Read an Excerpt
The History of American Electoral Behavior
By Joel H. Silbey, Allan G. Bogue, William H. Flanigan
PRINCETON UNIVERSITY PRESSCopyright © 1978 The Center for Advanced Study in the Behavioral Sciences, Stanford, California
All rights reserved.
Partisan Realignment: A Systemic Perspective
WALTER DEAN BURNHAM, JEROME M. CLUBB, AND WILLIAM H. FLANIGAN
During recent decades major progress has been made toward improved knowledge of American mass political behavior. Intensive investigations, relying primarily upon the techniques of survey research, have produced extensive systematic information bearing upon popular political attitudes and behavior and upon the electoral process more generally. Well developed conceptualizations which emphasize the social-psychological basis of mass political behavior have appeared. They are marked by considerable predictive and explanatory power and have gained widespread acceptance. Nor has research been limited to the phenomena of the contemporary era. Historians and other social scientists have increasingly explored popular political behavior in earlier historical contexts, in order to establish a broader, empirically grounded knowledge of political behavior.
But despite progress, knowledge of popular political behavior is currently marked by some disarray. Various investigators drawing on survey research analysis suggest that significant tensions have developed within prevailing theories of mass politics. Since the early 1950's "deviating" presidential elections have been the rule rather than the exception, and the imminence — indeed the occurrence — of "partisan realignment" has been frequently proclaimed and has been just as frequently rejected. There is no doubt that segments of the American electorate are today in an exceptional state of flux. Individual identifications with political parties have declined in recent years, and group partisan loyalties have apparently eroded. An increase in issue awareness and in ideological orientation on the part of the mass electorate has been observed and has been taken by some as incompatible with the prevailing theory of voting behavior. These developments have led several scholars to suggest that widely accepted views of mass electoral processes may be "time bound" and limited in their applicability to the narrow temporal period from the late 1940's to the early 1960's.
It can be expected that systematic investigations of mass political behavior in differing historical contexts will contribute to reduction of theoretical tensions and to movement from special to general theory. These goals have not yet been achieved. Historical investigations have been marked by conflicting findings and have, if anything, worked to increase conceptual confusion. Investigations of historical voting behavior have tended to focus rather narrowly upon periods of sharp electoral change. While partisan realignments have received major scholarly attention in these historical studies, the other election types — maintaining, deviating, and converting — that complete the typology suggested by V. O. Key, Jr., developed by Angus Campbell, and expanded by Gerald Pomper have been less carefully considered. In general, much less attention has been directed to the electoral phenomenon of interrealignment periods than to those of realignment eras.
The periodic recurrence of partisan realignments has been widely recognized. Most notably the recent work by James Sundquist, Dynamics of the Party System, concentrates on major and minor realignments with particular attention to the issues involved. Realignments have been dated as occurring in the 1820's, the late 1850's and 1860's, the 1890's, and in the late 1920's or 1930's. Considerable scholarly effort has been directed to determining the exact timing of these realigning electoral shifts. It is fair to say, however, that definitions of partisan realignments have been marked by a lack of conceptual clarity. Key provided what still remains one of the most clear and comprehensive descriptions of realignments when he spoke of "an election type in which the depth and intensity of electoral involvement are high, in which more or less profound readjustments occur in the relations of power within the community, and in which new and durable electoral groupings are formed." In his discussion he suggested that realignments occur in the course of a single critical election, but elsewhere in the same essay he suggested that realignments may occur over a longer span of elections. And, indeed, some scholars have expressed preference for the latter view of the realignment phenomenon. In an early essay, for example, Duncan MacRae, Jr., and James A. Meldrum submitted substantial evidence to demonstrate that in Illinois, partisan realignments could best be seen as occurring across a series of elections which they described as constituting critical periods. In a subsequent essay, Key also added further complications when he described a process of secular realignment by which particular groups — or, conceivably, a majority of the electorate — gradually shifted their allegiances from one political party to the other.
