Selection Method,Partisanship,and the Administration of Elections

Abstract

The methods used to select public officials affect the preferences they bring to office, the incentives they face while in office, and, ultimately, the policy goals they pursue. We argue that the preferences and actions of local election officials (LEOs) differ depending on whether they are elected or appointed. We test these expectations with a data set that includes the survey responses of 1,200 Wisconsin LEOs, structured interviews, census data, and returns from the 2008 presidential election. Drawing upon a natural experiment in how officials are selected, we find that, compared to appointed officials, elected officials express greater support for voter access and expressless concern about ballot security and administrative costs. For appointed officials, we find that voter turnout in a municipality is lower when the LEO’s self-reported partisanship differs from the partisanship of the electorate but only in cases where the official is a Republican.

Keywords

election administration voter turnout method of selection accountability

One of the most important procedural questions in a democratic republic is how public officials are chosen. Different methods of selection may attract varying types of officials and introduce different behavioral incentives in office. A fundamental dilemma in any political system is finding appropriate structures of control and accountability over administrative functions. Appointed officials may be more insulated from the public and, thus, motivated by their personal policy preferences or the preferences of the officials who appointed them, while elected officials should be more responsive to the demands of the electorate. Appointment versus election might also condition how much the partisan preferences of the official affect their actions. In short, the method of selection has the potential to distort administrative behavior and policy outcomes in systematic ways.

In this article we take advantage of unique data from Wisconsin to show that the method of selection affects the policy preferences of local election officials (LEOs) and voter turnout in their jurisdictions. We theorize that the two methods of selection—appointment and election—influence the major dimension of policy discretion in the field of elections, namely, voter access versus system security. We further hypothesize that appointment provides LEOs discretion to increase turnout if they share the partisan preferences of their constituents. There are three key findings. First, although partisanship does not influence LEO attitudes toward voter access, elected officials are more supportive than appointed LEOs of policies such as absentee voting and election-day registration (EDR) that promote access over security and are popular with voters. Second, elected LEOs produce slightly higher levels of voter turnout, suggesting that attitudinal differences caused by selection translate into behavior in office. Third, and more striking, correspondence between the LEO’s partisanship and that of her constituents has a substantial effect on voter turnout for appointed LEOs but only when they are Republican. We view these results as generally consistent with the notion that elected officials are more inclined than appointed officials to represent and pursue the preferences of the electorate, and we speculate about why partisanship only appears to affect Republican officials.

We begin with a discussion of how election officials in the United States influence election administration and point to evidence that some may serve partisan interests. Next we consider the importance of different selection methods and develop our theoretical argument that elected officials have stronger incentives than appointed officials to respond to public preferences by increasing voter access and turnout. Taking advantage of data on LEOs from Wisconsin, we then show that method of selection for LEOs affects their attitudes toward policies and levels of voter turnout.

Methods of Selection and Partisanship in Election Administration

Researchers have pointed to the fundamental principal-agent problem in election administration (Alvarez & Hall, 2007). Because of the discretion granted to them, election officials are perceived to have substantial influence on elections.¹ Yet a recent analysis by Tokaji (2009) concludes that “we don’t know what institutional arrangements will produce the best election administration [due] to the shortage of research on the subject” (p. 10). In particular, given the many calls by reformers to change how LEOs are selected, it is surprising that the impact of various methods has not been studied extensively (see Kimball & Kropf, 2006). We know little about how election versus appointment affects the manner in which elections are administered. The public clearly prefers that election officials be elected rather than appointed (Alvarez, Hall, & Llewellyn, 2008), but about half of the nation’s many LEOs are appointed (Fischer & Coleman, 2008). Is this disjuncture consequential?

In this section we outline a theory of how the method for selecting election officials influences those officials’ policy attitudes and, ultimately, voter turnout. We suggest that elected election officials will care more about perceived voter wishes for convenience in the voting process than will their appointed counterparts and that this difference will translate into behavior in office that affects voter turnout. We also hypothesize that the method by which we select election officials moderates the impact of partisanship on election administration. Before we outline this theory, we point to evidence on the relevance of methods of selection in other policy areas, the influence of LEOs on electoral outcomes, and the role of partisanship in the exercise of that influence.

Evidence from a variety of contexts already shows that the method for selecting public officials is consequential, with most analyses showing that officials who are directly elected by the public are more responsive to the perceived wishes of citizens. Research reveals that elected judges, prosecutors, and regulators are more likely than their appointed counterparts to be consumer oriented and punitive toward criminals (Besley & Coate, 2003; Gordon & Huber, 2007; Huber & Gordon, 2004; Kwoka, 2002; Primeaux & Mann, 1985). For example, Brace and Boyea (2008) find that the influence of public support for capital punishment on judicial willingness to uphold death sentences is contingent on whether a judge is elected or not. In states where judges were not elected, public opinion had no impact on judicial decisions.

Extending the empirical analysis of selection method to election administration raises two additional issues. First is the prominence of partisanship. Unlike many other areas of public administration, elections are inherently partisan affairs. Election officials often win their positions as a result of partisan connections, and they are responsible for conducting elections in which partisan candidates are seeking office. Moreover, the public constituencies they serve display partisan preferences in their voting patterns. The mere possibility that election officials administer elections in a way that serves their partisan interests feeds the general perception that relatively few election administrators are completely independent from political pressures (Tokaji, 2009). The solution might be to recruit nonpartisan election officials and to insulate them from political pressures altogether, a position strongly favored by the public (Alvarez et al., 2008) and endorsed by the Carter-Baker Commission’s (2005) landmark report. But this begs the question, does officially making election administration nonpartisan actually eliminate partisan action in office? Also, does the choice of election or appointment influence how much partisanship shapes election administration?

The second issue is the highly decentralized nature of the system. The American electoral apparatus is administered at the county and municipal levels by approximately 10,000 LEOs.² The local nature of election administration creates opportunities for variation in implementation (Ewald, 2009). This variation can come from a “street-level” choice taken where the law is silent or by deliberate or accidental disregard of the rules during the implementation process (Kimball & Kropf, 2006; Lipsky, 1980). A small amount of mischief, incompetence, or selective effort by a LEO could alter the likelihood that individuals turn out to vote, voters’ perception that that their vote will be counted, and ultimately, the electoral fates of candidates.

Third, like all administrators, LEOs have discretion in implementing election laws. Indeed, despite massive efforts at the state and federal level to provide more regulation of elections, lawmakers have yet to—and likely cannot—eliminate all of the administrators’ discretion. The passage of the Help America Vote Act (HAVA) in 2002 was intended in part to remedy these concerns by establishing common standards for election administration across the United States. Although HAVA resulted in better voting technology and improved voter registration processes (Stewart, 2011), it did little to dampen variation in administrative behavior at the local level. Election officials retain authority over a wide variety of policy choices, including ballot design, the use of voting technology, the treatment of provisional ballots, standards for counting absentee ballots, public education about registration, and the implementation of voter ID rules (Kimball & Kropf, 2006). Indeed, surveys of LEOs reveal that they have leeway in a number of other areas: hiring poll workers, overseeing recounts, authorizing budgets, policing misconduct at the polls, purchasing voting equipment, and maintaining voter registration databases (Fischer & Coleman, 2008). They enjoy substantial autonomy on the number, hours, and location of polling places, and the supply of workers and ballots at these locations. They often have latitude in how much effort to expend on particular goals. Officials may proactively engage the media and specific community groups such as students or senior citizens to educate voters about the requirements to vote. It is this discretion that makes the method of selection a key contingency.

