Smyth, Russell --- "Explaining Voting Patterns on the Latham High Court 1935-50'" [2002] MelbULawRw 5; (2002) 26(1) Melbourne University Law Review 88

	Home \| Databases \| WorldLII \| Search \| Feedback Melbourne University Law Review

[A number of studies in the United States have examined the reasons for variations in the level of agreement between judges using multiple regression. However, there are no studies of this sort for Australian courts. This study uses multiple regression analysis to explain differences in levels of agreement between the judges on the Latham High Court (1935–50) in terms of ideological differences, personality differences, legal complexity and the backgrounds of the judges. The finding of the study is that personality differences and the complexity of the legal issues in a case are the major reasons for differences in levels of agreement, while ideological differences and the socio-economic background of the judge are not statistically significant indicators.]

INTRODUCTION

A number of empirical studies of courts in the United States have examined either the reasons why judges dissent or why judges write separate concurring opinions rather than signing on to the majority opinion. These studies seek to explain consensual and non-consensual behaviour in terms of the relative importance of factors such as the social background of the judges, ideological and personal differences between the judges, and the institutional structure of the court.[2] There has, however, been little attempt to examine the relevance of findings from these studies of courts in the United States for courts in other countries that often have different institutional settings. As Atkins[3] and Tate[4] have argued on several occasions, social scientists have not been aggressive enough in examining the generalisability of most models of the judicial process that have emerged in the American context. This, in turn, has impeded attempts to build more general cross-national theories of the socio-politics of judicial decision-making.

There have been a few attempts by Blackshield,[5] Douglas[6] and Schubert[7] to analyse judicial decision-making in Australia using scalogram techniques. In the 1960s and 1970s, scalogram analysis was a popular quantitative methodology for examining the influence of background variables (such as political beliefs, religion and social upbringing) on judicial decision-making. However, there are various problems with using scalogram analysis. One of the main limitations is that it is impossible to consider the statistical significance of individual factors in explaining voting patterns. For this reason, over the last 25 years, as statistical methods have advanced, more precise statistical techniques such as multiple regression have been substituted for scalogram analysis in studies of judicial decision-making in the United States. In spite of the now widespread use of multiple regression analysis to examine differences in the level of agreement between judges in the United States, there are no studies of this sort for Australian courts.

This article examines the factors that determined voting patterns on the High Court in cases reported in the Commonwealth Law Reports during the period in which Sir John Latham was Chief Justice of the Court. This contributes to existing literature on the High Court in two ways. First, apart from the scalogram studies published in the 1960s and 1970s, there are few studies generally of the High Court which attempt to analyse decision-making from either an economic, political or sociological perspective.[8] This represents a significant deficiency in our understanding of how the High Court works on the eve of its centenary. Second, the history of the High Court and, in particular, the High Court under Latham, has received little research.[9] This is somewhat perplexing given that the Latham Court represents one of the most interesting periods in the Court’s history, in which there were both important constitutional cases of national significance and strong personalities on the Court who decided those cases.[10]

The data collected on voting patterns in the High Court covers the period from Sir John Latham’s appointment as Chief Justice in 1935 to the resignation of Sir George Rich and Sir Hayden Starke in 1950, shortly after the defeat of the Chifley Labor government. Specifically, the study uses multiple regression analysis to examine the extent to which differences in the levels of agreement between different pairings of judges over this period can be explained in terms of proxies for ideological differences, personality differences, the complexity of the case and the judges’ social backgrounds. The conclusion that emerges is that personality differences and the complexity of the case are the major reasons for differences in levels of agreement, while ideological differences and socioeconomic background are not statistically significant indicators.

This article begins by examining the theoretical basis for the study and discusses the empirical literature regarding courts in the United States that has tested the conceptual framework. On the basis of the theoretical framework, previous empirical literature in the United States and the different institutional context in which courts operate in Australia, four hypotheses are formulated to explain consensual and non-consensual judicial behaviour. Part III discusses how the data was collected and the methodology used in the study. The results from multiple regression analysis are presented in Part IV. Part V offers some concluding comments, including consideration of the limitations of the findings, and some suggestions for further research.

II EXPLANATIONS FOR CONSENSUAL AND NON-CONSENSUAL BEHAVIOUR

A Conceptual Framework

Ulmer suggests that a ‘judicial decision can be viewed as a response to certain stimuli after they have been filtered through a set of system properties.’[11] He suggests that the stimuli are of two sorts: (a) primary stimuli, which flow from the case being litigated; and (b) secondary stimuli, which are factors that ‘muddy’ or ‘modify’ the decision that might be made if it were based purely on primary stimuli.[12] One form of secondary stimulus is the environmental context in which the decision is made. This includes factors such as leadership exercised by the Chief Justice, personality differences on the court and the presence or absence of consensual norms (such as exchanging draft judgments or having a post-hearing judicial conference).

