Chapter I:        Introduction

         In The Semisovreign People, E.E.  Schattschneider argued against pluralist and conventional-wisdom notions of democracy as ‘government by the people’.  Schattschneider thought this idea reflected a naive belief that democracy entails a kind of  “spontaneous generation” of knowledge and interest in politics  “at the grassroots” (Schattschneider 1960, 138).  This “amazingly rigid and uninventive” understanding of democracy, he argued, was not only simplistic but resulted in policies for democratic reform that in the end are bad for democracy, especially the idea that democracy requires “making a great effort to educate everyone to the point where they know enough to make [democratic] decisions” (136).  Such a view, for Schattschneider, sets up norms and institutions that frustrate the democratic impulses of the people since they become insecure about their qualifications for participation, and institutions that can easily command participation (e.g. parties) are deprecated in favor of the ideal of the independent voter.  As an alternative, Schattschneider suggested that democracy happens through leadership and institutions that shape conflict in such a way that people are mobilized into politics.  This mobilization, for Schattschneider, is essential, for people cannot act politically in the way assumed by those who pine for a more rational and active grassroots citizenry.  As he put it, “[w]hat 180 million people can do spontaneously, on their own initiative, is not much more than a locomotive can do without rails” (139).  The scope of politics in a pluralist mass democracy simply makes the conventional picture of democracy  unworkable.
         The conception of mobilization here is especially important.  Schattschneider’s conception of democracy and mobilization is directly opposed to pluralist or any democratic theory that places some aggregation mechanism at the center of democracy.  Democracy, for Schattschneider, is not a product of appropriately registering democratic impulses.  Instead, it is a dialectical process by which the political values of the citizenry and democratic institutions are constantly constructing one another. In this argument pluralists draw the brunt of the criticism.  These theorists, Schattschneider believed, were improperly applying the causal models of  physics to political phenomenon by specifying public opinion as a force acting on institutions when the relationship is actually more the reverse (36-40).  As a result of this mistake, pluralists have focused on the relative strength of various political actors and the openness of the political system to various influences rather than how the entire system is structured to mobilize some kinds of participation and ideas and demobilize others.
         Schattschneider’s conception of mobilization is especially interesting and important to the theoretical view of this dissertation.  When Schattschneider talks about mobilization he uses the phrase “the mobilization of bias”.   This phrase has often been reduced to the argument that pluralists do not take into account the phenomenon of “non-decisions” (c.f. Gaventa 1980).  But I think it can be taken more broadly to embody a constructivist theory of politics.  For Schattschneider does use this term most noticeably with reference to the way the pressure group system in American politics is biased towards the interests of corporate elites, but he also applies the term to the mobilization of the working class and poor by the Democratic party.  For Schattschneider, “organization is itself a mobilization of bias” (30).  Thus the bias of organization is not pejorative per se.  It instead  embodies the fact, for Schattschneider, that democratic impulses are always already transmitted through institutions that themselves mobilize participation through inculcating or exploiting some kind of bias.  The term ‘bias’ is, therefore, a frank challenge to normative theories that try to eliminate bias from the political system in favor of rationality.
 Schattschneider argues that prejudice against bias in American politics amounts to a preference in the United States for pressure-group politics instead of party politics (20-36).  Pressure group politics is more rational, Schattschneider would acknowledge, since it mobilizes people through the rational bias of private self interest.   But it is not optimum for democracy since the pressure-group system disproportionately registers the preferences of economic and social elites.  Problems of collective action make mobilizing the majority of citizens on the basis of rational self interest impossible.  These citizens, however, can be mobilized by the non-rational biases exploited by political parties, biases of place, religion, ethnicity, class and race.  For this reason Schattschneider prefers party politics with the proviso that the party system must have at least two parties that are competitive across the entire electoral arena.  Competitive elections encourage parties to mobilize participation, and they make sure that the electorate is at least offered a clear set of alternatives in elections.  This kind of competition ensures that a collectively rational outcome is reached even if the mobilization of individual citizens is accomplished through non-rational biases.  For Schattschneider this is what happened during the New Deal when the Democratic party was able to nationalize political conflict through the mobilization of class, ethnic and racial biases (chapter V).
