This paper was prepared by Eric Limbs, Fiscal Analyst with the Senate Fiscal Agency. Pat Stinton-Harper prepared and formatted the tables for the paper.

System Implementation

Funding Requests and Expenditures

Local Response and Department of State Reaction

CONCLUSION

During 1994, the Michigan Legislature authorized the development and implementation of a new information system to control elections across the State. This new information system is referred to as the Qualified Voter File (QVF). The goal of the QVF was to move Michigan to a more centralized system of record-keeping and tracking of voter activity across the State. There are currently 413 counties and local jurisdictions on-line with the QVF server in Lansing. These 413 sites support election activities of the 1,776 cities, townships, and villages across the State of Michigan. Public Act 441 of 1994 required the QVF to be operational and available for use in all elections on or after January 1, 1998. On a statewide basis, the QVF was available for the first time for primary elections on August 4, 1998, and for the general election on November 3, 1998. This paper provides an overview of the implementation of the QVF, appropriation history, and reactions to the progress of the system from sample jurisdictions and the Department of State.

System Implementation

All cities, counties, and townships are served by the QVF. Local school districts also utilize the QVF for their elections. Michigan's 83 county clerks as well as local jurisdictions with a voting age population over 5,000 were provided with computer equipment to link to the QVF file server in Lansing. Jurisdictions with a voting age population between 3,000 and 5,000 had the option of purchasing equipment or gaining access to the QFV through their county clerk's office. Jurisdictions with a voting age population of less than 5,000 were reimbursed for their assistance in the data validation process for the implementation of the QVF. The File was loaded with every registered voter appearing in the Department of State's driver license/personal identification card file and the voter registration files held by the State's city, village, and township clerks.

The primary components of the system are the main file server in Lansing, the county and local installations of the system, and the telecommunications network of the QVF, as described below.

1) The main file server holds the voter registration database for the entire State. The system also holds the system database software. Voter registration information is captured in an electronic format at branch offices and is downloaded daily into the QVF database.

2) A total of 545 personal computers (PCs) and 512 printers are installed in the State's counties and local jurisdictions. These systems were placed in 413 communities. Among these communities are the State's 83 counties and the 236 largest cities and townships. Ninety-four additional townships opted to purchase QVF systems at their own expense. Some of the State's larger communities have multiple PCs and printers.

3 The method by which voter information is updated in the QVF is a record replication process. Replication updates the Lansing server with new information provided by local jurisdictions and updates the local jurisdiction with new information provided by the Lansing server. This replication process is performed through an Internet connection. For this purpose, the Department of State provided each QVF jurisdiction with an Internet account and Internet software. The Internet serves as the system's telecommunication network.

Funding Requests and Expenditures

Public Act 242 of 1995 appropriated $7,598,300 for the implementation of the QVF (Table 1). In the subsequent General Government appropriation bills for fiscal years (FYs) 1996-97 and 1997-98, the Legislature appropriated $1,069,500 and $1,473,400, respectively, for maintenance of the system. In FY 1998-99, the Legislature appropriated $4,863,700 for maintenance and software updates to the system (Table 2). In his FY 1999-2000 budget recommendation, the Governor proposed $1,154,300 for maintenance of the system. To date, the Legislature has appropriated $16,159,200 for the implementation and maintenance of the QVF. According to the Department of State, approximately $1,000,000 per year will be needed to maintain the system.
 

Table 1
QVF Original Appropriation Cost Allocation FY 1995-96
Payments to Local Jurisdictions - Payment to local units for file verification costs and reimbursement to locals in the form of hardware and software for the implementation of the QVF system.	$3,482,289
Jurisdiction/Precinct Assignment - Payment to the Department of Management and Budget for the Department of State's portion for usage of the Geographic Information System.	2,000,000
Telecommunications - Cost for creation of the QVF's telecommunication network so that each system in the respective local jurisdictions would be able to communicate with the main file server in Lansing.	265,950
Developmental Costs - Cost for file server hardware and software development, training, and help desk.	1,850,000
Total	$7,598,239

The FY 1998-99 appropriation for the QVF includes a supplemental appropriation of $3,743,800 for improvements to the QVF. The supplemental appropriation is requested for various items needed to facilitate the operation of the system. Table 3 provides a breakdown of the expenditures proposed from the FY 1998-99 supplemental appropriation proposal.
 

