Elections: A Framework for Evaluating Reform Proposals		 
(15-OCT-01, GAO-02-90). 					 
								 
As a result of the events surrounding the November 2000 	 
presidential election, public officials and various interest	 
groups have proposed a number of reform measures to address the  
perceived shortcomings of various election systems. When	 
considering election reform proposals, careful consideration	 
should be given to the way election systems have evolved	 
throughout history. If required to adopt federal election reform,
states and local jurisdictions have indicated their desire to	 
have reasonable flexibility and time to implement any changes.	 
The complexity and intricacy inherent in the American electoral  
system suggests that the success of an election system is	 
contingent on the appropriate integration of people, processes,  
and technology. Failure to coordinate these components can cause 
problems that may be hard to accurately characterize, diagnose	 
and correct. This report delineates an analytical framework that 
Congress could use as it weighs the merits of proposed reforms.  
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-02-90						        
    ACCNO:   A02301						        
    TITLE:   Elections: A Framework for Evaluating Reform Proposals   
     DATE:   10/15/2001 
  SUBJECT:   Congressional powers				 
	     Elections						 
	     Information technology				 
	     Voting records					 
	     Standards and standardization			 
	     Congressional/executive relations			 
	     DOD Federal Voting Assistance Program		 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-02-90
     
A

October 2001 ELECTIONS A Framework for Evaluating Reform Proposals

GAO- 02- 90

Letter 1 Introduction 4 Results in Brief 6 Section 1: States and Local
Jurisdictions Indicated the Desire for

Reasonable Flexibility and Time to Implement Reforms 9 Section 2: Challenges
Exist to All Parts of Election Systems: People,

Process, and Technology 18 Section 3: Framework for Assessing Election
Reform Proposals 43 Section 4: Conclusion 53

Appendixes

Appendix I: Results of Our Election Work 56 The Scope of Congressional
Authority in Election

Administration 56 Elections: Perspectives on Activities and Challenges
Across the

Nation 58 Voting Accessibility for Voters with Disabilities 61 Improving
Voting Assistance for Military and Overseas Voters 63 Status and Use of
Federal Voting Equipment Standards 65 Statistical Analysis of Factors
Affecting Uncounted Votes in the 2000

Presidential Election 66

Appendix II: List of Recently Issued Election Reports 69 GAO Reports 69
Other Recent National Reports 69 State Commission and Task Force Reports 70

Tables Table 1: Percentage of Jurisdictions That Collected Data on Accuracy,
by Type of Voting Equipment 40

Table 2: Election Issues and GAO Reports 56 Figures Figure 1: Oklahoma and
Pennsylvania Illustrate Differences

Between Statewide Election Systems and Locally Autonomous Jurisdictions 11
Figure 2: Size of Jurisdiction Affects Magnitude of Key Tasks for Election
Officials 12 Figure 3: Some States Require An Excuse Before Allowing Voters

to Cast An Absentee Ballot, But Others Do Not 14 Figure 4: Voting Methods
Used In November 2000 Election by

Registered Voters 16

Figure 5: Jurisdictions Offer Various Types of Voter Education Materials 23
Figure 6: Prevalence of Potential Impediments at Polling Places

and Availability of Curbside Voting 25 Figure 7: Persons Voting Absentee in
the November 2000 Election

Cast Ballots Both by Mail- in and In- Person Absentee Voting 28 Figure 8:
Example of An Absentee Ballot Secrecy Envelope That Must Be Completed
Correctly in Order For the Absentee Ballot to be Counted 30

Figure 9: Optical Scan Ballots Can Be Mismarked in a Variety of Ways 33
Figure 10: Voting Technologies Perform Roughly on Par With Each

Other 37 Figure 11: Percentage of Total Variation in Uncounted Presidential
Votes Explained By County Demographics, Voting Equipment, State Differences,
and Unknown Factors 38

Lett er

October 15, 2001 The President of the Senate The Speaker of the House of
Representatives Voting is fundamental to our democracy. Each year the
millions of people who go to the polls expect that their ballots will be
cast in private and accurately processed and counted. However, events
surrounding the November 2000 presidential election raised broad- based
concerns,

including such issues as the performance of different types of voting
equipment, the disqualification of absentee ballots, and the accuracy of
vote tallies and recounts. As a result, public officials and various
interest groups have proposed a number of reform measures to address the

perceived shortcomings of various election systems. As part of the broad
congressional interest in these issues, we were asked to review certain
aspects of elections throughout the United States. This capping report draws
upon the extensive body of recent work done by GAO at the request of several
congressional leaders, congressional committees and Members of Congress on
matters surrounding the November 2000

election. Our work, which culminated in six other reports that are
summarized in the appendices, addressed the following topics:

* the scope of Congressional authority in election administration, 
compilation and maintenance of voter registration lists,  absentee and
early voting,  voting assistance for military and overseas voters, 
election day administration,  voting accessibility for voters with
disabilities,  vote counts, recounts, and certification,  voting
technology,  Internet voting, and  status and use of the Federal Election
Commission voting equipment standards.

This capping report serves three purposes. First, we discuss how the
constitutional and operational division of federal and state authority to
conduct elections has resulted in great variability in the ways elections
are

administered in the United States. As a result, given this diversity and the
decentralized nature of election administration, careful consideration needs
to be given to the degree of flexibility and the planned timeframes for
implementing new initiatives.

Second, in order for election reform to be effective, reform proposals must
address all major parts of our election systems- its people, processes, and
technology- which are interconnected and significantly affect the election
process. We discuss the main challenges that election officials face in each
of these areas and the need to accurately characterize and diagnose related
issues.

Finally, as Congress considers if and how to address the challenges to our
national election system it will have to deliberate upon a number of reform
proposals. Based on our work, this report offers basic criteria for
assessing a range of election reform proposals. It is not our intention to
advocate any particular reform proposal, but we believe that these criteria
offer a useful analytical framework with which to consider changes to our
nation?s election system.

Copies of this report are being sent to the President, the congressional
leadership and all other Members of Congress. Copies will also be made
available to other interested parties via our web site at www. gao. gov. If
you or your offices have any questions about matters discussed in this
report, please contact me at (202) 512- 5500; Norman J. Rabkin at (202) 512-
3610, Managing Director, Tax Administration and Justice; or Richard M.
Stana, Director, at (202) 512- 8816. They can also be reached by e- mail at
rabkinn@ gao. gov and stanar@ gao. gov, respectively.

David M. Walker Comptroller General of the United States

Introduction On November 7, 2000, more than 100 million Americans cast their
votes for various candidates and ballot issues across the country. This
hallmark of American democracy unfolded in more than 10, 000 local election

jurisdictions, which used several different types of voting equipment,
ranging from hand- counted paper ballots to electronic touchscreen voting
machines. Staffing the precincts were some 1.4 million dedicated poll
workers who opened the polls, logged in and assisted voters, closed polls,

and in many cases, tabulated the votes cast at the precinct. These poll
workers were assisted by an army of unseen election workers who answered
phone calls from both voters and poll workers, delivered extra ballots to
precincts, replaced or repaired broken voting machines, tabulated absentee
ballots, and compiled the election results from individual precincts within
their jurisdiction. This highly decentralized, complex, and massive
logistical effort made it possible for American citizens to participate in
one of the most fundamental democratic traditions- that of eligible citizens
to cast their votes for candidates of their choice.

The controversy surrounding the presidential vote in the November 2000
election cast America?s election system in a new and revealing light,
spotlighting issues such as the

 accuracy of voter registration lists;

 procedures used to accept or disqualify absentee ballots for counting;

 variety of ways elections are administered across and within
jurisdictions;

 widely varying types of voting equipment used to cast and count ballots;
and

 many different methods of determining voter intent when voters improperly
or incompletely marked their ballots. Several congressional leaders,
congressional committees, and Members of Congress asked us to review our
nation?s election systems. Accordingly, we focused on issues that may affect
the ability of eligible U. S. citizens to cast their ballots in private and
have those ballots counted accurately.

This capping report draws on a considerable body of work recently done by
GAO on election systems. We address three main issues that Congress may wish
to keep in mind as it approaches election reform. First, we examine the
division of federal and state authority to conduct elections and the
resulting variation among election jurisdictions. Second, we describe the

challenges that election officials face as they work with the people,
processes, and technology involved in our administering our nation?s
election systems. And third, we suggest four criteria that Congress could
use as it weighs the merits of various reform proposals.

Our work on election systems, which is contained in this capping report and
six separate reports, took us across the United States and around the world
as we employed a variety of methods to answer Congress? questions. We
conducted a detailed analysis of relevant constitutional provisions, federal
statutes, and federal court decisions as well as state statutes and
regulations on selected election issues. We met with and reviewed documents
provided by local election officials in 41 election jurisdictions in 22
states and met with officials at the Department of State, the Department of
Defense, the Federal Election Commission (FEC), and the National Conference
of State Legislatures. We surveyed District of Columbia and

state election directors. 1 Moreover, we used a mail survey and a telephone
survey 2 and interviews with local election officials to obtain information
about the election process that would generally be representative of the
more than 10,000 local election jurisdictions in the United States. We also
visited 585 polling places and met with embassy and military personnel
abroad and overseas citizens as well as with manufacturers and testers of
voting equipment. And finally, we reviewed documents provided by state and
local election officials, voting equipment manufacturers and testers, and
obtained data on voting methods and election results for the November

2000 election from Election Data Services, Inc., and other sources. 1 In our
discussion of state election systems, we include the District of Columbia. 2
Our mail survey is generalizeable to 90 percent of the U. S. population, and
our telephone survey is generalizeable nationwide.

Results in Brief Principal findings from our work provide a context for
considering challenges and subsequent reforms of our nation?s election
systems. When

considering election reform proposals, careful consideration should be given
to the way election systems have evolved throughout history. First, if
required to adopt federal election reform, states and local jurisdictions
have indicated their desire to have reasonable flexibility and time to

implement any changes. Traditionally, election administration has primarily
been the responsibility of state and local governments which are guided by a
tapestry of federal and state laws and regulations. Historically, Congress
has exercised its constitutional authority to affect the way state and local
jurisdictions conduct elections, primarily focusing on prohibiting certain
discriminatory practices and providing statutory protections for the rights
of groups as racial and language minorities, elderly voters, and voters with
disabilities. Over time, our nation?s election system has evolved into

51 individual systems 3 that in turn are administered and principally funded
by more than 10, 000 counties, cities, townships, and villages. We found
considerable variation in how these jurisdictions register voters; arrange
for polling places; recruit, train and pay poll workers; buy and maintain
voting equipment; conduct the election; and count the votes. Given this
variability, states and local jurisdictions have indicated the need for

reasonable flexibility in tailoring reform proposals to accommodate local
needs. Because implementing reform may require states and local
jurisdictions to transform long held practices, they will also need
appropriate time to successfully integrate the changes into policies and
procedures. Second, our work also underscores the complexity and intricacy
inherent in the American electoral system. The success of an election system
is contingent on the appropriate integration of people, processes, and

technology. Our research disclosed various challenges that election
officials face in each of these areas.

People: The people involved in an election include the election officials
and legions of temporary poll workers who prepare for and conduct the
election as well as the voters who participate in them. Our work disclosed a
wide range of challenges election officials faced related to both election
workers and voters. For example, recruiting and training a sufficient number
of poll workers presented a challenge to many officials. We 3 Including the
50 states and the District of Columbia.

estimate that 51 percent of jurisdictions nationwide found it somewhat or
very difficult to recruit a sufficient number of poll workers. As a second
example, providing education to voters on election processes and equipment
also posed a challenge to election officials, in some cases due to a lack of
funds. We estimate that 38 percent of jurisdictions nationwide support
federal funding to subsidize voter education. We also found that making
polling places accessible to voters with disabilities or providing
alternative methods for voting to individuals with disabilities posed a

challenge to many election jurisdictions.

Process: Processes guide people involved in the election process. For
example, election workers rely on processes as they carry out their duties
such as registering voters, distributing and processing absentee ballots,

educating voters, conducting the election, and counting the votes. We found
that a wide variety of process- related challenges confronted election
officials. For instance, keeping registration lists accurate and current
presented a challenge. In some cases, difficulties maintaining current voter
registration lists were related to the National Voter Registration Act
(NVRA, which is widely known as the ?Motor Voter? Act) in part because local
election officials received incomplete or incorrect voter registration

information from motor vehicle authorities. We estimate that about 46
percent of jurisdictions nationwide had problems with NVRA during the
November 2000 election. Our work also surfaced process- related challenges
associated with providing assistance for military personnel and overseas
citizens absentee voters. These voters? absentee ballots were disqualified
at higher rates than absentee ballots cast by voters living in the United
States. We found the availability of assistance to military personnel and
overseas citizens lacking and recommend that the Departments of State and
Defense take action to improve voter education and assistance

efforts. As a final example, determining voter intent when ballots were
improperly marked presented a challenge to many election officials. We
estimate that nationwide about 32 percent of local election jurisdictions
had no written instructions to interpret voter intent when ballots are
mismarked. Technology: Technology, such as voting equipment, provides tools
for

officials as they administer elections and for voters as they participate in
them. We identified diverse technology- related challenges facing our
nation?s election systems. Assessing why voting equipment does not meet the
needs of a jurisdiction presented a challenge to many election officials. As
a result, election officials may face a related challenge-- a lack of

performance data to help them make informed decisions regarding

acquisition of new voting equipment. Overall, about 96 percent of
jurisdictions nationwide reported being satisfied with the performance of
their voting equipment, but this level of satisfaction was typically based
on election officials? subjective impressions rather than objective
performance data. We estimate that less than half of election jurisdictions
collected data on how well their voting equipment performed in the November
2000 election. In addition to these challenges, we found that federal
standards for voting equipment should be updated (the FEC has initiated
plans to issue revised standards in 2002). We are making recommendations to
the FEC aimed at improving its efforts to update its 1990 voting equipment
standards. Because development, maintenance, and implementation of voting
equipment standards are very important responsibilities, we also raise as a
matter for congressional consideration the explicit assignment of
responsibility in these areas.

