Herbert F. Weisberg

A multidimensional conceptualization of party identification

The standard treatment of party identification makes several untested assumptions, especially that citizens can identify with only a single party and that political independence is just the opposite of partisanship. A more general possibility is that several attitudes must be taken into account: attitudes toward the Republican party, the Democratic party, political independence, and political parties generally. A literal reading of the usual party identification is consistent with this multidimensional interpretation. Citizen ratings of the two parties turn out to be virtually uncorrelated, as are ratings of independence and political parties, confirming this multidimensional view. Strength of identification and strength of independence are separate in this model, which explains some of the anomalies in the current literature, including intransitivities in relationships with other variables and weak correlations involving independence. New questions included in the 1980 CPS National Election Study support this interpretation and provide a new understanding of political independence.

Dimensions of Candidate Evaluation

The story of a presidential election year is in many ways the story of the actions and interactions of those considered as possible candidates for their nations highest office. If this is true in the abstract, it certainly was true in the election of 1968. The political headlines of 1968 were captured by those who ran for the nominations of their parties, those who pondered over whether or not to run, those who chose to pull out of the race or were struck down during the campaign, those who raised a third party banner, and those who resisted suggestions to run outside the two-party structure. While 1968 may have been unusual in the extent to which many prospective candidates dominated the political scene, every presidential election is, in its own way, highlighted by those considered for the office of President. The political scientist has shown scholarly interest in the candidates. His interest, however, has been selective in its focus—mainly concentrating on the two actual party nominees and not the larger set of possible presidential candidates. Research in electoral behavior has detailed the popular image of the nominees in terms of the publics reactions to their record and experience, personal qualities, and party affiliation. Furthermore, attitudes toward the nominees have been shown to constitute a major short-term influence on the vote. Yet attitudes toward other candidates have been surveyed only to ascertain the behavior of those people who favored someone other than the ultimate nominees.

Evaluating Theories of Congressional Roll-Call Voting

Criteria are developed to evaluate recent theories of roll-call voting in the House of Representatives. Since tests of these very different theories find high levels of predictive success, we must decide how to choose among them. Baseline models are developed to show the extent to which the votes could be predicted with minimal information. The 80-90 percent of the individual votes correctly predicted by the theories is found to be little improvement over the baseline models predicting voting along with the House or party majority. Since the statistical criterion is found to be indeterminate, the importance of verisimilitude to the process being studied is stressed. Simulation studies have done a good job of portraying the process aspects, and they could be usefully combined with statistical studies of long-term forces and interviewing studies of short-term forces affecting the voting. Theory development in the field should move to a more longitudinal perspective, as well as combining the long-term and short-term elements into a single overarching theory.

Measuring Change in Party Identification in an Election Campaign

Partisan change in the 1984 election campaign is analyzed in the continuous monitoring studies by the National Election Studies and Decision/Making/Information. More change was found in partisanship than is suggested in the literature. There was a shift toward the Republican party throughout the campaign, but the change was not unidirectional. Instead, there was curvilinear change moving toward the Democrat party during parts of the campaign and toward the Republicans during other parts of the campaign. The changes in partisan strength correspond to changes in vote intention, suggesting responsiveness to short-term campaign events. All in all, partisanship is more volatile than we have been led to expect.

The Influence of the Economy on Party Identification in the Reagan Years

Utilizing frequent measurements taken during a period of considerable economic volatility, our research explores how short-term fluctuations in partisanship have operated and continue to operate within a limited range before permanent realignment occurs. Specifically, we examine monthly fluctuations in the aggregate distribution of partisan identifiers between 1981 and 1986. We show that there were systematic changes in the partisan balance during the period, and confirm the MacKuen, Erikson, and Stimson (1989) finding that changes in presidential popularity account for some of the movement. Further, we detail economic influences on macropartisanship using objective indicators. The results show that both inflation and unemployment bear directly on the partisan balance, and that each exerts an indirect influence through presidential popularity.

