ARE JUDGES CONSISTENT ACROSS AREAS OF LAW?
Polarization possibly means a politicized court, but the difficulty is that the source of the differences is unknown. For instance, judges may be basing their decisions on legal factors but disagree as to the relevant factors. We want to know whether there is a connection between judges’ personal preferences and how they decide appeals. There are a number of different ways to tease out this connection. We can look at how a judge’s decisions are related to how she tends to vote on other policy issues, how she was described in newspapers at the time she was appointed, how she decided past cases, or, as we will look at later, the political party of the government actor who appointed her. Although a judge’s views are often labeled as either liberal or conservative in particular areas of law, there are other factors such as gender, regional origin, or religion that affect decisions.93
№ot surprisingly, judges on the uS Supreme Court have been found to decide cases in line with their political views despite their frequent statements to the contrary during confirmation hearings in the Senate.94 For example, for judges on the uS Supreme Court between 1962 and 1998, ideology measures derived from newspaper editorials at the time of their appointment were able to predict 70 percent of their votes in search and seizure cases, which is much higher than the prediction just based on the facts of the case.95 In general, the ideology of uS Supreme Court judges is strongly connected to how they decide cases. The strength varies over time but tends to be fairly consistent across areas of law.96
This connection also holds in countries with more closed processes with appointments made by a prime minister or president such as Canada and Australia, though the connection appears weaker. In Canada, for example, judges’ votes were correlated with scores based on newspaper reports at the time of their appointment—judges who were described as conservative in newspapers tended to vote similarly in right to counsel, search and seizure, union, and tax cases.97 These scores were not, however, correlated with all decisions by the judges, of British Columbia Press, 2007) at 124 [Ostberg & Wetstein, Attitudinal Decision Making]; Donald R. Songer et al., Law, Ideology and Collegiality: Judicial Behaviour in the Supreme Court of Canada (Montreal and Kingston: McGill University Press, 2012) [Songer, Judicial Behaviour] for Canada; Weiden, “Judicial Politicization,” supra note 69 for Australia).
- 93. Epstein et al., “Ideology,” supra note 92.
- 94. Epstein, Landes & Posner, Behavior, supra note 74 (Justice Alito, for instance, stated that Supreme Court justices must “be careful not to inject their own views” into their decisions, and Justice Sotomayor said that “[j]udges must apply the law and not make the law” at 101).
- 95. Segal & Spaeth, Attitudinal Model, supra note 74 at 324.
- 96. Epstein et al., “Ideology”’ supra note 92. See also Mathew E. Wetstein, C.L. Ostberg, Donald R. Songer & Susan Johnson, “Ideological Consistency and Attitudinal Conflict: A Comparative Analysis of the U.S. and Canadian Supreme Courts” (2009) 42:6 Comparative Political Studies 763. See also Devins & Baum, supra note 8.
- 97. For a description of the scores, see Ostberg & Wetstein, Attitudinal Decision Making, supra note 92 at 49-58. See also Ostberg & Wetstein, Attitudinal Decision Making, supra note 92 (finding newspaper scores were significant in right to counsel, search and seizure, union, and non-unanimous tax cases but not equality or free expression cases over the period 1984-2003). C.L. Ostberg & Matthew E. Wetstein, “Search and Seizure Cases in the Supreme Court of Canada: Extending an American Model of Judicial Decision Making across Countries” (1999) 80:4 Social Science Quarterly 757 (examining search and seizure cases between 1984 and 1994 and finding votes were correlated with an ideology measure combining newspaper scores and the party of the appointing prime minister).
including in equality or freedom of expression cases.98 When we look over the post-Charter period (post-1982), a judge’s newspaper scores were also modestly related to his ideal point—a measure of where he sits on policy issues relative to everyone else on the Court at the time.99 Further, judges on the Supreme Court of Canada during a natural court period (where no one left or joined the Court) in 1992-1997 were influenced by similar factors to judges of the US Supreme Court in the areas of economic, criminal, and civil liberties cases.100 However, although a liberal/conservative divide was important for both courts, ideology seemed to play a different role or at least not as unidimensional a role in Canada, as Canadian judges could not be classified consistently as liberal or conservative in each area.101
Therefore, although the Supreme Court of Canada judges’ political views and their decisions are correlated, they are seen as less influenced by political views than judges on the US Supreme Court.102 They may also be influenced by other factors. For example, judges vote differently if the appeal is from their province of origin.103 Further, female judges vote differently than male in some areas. For
- 98. Ostberg & Wetstein, Attitudinal Decision Making, supra note 92. Recently, C.L. Ostberg & Matthew E. Wetstein, “Postmaterialist Values in the Supreme Court of Canada: Empirical Analysis of Environmental Cases, 1973-2010” (Paper delivered at the annual meeting of the Canadian Political Science Association, Victoria, BC, June 6, 2013) [unpublished] found that newspaper scores were not correlated with voting on the Supreme Court on environmental matters. See Alarie & Green “Policy Preference,” supra note 68, finding no connection between these scores and liberal voting rates of judges.
- 99. Alarie & Green “Policy Preference,” supra note 68.
- 100. Songer, Judicial Behaviour, supra note 92 at 114 (using factor analysis and finding that a liberal-conservative dimension was a strong predictor of splits in economic, criminal, and civil liberties cases).
- 101. Songer, Judicial Behaviour, supra note 92 at 115; Alarie & Green “Policy Preference,” supra note 68; Ostberg & Wetstein, Attitudinal Decision Making, supra note 92 at 191; Wetstein et al., “Comparative Analysis,” supra note 81 at 780 (finding a strong inverse relationship between criminal and economic liberalism between 1992 and 1997).
- 102. Alarie & Green “Policy Preference,” supra note 68; Ostberg & Wetstein, Attitudinal Decision Making, supra note 92 at 191; Songer, Judicial Behaviour, supra note 92 at 133.
- 103. C. Neal Tate & Panu Sittiwong, “Decision Making in the Canadian Supreme Court: Extending the Personal Attributes Model across Nations” (1989) 51:4 Journal of Politics 901 at 905-906 [Tate & Sittiwong] (an early study examining a range of judicial traits including regionalism, partisanship, religion, and career experience); Donald R. Songer & Susan W. Johnson, “Judicial Decision Making in the Supreme Court of Canada: Updating the Personal Attribute Model” (2007) 40:4 Canadian Journal of Political Science 911 [Songer & Johnson, “Judicial Decision”]; C.L. Ostberg, Matthew Wetstein & Craig R. Ducat, “Leaders, Followers, and Outsiders: Task and Social Leadership on the Canadian Supreme Court in the Early Nineties” (2004) 36:3 Polity 505; Songer, Judicial Behaviour, supra note 92 at 141-142.
example, female judges were more likely to vote in favor of rights claimants and economic underdogs and less likely to vote for criminal defendants.
Australia, another country with the prime minister in control of appointments, lies between the united States and Canada in terms of the judges following their political views. As we saw, Weiden looked at judges in these three countries, measuring the views of the judges by a score based on how each judge was viewed in newspaper editorials at the time of appointment. He compared this newspaper score with how likely the judge was to strike down a law, classifying the laws as either liberal or conservative. Judges in each country tended to vote in line with their political views (as measured by the newspaper scores) to some extent—that is, conservative judges were more likely to strike down liberal than conservative laws. However, judges on the uS Supreme Court were more likely to be influenced by these political views than those on the Australian High Court, who in turn were more influenced than judges on the Supreme Court of Canada.
Judges in the uK appear to be even less influenced in their decisions by political views. Much of the evidence relates to the House of Lords when the executive appointed judges, as opposed to the uK Supreme Court under the committee process. The Law Lords did differ in permissiveness toward the state in cases that challenge state actions; some were much more likely to uphold the state action than others. They also differed in ideal points—that is, when we look at how a judge decided, we can position him on a line with the other judges depending on whether he tends to agree or disagree with each judge. However, the judges may follow their political views less than other potential influences such as norm of professionalism or a greater or lesser willingness to dissent, although these norms may hide rather than replace differences in attitudes. As in other countries, judges on the UK Supreme Court may be influenced when making decisions not just by their political views but by other personal values. Although there is limited work on the issue, there is then less support for the presence of political voting by judges in the uK than in the united States, Canada, or Australia.
So far we have seen that where political actors control appointments, the judges tend to be influenced by their political views, though the extent of the influence varies by country. Is this also the case when a committee selects judges, as in Israel? Although justices on the Israeli Supreme Court did differ in how they voted in different areas, the differences were “not shockingly broad" The individual justices differed in the extent to which they favored the defendant in criminal cases. Further, in a random selection of freedom of religion cases before the Israeli Supreme Court between 1985 and 2008, justices who are religiously observant were more likely to support a freedom of religion claim. Moreover, female justices were more likely to vote for the defendant than their male counterparts, though the difference was very small and seemed to depend on one justice. The Israeli example therefore shows some attitudinal voting even where judges play a significant role in the appointment process.
If we use our data to compare across countries, one measure of whether judges rely on their personal preferences is their consistency in voting across areas of law. It is not a perfect measure as a judge may have other reasons for consistency, such as reliance on a particular doctrinal or interpretative approach to legal questions that is not reflected in what we would normally think of as ideology or personal preferences. Figure 3-5 shows the relationship between how judges vote across criminal, civil rights, and economic cases using long-term liberal voting rates in each of these areas. The line in each individual graph is the line that best fits how the judges on the particular court compare across the areas of law. An upward-sloping line means that there is a positive correlation between the votes of the judges in the different areas. If the judges voted exactly the same in both areas of law in the graph, the line would be the 45-degree line.
On this measure again, the uS Supreme Court justices clearly lead the pack with the most consistency in voting between each of these areas, with the lines of best fit becoming successively flatter as we move from the united States through to India, indicating less of a positive relationship. We may expect to find the highest correlation in voting for each judge between criminal and civil rights cases, and that was true for each of the countries. The correlation between criminal and civil rights cases for uS Supreme Court justices (with a correlation coefficient of 0.93) was far higher than for Australian (0.58), Canada (0.57), or the uK (0.5), with Indian judges having almost no correlation between votes in this area (0.03). The uS justices also ranked first ahead of Australian for the correlation between their voting in both economic and civil rights cases and economic and criminal cases. Canada and the uK were fairly far behind in both these comparisons, with again Indian judges evincing almost no correlation between votes.
Judges’ votes in these areas then appear to be correlated in some courts. Does this relationship continue to hold if we control for (hold constant) other factors that may affect voting, such as how the lower court decided the issue before it came to the high court? We need to control for such variables as otherwise an apparent correlation may be misleading. In the case of the direction of the lower court decision, for example, it may be that judges overall tend to reverse lower court decisions. We would want to know a judge’s liberal voting rate, taking into account the fact that in some cases they are reacting to liberal or conservative lower court decisions.
Figure 3-5 Scatterplots of long-term liberal voting rates for judges across Criminal,
Civil Rights, and Economic cases, 1970 to early 2000s. uS Supreme Court judges had the highest correlation between their votes across all three of these comparisons, followed by Australian High Court justices. Judges on the Supreme Court of India showed almost no correlation in their votes across areas of law.
Figure 3-6 Predicted liberal voting rate in non-criminal cases (percent) based on long-term liberal voting rates in criminal cases (percent), 1970 to early 2000s. The predictions are controlling for the direction (liberal or conservative) of the lower court decision. Justices on the uS Supreme Court are much more consistent across criminal and all non-criminal cases than are justices on the Supreme Court of Canada. India, the uK, and Australia are not included as the relationship between lifetime voting on criminal cases and voting on all other cases was not statistically significant.
Figure 3-6 shows the liberal voting rate in non-criminal cases that are predicted from the long-term liberal voting rates of judges in criminal cases (controlling for the direction of the lower court decision). The first point to notice is that the only two courts for which there is a statistically significant relationship between lifetime liberal voting rates in criminal cases and voting in all other cases are the uS Supreme Court and the Supreme Court of Canada. We cannot safely
Figure 3-7 Predicted liberal voting rate in civil rights cases (percent) based on long-term liberal voting rates in criminal cases (percent), 1970 to early 2000s. The predictions are controlling for the direction (liberal or conservative) of the lower court decision. Justices on the uS Supreme Court have a greater increase in voting on civil rights cases than justices on the high courts in the uK, Canada, or Australia for equivalent increases in long-term liberal voting rates in criminal cases. India is not included as the relationship between long-term voting on criminal cases and voting on all other cases was not statistically significant.
