State Parties Research: The Quest for Strong, Competitive State Parties
Abstract and Keywords
This article addresses how the two reformist traditions have shaped the party systems of the states and how they can be studied. It also provides a broad outline of research on partisanship in the state electorates, state party organizations, and the role of the parties in state government. Some commentary on future directions for research is presented. It then briefly reviews these recommendations, focusing in particular on what kinds of opportunities and challenges are given for state parties research by the changes in the ideological polarization of the parties which are so evident at the national level. Stronger, responsible parties are supposed to increase both government's accountability to the public and policy coherence. The future for state parties research is bright with the promise of new data resources which will permit stronger and more definitive tests of theories of party and citizen behaviour.
The contours of political parties in the states, how they have changed, and how we study them are very much a function of two rather contrary reformist traditions. One is political: the Progressive movement which generated regulations, variously adopted across the states, and state laws and court decisions which at once constrain and limit the parties, while also institutionalizing them and almost guaranteeing some form of two‐party competition across all the states. The second tradition is disciplinary: a reformist inclination within political science stemming from the mid‐twentieth‐century work which declared that many of America's political ills could and would be cured if only our weakly organized, poorly disciplined, and sometimes uncompetitive party systems could be greatly strengthened. The disciplinary movement set much of the normative and theoretical agenda for the study of parties in the states that followed.
To appreciate the journey of political scientists' study of state parties we need to appreciate some elemental truths which make the systematic study of state parties both vitally important and extremely challenging. The importance of parties is that they are so central; unlike the courts, for instance, which seem to sit off to the side (p. 404) of political battles much like the referees in a football game or boxing match, the parties, in the words of the authors of a major text, “permeate every aspect of state government” (Bibby and Holbrook 1996, 96). This is reflected in the commonly adopted framework that we use here, one that examines parties in the electorate, the party organizations, and the parties in government. Of course, as discussed below, the impact of parties in the political system can best be understood by focusing not just on these individual components, but on how they are interconnected.
The challenge in studying state parties stems from the complexity of the subject and the practical difficulty of gathering data across fifty states. First, obviously the states vary a lot—no one can confuse Alaska and New York or California and Mississippi. This means that the parties across the states also vary and operate in somewhat different ways. The full complexity of what is involved when we think about party activity in the states includes two competing parties at three levels—in the electorate, as formal organizations, and in the branches of government across all the fifty states. Optimally, we would have analyses of the interrelationships of these components as well. As we will see, a lot of bright ideas have been floated and methods developed to make sense of this complexity.
Second, if we deal with the conceptual complexity, practical limitations in data gathering mean that we lack the good‐quality measures that empirical research demands. While measures of a few aspects of party politics in the states have long been readily available, the field continues to suffer inadequate data on key features of the party system over time as well as across all or most of the states. The result is that too often scholars are left to make broad, imprecise generalizations (“some states do this, but most appear to do that”), or large differences are noted among the party systems—and they do vary a great deal in most respects—but we have not developed the theory or data resources to achieve convincing explanations. There is a lot left to do.
The plan for this chapter is as follows. In the next section I lay out the broad context for how the two reformist traditions have shaped the party systems of the states and how we study them. This helps to establish the broad evolution of the state parties over the last half‐century or so and the main research trends that we have charted. Then the next three sections review the broad outlines of research on partisanship in the state electorates, state party organizations, and the role of the parties in state government. Each section contains some commentary on future directions for research. The concluding section briefly reviews these recommendations, focusing in particular on what kinds of opportunities and challenges are presented for state parties research by the changes in the ideological polarization of the parties which are so evident at the national level.
(p. 405) Two Reformist Waves Shape State Parties and Parties Research
The Progressives Pushed Laws to Weaken Parties
The state parties have been shaped in good part by the reforms of the Progressive era around the beginning of the twentieth century. They sought to break the power of the traditional patronage‐based parties, reflecting early beliefs that parties as “factions” worked against the public good and contributed to corruption and inefficiency in governing (Duncan 1913; L. Gould 1986; Hofstadter 1955). Their reforms sought to break the hold of the parties over electoral politics. The most pervasive in its adoption, and perhaps its impact, was the direct primary, which removed control of nominations—a key source of power for traditional party organizations—and gave that power to the citizenry. That change allowed the development of what in the latter half of the twentieth century became known as “candidate‐centered” campaigns. The direct primary is the principal means of candidate selection in the country with only a handful of states either providing for or allowing nomination by state conventions. Within the realm of the direct primary the states vary in their permissiveness of who is allowed to participate (LaRaja, Chapter 9 in this volume) and whether the parties can endorse candidates (Morehouse and Jewell 2003b, 133–8). These provisions appear to have an effect on the ideological character of the primary and caucus participants (Carsey et al. 2006; G. Wright 2009).
