State and National Elections Sub-national elections do not occur in isolation. Studying state elections is not only about studying gubernatorial races or the outcome of contests for the legislature. The outcomes of these elections appear to be influenced by a political context dominated by national political concerns. Everywhere, national elections tend to engender higher levels of attention, interest and participation. Consequently, national elections and concerns about national politics can contaminate non-national races where information is less readily available, participation is lower and often affected by the political situation at the national-level when the state election is held. At the same time, sub-national elections are not often mirrors of the national election even when held at the same time, in the same district with the same electorate and the same competing parties. In certain circumstances, they might be entirely independent of national races, or a local race may even influence the national campaign. In this paper, I aim to do a comparative analysis of sub-national elections to see what circumstances systematically influence their outcomes relative to national elections. Some analogies can be made from sub-national elections to trans-national elections like those for the European Parliament. Certain characteristics are shared. Information cues derived from national politics and other heuristics might be used more extensively in the less-salient environment. Concerns and issues distinct from the office actually being contested can influence the vote. For instance, national economic conditions can hurt the party of the President in a gubernatorial election even if that state has an incumbent governor from the opposition party (Peltzman 1987). As a result, disproportionate changes in participation rates among one group of partisans might determine the outcome of a race. I focus on explaining this volatility over-time and across-cases. I am interested in testing theories about how institutions reduce this volatility or account for particular patterns of behavior. Specifically, I test whether strong party competition reduces or increases the observed volatility and whether the timing of the state election vis a vis the national polls has a systematic effect on the outcome of the race. Are non-national elections all alike? There may be varying degrees to which non-national election are insulated from national political pressures. Non-national elections are influenced by national elections and electoral conditions, and non-national elections might influence national races. A growing literature has explored how these non-national elections, and the behavior of both candidates and the electors, differ from national elections. Reif and Schmitt (1980) developed a typology of elections, calling national elections "first-order elections" since there is the most at stake in those contests. Second-order elections share party systems with first order elections, but do not determine the national government. While the terms refer to distinct categories, with third-order elections becoming akin to opinion polls, there is a continuum of elections, with the most decisive for the formation of a national government on one extreme and the less decisive elections acquiring attributes more similar to second-order elections. So, the extent to which national elections are decisive, the decisiveness of the second-order elections on the path of government, and the importance of policies enacted by the office being selected all determine the extent to which an election takes on attributes of a second-order election. In the U.S., I ask whether state elections can be properly understood as "second-order" elections similar to the ones in Europe where scholarly attention focuses on European Union elections. By examining how national political forces influence non-national elections, I hope to shed light on American politics as well as test theories and explore riddles posed by students of comparative politics. While I focus on non-national elections, the phenomena I am concerned with extends beyond the results of these elections. Elections are not actors in our political life, they are stages on which actors perform. Specifically, there are two main actors in the electoral drama: politicians and citizens (all those eligible to vote). In a critical supporting role, we find parties aiding the efforts of the politicians, integrating their efforts with a set of loyalties and attachments employed by the citizens, and organizing the competition. Only actors have the capacity to determine outcomes. By examining election outcomes, I am implicitly examining the dynamic interaction of these actors and searching for patterns in that interaction to better understand the outcomes. Citizens The citizen's set of behavioral options is small and appears quite simple. A citizen can choose to vote or not to vote in an election. If voting, a citizen chooses between the candidates in the race. However, if there are multiple races during one election, the combinatorics increase as the citizen must make a voting decision for each race. Since the voter can choose candidates for different parties or choose not to vote for anyone for some races, the number of observable modes of behavior is quite large. Consider an election, more restrictive than (but indicative of) most American elections that will determine one national executive, one representative to each house of the legislature, and a state executive with only two candidates per race: there are four races which the citizen can vote or not vote, and two choices per vote (assuming no write-in options). We can observe eighty-one different scenarios. Extend this analysis over time and the possibilities continue to increase exponentially. It is hardly surprising that explaining voting decisions over multiple offices over the course of time remains one of the outstanding puzzles in political science. Many voters do not change their behavior very much and act consistently with their partisan predispositions (Campbell et al 1960, Miller and Shanks 1996). Others systematically change their vote for a single office based on prevailing conditions of the previous years (Key 1966, Fiorina 1981). Enough fail to vote for different offices in a single election, split their tickets between candidates of different parties or neglect to turnout for every election (Burnham 