Political Ideology Measurement Project

Much political science research requires indicators of preferences that may be contrasted with behavior, for purposes of understanding strategy. One of the key impediments to such research lies in the need to establish preference indicators that are truly comparable between different individuals (perhaps serving in different institutions) and from one time period to another. Groseclose, Levitt, and Snyder (APSR 1999) and Bailey and Chang (JLEO 2002) highlight the pitfalls of using non-comparable preference indicators in political science research. Over the past few years, I have employed some estimated ideal points that satisfy the demands for my own projects, and may be useful to other scholars. This page contains replications of those measures, along with working papers and other materials.

All ideal points provided from this project are scaled in the exact same metric as Poole's first dimension common space scores. As a result, it is possible to make valid comparisons and perform valid algebraic calculations (such as linear distances, or variances) among and between the economic ideology scores for (virtually) every federal judge, justice, congressman, president or independent commissioner this century.

All the measures I have constructed are rooted in a straightforward data innovation - identify a collection of congressmen (for whom a wide variety of ideology measures are available) who have also been appointed to the executive or judicial branch; construct a predictive model of ideology for those individuals; use the model to extrapolate ideology indicators for all similar executive or judicial appointees. Thus, all of the ideology scores reported here are based on identification of "bridging" congressmen who have been appointed to some executive or judicial post.

Poole demonstrates that legislators do not change their voting behavior, even if their constituencies change due to demographic shifts or redistricting (see Poole and Romer 1993; Poole 1999 in Changing Minds, by Munger and Roberts, eds). I deduce that their floor voting behavior represents a stable indicator of their political preferences. I further assume that their personal political preferences while serving in their appointive capacity do not vary from the days of their legislative service. Thus, I treat those federal judges with prior congressional service as having the same (fixed) ideologies on the bench that they had while in Congress. Similarly, I treat those executive appointees with prior (or in a small number of instances, subsequent) congressional service as having the same ideologies at the head of an agency that they had while in Congress.

The assumption of stable preferences is difficult to test even within an institution, let alone across institutional settings. However, my work so far is consistent with the assumption of stability of preferences after congressional service. Among executive appointees who have served in Congress, appointment outcomes exhibit the patterns that one would expect if the president and senate engaged in a bargaining game over appointees whose ideology was identical to when they served in Congress. Additional ongoing research by me and by some graduate students further tests the assumption of ideological stability.

Poole (with Rosenthal, Oxford U Press 1997) also demonstrates that legislative voting behavior is largely unidimensional, and increasingly so. His recent work suggests that Supreme Court voting behavior is similarly unidimensional (working paper available from his website). Using different techniques over a longer time frame, Grofman and Brazill (Public Choice 2002) also demonstrate that Supreme Court voting is unidimensional. I assume that political preferences of lower court judges and independent commissioners are equally well captured by the traditional economic liberal/conservative dimension that seems to represent most disagreement in Congress and on the high court.

The data I present here has the virtue that it is NOT based on behavior of the appointees once in office. As a result, the scores may be used as non-circular predictors of appointee behavior. Segal and Spaeth rightly note the importance of identifying exogenous predictors of behavior (see Segal and Spaeth, 2002, SCAM-R, pp. 320-1). Other preference estimations have been based on behaviors, such as the useful recognation that presidents take positions on legislative bills, or that presidents take positions (through their Solicitor Generals) on Supreme Court cases (see Zupan Public Choice 1992; McCarty & Poole, 1995, LSQ; Bailey & Chang, 2002, JLEO). As a result, they may be more properly described as summaries of behavior, rather than predictors of behavior.


These datasets contain my ideal point estimates for the US Supreme Court (Black to Breyer), US Circuit Courts of Appeals (1891-2002), US District Courts (1933-2002), and various independent regulatory commissions. The scores are scaled in the identical metric to the first dimension of Keith Poole's "common-space" scores for all Presidents, Senators, and Representatives since 1937. For explanation of the estimation technique employed by Poole, see AJPS 1998 and Poole's website. The first dimension common space scores represent economic conservatism. I follow Poole's demonstrations that ideology of presidents and legislators is fixed throughout their career, and first dimension scores capture the vast majority of political preferences. I therefore estimate only a single ideal point for each judicial or executive branch appointee, based on a variety of personal and contextual political factors.

For explanation of the estimation technique I employ, see my brief white papers on estimating common space scores for judicial or executive branch positions. The estimations are based on collections of 63 federal judges and justices who had previously served in Congress since 1937 and 72 independent commissioners who had same experience. These data were collected in connection with my JLEO article and I have ongoing research on these two datasets.

Data on the full complement of judges, justices, and commissioners have been collected from the Supreme Court Compendium, The Federal Judicial Center, The Independent Regulatory Commissioner Database, and by generous provision from Shel Goldman, Gerry Gryski, and Gary Zuk.

Individual Common Space Scores

(one observation for each individual who has served, scaled in first dimension common space)

Aggregate Common Space Scores

(one observation for each day, includes many aggregation variables, including means, medians, variance, clipped ranges and more. See aggREADME.TXT)

Published Articles

Work in Progress

Related Sites

Go to the Independent Regulatory Commissioner Database Homepage

Return to David C. Nixon's Homepage

revised 12/17/2004

Copyright © 2004 David C. Nixon. All rights reserved.