Page 1Page 2Page 3Page 4Page 5Page 6Page 7Page 8Page 9Page 10Page 11Page 12Page 13Page 14Chapter 1CRITICAL ELECTIONS, REALIGNMENTS,AND PARTY SYSTEMSIt would be a serious exaggeration to suggest that what we willterm the "realignment perspective" dominates the study ofAmerican political history. However, since the appearance ofV. 0. Key's seminal article in 1955, an extensive literature hasappeared concerned in one way or another with critical elec-tions, partisan realignments, party systems, and related politicalphenomena and processes.' The terminology of this perspective,at least, has penetrated the accounts of traditional historians, aswell as the works of the more scientifically inclined of variousdisciplines, and it is also to be found in the writings of journal-ists and other political commentators. To a growing degree,moreover, political history is periodized, sometimes almost con-ventionally, in terms of party systems, or realignment eras. Therealignment perspective, in short, has attained something of the1920PARTISAN REALIGNMENTElections, Realignments, and Party Systems21status of an organizing or synthesizing framework for the studyand discussion of the American political past.Our task in this chapter is to develop a comprehensive andextended statement of the realignment perspective as a startingpoint for empirical investigation. Our effort is to state theperspective in a form that encompasses partisan control of theagencies of government, the behavior of partisan elites, and thepolicy products of government as well as the behavior andattitudes of the electorate. To do so, however, it 4is first neces-sary to consider relevant work in some detail both to identifyelements of the realignment perspective that are often eitherleft implicit or neglected and to suggest areas of neededresearch.THE REALIGNMENT PERSPECTIVEDiscussions of the realignment phenomenon often and appro-priately begin with V. 0. Key's article, "A Theory of CriticalElections." Key identified one category of election which hedescribed in an often quoted passage as "an election type inwhich the depth and intensity of electoral involvement are hiin which more or less profound readjustments occur in therelations of power within the community, and in which newand durable electoral groupings are formed."' Angus Campbellextended Key's conceptualization to include "maintaining" and:"deviating" elections and later added "reinstating" elections tothe scheme.3 Gerald Pomper modified the classification toallow for realigning chAnge that benefited the already dominantparty, an election type he labeled "converting."'Campbell's further contribution is less often noted. Heexplicitly linked this classification of elections to the social-psychological model of individual partisan attitudes and behav-ior developed in The American Voter and to the "normal vote"concept subsequently developed by Philip E. Converse.' In thisway, he both extended the analytical utility of the classificationand enriched its theoretical basis. It may be that a further andinadvertent consequence was to focus historical analyses evenmore narrowly upon realigning elections and periods to therelative neglect of other elections. While realigning electionsshifted both the balance of partisan identifications within theelectorate and partisan control of the agencies of government,maintaining, deviating, and reinstating elections did not. Hencethe latter categories of elections could be seen as intrinsicallyless interesting and indeed less important than realigning elec-tions.Whatever its specific source, interest in critical elections andpartisan realignment has led investigators in several directions.One endeavor has involved establishing the timing of realign-ments—that is, deciding which elections are critical—and hasproduced substantial disagreement. Virtually all analysts agreethat sometime between 1920 and 1940 a realigning electiontook place, but each presidential election from 1924 to 1936has been designated as the realigning election.6 Similar disagree-ments surround the elections of the Civil War era and of the1890s.7 Much of this discussion has occurred without clearconceptualizations of the realignment process, clear ideas ofrealigning electoral change, or an effective means to assess andmeasure such change. We address the latter two issues in thefollowing two chapters, while the first is in certain respects acentral topic of our entire study.One critical conceptual and empirical difficulty in researchinto historical electoral behavior, a source of disagreementamong investigators, is the task of inferring the underlyingdistribution of partisan loyalties—and shifts in that distribu-tion—on the basis of the actual vote. Realignments are seen asproducing new patterns of partisan loyalties and as ushering inperiods of relative electoral stability based upon a new distribu-,.tion or alignment of underlying loyalties. As the classification'of elections suggests, however, complete consistency in thepartisan distribntion of the actual yule i§iidrattraFiereiiirdperioa-Dt-stability-;-dintShiftr mTactirar-vo'reripartieularielections are--not_n.ecessarLlyindications6? lasting change inkpartisan identifications within the eleifol'aTE—ChirigiiTthedistributiiin—o-rMEWiii-61E-ei: -WORM: occurs both duringsupposedly stable periods as well as during realignments, and a22PARTISAN REALIGNMENTElections, Realignments, and Party Systems23critical task is to determine which of these changes in the actualvote are reflections of change in the underlying distribution ofelectoral loyalties..To accomplish this task, a seemingly straightforward proce-dure has often been followed. Elections in which the distribu-tion of the actual vote departs from the distribution of the votein preceding elections but resembles the distribution in follow-Mg elections are seen as realigning elections and as involvingchange in the basic partisan loyalties of the electorate. Theimplications of this operational definition of realigning electoralchange introduce both theoretical and empirical tensions whichare not as yet well reconciled. This definition implicitly identi-fies voting behavior with partisan attitudes and treats change inthe former as a direct indication of change in the latter, despiterecognition in other contexts that the partisan voting behaviorLAA- individuals often does not coincide with their partisan iden-tifications. The implication of