Other secondary stimuli include the ideological predisposition and social background of the judge. The social background of the judge, such as class origins, religion and schooling, as well as political beliefs, help to shape the judge’s role-perceptions and values which in turn shape behaviour. Of course, these factors are often interdependent. For instance, class origins might help to mould one’s political views later in life. The environmental context in which the decision is made might influence the extent to which behaviour consistent with the judge’s predisposition occurs.[13] To illustrate this point, assume a judge has a predisposition to dissent, or write separate judgments, because of their social background.[14] If the judge is a member of a collegial court, or if the Chief Justice exercises strong leadership, this predisposition might be tempered. However, if there are strong personality differences on the court, or an absence of consensual institutional norms, a predisposition towards dissent or individualism might be magnified. Finally, the secondary stimuli (social background and ideological position filtered through the environmental context in which the decision is made) modify the purely case-related factors, including the complexity of the issues, in producing voting patterns.

Many of the early studies of courts in the United States that attempted to explain consensual and non-consensual behaviour focused primarily on the social background of judges. More recent studies, however, have recognised that judicial decision-making is more complex and have attempted to examine the importance of social background as well as other primary and secondary stimuli in explaining judicial voting behaviour.[15] As Tate puts it:

While recent studies in the United States have taken a more holistic approach to explaining judicial decision-making, the actual factors that have been tested vary from study to study on a fairly ad hoc basis. This reflects differences in data availability — between courts and for different periods in a court’s history — as well as variations in methodology between studies. This study is not able to test all of the hypotheses that have been put forward for United States courts but instead tests four separate hypotheses that have been prominent in the existing empirical literature.

B Ideological Compatibility

Hypothesis 1: ceteris paribus, there is a positive relationship between judges’ ideological compatibility and their propensity to write joint judgments.

Ideological preferences are secondary stimuli which help mould the predisposition of the judge. Most of the existing studies for the United States equate the ideological position of the judge with that of the political party that appointed him or her. Many studies have found judges’ party identifications to be among the most important influences on their voting behaviour on the United States Supreme Court, United States Courts of Appeal and at the state Supreme Court level.[17] There is, however, a need for caution when considering the relevance of the results from the United States studies for the High Court of Australia. The results of these studies may not translate directly into the different institutional setting of the High Court. Previous researchers have made the point that the High Court is not as politicised as the United States Supreme Court.[18] Therefore, the effect of the political persuasion of the appointing government on voting patterns in the High Court might be expected to be dampened, compared with courts in the United States.

Putting aside the case of Piddington, who never sat on the Court, Sawer suggests that ‘the only appointments which were made with a fairly deliberate attempt to affect the political outlook of the Court were those of [Herbert Vere ‘Doc’] Evatt and [Edward] McTiernan by the Scullin ALP government in 1930’.[19] Winterton expresses a similar view. He states: ‘Political appointments [to the High Court] (in the sense that a judge is appointed because of his or her political opinions, to satisfy political party pressures, or to derive electoral advantage) have been rare’.[20] In contrast, in the United States, at the state level, several states use either partisan or nonpartisan direct election. Thus the judges tend to take on the appearance of ‘politicians in robes’.[21] At the federal level, the United States Senate Judiciary Committee rigorously examines the political credentials of all Supreme Court nominees. In some cases, such as the unsuccessful nomination of Robert Bork to the United States Supreme Court in 1987, the President’s preferred candidate has been rejected on party political grounds. The importance of partisan political considerations in the United States is also reflected in the telling statistic that, of the 216 judges appointed to the United States Courts of Appeals between the Johnson and Reagan administrations, 96 per cent were drawn from the same political party as the administration.[22]

C Complexity of the Case

Hypothesis 2: ceteris paribus, there is a negative relationship between the complexity of the legal issues in a case and judges’ propensity to write joint judgments.

The complexity of the legal issues in a case is a primary, or case-related stimulus. In cases where there are complex legal issues, it is more likely that judges will have divergent views about both the outcome of the case and the reasons for the outcome. In these cases, a judge is more likely to want to write a separate judgment expressing their view on the issues. As Sir Harry Gibbs puts it, ‘it is not wise to have only one judgment in an appellate court dealing with an important question of law’.[23] This is because ‘sometimes a joint judgment may lead to compromise, or to the omission of something that might have been useful to state, but that does not command general agreement.’[24] The result, as Justice Ginsburg of the United States Supreme Court describes it, is that ‘[h]ard cases do not inevitably make bad law, but ... often they produce multiple opinions’.[25] Studies of the United States Supreme Court are consistent with the hypothesis that there is a negative relationship between the complexity of the legal issues in a case and judges’ propensity to sign on to the majority opinion.[26]

D Personality Differences

Hypothesis 3: ceteris paribus, particularly on an acrimonious Bench, judges who have personal differences of opinion are less likely to write joint judgments.