         The moral I take from Schattschneider is that democratic politics requires a mobilization of bias.  This claim establishes a fascinating paradox for democratic theory that is the central preoccupation of this dissertation: a rationalized politics produces irrational politics.  By this I mean that a politics which demands that each citizen act like a independent, objective, and rational consumer of public policy will be irrational since it facilitates the political  demobilization of the citizenry that is essential for democratic politics.  Behind this paradox is the fact that only the mobilization of bias can overcome the collective action problems that stymie mass participation, and only the mobilization of bias can create categories and political ideas through which democratic discourse can happen.  What this means in practice, again echoing Schattschneider, is democracy can only work through the institutions of political parties that produce a mobilization of bias that political scientists call partisan identity.   What this dissertation explores is the possibility that the nature of partisan identity has been changed, changed because various forces have produced a political system that favors the construction of institutions and citizens according to the ideal of rational participation embodied by the pluralist model rather than the mobilization of bias of previous party systems.
       Central to this argument is the decline of party-centered elections and the rise of candidate-centered elections as a political system which favors the ‘rational’ consideration of individual candidate’s promises rather than the mobilization of bias through partisanship.  Candidate-centered elections have given candidates tremendous freedom in presenting themselves to the electorate.  Since party organizations have evolved to simply serve the needs of candidates rather than being an independent programmatic voice, there is no institutional check on candidates' ability to campaign according to their own political vision.  Only the ideological or affective inertia of the parties' constituents constrains presidential candidates from reconstituting the party around their own programs and personalities.  But even these attitudes may, in fact, be substantially molded by candidates given the array of media technologies that contemporary campaigns have at their disposal.  Such technologies may dominate the information environment to the extent that candidates have significant power to control how the electorate evaluates them (Orren and Mayer 1990; Arterton 1993; Popkin 1991; Jamieson 1992).
         This picture of American politics represents an important departure from the relatively party-centered eras before the 1960s, in particular the era of the New Deal which Schattschneider looked upon as a progressive development.  Of course romance for the past must be tempered, but in many ways Schattschneider was right in claiming that the United States’ old party system more effectively served the interests of democracy.  In this vein several scholars have commented on the party-centered system’s superior ability to connect people to democratic institutions in ways that better mobilized citizen participation, interest, and consent.  Political parties are the heroes of these stories, as they are the institutions which did the mobilizing.  The local and parochial nature of this political system may have fostered corruption, but that same parochialism provided a kind of communitarian support for the construction and exercise of citizenship that is lacking today.  The quality of this citizenship necessarily rested more on the cohesiveness of ethnic, religious and cultural affiliations than rational voting or administration, but these affiliations facilitated participation and provided a more stable basis for broader coalition building than we have in the present candidate-centered era (Kayden and Mahe 1985; Ehrenhalt 1995; Hays 1980; Burnham 1970; Key 1949; Aldrich 1995)
         There might be hyperbole in these descriptions, but there is enough truth to these journalistic intuitions to suggest the need to dig deeper into the implications of candidate-centered campaigns for American democracy.  Candidate-centered politics might, in fact, be a good development if it makes our democracy more responsive to public policy problems.  A political system that is more responsive and permeable to the development and exercise of political leadership could produce a government that is more rational and efficient.   But it also might be less felicitous if candidates use their newfound freedom and capabilities to manipulate political discourse for their own interests rather than in the interest of the public.  For instance, candidate-centered elections might erode the stable political values that are indispensable for representative government.  As a result, public opinion might become little more than a will  o’ wisp lacking in any substance that could produce public policy outcomes other than frustration and apathy.
