by Thomas R. Marshall. Albany: State University of New York Press, 2008. 269pp. Hardcover. $85.00. ISBN: 9780791473474.

Reviewed by Justin Wedeking, Department of Political Science, University of Kentucky. Email: justin.wedeking [at]


Should the US Supreme Court adopt a majoritarian stance and defer to lawmaking majorities? Or should the Court serve as an independent check against a tyrannous majority? While judicial scholars have focused much time and energy on those questions, within that debate is an empirical question: Is the Court consistent with public opinion? Resolution of that question holds implications for the larger normative debate. In PUBLIC OPINION AND THE REHNQUIST COURT, Thomas Marshall addresses two primary questions: Were Rehnquist Court decisions consistent with American public opinion? And, to the extent the decisions were consistent, what best explains why they agreed with American public opinion? To answer those questions, Marshall identifies fifteen potential theoretical explanations, gathers data and then empirically tests the theories. Those familiar with Marshall’s earlier book on this subject (Marshal 1989) will find similarities and comparisons throughout.

The book is full of interesting facts about the Court. Chapter 1 reviews how justices view public opinion and finds that the Rehnquist Court uses specific poll results in 15% of Court opinions, while 28% of opinions do not cite a specific source when referencing public opinion (Table 1.2, p.7). This is remarkable and provides fuel for normative debates (for both sides) on whether the Court should represent majoritarian positions. Marshall also examines how often the Rehnquist Court invokes four judicial theories of public opinion: (1) an informed public opinion; (2) judicial restraint; (3) contemporary public opinion; and (4) public opinion as a threat. Marshall shows how use of the theory, ‘law and policy should reflect contemporary or evolving public opinion,’ increases substantially during the Rehnquist Court compared to previous Courts. Conversely, the Rehnquist Court declined its usage of the theory, ‘public opinion alone is an adequate check on policy.’ These comparisons to earlier Courts are made throughout the book and are a great strength in demonstrating to the reader how the Rehnquist Court both changes and maintains continuity. Chapter 1 finishes with a brief introduction to 15 different linkage models tested throughout the book.

Chapter 2 discusses representation and two ways of studying it: the trend method that tracks policy and Court decisions over time; and the pairwise method that “matches” specific Court decisions to poll questions. Marshall uses the pairwise method and bases the analysis and results on 111 poll-to-decision matches during the Rehnquist era. This represents a tremendous data collection effort and should be [*730] commended. One cautionary note with using poll results, however (aside from the possibility that organizations may have different sampling and contact methods), is that respondents are sensitive to question wording. It would be beneficial for readers to have access to the actual question asked for each poll, response choices, sample size, frequencies and the margin of error. This is important given that Segal, Spaeth, and Benesh (2005, at 365) provide a clear example of how different sources and question wordings produced diametrically opposed poll results for RUST v. SULLIVAN (1991), which is, incidentally, one of the cases in the sample (Appendix 1, case # 23, p.164). Given that research shows even the smallest change in question wording can produce significant changes on topics like ideological constraint (Sullivan, Piereson, and Marcus 1978) and party identification (Burden and Klofstad 2005), displaying basic information is imperative for ensuring the facial validity of each poll.

Chapters 2 through 8 ambitiously tests 15 different linkage models, and while there is not enough space to summarize each model adequately, Marshall explains how each model generates specific predictions on how public opinion and the Court are linked. Table 1 contains a summary of the results, the left column identifies the linkage model, the second column lists the corresponding chapter, and the third column contains the key finding for each model. As Table 1 shows, six models explain the relationship between public opinion and the Court, and one receives partial support, while eight models fail to garner support. The last column of Table 1 pertains to Chapter 9 and whether Marshall included the model into his final empirical linkage model of public opinion. The purpose of this final model is not clear, other than the included factors were found to be useful in earlier analysis. I anticipated a multivariate test of the final model, but it never materialized. Instead, Marshall summarizes the relationship between the Rehnquist Court and public opinion as consisting of three ties: “the context of a controversy, the justices themselves, and the Court’s current norms” (p.162). Moreover, the criteria for inclusion into the final model are not clear. For example, four relationships were tested for the “state of public opinion model” in Table 2.4 (p.44) and only one reached statistical significance, yet the “state of public opinion model” was included in the final model. In contrast, the “realignment model” found support for one out of four statistical tests (Table 5.9, p.101), but was labeled as “mixed support” for the Rehnquist Court and excluded from the final model. A separate question also arises whether many linkage models might be useful at the individual level of analysis.

Marshall carefully walks the reader through each chapter’s analyses, and the book is meticulously footnoted, demonstrating extraordinary effort in covering the subject matter. The book’s main finding is that the Rehnquist Court continued the long tradition of earlier Courts (dating back to the 1930s) of being consistent with public opinion majorities in 64% of its decisions. Knowing this fact is helpful for scholars who want to evaluate normative claims about the Court. However, it would be beneficial to have a measure of certainty (e.g., 95% confidence interval) for the 64% estimate. The finding would be [*731] more persuasive if we were reassured that the lower bound does not cross the 50% threshold, which is crucial to sustain the claim that the Court aligns with public opinion a majority of the time.

Another quibble is the use of public opinion polls taken after the Court decision. While use of post-decision polls is ideal when examining whether the Court affects public opinion, it presents a challenge for answering the book’s main question – whether the Court is consistent with public opinion – because it suggests a different temporal ordering. Using post-decision polls to examine whether the Court is consistent with public opinion assumes that public opinion is stable across Court decisions. This concern is bolstered by Marshall’s results in Table 2.4 (p.44), where one of the few statistically significant findings indicates that Rehnquist Court decisions are consistent with public opinion 71% of the time for pre-decision polls, but only 52% with post-decision polls. Why does this gap appear, and is this research design issue worthy of further investigation? The gap is indirectly addressed in Chapter 7 with the test of the “Short-Term Manipulation Model.” In 16 cases where a decision had both pre- and post-decision polls, Marshall found no evidence the Court moved public opinion towards its decisions, but did find mild support for the public moving away from the Court’s position. This finding runs counter to the “positive response” theory investigated by others and suggests the possibility of a “negative response.” Fortunately, the shift works in favor of the book’s main finding and suggests that the 64% estimate is conservative.

Regardless, caution should be exercised when using post-decision polls to make inferences about the Court’s consistency with public opinion majorities, and it highlights a potentially new problem for researchers using polls that correspond to a single Court decision – a “pollster selection bias.” This would suggest that cases that have only a post-decision poll are fundamentally different to the public than cases with a pre-decision poll. In other words, public discussion of an issue sometimes increases after a decision (Johnson and Martin 1998) and may spark pollsters’ interest in the topic.

Marshall covers an extensive number of polls, and analyzes them across an array of explanatory variables, with the dominant mode of analysis being crosstabulations. If readers are interested in multivariate tests, few are mentioned and with little detail. Additionally, some analyses would benefit from a more transparent explanation. For example, when evaluating the “test-of-time model,” Figure 8.1 graphs the probability that a Rehnquist Court decision will survive the test of time and shows that unpopular Rehnquist Court decisions have an expected half-life of 17 years, but 24 years if popular. However, it is not clear how a decision can have an estimated half-life of 24 years if the period under analysis is only 19 years (1986-2005).

These criticisms by no means diminish Marshall’s main finding that the Court often agrees with the public a majority of the time. PUBLIC OPINION AND THE REHNQUIST COURT would be a welcome addition to any undergraduate political science course on public opinion, the judicial process, or a [*732] specialized course examining linkages between institutions and the public.

Burden, Barry C. and Casey A. Klofstad. 2005. “Affect and Cognition in Party Identification.” POLITICAL PSYCHOLOGY 26: 869-86.

Johnson, Timothy R., and Andrew D. Martin. 1998. “The Public’s Conditional Response to Supreme Court Decisions.” AMERICAN POLITICAL SCIENCE REVIEW 92: 299-309.

Marshall, Thomas R. 1989. PUBLIC OPINION AND THE SUPREME COURT. Winchester, MA: Unwin Hyman.

Segal, Jeffrey A., Harold J. Spaeth, and Sara C. Benesh. 2005. THE SUPREME COURT IN THE AMERICAN LEGAL SYSTEM. New York: Cambridge University Press.

Sullivan, John L., James E. Piereson, and George E. Marcus. 1978. “Ideological Constraint in the Mass Public: A Methodological Critique and Some New Findings.” AMERICAN JOURNAL OF POLITICAL SCIENCE 22(2) 233-249.

RUST v. SULLIVAN, 500 U.S. 173 (1991). [*733]

Table 1: Summary of Results for 15 Public Opinion-Court Linkage Models

Linkage Model


Marshall’s findings on whether there was any empirical support for a public opinion linkage with the Rehnquist Court?

Included in the final linkage model?

State of Public Opinion


Some support


Federal Policy


Less support than earlier Courts


State/Local Policy


Less support than earlier Courts




Strong support


Interest Groups


Strong support for Solicitor General

No support for other groups

SG - Yes

Others – NO

Political Parties & Ideology


No evidence for political parties

Moderate justices consistent support

Party- No

Ideology- No

Political Socialization


Strong support if justice moved between regions or had D.C. job experience


Appointment Process


Inconsistent support


Judicial Roles


Support for Chief Justice and intellectual leader; no support for oral argument leaders


Length of Tenure


Longevity decreases representation




Mixed evidence


Symbolic Representation


No support for race or gender;

some for religion and political party


Short-Term Manipulation


No positive shift, some negative shift


Long-Term Manipulation


Decisions did not grow in popularity




Public opinion affects longevity


©Copyright 2008 by the author, Justin Wedeking.