Dimensionland: An Excursion into Spaces

Scaling analysis is based on a geometric metaphor. This workshop paper examines how our understanding of the metaphor affects our use of scaling. Instances which appear to be multidimensional are shown to be unidimensional under other scaling models. Conversely, some apparently unidimensional cases are found to be better described as multidimensional. Particular attention is given to the difference between multidimensional scaling and factor analysis. The philosophical implication of our dependence on the definition of unidimensionality is that scaling seeks only partial images of a real world that may be fundamentally unknowable.

THE DEMOGRAPHICS OF A NEW VOTING GAP MARITAL DIFFERENCES IN AMERICAN VOTING

There is now a marriage gap in presidential voting in the United States, with married people voting about 10%-15% more Republican than unmarried people. This voting difference developed in the 1972 election, and it is also evident in party identification. The proportion of unmarried people has doubled since 1964, so this cleavage is important. The gap can be explained statistically in terms of race and income, but still may give politicians the opportunity to gain votes by appealing to voters on the basis of family issues. Survey researchers have long probed social group differences in voting. They have come to expect certain demographic variables to be related to the vote while others are seen as so irrelevant that they are never examined. Yet familiar voting cleavages can diminish in importance, while new differences can emerge. We generally think of social group differences in the vote as resulting from divergent issue appeals or from mobilization of groups into parties by group leaders, and such appeals and mobilization would be noticed enough for us to look specifically for their effects. However, group voting differences can also emerge more subtly because of the changing demographics of the population. As a result, it is important for us to examine periodically the full range of social group differences in voting. When one so examines voting in the 1984 American presidential election, one finds the largest voting differences on the basis of race and income, followed in order by differences based on marital status, religion, gender, and region. These differences are familiar ones, exHERBERT F. WEISBERG iS Professor of Political Science at the Ohio State University. This research was partially based on work supported by the National Science Foundation under Grant No. SES-8218873. The National Election Studies data were supplied by the Inter-university Consortium for Political and Social Research. Any opinions, findings, conclusions, or recommendations expressed in this paper are those of the author and do not necessarily reflect the views of the National Science Foundation or the ICPSR. The author would like to express appreciation to Terri Royed for assistance in the data

PARTISANSHIP AND INCUMBENCY IN PRESIDENTIAL ELECTIONS

Party identification is a standard part of our understanding of presidential voting, but the effects of presidential incumbency on presidential voting have not been recognized in most voting models. Democratic candidates in the twentieth century received 10 percent more of the two-party vote when Democratic incumbents were running for reelection than when Republican incumbents were running. National Election Studies surveys show that the effect of incumbency varies with individual partisanship, with the greatest effect, as expected, among independents. Opposition party identifiers defect at a higher rate than incumbent party identifiers when the incumbent is running for reelection. Even after controlling for retrospective and prospective economic voting, a 6 percent effect is found for incumbency. Incumbency thus conditions the impact of partisanship on presidential voting.

Interpreting the First Eigenvalue of a Correlation Matrix

The first eigenvalue of a correlation matrix indicates the maximum amount of the variance of the variables which can be accounted for with a linear model by a single underlying factor. When all correlations are positive, this first eigenvalue is approximately a linear function of the average correlation among the variables. While that is not true when not all the correlations are positive, in the general case the first eigenvalue is approximately equal to a lower bound derived in the paper. That lower bound is based on the maximum average correlation over reversals of variables and over subsets of the variables. Regression tests show these linear approximations are very accurate. The first eigenvalue measures the primary cluster in the matrix, its number of variables and average correlation.

Presidential Succession Effects in Voting

Attitudes towards a departing administration can help shape attitudes towards candidates, especially when the incumbent vice-president is one of the candidates. This succession effect was apparent in the 1988 presidential election, when Vice-President Bush benefited from the enduring popularity of retiring President Reagan. This article develops a model in which succession effects, the net candidate score and party identification affect the general election vote. Analysis shows that this effect remains when controls are instituted for retrospective voting more generally. Attitudes towards Reagan also had an indirect impact by affecting the net Bush-Dukakis candidate score; altogether the estimated impact of the Reagan effect in 1988 was to turn the vice-presidents predicted loss into his observed victory. Additionally, a succession effect was detected in the 1988 nominating campaign, with Bushs popularity over Dole benefiting from reactions to the Reagan administration. There is evidence of succession effects in other presidential elections, particularly a Johnson effect in 1968.