THE LEGITIMATE CONSEQUENCES OF FORM OF GOVERNMENT AND NONPARTISAN BALLOTS IN MUNICIPAL ELECTIONS

In this paper, the question of whether there is a difference in voter participation between partisan and nonpartisan ballots in municipal elections is addressed. This study will employ statistical regression to isolate and measure voter turnout in these two scenarios. Reforms which began in the late nineteenth century continue to have an impact on our daily lives. The Progressive Era, which championed much social equality for our country, has an oft overlooked darker side whose influences and consequences remain. Specifically, the municipal reforms of the early twentieth century. Much of the research on the topic of municipal elections has included nonpartisan ballots, as they are included in what are known as reform cities along with manager governments and at-large elections, to name a few. Research on nonpartisan elections and turnout has yielded support for the notion that the implementation of municipal reform has served to depress civic participation. However, there is not yet a study looking solely at these variables. The research design for this study is non-experimental. A random effects generalized least squares regression with robust standard error adjusted for clustering of municipalities over time was employed to test for an effect on voter turnout based on the type of municipal ballot, partisan or non-partisan. The dependent variable is voter turnout and is a quantitative variable. The independent variable is a categorical qualitative variable which is defined by the presence or absent of party label on municipal ballots. This paper addressed this deficit using Rhode Island as a case study, as nine of the 39 municipalities are nonpartisan. Comparing turnout over five elections, the results of this study will hopefully provide strong evidence about the impact of nonpartisan ballots in local elections and aid in the overall discussion.


THE LEGITIMATE CONSEQUENCES OF FORM OF GOVERNMENT AND
Before running my husband's city council campaign, I had a number of assumptions about local politics. Since I was working for a Republican in a heavily Democratic city, let alone state, I thought that the absence of party labels in a nonpartisan contest would be beneficial for the candidate. One of the problems with this assumption about nonpartisan municipal elections was that I concurrently assumed that the residents year of Adrian's paper is the same year that the nonpartisan ballot, as well as other Progressive reforms, was implemented in Newport. Adrian's register of general characteristics consists of eleven items, many of which have been assessed, questioned and built upon over the past 60 years. The resonance this particular paper held for me led me to pursue research on nonpartisan municipal elections. Further research on the origins of nonpartisan elections and Progressive Era reforms led to an expanded scope including municipal forms of government.
NYU political science professor, Costas Panagopoulos, in an editorial on nonpartisan municipal elections, stated "This is one debate the discipline has missed the boat on....
At best, political science data on the matter is scarce, outdated, and inconclusive" (Panagopoulos 2003). There have been a variety of studies about the impact of nonpartisan ballots on municipal elections and the impact of different forms of municipal government structures (Alford and Lee 1968, Caren 2007, S. P. Hays 1964, Hajnal and Lewis 2003, Karnig and Walter 1983, Lee 1960, Schaffner, Streb and Wright 2001, Welch and Bledsoe 1986, C. Wood 2002. To date, there have been comparative quantitative analyses between comparable municipalities but there has not been an empirical test of the effect of these municipal structures on voter participation within an entire state. This paper seeks to address Panagopoulos' point by utilizing the necessary, recent, and relevant data towards a more conclusive end.
The issue of voter participation is one which ties together threads throughout our history as a democracy, from our country's founding, through the evolution of our government, up to our current political system. The United States did not realize full voter suffrage until the ratification of the 19 th amendment in 1920 granting women the right to vote. Currently, voting rights is a subject being disputed and decided in municipalities, states, and even the Supreme Court. As will be examined in this study, parties and government structures impact voter participation. Voter participation is a crucial component of representation, a keystone of our conception of democracy.
From the inception of this experiment in democracy, the founders were wary of political parties. In his farewell address, George Washington warned against political parties casting them as potential impediments to the efficacy and permanency of the country as a whole. The fear was that if the populace segregated itself into factions or parties, this division would lead to instability and potentially the demise of the entire union. In Federalist 10, James Madison acknowledged the tendency for man to fall into factions or groups of interest. With this recognition of human behavior, he decided not to support prohibition of the inevitable but to create a series of mediating forces, ultimately through federalism, to impede the negative effects of such groups. In so doing, however, the structures and impediments the founders put in place ultimately set the course for modern political parties (Aldrich 1995). E. E. Schattschneider keenly notes that the founders clearly did not anticipate the real prospect of political parties, as we know them, as evidenced by "the provision in the original Constitution that the House choose a president from five candidates receiving the highest electoral vote" (Schattschneider 2009, 51). Similar distaste and mistrust of political parties, albeit due to real not theoretical concerns, was a driving force by reformers in the Progressive Era.
Another aspect of representation which was present amongst the founders and reformers is an elitist view of who exactly should govern. Both felt that the well-educated should represent the rest of the population. As will be discussed in a later section, reformers in the Progressive Era implemented mediating forces, or structures, in an effort to bring about that outcome.
In the following sections, I will present the historical context of the Progressive Era, including how and why reformers were spurred to action and the actions they took, specifically in municipal reform. The introduction of nonpartisan ballots, the Council-Manager form of government, and other municipal reforms will be discussed. This is followed by an assessment of the implications and consequences of these reforms. A review of the extant literature on participation will lay the groundwork for the data and methods used to test the stated hypotheses. After the finding and results of the analyses are presented, the paper will close with a discussion of said findings and their implications and relevance.

WHY RHODE ISLAND?
As a graduate student at the University of Rhode Island, primarily interested in American politics at the state and local levels, Rhode Island was an obvious choice to study. But there are other reasons which inspired my selection. One was my experience living in Newport while pursuing my degree. Another was the size of the state and its 39 cities and towns. While there have been studies evaluating the relationship between non-partisan elections and participation, there has not been one which studies an entire state in this manner. Accordingly, the relative scale of data collection was manageable. Thankfully, Rhode Island's Secretary of State provides easily accessible elections results. Additionally, there was a large enough sample of non-partisan municipalities for statistical analysis. Twenty-three percent of the state's municipalities elect their local officials without party designation on the ballot.