Observing the consequences of discretion is difficult. The myriad of ways in which election officials can make it harder or easier to vote creates a “black box” problem for researchers in that the actual behavior that leads to electoral outcomes may be difficult to observe. Moreover, in contrast with nefarious practices used in the South before the Civil Rights era (Key, 1949), officials need not exercise this decision-making authority transparently, maliciously, or even consciously for their decisions to affect outcomes. Administrators view many policies and procedures as burdensome (Burden, Canon, Mayer, & Moynihan, 2012); they have latitude to decide which to pursue wholeheartedly and which to give less than full support. These attitudes and actions of LEOs may well depend on whom they answer to and whether their own partisan preferences match those of the constituencies they serve.

Expectations About the Method of Selection

Our central claim is that the preferences and actions of LEOs will vary depending on their method of selection. This claim is consistent with Alesina and Tabellini’s (2007) observation that “different accountability mechanisms induce two behavioral differences between a bureaucrat and a politician” (p. 170). They contend that “[appointed] Bureaucrats want to signal their competence for career concerns, [elected] politicians for reelection purposes” (p. 177). That is, because elected and appointed officials have different methods of being held accountable, they face different returns on policy performance and will pursue different goals.

Specifically, the mayor or board members who make appointments are likely to prioritize different aspects of election administration than does the general public. Because they oversee all aspects of municipal budgets, these authorities are relatively more likely to focus on minimizing the costs of running elections. And because their public reputations depend on problem-free administration of municipal services, they are concerned about violations of security. By controlling costs and limiting security violations, the appointers demonstrate more risk aversion in election administration than would the typical voter. In subtle and perhaps more overt ways, they may induce these preferences in election officials they appoint.³

Appointed officials may be motivated by particular career concerns that lead them to value the perceptions of their professional peers and employers (Alesina & Tabellini, 2007). In the case of election administration, appointed LEOs are not entirely independent. They remain accountable to elected officials and, indirectly, to the voters. The general theory we present stipulates that the addition of an intermediary weakens voters’ ability to hold officials accountable. This theory implies that if election officials have policy goals that differ from those of the electorate, then those who are appointed may be better able than those who are elected to pursue these divergent goals without fear of losing their jobs. That is, the appointed official does not entirely “inherit” the preferences of voters through the municipal appointing authority because of slack in the principal-agent relationships. It also implies that elected election officials might have policy goals that better match the preferences of the electorate than appointed election officials because they have been selected precisely for their views on election administration. In other words, whether their policy goals and attitudes are hard-wired or induced through institutional incentives, we expect a divergence in preferences between elected and appointed election officials.

To formulate more precise empirical hypotheses, we further assume that voters are most concerned about making the voting process convenient (Katosh & Traugott, 1982). Voters want short lines and convenient practices such as EDR and early voting. Studies of Oregon’s convenient vote-by-mail system, for example, showed strong initial support (76.5%) and even stronger support 5 years later (80.9%; Southwell, 2004; Southwell & Burchett, 1997). When early voting was put to a vote in Maryland in 2008, the public endorsed it 72% to 28%, and voters in Maine recently repealed a state law, by a 20-point margin, that would have ended 38 years of EDR. When a recent survey in New Mexico asked respondents whether they preferred ensuring access for eligible voters or preventing fraud, a majority chose the former (Alvarez, Atkeson, Hall, & Sinclair, 2011). This is not to say that voters lack concern for the costs of elections or the possibility of election fraud. Indeed, voters in Mississippi recently endorsed a provision to implement voter ID. Rather, we posit that relative to municipal authorities, voters give higher priority to access and convenience and less to finances and security.

A somewhat weaker version of the theory, which produces the same empirical expectations, is that voters want both convenience and security, but they only observe the former (length of lines, ease of registration, etc.) and not the latter. This leads them to reward LEOs who provide convenience. Thus, LEOs who are elected will be more apt to administer elections in a way that favors great voter turnout. If in doubt, they have an incentive to enable rather than deny access for a potential voter. Because they need to satisfy the electorate to keep their jobs, elected LEOs will be less concerned than appointed LEOs about administrative costs and security. Appointed LEOs will have more discretion when it comes to affecting turnout.

In addition to their method of selection, there is evidence that the partisanship of election officials influences their actions. Kimball and Baybeck (2010) show that LEOs from opposing parties take divergent positions on practices that provide greater voter access and combat fraud.⁴ Stuart (2004) finds that Republican election registrars in Florida were more likely than their Democratic counterparts to purge voter rolls of the names of suspected felons provided by the state. The partisanship of LEOs also shaped how the HAVA’s provisional balloting rules were implemented, with officials often applying rules in ways thought to help their parties at the polls (Kimball, Kropf, & Battles, 2006).

If LEO partisanship matters, it should depend on whether the official’s preferences correspond or conflict with those of their constituents. One way to identify a divergence in the preferences of election officials and their electorates is by comparing the partisanship of officials with that of voters in their municipalities. Some evidence has already emerged showing the match between voters and LEOs to be consequential. Using this approach, one provocative analysis found that LEOs increased turnout among voters who share their partisan affiliations (Bassi, Morton, and Troustine, 2009). We thus hypothesize Democratic (Republican) officials in Democratic (Republican) municipalities should be more inclined to administer elections in ways that enhance voter turnout than Democratic (Republican) officials in Republican (Democratic) municipalities. This is in addition to the direct effects of partisanship in which Democrats are expected to put greater emphasis on voter convenience than Republicans.

There is one more wrinkle to the argument that brings both selection method and partisanship into a single explanatory framework. While there is some evidence to suggest that partisan preferences matter to election administration, such research has not considered whether partisan influence might be constrained or encouraged by method of selection. We argue the electorate is less able to hold appointed election officials accountable for administering elections in ways with which they disagree. Appointed LEOs whose partisan preferences differ from those of the electorate should therefore be better able to administer elections in ways that advantage their party. We hypothesize that a partisan mismatch will have more impact on voter turnout in municipalities with appointed election officials than in those with elected LEOs. Not only are appointed officials less directly accountable to voters, thus allowing their partisan preferences to shape behavior in office, but the municipal authorities who appointed them are likely to share their preferences to some degree, providing another motivation for actions to follow from personal views.

There are reasons to believe that the effects of selection method will be limited. The norms of professional behavior, limits set by federal and state law, and a paucity of resources will prevent LEOs from using their autonomy—intentionally or not—to produce large differences in turnout. Given the multitude of factors beyond the clerk’s control that shape turnout, their impact should be limited. In addition, election officials might not be able to predict the consequences of specific practices on voter turnout. For example, many “convenience voting” options do not appear to increase turnout (Berinsky, 2005). This will weaken the connection between clerk attitudes and behavioral outcomes. Further, as we explain below, municipal LEOs in our Wisconsin application are officially nonpartisan. Although it seems clear that many have partisan attachments—which they readily admit in our surveys—the lack of public partisanship should dampen effects. In short, the constraints imposed by institutions, resources, and professional expectations suggest that all LEOs should administer elections in the similar ways, regardless of their method of selection. Thus, there are strong expectations in support of a null result, which increases the significance of our findings.