The Progressives were less successful with other reforms, but these too have left lasting effects on the state parties. The institutions of direct democracy—the initiative, referendum, and recall—all sought to break the power of the parties and party‐controlled state legislatures and were adopted in various combinations in about half the states. Although in recent years the parties have sometimes learned to use the initiative process to achieve policy ends, more generally it is accepted that initiative and referendum provide meaningful extraparty access to the policy process in the states where they are used (D. Smith 2006). Perhaps one of the least successful of the Progressive reforms in terms of adoptions in the states was the idea of non‐partisan elections. Only Minnesota (until the early 1970s) and Nebraska have elected their state legislators without party labels, but the reform has gained much wider footing in local elections. Research on the effects of non‐partisanship show that, indeed, the simple rule of removing party labels from general election ballots has major consequences for voting decisions (increasing the value of incumbency, decreasing voter turnout, weakening the policy ties between representatives and constituents) (Schaffner, Streb, and Wright 2001), and in the structure of political conflict in the legislatures (Aldrich and Battista 2002; G. Wright and Schaffner 2002).
At the same time that the Progressives' reforms weakened the political parties, other state laws and court decisions worked to ensure the survival of the basic two‐party system. One such rule was the near universal adoption of single‐member plurality elections for legislative office. This “first‐past‐the post” system presents a strong obstacle to the success of minor parties. In addition, the parties are propped up by favorable ballot access laws, making it much easier for the two major parties to put up candidates than it is for independents or third parties. As a result, the parties have become what some refer to as “quasi‐public” entities (Bibby 2002, 21–2; Ware 2002, 89–90). The legal system lends a huge dose of stability in almost guaranteeing that the fundamental structure of political competition in the states will be between the Democratic and Republican parties. All of the incredible diversity of the states, and the conflicts that arise from this, must be accommodated within this relatively narrow competitive structure.
Political Scientists Back Reforms for Stronger Parties
As the Progressive reforms, the growth of civil service regulations, and Supreme Court rulings limiting patronage greatly weakened the parties, political scientists were arguing for change in the opposite direction. Beginning in the 1940s—and only sporadically challenged since—influential political scientists argued that stronger parties are better for democracy (Herring 1940; Key 1949; Schattschneider 1942). These scholars focused on the potential benefits of parties for providing voters with a clear choice, for bridging the problems inherent in a separation of powers system with powers divided even further by federalism, and for making the parties more responsive to citizens, particularly those denied participation in the existing one‐party states of the South (Key 1949). Academic arguments were converted to reformist calls for change in the report of the Committee on Political Parties of the American Political Science Association (1950).
This general perspective on parties has structured a great deal of the research since.1 The call for stronger parties came just at the dawn of the behavioral revolution with its goals of a more systematic and quantitative study of political processes. The assumption that stronger political parties are good is implicit through the great majority of research on state parties, although the efforts to demonstrate this position have been at best partially successful. What is remarkable is the virtual absence in the field of systematic efforts to demonstrate, or even test (p. 407) for, any kinds of ill effects of strong political parties.2 This perspective has certainly resulted in the illumination of some important aspects of party politics in the states, but it has also meant that some other, I believe increasingly important, aspects have not been as systematically incorporated into the ongoing research as they ought to be. My thesis is that research on parties, in the states as well as the nation, should attend at least equally to what the parties stand for as to their “strength” and competitiveness. For how else can we judge the impact of parties if we do not pay close attention to what they promise to do and what they actually do when in office?
Since The American Voter (Angus Campbell et al. 1960) surveys have provided the bread and butter data for the study of mass partisanship. Unfortunately, the cost of gathering comparative data on anything like a full set of the state electorates has been prohibitive. This is because the cost of a single‐state survey is not much less than that for a national survey. Such data collections would have allowed analyses of party in the electorate in the states to parallel the tremendous attention mass partisanship received for the national electorate.
Lacking survey‐based estimates of partisan preferences, scholars relied on that which is easily measurable: voting returns and election outcomes. The most enduring measure was developed by Austin Ranney (Ranney 1976) and incorporated election returns for governor and the percentages of each house of the state legislatures controlled by the parties. This index (and a number of variations) is simple to compute and provides the basis for classifications of the states as ranging from one‐party Democratic, through competitive, to one‐party Republican.3 The difficulty with the index for studying mass partisan tendencies is that it requires that we assume that which we often want to test, namely the relationship between party identification and voting outcomes. Indeed, it is often assumed that this is almost a one‐to‐one relationship in lower‐level contests like those for the state legislature, seemingly on the assumption that citizens probably know next to nothing about candidates for these offices. But in fact, there have been virtually no systematic (p. 408) studies of mass voting behavior for these contests, and certainly none done on a comparative basis.
By the 1980s enough national media polls had accumulated that researchers were able to patch together basic measures of partisanship in the states for comparative analyses. Erikson, Wright, and McIver (1993) used CBS–New York Times polls aggregated to the state level to achieve numbers of cases in the states that allow reliable estimates of state partisanship and ideology. Others have used the state‐based samples of the Senate Election Studies of the American National Election series (Norrander 2001), or the General Social Survey (Brace et al. 2002), and media exit polls (R. Jackson and Carsey 1999; Stein 1990; G. Wright and Berkman 1986) to gauge the partisanship and issue preferences of state electorates. The findings, even from these secondary uses of data collected for other purposes, have been informative. For example, just the incidence of identifying oneself as a Democrat or Republican varies as much across the states as does the direction of partisan preferences (Norrander 1989a) and whether or not people identify themselves as partisans is significantly influenced by state voter registration requirements: there are more people who say they are Democrats or Republicans in states that have party registration, but interestingly, there is also more defection from party in these states (Finkel and Scarrow 1985).