Table 3
FY 1998-99 Supplemental Appropriation Request
Software and Database Improvements	$1,124,800
Digitized Signature Technology	250,000
Reapportionment and Synchronization of Street Index	1,365,000
Maintenance of Current Street Index	804,000
Help Desk Support	200,000
Total	$3,743,800

These include funds to update software and improve the database design, fund positions for the QVF Help Desk, purchase digitized signature-scanning technology for the system for more efficient voter verification, cover anticipated reapportionment costs and synchronization costs for the system, and maintain and update the current Street Index database for the QVF (Table 3). A description of each segment of the FY 1998-99 supplemental appropriation request follows.

• •Software and Database Improvements. Software improvements are needed to make the system more compatible with the needs of larger voting jurisdictions. The Department will upgrade to new and far superior versions of database and application software.

 
• •Digitized Signature Technology. The Department wants to implement digital scanning technology for the QVF to enable jurisdictions to identify a voter's signature more efficiently by scanning the image from an existing voter registration card. The ability to scan signatures digitally will facilitate the maintenance of records in the system. Digitized signatures can be used to verify signatures on petitions and absentee voter applications.

 
• •Reapportionment and Synchronization of Street Index. The Department of State will incur costs associated with district reapportionment based on the year 2000 census. Reapportionment will require updates to the QVF Street Index database. Further, the QVF Street Index will need to be synchronized with the Michigan Information Center's Geographic Information System (GIS) mapping program. The GIS is the base map for the State of Michigan. The GIS already has a street index; however, the index is not currently based on updates and corrections made by local election officials. Once these corrections are made, the QVF Street Index will be synchronized with the GIS program.

 
• •Maintenance of Current Street Index. Until the systems are synchronized in fiscal year 2000-01, the QVF Street Index will need to be maintained and updated in its current form.
• •Help Desk Support. The Help Desk provides QVF training to local clerks, phone support, and e-mail support. The Department intends to fill 3.0 additional full-time equated positions (FTEs) to provide training support, as well as provide adequate resources for a presidential election year when every jurisdiction has elections.

Local response to the QVF has been mixed. Generally, local units of government have stated that the Qualified Voter File is a good concept that still needs improvement to be a reliable product. The Department of State is working with jurisdictions to improve the system by correcting inaccurate voter information in the QVF, improving the efficiency of the absentee voter portion of the system, as well as updating and correcting the Street Index database of the QVF.

According to the Department of State, there may have been some discrepancies between polling lists generated by the QVF and lists generated by communities' old systems. The Department says that it eliminated 800,000 duplicate records from the electronic files originally submitted for the conversion process (the procedure by which the Department properly converts information submitted by local units to QVF format). The elimination of the these duplicate records would make the lists generated by local units' old systems inaccurate. The Department states that it is working with jurisdictions to reconcile differences in QVF records and records maintained by communities in their old systems.

The City of Flint attempted to use the QVF as a stand-alone system for the November 1998 election. Delta Township and the City of Detroit used the QVF in a parallel fashion with the old systems each community had in place to generate poll lists. Other local units, such as Keweenaw County and Alger County, did not use the system at all to generate their lists.

For those that generated polling lists using the QVF, the number of names between the lists that did not match varied across communities. In the case of Delta Township (which has a voting age population of 31,000), it does not know the exact number of records that did not match, but stated that the number was likely in the few thousands. Detroit (which has a voting age population of 620,000) did not have an idea of the number of records that did not match. Detroit made clear, however, that it did not feel comfortable using the QVF because of the number of discrepancies between the records in the QVF and the city's own database. For discrepancies on election day, these two communities relied on the lists produced by their old systems rather than relying on the QVF-generated lists.

Flint, on the other hand, did attempt to use the QVF as a stand-alone system for the November election. One week prior to the general election on November 3, 1998, the City of Flint contends that it was assured by the Department of State of the accuracy of the records in the QVF, and that Flint should use the system on election day. On election day, Flint discovered that a number of names of voters did not appear on the QVF-generated polling list. Although Flint did not know the exact number of names that did not appear, the city estimated that several hundred names were missing from the QVF-generated list on election day. At the present time, Flint and the Department of State are working to refine the accuracy of the records in the QVF. Flint maintains, however, that there may be as many as several thousand records in the QVF that do not match those maintained on Flint's old system.

The Department of State has a different perspective from Flint. The Department of State contends that it did not require or urge Flint to use the QVF as a stand-alone system for the November election. The Department claims that subsequent analysis indicated that many of the omissions from Flint's QVF lists were due to Flint errors. The Department of State does not believe that differences between Flint's locally produced lists and QVF-produced lists indicate errors with the QVF system.

There were some communities that did not use the QVF at all. The rural counties of Alger and Keweenaw in the Upper Peninsula, for example, did not use the QVF for the November general election or the August primary. In the cases of both counties, the voting records of the names in the QVF did not match the records of those names maintained under the old systems of these respective counties. As in the case of the previous mentioned communities, there were records omitted from the QVF. Rather than use the new database parallel to that of these communities' old systems, they decided to wait until the records in the new system were more accurate before they felt comfortable using the QVF. The Department of State maintains that some omissions of data in rural counties were due to the failure of clerks to provide voter registration data for entry in the QVF and failure of local jurisdictions to complete their portions of data reconciliation.

Among the issues that the Department of State is working to improve is the absentee voter portion of the QVF. Due to the lack of user-friendliness of the absentee voter portion of the database, the design of this portion did not satisfy the needs of many larger jurisdictions. Because of this problem, jurisdictions relied on the old absentee voting systems they each had in place for the last two statewide elections. The absentee voter portion, as initially designed, is very labor-intensive in that it requires the local units to proceed through many computer screens for a single voter, thereby making this process impractical for many jurisdictions. In the examples cited above, Delta Township, Detroit, and Flint did not use the absentee voter portion of the QVF. As previously stated, neither Alger County nor Keweenaw County used any portion of the QVF, including the absentee voter portion. In the cases of Delta Township, Detroit, and Flint, it was impractical for them to use this portion because of the volume of absentee voters that they have to deal with. During the early part of this year, the Department worked with several local clerks to identify problems with the absentee voter portion of the QVF. The Department has implemented these recommendations into a redesign of the absentee voter portion of the software is now available.

Further, several jurisdictions stated that the Street Index portion of the database is not accurate. The QVF Street Index database is critical to the system because it is the index of all street addresses within a particular community's jurisdiction. For all of the examples discussed - Delta Township, Detroit, Flint, Alger County, and Keweenaw County - they reported differences between the QVF Street Index and the local records maintained by the communities. Keweenaw County, in particular, reported an issue of concern. This county has a number of registered voters who live "off-road" and do not have conventional street addresses.

The Department of State responds to the Street Index issue by explaining that the QVF relies on established and published postal standards to determine the exact names of streets and street types such as "Rd" and "Ave". The Department contends that the fact that local records differ from those in the QVF does not necessarily make the QVF Street Index incorrect. With regard to rural addressing, the QVF was designed to accept "off-road" or unconventional addresses. The Department says, however, that this process may be time-consuming because some communities may not have a designated person to resolve disputes regarding street address names. The title of the person who may handle such disputes varies from community to community. There are also cases in which a person who is responsible for making a final decision in these matters may not even know that he or she is the appropriate person to make such a decision. The Department of State is working with local units to address these issues so that street information can be accurately indexed in the system.

Public Act 441 of 1994 required the QVF to be operational and available for use in all elections on or after January 1, 1998. The QVF was available on a statewide basis for the first time for primary elections on August 4, 1998, and for the general election on November 3, 1998. Currently, the Department of State continues to work on data collection and data reconciliation. At present, the Department is missing records from approximately 15,000 voters from jurisdictions that have not yet submitted data to be entered into the QVF. This is down from 30,000 voters on November 3, 1998.

With regard to data reconciliation, many records remain to be reconciled, although there are fewer records that remain unreconciled now compared with those remaining in November 1998. For large jurisdictions, approximately 2,000 records are left to be corrected, down from 10,600 last November. For small jurisdictions, approximately 60,000 records are left to be corrected, down from 163,000. Finally, the Department must enter 10,000 records from communities that submitted their information in paper (rather than on floppy disks or other electronic media). This number is down from 100,000 records that needed to be entered in November 1998.

The Department of State continues to work with local units of government to improve the accuracy of the Qualified Voter File. These local units still feel uncomfortable about using the QVF as a stand-alone system for large elections. According to the Department, however, it will continue working to correct problems with the system if they arise.