Our research also indicates that failure to coordinate these components can
cause problems that may be hard to accurately characterize, diagnose and
correct. For example, many of the problems that received media attention in
the November 2000 election were attributed to the failure of voting
technologies to accurately record the voter?s selections or count the
ballots. Our analysis showed that the type of voting equipment used
(including equipment which allowed for error correction) explained a

relatively small percent of the total variation among jurisdictions in
uncounted presidential votes. Jurisdictions, however, may still wish to
obtain benefits from modernizing voting equipment. In addition, many
election officials pointed to greater voter education as a more immediate
way to resolve voter error and uncounted votes.

Third, this report delineates an analytical framework that Congress could
use as it weighs the merits of various reform proposals. This framework
entails the following four fundamental criteria:

 Criterion I: The Appropriate Role of the Federal Government in Election
Reform. Does the proposed change call for an appropriate federal role in
affecting election reform, given the historic balance struck between
Congress? constitutional authority to legislate election administration and
some states? laws and traditions that grant autonomy to local jurisdictions
as they administer elections?

 Criterion II: The Balance Between Accessibility and Integrity. 4 How are
the goals of providing citizens broad access to the voting process balanced
against the public?s interest in ensuring the integrity of our election
systems?

 Criterion III: Coordination and Integration of People, Processes, and
Technology. How does the proposed change affect both the discrete problem it
is intended to resolve and the election system as a whole?  Criterion IV:
The Affordability and Sustainability of Proposed Election Reforms.

Have the necessary resources been identified to institute the change and to
continually monitor and re- evaluate it over time? Section 1: States and

Election administration in the U. S. is guided by federal and state laws,
Local Jurisdictions

regulations and policies. Within the broad framework established by the
Constitution and federal statutes, each state sets its own requirements for
Indicated the Desire conducting local, state, and federal elections within
the state. for Reasonable Consequently, state requirements and processes
vary considerably, and the

Flexibility and Time to U. S. election system comprises 51 separate election
systems. In turn, states typically have decentralized this process so the
responsibility for Implement Reforms

administering and funding elections resides in thousands of local government
election jurisdictions, creating even more variability among our nation?s
election systems. Thus, in adopting federal election reforms, the degree of
flexibility and the timeframes for implementing new initiatives need to be
given careful consideration during deliberation and execution of related
reforms.

4 As used in this criterion, accessibility refers to voters? access to the
political process rather than the access of a voter with disabilities to the
polling place.

The Federal Government The constitutional framework for elections
contemplates both state and Establishes a General federal roles. With regard
to the administration of federal elections, Framework for Elections,

Congress has certain constitutional authorities over both congressional and
presidential elections. 5 Congress has passed legislation relating to the
but Elections Are Primarily

administration of federal elections, under its various constitutional the
Responsibility of State

authorities in certain areas, including the timing of federal elections,
voter and Local Governments

registration, accessibility provisions for the elderly and persons with
disabilities, and absentee voting. Congress has, however, been most active
with respect to enacting prohibitions against discriminatory voting
practices, which apply in the context of both federal and state elections.
The Voting Rights Act of 1965, for example, established the constitutional
guarantee that no person be denied the right to vote on account of race or

color. In addition, subsequent amendments to the Act expanded it to include
protections for members of language minority groups, as well as other
matters regarding voting registration and procedures.

Within the broad framework established by the Constitution and federal
statutes, each state sets the requirements for conducting local, state, and
federal elections within the state. For example, states regulate such
aspects of elections as ballot access, registration procedures, absentee
voting requirements, establishment of voting places, provision of election
day

workers, and counting and certifying the vote. The states, in turn have
typically delegated responsibility for administering and funding state
election systems to the thousands of local election jurisdictions- more than
10, 000 nationwide- creating even more variability among our nation?s

election systems. State election codes and regulations may be very specific
or very general. In particular, some states have mandated statewide election
administration guidelines and procedures that foster uniformity in the way
local jurisdictions conduct elections. It is common for state provisions to
furnish some guidance regarding voter registration requirements and
procedures, absentee voting requirements and procedures, performance
requirements for voting methods used within the state, establishment of
polling places,

provision of election day workers, and the count and certification of the
vote. Other states have guidelines that generally permit local election 5
For additional details, see Elections: The Scope of Congressional Authority
in Election Administration (GAO- 01- 470, Mar. 13, 2001).

jurisdictions considerable autonomy and discretion in the way they run
elections (see figure 1).

Figure 1: Oklahoma and Pennsylvania Illustrate Differences Between Statewide
Election Systems and Locally Autonomous Jurisdictions Oklahoma has

Pennsylvania has

 standard election day

 minimal state guidelines.

 67 counties with 67 different election

 standard poll worker training. systems.

 diverse voting technologies. election mamagement system.

 one type of voting machine.

Source: GAO analysis based on information from local election officials.

Great Variability Exists in The variability from state to state becomes even
more pronounced at the the Ways Various Local

local level, as local jurisdictions have used the flexibility afforded in
state Elections Are Conducted

provisions to create local elections systems that vary from county to
county, and even, in some cases, within counties. This variation stems from
several factors. One factor can be a consequence of the size of local
election jurisdictions, which varies considerably. For example, one rural
county has 208 registered voters in contrast with a large, urban county,
such as Los Angeles County, whose total number of registered voters exceeds
that of 41 states. The complexity of preparing for and conducting an
election in large jurisdictions is generally greater than in smaller

jurisdictions. For example, a rural county, with a few thousand voters who
share the same language, prints its ballot in one language. In contrast, in
a large, urban jurisdiction with a diverse population of 4 million
registered voters prints its ballots in 7 different languages. This can also
have an

effect on the processes and type of voting equipment used. As illustrated in
figure 2, the magnitude of other key administrative tasks in this large,
urban jurisdiction is a thousand times larger than for the small
jurisdiction.

Figure 2: Size of Jurisdiction Affects Magnitude of Key Tasks for Election
Officials Numbers Numbers 4,000,000 4 million

120,000 100,000

100,000 90,000

90,000 80,000

80,000 70,000

70,000 60,000

60,000 50,000

50,000 40,000

40,000 30,000

25,000

30,000 20,000

20,000 10,000

4,963

10,000 0 2,843 28 5

0

Registered voters workers

precinct Registered voters

workers precinct

Poll Voting Poll Voting Source: GAO analysis based on information from local
election officials.

Variability can also be a consequence of local needs. For example, a
jurisdiction with a large population segment that moves out of the location
each year might opt for certain voter registration and voter education
processes that reflect the need to address a large voter turnover. As a
second example, a jurisdiction might use a certain type of voting equipment

based on financial resource availability. More wealthy jurisdictions have
had the resources necessary to modernize equipment, while others need to
make do with what they have. Finally, variability can be a consequence of a
jurisdiction?s perceived need to maintain voting traditions that have been
in place for a long time. In two jurisdictions we visited, election
officials opted to replace lever machines with full- screen electronic
voting machines because this machine uses a

ballot that most closely resembles the type of ballot voters were used to
seeing on lever machines. This choice was possible because ballots did not

have to be printed in more than two languages in either jurisdiction so the
ballots could fit on a single page.

Variability among states and local jurisdictions was evident in each major
stage of an election-- voter registration, absentee and early voting,
preparing for and conducting election day activities, and vote counting and
certification. Some examples follow.

Who Could Vote Varied For the November 2000 election, the FEC reported that
nearly 168 million people, or about 82 percent of the voting age population,
6 were registered to vote. Registering to vote is not a federal requirement,
but in November 2000, all states except North Dakota required citizens to
register before voting. At a minimum, every state and the District of
Columbia required that a voter be a U. S. citizen, at least 18 years of age,
and a resident.

Additional requirements to vote, such as time in residence, varied. Due to
variations in voter eligibility requirements, different citizens with the
same qualifications would be eligible to vote in some states but not in
others including (1) those that had completed their sentence after a felony
conviction; (2) those who had been adjudged mentally incompetent; and (3)
those who met all of the qualifications to vote but who had not registered
in accordance with prescribed timeframes.

When People Could Vote Varied In November 2000, citizens had different
opportunities for obtaining and successfully casting absentee ballots due to
the differences in absentee and early voting requirements, administration,
and procedures. All states allow some provision for absentee balloting;
some, however, require a reason to vote absentee as indicated in the figure
below. 6 This number includes active and inactive voters. FEC defines
inactive voters as those who remain on the registration list but who have
moved, according to information provided by the Postal Service, have been
mailed a registration confirmation notice, but have neither responded nor
offered to vote in the subsequent federal election. All other persons on the

registration list are considered to be active voters. In The Impact of The
National Voter Registration Act of 1993 on the Administration of Elections
for Federal Office, 19992000, the FEC reported that for the November 2000
general elections, there were 149,476,705 active registered voters, about 73
percent of the voting age population.

Figure 3: Some States Require An Excuse Before Allowing Voters to Cast An
Absentee Ballot, But Others Do Not WA

MT ND

OR NH

MA SD

RI WY

CT N

NJ NV

DE UT

DC CO CA

KS AZ

NM OK

AL S

TX AK

HI

Allows no- excuse voting before election day Requires an excuse to vote
before election day

Source: GAO review of state statutes.

Who Conducted the Election For the November 2000 election, about 1.4 million
poll workers staffed

Varied polling places across the country on election day. Although poll
workers are usually employed for only one day, the success of election

administration largely hinges upon their ability to perform their jobs well.
Depending on state law and the organization and traditions of the local
jurisdiction, poll workers have different titles, levels of pay, training

requirements, and responsibilities. Some poll workers are elected, some are
appointed, and some volunteer. Levels of authority and responsibility that
jurisdictions grant to poll workers also ranged from significant autonomy
over the operation of the polling place with final authority to interpret
improper ballot markings to having limited discretion, functioning primarily
as clerks and facilitators who refer issues and problems back to

elections headquarters. How Elections Were Conducted

Jurisdictions followed various procedures on election day that created Vari
ed

differences in the way elections were conducted. For example, to determine
whether a citizen who appeared at the polls was eligible to vote, some
jurisdictions required voters to identify themselves by stating their names
and addresses to the poll workers, who also matched the signature on the
voter application with the voter registration records. Other

jurisdictions also required voters to present a valid photo identification
card. In other jurisdictions, a hunting or fishing license was sufficient to
verify one?s identity. Still other jurisdictions required no identification
other than the voter stating his or her name. If a voter?s name did not
appear on the list of registered voters, some jurisdictions accommodated
these individuals by automatically giving them a provisional ballot which
may have been counted if the voter?s eligibility was verified at a later
time. Others did not.

How the Votes Were Cast Varied Registered voters cast their ballots using
one of five voting methods in the November 2000 election: hand- counted
paper ballots and lever, punch card, optical scan, and Direct Recording
Electronic (DRE) voting equipment. Punch card and optical scan equipment was
most widely used by registered voters, as figure 4 shows.

Figure 4: Voting Methods Used In November 2000 Election by Registered Voters
Percentage 35

31 31

30 25 20

17

15

12

10

8

5

1

0

scan card

Lever DRE

Mixed ballots

i cal Punch Opt

Paper Source: GAO analysis of data from Electronic Data Services and states.

Election Reform Requires Although the states have traditionally had broad
authority to regulate and

An Appreciation for conduct elections, Congress also has broad authority to
regulate federal Variability Among

elections, and, in particular, congressional elections. State and local
Jurisdictions election officials generally use the same people, processes,
and technology

to conduct local, state, and federal elections. Consequently, as a practical
matter, congressionally directed changes in the conduct of congressional
elections are likely to affect the administration of state and local
elections. In addition, through the use of its spending power, Congress may
encourage state action by attaching conditions to the receipt of federal
funds. 7

7 For additional details, see GAO- 01- 470, Mar. 13, 2001.

Our work found wide variations across the country which have developed over
time in response to local economic, demographic, political, and cultural
traditions in each state and local jurisdiction. As might be expected given
this variety, local election officials do not share a common perspective on
election reform. For example, when asked their preferences for the use of
federal funds, should they become available, local election officials
identified a range of spending priorities, with no

clear consensus on the top priority. Among these priorities were voter
education, voting equipment, poll worker pay, and postage for mailing voting
materials and printing ballots. Nor was there a consensus among

local election officials on how involved the federal government should be in
state and local election administration and reform. About 27 percent of the
local jurisdictions we surveyed supported uniform standards for election
administration, and 30 percent supported updated federal standards for
voting equipment. However, many of the jurisdictions we visited did not want
federal funding for election administration if it meant sacrificing local
autonomy. Some supported the concept of a federal clearinghouse for sharing
information about election administration practices.

National Survey Results

We estimate that 54 percent of local election officials support federal
funding to subsidize postage for materials mailed to voters, 42 percent
support federal funding to help with the operational costs of elections, and
over 38 percent support federal funding for voter education.

GAO Mail Survey of Jurisdictions

Overall, our work suggests that a "one size fits all" approach may not be
suitable for every aspect of election administration. In jurisdictions with
a small number of voters, for example, hand- counted paper ballots may
produce accurate and complete vote counts. Conversely, full- faced DRE
equipment (this type of machine uses a large, single- page ballot) could not
readily accommodate Los Angeles County's long ballot printed in 7 languages.
Thus, the degree of flexibility afforded local jurisdictions in

implementing any reform should be given careful consideration when
deliberating specific reform initiatives. Historically, changes in election
administration have been evolutionary rather than revolutionary. Both

election officials and voters become accustomed to and comfortable with how
elections are planned and conducted in their individual jurisdictions. Many
of the underlying conditions associated with the variations are not likely
to change in the short term. Large- scale, immediate changes in the use of
people, processes, or technology will not necessarily change those
conditions. Thus, appropriate time may be needed for state and local
jurisdictions to determine how to implement changes effectively in their

specific jurisdictions. Section 2: Challenges

The second principal finding that emerged from our work is that election
Exist to All Parts of officials face a range of challenges connected with
all parts of the election

system: people, process, and technology. The people involved in an Election
Systems:

election include election officials, legions of temporary poll workers, and
People, Process, and

voters. Processes guide people as they carry out their duties, such as
Technology registering voters and conducting the vote. Technology, such as
voting equipment, provides tools for officials to administer elections and
for voters to participate in them. National Survey Results

We estimate that 57 percent of jurisdictions nationwide had major problems
in one or more areas on election day.