predict the voting on non-criminal cases from the voting on criminal cases for judges in the UK, India, or Australia. Second, justices on the US Supreme Court have more consistent liberal voting rates between criminal and all other cases than do justices on the Supreme Court of Canada. For the voting to be perfectly consistent, the predicted votes would be on the 45-degree line. For justices on the uS Supreme Court the relationship is positively correlated, with an increase from 40 percent to 50 percent in the long-term liberal voting rate in criminal cases related to about a 7 percent higher liberal voting rate in all other cases. Justices on the Supreme Court of Canada, on the other hand, have less of a close connection between the rates of voting, as indicated by the flatter line. An increase from 40 percent to 50 percent in lifetime liberal voting in criminal cases only corresponds to an increase of only about 1 percent in liberal voting in all other cases.
We could imagine, though, that this relationship is stronger across more controversial areas such as civil rights cases. Figure 3-7 looks at the relationship between long-term liberal voting in criminal cases and how judges vote in civil rights cases (controlling for the direction of the lower court decision). Justices on the uS Supreme Court have the greatest increase in predicted liberal voting in civil rights cases for equivalent increases in long-term liberal voting rates in criminal cases. An increase in long-term liberal voting rates in criminal cases from 40 to 50 percent leads to an increase in liberal voting in civil rights of about 9 percent for the uS Supreme Court (51 to 60 percent) but only 6 percent for the Supreme Court of Canada (49 to 55 percent) and about 7 percent for the uK and Australia (about 43 to 50 percent and 47 to 54 percent, respectively). So justices in all these countries (other than India) have a strong positive connection in liberal voting rates between these two areas of law. In the united States judges who are conservative on criminal cases tend to be equivalently conservative in civil rights cases whereas judges who are very liberal in criminal cases are also very liberal for civil rights. For the other countries, judges who are very conservative in criminal cases tend to be relatively more liberal in civil rights cases, whereas judges who are very liberal in criminal cases tend to be relatively less liberal in civil rights cases. In none of the countries, then, is the liberal voting rate identical across the two areas of law.
In sum, judges in the united States are very consistent in their voting, based on how much of a relationship exists between their votes in criminal cases and other cases. Judges in other countries tend to be less consistent, particularly when we look at the relationship between voting in criminal cases and in all other areas. In the more controversial cases judges who vote liberally in criminal cases also tend to vote liberally in civil rights cases, and judges who tend to vote conservatively also tend to vote conservatively in civil rights cases.
Judges on the courts we study then appear to be influenced to some extent in their decisions by their political views. However, greater involvement by political actors is only weakly correlated with judges being more likely to be influenced by these views, with the main support for such a connection coming from the united States. The united States is an outlier both because its legislative branch has such a large involvement in high court appointments and because it has such a politicized appointments process. In other countries where the legislative branch has little or no power over high court appointments, the influence of political ideology over judicial decision-making is weaker. Again there is correlation but not necessarily causation.
-  For a discussion of issues around measuring ideology, see Lee Epstein, Andrew D. Martin,Kevin M. Quinn & Jeffrey A. Segal, “Ideology and the Study of Judicial Behavior” in JonHanson, ed, Ideology, Psychology, and Law (Oxford: Oxford university Press, 2012) [Epsteinet al., “Ideology”]. The newspaper scores were initially devised by Segal and Cover who codednewspaper editorials at the time of the judges’ appointments on a scale denoting more or lessconservative/liberal (Jeffrey A. Segal & Albert D. Cover, “Ideological Values and Votes of u.S.Supreme Court Justices” (1989) 83 American Political Science Review 557). These scores havebeen adopted by researchers on some other high courts (see e.g. C.L. Ostberg & Matthew E.Wetstein, Attitudinal Decision Making in the Supreme Court of Canada (Vancouver: university
-  Songer, Judicial Behaviour, supra note 92; Songer & Johnson, “Judicial Decision,” supra note103; Susan W. Johnson, Donald R. Songer & №dia A. Jilani, “Judge gender, Critical Mass, andDecision Making in the Appellate Courts of Canada” (2011) 32:3 Journal of Women, Politics &Policy 11; Ostberg & Wetstein, Attitudinal Decision Making, supra note 92 at 120 (“[r]emarkably,all five female justices serving on the Court in the period studied can be found at the liberalend of the civil rights and liberties spectrum, with five of the six most liberal voting patternson the Court”).