These characterizations of the realignment process are, of course, not necessarily in contradiction with each other, nor should they in themselves lead to a lack of conceptual clarity. It is obviously possible to conceive of partisan realignments as occurring in all three ways in different areas and time periods. But much of the literature has been in the form of case studies aimed at dating the precise temporal location of partisan realignment in particular and often highly limited areas. Nowhere is this characteristic of the literature more clear than in the debate that has raged over the status of the presidential election of 1928. Since the appearance of Samuel Lubell's Future of American Politics in 1952, numerous scholars have sought to demonstrate that 1928 was or was not a critical election in particular areas. Others have persisted in the view that realignment occurred in the 1930's — again, usually in particular areas. It is fair to say that each of the elections of that decade has its supporters as the critical election of the period. Furthermore, various scholars have also suggested that partisan realignment actually occurred, once again in particular areas, during elections prior to 1928.
This is not the place for extended commentary on the methodological characteristics of these studies. However, several issues do require discussion at this point. Since each of these studies has tended to focus on a different geographical area or population group, direct confrontations between conflicting findings for the same population group or area have rarely occurred. Even more important, the findings for different particular areas do not aggregate to a view of partisan realignment as a national phenomenon. With some exceptions, such studies have focused exclusively upon presidential contests, while neglecting elections to lesser federal and to state and local offices. These comments refer, of course, primarily to studies of the so-called New Deal realignment whenever that realignment is dated. The growing body of literature on earlier realignments suggests much the same pattern of conflicting findings and noncumulative results.
With some important exceptions, critical realignments have rarely been studied empirically as a national systemic process. The national-level consequences of realignments are sometimes asserted, but rarely are the linkages between grass roots behavior and these national-level consequences investigated. Furthermore, it is often asserted that realignments are a product of national crises, but such crises have been more common historically than realignments. Very little effort has been devoted to explaining why some crises give rise to realignments while others do not. Few scholars have attempted to explain, much less to demonstrate, why the American political system periodically undergoes these large-scale readjustments or to discover their effects on the political system as a whole. From any perspective, moreover, realignments have constituted only brief episodes in the political life of the nation. Narrow preoccupation with the realignment phenomenon leads, in other words, to neglect of by far the largest part of Amercan political life and development. Narrow concern with realignments has inhibited exploration of the relationships between the interrealignment period and the realignment process itself.
In this chapter we suggest a more comprehensive definition of partisan realignment. Generally, we argue that permanent electoral changes favoring a party and sustained control of policymaking institutions by that party should be viewed as central defining elements of critical realignments. We attempt to report on research in progress and to develop a broader and more comprehensive conceptualization of American political processes. Our definition of partisan realignment is an element in that larger conceptualization. In suggesting that definition, and in attempting to demonstrate its congruence with basic empirical data, we do not mean to rule out other definitions or to argue that realignments can only occur in a single way. We do mean to assert two propositions. First, realignments should be viewed as parts of a longer sequence which is marked by more or less considerable electoral variability at all times. Second, the linkage between electoral change — even marginal change — and durable transitions in partisan control over policy institutions has been seriously underexamined.
For the purposes of this chapter we have employed data bearing upon the patterns of partisan control over the federal and state governments and aggregate election returns at the state level. The reasons for choosing this level of analysis are obvious. The American constitutional structure is such that states are both constituencies and major policymaking units in their own right. Our purposes here, moreover, are explicitly macroanalytical; for such a system-level strategy, states are peculiarly appropriate units of analysis.
I Electoral Change
Traditional political historiography treats elections as unique events determined by the reactions of an informed citizenry to the specific issues, candidates and occurrences of the day. More recent investigations of historical voting behavior have placed less emphasis upon the unique character of elections. Such a shift in emphasis is a valuable corrective to the view of traditional historiography, but it can be carried to an extreme. Both partisan continuity and change have characterized American electoral history. Partisan change, including lasting change, has not been confined to periods that have been identified as eras of critical realignment. It is true, of course, that the major-party distribution of the presidential vote, for example, has tended to fluctuate within a relatively narrow range across the course of American history. Usually the winning candidate has received lass than 60 percent of the total vote; by the same token, the losing candidate's share of the total vote has seldom fallen as low as 40 percent. Even so, it is clear, that the partisan distribution of the presidential vote has consistently fluctuated, frequently by a considerable magnitude, from one election to the other.