HISTORICAL SETTING: Progressive Era
If one took but a cursory review of the Progressive Era, perhaps in textbooks or online, one would read about good government reformers working in the public interest and against corruption (Buhle, et al., 2000, Glencoe McGraw-Hill, 2009. While this assessment is not necessarily incorrect, it is hardly the whole story (Judd and Swanstrom 2004, p. 75). The Progressive Era, or the Progressive Movement, came about in response to the growing pains of a young nation which was in the midst of rapid industrial and population expansion (Frederickson, et al., 2004, Schiesl 1977.
This paper focuses on local and municipal elections, whose present day structures were in large part constructed during the Progressive Era. To understand current forms of government and elections, I will first review how these reforms. The following section reviews the history of the Progressive Era, focusing primarily on changes to local governance, and sets the stage for present day municipal government and elections.
The Progressive Era took place roughly between 1880and 1930(Hofstadte, 1955, Morgan, et al., 2007, Welch and Bledsoe 1988. At the national and federal level, there were calls for reform of industry and for protection of workers. Charles Beard, a prominent scholar and pro-reform advocate, described the Progressive Era by saying, "The country was then in the midst of a periodic depression. Strikes and unrest were abroad in the land. By leaps and bounds cities had been growing in population, wealth, poverty, slums and degradation. During the preceding years scandals and frauds in national, state and local government had been unearthed in shocking forms" (National Civic League 2009). As a result, there was a populist thread to the national Progressive Movement which called for a more equal playing field between businesses and workers as well as for civil service reforms which would make government employment more merit based. Leagues and associations sprung up to tackle such issues, including the Civil Service Reform Association and the Massachusetts Reform Club.

THE PROGRESIVE MOVEMENT AND POLITICAL PARTY MACHINES
The changes in population and the influx of immigrants experienced during the late 1800s were felt acutely in the cities and urban areas. In the 1880s New York City's population increased by 75 percent from 171,750 to 3 million, Philadelphia's population topped one million, and Midwestern cities such as Cleveland and Chicago doubled their populations (Frederickson, et al., 2004, Schiesl 1977, Wood 1958). These surges in population left cities with more people than they were prepared to manage: in short, populations soared faster than city planners could plan for them. The lack of infrastructure and systems for delivering services by the local government left a vacuum which was eagerly, and sometimes effectively, filled by local partisans.
Seizing the opportunity, the local parties grew and took advantage of their position in the community (Hofstadter 1955). According to Welch and Bledsoe (1988, p.2), "the turn-of-the-century political machines were led and backed by members of the lower class and the working class, most of whom were of immigrant background." Leading the machines were the party bosses who used a system of patronage and wielded power over many jobs in city government. Tammany Hall and Boss Tweed are infamous examples of the party machine and ward boss systems (Riordan and Quinn 1995). The system of "spoils," 1 while serving some benefit to their communities, was not without its significant downsides. Corruption, waste, and a lack of improvement in living conditions were hallmarks of these parties.
The old guards of the cities, the Brahman and Anglo-Protestant classes, were unnerved by these rapid changes in their communities. Specifically, they were alarmed by their loss of political power to the machines and party bosses. While they were aware of the need of administrative structures to handle the needs of the community, they were keenly aware of the potential for loss of status and command. The main stated goal of reform was to eliminate corruption, waste, and inefficiency (Banfield andWilson 1963, p. 138, Morgan, et al., 2007, p. 63), which reformers saw manifested in corrupt party officials and partisan patronage. The main source of such excess was the party machine and it thereby became the direct target of the reformers' ire.

REFORMERS
The tumultuous situations in the growing urban areas eventually became untenable for many and calls for reform of the municipal system increased. Many of those who went on to be reformers hailed from the middle and upper class, who had previously held power in local government (Frederickson, Johnson and Wood 2004, 36). Melvin G.
Holli identified two main types of reformers, social and structural (Hawley 1973, Welch andBledsoe 1988). While both were distressed by the state of affairs, social reformers were also interested in the well-being of the less fortunate and the overall welfare of the general populace. Structural reformers were more concerned with returning their cities to a pre-immigrant era and the associated benefits like lower taxes (Morgan, England and Pelissero 2007, 64).
Municipal reformers were united by the belief that the political machines were the primary root of the present day evils. A logical antithesis of corruption and waste is efficiency. Early on in the reform movement, efficiency was synonymous with good government (Judd and Swanstrom 2004). The drive for efficiency was tied to a desire to lower the cost of running a city, including lowering taxes. Therefore the more efficient a government, the better it was. In the same vein, reformers argued that a municipality was a corporation, as opposed to a political entity, with citizens as shareholders. A new system or structure of governance was sought to take the politics and waste out of city government (Frederickson, et al., 2004, Schiesl 1977. The practicality and efficiency of a business model was a hallmark of the reformers argument (Frederickson, et al., 2004, Hays 1964, Holli 1974, Judd and Swanstrom 2004, Link 1959, McCormick 1981, Morgan, et al., 2007, Schiesl 1977, Weinstein 1962, Welch and Bledsoe 1988. Another important component of the reformers position was their ideology, which included a firm belief in their purpose because it was good and moral. Righteously sure, reformers declared their new structure of municipal government to be "more moral, more rational, and more efficient and, because it was so, self-evidently more desirable" (Hays 1964, p. 157). Elitism was included in this moral superiority, stemming from discord and discomfort between the WASP reformers and growing immigrant population. A natural next step from this combination of ideological zeal, elitism, and business principles, was the "pervasive belief that certain citizensprofessionals, experts, and the well-educatedwere more fit to govern than others" (Hawley 1973, 9).
Municipal leagues sprung up in cities and states to tackle the municipal problems of the day (Holli 1974, Tolman andParkhurst 1895). The National Municipal League, which was founded in 1894, was "primarily concerned with structural, not social, reform, promoted 'good government' practices of various kinds and provided a network for local reform groups" (Welch and Bledsoe 1988). Membership in the League was largely made up of local businessmen and upper-middle class professionals (Hawley 1973, Link 1959, Schiesl 1977. These businessmen and members of the upper and middle class understood the opportunities available to them in reforming the local government. It was logical for them to support good government reforms which would increase access of government services and benefit their personal financial interests (Frederickson, et al., 2004, Hays 1964, Link 1959.