The Setting: Election Administration in Wisconsin

Wisconsin is a medium-sized “battleground” state with approximately 4 million eligible voters. There are over 1,830 municipal clerks serving communities whose populations range from 37 to nearly 600,000. Wisconsin law mandates that each municipality have a nonpartisan clerk who is responsible for conducting elections and other administrative duties.⁵ This decentralized structure offers advantages in testing our hypotheses. The large sample, range of backgrounds, and mix of selection methods provides us with an unusual opportunity to explore the impact of selection method on election officials’ policy preferences and the turnout they help produce. For all of the differences in scale and context, most of the electoral responsibilities of these LEOs are similar. The decentralized system also means that approximately one in every five LEOs in the United States works in Wisconsin, providing us with a large population to study. By studying a single state, we have the advantage of being able to hold state-level variables such as election laws and political culture constant (Hanmer, 2009; Keele & Minozzi, forthcoming; Nicholson-Crotty & Meier, 2002).⁶

As in many other states, LEOs in Wisconsin have significant discretion that can affect electoral outcomes. Although constrained by state and federal laws, they are also accountable to their own jurisdictions, either the local electorates who voted them in or to the boards or officials who appointed them. Clerks’ election duties include maintaining voter records and registration lists, recruiting and training poll workers and election inspectors, choosing how many ballots to print, judging provisional ballots, issuing absentee ballots, and establishing and equipping poll locations, including acquiring voting equipment.⁷

Particularly helpful in our application is that clerks vary in how they are selected. About 60% are elected and 40% are appointed. Moreover, the method of selection is reasonably exogenous to the outcome variables we consider. The “treatment” variable of selection method was largely set by the form of governments established by the state constitution in 1848 and to a much lesser degree by local incorporation decisions in the 19th century and early 20th century. To the degree that we can identify variables that distinguish between municipalities with elected and appointed LEOs, our multivariate models control for them. Before moving to those models, further discussion of the history behind the method of selection is necessary.

Clerks work in one of three types of Wisconsin municipalities: city, village, or town. Elected clerks are chosen in nonpartisan elections in the spring of odd-numbered years and typically have terms of 2 years. By constitutional design, cities and villages are incorporated and as such are free to appoint or elect their clerks without much interference; most opted for appointment early in their histories. In contrast, the default procedure for towns is to elect their clerks.⁸ As a result, 84% of clerks in villages and cities are appointed, but 87% of clerks in towns are appointed, as shown in Table 1. As we explain below, this somewhat arbitrary linkage between municipality types and selection methods provides a situation that closely approximates a natural experiment.

Table 1.

Number of Clerks by Municipal Type and Local Election Official (LEO) Selection Method.

	Town	Village	City	Total
Appointed, n (%)	147 (13.0)	293 (81.6)	154 (88.0)	594 (35.7)
Elected, n (%)	981 (87.0)	66 (18.4)	21 (12.0)	1,068 (64.3)
Total n	1,128	359	175	1,662

Whether a municipality is designated as a town or village is largely a historical matter of constitutional design. Selection methods have varied little enough over the ensuing 160 years that they can be treated as nearly randomly assigned to municipalities. There has been some switching between municipal types over time, but there is no evidence that the desire for a different method of LEO selection was the motivation. Instead, some cities and villages incorporated out of towns as populations grew and their residents wanted more municipal services (Paddock, 1997).⁹ This means that whether a clerk is appointed or elected is nearly exogenous, largely an accident of history, and in any event, the practice of switching municipal form largely ended during the Progressive era, fixing the form of government and selection method a century before our analysis. That 96% of municipality types have been in place since 1960 leaves us reasonably confident that cross-sectional analysis provides sufficient evidence of the causal effects of selection method (Paddock, 1997).

Because only about 10% of Wisconsin municipalities are cities, we consider whether towns and villages differ in ways that might produce spurious relationships between partisanship, method of selection, clerk attitudes, and turnout. One potentially important difference between towns and villages for our analysis of turnout is population density (Preuss, 1981). In Wisconsin, towns are much less densely populated than villages because they generally cover a much larger area of land. Density may also be correlated with economic, cultural, or other factors that influence attitudes about voting. We therefore control for this potential confound in our statistical models. Alternatively, one could employ a dummy variable for towns or villages as an instrument for the method of selection. In models not reported here, we find similar results using this statistical approach.

Tables 2 and 3 show that neither clerk selection method nor municipal form by itself has a strong effect on political outcomes. In Table 2, we report the number of clerks who self-identify as Democratic, Independent, or Republican in each municipality type. While there are some differences between towns, cities, and villages, they are not large. Most of the time, partisanship differs by only a few percentage points across municipal type. The one exception is that city clerks are less likely to identify as Republican (and somewhat more likely to identify as Democratic) than town or village clerks.

Table 2.

Clerk Self-Reported Partisanship by Municipal Type.

	Town	City	Village	Total
Democrat, n (%)	232 (31.6)	43 (35.2)	77 (30.8)	352 (31.8)
Independent, n (%)	229 (31.2)	46 (37.7)	86 (34.4)	371 (33.5)
Republican, n (%)	274 (37.3)	33 (27.0)	87 (34.8)	394 (35.6)
Total n	735	122	250	1,117

Table 3.

2008 Turnout by Municipal Type and Local Election Official (LEO) Selection Method.

	Town (%)	Village (%)	City (%)	Total (%)
Appointed	68.5	66.6	62.3	68.3
Elected	73.0	67.6	62.8	67.7
Total	69.1	67.4	62.7	68.1

Table 3 shows the mean turnout, a key dependent variable in our subsequent analysis, by municipal type and selection method. Overall, turnout was 68.1% in all municipalities, with a relatively minor difference between those with appointed clerks (68.3%) and elected clerks (67.7%). Cities had a lower turnout than either villages or towns. We note that our substantive results reported below do not change if we exclude cities from the analysis.

None of this should be read as indicating that there are no differences between clerks in the various municipal types. Rather, the data suggest that the differences are small enough that we are able to identify the effect of selection method independent of the effects of municipal type or partisanship on turnout.

A final advantage of the Wisconsin system is that it allows us to test the impact of employing nonpartisan elected officials, a system strongly preferred by the public (Alvarez et al., 2008). This preference reflects, no doubt, a desire for a strong, direct form of accountability, while at the same time avoiding the negative aspects of partisanship. Yet the uncertainties of this selection model lead Alvarez et al. to conclude that “additional research is needed to determine if the public’s choice for elected, non-partisan election boards is also the electoral governance structure which is best able to prevent electoral fraud and to instill confidence in voters that the process is fair” (p. 341) As we explain below, our survey also asked clerks for their self-identifications on a standard scale of political partisanship. This allows us to contrast election and appointment directly and analyze whether the official absence of party labels truly translates to nonpartisan administration.

Data Sources

To investigate the impact of method of appointment and partisanship on election administration, we created a data set on clerks and their municipalities from five sources. First, we conducted an original survey of all clerks serving 1,850 Wisconsin municipalities in the spring of 2009. Because the survey had the endorsement of state election officials as well as financial incentives to encourage participation, it yielded a 71% response rate with no obvious selection effects that would render the sample unrepresentative. The survey included a battery of questions about clerks’ backgrounds, training, career ambitions, methods of selection, and partisanship.¹⁰ The survey also asked clerks’ views about a number of election administration processes such as EDR and absentee voting.

A second source of information is qualitative data from clerk responses to open-ended questions in the survey and from 100 personal interviews with selected clerks. For the in-person interviews, we developed a semistructured interview protocol using the patterns that emerged from the survey. We selected clerks for these interviews by including the state’s 15 largest municipalities and then randomly selected the remaining 85 using a process based on the distribution of the state’s voting-age population. These interviews informed our quantitative analysis; we highlight significant responses below to provide some illustrative examples of the mechanisms we are investigating.

Third, we utilize data from the 2008 Election Administration and Voting Survey submitted by state election officials to the federal Election Assistance Commission. The survey is a large-scale data collection effort authorized by HAVA and has been conducted after each federal election since 2004. A key feature of the survey is that it collects information about election-related processes and outcomes at the local level, including information on voter participation. Combining this with Census data on the size of the potential voter population provides a measure of turnout, which we treat as an output that may be the product of clerk and municipality characteristics.