We have never had the surveys that would allow in‐depth comparative analyses of the processes of party identification and change in the states. However, exploiting the limited information available in media polls, Robert Brown's decomposition of the demographic characteristics of the state electorates showed that the social and economic compositions of the partisan coalitions in the states varied tremendously through the 1980s (R. Brown 1995). The state parties and their candidates seemed to adapt to their individual demographic terrains. Brown developed a typology of state party systems; it will be informative if future research will replicate those analyses to see if and how recent changes in partisanship in the states have altered the character of the demographic‐based classifications.
New technologies including Internet‐based surveys, and automated computer telephone interviews are collecting information on state‐based samples at reasonable costs. These will expand our ability to do comparative analyses of the partisanship and electoral decision making of state electorates. See, for example, the results of Survey USA or the data being collected by Polimetrics for the Cooperative Congressional Election Studies.4
The big story in partisanship at the national level has been the transition from what looked in the 1970s and 1980s like a period of “dealignment,” with the gradual disintegration of mass partisanship (Wattenberg 1994), to a period of partisan resurgence or reinvigoration (M. Brewer 2005; Hetherington 2001; Knuckey 2006). One aspect of the current system is the greater parity of the parties nationally. (p. 409) Of course, much of this has been fueled by the decline of the one‐party Democratic South. The overwhelming Democratic majority produced by the New Deal alignment is no more; rather we have a system of essential parity. Research drawing on media polls taken over a twenty‐five‐year period and aggregated into presidential administration time chunks at the state level allows us to chart changes in state partisanship and what moves it (Erikson, Wright, and McIver 2006). Looking at the individual states reinforces the general view based on regional analyses that most of the movement has been in relative growth of Republican identifications in the South. But these data also show a less dramatic but still significant growth in Democratic identifications in a number of northern states, adding to the images of the solid “blue” coastal voting we see in presidential elections.
The most important finding from the longitudinal analyses of state partisanship is that the causal agent driving changes in state party identifications is state ideology. At the aggregate level (as well as the individual level; Abramowitz and Saunders 2008; M. Brewer 2005; Hetherington 2001; Layman, Carsey, and Horowitz 2006) there has been a sharp increase over time in the correlations of partisanship and ideology. This means that today much more so than in earlier decades, party identification and ideological and policy preferences push voters in the same direction. This alignment of state partisanship and ideology has been a product of party identifications coming into alignment with relatively stable ideological orientations (Erikson, Wright, and McIver 2006).
At the foundational level of citizen political attachments, we see an important transformation of the state party systems. Whereas in the 1970 and early 1980s (and before) state partisanship and ideological preferences were virtually unrelated; today they are increasingly correlated, and thus reinforcing in their influence on most voters. This appears to be a good part of the explanation for perceptions of Americans dividing into sets of “red” and “blue” states. Ideology has been the driving force in these changes. The origins of this, of course, lie in the parties' efforts to win elections as well as changes in the motivations and preferences of the party elites (Carmines and Stimson 1989).
Transformed State Party Organizations: In Service of What?
Changes in the State Parties
There are few things that almost all researchers in an area agree upon, but the character of changes in state party organizations is one of them. The golden era of (p. 410) political parties has been dramatically changed. The strong party organizations of the Great Lakes, Middle Atlantic, and New England states which controlled nominations, elected officials, and had substantial patronage packages to dispense to supporters have largely vanished (Mayhew 1986). This led to the perception that the party organizations were all but dead, consistent with the growing consensus among party observers that the parties were dying; mass partisanship was declining and the state parties as forces in American politics had seen their day (Broder 1972; Wattenberg 1994).
The ideal of what Mayhew (1986) calls the “traditional party organizations” were clearly the exception rather than the rule by the time of his impressive survey of state party organizations of the late 1960s. Only a few of the states scored toward the top of his “TPO” scale, with the majority receiving the lowest score of a “1,” indicating no evidence at all of the traditional, hierarchical, nomination‐controlling, patronage‐based organizations. It was something of a surprise for many scholars whenCotter et al. (1984) reported a major resurgence of party organizations based on their surveys of state and county party officials. In contrast to the expectations of many that state party organizations were moribund, Cotter et al. found healthy levels of professionalism and staffing, and extensive campaign activities, for many of the state and county organizations. The parties were providing financial support to gubernatorial, congressional, and state legislative candidates in the vast majority of states, as well as help in matching appropriate PAC donors with nominees (Aldrich 2000; Reichley 1992).
The party organizations were transformed. They no longer controlled nominations, although a handful still mustered the ability to make pre‐primary endorsements, and the state parties made no pretense at controlling the actions of elected officials. Rather, their new activities earned them the label of “parties in service” of electing candidates who generally won election and re‐election on their own, but now received assistance in increasing measure from the state parties. However, this help in the vast majority of cases is clearly supplemental; very few candidates for the state legislatures depend on the state parties for the bulk of their support (Francia et al. 2003b). In addition, the official state party organizations provide linkage with the increasingly well‐funded national organizations which have provided a great deal of support for the state‐level organizations, especially for party‐building activities like registration and get‐out‐the‐vote (Morehouse and Jewell 2003a). State party organizations are active in providing electoral support to candidates in the states, frequently with significant funding from the national party organizations.