GAO Mail Survey of Jurisdictions

Our work identifies several challenges that can be categorized as primarily
people, process, or technology- related issues. Although we classify
problems as falling into one of these three areas, what becomes apparent is
that a problem in any one of these three broad categories is related in some
way to another part of the system. For example, people- related challenges

impinge upon process issues, process challenges affect technology, and
technology challenges affect people issues.

Specifically, several challenges emerged from our work on elections. Issues
with recruiting and training poll workers, educating voters, and

addressing needs of voters with disabilities posed major people- related
challenges in the November 2000 elections. Election officials confronted
process- related challenges that included maintaining accurate voter
registration lists, completing and processing absentee ballots, and
canvassing the vote. Technology- related challenges that faced our nation?s

election systems included assessing why voting equipment may fail to meet
jurisdictions? needs, collecting useful performance data to make informed
investment decisions, and needing federal leadership in updating and
implementing standards for voting equipment. Below we highlight the main
challenges identified through discussions with election officials and

our analysis. People, Both Election

Issues with recruiting and training poll workers, educating voters, and
Workers and Voters,

addressing needs of voters with disabilities posed major challenges in the
Presented Challenges

November 2000 elections. We estimate that over half of the jurisdictions in
the United States found it somewhat or very difficult to recruit and train a
sufficient number of poll workers. The major challenge many jurisdictions
identified regarding voter education was finding sufficient funding. On the
basis of our mail survey, we estimate that over a third of the jurisdictions
nationwide believed that the federal government should provide monetary
assistance for voter education programs. Limited availability of accessible
buildings and other constraints create obstacles to election officials?
efforts

to make polling places accessible to voters with disabilities. We estimate
that, from the parking area to the voting booth, 16 percent of all polling
places have no potential impediments, 56 percent have one or more

potential impediments but offer curbside voting, and 28 percent have one or
more potential impediments and do not offer curbside voting.

Recruiting and Training Qualified Many election officials told us recruiting
and training a sufficient number of Poll Workers Presents a Major poll
workers with appropriate skills to open, operate, and close polling

Challenge for Many Jurisdictions places was a major challenge on election
day. Factors that can work in concert to complicate an already difficult
task for election officials include

an aging work force, low pay, and little or no poll worker training.

Recruiting Enough Poll Workers Is Difficult for Many On the basis of our
mail survey, we estimate that 51 percent of jurisdictions nationwide had a
somewhat or very difficult time getting enough poll workers. For these
jurisdictions, obtaining enough poll workers (27

percent) was the most frequently identified major problem the jurisdictions
faced. National Survey Results

We estimate 51 percent of the jurisdictions nationwide reported that it was
somewhat or very difficult to get a sufficient number of poll workers.

GAO Mail Survey of Jurisdictions

Poll Workers Drawn From Aging Labor Pool

Many people who are available for occasional full- day employment as poll
workers are older, perhaps retired, and likely attracted to the work because
of something other than the pay because poll workers are generally paid low
wages. For example, an election official in a small jurisdiction said that
over 70 percent of their poll workers are over 65 years old. One official
remarked that volunteering is characteristic of an older generation. Several
officials echoed the statement of an official in a small jurisdiction

that ?[ o] ur election workforce is aging and we are having difficulty
recruiting younger workers.? Low Pay, Long Hours May Discourage Younger
Workers

The pool of potential poll workers may be shrinking because poll worker pay
is inadequate to attract employed or more skilled workers and poll workers
often are required to complete a 15- to 18- hour day. One election

official reported that ?[ s] ince compensation for this job is only $80 to
$135 per day, depending upon the election district, it is not sufficient to
attract a younger workforce.? The length of the day is a complaint of many
poll workers and may even pose an obstacle for younger workers. Another

official said that, ?[ w] hat they (the election judges) used to consider as
a fun and interesting day and an American duty has become ?heavy duty?.? In
one large jurisdiction, election officials asked poll workers to provide
feedback on their experience in the November 2000 election. One poll worker
responded that it was ?[ a] bsolutely, positively too long a day. I am

26 years old and very athletic and still went home at night and fell asleep
with my clothes on. With the majority of helpers either older or disabled, I
have no idea how they survived the day.?

Poll Workers With Specialized Skills Are Often Difficult to Recruit

Another problem is addressing the specialized labor needs unique to
particular polling sites, according to several local election officials.
Some polling places required poll workers to have specific language skills.
Finding qualified bilingual workers, specifically workers fluent in Asian
languages, is one very large jurisdiction?s biggest recruiting problem. Some
places had trouble finding poll workers who are able to learn the technical
skills necessary to operate voting equipment. Officials in one very large

jurisdiction said they have no scarcity of people willing to serve, but
finding people to meet specialized needs is the issue. Obstacles to
recruiting poll workers may overlap. One election official wrote that ?[ i]
t is increasingly difficult to find folks to work for $6 an hour. We are
relying on older retired

persons- many who can?t/ won?t keep up with changes in the technology or
laws. Many of our workers are 70+.? Minimal Training May Not Have Adequately
Prepared Poll Workers

for Election Day We estimate that 87 percent of jurisdictions nationwide
provided some training for poll workers. Poll worker training courses
generally span a few hours? time and focus on the key processes that poll
workers should follow, including how to operate voting equipment. Although
most of the jurisdictions we visited required some poll worker training,
election officials cited instances where poll workers who had attended
training still either did not understand what they were to do or chose not
to follow

specific instructions on how to run the polls. For example, to handle
unregistered voters in one very large jurisdiction, the poll workers were
instructed to provide those voters with questionable credentials a
provisional ballot. However, some poll workers failed to follow these rules
and turned away some voters from the polling place. Poll worker training in
the sites we visited rarely included discussion of the interpersonal skills
that poll workers should employ when dealing with frustrated citizens or

with each other. Educating Voters About

Another people- related challenge concerns educating voters about Processes,
Voting Equipment is a

particular processes, such as voter registration and how to operate voting
Challenge

equipment. Jurisdictions place varying degrees of emphasis on educating

voters about election processes and procedures. A lack of funds is the
primary challenge that election officials said they face in expanding their
efforts to educate voters about elections. Further, spending for voter
education is considered discretionary. Some local officials must first take
care of mandatory items such as equipment, supplies, poll worker salaries
and polling places. Many officials said that they see voter education as an
area where federal funds could be particularly helpful. On the basis of our
mail survey, we estimate that over 38 percent of jurisdictions nationwide
believed that the federal government should provide monetary assistance for
voter education programs.

Voter Education Needed Regarding Processes, Use of Voting Equipment

How well a jurisdiction educates voters about election processes and how to
use voting equipment can effect how well an election system functions. For
example, a number of problems associated with processes, such as requesting
and completing absentee ballots and registering to vote, are

precipitated by voters failing to provide complete information or to meet
deadlines. Voter education can be used to help remedy some of these
difficulties. How well jurisdictions educate voters on the use of voting
equipment can affect how easy voters find the equipment to use and integrity
of the vote. Jurisdictions provide various types of voter education
materials to help voters correctly use voting equipment.

Figure 5: Jurisdictions Offer Various Types of Voter Education Materials A

B

Tulsa uses the front side (A) of wallet cards to educate voters on the
proper way to complete optical scan ballots and the reverse side to provide
(B) schedules for key events.

Detroit warns voters against choosing more than one candidate in a single
race on one page of its voter guide.

Los Angeles County advises voters to remove hanging chads from their punch
card ballots.

Cuyhoga County provides instructions on how to vote a punch card ballot.

Source: Local election officials.

How frequently the voting equipment counts votes as intended by voters is a
function not only of equipment design, but of how well poll workers properly
instructed and guided voters, how well voters followed applicable
instructions, and what type of assistance was available to help voters who

have questions or make mistakes in voting. To illustrate this point,
officials from a very large jurisdiction stated that in the November 2000
election 1,500 voters had inserted their punch cards in the recording device
upside down, thus causing their votes to be inaccurately recorded.
Similarly, at a small jurisdiction that we visited where optical scan
equipment was used,

officials reported that some voters incorrectly marked the ovals or used a
nonreadable pen to mark the ballot, resulting in partially read ballots. In
a medium- sized jurisdiction that we visited, voters selected a candidate on
the optical scan ballot and then wrote the candidate?s name in the write- in
section of the ballot, thus overvoting (making more choices than are
permitted per contest) and spoiling the ballot. The election officials
stated that they believed that this misunderstanding contributed to the
jurisdictions? almost 5- percent overvote rate. In each of these cases, the
way that the voter completed the ballot caused the vote to be recorded
inaccurately, even though the voting equipment correctly counted the votes
as recorded.

Officials Face Challenges A third people- related challenge surfaced during
the November 2000 Addressing Needs of Voters with election- making polling
places accessible to voters with disabilities or Disabilities providing
alternative voting methods such as curbside voting. The extent

to which any given feature may prevent or facilitate access to a polling
place is unknown; however, based on our onsite work during the November 2000
election, we estimate that, from the parking area to the voting room, 16
percent of all polling places have no potential impediments, 56 percent have
one or more potential impediments but offer

curbside voting, and 28 percent have one or more potential impediments and
do not offer curbside voting (see fig. 6). 8 These potential impediments
would primarily affect individuals with mobility impairments and occur most
often on the route from the parking area to the building or at the entrance
to the polling place. Inside the voting room, the types and arrangement of
voting equipment used may also pose challenges for people with mobility,
vision, or dexterity impairments. 8 Although curbside voting is not
available at a number of polling places with potential impediments, as noted
earlier all states have provisions for absentee voting, and many states
provide for other alternative voting methods or accommodations, which may
facilitate voting by people with disabilities on or before election day.

Figure 6: Prevalence of Potential Impediments at Polling Places and
Availability of Curbside Voting

Percentage of polling places with no potential impediments

16%

Percentage of polling places with

56% one or more potential impediments

28%

that offer curbside voting Percentage of polling places with one or more
potential impediments that do not offer curbside voting

Note: These potential impediments are located along the route from the
parking area to the voting room.

Source: GAO analysis of polling place data collected on Nov. 7, 2000.

A number of efforts have been made by states and localities to improve
voting accessibility for people with disabilities, such as modifying polling
places, acquiring new voting equipment, and providing curbside voting. State
and county election officials we surveyed cited a variety of challenges to
improving access, including limited availability of accessible facilities
and funding constraints at the local level. Some disability advocates
believe that alternative voting methods and accommodations should not be

viewed as permanent solutions for inaccessible polling places because these
remedies do not provide the same opportunity for voting afforded the general
public, that is, in a polling place and in private.

Processes for Election Election officials confronted process- related
challenges that included

Activities Present maintaining accurate voter registration lists, completing
and processing

Challenges to Election absentee ballots, and interpreting voter intent. A
number of jurisdictions

Administration Officials reported they had trouble maintaining accurate
voter registration lists

because of the NVRA. This difficulty, in turn, may have exacerbated problems
related to qualifying voters at the polls on election day. As the

number of voters at home and abroad who cast absentee ballots grows,
challenges related to absentee voting are increasing. Election officials
reported difficulties with processing millions of absentee ballots cast in
the weeks and days before election day and noted the added financial burden
of processing these ballots. Military personnel and overseas citizens?
absentee ballots were disqualified at a higher rate than voters at home

which presents another challenge. And finally, interpreting improperly
marked ballots to determine the voter?s intent was a challenging process
because local jurisdictions often lack specific, written guidance and the
task itself can be inherently difficult. NVRA May Complicate Maintenance of
Voter Registration Lists, Qualifying Voters on Election Day

National Survey Results

We estimate that about 46 percent of jurisdictions nationwide had problems
with the National Voter Registration Act (NVRA or motor voter) during the
November 2000 election.

GAO Telephone Survey of Jurisdictions

Primarily, a citizen?s access to voting was based on the appearance of his
or her name on a voter registration list, which is compiled and maintained
by election officials. Officials encountered several process- related
challenges connected with compiling and maintaining accurate voter
registration lists, chief among them processing applications submitted
through sources other than elections offices, such as motor vehicle
authorities. Specifically, local election officials around the country
expressed concerns about processing

voter registration applications submitted at state motor vehicle
authorities, as permitted by NVRA. They claimed motor vehicle authorities
forward incomplete, illegible and late applications. Some election officials
attribute not only missing or incomplete applications to NVRA, but a new
problem, duplicate registrations. In the words of one official:

?You can ask any county clerk in the state and they will tell you that the
biggest problem is motor voter [NVRA]. Residents can register at the welfare
office, the health department, the motor vehicle authorities, and they do,
time and again. This results in tons of registrations which are costly and
time- consuming to sort through and check against records.?

Inaccurate registration lists affect other parts of the election system,
especially qualifying voters on election day. Officials reported that voters
appeared at the polls on election day claiming to have registered to vote
through the motor vehicle authority, but their applications never arrived in
the elections office. These individuals were sometimes turned away from the
polls. Dealing with voter eligibility issues can be a major problem for

some jurisdictions. We estimate that 30 percent of jurisdictions considered
dealing with unregistered voters at the polls to be a major problem.
Processes for Assisting Citizens All 50 states and the District of Columbia
allowed some form of absentee or at Home and Abroad with

early voting to increase voter access, convenience, and participation, and
Absentee Voting Present

the number of American voting absentee is growing. Using Census data,
Challenges

we estimate that for the November 2000 election about 14 percent of voters
nationwide cast their ballots before election day. 9 Of these voters, about
73 percent used mail ballots and 27 percent voted in- person, as seen in
Figure

7. This represents an increase from the 1996 presidential election in which
a total of about 11 percent of voters cast ballots before election day. We
estimate that nationwide local election officials received about 14. 5
million

applications for mail- in absentee ballots (plus or minus 3 million) for the
November 2000 election.