1970) to affect the outcome of the election. It is this "drop-off" and "roll-off" that is particularly important to non-national elections since these phenomena appear to fluctuate systematically with past election results, short-term economic factors and varying levels of intensity in party competition. Politicians One can conceptualize elite political strategy as comprising two sets of tactics: efforts to influence factors that the politicians have direct control over, and efforts to insulate politicians from factors that they have little or no control over. The first set includes much of the tactics that one sees when one observes campaigns, fund-raising, advertising, media strategies, etc. Each is an example of the politician exerting maximum effort to capture the loyalties of enough voters to emerge victorious. While there is considerable variance in the quality of these efforts, even the best, well-financed campaign of the most likable politician might fail due to factors beyond that particular politician's control. These include adverse economic conditions which hurt incumbents, coattails from another race that abets the opposition, or low turnout or interest in an election. While there are surely appropriate strategic responses to each of these conditions, the conditions themselves are not dependent on the actions of the politician in the same way that, say her fundraising is dependent on the skill and effort devoted to raising money. The best a politician can aim for in these circumstances is to maximize their chances of winning given these conditions. To maximize their chance of winning (or keeping) office, politicians can choose to insulate themselves from electoral conditions which they have little control over, they can gamble that particular conditions will systematically benefit them over the distance of their relevant time horizon, and/or seek to control the conditions under which elections take place. Much scholarly attention has focused on control over the conditions under which elections take place. These strategies might include the ability to manipulate the social and economic conditions that influence voting behavior (Tufte 1978, Nordlinger 1981). However, as nations feel the constraints of globalization on their ability to manage their local economies or cede power of the money supply to independent central banks or supranational institutions that are beyond their control, the ability to effect changes in the economic or political cycle will diminish. Alternatively, success at managing the economy is becoming so limited that politicians might choose to relinquish responsibility for the economy to diminish the risk of being blamed for failures. Consequently, one might expect that politicians are less likely to seek control of the conditions under which elections take place. Unable to exert much influence on the external environment, strategic politicians in parliamentary systems may choose to time elections, or their entrance into governing coalitions, to coincide with periods which are most amenable to their electoral chances. A growing literature in comparative politics exhaustively investigates this phenomena. In non-parliamentary systems that do not give governments control over the timing of their re-election, there are fewer resources politicians can employ to link their prospects with conditions that will benefit them. That does not mean they are nonexistent: party labels, for instance, links one politician to others in the same party system, allowing that politician to share the benefits like a fund-raising base and a loyal, dependable core group of voters. However, this exposes them to swings in support for the party independent of their own performance in office. Often the only recourse for these politicians is to protect themselves from adverse conditions from which they have no control. Lacking extensive control of monetary policy, sub-national politicians have been limited to this strategic option for a long time. Many of the progressive era reforms that resulted in secret balloting, non-partisan elections, and changes in ballots to discourage straight-ticket voting, whether deliberate or not, had consequences for the degree to which candidates could insulate themselves from national party politics. Similarly, more recent innovations like the institutionalization of legislatures and governorships (with longer terms, reelection opportunities and personal staffs devoted to their particular legislators), blanket primaries and the scheduling of the election relative to national elections have had similar effects. American politics These are not entirely new concepts to be applied to American politics. Research along similar lines has focused on mid-term elections for Congress and the nearly-inevitable loss of seats by the President's party. Angus Campbell's (1960) seminal work on electoral change, differentiated between high-stimulus and low stimulus elections. Low stimulus elections are marked by less lower levels of turnout as peripheral voters stay at home. Since Campbell thought these peripheral voters were mainly affected by short-term forces which disproportionately benefitted the victor of the previous Presidential election, the absence of those forces two years later caused the decline in support for the President's party. Converse (1966) suggested that variations in political information provide the individual causal mechanism for surges and declines in voting patterns. In particular, he thought the high-information levels of presidential election years explain why Presidential elections are so volatile over time as the number of defections increases. However, his work did not explain why Congressional candidates of the President's party benefitted in Presidential election years. Consequently, work focusing on the strategic behavior of the office-holders and their opponents were used to explain incumbency advantage (eg Jacobson, Zaller 1997). High quality challengers are deterred from entering races which they do not have a good chance of winning. Presumably, these challengers are less likely to enter races where short-term forces would not benefit their campaigns. Other work explicitly focused on the relationship between the national, Presidential election and the district races held at the midterm. For instance, Tufte (1975, 1978) focused on the importance of economic conditions, while Kernell (1977) suggested that midterm