At the very least the strengthened, more active, more professionalized state party organizations would seem to be an important ingredient in moving the parties toward the goal of a stronger, more responsible party system. However, if we reflect on the function of strong parties, a key factor is missing. The contemporary party organizations have little say in the nomination process. Although some state (p. 411) parties produce policy platforms (Coffey 2007; Paddock 1998), because these are not made by, nor binding on, those who make policy, the effective “platform” is arguably defined by the campaign promises of the candidates winning the primaries. Thus, in order to assess whether stronger parties are really helpful to democratic governance, it would seem prudent to focus on what it is the stronger and more able parties want to do. That is, the formal party organizations are impressively active and professional, even smartly strategic in supporting their candidates, but in support of what?
State Party Ideology
This is where the huge and increased role of ideology in the transformation of state parties comes into play. Contributors and activists in the parties are no longer motivated by prospects of a job but by issues that touch on deeply held values. Wilson (1962) was one of the first to identify the ideologically motivated activist in contrast to the older traditional job‐seeking party regulars of the traditional party organizations. More generally, “pragmatists” who are primarily interested in winning elections have been replaced by “purists” (Wildavsky 1965) who are motivated by issues and ideology. The latter are more likely to be deeply committed and less willing to compromise in order to win (Layman, Carsey, and Horowitz 2006; Pomper 1999). Studies of campaign activists and convention delegates show the development of deep ideological polarization between the parties (Abramowitz, McGlennon, and Rapoport 1983; John S. Jackson, Brown, and Bositis 1982), mirroring that found in Congress (Poole and Rosenthal 1997).
One potentially fruitful avenue of research on state party organizations is to address the question of what determines the character of state party activists' preferences. Aldrich (1983a) developed an insightful model of how the ideological contours of party activists' communities may change. His model relies on the entrance and exit of activists with differing preferences. Some evidence of the processes he discusses can be seen in the “takeovers” of state party organizations. Wilson's (1962) depiction of the impact of the political clubs fits, as do more current efforts by the Christian Right in capturing the Republican Party apparatus in several states (J. Green, Rozell, and Wilcox 2000, 2003; Rozell and Wilcox 1995, 1997; Wilcox 1996, ch. 3). We know that state party activists have generally become more ideological, and a few efforts have been made to obtain descriptive information on the differing ideological or issue preferences of activists in the states. Some of these rely on surveys of convention delegates (Abramowitz and Stone 1984; John S. Jackson, Brown, and Bositis 1982; W. Miller, Jennings, and Farah 1986; W. Stone and Abramowitz 1983). Convention delegates have the advantage of providing a known universe for sampling and in some cases of getting that population together (p. 412) for the logistics of interviewing. However, they are limited because they are selected not as random samples of the states' party elites, but in presidential primaries and caucuses which, of course, are contests between alternative teams of activists within the parties committed to different candidates, frequently representing competing ideological camps within the parties. There is no guarantee that the state delegations attending the national conventions, are representative of the larger body of activists within the state parties. Biases are most likely to crop up on the Republican side, which still permits winner‐take‐all primaries and caucuses.
Others have used surveys of county or state party officials (Cotter et al. 1984; Uslaner and Weber 1984), while some have sought to use the readily available interest group ratings of members of Congress with various weights to measure “government ideology” (W. Berry et al. 1998). Erikson, Wright, and McIver (1993) used a combination of these as well as surveys of candidates to develop measures of state party elite ideology. Finally, another window into the ideological preferences of party activists is provided by the sometimes sporadic platforms of the state parties which can be content‐analyzed to scale the state parties (Coffey 2007; Elling 1979; Paddock 1992, 1998).
Erikson et al. find a weak relationship between state party elite ideology and the ideological preferences of the mass electorate, while Coffey (2007), using party platform measures of elite preferences, finds absolutely no relationship between the preferences of elites and the mass public. This difference may be a function of timing. With greater polarization of elites the parties may have become internally more homogeneous across the states, thus losing an anchor to the values of the mass public. If so, then the parties have become detached ideologically from the mooring of local values. Indeed, that is exactly the relationship implied by a system of national responsible parties. A second possibility is that these are measurement–methods differences. Paddock (1998) uses both measures of party committee members' issue preferences and scales derived from state party platforms and produces an important finding. He relates the levels of polarization to several factors and finds that states with histories of strong traditional party organizations continued in the 1990s to have less ideological parties. He also found some relationship to ideological differences in the mass electorate.
The relationship between elite and mass polarization is causally problematic. Work on party changes which link party elites and the public find that elites are the independent movers, with the public responding to what they see elites and elected official do (Carmines and Stimson 1989). Among the states, elite ideology has had a clear impact on party identifications of the states' electorates. As state party elite opinion gets more extreme relative to state opinion, that party loses identifiers (Erikson, Wright, and McIver 1993, ch. 6). This ought to provide a brake on the ideological polarization of the parties: as they polarize, they shed more and more of their base. The interesting thing is that in the era of party “purists” seemingly fewer of these party activists care enough about winning to seriously compromise on the issues which motivated their participation in the first place (Carsey et al. 2006).
What we do not have a good handle on yet is the relative contributions of national versus state party elites (and here I include elected officials) to changes in state partisanship. Clearly the trend has been toward greater polarization and this has been documented repeatedly for the national scene. A good research question is whether individual states can stem this tide for their electorates and the extent to which ideologically motivated party elites even want to field electorally more attractive and less ideologically extreme candidates.