Fourth, we use the 2008 presidential election returns in each municipality to gauge whether the partisan preferences of the clerks are in alignment with those of the voters they serve. Finally, we use demographic information such as income, population density, education, and racial composition at the municipal level from the 2000 Census as control variables.

We combined all of the data at the municipal level, giving us a database of clerk characteristics, clerk attitudes, local election practices, voting data, and census data. Our final merged data set consisted of 1,388 observations. Missing data reduce the effective sample size to 1,225 for most of our statistical models. These data give us a unique vantage point for examining how clerk characteristics and attitudes can affect election outcomes.

Qualitative Data on Clerk Attitudes

To provide a flavor of the qualitative data from clerk surveys and interviews,¹¹ we offer responses from clerks on two common practices permitted under Wisconsin law: EDR and absentee voting. EDR provides access by allowing voters to register and vote at the polling place. Absentee voting also provides convenience by allowing voters to cast ballots in advance of election day and to do so by mail or in person.¹² In providing greater access, however, there are concerns that EDR and absentee voting increase the costs of running elections and provide more opportunities for security breaches and other forms of fraud. Thus, EDR and absentee voting provide useful windows into the tension between access, which voters want, and efficiency and security, which municipal officials desire.

The interviews and surveys show that clerks often have strongly held views about election processes, voter responsibilities, and the appropriate balance between voter convenience and administrative costs.¹³ On the whole, and consistent with our expectations, we found that elected clerks were more supportive of efforts to expand access and make voting more convenient, while appointed clerks were more concerned about the administrative burdens imposed by EDR and absentee voting.

EDR creates administrative headaches for the election officials who must make available separate processes to register and vote on election day. This comment from an appointed clerk was typical:

Election-day registration is a nightmare. It increases pressure at the polls, where we have a hard time finding adequate numbers of staff to work. The workers tend to be elderly and get flustered under the stress of a large election, or have a hard time simply reading and following the forms. There are so many mistakes made that would not happen if we closed registration three-plus weeks before the election.

To a greater extent than elected clerks, appointed clerks explicitly criticized voters for wanting additional convenience in voting, as the following quotes illustrate.

Election-day registration is being abused by people who have begun to presume that it is their right. I think there should be a provision to allow for only certain limited EDR. There is no reason that the vast majority of the voters cannot register at least 30 days prior to the election. I believe that voting is both a privilege and a right and more people need to act responsibly and try to be better prepared. There is enough information available that people can easily find out where to register and what proof of residency they need to bring with them.

Absentee voting has become the lazy person’s way of life. . . . The “old” rule of needing a reason to vote absentee should be put back in place so that only those who need to vote by absentee would vote by that means.

An elected clerk, on the other hand, supported EDR as a convenience to voters, despite the crunch it created on election day:

I think it’s a good thing for the voters because they don’t have to plan ahead. And it probably does increase the number of people voting, coming out to vote. On the administrative side, it’s difficult to manage hundreds and hundreds of registrations very close to an election day.

Elected clerks also seemed more supportive of the need for absentee voting, despite the increase work load and mailing costs: “It definitely takes up more of my work time. I like people being able to vote absentee. I get very disappointed when they do not return their ballot.”

Appointed clerks also seemed to be more concerned about the implications of EDR for ballot security and voter fraud than did elected clerks:

Election-day registration creates such a large post-election burden. If Wisconsin wants to make changes to elections in Wisconsin this should be eliminated. By doing so I think it could reduce voter fraud and potential errors by poll workers. The day before the election should be the last day to register in the clerk’s office.

If nothing else, these responses suggest that clerks are not neutral in how they assess their role. They have clear ideas about appropriate policy, even if their views are inconsistent with existing laws. Although clerks’ attitudes are the result of many factors, including partisan and ideological preferences, the differences in views between elected and appointed clerks suggests that they are at least partly connected to the principals who employ them. It seems reasonable to expect that these views will influence the execution of their jobs. However, qualitative data cannot establish the existence of systematic differences between elected and appointed clerks’ attitudes about voting practices and voter turnout. For that we turn to quantitative analysis.

Measures

We present two models, estimating the effects of selection method and partisanship first on clerk attitudes about policies and then on voter turnout. We begin with a working assumption that clerk attitudes toward election policies will be a function of clerk and municipality characteristics and method of selection. Above we hypothesized that the enhanced accountability to voters provided by direct election should lead elected clerks to be more supportive than appointed clerks of initiatives that facilitate voting, such as EDR and absentee voting, and that these preferences should translate to increased voter turnout. To measure clerk attitudes about EDR we asked clerks to rate on a 7-point scale their agreement with the following statements: “Election-day registration makes it more difficult to protect the security of the voting process,” “Election-day registration should be considered a voter’s right,” “Election-day registration increases the administrative burden on election officials like me,” and “The benefits of election-day registration outweigh the costs.” For mail-in absentee voting, we asked clerks to rate their agreement with the following statements: “The authenticity of a mail absentee ballot is difficult to verify,” “Mail absentee voting should be considered a voter’s right,” and “The benefits of mail absentee voting outweigh the costs.”

These seven measures capture many of the elements that might divide clerks serving principals with differing priorities. The upper half of Table 4 shows descriptive statistics for these variables.¹⁴ All of the attitudinal measures have been rescaled so that higher values indicate positions that favor voter access convenience over security and cost. As the means indicate, clerks as a group have varying views on these matters. The item with the most support is that EDR is a voter’s right, whereas the item asking about the administrative burden imposed by EDR leans most strongly in the other direction.¹⁵ The two items that ask directly about the benefits versus the costs place LEOs about one point above the scale’s midpoint.

Table 4.

Descriptive Statistics.

Variable	Mean	Minimum	Maximum
Dependent variables
EDR security is a problem	3.74	1	7
EDR is a right	5.19	1	7
EDR is worth the costs	4.87	1	7
EDR increases administrative burden	2.38	1	7
Absentee votes are hard to verify	3.46	1	7
Absentee voting is a right	5.04	1	7
Absentee voting is worth the costs	4.31	1	7
Turnout %	69.1	22.2	100
Independent variables
Elected	0.61	0	1
Democrat	0.32	0	1
Republican	0.35	0	1
Constituency partisanship (Obama vote %)	53.5	15.4	91.9
Relier	0.74	0	1
Seeks higher office	0.22	0	1
Years in office	12.4	1	57
Treasurer	0.32	0	1
Median family income (in thousands of U.S. dollars)	49.258	22.188	181.443
% with college degree	15.3	0	81.8
Population (in thousands)	2.87	0.04	596.97
Population density (per square mile)	368.61	0.7	8,599.5
% black	0.41	0	37.34

Note: EDR = election day registration; descriptive statistics are based on full sample (N = 1,851). Sample sizes vary slightly for subsamples used in each analysis.

As we expected, there are differences in the attitudes of elected versus appointed clerks. Elected clerks, to a far greater degree than appointed clerks, were supportive of both EDR and absentee voting and less concerned with administrative costs. The differences between the two groups were statistically significant at .01 or above for all of the attitudinal items (based on simple t tests, not shown). These results provide initial support for our claim that elected officials are more responsive to voter preferences than are appointed officials.