elections reflected the disappointment shared by many voters when the high expectations generated during the president's honeymoon failed to be realized. James E. Campbell (1993) integrated these theories with Angus Campbell's work by dividing the electorate into three identifiable groups, independents, partisans of the advantaged party and partisans of the disadvantaged party. Independents focus more on personal characteristics and local conditions during mid-term years. "Advantaged partisans" are more likely to turn-out for their party during the Presidential year, but "disadvantaged partisans" are less likely to turn-out in the Presidential election. During off years, these partisans revert to normal turnout patterns (Ch. 5). These forces result in the regular vote loss by the party of the President at the midterm. Campbell (1986) duplicated his analysis on state legislative elections and found that Presidential coattails mattered during mid-term state elections, but were modest in those states that lack highly competitive parties. The most important study of state level election results was Chubb (1988). Chubb looked at the relative impact of national economic conditions, state economic conditions, presidential, senatorial and gubernatorial coattails, and turnout (surge and decline) on state-legislative and gubernatorial elections. He examined these elections in light of the "institutionalization" of many state legislatures. As salaries and support-staff increase, legislators have greater incentive to maintain their office, so Chubb hypothesized that they were increasingly taking steps to insulate themselves against all conditions over which they have limited control. States have been allowing governors to run for four year terms elected during non-presidential years, re-election and granting them increases in salary, staff and authority. He found that state-legislators had effectively insulated themselves from many factors out of there control, but both governors and legislators were vulnerable to swings in the national economy and attitudes towards the party of the President. It is an exciting time to study the influence on national politics on state politics. Few states are dominated by one party, but there are still substantial variation between the strength and organizational prowess of state parties. Many states reformed their process of electing officials. Blanket primaries, run-offs, mid- and odd year state elections have all become more commonplace. Today these reforms have increased in significance as Washington has pursued efforts to devolve more authority to the states while encouraging more diversity within state policies. Meanwhile, states increasingly are becoming an important launching pad for federal office. Many of the strongest congressional candidates are state office-holders, and three of the past four Presidents of the United States occupied gubernatorial mansions. The outcome of state elections has the potential to impact on the lives of citizens more so today than in years past. Aside from a growing literature examining differences in the informational context of the choice and the extent to which national or state-specific economic conditions affects voting, we do not know very much about the context under which sub-national elections are held. The extent to which those national elections influence state elections has not been well explored either in the literature on American politics nor from a comparative perspective. Similarly, we do not know much about the prior beliefs of office-seekers at the state-level and how those beliefs influenced decisions to reform the electoral system. By studying non-national elections, the behavior of the office-seekers in those elections and the voters that participate in the polling we can shed light on national politics in federal systems. Traditional political parties in federal systems like the U.S. and Canada were organized by state, often with much variance in organizational strength and ideological attachment across the states. However, only a few minor parties such as Farmer-Labor in Minnesota and the LaFollette's Progressives in Wisconsin, became major (but ephemeral) players on the state level in the U.S. In most cases, across most time periods, politics have been organized along the same partisan lines as the national political arena. As a result, the party of the incumbent governor may be the same as the incumbent president and share his baggage. Or, the party of the governor may be out of the White House, and they may benefit from protests against the national party being expressed at the local level. Much of the literature of federalism has focused on policies and policy outcomes. Recent welfare reforms in the U.S., growing popularity for block grants, and an ideological preference for localized decision-making have all led to an increase in the extent to which states can once again exercise their prerogative as laboratories of democracy. As a result, one might argue that state elections have increased in significance. In the age of globalization, it is hard to argue that states enjoy less contraints on policy and taxation. However, these developments in federal relations suggest that state politicians have more latitude to implement novel policies. Recent governors like Thompson, Ryan and Weld can carve out niches (and national reputations) in policy areas, as well as establishing reputations for themselves distinct from their party. Concurrently, the number of states dominated by one of the two major American parties has declined precipitously. The one-party south has disappeared. As of the 1999 elections, 35 out of 50 states were served by a governor or a senator from different parties, some elected on the same day. State legislatures are now full-time occupations for most law-makers. Mirroring similar developments on the national level, state campaigns are increasingly expensive. Today, state elections are both more important and more competitive. While there has been some important articles and textbooks published on state elections in the past decade, these analyses have not explored the impact of two of these developments on elections. In the past decades several states have changed their elections to a mid-term or odd- year cycle rather than hold the polls on the same day as the national election. I have not found any analysis of why these states changed the date of their