Based on simple spatial theories of elections, parties ought to face a trade‐off in terms of ideological extremism and electoral success. Several have pointed out that the conservatism of the Republicans has provided a brake on their attaining a sustained majority (Hurley 1989; Stonecash, Brewer, and Mariani 2003). This tension, especially in an era where we see the parties polarizing, suggests what may be the prime question for scholars trying to discern the future of the parties and their role in American politics. What forces drive party elites (Carsey et al. 2006)? When does losing elections because of being out of touch with the median voter cause parties to moderate? We have fifty state party laboratories providing varying answers to these questions. What seems clear, however, is that the prospect of winning or losing alone is not enough; as we shall see below, systems with high levels of competition tend to be the most polarized, and it is clear that culture plays a substantial role. Systems that long ago were hierarchical, patronage‐based systems continue to produce ideologically muted versions of political combat in their states. Will this historical brake continue? And can we identify other factors that influence levels of party elite polarization that have evolved?
The study of party organizations in the post‐TPO era presents huge challenges. The party organization in the states is no longer performing all or even most of the roles of recruitment, nomination, electoral support, and party discipline of elected officials. The activities of the formal state party organizations are more supplemental than controlling. The effective “party” is now more accurately seen as a network of the formal party, allied interest groups, deep‐pocketed donors, issue activists, and legislative leadership PACs. This picture of the presidential parties (M. Cohen et al. 2008) applies to the state parties as well. The glue for the individual and highly substitutable parts then becomes shared policy goals as much as any overarching loyalty to the parties. And as such, the shifting salience of different policies and the circulation of activists as the issue agendas of the states change should be a profitable area for future research.
This research agenda for state parties scholars should proceed arm in arm with the parallel search for the factors influencing change in the national parties. Indeed, if there is merit to the idea that party elites in the states at some point feel the pressures of a trade‐off between ideological purity and winning, this in itself should provide a constraint on the national parties which continue to elect officials in contests defined in terms of local or state jurisdictions.
(p. 414) What Does Strong Party Government Achieve?
As mentioned above, the underlying and sometimes implicit theme of a great deal of the research on parties in the states is the familiar idea that strong, competitive parties are good both for getting politicians to attend to the interest of “have nots” (Key 1949) and for countering the power‐splintering forces of federalism and separation of powers. In addition, stronger, responsible parties are supposed to increase both government's accountability to the public and policy coherence (American Political Science Association 1950).
This general perspective on parties in the states has given rise to several lines of research. One of the most obvious has also been the most frustrating. This is the simple question of what difference it makes if Democrats or Republicans win office. The most frequently used measure of party control has been the Ranney index, discussed above, which provides an overall indicator of the extensiveness of party control of the governorship and state legislature. It would seem elementary that if parties are meaningful entities, and are fundamental to how voters control the broad contours of government policymaking (Angus Campbell et al. 1960; Erikson, MacKuen, and Stimson 2002), party control of government would have clear results for state policy. Initial forays to demonstrate this were not very successful. Thomas Dye (1966) used something of a shotgun approach in correlating party control and other factors across many measures of state spending, coming up, with few exceptions, with the message that party control did not matter. In another early but more focused effort, Winters (1976) found no impact of party control where it should have mattered the most, on redistributive policies. Several authors (R. Brown 1995; Dye 1984; Garand 1985; Jennings 1979) subsequently found modest support for the party control hypothesis by isolating those systems where the parties were especially polarized. Of course, this just backs up the question of why the parties were polarized in some states but not others. The diversity of state populations is part of the explanation (Aistrup 1993; Barrilleaux 1986; S. Patterson and Caldeira 1984), but the relationships have not been constant over time nor fully accounted for by combinations of environmental measures.
(p. 415) Interparty Competition
The frustration over the lack of party control effects was attributable in part to the odd situation of the southern states, which were at once solidly Democratic and also ideologically conservative. In fact, it was the observation of politics in the one‐party states of the South that produced the most enduring hypothesis of parties in state politics. Key (1949) noted that in the non‐ideological competition among party factions in these one‐party states the “have‐nots” were commonly left out. The fluidity of the factions prevented the disfranchised and poor from gaining any kind of electoral toehold in the system. Thus, Key hypothesized that if competition were organized around parties (presumably still non‐programmatic), then the parties would have an incentive to develop policies that appealed to the (poor) non‐voters who could supply the votes the parties' candidates would need to win. Thus, competition between organized parties was argued to benefit society's have‐nots.
An early article by Dawson and Robinson (1963) set the stage for an explosion of work in the late 1960s and 1970s that became the field of comparative state policy. However, work in the field dropped off dramatically by the 1980s.5 The early studies found that indeed states with greater interparty competition also had policies that favored the working class or poor, but this apparent relationship generally vanished when controls were brought into the analysis for such things as state income, urbanization, or industrialization. From a question focusing on Key's interparty competition hypothesis the field grew to address the more generic question of “does politics matter?” Repeatedly, the answer was “no” or, “maybe, just a bit under some circumstances.” In short, whether party control matters or levels of party competition matter “depends,” but in general, the conclusion was that the economic environment of the state political systems was a lot more important in explaining state policies.