The harmony, or lack of harmony, on the Bench is an environmental factor which modifies a judge’s predisposition towards deciding the case in a certain fashion. As Douglas puts it, ‘personal factors have some role to play, especially where there is no obvious policy issue at stake. Prestige, friendship and antipathies may influence the way judges respond to their fellow judges’ ideas.’[27] This was a particularly important factor during the period in which Latham was Chief Justice, primarily because of personal animosities between Starke and the other members of the Court, especially Evatt. Starke’s antipathies were manifest in a range of petty acts. For example, Crockett records that ‘Starke enjoyed seeing [Evatt] distraught as he blew cigar smoke into his face.’[28] Cowen notes: ‘It was no secret that this was not an harmonious Bench, and there were unhappy personal relationships between some of its members. Starke was a formidable personality and he made no attempt to conceal his personal animosities.’[29] Lloyd agrees with this assessment, stating that ‘deeply-ingrained personal antagonism ... permeated the Bench’[30] and that there were ‘virtually irreconcilable personality differences’.[31] He claims that Starke’s personal relationship with Evatt dominated the internal workings of the Court.[32]

Starke was also hostile towards Dixon, McTiernan and Rich. Starke thought that Sir Owen Dixon exerted undue influence over Evatt and McTiernan, whom he caricatured as ‘parrots’. For example, in a letter to Latham, Starke complained: ‘I am quite convinced that Evatt pays no attention to the facts and is merely a parrot. ... Dixon suddenly alters his mind and to me [gives] a most confusing judgment and the parrots at once agree’.[33] Starke accused Dixon of playing on this influence. In another letter to Latham, Starke stated: ‘I blame [Dixon] a good deal for he angles for their support and shepherds them into the proper cage as he thinks fit.’[34] Starke, however, saved some of his most venomous criticisms for Rich’s legal abilities. For instance, when Rich was appointed to the Judicial Committee of the Privy Council, Starke asserted: ‘Rich will be like a dog with two tails. ... I thought the Privy Councillorship was reserved for those who had rendered distinguished political, judicial or other services. It is a pity to degrade the rank by such an appointment.’[35] Starke’s personal animosities impeded Latham’s attempts to foster institutional practices designed to build consensual norms, such as the exchange of draft judgments. According to Lloyd, Evatt and Starke refused to exchange draft judgments and Starke normally refused to supply Dixon with copies of draft judgments.[36]

E Social Background

Hypothesis 4: ceteris paribus, judges who have a similar socioeconomic background and who share similar upbringings are more likely to have a shared system of values, and are therefore more likely to write joint judgments.

Ulmer describes the hypothesised causal relationship between social background and judicial decision-making. He states that

Schmidhauser was among the first to argue that the social background of a judge is related to their subsequent voting behaviour on the Bench.[38] Some subsequent studies have found social background to be an important predictor of voting behaviour.[39] However, overall, in the United States, social background has been found not to be an important factor. After reviewing the literature, Goldman and Sarat conclude: ‘In general, the background-behavior studies of aggregates of judges have not satisfactorily established clear-cut linkages between most background variables and decision making.’[40]

In this study, it was impossible for statistical reasons to test all of the factors that have been examined in the United States context.[41] The study instead focused on three characteristics that are prominent in the empirical literature: (a) the judge’s religious affiliation; (b) whether the judge attended a private, Catholic or government state school; and (c) the occupation of the judge’s father. Studies in the United States have found that religion is the background variable with the greatest impact on judges’ voting patterns. Ulmer finds a positive correlation between being Catholic and a propensity to dissent.[42] More generally, religion has been found to be an important factor in cases involving either civil liberties or economics issues, although the magnitude of its impact is greater in civil liberties cases than economics cases.[43]

Whether the judge attended a private or state school tests for the existence of network effects. This is reflected in the ‘old-boys’ network, with private school graduates often coming from conservative backgrounds. As Douglas puts it, ‘old (and current) Public Schoolboys are inclined towards social conservatism and their environment will often be one which reinforces and inculcates this’.[44] This was especially so when the Latham Court judges attended school. The other background variable to be tested in this study is the occupation of the judge’s father, which is a proxy for social class. Douglas suggests that

While allocating class origins to judges of the High Court is bound to be somewhat subjective, the majority view in the judicial biography literature suggests that Douglas’s view is somewhat misleading. On the Latham Court there were three judges (Latham, Evatt and McTiernan) with what one might reasonably regard as working class origins.[46] Douglas’s own scalogram analysis of voting patterns on the Latham Court suggests that the decisions of Evatt and McTiernan were more favourable to workers than those of their colleagues.[47] One of the difficulties with scalogram analysis, though, is that it is impossible to isolate whether the sympathies of Evatt and McTiernan were due to their working class backgrounds, their ideological leanings towards the Labor Party, or a combination of both.