9 Based on GAO analysis of U. S. Census Current Population Survey, November
2000 Voting Supplement.

Figure 7: Persons Voting Absentee in the November 2000 Election Cast Ballots
Both by Mail- in and In- Person Absentee Voting

Voted before election day

27%

Voted in person

14% 73%

Voted by mail Source: GAO analysis of U. S. Census Bureau Current Population
Survey, November 2000 Voting Supplement.

As more voters, at home and abroad, cast absentee ballots, officials from
several local election jurisdictions reported costs and workload involved in
reviewing the volume of ballots have grown. Each of the millions of mail- in
absentee ballots received by local election officials had to be qualified
before being counted. Officials from one very large jurisdiction stated that
the sheer volume of mail- in ballots received creates a greater potential
for

errors.

Military and Overseas Citizens? Absentee Ballots Disqualified at Higher
Rates

Because military and overseas citizens? absentee ballots are disqualified at
higher rates than those of citizens voting absentee at home, processes for
assisting military personnel and overseas citizens need to be improved.
Although precise numbers are not available, we estimate that counties having
a voting age population of less than 60, 000 nationwide disqualified about 8
percent of ballots cast by military and overseas voters. In contrast, the
ballot disqualification rate for civilian voters not living overseas was
less than 2 percent. 10 While counties having a voting age population of
more than 60,000 that responded to GAO's survey showed a similar pattern,
the data was insufficient to make a national estimate. 11 The survey showed
that for all absentee ballots cast, almost two- thirds of

the disqualified absentee ballots were rejected because the ballots arrived
too late to be counted or the envelopes or forms accompanying the ballots
were not completed properly. The figure below describes the forms that must
be completed in order for the mail- in absentee ballot to be qualified.

10 The confidence intervals in this report are calculated at the 95- percent
confidence level. For example, we are 95 percent certain that the actual
disqualification rate for civilians not living overseas was between 1.2 and
2. 4 percent. In this analysis, GAO uses the term "disqualified ballots" to
refer to absentee ballots that, in the judgement of local election
officials, did not meet state requirements and that were rejected prior to
the vote counting process. For instance, the ballot may have been received
after the deadline or may have lacked certain required information on the
ballot/ return envelope, such as the voter's signature. Disqualification
does not refer to ballots that were rejected during ballot

counting due to problems in reading the ballot and/ or determining a voter's
actual preferences. 11 Not enough counties in the sample provided data on
the reasons ballots were disqualified from specific groups of voters, such
as military versus non- military, to enable GAO to make reasonable estimates
for these groups.

Figure 8: Example of An Absentee Ballot Secrecy Envelope That Must Be
Completed Correctly in Order For the Absentee Ballot to be Counted A

B

(A) A voter is to complete

his or her ballot and place it in the secrecy envelope.

C

(B) The secrecy envelope

containing the ballot is then to be placed into the ballot/ return envelope.

(C) The voter must sign

the envelope for the vote to be counted.

Source: Local election officials.

Processes for Assisting Military and Overseas Citizens Need Improvement

The Uniformed and Overseas Citizens Absentee Voting Act of 1986 12 protects
the right to vote by absentee ballot in federal elections for more than 6
million military and overseas citizens and recommends that states adopt a
number of provisions that facilitate absentee voting by these populations.
The Federal Voting Assistance Program, established within the Department of
Defense (DOD), is responsible for implementing the act by informing U. S.
citizens worldwide about their right to vote, fostering voting
participation, and working with states to simplify the registration and
absentee voting process. Also, the State Department works with DOD

to provide voter assistance to overseas citizens. 12 42 U. S. C. 1973ff-
1973ff- 6.

The extent and quality of federal voter assistance for military personnel
and overseas citizens varied considerably in the November 2000 election.
While the Federal Voting Assistance Program developed a number of useful
tools for voters and some installations GAO visited had well run programs

providing assistance and information to potential voters, other
installations did not meet DOD and service requirements. The variability in
executing the program is due to incomplete service- level guidance that does
not reflect DOD's directive, a lack of command support at some
installations, and a lack of program oversight by some DOD components.
Finally, the State Department provided citizens abroad with a variety of
useful

assistance, according to overseas citizens and federal employees GAO spoke
to, although both groups believed more outreach could be beneficial. Also,
State Department Headquarters has not played an active role in sharing best
practices and lessons learned or in overseeing the program.

We recommend that the Secretaries of Defense and State improve (1) the
clarity and completeness of service guidance, (2) voter education and
outreach programs, (3) oversight and evaluation of voting assistance
efforts, and (4) sharing of best practices. 13

Processes for Interpreting Voter Processes for handling improperly marked
ballots present a challenge for

Intent Posed a Challenge to many election officials, especially when an
election is close. Many states Many specifically require election officials
to count ballots if the ?intent of the voter? can be determined. Thirty- one
states and the District of Columbia reported to us that they make some
determination of voter intent. Voter intent issues arise with paper, optical
scan, and punch card ballots, not when the ballots are marked properly for
the type of ballot used, but when there are variations from proper marking.
During the canvassing stage (when votes are counted and totals calculated),
election officials are tasked with reviewing ballots that are not properly
marked and sometimes required to determine how those voters intended to cast
their votes. After the polls close and ballots are returned to election
headquarters, workers canvass the votes, a process that entails reviewing
all votes by precinct, resolving problem votes, and counting all valid
votes. At this point, workers deal with ballots that are either unclearly or
improperly marked.

13 The recommendations made to the Secretaries of Defense and State were
made in Elections: Voting Assistance to Military and Overseas Citizens
Should be Improved (GAO- 01- 1026, Sept. 28, 2001)

Ballots can be improperly marked in variety of ways that differ according to
the type of voting equipment being used in a jurisdiction. Because the DRE
and lever machines voters record the vote directly on the equipment rather
than a separate ballot, there is no opportunity for a mismarked ballot.
Paper, punch card, and optical scan ballots, however, can be

improperly marked. For example, on an optical scan ballot voters may have
circled a candidate?s name, instead of completing the oval, box, or arrow
next to the candidate?s name as illustrated in figure 9.

Figure 9: Optical Scan Ballots Can Be Mismarked in a Variety of Ways B

A. Optical Scan Ballots

A

may be improperly marked when voters

(a) do not complete the arrow,

(b) circle nominee's names, and

(c) both fill in the arrow and write in the candidate's name.

B. Optical scan ballots are properly marked when the ends of the arrow are
connected.

( (a) ( (b) (c) (

John Doe

Source: Local election officials and GAO analysis.

Interpreting a mismarked ballot to determine the voter?s intent can be a
challenging process. While states may instruct officials to determine
voter?s intent on mismarked ballots, states do not always provide guidance
on how to do so. Our work indicates that nationwide about 30 percent of

local jurisdictions had no instructions, either from the state or local
jurisdiction, on how to interpret voter intent, for example, how to read

stray marks on paper ballots or dimples or partially punched chads on punch
card ballots. We estimate that about 15 percent of jurisdictions had
instructions developed by the jurisdiction and 23 percent had both state and
local written guidance. National Survey Results

We estimate that nationwide about 32 percent of local election jurisdictions
had no written instructions, either from the state or local jurisdiction, on
how to interpret voter intent, such as stray marks on ballots or partially
punched punch card boxes.

GAO Mail Survey of Jurisdictions

Developing processes to interpret a voter?s intent can be challenging, and
local jurisdictions vary in how they approach this task. Processes for
handling punch card ballots illustrate this point. Jurisdictions we visited

reported various ways to handle problem punch card ballots. For example, in
one jurisdiction, election officials told us if the punch card ballot
contains a dimple with a pinhole, employees put the original ballot over a

pink (or duplicate) ballot and held it up to the light. Where they saw
light, they punched. The employee also turned over the ballot and looked for
bumps, which indicated the voter inserted the ballot backwards. If a ballot
contained bumps on the backside, the ballot could be duplicated.

In another jurisdiction, a vote on a punch card consisted of any removed
chad plus any chad that freely swung by one side. The person scanning the
ballot inspected it for improperly punched chads by running the ballot
through their fingers. In another jurisdiction, the ballot inspection teams

are given a pair of tweezers and told to remove any chads remaining on the
punch card. One jurisdiction used persons called ?scanners? to go over the
ballots before they are counted. Each ballot is inspected for improperly
punched chad by running the ballot cards between the scanners? fingers. Very
loose chad will be removed through this process. If the chad does not come
off and freely swings by one side, it may be removed. Problem ballots, such
as

those that are unreadable because of incompletely removed punches or
incorrect punches, can alter the counting results or create problems with
the computer processing. They are given to ?makeover scanners? to be remade.
Technology Challenges While problems related to voting equipment performance
during the Include Assessing Why

November 2000 election received a great deal of media attention, the
Equipment May Not Meet performance of voting equipment is not only a
function of the technology

Needs, Collecting Useful design itself. The people who interact with the
technology and the

processes governing this interaction can also affect whether voting
Performance Data, and

technology meets the needs of a jurisdiction. As a result, assessing why
Updating Standards voting equipment may not meet needs of some jurisdictions
can be difficult. Another challenge facing election officials involved
obtaining

reliable measures and objective data to make informed decisions about
whether to invest in new voting equipment or to invest in measures to
improve performance of existing equipment, such as maintenance personnel.
Local jurisdictions do not always have the information they need to select
the most appropriate investment option given their needs

and resource constraints. Although 96 percent of local jurisdictions report
that they are satisfied with their voting equipment, less than 50 percent of
them collect data on how well their equipment performed. This information is
vital for jurisdictions considering modernizing their equipment. Another
challenge relates to developing and maintaining

updated standards for voting equipment. Although the FEC is in the process
of updating voting equipment standards issued in 1990, responsibility for
establishing, maintaining, and implementing up- to- date standards for
voting equipment has not been explicitly assigned. As a result, the 1990
standards have become dated. Assessing Why Voting Equipment Understanding a
jurisdiction?s voting equipment needs and why voting May Not Meet
Jurisdictions? equipment may not meet those needs can pose a challenge. In
assessing

Needs Poses a Challenge whether voting equipment meets the needs of a
jurisdiction?s user communities (both the voters and the officials who
administer the

elections), election officials must have reliable measures and objective
performance data. When voting equipment does not meet the needs of a
jurisdiction, officials must also understand the cause or causes of the
problem before they can choose an appropriate solution, such as more voter
education, increased training for election workers, or acquiring new

equipment. These causes can be difficult to identify because performance of
voting equipment is not only a function of the technology design itself,

but also of the people who interact with the technology and the processes
governing this interaction. To illustrate this point, our survey of vendors
showed little difference among the basic performance characteristics of DRE,
optical scan, and

punch card equipment. 14 However, when local election jurisdictions?
experiences with the equipment are considered, performance differences among
voting equipment became more evident. These differences arise because a
real- world setting- such as an election in which equipment is operated by
actual voters, poll workers, and technicians- tends to result in performance
that differs from that in a controlled setting (such as in the

manufacturer?s laboratory). This difference demonstrates the importance of
the effect of people and process on equipment performance.

While Some Voting Equipment Is Easier to Use, No Clear ?Best Performer?

Figure 10 shows a relative comparison of certain characteristics- accuracy,
ease of use, efficiency, and security- of the various types of voting
equipment used in the November 2000 elections. The comparison reflects the
results of our analysis of data provided by voting equipment vendors that
responded to our survey and survey responses of 513 local

election jurisdictions. 15 With appropriate maintenance and proper
operation, most equipment performs on par with each other. Some voting
technology is easier to use thus eliminating some opportunities for voter
error. Overall, our analysis of both the vendor and jurisdiction data showed
that DREs are slightly easier to use and slightly more efficient than

the other types of equipment. In the area of security, DRE and optical scan
are relatively equal, and in the area of accuracy, all equipment is
relatively the same.

14 Our vendor survey did not include lever machines because these machines
are no longer manufactured. 15 In our survey of jurisdictions, we grouped
those that used punch card, lever, and handcounted paper ballots, and placed
them in an ?other? category. In our vendor survey, we excluded lever
equipment because it is no longer manufactured and, of course, handcounted
paper ballots, for which no equipment is needed.

Figure 10: Voting Technologies Perform Roughly on Par With Each Other GAO
analysis of

GAO analysis of vendor- reported data jurisdiction- reported data

Technology Technology Characteristics

DRE Optical

Punch DRE

Optical Othera

Accuracy Ease of use Efficiency Security

= Technology type performs slightly better relative to other technology
types. = Technologies have relatively equal performance, or data were
insufficient to judge. = Technologies have relatively worse performance. a
Other includes punch card, lever machines, and paper ballots

Source: GAO analysis.

The differences among voting equipment can be attributed, in part, to the
differences in the equipment itself. However, they also can be attributed to
the people who use the equipment and the rules or processes that govern its
use. Further, all voting equipment is influenced by security, testing,
maintenance, and cost issues, each of which also involves people and
processes. In addition, the accuracy of voting equipment (as measured by how
reliably the equipment captures the voter?s intent) can be affected by the
processes and procedures that govern how voters interact with the
technologies. Differences in these procedures can have noticeable effects on
the

prevalence of undervotes (votes for fewer choices than permitted, such as
not voting for president) and overvotes, for example. In particular, we
found that some precinct- count optical scan voting equipment can be
programmed to return a voter?s ballot if the ballot is overvoted or
undervoted. Such programming allows the voter to make any changes

necessary to ensure that the vote is recorded correctly. However, not all
states allow this. For example, election officials in one Virginia
jurisdiction state that Virginia jurisdictions must accept ballots as cast.