Because previous work has suggested that partisanship can affect LEO behavior, we asked clerks to specify their partisanship using the American National Election Study (ANES) 7-point scale running from strong Democrat to strong Republican. For some of the statistical analyses we also created party dummy variables, collapsing the strong partisans, weak partisans, and leaners into a single group. The models include a dummy for each major party, with pure independents as the reference category. Despite the official nonpartisan nature of the offices and professional norms against partisanship, LEOs were quite willing to admit their partisan attachments.¹⁶ Fewer than a third of clerks put themselves in the “pure independent” category. Another third called themselves “strong” or “weak” partisans, with the remainder self-identifying as leaners. As Table 2 shows, our categorical coding produces a breakdown that is roughly equally divided among self-identified Democrats, Republicans, and independents.¹⁷

We measured the partisan preferences of constituents using the 2008 presidential vote. Interacting this measure with the clerk’s partisanship provides a test of whether LEO discretion is exercised differently when the preferences of the principal differ from those of the agent. Using this measure, we conclude that 29% of clerks are “out of step” with their constituents politically.¹⁸

The specification allows us to determine whether the relationship between clerks’ attitudes and local voter turnout is shaped directly by clerk partisanship, by constituency preferences, or a mismatch between the two. Because we use separate interaction terms for Democratic and Republican LEOs, we can also observe whether the correspondence in preferences has a greater impact for one party or the other. Moreover, in our models of voter turnout, we allow this effect to vary with the method of selection to determine whether divergence between the principal and agent is more consequential when the agent is elected or appointed.

Four other clerk control variables also are included. In Wisconsin many small-town municipal clerks rely on county clerks for administrative support, such as for the processing of voter registration forms. We include a dummy variable to account for these clerks, who are known as “reliers” by statewide election officials. We also cluster the standard errors by county in the event that some unmeasured characteristic generates correlations among municipal clerks from the same county. The survey also asked whether or not the clerk plans to seek higher office in the future. This allows us to account for the possibility that officials with political career ambition might be more likely to support initiatives that are popular with voters. We controlled for whether a clerk also serves as treasurer for her municipality, a somewhat common occurrence in Wisconsin. This avoids a confounding association with the method of selection and tests the idea that clerks who also keep the municipal books will be more attentive to the costs of voting. Finally, we include a measure of clerks’ tenure to check whether longer-serving clerks believe and act differently, either as a result of generational differences or because of accumulated experience (Moynihan & Silva, 2008). The data in Table 4 show that clerks are mostly reliers, have about 12 years of experience on average, and rarely plan to pursue higher office (about one in five).

The models also control for demographic characteristics of the clerk’s municipality. We use Census measures of median family income, population density, the percentage of the population with a college degree, the overall population size, and the percentage of the population that is black. We also include Obama’s vote share in the municipality, which is necessary for the interaction term between constituency partisanship and LEO partisanship. It is also an interesting control in its own right because it tests whether more Democratic jurisdictions disproportionately prioritize voter access.¹⁹ There is tremendous variation in these measures, which will help isolate how the method of selection is associated with clerks’ attitudes toward election policy.

Clerk Attitude Results

Table 5 presents the results of ordered logit models for each of the seven attitudinal measures.²⁰ The findings are straightforward. In all seven models the method of selection exerts a strong and statistically significant effect in the expected direction. Elected clerks are more likely to see EDR as a right, worth the costs, and engendering relatively low administrative burdens, whereas appointed clerks are more likely to view EDR as jeopardizing the security of the voting process and increasing administrative burdens. Similarly, elected clerks are more likely to view absentee voting as both a right and worth the costs and are less concerned about the difficulty of verifying the identities of absentee voters. We display these effects in Figure 1, which plots the point estimates and 90% confidence intervals for each of the seven dependent variables. The figure makes plain how difference in method of selection consistently affects the placements of LEOs on the continuum between access and security.

Table 5.

Clerk and Municipal Characteristics Affect Local Election Official (LEO) Attitudes.

	EDR security is a problem	EDR is a right	EDR is worth the costs	EDR increases admin burden	Absentee votes are hard to verify	Absentee voting is a right	Absentee voting is worth the costs
Elected	0.629*** (0.134)	0.637*** (0.137)	0.690*** (0.141)	0.511*** (0.148)	0.355* (0.185)	0.492*** (0.150)	0.504*** (0.157)
Democratic LEO	0.596 (0.573)	0.762 (0.697)	0.951 (0.580)	−0.662 (0.607)	0.653 (0.797)	1.093 (0.691)	0.462 (0.834)
Republican LEO	0.346 (0.894)	0.415 (0.810)	0.913 (0.562)	−0.747 (0.732)	−0.045 (0.669)	−0.090 (0.561)	0.583 (0.629)
Constituency partisanship (Obama vote %)	0.019** (0.010)	0.017** (0.008)	0.022*** (0.008)	0.010 (0.008)	0.016 (0.010)	0.017 (0.011)	0.011 (0.010)
Democratic LEO × Constituency partisanship	−0.011 (0.010)	−0.009 (0.012)	−0.013 (0.011)	0.008 (0.011)	−0.011 (0.014)	−0.017 (0.013)	−0.006 (0.016)
Republican LEO × Constituency partisanship	−0.011 (0.016)	−0.009 (0.014)	−0.018* (0.010)	0.008 (0.014)	0.0001 (0.012)	0.001 (0.011)	−0.011 (0.012)
Relier	0.322** (0.152)	0.430** (0.170)	0.213 (0.160)	0.361** (0.170)	0.028 (0.140)	0.069 (0.161)	0.196 (0.144)
Treasurer	0.326** (0.141)	0.147 (0.149)	−0.011 (0.163)	−0.034 (0.148)	0.217 (0.153)	0.287** (0.137)	0.282** (0.143)
Seeks higher office	−0.016 (0.136)	0.006 (0.133)	−0.034 (0.138)	−0.134 (0.120)	0.171 (0.131)	0.051 (0.120)	−0.204 (0.140)
Years in office	0.006 (0.005)	−0.002 (0.006)	0.001 (0.006)	−0.010* (0.005)	0.009** (0.004)	0.0001 (0.005)	0.011** (0.005)
Median family income (in thousands of dollars)	−0.013** (0.006)	−0.016** (0.007)	−0.011* (0.007)	0.003 (0.007)	−0.010 (0.006)	−0.012* (0.006)	−0.017** (0.007)
% with college degree	−0.003 (0.008)	0.003 (0.008)	0.006 (0.009)	−0.008 (0.009)	0.025*** (0.009)	0.014* (0.007)	0.013* (0.008)
Population (in thousands)	−0.010 (0.008)	−0.032*** (0.011)	−0.026** (0.011)	−0.034* (0.019)	0.002 (0.009)	−0.004 (0.007)	−0.006 (0.011)
Population density (per square mile)	−0.0003*** (0.0001)	−0.0002 (0.0001)	−0.0002* (0.0001)	−0.0002* (0.0001)	−0.0001 (0.0001)	−0.0003*** (0.0001)	−0.0002* (0.0001)
% black	−0.024 (0.04)	−0.002 (0.050)	−0.001 (0.039)	−0.054 (0.035)	0.020 (0.030)	0.017 (0.011)	0.044 (0.041)
N	1,218	1,216	1,217	1,219	1,215	1,215	1,211

Note: All dependent variables have been rescaled so that higher values indicate greater emphasis on access than security. All are 7-point scales. Models are ordered logits. The estimated cutpoints are not reported. Standard errors are clustered by county.

p < .10. **p < .05. ***p < .01 (two tailed).

Figure 1.

Elected local election officials (LEOs) are more supportive of voter access.