The following evidence supports the view that Evatt, McTiernan and Latham should be regarded as being of working class or humble social origins. Schubert treats McTiernan’s father’s occupation (bookkeeper or clerk) as working class.[48] Neumann states that Evatt’s father died when Evatt was seven years old and ‘Evatt and his brothers [were] brought up in humble circumstances by his mother’.[49] Neumann suggests ‘certainly ... Evatt and McTiernan came from socially humble circumstances — perhaps also Latham’.[50] Latham’s father was employed as the Secretary of the Victorian Society for the Protection of Animals and Latham attended Scotch College in Melbourne, which suggests a very middle class background. However, Lloyd notes that Latham was born ‘in relatively humble circumstances’.[51] Fricke’s description of Latham’s upbringing is similar. He states bluntly: ‘Although he was to achieve high office in conservative politics [Latham] was not born with a silver spoon in his mouth. The home of his parents ... was located in the unpretentious Melbourne suburb of Ascot Vale. It was austerely furnished’.[52] As far as his education was concerned, Fricke also points out that Latham began his schooling at Gore Street State School in the distinctly working class suburb of Fitzroy, and that it is was necessary for him to win a scholarship to attend Scotch College and later the University of Melbourne.[53]

III DATA SET AND METHODOLOGY

A Data Set and Time Frame

The study covers all cases reported in the Commonwealth Law Reports from the appointment of Latham as Chief Justice on 11 October 1935 until the retirement of Starke on 31 January 1950. This period spans volumes 54–79 of the Commonwealth Law Reports inclusive. Starke’s retirement was shortly followed by that of Rich on 5 May 1950. Latham did not officially retire until 1952, but he did not take any part in the decision-making of the Court after 11 May 1951. The practical advantage of only considering cases up to the retirement of Starke, and not considering cases in 1951 and 1952, is that in effect we have two stable courts of six Justices over the 15 year period.

From 1935 to 1940, there were six Justices (Latham CJ, Rich, Starke, Dixon, Evatt and McTiernan JJ). In 1940, Evatt retired and was replaced by Sir Dudley Williams. From 1940 to 1950, there was, for practical purposes, also a stable Court of six Justices (Latham CJ, Rich, Starke, Dixon, McTiernan and Williams JJ). Sir William Webb was appointed to the High Court in 1946, bringing the Bench to seven Justices. However, he continued to serve in Japan as President of the International Military Tribunal and did not assume his duties in the High Court until 1948. As a result, he sat in few cases in the sample. This comes close to two ‘natural courts’.[54] In the sample there were 703 reported cases in which there was a Bench consisting of three, four, five or six Justices. A total of 54 of these cases, or less than 8 per cent, involved unanimous decisions. There were 420 cases in which either all of the Justices delivered a separate judgment or there was no clear agreement. This left 229 cases in which there were judicial coalitions consisting of two, three, four or five Justices.

B Constructing the Dependent Variable

To use multiple regression analysis to test the hypotheses set out in the previous section, it was first necessary to construct a dependent variable measuring the level of agreement between different pairings of judges. Consistent with previous studies of judicial voting patterns in Australia and Canada, the main assumption made in constructing the dependent variable was that the reasons for judgment are more important than the outcome of the case.[55] McCormick

There is an important qualification to this statement. Relative to the current High Court, when Latham was Chief Justice a high proportion of the Court’s workload was concerned with straightforward appeals. In such appeals, the ratio decidendi will often have little legal significance. The assumption, however, is still reasonable given that, although it heard relatively more mundane appeals, the High Court, at the time, was the final Australian court of appeal and, as such, its decisions constrained and directed decision-making in the lower courts. This is reinforced in this study by the fact that the data set is restricted to reported decisions, which contain a high proportion of cases with precedent value.