For example, two important studies locate the impact of party control, but only after specifying both the liberal or conservative character of the parties and the condition of party competition. Plotnick and Winters (1990) show that the guarantees under the old Aid to Families with Dependent Children program were responsive to party control, but only when they tested for the statistical interaction of liberal–conservative party ideology and the competitiveness of the parties. Later, Barrilleaux and his collaborators (Barrilleaux 1997, 2000; Barrilleaux, Holbrook, and Langer 2002) used a measure of competition based on electoral competitiveness and reported that liberal parties in power do yield more generous (p. 416) policies for the poor when facing electoral competition. However, contrary to the Key hypothesis, their results indicate that Republican majorities faced with competition produce more conservative welfare policies.
These interactive effects do not appear to be consistent across studies, and more to the point, it is not clear why party elites would fall back to the values of their core supporters when they are in danger of losing office, but play to the median voter when they are electorally safe (Barrilleaux, Holbrook, and Langer 2002). Jackson (1992) did similar tests in looking at the interaction of public opinion and party competition (as well as several other “political” factors that might mediate how public opinion is translated into policy outcomes). He found very little evidence for such mediating effects, reinforcing the at best murky support to the long‐running Key hypothesis about the policy effects of party competition.
One difficulty that researchers have had in nailing down the exact effects of party control and party competition lies with failure to take state electorates into account. When the ideological preferences of the state electorates are brought into the analysis, we find that electorates simply do not allow the “experiment” of having a legislature that is very liberal or very conservative, given state opinion (Erikson, Wright, and McIver 1993). When state party elites and the candidates they nominate get too liberal or too conservative, their numbers in the mass electorate shrink. The impact of party control has been hard to substantiate because in state policy research we are, for the most part, comparing policies made by relatively moderate Republican regimes with those made by relatively moderate Democratic regimes. The obvious question is how this pattern of electing more moderate governments stands up in an age of increasing party polarization. Are the ranks of party moderates in the states being decimated in the same way as appears to be the case in Congress? And is electoral—or even legislative—competition the factor to keep such polarization of party elites in check? Those are important questions for future work.
Strong Parties in Government
The interest in responsible parties was evident in much of the early behavioral work on state legislatures, where one of the principal questions was the character of party voting in the state legislatures (Buchanan 1963; Derge 1958; Flinn 1964; Keefe 1954; LeBlanc 1969; Sorauf 1963). They found remarkable variation in the levels of party voting. Party conflict tends to be higher in more urban states, but the striking feature even among these is the wide variation in the centrality of party in roll call voting (Dye 1965; Keefe 1956; Lockard 1959; Zeller 1954). One of the main findings of this early research was that party voting increases with the competitiveness of the parties (Broach 1972; Key 1956; S. Patterson 1962; S. Welch and Carlson 1973), (p. 417) although LeBlanc (1969) found little relationship between party voting and legislative competitiveness in the twenty‐six chambers he studied. More recent work indicates that higher levels of party voting are associated with, if not causally connected to, levels of competitiveness of the parties in the state legislatures (Aldrich and Battista 2002; S. Jenkins 2006; G. Wright and Schaffner 2002).
There was a sharp drop‐off in studies of state legislative roll call voting beginning in the 1970s. This coincided with the Inter‐University Consortium for Political and Social Research making roll call votes of the US Congress available to scholars. Getting sets of roll calls already collected, coded, documented, and ready for analysis is much easier than traipsing to state capitols and copying roll calls from legislative journals and then getting them keypunched using the technology of the day. Many of the early efforts were generally motivated by behavioral assumptions about the importance of attitudes and group pressures (rather than institutions and rules) and concerns for responsible party government rather than state politics or parties per se. Hence, analyzing the more readily available congressional roll calls to deal with such concerns was an understandable choice and helps to explain the dramatic drop in party–roll call studies in the states.
In the congressional literature questions about the role of party in roll call voting decisions evolved from one of gauging degrees of loyalty and cohesion to the effects of leadership and rules in bringing about those patterns (see, for example, the discussions in McCarty, Poole, and Rosenthal 2001; Snyder and Groseclose 2001). The problem came to be framed as one of “partisanship,” meaning party leaders' influence over members' votes, versus simple “preferenceship,” in which levels of party loyalty reflect the coincidence of shared interests but no special efforts on the part of party leaders.6 From the perspective of state politics or even that of advocates of responsible parties, the controversy is less central than its prominence in recent congressional scholarship. In fact, two studies have gone to the heart of the problem faced by congressional scholars, which is that roll call votes were used to measure both the preferences of members of Congress (ideal points) and the effects of party (measures of cohesion, party voting, or polarization). Using independent measures of attitudes toward the parties and members' ideological preferences, these studies showed that, not surprisingly, members' attitudes are important (Scully and Patterson 2001), as are the powers of the majority party leaders and some constituency factors (S. Jenkins 2008).
Two points should be remembered in thinking about this as a research question on the role of parties in state legislatures. First, the assumption that party matters (p. 418) only if it can be demonstrated that members are persuaded or influenced in their voting by party leaders is misleadingly narrow. In fact, comparing the non‐partisan Nebraska legislature with the similar but partisan Kansas Senate shows that having partisan elections and organizing the chambers by party has a major influence on the structure of conflict. Parties, when in play as alternative governing teams, lead to the bundling of issues and reduce the dimensionality of conflict as evidenced in roll call voting (Aldrich and Battista 2002; G. Wright and Schaffner 2002).