MUNICIPAL REFORM
At the local level, reforms were promoted to make government more efficient, more accessible and more representative. Richard Childs, thought of as the founder of council-manager government, conceived of and convincingly argued for the new government structure. His plan was "patterned after a business model of an organization" (East 1965, p. 36). Eventually, "progressivism, scientific management and public administration had accepted the model of the private business corporation as the form most likely to produce that prized commodity, efficiency" (East 1965, p. 74). Childs' work was and still is used by organizations promoting municipal reform.
In the late 1890s, the National Municipal League was founded to advocate for municipal reform and is still in operation today as the National Civic League. In 1898, the League adopted its first Municipal Plan which included a provision known as a "home rule charter" to "give more power and autonomy to local officials, a unicameral city council with nonpartisan elections, and a hands-on managerial mayor to appoint and remove department heads" (McGrath 2011). In 2003, the League published its eighth edition of the Model City Charter and continues to promote many of the same reforms and platforms as earlier editions, including nonpartisan elections.
James Bryce, and his 1888 The American Commonwealth, laid the groundwork for the structural reform that would take place over the following decades (Holli 1974).
Bryce's outline discusses corrupt officials, problems with local partisanship, the intervention of state government in local affairs, and "mechanical defects" in the structure of local government. Banfield and Wilson argue that the "program of reform" is best viewed in four categories or objectives. These objectives include: 1.
"putting the electorate in the position to assert its will despite professional politicians," which encompasses the initiative, recall and referendum; 2. "simplifying the voter's task" with the use of a short ballot and information provided by municipal associations; 3. "checking the tide of immigration," which speaks to the contrast between the demographics of the reformers and immigrant populations; 4. "separating the 'business' of city government from state and national politics" by implementing home rule, nonpartisanship, and timing local elections apart from state and national ones (Banfield and Wilson 1963, 140-141).
To the reformers, party machines and bosses were the embodiment of the problem belying the cities . Thus, the focus of reform originated with removing the parties from power. Call for nonpartisanship began as early as 1881 (Hawley 1973, Holl, 1974, Lee 1960, Morgan, et al., 2007, Schies, 1977, Welch and Bledsoe 1988.
By removing the partisan influence, the reformers could remove the bad actors and make room for those who they saw as more fit to govern.
To tackle corrupt officials, reformers were in favor of merit based placement for government jobs. This was part of civil service reform which was a concurrent movement at both the local and national levels. The removal of party loyalty and patronage blended well with the argument for merit based placement in the government (Frederickson, et al., 2004, Holli 1974. In response to the relationship between the state legislature and municipal government, Home Rule set the legal basis for local autonomy for a city or municipality within its home state (Vanlandingham 1968, 270). To remedy the perceived decline in representation of the reformers in local government, at-large elections were endorsed over ward elections (Banfield and Wilson 1963, Hays 1967, Judd and Swanstrom 2004, Morgan, et al., 2007, Welch and Bledsoe 1988. Armed with these and other prescriptions for taking the politics out of local government, the National Municipal League published its first Model City Charter in 1899 (Gates and Loper 2003).
Charter revision was already underway in Galveston, TX by 1899. Four years earlier, a Good Government club had been formed by local businessmen. In 1900, the opportunity for a new form of government presented itself when a devastating hurricane and tidal wave all but destroyed the city (National Civic League 2009, Schiesl 1977. Unable to properly supervise the resulting circumstance, the sitting council resigned and a commission of, primarily, businessmen took up the mantle of governing the city. After some legal fine-tuning with the state legislature, a legitimate Commission plan was officially in place. Part of the appeal was that a city could function as an administrative entity without conflict of interest (Schiesl 1977, 136).
The new system of government spread quickly, gaining notoriety as the Des Moines Plan in 1907, and was implemented in 300 cities by 1913 (Weinstein 1962, 169).
Unfortunately, the Des Moines Plan proved problematic in practice. The main issue was the combination of legislative and executive functions into one body. Efficiency suffered (Schiesl 1977, 146). Another result of the commission form of government was the increasing presence of businessmen and business interests (Schiesl 1977, Weinstein 1962. The consequence of this was policies which favored middle and upper class values (Judd and Swanstrom 2004, Schiesl 1977, Welch and Bledsoe 1988. Opportunely, Richard S. Childs and the National Municipal League were equipped with a solution. Their answer was a new form of government which remedied the problem of commission government by placing a city manager at the helm of the administrative needs of the council. Staunton, Virginia was the first locale to hire a city manager in 1908. However, Sumter, South Carolina was the first city to adopt the full platform in 1912. The full platform required alteration to the city charter. Staunton adopted it in 1920 (Frederickson, et al., 2004, Schiesl 1977. Dayton, Ohio followed suit in 1913 and ten years later the system was in pace in 270 cities. As of 2006, the International City/County Management Association states this to be the most common form of government with 55 percent of American municipalities using it (National Civic League 2009, National League of Cities 2013).

REFORM CITIES: Implementation and Effects
The most common implementation of the city manager platform, or Council-Manager form of government, involves charter reform which was codified in the National Municipal League's second edition of the Model City Charter in 1915. The most recent version, the Eighth Edition, was released in 2003. Presently, municipal government is described as being reform or non-form, or administrative or political.
Reform, or administrative, municipalities often include the council-manager form of government, nonpartisan elections, and at-large elections. Non-reform or political governments consist of a strong, directly elected mayor, partisan elections, and ward elections (Frederickson, et al., 2004, Morgan, et al., 2007.
Clearly, the primary goals of the reformers were to end corruption in municipal government. However, the normative goals also included making the government more accessible and responsible "to the people." Reformers viewed machines as barriers to participation because of national and state influence on local elections, patronage, and generally corruption elections (Banfield and Wilson 1963, Hawley 1973, Lee 1960, Wood 1958. Therefore, removing these barriers, it would be easier for citizens to become informed and make rational electoral decisions. By making these informed and rational decisions, they would see the fruit of their labor and political and civic engagement would take hold (Gates 1999, Keller 2002, McGrath 2011, National Civic Review 1994. Instead, the reformers removed one set of barriers only to put others in place. In this next section, I will review the extant literature on the effects of nonpartisanship, council-manager structure, and at-large elections on voter participation. The important point here is that while most political scientists believe that high levels of participation are required in order for there to be responsive executives and legislatures (Lijphart 1997, S. M. Lipset 1963, Lipset and Schneider 1983, the evidence is that these reforms produced the opposite results.