Interaction Between People and Technology Affect Uncounted Votes

Our analysis showed that the type of voting equipment used (including
equipment which allowed for error correction) explained a relatively small
percent of the total variation among jurisdictions in uncounted presidential
votes. The state in which counties were located had more of an effect on the
number of uncounted presidential votes than either a county?s voting
equipment or demographic characteristics. Figure 11 shows the results of our
analysis. 16 Figure 11: Percentage of Total Variation in Uncounted
Presidential Votes Explained

By County Demographics, Voting Equipment, State Differences, and Unknown
Factors

Unknown State differences

52% 26%

County demographics

16%

overall Voting equipment 2%

Error correction 4% a a This contribution of 4% results from an analysis of
the 404 counties for which we had error correction information. Source: GAO
analysis.

Counties? demographic characteristics also affected their percentages of
uncounted presidential votes. Specifically, counties with higher percentages
of minority residents tended to have higher percentages of uncounted
presidential votes, which counties with higher percentages of younger and
more educated residents tended to have lower percentages of 16 Our findings,
which are based on aggregate statistics and only those data that were
comprehensively available for the more than 2,000 counties included in our
analyses, have methodological limitations that are inherent to statistical
studies of this type.

uncounted presidential votes. Counties that used punch card equipment did
not generally have higher percentages of minority, less educated, or lower
income residents.

We found that the state in which counties are located had a greater effect
on counties? percentage of uncounted presidential votes than did counties?
voting equipment or demographic characteristics combined. State differences,
which may have included such factors as statewide voter education efforts
and state standards for determining that is a valid vote, accounted for 26
percent of the total variation in uncounted presidential votes across
counties. County demographic characteristics accounted for 16 percent of the
variation. Voting equipment, including the use of optical scan error
correction technology, accounted for a total of about 6 percent of the
variation in counties? uncounted presidential votes. The largest percentages
of uncounted presidential votes tended to occur in counties

that used punch card equipment. Counties that used optical scan equipment
with error correction had about 1.1 percentage points fewer uncounted
presidential votes than did counties with punch card equipment. The
remaining 52 percent of variation was due to unknown factors such as whether
a county switched to a new type of voting equipment or the

number of inexperienced voters in a county. Absence of Data Limit Ability to
Looking back to the technology used in the November 2000 elections, our Make
Informed Investment

survey of jurisdictions showed that the vast majority of jurisdictions were
Decisions satisfied with the performance of their respective technologies.
However, this satisfaction was in most cases not based on hard data, but on
the

subjective impressions of election officials. While these impressions should
not be discounted, informed decision- making regarding where to make the
most appropriate investments, for example, in new equipment, training for
election workers, or voter education, requires more objective data.
Acquiring new voting equipment is not the only investment option
jurisdictions may consider, and, in some cases, may not be the most
appropriate solution for jurisdictions who find their voting equipment does

not meet their needs. Making wise technology investment decisions present a
challenge to our election systems. It is extremely important that election
officials be able to define, measure, evaluate voting equipment performance
so that they may

properly assess whether their current technology is meeting their needs.
This information is also important as election officials consider the
suitability of available technology options to get the best return on their
investment if they choose to modernize their voting equipment.

However, we found that about half of the jurisdictions did not collect
actual performance data for the voting equipment that they used in the
November 2000 election. Table 1 shows the percentage of jurisdictions that
collected

data on accuracy (which is one measure of performance) by type of voting
equipment.

Table 1: Percentage of Jurisdictions That Collected Data on Accuracy, by
Type of Voting Equipment

Technology Percentage of jurisdictions that collected accuracy data

DRE 44 a Optical scan 54 b Other c 42 b a The 95- percent confidence
interval is plus 7 or minus 8 percentage points. b The 95- percent
confidence interval is plus or minus 7 percentage points. c Other includes
punch cards, lever machines, and paper ballots.

Source: GAO mail survey of jurisdictions.

Further, it is unclear the extent to which jurisdictions have meaningful
performance data. For those local election jurisdictions that we visited
that stated that their voting equipment was 100- percent accurate, none was
able to provide actual data to substantiate these statements. Similarly, the
results of our mail survey indicates that only about 51 percent of
jurisdictions nationwide collected data on undervotes, and about 47 percent
of jurisdictions nationwide collected data on overvotes for the

November 2000 election. Jurisdiction election officials were nevertheless
able to provide their perceptions about how the equipment performed. For
example, our mail survey results indicated that 96 percent of jurisdictions
nationwide were

satisfied with the performance of their voting equipment during the November
2000 election.

National Survey Results

We estimate that 96 percent of jurisdictions nationwide were satisfied with
the performance of their voting equipment during the November 2000 election.

GAO Mail Survey of Jurisdictions

These perceptions aside, a lack of performance data may limit jurisdictions?
abilities to select the most appropriate voting equipment that gives them
the best return on their investment. Thus, without reliable performance
data, were federal funds to be made available for the purchase of new voting
equipment, over half of the jurisdictions in the U. S. would not be in the
best position to make wise investment choices.

Updating and Implementing While no federal agency has been assigned explicit
statutory responsibility Voting Equipment Standards for developing voting
equipment standards, the FEC assumed this role Presents a Challenge

developing voluntary standards in 1990 for computer- based systems, and
Congress has supported this role with appropriations. These standards
describe specific performance benchmarks and address many- but not all-
types of systems requirements. However, these standards have not

been maintained and are now out of date (the FEC initiated plans to issue
revised standards in 2002).

According to FEC officials, the Commission has not proactively maintained
the standards because it has not been assigned explicit responsibility to do
so. Without current, relevant, and complete voting equipment standards,
states may choose not to follow them, resulting in the adoption of disparate
standards that could drive up the cost of voting equipment and produce

unevenness among states in the capabilities of their respective voting
equipment.

No federal agency has been assigned responsibility for or assumed the role
of testing voting equipment against the federal standards. Instead, NASED,
through its Voting Systems Committee, has assumed responsibility for
implementing the federal voting equipment standards by accrediting
independent test authorities, 17 which in turn, test voting equipment
against the standards. To this end, the committee has developed procedures
to

accredit the independent test authorities. 18 When testing is successfully
completed, the independent test authorities notify the NASED that the voting
equipment has satisfied testing requirements. 19 As of July 3, 2001, the
Association had qualified 21 voting equipment, representing 10 vendors.
Because development, maintenance, and implementation of voting

equipment standards are very important responsibilities, we are raising
matters for congressional consideration regarding the explicit assignment of
responsibility in these areas. Additionally, we are making recommendations
to the FEC aimed at improving its efforts to update its 1990 voting
equipment standards. 20

The Interaction of People, Our nation?s election systems are complex and
intricate. Successful Process, and Technology

election administration requires the appropriate integration of people,
Needs To Be Considered

processes and technology. In the challenges that we have categorized as
related primarily to this part of the system, it is important to note that
each of them is related in some way to another part of the system.
Therefore, in considering election reform proposals, it is important to
remember that people, processes, and technology issues should not be
addressed in isolation. Any reform proposal that influences one part of the
system, for example, a process, may have an unforeseen and perhaps
undesirable

17 Independent test authorities are contractors independent from the voting
equipment vendors and are responsible for testing voting equipment to ensure
that they meet the Commission's standards.

18 According to the NASED, accreditation signifies formal recognition that
the independent test authority possesses or will acquire the competence to
design and perform specific test methods applicable to voting equipment
hardware and software, and that the test authority has adequately
demonstrated its competence for voting equipment testing. 19 Independent
test authorities notify the Election Center, which serves as the NASED?s
Secretariat and maintains the list of the association's approved voting
equipment. 20 The matters for congressional consideration and
recommendations to the FEC were made in Elections: Status and Use of Federal
Voting Equipment Standards (GAO- 02- 52,

Oct. 15, 2001).

effect on another part of the system, such as the people. Additionally, any
problem attributed to people, processes, or technology might actually have
its root cause in a different part of the system or be precipitated by a
lack of integration among the components of the system.

For example, many problems that surfaced in the November 2000 election were
attributed to faulty technology, more specifically to punch card machines.
While voting technology may need to be modernized, our work showed that any
of the types of voting equipment, if used properly, can

reliably record a voter?s selections. In most cases, technology was not the
dominant factor related to voter error such as mismarked ballots or
uncounted votes. Rather, problems were more closely related to voter error
included processes that did not allow for poll workers and voters to
recognize when errors occurred. In fact, one of the jurisdictions we visited
that used punch card machines had a voter error rate of 1.2 percent, which
election officials attributed in part to voter education efforts. Greater
voter education on voting processes and equipment, rather than the purchase
of new voting equipment, can be a more immediate way to resolve issues

related to voter error. Section 3: Framework

Our work and the work of others have disclosed a number of challenges to for
Assessing Election

our national election system. As Congress considers if and how it may wish
to address these challenges, it may turn to a number of reform Reform
Proposals proposals put forth by commissions or by proposed legislation. The
proposals to date may be grouped into broad categories such as those listed
below:

 providing federal funds for replacing voting equipment,

 providing federal funds for state or locally determined election
administration needs,

 creating special postal rates, or requiring no postage, for election
materials,

 creating federal election administration standards, mandatory or
voluntary,

 updating FEC voluntary voting equipment standards and developing operation
standards for voting equipment,

 developing or improving electronic voter registration systems and
statewide information sharing capabilities,

 reforming absentee or early voting requirements,

 creating uniform statewide standards for what constitutes a vote and how
votes are counted and recounted, and

 mandating the availability of provisional ballots for all jurisdictions.
Variation in specific proposals may occur because of many factors, including
the source of the proposal.

Proposals may be crafted with various goals in mind- among them enhancing
the accessibility, integrity, fairness, consistency, affordability, and
sustainability of election systems. While all of these goals are

consonant with our democratic traditions, some reform proposals may advance
one goal at the expense of the other. For example, some officials promote
reforms such as early voting to enhance the accessibility of the electoral
process to the general public, while others claim such a move could open the
door to voter fraud and thus may come at the price of the integrity of the
election system. When reform proposals forward

competing goals, the debate over election reform becomes more complex and
assessing different reform proposals more difficult. We do not presume to
endorse any particular election reform proposal or

package, because this is best left to the Congress and other elected
officials. However, our review of state and local practices, as well as our
analysis of input from state and local officials, suggest criteria that
Congress could use as it weighs the merits of reform proposals.

 Criterion I: The Appropriate Role of the Federal Government in Election
Reform. Does the proposed change call for an appropriate federal role in
affecting election reform, given the historic balance struck between
Congress? constitutional authority to legislate election administration and
some states? laws and traditions that grant autonomy to local jurisdictions
as they administer elections?

 Criterion II: The Balance Between Accessibility and Integrity. 21 How are
the goals of providing citizens broad access to the voting process balanced
against the public?s interest in ensuring the integrity of our

election systems?

 Criterion III: Coordination and Integration of People, Processes, and
Technology. How does the proposed change affect both the discrete 21 As used
in this criterion, accessibility refers to voters? access to the political
process rather than the access of a voter with disabilities to the polling
place.

problem it is intended to resolve and the election system as a whole?

 Criterion IV: The Affordability and Sustainability of Proposed Election
Reforms. Have the necessary resources been identified to institute the
change and to continually monitor and re- evaluate it over

time? We believe using these criteria will help clarify the debate and
provide a framework to evaluate the potential effects of various election
reform proposals. The following sections further elaborate how Congress
might use each of the four criteria.

Criterion I The Appropriate Federal Role in Election

Reform A threshold consideration in assessing various proposals and
legislation is the appropriate role of the federal government in effecting
election reform. Pursuant to its constitutional authority, Congress has
periodically enacted legislation that mandates elections be conducted in
particular ways. For example, Congress has prohibited discrimination based
on certain voter characteristics, such as race or age for both state and
federal elections. In addition, Congress has broad authority to establish
requirements for congressional elections that are binding on the states. As
a practical matter, such requirements may also affect state and local
elections held in conjunction with elections to federal office. Congress has
enacted legislation affecting the timing of federal elections, voter
registration, absentee voting for military and overseas civilian citizens of
the United States, and voting accessibility for the elderly and the disabled
in federal

elections. These statutes have basically focused on facilitating the
opportunity for voters to participate in the voting process and ensuring
fair and equitable treatment of voters. Aside from direct regulation of
election

administration, Congress may also, in exercising its spending power,
encourage state action by attaching conditions to the receipt of federal
funds. The scope of congressional authority in election administration is
discussed in our March 2001 report. 22

22 GAO- 01- 470, Mar. 13, 2001.

This constitutionally- derived authority notwithstanding, election
administration has principally been the responsibility of state and local
jurisdictions that conduct elections for local, state, and federal offices.
As

discussed elsewhere in this report, states and local jurisdictions have
determined voter qualifications, types of voting equipment to be used,
ballot design, selection of poll workers, and what constitutes a vote. This
historical balance between Congress? constitutional authority to prescribe
change and the states? and localities? traditional roles in defining the
terms of election administration raises fundamental policy issues that must
be confronted in the debate about the federal government?s efforts to pursue

election reform. Various reform proposals offered to date differ in the role
envisioned for the federal government. These can be categorized into
essentially four distinct options for federal action that fall along a
continuum of low- to high- federal involvement. The first option, falling on
the low end of the continuum, calls for the federal government to provide
information, guidance, and encouragement to states and local jurisdictions
to take action in specific areas. At the second point on the continuum, a
reform

proposal may envision a more involved federal role that calls for the
federal government to provide funds to states and localities to improve
election administration, allowing each jurisdiction to use the funds where
it believes they are most needed. At the third point on the continuum,
proposals may go a step further suggesting that the federal government
provide funds, contingent on states and local jurisdictions taking specific

actions or achieving specific results. In this role the federal government
uses a ?carrot? to encourage a desired behavior by states and local
jurisdictions. However, the states and local jurisdictions still have the
ability to opt for the status quo by refusing the federal funds. At the
fourth point on the continuum, where election reform would involve the
greatest use of federal authority, the federal government would mandate that
state

and local jurisdictions take specific actions or achieve specific results--
with or without accompanying funding. Historically, this option has been
used when the federal government wishes to guarantee a voter right or
protection.