-  Weiden, “Judicial Politicization,” supra note 69.
-  Weiden, “Judicial Politicization,” supra note 69 at 340.
-  Arvind & Stirton, supra note 86 at 27.
-  Chris Hanretty, “The Decisions and Ideal Points of British Law Lords” (2012) 43 BritishJournal of Political Science 703 at 703 [Hanretty, “Ideal Points”] (examining ideal points of LawLords between 1969 and 2009 and finding that the ideal points are not strongly related to ideology or to the liberal voting rate of judges).
-  Arvind & Stirton, supra note 86 at 27; Hanretty, “Ideal Points," supra note 108 at 715.
-  Rachel J. Cahill-O’Callaghan, “The Influence of Personal Values on Legal Judgments"(2013) 40:4 Journal of Law and Society 596. O’Callaghan undertook a content analysis of 18 UKSupreme Court decisions and found a connection between certain personal values of judges(such as universalism or traditionalism) and their decisions.
-  Judges were also influenced by political ideology in South Africa. See Glendon Schubert,“Political Culture and Judicial Ideology: Some Cross- and Subcultural Comparisons" (1977)9:4 Comparative Political Studies 363; Glendon Schubert, “Subcultural Effects on JudicialBehaviour: A Comparative Analysis" (1980) 42 The Journal of Politics 952. Schubert foundEnglish-speaking and Afrikaans-speaking judges on the South African Appellate Division,South Africa’s highest court of appeal during the apartheid era, voted very differently. See alsoHaynie, Black and White, supra note 85 at 77 (finding that Afrikaans-speaking judges on theSouth African Appellate Division during the apartheid era who were appointed by the apartheid government had a more conservative voting record than their colleagues).
-  Theodore Eisenberg, Talia Fisher & Issie Rosen-Zvi, “Does the Judge Matter? ExploitingRandom Assignment on a Court of Last Resort to Assess Judge and Case Selection Effects"(2012) 9:2 Journal of Empirical Legal Studies 246 [Eisenberg et al., “Does the Judge Matter"].
-  Eisenberg et al., “Does the Judge Matter," supra note 112.
-  Weinshall-Margel, “Attitudinal," supra note 87 at 570. Weinshall-Margel employed thismethod to measure ideological voting because in Israel, politicians are only a minority in theappointment process, and the discussions of the nominating committee are kept secret, makingmore conventional methods of measuring ideology inapplicable.
-  Eisenberg et al., “Does the Judge Matter," supra note 112 at 248.
-  Joshua Fischman & Tonja Jacobi, “The Second Dimension of the Supreme Court” (2016) 57Wm. & Mary Law Review 1671.
-  Recall that the Indian Supreme Court data was based on a sample of 100 cases for each year,leading to lower numbers of decisions by each judge in the database.
-  Based on logit regressions with vote in non-criminal cases (liberal versus conservative) asthe dependent variable and long-term liberal voting rates in criminal cases and direction of thelower court decision as the independent variables, the coefficients on long-term liberal votingrates in criminal cases were significant at the .01 level for the uS Supreme Court and .05 level forthe Supreme Court of Canada. The results were similar if the analysis also controlled for area oflaw. See Appendix 2 Table 1 for a summary of regression results.
-  note that there is a trade-off in specifying ideology—the party of the appointer is quite acrude measure of ideology but it can be applied across all areas of law whereas judges’ voting incriminal cases can only be applied to non-criminal cases.
-  Based on logit regressions with vote in civil rights cases (liberal versus conservative) asthe dependent variable and long-term liberal voting rates in criminal cases and direction of thelower court decision as the independent variables, the coefficients on long-term liberal votingrates in criminal cases were significant at the .05 level except for India.