         This is a potentially dangerous development, but the extent to which the change to media-intensive candidate-centered elections has had an impact on American politics is unclear at this point.  An effect is often assumed in arguments that trace presidential or congressional failures to the lack of stable governing coalitions formed by responsible parties (e.g. Lowi 1985; Burnham 1970; Wattenberg 1991; Pomper 1992), but if one turns to the voting behavior literature for confirmation of this intuition there is a puzzling lack of evidence.  An important recent book, Issue Evolution, argues that politicians drive changes in the way voters evaluate politics (Carmines and Stimson 1989).  Similarly, there is a growing emphasis on the role of strategic politicians and candidacy in shaping democratic discourse (Fowler 1993).  And there is much discussion about the decline of party-in-the-electorate and the fact of candidate-centered elections (Wattenberg 1994; 1991).  But evidence of a direct tie between voting behavior and candidate-centered elections is limited.
         Several studies, however, do point to data that hint candidate-centered elections are having interesting effects on voting behavior,  effects which suggest that understanding the changing character of partisanship is the key to understanding the current candidate-centered system. Jacobson (1991), for instance, argues that congressional elections have become much more volatile recently because voters are less tied to their party identifications.  This is supported by Fiorina (1994) who finds that the importance of partisan considerations in congressional elections has declined as of late.  But Fiorina also finds that the importance of partisanship in presidential elections has increased recently.  How could the effect of partisanship be increased by candidate-centered campaigns at one level but decreased at another?
         Such an effect directs me to consider the possibility that partisanship in candidate-centered campaigns is still central because of  another intriguing paradox: partisanship might be of continuing importance in political discourse precisely because it is less important in social life, as a social fact. The decline of parties as institutions means voters no longer have the close social ties to the parties that have established deep affective partisanship in the past.  This has opened up the party system, perhaps even increasing the importance of parties since discourse concerning them is no longer confined and defined by the more local knowledges that constituted discourse about parties in the past.  It is easy to see how candidate-centered campaigns may have fostered this change.  The decline of party identification and the rise of candidate-centered elections is ironically the impetus to the invigoration of a new kind of highly competitive media-based politics in which parties take on new roles: the effective manipulation of partisanship and partisan symbols (Schlessinger 1985).
         Partisanship is still important in this party system; however, it lacks a connection with the salient social identities that defined partisanship in the old party-centered system. The implication of this change in the social bases of partisanship certainly must be that the character of partisanship has changed as well.  The psychological mechanisms through which partisanship operates may be such that that it is now more responsive to the political environment.  Partisanship that acts less like a closely held social identity may be less stable since voters seek new sources of information to make rational decisions at the ballot box rather than partisan solidarity.
         Clearly what is needed to explore these ideas is research that clarifies the role of partisanship in American politics.  This will be the goal of this dissertation.  My central research question asks whether the character and  stability of partisanship has decreased with the advent of candidate-centered campaigns.  But this question will be pursued from somewhat outside the dominant traditions in  the study of American political behavior: “Michigan School” or mainline voting behavior and political cognition.  While they have deeply influenced my own work, I argue that these perspectives suffer from flaws that limit their ability to adequately describe the changing nature of partisanship.  Mainline voting behavior, for its part, has ignored the contention from political cognition scholars  that their arguments must be firmly grounded in psychologically realist theories of cognition (Lodge, Stroh and Wahlke 1990).  They have, therefore, focused too narrowly on the American National Election Studies (ANES) party identification conception of partisanship and voting models that rest on unsubstantiated ontological commitments.  Political cognition researchers, on the other hand, have been too content to develop individual-level generalizations in largely experimental conditions.  They have developed refined conceptual tools from careful research into how people actually think, but their (so far unmet) challenge is to "integrate the now-emerging constructivist view of political cognition with the aggregate regularities that are observed in real voting behavior (Rahn 1993: 63)."  The result has been theories which conceptualize voters more as computers than as citizens.  The dominant theory of the political cognition school, the information processing model, has emphasized that conceptualizing voters as cognitive misers is the key to understanding behavior.  I will argue, following the social identity approach developed by John Turner and his colleagues (Tuner 1987; Oakes, Haslam and Turner 1994), that this approach ignores the degree to which cognition is driven as much by the interaction  between social identity and the political environment as the computational limitations and hard-wired heuristics of the mind.