It is clear that method of selection affects the kind of people who take office, shapes their preferences by requiring that they be accountable first to the voters who elected them, or both. Elected clerks are more supportive of policies promoting voter access and less concerned about vote security and administrative costs than are appointed clerks.²¹

LEO partisan sympathies, in contrast, show no direct effects on attitudes toward security versus access. Self-identified Democrats and Republicans do not differ when it comes to their positions on the use of EDR and absentee voting as they relate to voter access and system security. This is not just a matter of collinearity with method of selection; the bivariate correlations between clerk partisanship and method of selection are also insignificant. There is some evidence that the preferences of constituents influence LEO attitudes. When it comes to EDR, LEOs in municipalities with strong Democratic tendencies—as measured by the Obama vote share—are more supportive of access and less concerned about security.²² But there is little support for the hypothesis that LEO partisanship and constituency partisanship interact. In only one case is the “mismatch” between the clerk and her constituents statistically significant (Republican LEOs in Democratic municipalities, on attitudes about the costs of EDR), and only weakly at that. It is unsurprising that the congruence between LEO and constituent preferences would have limited influence on LEO attitudes about election practices. This suggests that these attitudes are a fairly straightforward function of the method of selection and clerk partisanship itself and have little to do with how clerks’ partisan views compare to those of the voters they serve.

Some of the control variables are also of interest. Constituency demographics show inconsistent results, with the exception of median income. Municipalities with lower incomes have clerks who are more supportive of voter access over security. This coefficient could be a proxy for allegiances with the Democratic Party that go beyond pure voting habits or it might reflect more general concerns of those with lower incomes about exclusion from the political process. LEOs serving higher educated electorates are more likely to support access when it comes to absentee voting but not EDR. Population itself has no effect, but communities with greater population density are represented by clerks who give greater priority to security rather than access. The racial composition of a municipality has no impact.

Few LEO characteristics are significantly related to attitudes, with the exception of relier status. Relier clerks are more supportive of voter access in three of the seven models, perhaps because they shift the burdens of EDR onto county clerks. In several cases it appears that clerks who also serve as treasurers are actually less concerned about security and costs. Ambition for higher office is unrelated to any of the attitudes. Time in office has little consistent effect on any attitudes. The general absence of effects associated with clerk backgrounds, responsibilities, ambition, and especially municipality type, makes it all the more striking that method of selection exhibits such a strong and consistent influence on attitudes toward EDR and absentee voting.

Turnout Results

Knowing that the attitudes of LEOs are shaped by their method of selection is interesting, but it would carry more weight if we can identify tangible consequences of these differences. Voter turnout is an important indicator because it is frequently viewed as the best overall measure of mass political participation and because it could demonstrate the influence of LEO discretion. It is one thing to learn that the method of selection affects how clerk preferences and even their actions in office. It is a higher hurdle to show that variation in selection method influences levels of voter involvement.

We draw upon the EAC survey to analyze voter turnout in the 2008 presidential election. We calculate turnout as a percentage of the voting-age population, dividing the number of votes by municipal-level voting-age population estimates made by the Demographic Services Center of the Wisconsin Department of Administration (Wisconsin Demographic Services Center, n.d.).²³ We employ OLS (ordinary least squares) models using the same specifications as the attitudinal models above, but with percentage turnout as the continuous dependent variable.

The first column of Table 6 reports results for all of the municipalities in our data set. The coefficient for the method of selection variable indicates that turnout is about 1.5 points higher in municipalities with elected clerks. This is consistent with our finding that elected clerks are more concerned than appointed clerks with enhancing access to the polls. While an effect of a point or two might seem small compared to the impact of other forces such as demographics, campaign competitiveness, and political culture, we find it remarkable that there is enough discretion for the method of selection to affect levels of participation, despite the many more obvious forces influencing voting turnout.

Table 6.

How Local Election Officials (LEOs) and Municipal Characteristics Affect Turnout.

	All	Appointed	Elected
Elected	1.450** (0.672)	—	—
Democratic LEO	1.363 (3.351)	1.973 (4.061)	1.776 (4.911)
Republican LEO	7.781*** (2.696)	13.417*** (3.437)	2.351 (3.674)
Constituency partisanship (Obama Vote %)	0.009 (0.051)	0.020 (0.085)	−0.015 (0.070)
Democratic LEO × Constituency partisanship	−0.033 (0.061)	−0.070 (0.069)	−0.042 (0.093)
Republican LEO × Constituency partisanship	−0.144*** (0.052)	−0.245*** (0.062)	−0.045 (0.072)
Relier	1.660** (0.768)	1.451 (0.939)	2.336** (0.977)
Treasurer	−0.493 (0.631)	−0.799 (0.800)	0.448 (1.078)
Seeks higher office	−0.074 (0.488)	−0.605 (1.399)	0.073 (0.498)
Years in office	0.010 (0.024)	0.003 (0.052)	0.008 (0.027)
Median family income(in thousands of U.S. dollars)	0.278*** (0.064)	0.293*** (0.102)	0.292*** (0.063)
% with college degree	0.384*** (0.063)	0.289*** (0.066)	0.520*** (0.096)
Population (in thousands)	0.0006 (0.00002)	0.006 (0.018)	−0.125 (0.078)
Population density (per square mile)	−0.0007* (0.0004)	−0.0003 (0.00004)	−0.0001 (0.001)
% black	−0.923*** (0.235)	−0.768*** (0.173)	−1.285*** (0.407)
Constant	46.453*** (4.633)	46.553*** (8.142)	46.337*** (3.865)
N	1,221	509	712
R ²	.426	.511	.362

Note: Models are OLS (ordinary least squares) regressions. Standard errors clustered by county.

p < .10. **p < .05. ***p < .01 (two tailed).

Of equal interest is the interaction between LEO and constituency partisanship. The model suggests that the effect of constituency preferences is shaped by the clerk’s partisanship, but only when he or she identifies as a Republican. Among GOP clerks, the more Democratic the constituency, the lower turnout is. This is a striking result, especially because the magnitude of the interaction term is substantial. An increase of 1 standard deviation in the Obama vote (roughly 10 percentage points) results in a decrease in turnout of 1.5 points. It is not immediately clear why the effect is limited to Republican office holders.

It is possible that this effect is due to the actions of citizens rather than the LEO. Evidence that people are more likely to vote when represented by politicians who are like them demographically (e.g., Burns, Schlozman, & Verba, 2001) would lead to a hypothesis that people are more motivated to vote if their election administrator shares their partisanship. We think this mechanism is dubious, however, for two reasons. First, it requires constituents to know the partisanship of their LEO, which is unlikely, especially in a state where clerks are officially nonpartisan. Clerks are far more likely to know the partisan tendencies of their constituents than the other way around. Second, below we will show that this effect only holds for appointed clerks. Our theory explains why the interaction effect would be conditional on the method of selection. There is no plausible mechanism that would explain why voters would condition their participation on both the clerk’s partisanship and the clerk’s selection method.

Before turning to that analysis, we observe that control variables in our turnout model operate largely as expected. Turnout is significantly higher in municipalities where residents have higher incomes and more formal education, two variables known to influence participation at both individual and aggregate levels. Turnout is also higher where the constituency is more racially homogeneous and where the population is less dense.

Our theory posits that the method of selection should shape the degree to which LEOs are faithful to the preferences of the people who hold them accountable. We conjectured that being elected would motivate LEOs to prioritize actions that favor voter access and convenience, whereas appointment by a municipal authority would result in clerks placing greater weight to security and cost considerations, and found that the attitudinal results reported in Table 5 support this hypothesis. Whereas election forces accountability to the electorate generally, appointment allows a clerk to be attentive to their own views. In short, the divergence between clerk and constituent partisanship will be more consequential when the clerk is not directly elected by those same constituents.

