Panel Assignment in Appellate Courts: Strategic Behaviour in the South African Supreme Court of Appeal

Abstract

Numerous studies on judges in various countries have found that judges behave strategically in order to obtain politically favourable case outcomes. In this study, we examine whether chief justices of the South African Supreme Court of Appeal1 strategically assign judges to panels to maximize the political favourability of case outcomes. We confirm what other South African scholars have found, that chief justices do strategically make panel assignments. More critically, our analysis shows that they take into account the cumulative composition of the final panel rather than solely considering the individual judges. Specifically, we find that chief justices attempt to make panel assignments in order to maximize the ideological proximity of a minimum winning coalition on the panel, especially in highly salient cases. We assert that by strategically appointing the panels, the chief increases the likelihood that the minimum number of judges necessary will vote in his preferred direction. Thus, we provide further evidence that judges engage in strategic behaviour during the decision-making process.

Notes

Prior to the 1994 democratic dispensation the court was known as the Appellate Division and was the titular head of the judicial hierarchy. Under the new constitution, the newly formed Constitutional Court now sits at the apex of the judiciary. For consistency we will refer to the court as the Supreme Court of Appeal despite the 31-year timeframe of our analysis.

In 2002 the titles of the heads of courts were switched so that the SCA is now formally headed by a President while the Chief Justice presides over the Constitutional Court. Again, for consistency, we will refer to the head of the SCA as Chief Justice.

It should be noted that recent controversy has arisen over the appointment of members of the JSC, its recent appointments to the Constitutional Court, and allegations of misconduct related to a sitting High Court Judge.

Judges retire at age 70 with 15 years of service but may continue on the bench until age 75 to reach 15 years of service. Judges are also generally obliged until 75 to work for three months a year if asked.

Haynie, S.L., Tate, C.N., Sheehan, R. and Songer, D. (2000), ‘Fitting More Pieces into the Puzzle of Judicial Behaviour: A Multi-Country Data Base and Program of Research’, Grant Proposal National Science Foundation (SES #9975237); and Tate, C.N., Haynie, S.L., Sheehan, R. and Songer, D. (2002), ‘Extending a Multi-Country Data Base and Program of Research’, Grant Proposal National Science Foundation (SES #0137055). This database comprises the decisions of the highest courts of appeal in Australia, Canada, India, the Philippines, South Africa, Tanzania, the United Kingdom, Zambia and Zimbabwe from 1970 to 2000. These data should not be confused with the South African High Courts, which are not included in the data.

In addition to the factors we describe below, prior studies have indicated that panel and opinion assignment may be affected by the legal expertise of the justices (Maltzman, Spriggs & Wahlbeck, 2000). Unfortunately, we do not have a measure of expertise of the SCA justices. Thus, while we are unable to test for its effect, we acknowledge that expertise may affect the assignment behaviour of the Chief Justice.

One potential implication of our argument that the chief justice is primarily concerned with the minimum winning coalition is an expectation of frequent dissent; however, non-unanimous decisions are exceptionally rare in the SCA. The lack of dissension does not necessarily indicate a lack of disagreement. Rather, empirical studies demonstrate that, in deciding to dissent from a case, justices weigh numerous factors, in addition to their level of agreement, including the norms of the court (Epstein, Walker & Dixon, 1989; Haynie, 1992) and workload (Maltzman, Spriggs & Wahlbeck, 2000). Since the SCA has a norm of unanimity and substantial workloads, judges may be more willing to acquiesce to the majority when they find themselves in conflict. This, indeed, highlights the importance of the minimum winning coalition because the chief justice may realise that once he has the critical mass of a panel, the minority judge will defer. As such, the lack of dissent on the SCA does not de-emphasize the potential importance of the minimum winning coalition.

We note that political preferences may be dynamic, in which case a static measure of those preferences may be inaccurate. However, the discipline's understanding of dynamic policy preferences of judges is relatively unexplored and, as such, few measures have been developed to capture potentially changing preference. We, therefore, follow the norm in the field and use a static measure of political preferences.