 The approach I follow in the dissertation tries to avoid the problems associated with mainline voting behavior and political cognition while drawing heavily on their contributions to voting behavior research.  Specifically, I follow some researchers in political cognition by reconceptualizing partisanship as party stereotypes.  While party identification emphasizes the long-term affective predisposition towards a political party (Miller 1976), the party stereotype concept connotes a wider view of partisanship.  It defines partisanship by the knowledge a person holds about the parties as well as the affect associated with parties. In this way party stereotypes may be more sensitive to partisan change, as the contents of a person’s stereotypes of the parties may change while their long-term affect may remain constant.  Such sensitivity should make inquiry into party stereotypes more appropriate for exploring the changing construction of partisanship associated with candidate-centered campaigns.
         Stereotype theory will prove to be especially appropriate for detecting the relationship between context and cognition when supplemented by the social identity approach of Turner and his colleagues.  The social identity approach to stereotyping emphasizes that party stereotypes are a product of the relationship between the political environment and the social identities that constitute a person’s partisanship.  In this emphasis on social identity my approach could be considered as a return to the sociological model of the Columbia studies (Lazarsfeld, Berelson, and Gaudet 1948).   This means that the theory provides us a diagnostic tool for exploring the changing meaning and character of partisanship.  If stereotype changes are evidence that the political environment is cueing the social identities that constitute partisanship in a new way, then tracking these changes gives us an indication of which social identities are helping to construct the party system.
         This tool could thus be useful for exploring the impact of candidate-centered campaigns.  In a party-centered system, party stereotypes should be strongly connected to the ethnic, religious and cultural identities that are mobilized by local party organizations.  Candidate-centered elections, on the other hand,  should signal a loosening of those ties to local identities, either introducing new cosmopolitan bases for partisan identity or leaving partisanship more subject to the political environment.  This dissertation explores these possibilities.
         This inquiry proceeds in chapter II with an expanded argument for the appropriateness of the party stereotype conception of partisanship.  This argument takes the form of an extended critique of partisanship theories as presented in the mainline voting behavior and political cognition approaches.  The chapter emphasizes the epistemological commitments that underlie each approach and argues that the party stereotype approach rests on firmer theoretical grounds.  The chapter also reviews the development of stereotype theory in social psychology to support the conception of party stereotypes.  Chapter III sets out  a discussion and defense of the self categorization approach to stereotype research and justification for the data and methods used in operationalizing party stereotypes. Chapter IV employs a cluster analysis of the party likes and dislikes questions in ANES cross-sectional data to explore the contents of party stereotypes in presidential elections from 1952 to 1992.  The chapter describes the broad changes in the character and contents of party  stereotypes, and it demonstrates how party stereotypes  reflect the dominance of the New Deal in the 1950s and 1960s, the decline of the New Deal in the 1970s and 1980s, the rise of political Independence, and the development of new party stereotypes, including those emphasizing social issues,  in the 1980s and 1990s.  The cluster analysis elucidates the degree of instability in partisan stereotypes and the degree to which party stereotypes have become separated from party identification.  Chapter V develops interactive regression models to explore the shifting bases of partisanship in social identities suggested by the self categorization model of stereotyping.  The chapter argues that while stereotypes in the 1950s were connected with the social identities that made up the New Deal party system, party stereotypes do not exhibit such connections after 1960.  These findings lead me to question whether contemporary partisanship is really an identity at all in the sense it was when the concept was first developed.  Chapter VI explores the possibility that the changing constitution of partisan identity has resulted in less stable party stereotypes in recent years.  This analysis concludes that contemporary party stereotypes are less stable than in the 1950s, but new narratives about parties provide a basis for party stereotype stability that may or may not herald problems for democratic theory.