We further test for this conditional relationship in the final two columns of Table 6 by conducting separate analyses for appointed and elected LEOs. Here we find that the interaction between clerk and constituent preferences only affects turnout when the clerk is appointed, but again only when the clerk is Republican. That is, the significant interaction term in the full model appears to be a result of an even larger effect among appointed clerks offset by an insignificant effect among elected clerks. In contrast, the party mismatch variable is insignificant for municipalities with elected clerks.²⁴ This finding is consistent with the notion that elected clerks were reluctant to appear as if they are administering elections in a manner that is biased against the partisan interests of the electorate. Appointed clerks are at least a step removed from voters. Taken in conjunction with the finding that municipalities with elected clerks have higher turnout generally, this finding strengthens the conclusion that method of selection shapes clerk behavior and that appointed clerks are freer to administer elections in a way that conflicts with the wishes of the electorate.

To demonstrate more clearly what these results imply, Figure 2 displays predicted turnout using the model estimated among appointed clerks, showing how constituency partisanship and clerk partisanship affect turnout. When the clerk is a Democrat, the preferences of the constituency have no effect on turnout. Indeed, this is true regardless of whether the Democratic clerk is appointed or elected. In contrast, turnout is sharply responsive to constituency preferences when the clerk is a Republican appointee. Across the full range of the Obama vote share variable, turnout in municipalities with Republican clerks drops about 18 points. The two lines cross near the statewide mean of 57%. This means that turnout is raised above the average when a Republican clerk represents a Republican constituency but depressed below the average when the clerk represents a Democratic constituency.²⁵

Figure 2.

Turnout is correlated with constituency partisanship only when the local election official (LEO) is a Republican.

It is not immediately clear why the effect holds for Republican LEOs but not for their Democratic counterparts. The asymmetry could be specific to the 2008 election context, a year with high voter turnout and a strong showing for Democrats from Obama to down-ballot races. In this environment, Republican LEOs might have acted under the threat of electoral losses for their party. Viewing this relationship from the other side, one might infer that Democratic and independent clerks are especially facilitative of turnout in municipalities where Obama was running strongly. If this is a general phenomenon, one would expect the reverse pattern in a year when electoral winds favor the Republicans. On the other hand, the asymmetry could be a general difference between the parties. Compared to Democrats, Republican identifiers in the electorate are widely believed to be more loyal to their party and more consistent voters (e.g., Martinez & Gill, 2005). Whether this stereotype is true or not, LEOs might well act on the assumption that it is, and consequently believe that turnout of Democratic voters can be influenced more easily than that of Republicans. These are fruitful questions for future research.

In additional analyses we attempted to verify the robustness of the result by examining the connection between clerk attitudes and turnout—our two dependent variables—to determine whether the attitudes about access, security, and costs expressed in Table 5 might explain levels of voter turnout in Table 6. It is possible that clerks who are more concerned about the cost and security of EDR and absentee voting might take actions to lower voter turnout. We examined this possibility by conducting both exploratory and confirmatory factor analysis on the seven attitudinal items and then adding the resulting factor scores as independent variables in the turnout models. None of the attitudinal factors had a statistically significant effect on turnout, even when the method-of-selection variable was omitted. Nor did their inclusion in the Table 6 models alter coefficient values of other variables in the models.²⁶ This nonfinding is revealing. It suggests that clerks’ efforts to increase or decrease turnout were not driven by their philosophical views on voter access and ballot security, but by methods of selection and partisan mismatch. Even after we control for clerk attitudes, voter turnout is higher when a clerk is elected, when the clerk’s partisanship is congruent with voter partisanship, and when a clerk matches the partisanship of the municipality and is appointed.

Conclusion

Given the centrality of fair elections to democratic legitimacy, it is vital to understand the impact selection method has on election officials’ policy preferences and voter turnout. As the individuals who determine the staffing and functioning of polling places, LEOs are indeed the “administrators of democracy” (Moynihan & Silva, 2008). Taking advantage of a natural experiment and original data on Wisconsin election officials, we have shown that the method of selection matters. Our results indicate that elected officials are more in favor of policies that are thought to promote turnout than appointed officials and that their jurisdictions are associated with higher voter turnout.

Additionally, we find that voter turnout is lower when an official’s partisanship differs from the partisanship of the electorate but that this effect holds only in municipalities that appoint their election officials and when the official is a Republican. Municipalities with appointed officials on average have lower turnout when the LEO is a Republican representing mostly Democratic constituents; turnout in municipalities with elected officials are unaffected by such a mismatch in partisanship. These results are consistent with the notion that elected officials are more responsive to the preferences of the electorate than appointed officials, if we assume that the electorate generally favors policies that reduce the costs of voting. Appointed officials, who serve at the pleasure of local municipal officials, are concerned less with voter rights and turnout and more with administrative burdens, costs, and security.

Although we are confident that our results have shown how selection method affects election administration, identifying the precise mechanisms through which this occurs remains a question for future research. Neither our theory nor our empirical model makes a claim about the degree to which the impact of selection method is attributable to a selection effect (i.e., election officials are attracted to the method of selection that reflects their preferences) versus incentives and socialization effects once an individual takes a position. In addition, we cannot determine whether the turnout effect is caused by acts of omission or acts of commission. An elected clerk who is motivated to facilitate voter access might engage in proactive efforts such as educating voters, increasing the ballot supply, and locating the most convenient polling places. On the other hand, an appointed clerk who is less directly accountable to the public might simply not do these things. What we can say is that it appears that partisanship is driving the result for appointed officials. Including attitudes toward EDR and absentee voting in the turnout models did not alter the fundamental relationship between method of selection and turnout. Moreover, all of this occurred in an administrative environment where all clerks, whether elected or appointed, are officially nonpartisan. One might imagine even larger effects in states where partisanship is more overt.

Our findings inform an ongoing policy debate about the strengths and weaknesses of various institutional designs in the administration of elections, a subject about which little is known. Descriptive statistics indicate that appointed clerks may possess greater skills and expertise, as they tend to be better educated, have greater experience, and are more often full-time professionals. But we provide some evidence that elected officials administer elections in a less biased fashion than appointed officials.

All of this brings into sharper relief the inconsistency between how LEOs are actually selected and what selection method voters would prefer. Unwittingly or not, public desire for choosing LEOs via elections tilts administration in a direction that favors access over security. Insulation from political pressures has its benefits, but it may also make administrators less concerned with the wishes of the public, and overly concerned with administrative burdens. While there are problems in assuming that the public actively monitors LEOs, the findings do suggest that elected LEOs pursue the perceived wishes of the public. In nonpartisan positions, which the public also prefers, election also deters actions that permit LEOs to increase voter participation only when it favors their party.

Footnotes

Acknowledgements

We thank Leticia Bode, Hannah Goble, Matt Holleque, Patrick Moran, Jacob Neiheisel, David Nelson, and Sarah Niebler for research assistance. We also thank Michael Hanmer, David Kimball, Susan Paddock, Nasos Roussias, Julie Ruder, and participants in the American Politics Workshop at the University of Wisconsin–Madison for their helpful comments.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Earlier versions of this article were presented at the 2010 annual meetings of the American Political Science Association and the Midwest Political Science Association. This article is part of a larger project funded by the Wisconsin Government Accountability Board and the Pew Charitable Trusts.

Notes

Author Biographies

Barry C. Burden is a professor of political science at the University of Wisconsin–Madison. His research focuses on election administration, elections and campaigns, public opinion, federal government programs, and legislative representation.