Second, the advocates of the pro‐party effects position are really positing that strong parties distort policy outcomes from the preferences of the median legislator (which presumably would be the optimally representative outcome for the polity). For example, Cox and McCubbins (1993) argue that strong majority parties will seek to pass legislation at the ideal point of the median majority party legislator. Thus, a hypothesis for further research is whether responsible party government increases the likelihood that policy will be unrepresentative of the preferences of the median citizen. When the parties are not so far apart, as in the 1950s and 1960s, this probably is not a great concern; it would entail movement a bit this way or that depending on the parties' recent electoral fates. However, when the parties are ideologically polarized, such majority power would appear to be a recipe for clearly unrepresentative policy outcomes.
Future work on party government in the state legislatures will be able to take advantage of significant new data resources that are becoming available.7 These data collection efforts will enable significant expansion of questions about parties in the legislatures while adding crucial analytic leverage of comparative analysis across the states. These opportunities will hopefully prompt researchers to take a more complete accounting of the effects of party than simply whether legislative leaders influence their members' behavior. A full understanding and evaluation of party effects should look at the impact of party in the mass public, certainly among activists through the nomination processes as these filter who even gets into the legislatures. In terms of the controversy among congressional scholars, the comparative perspective of the states suggests if the nomination process is sufficiently efficient in yielding ideologically distinct sets of party nominees, there would be little need for arm twisting or major agenda manipulation by legislative leaders. (p. 419) Party effects may well be pervasive, as in the quote by Bibby and Holbrook at the outset, even in a legislative world of apparent pure preference voting.
We have had an interesting pair of contradictory trends in party control of the states governments—contradictory at least from the perspective of strong responsible parties. On the one hand, as we have noted, the parties appear to be more ideological today than was the case fifty or sixty years ago. That is clearly a step toward responsible party government. However, headed in the other direction over this period is the decline of unified government wherein one party clearly controls the executive and legislative branches, thus allowing the electorate to know which party to hold accountable for how things are going (Fiorina 1996). The trend toward divided government has been evident at both the state and national levels. We know that most of the instances of divided government stem from the success of minority party governors. This appears to be a reflection of the strong relationships between state partisanship and the partisan composition of the legislatures on the one hand, and the ability of governors to develop a “personal vote” less tied to state partisanship (Erikson, Wright, and McIver 1993, 128–30, 194–5).
An open research question is whether these patterns of divided government have continued. It would be reasonable to entertain the hypothesis that the growing polarization of American political parties reaches beyond Congress and the division of presidential voting patterns into “red” and “blue” state governments. This returns us to the basic question of what the parties stand for in an era of party polarization; how much homogenization has occurred, and do the centripetal forces pulling the parties apart from one another affect governors?8
One important question is whether it matters that government is unified. The answer is an unequivocal yes from the perspective of responsible party government, and indeed, Sundquist (1988) argued that the rise of divided government has created something of a theoretical crisis for parties scholars. In the states there is evidence that the governor clearly has more difficulty passing his or her legislation when the governor's party does not control the legislature, especially when the opposition has a majority in both chambers (Bowling and Ferguson 2001; Clarke 1998), and similarly, that changes in the expected directions of fiscal policy are more consistent with unified government in the states (Alt and Lowry 1994). Perhaps most surprising (p. 420) concerning divided government in the states is Squire's (1993) discovery that citizens in states with divided government are significantly more content with their government than citizens living in states with unified government. One hypothesis, heretical from the responsible parties viewpoint, is that citizens prefer divided government because it produces policies offensive to fewer people than governments under unified control of parties that are clearly too liberal or too conservative for their average citizens. A similar challenge to the advocate of responsible parties comes from Daniel Shea (2003), who points out that the polarization of the current era has been accompanied by growing public alienation from government and high levels of disapproval of Congress in particular.
Policy gridlock is one of the problems parties scholars attribute to divided government, but this may not be much of a possibility for the important issues of state government: budgets have to be passed, and while much legislation no doubt faces rougher waters under divided control, we have few instances of government shutting down. In short, as Fiorina observed (1991), there is really no theoretical necessity that policies under divided governments will be inherently less desirable for the citizenry than those made by unified governments.
Where Do We Go from Here?
One of the challenges I mentioned was the complexity of state party systems. Much past research has sought to correlate variations in the “strength” of the parties in the electorate, the party organizations, and in government. An appropriate direction for future work may be to incorporate more systematically the ways in which ideological preferences tie the components of the parties together. Can strong ideological party networks successfully promote candidates without also winning over major portions of the mass electorate? And reflecting on the work on Congress, we should address the extent to which party polarization in government is endogenous, a function of rules and leadership powers, versus environmentally determined as a function of mass and party elite preferences, or even the cultural traditions of the states. Finally, can we find an equilibrium in the process of polarizing parties? Specifically, under what conditions do such factors as losing elections, or efforts to maintain majority status, lead to greater party moderation?
The second major challenge in studying the state party systems lies in the very heavy data demands for systematic comparative analyses. Fortunately, it appears that the future has some good news on this front. New polling technologies are likely to make increasing amounts of survey data with large state samples available to scholars. As these resources come on board, scholars will be able to delve deeper (p. 421) into the relationships between citizen partisan attachments on the one hand, and their values and how they perceive both the state and national parties (and their candidates) on the other. Ultimately, what happens to the state parties is rooted in the attachments of their mass electorates.