Criminal cases in the SCA can run the gamut from a perfunctory review of the facts of the case to decisions involving substantive procedural or due process issues. However, we believe our measure captures those cases that represent more substantive claims of the accused because we code only published opinions. Cases noted for publication should represent some new statutory interpretation, alteration of precedent or other substantive issue. Criminal cases also provide sufficient numbers of votes across the judges to ensure a robust measure of a judge's support for the underdog (ie the criminal defendant) as opposed to the top dog (ie the government), the underlying concept generally associated with liberal voting patterns. Since the votes on the merits may influence a chief justice's panel appointments, it may be that our measure is correlated with the appointment in criminal panels. To test for this endogeneity, we ran the model without criminal cases, and the results are the same, with one exception. For the appointment to the entire panel, the 1994 Constitutional Transition variable is positive and significant, meaning the chief justice selected ideologically similar justices for the entire panel. However, this does not affect the criminal issues variable or substantively change the variables in the minimum winning coalition panel. As a result, we believe the variable provides a measure that is both valid and reliable, if not ideal.

Because it would require substantial space, we do not present the ideological scores of each justice on the court. during the period of analysis; however, the data are available on request.

Mahomed was the first Chief Justice of the SCA selected post-apartheid and served from 1998–2000.

Milne served in the period 1988–93.

Furthermore, studies of opinion assignment have shown that chief justices are concerned with non-ideological factors such as workload and judicial expertise (Maltzman & Wahlbeck, 1996a, 1996b, 2004; Maltzman, Spriggs & Wahlbeck, 2000).

Unfortunately in this study it is not possible to include separate controls for workload in the analysis. Since judges are assigned to panels immediately before the start of each session, there is no way to empirically assess whether equalization plays a role in panel assignment, though interviews with clerks anecdotally indicate that workload concerns do indeed influence panel assignment. Furthermore, workload concerns should not play as large a role in panel assignment as they do in opinion assignment since sitting on a panel demands a significantly smaller time commitment than opinion writing.

We recognise that what constitutes salience may change over time. However, we believe, particularly in the context of South Africa, rights and liberties, including the rights of the accused, represent significant issues to which constituents were attentive during apartheid and are attentive in the new constitutional order. During apartheid, racial tensions permeated rights and liberties challenges (Wacks, 1984; Abel, 1995) as well as the perception of fairness in criminal sentencing (see Van Niekerk, 1970). Because criminal cases are the category from which our ideology variable was drawn, separating civil rights and liberties from those cases involving criminal defendants provides a cleaner test of the effect of salience.

Freshman is defined as a judge sitting in their first year on the Court.

For both the analysis examining whole panels and that examining minimum winning coalitions, we code whether the freshman was on the whole panel.

The interim constitution (Constitution of the Republic of South Africa Act 200 of 1993) came into effect in 1994 followed by the passage of the Constitution of the Republic of South Africa Act 108 of 1996. We also tested the effects of the adoption of the new constitutional order by separating our data into pre-constitution and post-constitution and running the analysis on each subsection of the data. We choose not to present the results because the lack of variation in the dependent variable post-1994 rendered the standard errors unstable.

In the first year of the study, nine judges comprised the SCA, and 15 served by 2000. Twenty judges currently serve on the court.

We also ran the model as a fixed effects model with dummy variables for each year in the analysis. The results and significance levels were unchanged; however, the beta coefficients were relatively uninterpretable, so we opted to report the model with the year variable.

Since five-judge panel cases are repeated 10 times while three-judge panel cases are entered three times, potential weighting problems exist; however, we ran individual analyses on three-judge and five-judge panel cases and there was no substantive change in the findings. Thus, we feel that grouping three and five-judge panels into a single analysis does not produce any substantive problems.

Because of space constraints we are unable to present which panels and minimum winning coalitions repeated, which total in the thousands; however, the data are available on request.

Additional information

Notes on contributors

Kaitlyn L. Sill

Department of Political Science, Pacific Lutheran University, Tacoma, USA.

Stacia L. Haynie

Department of Political Science, Louisiana State University, Baton Rouge, USA. Email: [email protected].