The A-Political Government
Research on nonpartisan elections and turnout suggests that municipal reform depresses civic participation. To better understand the relationship between voters and participation, it is helpful and important to review the existing scholarship on the subject. The most basic and common way to explain why people do, or do not vote is through what was originally presented by Anthony Downs in 1957 as the "calculus of voting." This calculus provides a simple cost/benefit analysis of the rationale people use for going to the polls. The basic equation states that the reward the voter receives from the act of voting is equal to a combination of both how much the voter prefers one candidate over another and the probability that candidate will win, minus the costs to the voter for this act. The costs involved include time and information which can be quite high. Based on this formula, the costs usually seem to outweigh the benefits for "rational" people. Thus, voting turnout can be understood to be low. Other scholars have sought to enhance the calculus by adding other factors which better represent the human and social qualities involved in the act of participation, such as benefits which include a sense of fulfillment from exercising a civic duty (Riker and Ordeshook 1968). The significance of the calculus of voting here is that costs associated with voting, including the time is takes to register as well as the time and effort it takes to become informed on the issues and candidates and the time it takes to get to the polling place, can easily be too high for people to overcome.
Efforts to increase voter turnout and participation invariably include ways which aim to decrease the costs of voting. For example easing the registration process either by allowing people to register the same day as they vote or allowing people to vote by mail have been shown to do just that (Wolfinger and Rosenstone 1980, Highton 2004, Hill 2006. Another method voters use to decrease the transaction costs associated with voting is by employing "information shortcuts." People use information shortcuts every day in their decision-making processes. It is costly to spend time gathering all the available information about any given decision, whether it be which car to purchase or which candidate to favor. Information shortcuts are informed by a variety of sources including friends and neighbors, the media, and opinion leaders (Popkin 1995). One particularly important shortcut, or heuristic, is political party. While political parties at the state and local levels can vary from the national parties on some messages and platforms, party labels can quickly help a voter decipher information about candidates (Conover and Feldman 1989). Party preference is likewise influenced by friends and neighbors, the media, and opinion leaders in addition to past events and current facts about a campaign or candidate (Popkin 1995).
One of the arguments for reform was that by equalizing the ballot, by removing partisan labels, voters would take it upon themselves to become more informed in their decisions. What has been shown to happen, though, is that by removing this information shortcut the costs associated with voting increase. This increase in cost causes some voters to disengage and thereby lowers voter turnout and participation.
Charles Adrian was one of the first to assess and critique the impact of nonpartisan municipal elections and point to this unfortunate consequence.
Adrian begins his 1952 paper by describing the context within which the reformers made their case for "a series of innovations designed to place government 'on a business basis' and to weaken the power of the political parties" (Adrian 1952, p. 766). He describes eleven characteristics 2 which he saw as the negative consequences and outcomes from the implementation of the specific reform of nonpartisan ballots and elections. Ultimately, he presents consequences which undermine some of the arguments reformers made in advocating such amendments. For example, rather than producing a more informed and energized citizenry which chooses the most qualified leaders for their community, Adrian points to the proliferation of incumbents and the lack of accountability for the elected officials. Adrian (1959) also categorized a typology including four generalized municipalities which vary by presence and/or strength of political parties and other groups in the community, either politically affiliated or not. Some communities have no semblance of political parties in their communities, whereas other communities have other groups which operate in very similar ways to parties but do not have a political label, per se.
Specifically, Type I includes elections where candidates with the best chance of election are those which have partisan backing, despite the lack of party affiliation on the ballot. Type II includes elections where slating groups, partisan and nonpartisan, 2 The eleven characteristics are: 1. Nonpartisanship serves to weaken the political parties in those areas where it is in effect; 2. Segregation of political leaders strictly to either partisan or nonpartisan areas is the general rule; 3. Channels for recruitment of candidates for partisan offices are restricted by nonpartisanship; 4. Channels for recruitment of candidates for nonpartisan offices are restricted by nonpartisanship; 5. Limited new channels for recruitment of candidates for nonpartisan offices are opened by nonpartisanship; 6. Segregation of funds for financing nonpartisan and partisan election campaigns is nearly complete; 7. Facilities for fund-raising by candidates for nonpartisan offices are restricted by nonpartisanship; 8. Nonpartisanship encourages the avoidance of issues of policy in campaigns; 9. Nonpartisanship tends to frustrate protest voting; 10. Nonpartisanship produces a legislative body with a relatively high percentage of experienced members, making for conservatism; 11. There is no collective responsibility in a nonpartisan body. support various candidates. Type III includes elections where various slating groups support candidates, but there is little party involvement. Type IV includes elections where neither slating groups nor parties are active in campaigns, and if they are it is only sporadically. This study illustrates some of the ways communities and their leaders adapted to the structural absence of party labels. Adrian concludes by writing, "with a few exceptions, nonpartisan elections have accomplished what they were originally designed to do; they have effectively removed regular political machinery from involvement in certain kinds of local, judicial and state elections" (Adrian 1959, p. 458). Pomper (1966) studied the effect of other groups in nonpartisan municipal elections in Newark, New Jersey. The author surmises that Newark fits into Adrian's third type of nonpartisan election. He found that where there are other groupseither religious or ethnic or racialto be organized along such cleavages, they can and will do so. What is striking about this study, as well as some others from that time period, is that it speaks to the social aspect in communities as well as to the history and formation of parties in the U.S. (Freeman 1958, Salisbury and Black 1963, Hagensick 1964. The United States was conceived without parties, but they emerged out of necessity and out of the social beings that operated the government (Aldrich 1995). On one hand, this is a hydra-like problem for the reformers. By cutting off the name of the organizations or parties, people moved to organize in a similar way. On the other hand, once the organization and structure of parties or like-groups is gone it is hard to reconstruct (Karnig and Walter 1983, p. 503).
In his research on California city elections, Lee (1960) discusses the decrease in voter turnout in local elections from general elections. While his findings do support the consensus that local elections draw fewer voters to the polls, his findings do not point directly to nonpartisan ballots as the culprit. Karnig and Walter, however, found that both nonpartisan ballots and city-manager structures impede voter turnout (Karnig and Walter 1983). Supporting Adrian and Pomper's findings, Squire and Smith show that nonpartisan elections can easily be viewed as partisan or partisan-like in the minds of the electorate based on the information available to them (Squire and Smith 1988).
More recently, Schaffner, Streb and White (2001) present further evidence that nonpartisan elections depress voter turnout and that voters use information based on incumbency rather than party, thus furthering incumbents careers.