By way of illustration, consider how Congress would reach different
conclusions on the issue of replacing existing voting equipment given how it
might use its authority. Many recommendations from recently completed
studies and congressional legislative proposals call for providing federal
assistance to state and local election jurisdictions for replacing voting
equipment.

 Under the first option, Congress could require the FEC to act as a
clearinghouse to gather and disseminate information and to sponsor research
on the various types of voting equipment. The federal government would
provide information, assistance, and advice that would make it easier for
that state and local election jurisdictions to examine their choice of
voting equipment in light of national data on

specific practices or issues. This approach provides a minimal federal role
in affecting change, leaving the greatest discretion and control to states
and local election jurisdictions. This federal role would not entail
disbursing federal funds to support purchase and installation of new
technologies, leaving acquisition decisions up to local jurisdictions.

 Under the second option, the federal government could create a grant
program that would make federal funding available to states to support
purchase and installation of new voting equipment. Funds would be provided
with no ?strings? attached regarding which type of equipment the state could
buy.

 Under the third option, the federal government could create a similar
grant program to that in the second option, except that strings would be
attached. For example, funds would only be provided for states to buy
equipment that meets federal standards, or only for certain types of
equipment (e. g., precinct- based optical scanners). This further limits
local jurisdictions? discretion in choosing appropriate equipment and

allows states to opt out of the program.

 Under the fourth option, the federal government could mandate that only
certain types of voting equipment could be used in federal elections.
Congress might or might not provide funding to enable states without this
type of equipment to purchase and install it. Either way, jurisdictions
would have no choice but to comply with the law and

acquire that voting equipment, regardless of whether this choice best meets
local needs. This option provides the least discretion. Because federal
elections are usually conducted in conjunction with state and local
elections, congressional mandates regarding the conduct of federal

elections would likely involve changes in many, if not all, state and local
elections.

Criterion II Balancing Accessibility and Integrity

In addition to the appropriate federal role in election reform, the degree
to which reform proposals may affect the accessibility and integrity of an
election system, our work suggests, is an important criterion for Congress
to use in assessing the effect of reform proposals on our election systems.

Some proposals may seek to increase the general public?s access to the
election system. Accessibility describes the degree to which an election
system promotes inclusiveness, thus making it as easy as possible for the
general population to register to vote and to cast their votes. For example,

reform proposals that attempt to (1) make voter registration less
cumbersome, (2) give voters more opportunity to cast absentee or early
ballots, or (3) provide voting equipment that all voters can use with ease
can be considered as affecting an election system?s accessibility. Other
proposals may attempt to increase the system?s integrity, that is, the
degree to which the system is impervious to voter fraud. For instance,
proposals may implement controls to ensure that (1) voters present
identification or proof of eligibility at the polls on election day or (2)
all eligible votes are

counted can be said to affect an election system?s integrity. The goal of
making the election system more accessible to voters can run at cross
purposes with the goal of ensuring the election system?s integrity. This
tension suggests decision makers should ask how reform proposals

balance the goals of providing all citizens broad access to the voting
process against the public interest in ensuring the integrity of the vote.

Most election reform proposals address one or both of these concepts to some
degree, with some placing more emphasis on one or the other. The weight
individual policymakers may place on different concerns could vary,
depending on how they value different attributes. For example, increasing
the opportunity to use absentee ballots may improve access to the vote, but
it also might negate the possibility of using some of the controls that may
be used at a polling place to assure voter identification and eligibility.
If increasing access to the vote is deemed more important than ensuring the

presence of rigorous controls, then reform proposals emphasizing
accessibility considerations might be preferred.

In the past when Congress has taken action to change the election system, it
considered both accessibility and integrity issues. Constitutional
amendments and federal legislation affecting the election process opened

access to those whose access was either denied or circumscribed- e. g.,
African- Americans, women, language minorities, people with disabilities.
All of these reforms assumed the existence of controls to ensure that only
those who were otherwise eligible among these groups would be able to
register to vote and cast their ballots. The most recent federal statute

affecting the election process- NVRA, or motor voter act- specifically
recognized the dual goals of access and integrity. The act established
registration procedures designed to ?increase the number of eligible
citizens who register to vote in elections for Federal office,? without
compromising ?the integrity of the electoral process? or the maintenance of

?accurate and current voter registration rolls.? 23 Creating a proper
balance between accessibility and integrity is sometimes difficult, as seen
in the following examples which illustrate the inherent tension between
these competing goals:

 Requiring that citizens who register to vote present a form of picture
identification with their residence?s address provides some proof of
identify and some assurance that the person resides in the voting
jurisdiction and is therefore eligible to vote in the jurisdiction in which
he or she is registering. However, this procedure makes it more difficult

for persons to register to vote if they meet registration qualifications but
do not have a driver?s license or other picture identification indicating
their place of residence.  Providing a provisional ballot to every person
who wishes to vote but is not listed in the poll books at the polling place
maximizes the

opportunity of every person to cast a ballot. 24 The purpose of checking
whether these persons are in fact registered to vote prior to counting 23
National Voter Registration Act of 1993, 42 U. S. C. 1973gg to 1973gg- 10.
Also see GAO- 01470, p. 12 and footnote 65). 24 Provisional ballots are not
an issue in those jurisdictions that do not require registration, such as
North Dakota, or that permit eligible voters to register and vote on
election day. Election day registration is permitted in six states.

their ballots is to assure that the vote count will include only ballots
cast by eligible voters.  Election officials may go to nursing homes to
review absentee ballot

requests, and accept ballots cast by nursing home residents while the
officials are present to supervise the voting process. In effect, the
nursing home becomes an unofficial polling place. Residents are provided a
greater opportunity to vote but with better controls in place to address the
potential that the absentee ballot might be voted by someone other than the
registered voter.

 Allowing remote voting via the Internet may improve some voters?
opportunities to cast a ballot. Although this method is in an experimental
stage, unresolved questions about its impact on the integrity of election
systems remain. As with other forms of remote voting, such as absentee
voting by mail, there is a need to ensure that

only eligible voters cast ballots, voter privacy is protected, and voters
are not subject to coercion.

Criterion III Integration of People, Processes and Technology

As Congress assesses various reform proposals, it may consider both reforms
that address a discrete problem and that address the election system more
broadly. Congress may also be asked to choose among proposals that address
specific parts of a perceived problem or address perceived problems in a
more systematic fashion.

Effective election administration requires the appropriate coordination and
integration of people, processes, and technology. For example, successfully
registering a new voter, whether the person registers by mail, at the
Department of Motor Vehicles, or at the registrar?s office, involves the
coordination and integration of (1) voters and registration workers who

know and follow the registration process, including obtaining the
information required to register successfully; (2) the process for
registering new voters that guides election workers as they supply the
correct forms to voters, compile and update voter information, and notify
voters of their registration status; and (3) a computer system or other
means of creating and updating a voter registration list to assure an

accurate, current list of registered voters. Shortcomings in any of three
areas can affect the ability of persons to register successfully and the
accuracy of the registration rolls.

To illustrate the difference between approaching reforms from a discrete
versus system- wide perspective, consider election reform proposals that
recommend Congress provide assistance to states and localities to purchase
new voting equipment in order to reduce voter error. Some of these proposals
approach voter error as if it were precipitated by a single cause, such as
the type of voting equipment used. However, introducing new technology alone
may not necessarily reduce voter error. In fact, switching equipment
actually may introduce new opportunities for voter error unless the
jurisdiction deals with the people aspects of successfully fielding new
voting technology and offers voter education on how to use

the new equipment effectively. Moreover, successful implementation must
include processes for dealing with machine failure, ensuring that the
equipment is programmed properly to accurately count the votes for each
office on the ballot and that the ballots and machines are secure. Failure
to consider the interaction of people, processes, and technology in fielding
new voting equipment may result in increased voter or counting error, rather
than a decrease.

As a second example, some proposals suggest a change in process that
establishes standard voting hours such that all polling places across the
nation are opened and closed simultaneously, regardless of time zone. With
this proposal, voters in every part of the country would cast their votes
while the outcome of the election is still unknown, thus negating any
influence that media reports of election results may have. Although voter
participation might increase as a result, this proposal might also have

unintended consequences for other parts of the election system. For example,
keeping polling places open at earlier or later hours may increase the
burden placed on poll workers. Some elections officials currently report
difficulties in securing poll workers who are willing to work 15 to 18 hour
days, a situation that might be exacerbated if poll workers were asked to
work even earlier or later.

Criterion IV Affordability and Sustainability of Proposed Election Reforms

The implementation of election reforms will likely increase the overall cost
of our nation?s election systems. Choosing among election reform proposals,
therefore, should include a careful assessment of the affordability and
sustainability of the reform as well as who is expected to shoulder the
costs. Simply making funding available to state and local governments to
implement a reform without consideration of whether all associated lifecycle
costs have been considered or how the reform is to be

sustained could result in having to revisit reform issues. Historically,
incurring the costs of election administration and equipment has largely
fallen to local jurisdictions, with some support provided by the state
governments. Elections have been conducted over the years with relatively
small budgets, and election officials consistently find themselves

competing for funds with other local government priorities. Our work
confirmed that election administration is not usually at the top of state
and local funding priorities. As an official in a large election
jurisdiction told us, election administration is often number 11 in the top
10 priorities of local government budgets.

As a result, realistic reform proposals are those that not only identify
solutions to the issues at hand, but are also affordable and sustainable
with achievable financial commitments for the federal, state, and local
government stakeholders. Along these lines, as Congress assesses reform
proposals, it should consider three factors related to affordability and
sustainability. The first factor is the initial outlay required to fully
implement the proposal. In this regard, the assessment should consider
whether the initial outlay for the proposed reform would be affordable to
the state and

localities, including all associated and transition costs (e. g., training
workers and voters to use new equipment and any changes to voting processes
necessitated by new equipment). For example, were Congress to implement a
proposal that requires all states to develop a state- wide voter
registration system, some states might find themselves unable to comply
unless federal funding sources were forthcoming. States with

budgetary pressures that prevented them from quickly implementing the
federal requirement could be even less likely to supply the funding to
comply with the federal requirements. In addition, the costs to the other
components of election administration (e. g., the cost of training election
workers to use the new voting technology) should also be considered.

The second factor is whether the federal government and/ or state and local
jurisdictions could afford the long- term costs of sustaining the proposed
reform over time. Reform proposals that provide funding for purchasing new
technology could enable some jurisdictions to upgrade their voting
equipment. To the extent that a local jurisdiction could sustain the funding
needed to continue its use, the performance of voting equipment would be
improved. For example, if the federal government were to make funds
available to purchase different voting technology (e. g., replacing punch
cards with electronic equipment), it should have some assurance that the
additional resources necessary to sustain the reform (e. g., software,

programming capability, vendor support, and updates) would be available.
However, not all jurisdictions are in a position to make that commitment.
The third factor is assigning responsibility for costs, and whether all
levels of government could commit to implement and sustain the reform. As
mentioned above, it is doubtful that every local jurisdiction could alone
commit the resources necessary to fund many of the reforms envisioned in
several proposals. However, they might be in a position to fund some of

them, thereby making a commitment to the success of the reform. On the other
hand, the question arises as to how much of a federal or state presence in
local election administration is perceived as desirable or financially
possible.

Section 4: Conclusion Events surrounding the November 2000 election brought
into question the integrity of our nation?s election systems. Although not
all states and jurisdictions reported experiencing major problems during the
November 2000 election, important concerns were raised in most jurisdictions
related to each stage of the election process- registration, absentee and
early voting, preparing for and conducting election day activities, and vote

tabulations. Congress has the opportunity to address these challenges now,
to avoid similar problems in the future. However, addressing these
challenges involves complex considerations and even more difficult choices
when considering the range of proposals for election reform.

Accordingly, we have offered four criteria against which any election reform
proposals may be measured. These may not be the criteria that every analyst
would suggest, and each policymaker would not give the same weight to each
criterion. However, if election system reform proposals were to be evaluated
as to the (1) appropriate federal role in election reform; (2) balance
between accessibility and integrity; (3)

integration of people, process, and technology; and (4) affordability and
sustainability of election reforms, Congress would have a good foundation
for devising sustainable solutions that will meet the needs of future
generations of U. S. citizens.

Appendi Appendi xes x I

Results of Our Election Work This section summarizes the major issues
contained in the other six reports that we prepared on our nation?s election
systems. Table 2 lists the issues that we addressed in our elections work
and the reports that discuss them in further detail.

Table 2: Election Issues and GAO Reports Election Issues Addressed GAO
Election Reports

The scope of congressional authority in election administration Elections:
The Scope of Congressional Authority in Election Administration (GAO- 01-
470, Mar. 2001) Compiling and maintaining voter registration lists, absentee
and Elections: Perspectives on Activities and Challenges Across the

early voting, election day administration, voting technology, Nation (GAO-
02- 03, Oct. 2001) Internet- based voting technologies, vote counts,
certification, and recounts

Voting accessibility for voters with disabilities Voters with Disabilities:
Access to Polling Places and Alternative Voting Methods (forthcoming)

Improving voting assistance for military and overseas voters Elections:
Voting Assistance to Military and Overseas Citizens Should be Improved (GAO
01- 1026, Sept. 2001) Status and use of federal voting equipment standards
Elections: Status and Use of Federal Voting Equipment Standards (GAO- 02-
52, Oct. 2001) Factors affecting uncounted votes in the 2000 presidential
election Elections: Statistical Analysis of Factors that Affected Uncounted

Votes in the 2000 Presidential Election (GAO- 02- 122, Oct. 2001)

Collectively, our extensive research shows that election systems vary widely
across states and jurisdictions. It also shows that federal, state, and
local governments face daunting, often long- standing challenges. In the
following sections we summarize our findings and insights from each of our
reports.