Perhaps the simplest way to assess the variation in a series of proportions, such as the Democratic share of the vote for president, is to examine the first differences in the series. Figure 1.1A displays the average of the absolute values of the first differences in the percentage of the vote received by Democratic candidates in all states from 1828 through 1968. Figure 1.1B presents the same data for the Republicans from 1860 through 1968 and for Whig candidates from 1840 through 1852. The higher the scores in Figure 1.1, the more the states departed, on the average, from the previous election in the percentage of the vote cast for the presidential candidates of a given party. The departure, or first difference for a particular state, may be either an increase or decrease in the share of the vote received in the previous election; the direction of this change is ignored. In some instances, such as the Democrats in 1932, the difference scores reflect increases in the vote received by the party, in this case the Democrats, in all states in 1932 as compared with 1928. In other instances the difference scores reflect decreases in all states as compared with the preceding election, and in still others the scores reflect a combination of decreases in some states and increases in others.
The data presented in Figure 1.1 convey an overall impression of considerable electoral volatility. Of course, some of this variation could be eliminated. The high Republican score in 1912 reflects the Bull Moose schism and could be greatly reduced, if not eliminated, by combining the loyal Republican vote with the Progressive vote. Republican division, however, decided the outcome of the election of 1912, and the difference score is a valid reflection of shifts in partisan voting behavior. Other elections marked by high difference scores are less susceptible to such combinatorial remedies; the variability remains. It should be clear, moreover, that the procedures and data employed underestimate variations in voting behavior. Comparison of the aggregate vote at the state level undoubtedly masks shifts in voting behavior on the part of individuals and smaller units within the various states. The aggregation process, however, cannot introduce spurious variation.
Examination of Figure 1.1A indicates sharp changes in the Democratic vote in 1836, 1860, 1896, 1932, and 1968, as might be expected, with above-average variability in the elections of 1864, 1920, 1928, 1948, 1952, and 1964. In the Republican case the elections of 1912, 1916, 1920, 1932, 1952, and 1964 are all marked by high levels of variability, and only slightly lower levels of variability are found in the elections of 1864, 1896, and 1968. In some elections, in other words the vote for both parties was marked by considerable variability; in others shifts are observable only in the vote for one party, reflecting in most instances the intrusion of third party movements. If we summarize the patterns of variability for both parties, we can also observe periods of relative stability in presidential voting which include the elections from 1868 through 1892, from 1900 through 1908, from 1936 through 1944, and in the elections of 1956 and 1960. The period prior to the Civil War also appears stable in the case of the Democratic and Whig vote; however, the other parties of those years, which are not displayed, were marked by considerable volatility. It can also be noted that each of the presidential elections that have been identified as partisan realignments — 1860, 1896, and 1932 — involved sharp electoral change, usually more marked in the case of one party than the other. However, other elections were also marked by electoral change of at least approximately equal magnitude.
The first differences were also computed for the Democratic and Republican congressional vote on a biennial basis (Figures 1.2A and 1.2B). The congressional returns were aggregated to the state level. In most cases, this means that votes from more than one district are combined. These data were treated in the same fashion as the state-level presidential returns above. When compared with the fluctuations characteristic of the presidential vote, the congressional series appear considerably more stable. Perhaps the most striking difference between the presidential and congressional series is to be found in the decades following the 1930's. During these years both congressional series are characterized by a high degree of stability, and, indeed, for both parties these decades were marked by the greatest and most prolonged stability of the entire period. In sharp contrast, the presidential vote during these same years was marked by increasing volatility beginning in the 1950's.
The congressional series for both parties reflect variation during periods that have been identified as partisan realignments. As in the case of the presidential series, however, other periods were also marked by considerable variation. The highest levels of variation in Democratic voting occurred in the middle years of the nineteenth century especially in the years before the Civil War. Subsequently, only the elections of 1894 and 1896 reflected comparably dramatic shifts in the Democratic vote for Congress. Substantially less variation characterized the peaks of twentieth century Democratic variability that occurred during the elections from 1920 through 1932. In general, the shorter Republican series reflects lower levels of variability than does the Democratic series. The peak of Republican variability came in 1860 and in several of the elections during the two decades that followed. Several elections during the years from 1912 through 1932 were also marked by variability of about the level of the 1860's and 1870's.
Excerpted from The History of American Electoral Behavior by Joel H. Silbey, Allan G. Bogue, William H. Flanigan. Copyright © 1978 The Center for Advanced Study in the Behavioral Sciences, Stanford, California. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.