Background

Since the election controversies of 2000, scholars of American politics have invested research time and effort into the study of election administration and election performance. These include studies on residual vote analysis (Ansolabehere and Stewart, 2005; Mebane, 2004; Wand et al., 2001), election auditing (Alvarez, Atkeson and Hall 2012; Atkeson et al., 2008), the role of poll workers (Alvarez and Hall, 2006; Atkeson, Alvarez, and Hall., 2009b; Claassen et al., 2008; Hall, Monson, and Patterson, 2009), the role of technology (Alvarez and Hall, 2004; Kimball and Kropf, 2005, 2008; Knack and Kropf, 2003b; Stein et al., 2008; Tomz and Van Howling, 2003), provisional votes (Alvarez and Hall, 2009; Atkeson, Alvarez, and Hall, 2009a; Kimball and Foley, 2009; Kimball, Kropf, and Battles, 2006; Pitts and Neumann, 2009), voter identification (Ansolabehere, 2009; Atkeson, et al., 2010; Cobb, et al., 2012; Pitts and Neumann, 2009), and voter confidence (Alvarez, Hall, and Llewellyn, 2008a; Atkeson and Saunders, 2007; Bullock, Hood, and Clark, 2005; Claassen et al., 2008; Murphy, Johnson, and Bowler, 2011), among others. These studies have been in direct response to the presidential election meltdown in 2000, which for the public focused largely on Florida, but was also seen in other states, especially those where the race was very close, including New Mexico and Ohio (Caltech/MIT Voting Technology Project, 2001). This work has been highly productive, creating new linkages between political scientists, local election officials, and legal professionals to create a data-driven approach to election reform and a push to improve and modernize the local election systems across the nation (Alvarez et al., 2009; Atkeson, Alvarez, and Hall 2010; Atkeson et al., 2011; Gerken, 2009; Liebshutz and Palazzolo, 2005).

Understanding the factors that contribute to voter confidence is an important component of the election performance literature. The interest in voter confidence stems from very visible problems in the election process observed since 2000 (Atkeson et al., 2010; Hall et al., 2009; Sinclair and Alvarez, 2004; Tomz and Van Houweling, 2003; Wand et al., 2001), attention by the mass media to the possibility that voting machines may not be counting the votes correctly, weak computer security in many systems that allows for break-ins (Kohno et al., 2004), the specter of voter fraud by citizens (Griffin and Johnston, 2008), and procedural manipulations by election officials to potentially change turnout (Kennedy, 2006; Koppelman, 2010).

Providing useful information to interested parties about the election process is the responsibility of election officials. Useful information is easily accessible and transparent. This implies openness, communication, and accountability. By accessible and transparent, I mean information or data that are relatively easily available and easy to understand. Transparency invites verification. Election officials are responsible for conducting elections in a free and fair manner. To help determine that an election is free and fair, data produced during the election must be made available for independent examination. That information must be transparent.

The concepts of accessibility and transparency focus on the ease and availability of information so that those involved in the election can participate in and evaluate it. These include voters, perhaps the primary client in an election, as well as other stakeholders, including election administrators, policy makers, legislators, policy advocates, election activists, and scholars. For voters, material that assists them in the voting process, including information about the ballot, how to register, if their votes were counted, where to vote, and how to obtain an absentee ballot, are all critical pieces of information. For other stakeholders, information is needed to assess the quality of the election, for example, to determine that the election was legitimate, to verify that the candidate who received the most qualified votes won the election, and to suggest ways in which the process can be improved.

The first question that arose in the immediate aftermath of the 2000 presidential election, when we started to study the performance of election processes in the United States, was a simple one: what data were readily available that could be used to assess the problems seen in the election, especially in Florida? That question led those of us who were involved in the early stages of the Caltech/MIT Voting Technology Project (VTP) to focus on the measure called the residual vote – a measure that we now see as a vital and readily available gauge of the accuracy and reliability of voting technologies.

However, as we discussed in the first two chapters of this book, there are other readily available sources of election performance data that election officials routinely collect but typically do not employ in a broader and more comprehensive performance-based management and evaluation process. These data form the basis for developing a set of indicators that can be used to assess the performance of elections in a given jurisdiction and the backbone for a performance-based management system that can be used to engage in ongoing quality improvements of the electoral process.

There are a common set of questions journalists, election administrators, and candidates sometimes ask about the administration of elections in the United States and internationally. A reporter from a local newspaper will ask whether recent elections in the area were run well. During an interview with a national media outlet, one of us will be asked, in our roles as academics who study election administration, to opine about whether some state has recently done a good job administering a presidential, primary, or other federal, state, or local election. More broadly, we have been asked by journalists from other countries to evaluate how well elections across the United States have been run since the 2000 presidential election.