The Scope of This report describes Congress? constitutional authority to
regulate Congressional congressional, presidential, and state and local
elections and identify major federal statutes enacted in the area of
election administration. Under the Authority in Election

Constitution, states are responsible for the administration of both their
Administration own and federal elections. Accordingly, states and localities
incur the costs associated with these activities.

Notwithstanding the state role in administering elections, Congress has
authority to affect the administration of elections in certain ways.
Congressional authority to legislate in this area derives from various
constitutional sources, depending upon the type of election. With regard to

the administration of federal elections, Congress has constitutional
authority over both congressional and presidential elections.

Congress? authority to regulate congressional elections derives primarily
from Article I, Section 4, Clause 1 of the Constitution (known as the
Elections Clause). The Elections Clause provides that the states will
prescribe the ?Times, Places and Manner? of congressional elections, and

that Congress may ?make or alter? the states? regulations at any time,
except as to the places of choosing Senators. The courts have held that the
Elections Clause grants Congress broad authority to override state
regulations in this area. Therefore, while the Elections Clause contemplates
both state and federal authority to regulate congressional

elections, Congress? authority is paramount to that of the states. With
respect to presidential elections, the text of the Constitution is more
limited. Specifically, Article II, Section 1, Clause 4, provides that
?Congress may determine the Time of chusing the Electors, and the Day on
which they

shall give their Votes; which Day shall be the same throughout the United
States.? Despite this limited language, the Supreme Court and federal
appellate courts have upheld certain federal statutory provisions regulating
presidential elections that go beyond regulating the ?time? of choosing the
electors. However, because federal legislation that relates solely to the
administration of presidential elections has been fairly limited, case law
on this subject has been sparse. Consequently, the precise parameters of
Congress? authority to pass legislation relating to presidential elections
have not been clearly established. With regard to state and local elections,
although Congress does not have

general constitutional authority to legislate regarding these elections, a
number of constitutional amendments authorize Congress to enforce
prohibitions against specific discriminatory practices in state and local
elections, such as discrimination on the basis of race or color, in all
elections- federal, state, and local.

Historically, Congress has passed legislation related to the administration
of both federal and state elections in several major functional areas of the
voting process, including: (1) timing of federal elections; (2) voter

registration; 1 (3) absentee voting requirements; 2 (4) accessibility
provisions for elderly and disabled voters; 3 and (5) prohibitions against
discriminatory voting practices. 4 In general, the purpose of these federal
statutes has been either to prohibit discrimination on the basis of specific
voter characteristics or make it easier for citizens to register to vote.

Elections: Perspectives Our report on the administration of the 2000
elections presents the results

on Activities and of our review of aspects of elections in the United
States. Specifically, we

(1) describe elections in the United States and the activities and
challenges Challenges Across the associated with each of the four major
stages of election administration-

Nation voter registration, absentee and early voting, preparing for and
conducting

election day activities, and vote tabulation; (2) identify the types of
voting methods used, their distribution in the United States, and any
associated challenges; assess such characteristics of voting equipment as
accuracy,

ease of use, efficiency, security, and cost; and estimate the cost of
replacing existing voting equipment in the United States with either optical
scan or electronic voting equipment; and (3) identify issues and challenges
associated with the use of the Internet for voting.

Although registration is a prerequisite to voting in nearly all states, we
found that different citizens with the same qualifications would be eligible
to vote in some states but not in others because of variations in voter
eligibility requirements. A citizen?s access to voting is primarily based on
the appearance of his or her name on a voter registration list, which is 1
The National Voter Registration Act of 1993, commonly referred to as ?motor
voter.? The Act requires states to establish certain procedures to
facilitate the registration of voters in

federal elections, both congressional and presidential, including permitting
voters to register by mail or in person when applying for a driver?s
license. 2 Uniformed and Overseas Citizens Absentee Voting Act of 1986
requires states to permit specified domestic and overseas military personnel
and their dependents and citizens living overseas to vote by absentee ballot
in federal elections. 3 The Voting Accessibility for the Elderly and
Handicapped Act of 1984 requires, with some exceptions, local election
jurisdictions to assure that polling places and registration sites are
accessible to elderly and disabled voters. 4 The Voting Rights Act of 1965,
as amended, provides, in general, that no person shall be denied the right
to vote on account of race or color.

developed from registration applications and compiled and maintained by
election officials using various technologies and information sources.
Election officials nationwide expressed varying degrees of confidence in

the accuracy of their voter registration lists; however, information about
list accuracy and currency, as well as the extent of error, was difficult to
obtain. Among the challenges identified were processing applications

submitted through sources other than elections offices, such as state motor
vehicle authorities; obtaining accurate and timely information from numerous
sources to update voter registration lists; and using technology to help
process applications and compile registration lists.

All states allowed some form of absentee or early voting in the November
2000 election; using U. S. Census data, we estimate that for the November
2000 election, about 14 percent of voters nationwide cast their ballots
before election day, three- fourths of them using mail- in ballots and
onefourth voting in person. However, we found that no national data are
currently maintained on the number of mail- in absentee ballots
disqualified. Differences in requirements, administration, and procedures
resulted in citizens having different opportunities for obtaining and

successfully casting absentee ballots. For example, the likelihood of a
ballot being disqualified due to voters? errors in completing and returning
mail- in absentee ballots varied even, in some instances, among
jurisdictions in the same state. Among the challenges local election
officials face are deciding whether and how to process incomplete and late
mail- in absentee applications and ballots; processing large numbers of

mail- in absentee applications and ballots in a timely manner; and obtaining
adequate staffing, ballots, and locations for conducting early voting.

Election officials across the country, with some variation, performed
similar duties to prepare for and conduct the November 2000 election. Our
survey indicated that 57 percent of voting jurisdictions nationwide
encountered major problems in conducting the 2000 election. While

jurisdictions did not experience the same problems, more than half cited
problems with recruiting enough qualified poll workers. However, because few
jurisdictions systematically collected information on how they administer
elections, what they consider to be major problems may be based on anecdotal
information and limited analysis. From the perspective of election
officials, a major election day challenge is resolving questions about voter
eligibility. Large numbers of ineligible voters can create long lines, voter
frustration, and problems for poll workers. Many eligibility issues stem
from the reliability of voter registration lists.

Counting votes is not a simple task. It involves counting votes cast before
and on election day and may be carried out at the precinct, at a central
location, by hand, or by some type of counting equipment. Vote counting
problems are highlighted when the election results are close. A ballot may
not be counted when a voter overvotes- marks for two candidates-- or when
the ballot cannot be read by the counting equipment. What constitutes a
proper mark on a ballot differs based on the type of voting

method used. According to our survey of state election directors, 31 states
and the District of Columbia have a state law or other provision specifying
what is a proper ballot marking for each voting method, but state guidance
also varied from general to specific. Forty- seven states and the District
of Columbia have laws with provisions for recount, but they vary. For
example, 17 states provide for a mandatory recount, but two of these require
a tie vote and another requires a 1 percent or 200 vote difference.

According to officials in 42 of the jurisdictions that had recounts in the
2000 election, none of the recounts altered the original election outcome.
Counting ballots posed several challenges for election officials, including
counting only votes cast by eligible voters; interpreting variations when
ballots are not properly marked; and completing the results of a recount in
a close or contested election in a fair, accurate, and timely manner.

Four types of voting equipment- punch card, optical scan, lever, and DRE-
were used in 98 percent of all election jurisdictions in 2000. 5 While a
survey of vendors showed little performance difference among DRE, optical
scan, and punch card equipment, local election officials we

contacted rated DRE as more easy to use than other voting methods. Only
about 50 percent of jurisdictions collected data on accuracy and few of the
jurisdictions we visited had collected actual performance data on the voting
equipment used in the 2000 election. Nevertheless the vast majority of
jurisdictions across the nation were satisfied with their respective voting

equipment, based largely on officials? perceptions of how their equipment
performed. The cost to replace existing voting equipment depends on the type
of equipment purchased and the number of jurisdictions for which it is
purchased. We estimated the cost of purchasing optical scan units, not
including certain software costs, could range from $191 million for optical
scan machines that use a central- count unit to about $1.3 billion for
optical scan equipment that counts ballots at the precinct and thus allows
for voter error correction. We estimated the cost of purchasing touchscreen
DRE units to be about $3 billion (including one DRE touchscreen unit per

5 Two percent of jurisdictions used paper ballots; no ?equipment? was
required.

precinct equipped for voters with disabilities and one central count optical
scan unit per county for absentee ballots). Among the voting equipment
challenges identified by election officials were having reliable measures
and objective data to know whether the technology used is meeting the
jurisdictions? needs; ensuring that necessary security, testing, and

maintenance activities are performed; and ensuring that the technology will
provide benefits over its useful life commensurate with life cycle costs and
that these collective costs are affordable and sustainable. Our review
identified three kinds of internet voting-- at a polling place; in a voting
?kiosk? at public places, such as malls or libraries; or at any location,
including the voter?s workplace or home through a personal computer.
Although opinion is not unanimous, security is seen as the primary challenge
for Internet voting. The cost effectiveness of Internet voting remains
unclear because reliable cost data are not available. The broad application
of Internet voting presents several social and technological challenges,
including providing adequate voter privacy safeguards, security for voting
equipment to ensure that it is adequately safeguarded, and equal access to
all voters.

Voting Accessibility for This report examines state and local provisions and
practices for ensuring

Voters with Disabilities voting accessibility, both at polling places and
with respect to alternative

voting methods and accommodations; estimates the proportion of polling
places with features that might facilitate or impede access, including
features of polling booths and voting accommodations; and identifies efforts
and challenges to improving accessibility.

All states 6 have provisions (in the form of statutes, regulations, or
policies) that specifically address voting by people with disabilities.
However, consistent with the broad discretion afforded states, these
provisions vary greatly. State laws and policies also vary on how counties
are to assure accessibility of polling places. Our survey of counties
confirms that most counties inspect all polling places for accessibility,
although county

practices for ensuring accessibility vary. All states provide for one or
more alternative voting methods or accommodations that may facilitate voting
by people with disabilities 6 For analytical purposes we treated the
District of Columbia as a state, resulting in a total of 51 states.

whose assigned polling places are inaccessible. For example, all states have
provisions allowing voters with disabilities to vote absentee without notary
or medical certification requirements, although the deadlines and methods
(for example, by mail or in person) for absentee voting vary among states.
In addition, many states, but not all, have laws or policies that provide
for other accommodations and alternatives for voting on or before election
day- such as reassignment to a polling place that is accessible, curbside
voting, or early voting.

Our onsite work on election day 2000 found that polling places are generally
located in schools, libraries, churches, and town halls, as well as other
facilities. Although the extent to which any given feature may prevent or
facilitate access is unknown, we estimate that, from the parking

area to the voting room, 16 percent of all polling places in the contiguous
United States have no potential impediments, 56 percent have one or more
potential impediments but offer curbside voting, and 28 percent have one

or more potential impediments and do not offer curbside voting. 7 These
potential impediments would primarily affect individuals with mobility
impairments and occur most often on the route from the parking area to the
building or at the entrance to the polling place. Inside the voting room,
the types and arrangement of voting equipment used may also pose challenges
for people with mobility, vision, or dexterity impairments. To facilitate
voting inside the voting room, polling places generally provide

accommodations, such as voter assistance, magnifying devices, and voting
instructions or sample ballots in large print. However, none of the polling
places that we visited had special ballots or voting equipment adapted for
blind voters. 8

A number of efforts have been made by states and localities to improve
voting accessibility for people with disabilities, such as modifying polling
places, acquiring new voting equipment, and expanding voting options.
Nevertheless, state and county election officials we surveyed cited a 7
Although curbside voting is not available at a number of polling places with
potential impediments, as noted earlier all states have provisions for
absentee voting, and many states provide for other alternative voting
methods or accommodations, which may facilitate voting by people with
disabilities on or before election day.

8 Although we did not observe such aids on election day, some county
officials told us that, upon request, they try to provide special aids so
that blind individuals can vote independently. We may not have observed
these aids on election day because they may not

have been requested in advance by voters in the polling places that we
visited or the local poll workers we interviewed may not have been aware of
these aids.

variety of challenges to improving access, including limited availability of
accessible facilities and funding constraints at the local level. Some
disability advocates believe that although alternative voting methods and
accommodations, such as curbside voting, expand options for voters with
disabilities, they do not provide the same voting opportunities afforded the

general public (that is, the opportunity to vote independently and privately
at a polling place) and should not be viewed as permanent solutions for
inaccessible polling places.

Improving Voting This report describes the extent and quality of voter
assistance provided Assistance for Military for uniformed and overseas
citizens; the challenges that state and local

requirements may pose to these voters; and the extent of and reasons for and
Overseas Voters

disqualification of ballots cast by these voters. The Uniformed and Overseas
Citizens Absentee Voting Act of 1986 9 protects the right to vote by
absentee ballot in federal elections for more than 6 million military and
overseas citizens. The act also recommends that states adopt a number of
provisions that facilitate absentee voting by these populations. The Federal
Voting Assistance Program, established within the Department of Defense
(DOD), is responsible for implementing the act by informing U. S. citizens
worldwide about their right to vote, fostering voting participation, and
working with states to simplify the

registration and absentee voting process. Voter education and assistance
efforts for military personnel are largely implemented by the military
services through Voting Assistance Officers. Also, the State Department
works with DOD to provide voter assistance to overseas citizens. The extent
and quality of federal voter assistance for military personnel and overseas
citizens varied considerably for the 2000 general election. The Federal
Voting Assistance Program developed a number of useful tools

for voters and Voting Assistance Officers, but many potential voters we
spoke to were unaware of them. While some installations we visited had well
run programs providing assistance and information to potential voters, other
installations did not meet DOD and service requirements because they did not
provide sufficient numbers of trained Voting Assistance

Officers, voter training, and voting materials. The variability in executing
the program is due to incomplete service- level guidance that does not

9 42 U. S. C. 1973ff- 1973ff- 6.

reflect DOD?s directive, a lack of command support at some installations,
and a lack of program oversight by some DOD components. Finally, the State
Department provided citizens abroad with a variety of useful

assistance, according to overseas citizens and federal employees we spoke
to, although both groups believed more outreach could be beneficial. Also,
State Department Headquarters has not played an active role in sharing best
practices and lessons learned or in overseeing the program.