David T. Canon is a professor of political science at the University of Wisconsin–Madison. His research interests include election administration and various aspects of the U.S. Congress, including race and representation, redistricting, political careers, and the committee system.

Stéphane Lavertu is an assistant professor in the John Glenn School of Public Affairs at the Ohio State University. His research interests include bureaucratic politics and education policy.

Kenneth R. Mayer is a professor of political science and faculty affiliate of the La Follette School of Public Affairs at the University of Wisconsin–Madison. His research interests are the presidency, campaign finance, and election administration.

Donald P. Moynihan is a professor of public affairs at the University of Wisconsin–Madison. His research interests include election administration, governmental performance, and homeland security.

References

Alesina

Tabellini

(2007). Bureaucrats or politicians? Part I: A single policy task. American Economic Review, 97, 169-179.

Alvarez

R. M.

Atkeson

L. R.

Hall

T. E.

Sinclair

J. A.

(2011). The balance between preventing fraud and ensuring participation: Attitudes towards voter identification in New Mexico (Voting Technology Project Working Paper 106). Retrieved from http://vote.caltech.edu/content/balance-between-preventing-fraud-and-ensuring-participation-attitudes-towards-voter

Alvarez

R. M.

Hall

T. E.

(2007). Controlling democracy: The principal-agent problems in election administration. Policy Studies Journal, 34, 491-510.

Alvarez

R. M.

Hall

T. E.

Llewellyn

(2008). Who should run elections in the United States? Policy Studies Journal, 36, 325-346.

Bassi

Morton

Trounstine

(2009). Local implementation of state and federal election law. Retrieved from http://faculty.ucmerced.edu/jtrounstine/BMT_Empirical_May09.pdf

Berinsky

A. J.

(2005). The perverse consequences of electoral reform in the United States. American Politics Research, 33, 471-491.

Besley

Coate

(2003). Elected versus appointed regulators: Theory and evidence. Journal of the European Economic Association, 1, 1176-1206

Brace

Boyea

B. D.

(2008). State public opinion, the death penalty and the practice of electing judges. American Journal of Political Science, 52, 360-372.

Burden

B. C.

Canon

D. T.

Mayer

K. R.

Moynihan

D. P.

(2012). The effect of administrative burden on bureaucratic perception of policies: Evidence from election administration. Public Administration Review, 72, 741-751.

10.

Burns

Schlozman

K. L.

Verba

(2001). The private roots of public action: Gender, equality, and political participation. Cambridge, MA: Harvard University Press.

11.

Commission on Federal Election Reform (Carter-Baker Commission). (2005). Building confidence in U.S. elections: Report of the commission of federal election reform. Retrieved from http://www.american.edu/ia/cfer/report/full_report.pdf

12.

Ewald

A. C.

(2009). The way we vote: The local dimension of American suffrage. Nashville, TN: Vanderbilt University Press.

13.

Fischer

E. A.

Coleman

K. J.

(2008). Election reform and local election officials: Results of two national surveys. Washington, DC: CRS Report for Congress.

14.

Gordon

S. C.

Huber

(2007). The effect of electoral competitiveness on incumbent behavior. Quarterly Journal of Political Science, 2, 107-138.

15.

Hajnal

Z. L.

Lewis

P. G.

(2003). Municipal institutions and voter turnout in local elections. Urban Affairs Review, 38, 645-668.

16.

Hanmer

M. J.

(2009). Discount voting: Voter registration reforms and their effects. New York, NY: Cambridge University Press.

17.

Huber

G. A.

Gordon

S. C.

(2004). Accountability and coercion: Is justice blind when it runs for office? American Journal of Political Science, 48, 247-263.

18.

Katosh

J. P.

Traugott

M. W.

(1982). Costs and values in the calculus of voting. American Journal of Political Science, 26, 361-376.

19.

Keele

L. J.

Minozzi

(forthcoming). How much is Minnesota like Wisconsin? States as counterfactuals. Political Analysis.

20.

Key

V. O.

Jr. (1949). Southern politics in state and nation. New York, NY: Alfred A. Knopf.

21.

Kimball

D. C.

Baybeck

(2010). Is there a partisan way to administer elections? Retrieved from http://www.umsl.edu/~kimballd/dk_bb_MPSA10.pdf

22.

Kimball

D. C.

Baybeck

(2012). Size matters in election administration (Prepared for HAVA at 10 Conference at Moritz College of Law, the Ohio State University). Retrieved from http://moritzlaw.osu.edu/electionlaw/hava-at-10/docs/files/KimballBaybeck-HAVA10-Draft.pdf.

23.

Kimball

D. C.

Kropf

(2006). The street-level bureaucrats of elections: Selection methods for local election officials. Review of Policy Research, 23, 1257-1268.

24.

Kimball

D. C.

Kropf

Battles

(2006). Helping America vote? Election administration, partisanship, and provisional voting in the 2004 election. Election Law Journal, 5, 447-461.

25.

Kwoka

J. E.

Jr. (2002). Governance alternatives and pricing in the U.S. electric power industry. Journal of Law, Economics, and Organization, 8, 278-294.

26.

Lipsky

(1980). Street-level bureaucracy: Dilemmas of the individual in public service. New York, NY: Russell Sage.

27.

Martinez

M. D.

Gill

(2005). The effects of turnout on partisan outcomes in U.S. presidential elections 1960-2000. Journal of Politics, 67, 1248-1274.

28.

Mayer

K. R.

(2007). Comparative election administration: Can we learn anything from the Australian Electoral Commission? Retrieved from http://users.polisci.wisc.edu/apw/archives/mayer2.pdf

29.

Moynihan

D. P.

Silva

(2008). The administrators of democracy: A research note on local election officials. Public Administration Review, 68, 816-827.

30.

Nicholson-Crotty

Meier

K. J.

(2002). Size doesn’t matter: In defense of single-state studies. State Politics & Policy Quarterly, 2, 411-422.

31.

Preuss

G. C.

(1981). The effects of density and urban residence on voter turnout. Population and Environment, 4, 246-265.

32.

Primeaux

W. J.

Jr. Mann

P. C.

(1985). Voter power and electricity prices. Public Choice, 47, 519-525.

33.

Southwell

P. L.

(2004). Five years later: A re-assessment of Oregon’s vote by mail electoral process. PS: Politics & Political Science, 98, 89-93.

34.

Southwell

P. L.

Burchett

(1997). Survey of vote-by-mail Senate election in the state of Oregon. PS: Political Science and Politics, 91, 52-57.

35.

Stewart

III (2011). What hath HAVA wrought? Consequences, intended or not, of the post-Bush v. Gore reforms. Retrieved from http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1843583

36.

Stuart

(2004). Databases, felons, and voting: Bias and partisanship of the Florida felons list in the 2000 elections. Political Science Quarterly, 119, 453-475.

37.

Tokaji

D. P.

(2009). The future of election reform: From rules to institutions. Yale Law & Policy Review, 28, 125-154.

38.

Walker

(2012, July 18). Second Waukesha County panel OKs more election money. Milwaukee Journal Sentinel. Retrieved from http://www.jsonline.com/news/waukesha/second-waukesha-county-panel-oks-more-election-money-hm660l3-162940566.html

39.

Wisconsin Demographic Services Center. (n.d.). Population estimates program description. Retrieved from http://www.doa.state.wi.us/docs_view2.asp?docid=4165

40.

Wood

(2002). Voter turnout in city elections. Urban Affairs Review, 38, 209-231.

41.

Wright

G. C.

Schaffner

B. F.

(2002). The influence of party: Evidence from the state legislatures. American Political Science Review, 96, 367-379.