Although these questions may seem simple, they are inherently difficult and complicated because (1) election administration involves a complex set of procedures, (2) there are many possible aspects of an election to consider to determine if it was “run well,” and (3) currently there is no accepted framework to assess the general quality of an election. More troubling, the question is focused solely on making a snap judgment about a given election – without taking the context of the election or the jurisdiction into account – and is not concerned with improving election management.

It has been more than a decade since the events in the 2000 presidential election, especially in Florida, focused attention on voting technologies and election administration in the United States. During that time, researchers have made great strides working to understand what works, and what does not, in how elections are run in the United States and across the world. It is easy to forget that in the aftermath of the 2000 election, the book on election administration that was most current was a text written by Joseph Harris in 1934. Political scientists had clearly studied issues such as the effects of voter registration on turnout and residual votes (ballot roll-off in down ballot races), but the idea of studying elections for the purpose of improving the mechanics of our democracy and improving administration and implementation was not something on which many people were focused.

Over the past decade, researchers have generated many reports, articles, and books on election administration and voting technology. However, although much of this research has been quite valuable, it has not been accompanied by the development of a set of tools that can easily be applied to evaluate the performance of or effectiveness of election administration, either in a particular area (e.g., voter registration) or across the entirety of an election. This book has discussed a variety of evaluative tools, applied them to different electoral contexts, and discussed different examples of how these tools can be used to effectively gauge whether elections are well run in a particular place.

In today's business world, measurement of the quality of the consumer's experience is essential. Businesses, especially those in areas like retail, spend untold sums of money researching what their consumers want, how they enjoy their shopping experiences, and what new products and services they may want. Think about all the times you have made an online purchase of a product or service and, at the end, you get a request to complete a short survey. In today's busy marketplace, where individuals have many potential outlets for their purchases, corporations are always studying what they can do to make each transaction better.

Although such consumer-oriented research is ubiquitous in the private sector, such tools have not been largely adopted by government entities. This is particularly true for election administrators, who rarely try to obtain systematic feedback from those who are the primary consumers of their services: voters. The other key stakeholders of the services of election officials include candidates running for office, political parties, and voter advocacy groups. Because these groups are generally well-organized, formal entities, they often have structured methods of providing feedback to election officials about the quality of the election-related services they received and the performance of their administrative practices.

Election administration is a highly complex process that involves multiple actors all working to achieve the goal of running an effective election. Accomplishing this goal requires election officials coordinating the efforts of contractors (from ballot printers to voting machine companies), third parties (like the U.S. Postal Service, which transports absentee ballots, and the entities that agree to house polling places), and the poll workers who actually implement the election at the polls. Managing this vast enterprise of actors within the organization and across a network of outside actors requires election officials to evaluate their election activities so that they can improve the implementation of the process over time (O’Toole and Meier 1999).

One important way to gather the performance data needed to improve election management is through an election ecosystem audit (EEA). These audits are evaluations of an election from start to finish that especially emphasize the administrative aspects of the election process. This type of evaluation uses both existing data that are collected as a matter of course in the election process and new data that are specifically generated for this purpose. Postelection voting machine performance audits – where the original count of the election is compared with a postelection hand count of ballots – is an example of this latter type.

For a majority of voters in a majority of states, the voting experience involves going to a polling place – either in early voting or on Election Day – and casting a ballot. In that visit to the polling place, it is with the poll workers whom the voter interacts. As several authors have noted (e.g., Hall et al. 2007, 2008, 2009), poll workers are what are typically referred to as street-level bureaucrats. The term bureaucrat is not meant here as a pejorative; rather, it reflects that fact that, in a polling place, the poll worker makes an array of decisions that determine the experience that each and every voter has. Perhaps the most important decision a poll workers makes is whether a person is an eligible voter and whether she will get a regular ballot, a provisional ballot, or no ballot at all. Poll workers also determine whether to ask a voter to show identification – a decision that may or may not follow the state's law (Atkeson et al. 2010e). The ability of a poll worker to follow the directions and complete a provisional ballot form correctly may determine whether a voter's ballot gets counted. This is just part of a larger list of ways in which poll workers make decisions that directly affect the experience that voters have at the polling place and the general functioning and operations of the precinct.

So far, we have focused on the use of quantitative data to study election system performance: voter and poll worker surveys, studies of residual votes, and how such data can be used for performance-based management. In this chapter, we turn to an entirely different source of data, but one that is critical for determining electoral performance. Here we focus on direct personal examination of an election by trained observers.

Election observation is a type of qualitative data analysis; collection of these data normally focuses on a relatively small number of precincts that are studied closely and intensely. It is qualitative because it relies primarily on the eyes and ears of observers and on their perceptions, intuition, and experience. It is qualitative because what observers focus on in one precinct might differ dramatically from what they might focus on in another precinct. Although all of this can make qualitative data difficult to analyze in a systematic manner, it does not mean that qualitative analysis of election performance by observers is inferior to other types of data analysis. This also does not mean that no systematic data are collected during the process. Indeed, we advise that the collection of qualitative, observational data includes some more systematic components.