polls, nor any analysis that has explored differences between those states with presidential year elections and those with off-year races. Secondly, while several studies have been done on political parties and state-level organizations, and the various party systems in the states, there have been few attempts to seek to explain how these variations might impact on electoral outcomes across the states. Comparative politics Reif and Schmitt (1980) first articulated the concept of "second-order" national elections when analyzing elections to the first European Parliament in 1979. They devised a typology to explain lower turnout during European elections, shifts against the incumbent party (-ies) and the improved performance of small parties. Since less is at stake, Marsh (1998) explained, voters are encouraged to shift between participation and non-participation, casting a "sincere" ballot and strategically or expressively voting for a party other than their true, ideal preference. As a result, large parties lose some of the support they attract in "first order" elections and unpopular incumbents often lose even more to expressions of protest. The popularity of the governing party is influenced by the national election cycle, with broad tendencies for unpopularity peaking around the midpoint of a government's term in office (Shugart and Carey 1992). If the source of change in voting behavior is the consequence of different sincere preferences, then the election might not be understood as a "second-order" election (Marsh 1998). This would occur if the election was fought over different issues, so the outcome of the non-national election is independent of the outcome of the national campaign. It is unclear whether state elections should be properly considered "second-order" elections. The candidates change, and even when the party system is unchanged, the salient issue dimensions that are foremost on voter's minds may be different. However, the important issues in European elections are also different than the national elections in each member country, yet Marsh (1998 and van der Eijk, Franklin and Marsh 1996) continues to find second-order election effects in European elections. Examining state elections could be an even more interesting exploration since in some polities, state governments are more consequential than the European parliament. In countries like the U.S., where state governments are becoming more important in their federal systems compared to several decades ago, would we find diminishing second-order election effects? What other factors intervene to counter-act the contamination of second order elections by the first-order arena? Do particular economic conditions make second-order elections vehicles for protests against governing parties while other conditions enable the focus of the electors to shift elsewhere? Has the rise of personality-centered candidacies at the expense of party-centered campaigns diminished the swing against governing parties by disentangling a party's nominees from other elected office-holders of the same party? Or, do weak party organizations breed environments where weak candidates are recruited, supporters fail to mobilize and as a result, the party's election results are highly volatile? Complicating matters further is the federal structure of governments with state-level elections. In each polity, states reserve some powers and there can be divided government. In practice, divided government may be the norm. Under conditions of divided government, the policies the state enacts may not be strongly dependent on the outcome of the sub-national election. Alternatively, the power that party or its officials can wield could make state elections more important than second-order elections like the European Parliamentary races which are widely perceived to have little impact on the national polity. This is an issue of growing importance as countries like Britain, Canada and Spain devolve (or discuss devolution) more authority to regional governments. How much power must be devolved in order to ensure that a large number of voters are participating and using the regional elections as a forum for something other than voicing opinions about the national government? A comparative project design should shed some light on these questions. The U.S., Canada, Germany and Australia all have very different federal systems, and the states, provinces or lander within their borders do not enjoy the same powers. I hope to explore whether second order election effects encourage voters to support smaller parties in these countries, much as minor parties gain support in European elections. Do state elections exhibit patterns of anti-government swings consistent with the cyclical nature of second-order elections elsewhere? Do these patterns vary across countries? Possible outline: A chapter (or chapters) comparing midterm, odd-year and presidential year state elections relying heavily on pooled time series analysis. One analysis can look at the outcome of the elections (winners, the competitiveness of the races and the volatility of the results). Another can look at differential turnout patterns. ICPSR has data on all the election returns. I should be able to add data like party strength indicators from other studies as well as economic data. Finally, I may need to put these together to look at candidate quality, perhaps by comparing several states or a particular time period. A chapter exploring why several states have changed the timing of their state elections.. This would be more qualitative. A comparative analysis that exports the American analysis to other federal systems. Here I will most directly explore the question of the extent to which sub-national elections are second order elections. Different countries can be compared on the basis to which state elections are more or less similar to national elections. Similarly, by bringing in other countries, I can continue to explore hypotheses related to the timing of the elections to see if the honeymoon-midterm-tune-up cycle is consistent across many countries. Finally, I can add one of my currently-shelved experimental/behavioral decision theory projects about information search or status quo bias. The status quo bias would be particularly interesting, since the second-order election and the older political-cycle theories suggest strong patterns of behavior opposing the status quo in these elections.