Party organizations are more diverse. As scholars, we should probably talk about state party networks and work to chart these rather than focus on the formal organizations. In doing so, we need to attend carefully to the currents of ideological change and cleavage, both within and between the parties. Extant work seems to establish that the main motors of change in American politics are the activists and contributors whose efforts frequently make or break candidates of different policy persuasions.
Policies are made by governments, and in responsible party systems these should reflect the preferences (and promises) of the majority party. Here too new resources should be available with much greater depth to legislative voting records, in committees and floor roll call votes, as well as better measures of state policy. These resources will allow scholars to gauge both what it is the parties want to do as well as whether they actually deliver responsibly.
The point is that to assess responsible party government in the states we need improved measures of party in the electorate, the party networks (as organization), and party in government and the policies they enact. At that point we should be able to provide a solid evaluation of the normative promise of strong competitive parties.
There are several reasons to expect that such conclusions might be mixed. Stronger competitive parties are clearly associated with some positive outcomes. One of the most consistent is voter turnout, which is consistently higher in more competitive contests (Grofman, Collet, and Griffin 1998; Hofstetter 1973; S. Patterson and Caldeira 1984). We also know that party voting is higher in chambers where there is near parity between the parties and that these systems are associated overall with more liberal policies. However, the analyses of the independent impact of the party system over and above the effects of state wealth, education, and particularly public opinion have been inconsistent and often non‐existent across studies (Treadway 1985).
One of the remaining challenges is revisiting our thinking about how competition affects parties, candidates, and policymakers. The Key hypothesis presumed non‐programmatic parties which were willing to adapt policy to the end of winning elections. That means making policy appeals to the non‐voters who are disproportionately have‐nots. But this logic needs to be amended in an era of highly programmatic parties where policy concerns are at least as much a driving force for activists and candidates as is winning. We have seen repeatedly that competition does not yield policy convergence; if anything as the parties have gotten more competitive we have seen polarization, although the causal connection here is far from established.
Another puzzle is the seeming contradiction between strong programmatic party policy platforms on the one hand and the fact that optimal policy should match well with the preferences of the great bulk of the population whose preferences are decidedly more moderate than the ideologues who define what the parties stand for on the other hand (Poole and Rosenthal 1984). In this situation voters have a clearer choice, but for the average citizen their ideal is not on the ballot. This by itself should form the basis for a strong argument that ever stronger, more cohesive parties are not an unmitigated good (John Coleman 1994).
The future for state parties research is bright with the promise of new data resources which will permit stronger and more definitive tests of theories of party and citizen behavior. But we are still dealing with concepts developed in reaction to the traditional party organizations of an earlier era. Either rethinking those or developing entirely new ideas is a necessary step to significantly improving our understanding of why the state party systems are what they are and what they are likely to become.
(1) There were clearly dissenters from the advocacy of the responsible parties model, but as often because they were deemed unworkable in the US as that they were deemed undesirable (E. Kirkpatrick 1971; M. S. Steadman and Sonthoff 1951; Ranney 1951; Turner 1951).
(2) An exception to this statement are reactions of political scientists to the alarms of some observers concerning the heightened levels of party polarization. See, for example, the essays in Nivola and Brady 2008.
(3) This index is frequently “folded” to make a measure of interparty competition running from very competitive to safe for one of the parties.
(5) This is a severely abbreviated account of the development of a rather substantial literature. Treadway (1985), for example, devotes an entire book to a review of the comparative state policy studies. See also the surveys by Brace and Jewett (1995) and Stonecash (1996). Throughout, the hypothesis of the effects of interparty competition on state policy is prominent.
(6) This controversy pits Keith Krehbiel's (1993, 1998, 2000) position that a simple model of members pursuing their own preferences is able to account for most patterns of congressional behavior, including roll call voting. Seeking to demonstrate the importance of party in Congress, an array of scholars have used different datasets and methods (Ansolabehere, Snyder, and Stewart 2001a; Binder, Lawrence, and Maltzman 1999; McCarty, Poole, and Rosenthal 2001; S. Smith 2007).
(7) These efforts include the Election Dynamic Project, which includes roll calls and interviews with legislators from five states over several sessions, 1992–6 (see S. Jenkins 2006 for a project description); Gerald Wright's Representation in America's Legislatures Project, which has collected comprehensive sets of roll calls for all the state legislatures for two sessions, 1999–2000 and 2003–4 (<http://www.indiana.edu/~ral>); and a parallel project by Nolan McCarty and his collaborators. They have collected roll calls over more sessions but include fewer states. The goal of their effort is to achieve a common scale for legislator ideal points across legislative chambers (<http://www.indiana.edu/~ral>) and a parallel project by Nolan McCarty and his collaborators. To date the major goal in that project has been to construct a common scale for legislator ideal points across legislative chambers (Shor 2009; Shor, Berry, and McCarty 2007).
(8) If the polarization of the parties has only a weak basis in the state publics, then we might find more divided government as voters confront party choices between too liberal and too conservative. On the other hand, if there is a strong mass basis to the polarization, the clearer brand labels would make it more difficult for minority party governors to win, as Fiorina pointed out happened with considerable frequency in places like Utah and South Dakota (Fiorina 1991).