)(a), Florida Statutes.
56. Florida Division of Elections, Florida Voter Registration, http://election.dos.state.fl.us/voterreg/vrchngs.shtml, March 2000.
57. Kleindienst, Linda, Democrats Seek Inquiry into Firm that Was Supposed to Purge Florida Voter Rolls, South Florida Sun-Sentinel, February 23, 2001.
58. Associated Press, DEA, FBI Suspend Online Contracts with Data Company, http://www.polkonline.com, 07/04/99.
59. Pennsylvania Department of Transportation, Press Release, http://insurance.state.pa.us, December 27, 2000.
60. NASDAQ, see symbol CPT, Holdings/Insider Reports, http://nasdaq.com (accessed April 2001).
61. NAACP, Case No. 01-CIV-120-GOLD, US District Court Southern District of Florida, January 11, 2001.

Deb Cupples received her Masters Degree in Political Science from the University of Florida in May for her research on voting roll purging in Florida. She spoke at the August 18 election workshop.

Rep. Lois Frankel addresses an election workshop at the Thelma Boltin Center in Gainesville, August 18. Frankel, UF Professor of Law Ken Nunn, political scientist Deb Cupples and several others addressed the crowd about the stolen election of 2000 and discussed plans for the 2002 elections.

previous article [current issue] next article
Search | Archives | Calendar | Directory | About / Subscriptions |

Valid HTML 4.01 Transitional eXTReMe Tracker