The Council-Manager Government
The literature on the structure of local governments, namely the form of government, indicates that those with reform governments have lower voter participation in the form of voter turnout. The following section reviews the findings from relevant scholarship.
In their 1968 article, Robert Alford and Eugene Lee examined turnout in American cities looking at the categories of "political structure, social structure, and community continuity" (Alford and Lee 1968, 800). Of interest for this study is their focus on political structure, namely form of government. The authors also included ballot type, partisan versus nonpartisan, in their calculation of structure. They found that reformed governments, those with the council-manager form of government and nonpartisan ballots, have lower voter turnout than those without reformed strcutures (Alford and Lee 1968, 802). In attempting to assess the independent impact of form of government on turnout, the authors found it to be more significantly correlated with turnout than ballot type (partisan or nonpartisan) (Alford and Lee 1968, 805).
As referenced above, Karnig and Walter examined the impact of structure on municipal turnout and found reformed governments, with city-manager forms and nonpartisan elections, to have lower turnout than non-reformed governments (Karnig and Walter 1983). The authors looked specifically at these two components of reform, city-manager and nonpartisan ballots, and used cross-sectional and longitudinal analysis of elections from the 1930s, 1960s, and 1970s. Across these three time frames, the analysis shows council-manager governments lagging in turnout by about 10 percent, as compared to commission and mayor-council forms (Karnig and Walter 1983, 497).
In 1997, Amy Bridges expanded the scope of study on the effects of progressive reforms by moving beyond the geographical areas previously studied. While some literature included analysis of region and geography, the south-western United States had not been extensively evaluated (Alford and Lee 1968, Karnig and Walter 1983, Lineberry and Fowler 1967. Bridges reviewed reform structures in seven southwest cities which implemented reform in the 1950s, comparing them to what she described as "machine descendent" cities from the Northeast and Midwest . Her findings show significant differences between municipal turnout with reform cities having roughly 20 percent average turnout and machine descendants having between 43.6 percent and 57.3 percent (Bridges 1997, 101).
Concerned with a broader conception of voter participation, J. Eric Oliver studied the state of democracy in suburban areas (Oliver 2001). The author compared civic participation levels, including contacting local officials, attending community board meetings, participating in voluntary organizations, working informally with neighbors, and voting between reform and non-reform municipal governments. The only civic participation which was significantly different was voting, with non-reform governments showing a higher rate of participation (Oliver 2001, 183). Curtis Wood's examination of turnout in city elections also supports the argument that form of government impacts turnout (C. Wood 2002). Using a random sample of 57 cities, the author categorized cities based on their political or administrative structure.
Wood employed five categories for structure, taking into account the varying types of government which have evolved over time. These categories include political, adapted political, conciliated, adapted administrative, and administrative. Controlling for socioeconomic factors and the timing of elections, Wood found that cities with more political structures, i.e. mayor-council and partisan ballots, have higher voter turnout than cities with more administrative structures, i.e. council-manager and nonpartisan ballots.
Seeking further insight into voter turnout and municipal structures and potential remedies to lower participation rates, Hajnal and Lewis examined cities in California.
While their key recommendation for tangible results to increase voter participation in municipal elections is to have local elections coincide with state and national elections, their findings provide more support for the argument that structure matters. Specifically, they found that council-manager government structures dampen turnout (Hajnal and Lewis 2003, 658).
Neal Caren sought to explain the variation in municipal turnout in cities by examining mayoral contests. Studying election data from 332 elections in 38 large U.S. cities over a 35 year period, the author employed a GLS random effects model and found more evidence to support the argument that the council-manager form of government is linked to lower turnout. Specifically, "cities with council managers have a turnout 7.5 percentage points lower than cities with strong mayors" (Caren 2007, 41). In another model in the same study, the author looked at the interaction between the mayoral margin of victory and the form of government and found that even in competitive elections, there was lower turnout in cities with the council-manager form of government.

Electoral Systems
Another component of reform government is at-large or city-wide election of city councilors. Banfield and Wilson categorize the arrangements and procedures by which city officials are elected, at-large or district, as electoral systems (Banfield and Wilson 1963, 87). While the bulk of the literature on the topic focuses on representation as opposed to turnout, it is important to understand the reformers' arguments for this structural change and the potential effects of this reform. The following section will review the evolution of the electoral systems from district, or ward, to at-large and the scholarship on the effects.
Wards, or electoral districts within a city or municipality, were synonymous with the party boss and thus with corruption, graft, and greed in the same way partisanship was (Bridges 1997, Welch andBledsoe 1988). Wards were also seen as bastions of the poor and immigrants classes which found representation through the ward system. So, it is not surprising that reformers moved to eliminate them soon after they argued for nonpartisanship. The reformers saw multiple reasons for moving to city-wide elections. It would weaken the power of the machine boss while encouraging those who ran to seek approval of the entire city. Like-wise, the reformers hoped this new method would produce the desired outcome of the election of "better people" (Judd and Swanstrom 2004, Morgan, England and Pelissero 2007, Welch and Bledsoe 1988. The impact of the elimination of district electoral systems was clear, and in some places almost immediate. After the 1895 city council elections in Galveston, TX, the three African-American councilors lost their seats in the city-wide contest (Bridges 1997, 74). Welch and Bledsoe found that at-large contests result in the election of better educated and wealthier councilors than those in district elections, holding other variables constant (Welch and Bledsoe 1988, 42). In sum, the move from district to atlarge elections appears to affect who is elected and tends to disadvantage minority populations.

IMPORT OF POLITICAL PARTIES
The literature review up to this point has included discussion of political parties with regard to their presence or absence in local elections. The following section will review scholarship on the importance and utility of parties, both generally in American politics as well as specifically as they pertain to state and local political activity.
Various scholars have cited the importance of political parties. The import and value they see ranges from the state and local to the federal and even to the normative underpinnings of how the American political system developed and functions. E. E.
Schattschneider wrote in 1942 that, "the rise of political parties is undoubtedly one of the principal distinguishing marks of modern government...political parties created modern democracy and modern democracy is unthinkable save in terms of parties" (Schattschneider 2009, 1). V. O. Key shows support for this view when he wrote that "political parties perform functions essential to the operation of a democratic order" (Key 1958, 311). Sarah McCally Morehouse makes a strong case for parties at the state level asserting that "the single most important factor in state politics is the political party. It is not possible to understand the differences in the way sovereign states carry out the process of government without understanding the type of party whose representatives are making decisions that affect the health, education, and welfare of its citizens" (Morehouse 1981, 29).
One way political parties demonstrate their utility is by solving the collect action problem. Collective action is the organized and planned behavior of and by individuals and groups. The problem of collective action occurs when individuals and groups work only in their self-interest and not with those who are like-minded. This can result in the least optimal results for those involved. Aldrich provides an excellent example of a collective action problem which highlights the incentives for party formation. In his example, there are three legislators in a unicameral legislature encountering three bills which would provide distributive benefits. The benefits for any given bill would serve some districts more than others but cost all involved. Assuming each legislator is acting rationally, Legislator A would favor Bill X which most benefits his district, Legislator B would favor Bill Z which most benefits his district, and Legislator C would favor Bill Y most benefits his district. If each bill passes by a 2-1 vote, the total benefit or payoff to each legislator is negative. This is the least optimal outcome but one that is likely if the legislators act independently. The least optimal outcome is avoided and the problem of collective action is overcome when Legislators A and B agree to work together and vote only for Bill X. This rudimentary scenario shows the incentives for party formation. This problem of collective action is manifested in the electorate when voters act as individuals with their own self-interest. For example, say there are three candidates representing three diverse opinions on a salient issue. One supports eliminating roads altogether, one supports building new roads while neglecting those which are in need of repair, and a one supports fixing the roads in need of repair. Functional roads are a public good and one from which everyone benefits. If voters in this election are split between the candidate who wants to repair the existing roads and the candidate who wants to build new roads, the candidate who wants to abandon roads altogether could win; resulting in the least optimal outcome. Parties overcome this problem by aggregating the choices available so that the majority preference can win. Voter turnout is the "quintessential example" of the collective action problem with regard to the logic of voting (Aldrich 1995, 46). As Schattschneider stated, "Other things being equal, any unorganized group will exhibit a tendency toward dispersion of its voting strength which can only be overcome by planning, consultation, and organization" (Schattschneider 2009, 41-42).
Parties help solve the collective action problem through such planning, consultation, and organization. As discussed above, the calculus of voting indicates that an individual's perception of making a difference by voting can be low. Aldrich adds that the calculus of voting is preceded by a collective action problem of being informed (Aldrich 1995, 47). Parties work to decrease the cost of voting by informing voters about the issue, particularly the importance of the outcome. Parties also work to mobilize the electorate by aggregating similar preferences amongst voters (Huckfeldt and Sprague 1992, 70). Lipset (2000, 48) stresses the importance of institutionalized party competition and therefore political parties through their role of "a social mechanism which permits the largest possible part of the population to influence

