The Attitudinal Model, Political Science, Ecological Fallacy and Exaggeration

33 Pages Posted: 8 Aug 2006

Date Written: August 3, 2006


Empirical scholars of the United States Supreme Court, Jeffrey Segal and Harold Spaeth, have long contended that Supreme Court decisions are based primarily upon the ideological beliefs of the justices, and that ideology alone accounts for the bulk of choices made in civil liberties cases. However, this conclusion results from the misinterpretation of an ecological regression model. The researchers never modeled the votes of the justices; they only analyzed an index of grouped aggregates. When announcing conclusions, however, scholars equated variation in a voting index with the frequency distribution of binary observations that comprised it. As a result, model conclusions were exaggerated and disciplinary misinformation was created. This work exposes and corrects this problem by re-estimating the relationship between justice ideology and votes with a multilevel approach that uses a logistic regression to directly examine the dependent variable prior to its manipulation into grouped data. The findings demonstrate that ideology models lose about two-thirds of the level of explanation researchers previously proclaimed. This new understanding supports a more limited critique of the role that ideology plays on the Court - one that has a long history in political science that predates the more value-dominant "attitudinal" framework.

Keywords: Attitudinal Model, Segal and Spaeth, Political Science, Ecological Regression, Bivariate Ideology Model, Ecological Fallacy, Exaggeration

Suggested Citation

Wilson, Sean, The Attitudinal Model, Political Science, Ecological Fallacy and Exaggeration (August 3, 2006). Available at SSRN: or

Sean Wilson (Contact Author)

Wright State University ( email )

Department of Political Science
306 Millett Hall
Dayton, OH
United States
(937) 775-2942 (Phone)


Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics