Scholars of political science often lament the difficulties associated with finding the most appropriate data to test theories about how the political world works. The study of local political participation and, more broadly, elections is certainly no stranger to this quandary. However, considering that the nearly 90,000 units of local government generally hold regular elections for a wide array of political offices, this field of study should offer an embarrassment of riches; we should never be at a loss for a way to examine the questions that are central to the functioning of democracy.
Ironically, though, these lines of research are instead plagued by a paucity of large-N studies, attributable not to a lack of quantity of data, but rather to a dearth of quality data. The greatest frustration of this endeavor is not whether appropriate data exist, but how to gain access to them in a useful format. In this article, I briefly reflect on my personal experiences with a major data collection effort on local elections. I then address how a centralized dataset on local elections would afford researchers multiple avenues of inquiry to more fully investigate the relationships between jurisdiction size, political institutions, and political participation from a variety of angles that are now impossible because of data constraints.
In 2004, David Lowery and I published an article in Urban Affairs Review entitled “Political Participation and Metropolitan Institutional Contexts” (Kelleher and Lowery Reference Kelleher and Lowery2004). We tested our expectations on the most recent municipal legislative elections in 336 municipalities in 12 metropolitan areas. The dependent variable was turnout, the most basic measure of political participation.
Because there is no centralized warehouse for local election data, the returns necessary to construct the dependent variable were obtained from a number of different sources—county and city websites, county and state boards of elections, local newspapers, and direct contact with municipalities. Collecting this information was an exercise in patience, fortitude, and perseverance that took nearly a full year to complete. Despite countless efforts to obtain full information for our sample, many difficulties precluded the securing of the necessary legislative election returns for all of the municipalities on which we intended to focus. Our final dataset included returns from 72% of our initial goal, for a variety of reasons discussed in the following paragraphs.
First, election data in select places has a very short shelf life. Despite targeting the most recent elections in the localities studied, in many instances, records had already been purged and data were simply no longer available. A second, related obstacle was the level of detail and aggregation that existed for data that did “stand the test of time.” In other words, although some information was available, it had not been preserved at the level of detail needed for our particular investigation. For example, although we were able to obtain official turnout figures for the election as a whole, we could not separate ballot returns across local races for mayor or town president, local legislators, and local judges, and thus, we were unable to investigate patterns in roll-off. Clearly, the lack of any systematic and timely retrieval of local election data results in a great loss of information for policymakers, administrators, and scholars. Third, in some cases, local registrars simply refused to provide voting returns, and newspapers failed to report complete election results. Any effort to partner with these individuals and organizations to ensure timely collection efforts would definitively improve the ability to collect reliable election returns at the local level.
A centralized local elections database has the potential to make a great contribution to the study of local elections and will enable scholars to ask a myriad of questions that are now stifled by the enormous time investment required to collect the requisite data. For scholars like me who are interested in the effect of metropolitan institutional structures on patterns and trends in political participation, this possibility is particularly exciting. Local governments are among the most understudied areas in political science, which is a terrible oversight, considering the diversity of their populations, elected leadership, political institutions, and metropolitan contexts.
With a vast, nationwide dataset (as opposed to the purposive samples employed in many studies, including my own), we would be able to leverage this variation much more effectively. Through the availability of larger samples over multiple years, we could revisit the conflicting evidence in prior research, expand on current participation research that relies on self-reported election turnout data, and also begin to explore a host of other timely puzzles and hypotheses with both academic and practical application. For example:
• How do regional partnerships influence patterns of turnout across participating municipalities?
• Does the act of “exit” influence the strength of “voice” when one moves from one place to another? If so, how? And is this influence different based on the nature of the office that is contested?
• What are the most effective mechanisms for mobilization of citizens? How do these mechanisms interact with institutional variation, leadership patterns and trends, and other municipal-level indicators?
As Oliver wrote, “To fully understand the mechanisms of democratic governance, me must consider both individuals and their environments” (Reference Oliver2001, 189). Improving the availability of data from elections at the local level will enable scholars to do just that—make important and necessary continued theoretical advancements in understanding the intersections among jurisdiction size, diversity, and other indicators of metropolitan institutional structure and design. In light of the growing prominence of local governments in the provision of services to citizens; resonating calls for regionalism and cooperation between municipalities; and discussions about efficiency, effectiveness, and economies of scale in these troubled fiscal times, the “black hole” of local election studies must be addressed. The construction of a local elections data repository that will stand the test of time is critical. Such an archive would afford the discipline of political science a unique opportunity to collect data that otherwise might be lost or underused and, in doing so, to inform policy and the processes of democratic governance and representation.