DATA AND HYPOTHESES
Based on the extant literature regarding voter participation in municipalities with reform governments and nonpartisan ballots, two hypotheses will be tested in this paper.
H1: Municipalities with partisan ballots will have higher voter participation than those with no party identification.

DEPENDENT VARIABLES
The dependent variable of interest is voter participation. Voter participation is measured using two variables, voter turnout and voter registration. Voter registration (e.g. percent registered) is calculated as the number of individuals registered to vote over the total population of residents age 18 and over, for each municipality. This value represents those who are technically old enough to vote in each election. Voter age population (VAP) was chosen instead of voter eligible population (VEP) 5 because information on voter eligibility by city was not available whereas basic population was available from the U.S. Census. VEP is more precise and includes an estimate of those within the population who are actually eligible to vote. For example, VEP takes into account non-citizens, felons, and other legally excluded sections of the population.
While VEP may offer a more precise count, VAP was chosen because VEP is unfortunately unavailable. While VEP would be preferable, using VAP biases the models away from getting support for my hypotheses and therefore is acceptable.
Population data, including total population and population over the age 18, is from the Voter turnout is derived by dividing the number of individuals who did turn out to vote by the number of individuals who were registered to vote, for each municipality.
Both turnout and registration data was obtained from the city or town clerk of the municipality's canvass office.
One issue with the data worth noting pertains to all observations for the town of New Shoreham. Using the above described method to derive voter registration, the resulting figures are more than one. This would mean there is more than 100% voter registration, which is not possible. There are three plausible potential explanations for this outcome. One is that the town officials have not cleaned their voter rolls, meaning they have old and inaccurate data. Another possibility is that the U.S. Census count for the town is not as current as the voter registration counts. The data for New Shoreham was included in the models.

INDEPENDENT VARIABLES
There are two primary independent variables of interest to this study. The first is a categorical variable which is defined by the presence or absent of party label on municipal ballots. This is a naturalistic observational study where the groups, nonpartisan and partisan, have already been assigned. Data on ballot type, partisan or nonpartisan, for each municipality was gathered from the Rhode Island Secretary of State, as well as individual municipal canvass or clerks' offices. In the data, partisan cities are coded as 1, and nonpartisan as 0. In Amy Bridges' assessments of municipal reform in the Southwest, she terms chosen municipalities form the Northeast and Midwest as "machine descendants" (Bridges, Morning Glories: Municipal Reform in the Southwest 1997). While in theory the municipalities which I am studying may fall into this categorization, I am reluctant to state that they are due to a lack of specific empirical definitions. Bridges' definition states that "Machine descendants all had strong party organizations, substantial patronage resources, and concurrent elections. Another difference between big-city reform and the machine descendants is that all of the big-city reform cities had citywide elections for the city council members and machine descendants had district representation on the city council" (Bridges 1997, 131). Without specific criteria for what constitutes strong party organizations or significant patronage resources or a timeframe for when concurrent elections began or ceased, I cannot confirm that any of the cities and towns in Rhode Island qualify.
The final variable employed and tested is one which combines the absence of party label, or nonpartisanship, and presence of the Council-Manager and Administrator-Council form of government. In the data, municipalities with both nonpartisan ballots and the Council-Manager and Administrator-Council forms of government are coded 1, and those which do not have this combination as 0. Alford and Lee (1968), Karnig and Walter (1983), and Wood (2002) each included a similar variable in their analyses and found mayor-council and partisan ballots, have higher voter turnout than cities with council-manager and nonpartisan ballots.

Control Categorical Variables
There are a number of other control variables which are included in this analysis.
Because there are many factors which can influence voter participation, it is important to include those which are relevant. Of course, all potentially influential variables cannot and are not included; however, I have tried to control for the major known predictors of voter turnout. Those which pertain to this study are discussed below.
Another variable which relates to the structural reforms discussed above, such as the form of government, is the timing of elections. For the most part, municipal elections in Rhode Island are held during even years, coinciding with national and state elections in November. However, some are held during odd-years and/or during times of year that are not generally associated with other elections. There is considerable extant literature and historical evidence which shows voter participation is higher in years with national, Presidential elections than those in which midterm, state-wide elections occur (Burden 2000, Campbell 60, Conway 1981, Gilliam 1985, Shields & Goidel 1997, Wolfinger, et al. 1981. Similarly, there is evidence which shows that elections which take place during odd-years have lower turnout and participation than those in midterm years (Hajnal & Lewis 2003, Wood 2002. To take into account the variation in timing, each municipal observation was coded to correspond with when the election occurred and included General Election, Mid-Term Election, Odd-Year Election, or Other Time (e.g. non-second Tuesday in November election). Information on when each municipal election occurred was gathered from a few different sources.
For those elections which occurred in even years, both Presidential and mid-term election years, the information was readily available online at the state Board of Elections website. For those municipalities which held elections either during oddyears and/or at different times of year, the respective town clerk or canvass office was contacted to verify when the election was held.
How competitive an election is may have an impact on turnout. The more interesting, exciting or talked about an election is may compel those who might ordinarily sit out an election to pay attention and cast a ballot. Similarly, if an election is perceived to be close, the idea that one's vote will matter increases (Conway 1981, Wood 2002, Hajnal & Lewis 2003. To account for any potential influence from close or competitive elections, a variable to represent electoral competition was derived and employed. To capture overall electoral competition for each municipality in a given election, the percent was taken using the number of challengers divided by the total number of council seats. For example, if there were 5 seats on a council and 7 challengers for the open seats that election, the electoral competition value would be 7/5 or 1.4. An overall competition value was chosen as there is variance in the construct of each municipality's city or town council. In some, all councilors are elected at-large, in others all are elected from wards, and others still have a mix of ward and at-large representation. This competition value takes into account the number of seats, the number of challengers, and allows for there to be one value per municipality for each observed election.
Socioeconomic and demographic factors are correlated with participation 7 . It has been clearly, significantly and robustly documented that "citizens with higher social and economic status participate more in politics" (Conway, 1985, p. 125, Verba and Nie 1972, Wolfinger and Rosenstone 1980. To ensure that demographic differences across the municipalities were controlled, the following variables were included as percents of the population: marital status, households with children under the age of It is important to note that the ACS data comes from a survey and the values for each municipality are estimates. These estimates can have varying margins of error, depending on the data set. The margins of error can be larger for smaller geographic areas. The relationship between socioeconomic and demographic variables and voter participation is not the focus of this study yet they are important to take into account.
The ACS data was used because it is the best and most reliable available data.