Despite progress made by states to facilitate absentee voting, many military
and overseas voters we spoke to believe that challenges remain, including
helping voters understand and comply with state requirements and local
procedures for absentee voting, such as deadlines for registering and

returning ballots. Continued efforts by DOD officials to work with the
states to simplify procedures, modify election schedules, or allow more use
of technology, such as faxing and the Internet, to speed some portions of
the voting process may help alleviate the challenges, but state legislative

actions may be required. Although precise numbers are not available, we
estimate that small counties (having a voting- age population of less than
60,000) nationwide disqualified 8. 1 percent (plus or minus 3.2 percent) of
ballots cast by military and overseas voters. In contrast, the ballot
disqualification rate for civilian voters not living overseas was 1.8
percent (plus or minus 0. 6

percent). 10 While larger counties (having a voting age population of more
than 60, 000) that responded to GAO's survey showed a similar pattern, the
data were insufficient to make a national estimate. 11 The survey showed
that for all absentee ballots cast, almost two- thirds of the disqualified
absentee ballots were rejected because the ballots arrived too late to be
counted or the envelopes or forms accompanying the ballots were not
completed properly. This report includes recommendations to the Secretaries
of Defense and State to improve (1) the clarity and completeness of service
guidance, (2)

10 The confidence intervals in this report are calculated at the 95- percent
confidence level. For example, we are 95 percent certain that the actual
disqualification rate for civilians not living overseas was between 1.2 and
2. 4 percent 11 Not enough counties in the sample provided data on the
reasons ballots were disqualified from specific groups of voters, such as
military versus non- military, to enable GAO to make reasonable estimates
for these groups

voter education and outreach programs, (3) oversight and evaluation of
voting assistance efforts, and (4) sharing of best practices. Status and Use
of

This report identifies the Federal Election Commission?s (FEC) role Federal
Voting regarding various voting equipment and assesses how well the FEC is
fulfilling its role. Our work also identifies the National Association of
State

Equipment Standards Election Directors? (NASED) process for testing and
qualifying voting equipment.

No federal agency has been assigned explicit statutory responsibility for
developing voting equipment standards; however, the FEC assumed this role by
developing voluntary standards in 1990 for computer- based

equipment, and Congress has supported this role with appropriations. These
standards describe specific performance benchmarks, and address many- but
not all- types of systems requirements. In 1997, the FEC initiated efforts
to evaluate the 1990 standards to identify areas to be updated, and in 1999,
initiated efforts to update the standards. The FEC plans to issue revised
standards in 2002. This update is necessary because the FEC has not
proactively maintained them, thus allowing them to become out of date.
According to FEC officials, the FEC has not proactively maintained the
standards because it has not been assigned

explicit responsibility to do so. Unless voting equipment standards are
current, relevant, and complete, states may choose not to follow them,
resulting in the adoption of disparate standards that could drive up the
cost of voting equipment and produce unevenness among states in the

capabilities of their respective voting equipment.

No federal agency has been assigned responsibility for or assumed the role
of testing voting equipment against the federal standards. Instead, NASED,
through its Voting Systems Committee, has assumed responsibility for
implementing the federal voting equipment standards by accrediting
independent test authorities, 12 which in turn, test voting equipment
against the standards. To this end, the committee has developed procedures
to

accredit the independent test authorities. 13 According to the test
authorities, testing is generally iterative, in which the voting equipment
vendors are provided an opportunity to correct deficiencies identified
during testing and resubmit the modified voting equipment for retesting.
When testing is successfully completed, the independent test authorities

notify the NASED that the voting equipment has satisfied testing
requirements. 14 As of July 3, 2001, the NASED had qualified 21 different
types of voting equipment, representing 10 vendors. Because development,
maintenance, and implementation of voting

equipment standards are very important responsibilities, we are raising
matters for congressional consideration regarding the explicit assignment of
responsibility in these areas. Additionally, we are making recommendations
to the FEC's Commissioners aimed at improving its efforts to update its 1990
voting equipment standards. Statistical Analysis of

Our analysis of data from 2,455 counties shows that the type of voting
Factors Affecting

equipment that counties used in the 2000 general election had an effect on
uncounted presidential votes. Specifically, counties that used punch card
Uncounted Votes in the

equipment had roughly 0. 6 percentage points higher percentages of 2000
Presidential

uncounted presidential votes than counties using electronic, paper, or
Election

optical scan voting equipment. Counties using lever equipment had 0.7
percentage points lower percentages of uncounted presidential votes than
counties using electronic, paper or optical scan voting equipment. When

12 Independent test authorities are contractors independent from the voting
equipment vendors who are responsible for testing voting equipment to ensure
that they meet the FEC standards

13 According to the NASED, accreditation signifies formal recognition that
the independent test authority possesses or will acquire the competence to
design and perform specific test methods applicable to voting equipment
hardware and software, and that the test authority has adequately
demonstrated its competence for voting equipment testing 14 Independent test
authorities notify the Election Center, which serves as the NASED?s
Secretariat and maintains the list of the association's approved voting
equipment

we supplement this analysis with information about the performance of
optical scan equipment with error correction from our sample of 404
counties, we found that counties using punch card equipment had
significantly higher percentages of uncounted presidential votes than
counties using error corrected optical scan equipment. If we apply the
relationship we found in these 404 counties to the larger set of 2,455

counties, an estimated 300,000 additional votes may have been counted if
counties that used punch card equipment had, instead, used optical scan
equipment with error correction. Overall, county voting equipment accounted
for 2% of the variation in uncounted presidential votes across counties.
Additionally, the analysis of the subset of 404 counties showed

that the use of error correction accounts for another 4% of variation in
uncounted presidential votes across counties. We found that counties?
demographic characteristics accounted for about 16 percent of the total
variation in uncounted presidential votes. Counties with higher percentages
of minority residents were more likely to have higher percentages of
uncounted presidential votes. Counties with higher percentages of 18- to -24
year olds and higher education were more likely to have lower percentages of
uncounted presidential votes.

The state in which counties are located accounted for about 26 percent of
the total variation in uncounted presidential votes. Data were not available
to examine the extent to which specific factors that were common to counties
within a state but varied across states affected uncounted presidential
votes. However, such factors may include statewide voter education efforts,
the number of candidates on the ballot, the extent to which absentee or
early voting occurred, and the state?s standards for

determining that is a valid vote. Non- election specific factors, such as
the percentage of the state?s population for which English is a second
language, may have also contributed to the variability in uncounted
presidential votes.

Our statistical models left about half of the variation in uncounted
presidential votes unexplained. Several factors may have contributed to this
remaining variability, including differences among counties, precincts, and
people. An example of this type of difference is whether a county had
switched to a new type of voting equipment that voters found difficult to
operate.

Our findings, which are based on aggregate statistics and only those data
that were available for our sample of 2,455 counties, and the subset of 404

counties, have methodological limitations that are inherent to statistical
studies of this type.

Appendi x II

List of Recently Issued Election Reports GAO Reports Elections: Perspectives
on Activities and Challenges Across the Nation

(GAO- 02- 03, Oct. 2001).

Elections: Status and Use of Federal Voting Equipment Standards (GAO02- 52,
Oct. 2001).

Voters with Disabilities: Access to Polling Places and Alternative Voting
Methods (forthcoming).

Elections: Voting Assistance to Military and Overseas Citizens Should be
Improved (GAO- 01- 1026, Sept. 2001).

Elections: Statistical Analysis of Factors that Affected Uncounted Votes in
the 2000 Presidential Election (GAO- 02- 122, Oct. 2001).

Elections: The Scope of Congressional Authority in Election Administration
(GAO- 01- 470, Mar. 2001).

Bilingual Voting Assistance: Assistance Provided and Costs (GAO/ GGD97- 81,
May 1997).

Other Recent National The Constitution Project, Building Consensus for
Election Reform (Aug. Reports

2001). National Association of Secretaries of State, State- by- State
Election Reform Best Practices Report (Aug. 2001).

The National Commission on Federal Election Reform, To Assure Pride and
Confidence in the Electoral Process (July 2001). National Conference of
State Legislatures, Voting in America: Final Report of the NCSL Elections
Reform Task Force (Aug. 2001). National Task Force on Election Reform,
Election 2000: Review and Recommendations by The Nation's Elections
Administrators (Aug. 2001)

Caltech- MIT Voting Technology Project, Voting: What Is, What Could Be (July
2001).

State Commission and (alphabetical order by state)

Task Force Reports The Governor?s Select Task Force on Election Procedures,
Standards, and

Technology, Revitalizing Democracy in Florida (Mar. 2001). Hon. Cathy Cox,
Georgia Secretary of State, The 2000 Election: A Wake- Up Call For Reform
and Change - Report to the Governor and Members of the General Assembly
(Jan. 2001).

Hon. Chet Culver, Iowa Secretary of State, Commissioner of Elections, and
Registrar of Voters, Iowa?s Election 2000: Facts, Findings, and Our Future
(Mar. 2001).

Hon. Ron Thornburgh, Kansas Secretary of State, Kansas Secretary of State?s
Six- Point Election Improvement Plan (Jan. 2001). Special Committee on
Voting Systems and Elections Procedures in Maryland, Report and
Recommendations (Feb. 2001).

Hon. Matt Blunt, Missouri Secretary of State, Making Every Vote Count:
Report of Secretary of State Matt Blunt to the People of Missouri (Jan.
2001).

Hon. Bob Brown, Montana Secretary of State, 2001 Election Reform Plan

(2001). Office of New York Attorney General, Eliot Spitzer, Voting Matters
in New York: Participation, Choice, Action, Integrity (Feb. 2001).

Hon. Henry Cellar, Texas Secretary of State, Texas Overvote/ Undervote Study
(Jan. 2001). Secretary of State Deborah L. Markowitz, Review of Vermont's
Election Administration and Proposals for Improvement (Jan. 2001).

a

GAO United States General Accounting Office

Page i GAO- 02- 90 Elections

Contents

Contents

Page ii GAO- 02- 90 Elections

Page 1 GAO- 02- 90 Elections United States General Accounting Office

Washington, D. C. 20548 Comptroller General

of the United States A

Page 2 GAO- 02- 90 Elections

Page 3 GAO- 02- 90 Elections

Page 4 GAO- 02- 90 Elections

Page 5 GAO- 02- 90 Elections

Page 6 GAO- 02- 90 Elections

Page 7 GAO- 02- 90 Elections

Page 8 GAO- 02- 90 Elections

Page 9 GAO- 02- 90 Elections

Page 10 GAO- 02- 90 Elections

Page 11 GAO- 02- 90 Elections

Page 12 GAO- 02- 90 Elections

Page 13 GAO- 02- 90 Elections

Page 14 GAO- 02- 90 Elections

Page 15 GAO- 02- 90 Elections

Page 16 GAO- 02- 90 Elections

Page 17 GAO- 02- 90 Elections

Page 18 GAO- 02- 90 Elections

Page 19 GAO- 02- 90 Elections

Page 20 GAO- 02- 90 Elections

Page 21 GAO- 02- 90 Elections

Page 22 GAO- 02- 90 Elections

Page 23 GAO- 02- 90 Elections

Page 24 GAO- 02- 90 Elections

Page 25 GAO- 02- 90 Elections

Page 26 GAO- 02- 90 Elections

Page 27 GAO- 02- 90 Elections

Page 28 GAO- 02- 90 Elections

Page 29 GAO- 02- 90 Elections

Page 30 GAO- 02- 90 Elections

Page 31 GAO- 02- 90 Elections

Page 32 GAO- 02- 90 Elections

Page 33 GAO- 02- 90 Elections

Page 34 GAO- 02- 90 Elections

Page 35 GAO- 02- 90 Elections

Page 36 GAO- 02- 90 Elections

Page 37 GAO- 02- 90 Elections

Page 38 GAO- 02- 90 Elections

Page 39 GAO- 02- 90 Elections

Page 40 GAO- 02- 90 Elections

Page 41 GAO- 02- 90 Elections

Page 42 GAO- 02- 90 Elections

Page 43 GAO- 02- 90 Elections

Page 44 GAO- 02- 90 Elections

Page 45 GAO- 02- 90 Elections

Page 46 GAO- 02- 90 Elections

Page 47 GAO- 02- 90 Elections

Page 48 GAO- 02- 90 Elections

Page 49 GAO- 02- 90 Elections

Page 50 GAO- 02- 90 Elections

Page 51 GAO- 02- 90 Elections

Page 52 GAO- 02- 90 Elections

Page 53 GAO- 02- 90 Elections

Page 54 GAO- 02- 90 Elections

Page 55 GAO- 02- 90 Elections

Page 56 GAO- 02- 90 Elections

Appendix I

Appendix I Results of Our Election Work

Page 57 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 58 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 59 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 60 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 61 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 62 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 63 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 64 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 65 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 66 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 67 GAO- 02- 90 Elections

Appendix I Results of Our Election Work

Page 68 GAO- 02- 90 Elections

Page 69 GAO- 02- 90 Elections

Appendix II

Appendix II List of Recently Issued Election Reports

Page 70 GAO- 02- 90 Elections

United States General Accounting Office Washington, D. C. 20548- 0001

Official Business Penalty for Private Use $300

Address Correction Requested Presorted Standard

Postage & Fees Paid GAO Permit No. GI00
*** End of document. ***