James A. Stimson

Regression in Space and Time: A Statistical Essay

Regressions on data jointly structured in space and time, commonly referred to as the pooling of cross sections of time series, can be formidable both in the strength of their design properties and in the number of special statistical problems encountered with them. This essay deals briefly with the potential applications of pooled design and more extensively with the special statistical problems commonly associated with analysis in space and time together. Four estimators-ordinary least squares, least squares with dummy variables, error components, and an adaptation of Box-Jenkins ARMA models to the pooled estimation problem are reviewed, with an effort to suggest where each may find application in political science research. The four estimators are then illustrated by analysis of the regional dynamics in party issue polarization over issues of racial desegregation in the U.S. House of Representatives.

The Two Faces of Issue Voting

Both implicit democratic norms and the reconstructions provided by theorists of rational choice suggest that issue voters are more sophisticated–educated, informed, and active in politics–than other voters. But some issues are clearly more difficult than others, and the voters who respond to “hard” and “easy” issues, respectively, are assumed to differ in kind. We propose the hypothesis that “easy-issue” voters are no more sophisticated than non-issue voters, and this is found to be the case. The findings suggest a reevaluation of the import of rising and falling levels of issue voting and suggest a prominent role for “easy” issues in electoral realignments.

Public opinion in America : moods, cycles, and swings

* Public Opinion? * The Concept of Policy Mood * Developing a Measure of Mood * The Components of Mood * An Electoral Connection? * Reflections on the Present and Future of American Politics

Tides of consent: How public opinion shapes American politics

List of tables and figures Preface Acknowledgements 1. Opinion flows 2. What the public wants of government 3. Left and Right movements in preference 4. The great horse race: finding meaning in presidential campaigns 5. Between the campaigns: public approval and and disapproval of government 6. On politics at the margin Bibliography Index.

The Least Dangerous Branch Revisited: New Evidence on Supreme Court Responsiveness to Public Preferences

With competing assumptions and alternative empirical models, scholars have come to rather different conclusions about the impact of public preferences on the decisions of the U.S. Supreme Court. Some have found the justices to be attentive to mass opinion, while others have judged it to be irrelevant. Across this divide, however, one assumption is widely shared; that is, political scientists generally agree upon how best to measure the Courts outputs. In this analysis, we employ an alternative estimate of the justices’ liberalism, one which we think better reflects the underlying ideological tenor of their policies. With data from 1953 to 1996, we compare time-series models using different indicators of the Supreme Courts aggregate liberalism. Our results suggest that, in addition to being motivated by their own preferences, the justices are highly responsive to public mood, as well.

Bankers or peasants revisited: economic expectations and presidential approval

Abstract This paper updates and expands the argument that the US electorate is forward-looking when evaluating presidential performance on the basis of the economy. The electorates approval of the president depends on its economic expectations regarding the future level of prosperity. More is incorporated in this evaluation than a simple extrapolation from the past. When evaluating the president, the US electorate efficiently combines its knowledge of the current economy with its informed expectation of economic change.

The moving centre : preferences for government activity in Britain, 1950-2005

The political ‘centre’ is often discussed in debates about public policy and analyses of party strategies and election outcomes. Yet, to date, there has been little effort to estimate the political centre outside the United States. This article outlines a method of estimating the political centre using public opinion data collected for the period between 1950 and 2005. It is demonstrated that it is possible to measure the centre in Britain, that it moves over time, that it shifts in response to government activity and, furthermore, that it has an observable association with general election outcomes.

Question Wording and Macropartisanship

MacKuen, Erikson and Stimson used quarterly Gallup poll data to show in this Review in 1989 that changing levels of macropartisanship, the two-party division of partisans, responded to presidential approval ratings and perceptions of the economy and predicted national election results. In a 1991 Review research note Abramson and Ostrom argued that the NES and GSS questions more commonly used by scholars generated macropartisanship measures less sensitive to short term factors and less predictive of election outcomes. In this Controversy , Erikson and Stimson respond to the challenge and present new data from CBS News and New York Times telephone surveys to buttress their earlier analyses, arguing against substantial effects of the different question wordings. Abramson and Ostrom explain their continuing reservations.

Congressional Response to Mandate Elections

Elections from time to time are widely believed to carry a mandate, to express a message about changed policy preferences of the electorate. Whatever the accuracy of such beliefs—a matter about which we are skeptical—perceptions of a mandate should affect the behavior of actors in government. Politicians lack the scholarly luxury of waiting for careful analyses. They must act in the months following elections. We postulate that many will act as if the mandate perceptions were true, veering away from their normal voting patterns. This is driven by election results and interpretations that undermine old calculations about what voters want. As the flow of information gradually changes these perceptions, and the election becomes more distant, members of Congress return to their normal position. We first ask, how would members observe an emerging consensus of mandate? And then we model the duration of the change in behavior in an event-history framework. That permits a depiction of important movements of the median member and, from this, inferences about policy impact.

The evolution of policy attitudes in France

The theory of issue evolution predicts that the dimensional space of party competition is simple. We contrast this prediction with the expectation that a complicated multi-party system, such as the one in France, produces a more complicated dimensional structure. To test this claim, we examine the longitudinal structure of the policy preferences that underlie public opinion in France. Using surveys of preferences as a basic data source, we are able to extract two latent dimensions that almost fully explain the reported preferences. Both dimensions are defined by the left–right structure of the French party system. Whereas one is the traditional socioeconomic domain, the other comprises a wide array of new cultural issues. The orthogonal solution, however, does not produce the expected socioeconomic and cultural dimensions. Thus we impose our prior belief in the socioeconomic and cultural content and rotate the two dimensions independently to maximize fit with the two prior dimensions. We show that the same two-dimensional structure is also present in cross-sectional data, and can be used to position parties in the two-dimensional space. Moreover, we find that the two dimensions are closely connected, despite their completely different content. The explanation, which arises from the theory of issue evolution, is that the meaning of left–right is dynamic as well as elastic and incorporates new issues as they arise.