METHODOLOGY
A random effects generalized least squares regression with robust standard error adjusted for clustering of municipalities over time was employed. A total of four models are run. They include a registration model and a turnout model for ballot type and form of government and a registration model and a turnout model for the combined variable of nonpartisan ballot type and council-manager and administratorcouncil forms of government. All four models were run in STATA using xtreg and the syntax code used for each is available in Appendix II.
The interclass correlation (rho) for the two turnout models are .14 and .23, respectively. The interclass correlation (rho) for the two registration models are both .93. The tables for each model are presented in the following results section.

Statistical Model
A total of four models are run. While the dependent variables change with each model, the control variables in each include the percent estimated married, the percent of households with children under the age of 18, the percent with educational attainment through high school, the percent with educational attainment of some college, the percent foreign born, the percent with a language other than English, median household income, percent race white, percent race black, percent race other, percent race other, the variable for electoral competition, the time of election in midterm years, the time of election in off-years, and the time of election in other times with the time of election in general elections held constant.

FINDINGS
The following section reviews the findings and results. Bivariate analyses of the hypotheses are discussed, followed by findings in the random effects generalized least squares regressions.  After a slight dip in the second observation period, registration appears to be increasing slightly over time. Turnout appears to be higher in the first, third and fifth observations, which corresponds to presidential election years. There is a slight drop in turnout in the second observation and a marked decreased in turnout in the fourth observation.    Figure 3 shows the comparison of average turnout in municipalities with partisan ballots to those with nonpartisan ballots. Average turnout for partisan municipalities over the five observations was 61.88%. Average turnout for nonpartisan municipalities over the five observations was 56.7%. Municipalities with partisan elections show a higher rate of registration, 10.81%, and a higher rate of voter turnout, 5.18%. This is consistent with the existing research comparing the two types of elections and provides some support for the hypothesis that municipalities with partisan ballots will have higher voter participation than those with no party identification. And again, the bivariate relationships suggest support for my hypothesis.    Reform Cities was 70.80% and for municipalities using partisan ballots and other forms of government was 74.54%. The bivariate analysis in Figure 6 supports the assertion that Reform Cities have lower participation, here specifically registration.  Over the five observations, average turnout for Reform Cities was 62.62% and for municipalities using partisan ballots and other forms of government was 60.32%. The 47 bivariate analysis in Figure 7 does not support the previous literature on the subject or the expectation that Reform Cities have lower turnout. In the following section, I present the random effects generalized least squares regressions. As previously mentioned, the model allows for the comparison of different population groups at specific points in time. It also allows for the comparison of many variables at the same time, including those which are time-varying and timeinvariant. A random effects model was included as it is assumed that the differences among the variables are not correlated with the independent variables.

RESULTS
The multivariate results presented in Table 2 do not provide statistically significant support for either of the stated hypotheses. The coefficient for the partisanship variable is positive but small and misses the threshold for statistical significance. Likewise, the coefficient for Mayor-Council municipalities is positive, but only slightly larger than that for Council-Manager and Administrator-Council. They also fail to make the threshold for statistical significance.  Table 3 replicates the model from Table 1, but substitutes turnout as the dependent variable. The results presented in Table 3 provide mixed results. The coefficient for the Partisanship variable is statistically significant at the .05 level but is negative, meaning that municipalities with partisan ballots, when controlling for all other factors in the model, turnout at a lower rate than those with nonpartisan ballots. This does not support the first hypothesis or the literature on this topic. The coefficient for the Mayor-Council variable is also statistically significant at the .05 level and negative, meaning municipalities with that form of government turnout at a lower rate than the control. The coefficient for the Council-Manager and Administrator-Council variable is the strongest statistically at the .05 level and is negative, which supports the second hypothesis and the literature. Turnout in those municipalities occurs at a rate 4.3% lower than the control.  Table 4 replicates the model from Table 2 Table 4 are consistent with expectations. Alford and Lee (1968) and Wood (2002) both employed a similar variable in their analyses and found the combination of nonpartisan ballots and the Council-Manager form of government to correlate with lower turnout. Their analyses, however, only looked at turnout and not registration.
The variable representing the combination of ballot type and form of government is statistically significant at the .05 level and negative, meaning that municipalities which have both nonpartisan ballot and the Council-Manager or Administrator-Council forms of government register to vote at a lower rate 3 than those with partisan ballots and reform governments.  Table 5 replicates the model from Table 4, but turnout is substituted as the dependent variable. The results are counter to expectations and the existing literature. The coefficient for the combination variable just misses the threshold for statistical significance but is positive, meaning that Reform Cities tend turnout to vote at a higher rate than those with partisan ballots and other forms of government. Taken together, the bivariate and multivariate analyses offer a mixed bag with regard to support for either hypothesis. While the bivariate analysis of turnout and registration between partisan and nonpartisan ballots supports the first hypothesis, the multivariate analyses for registration finds some support but the multivariate analysis for turnout yields contradictory statistically significant results. Similarly, the second hypothesis concerning form of government does not find support in the bivariate analysis but finds some support in the multivariate analyses.

DISCUSSION
This analysis of municipal elections has produced a range of results, some of which seem contradictory. The following discussion will review the major findings, how they pertain to the stated hypotheses and questions posed at the beginning of the paper, and offer explanation for some of the apparent inconsistencies in the findings.
The existing literature on the effects of ballot type, form of government, and combinations of the two, on voter participation uses voter turnout as the dependent variable representing participation. This study looks at two forms of participation, voter registration and voter turnout. Because voting is contingent on registration, it is logical that each represent participation. Likewise, it is logical that there would be differences between registration and turnout because they are two specific and distinct actions 8 .

Major Findings
The first hypothesis, testing the relationship between nonpartisan municipal ballots and voter participation, found support in the initial bivariate analysis which shows that municipalities utilizing partisan ballots have higher rates of registration and turnout.
The multivariate analysis of voter registration, while missing the threshold for statistical significance, shows additional meager support. The multivariate analysis of turnout, however, suggests that municipalities with partisan ballots turnout at rates lower that those with nonpartisan ballots. This is counter to the expectations and literature on this topic. Multiple studies examining the relationship between ballot type and turnout have found lower voter turnout in municipalities using nonpartisan ballots (Alford and Lee 1968, Hawley 1973, Karnig and Walter 1983, Schaffner, Streb and Wright 2001, C. Wood 2002.
The second hypothesis, testing the relationship between forms of municipal government and voter participation, did not find support in the initial bivariate analysis. This analysis shows that, counter to expectations, municipalities with Council forms of government tend to have higher rates of registration than those with Council-Manager and Administrator-Council forms and markedly higher rates than those with Mayor-Council forms of government. Similarly, the bivariate analysis of turnout shows Council forms of government to have higher rates of turnout than the Council-Manager and Administrator-Council governments and that the latter forms have higher rates of turnout than the Mayor-Council municipalities. These results are counter to both expectations and extant literature on the topic. Multiple studies examining the relationship between forms of government and turnout have found municipalities which have a Council-Manager form of government tend to have lower turnout than those with the Mayor-Council form (Alford and Lee 1968, Karnig and Walter 1983, C. Wood 2002. does not appear to influence either registration or turnout.

Practical Relevance
The findings in these analyses provide further support for a number of assertions about voter turnout in municipal elections and offer additional information on registration. Island has also cast its Electoral College votes for Democratic presidential contenders in twelve of the past fourteen contests. All of this is to say that there is a lack of strong competitive parties in the state. This lack of partisan competitive results in a lack of political innovation by the parties and a lack real alternatives for the electorate (Hawley 1973). This lack of competition may influence the political behavior of the electorate. V. O. Key notably articulated the matter of one-party states when he wrote, "Over the long run the have-nots lose in disorganized politics" (Key 1949, 307).
Another limitation is geography. The data comes from the smallest state in the union, and one which is located well within a single region. While the similarities within and between the municipalities may decrease the overall variance in the data, the lack of variety may play an important and unmeasured role in the analyses.
As previously mentioned in the Data and Methods section, the data for New Shoreham is problematic. It is impossible for a city or town to have over 100% registration. This is a problem one finds when collecting such data in any part of the United States. 9 The National Conference of State Legislatures provides historical data going back to 1978. http://www.ncsl.org/research/about-state-legislatures/partisan-composition.aspx 10 That Republican, Lincoln Chafee, went on to change his party affiliation twice. At the time of this writing, he is a Democrat and the Governor of Rhode Island. http://www.politico.com/story/2013/05/lincoln-chafee-to-switch-parties-sources-say-91994.html Although federal elections are regulated at the federal level, all elections are administrated at the local and county levels. This means there can be wide variations in the collection, filing, storing, cleaning, and saving of said data.
One possible limitation of the statistical analysis is that the random effects generalized least squares model used only a random intercept which assumes that variance and covariance of the variables over time are equal. However the model used was shown to account for a substantial amount of clustering within communities which provides does provide robust test of model effects. A more precise estimate might be obtained using models with covariance structure which allow for variation over time.

Further Research
In keeping with the previous section on limitations, there are two areas for further research and study which would provide immeasurable assistance to the study of municipal elections and voting behavior. One possibility is the creation of a central database of municipal data. The lack of consistency in collection and storage of such data, as well as the difficulty in accessing the data create barriers. Luckily, Rhode Island's electoral results are readily available on an easy to access website. The relatively low number of cities and towns made it feasible not only to collect the data but to contact local canvass offices for additional information when needed. However, it still took a significant amount of time to collect all of the necessary data. When I initially began research for this study, I was surprised to learn that there is not one readily accessible place which can definitely tell you how many municipalities in the United States use nonpartisan ballots, or specifically have a Council-Manager from of government. There are resources, such as the International City/County Management Association (ICMA) which produces the Municipal Yearbook, but they rely on survey data. If there was a central database which housed municipal data, I would venture the number of studies would increase, thereby increasing understanding in a variety of topics which could have important policy implications.
Another opportunity for further research would be to replicate this study in other states, even the entire remaining forty-nine. Replication of the study in other states would offer valuable scholarship on the topic. By increasing the number of locations studied, more would not only be known about those individual states, but more comparison and analysis between states could occur. Such an undertaking would, of course, be made much easier is a central database of easily accessible municipal data existed.

CONCLUSION
The purpose of this study was to test the effect of two structural reforms on voter participation in municipal elections. The findings and analysis add to the current literature on the topics of urban reform and voter participation and also contribute to a greater understanding of municipal elections in Rhode Island. It marks the first study to evaluate ballot type and form of government over five electoral observations in an entire state and in Rhode Island.
It is clear from this research that structure matters when it comes to voter participation in local elections. The kind of ballot, partisan or nonpartisan, the municipal form of government, and the timing of elections each impact the rates at which citizens register and turnout to vote. For policy makers with an interest in boosting voter participation at the local level, there are lessons in these findings.
Earlier in 2014, I was asked to consult on the proposals for charter reform in the city of Newport, RI. Generally, they were interested in ways to increase civic engagement and voter participation. More specifically, they were evaluating a number of structural reforms including the way in which the mayor is elected, whether or not to change from a hybrid system of ward and at-large elections of councilors to all ward or all atlarge, and what the current form of government really means with regard to accountability and responsibility. Due to my work on this thesis, I was able to provide the charter reform committee with information about the current literature on the areas of interest which I have studied. I also offered ballot type, moving from nonpartisan to partisan, as another element to consider. Now that this study is completed, I can offer the committee research on these topics which was conducted on Rhode Island. The findings go beyond a single state, however. The scope of this study allows the findings analysis to be used in future research on the topics covered.
"About all that can be concluded about voting in state and local elections is that scholars have a wonderful opportunity to narrow our focus of ignorance" (Key 1958, 614).

Descriptive Statistics
Municipalities in Rhode Island

Electoral System
At-Large 26 Ward 8 Hybrid 5