The implication of assuming that the reasons are more important than the outcome is not to ‘treat two Justices as having agreed if they did not join the same opinion, even if they agreed in the result of the case and wrote separate opinions revealing very little philosophical disagreement.’[57] This approach, at first, might appear too restrictive. If a Justice dissents from the outcome of the case it is clear that he or she is not part of the successful coalition that decided the case. However, it might be less obvious that a Justice who writes a separate judgment agreeing with the outcome, but not the reasons, of the other Justices should be treated the same. But this follows once it is accepted that the reasons are more important than the outcome. This is because ‘a separate concurrence by definition accepts the outcome of the appeal (allowed or dismissed) but explicitly distances itself from the reasons with which the majority or plurality decision of the Court justifies the result.’[58] The exception to this statement is a very short concurring judgment in the specific form of ‘I concur with Justices X, Y and Z’. In this study, the concurring Justice was treated as part of a coalition with X, Y and Z because he or she was agreeing with both their reasons for judgment and conclusion.

	Dixon		Rich	McTiernan	Latham	Starke
Dixon	_	64.9 (84)	55.6 (53)	48.8 (93)	1.6 (81)	0.0 (95)
Evatt	64.9 (84)	_	50.0 (40)	65.3 (70)	2.1 (64)	0.0 (91)
Rich	55.6 (53)	50.0 (40)	_	56.8 (51)	17.1 (50)	0.0 (45)
McTiernan	48.8 (93)	65.3 (70)	56.8 (51)	_	24.2 (79)	1.3 (85)
Latham		2.1 (64)	17.1 (50)	24.2 (79)	_	3.9 (65)
Starke		0.0 (91)	0.0 (45)	1.3 (85)	3.9 (65)	_

Note: figures in parentheses are the number of divided Benches on which both Justices sat.

	Dixon	McTiernan	Williams	Latham	Rich	Starke
	_	50.0 (63)	36.6 (51)	17.9 (66)	30.2 (53)	2.0 (62)
McTiernan	50.0 (63)	_	10.3 (80)	53.2 (88)	11.0 (97)	1.2 (95)
Williams	36.6 (51)	10.3 (80)	_	20.7 (71)	35.9 (77)	4.1 (94)
Latham	17.9 (66)	53.2 (88)	20.7 (71)	_	18.8 (83)	2.8 (90)
Rich	30.2 (53)	11.0 (97)	35.9 (77)	18.8 (83)	_	4.0 (88)
Starke		1.2 (95)	4.1 (94)	2.8 (90)	4.0 (88)	_

Note: figures in parentheses are the number of divided Benches on which both Justices sat.

Tables 1 and 2 present explicit agreement ratios for 1935–40 and 1940–50. These depict the number of times that each pairing of judges voted together as a percentage of the number of times each pairing sat on one of the 229 divided Benches in the sample cases.[59] Here, ‘voted together’ means that Justices X and Y participated in a joint judgment, Justice X wrote a short concurring judgment of the form ‘I concur with Justice Y’ or vice versa. The explicit agreement ratios standardise the results for the number of times each pairing of Justices sat together. In Tables 1 and 2, the number of observations on which the percentage is based is given below each explicit agreement ratio in parentheses. It is important to standardise the results because, while Latham was Chief Justice, there were several periodic absences from the Court. For example, Rich was absent for several months after Latham became Chief Justice as he was sitting on the Judicial Committee in London; Latham was Australian Ambassador to Japan in 1940–41; and Dixon was Australian Ambassador to the United States in

C Specification of the Empirical Model

The explicit agreement ratios in Tables 1 and 2 were regressed on a series of proxies to test the hypotheses in the previous section. There were 30 observations altogether (15 observations from 1935–40 and 15 observations from

1940–50). This meant that it was not possible to test all of the hypotheses in the same specification because the number of independent variables would severely restrict the available degrees of freedom. Thus, two separate specifications were run. As there is no a priori theoretical justification for using either a linear or log-linear functional form, the results using both are presented. The first specification regressed the explicit agreement ratios and natural log of the explicit agreement ratios over the two periods on proxies for ideological compatibility, personality differences and the complexity of the case. The second specification regressed the explicit agreement ratios and natural log of the explicit agreement ratios over the two periods on a series of dummies depicting social background. For each pairing, these were whether the judges: (a) had the same religious affiliation; (b) both went to wealthy private schools, or both went to Catholic or state schools; and (c) both had fathers whose occupation can be described as ‘working class’ or ‘middle or upper class’.

COMPLEXITY is the proportion of cases with divided Benches both judges sat on, where either one or both of the judges gave a dissenting judgment;

IDEOLOGY is a dummy variable set equal to one if both judges in the pairing have the same ideological position and is set equal to zero otherwise;

PERSONALITY is a dummy variable set equal to one if Starke is one member of the pairing of judges and is set equal to zero otherwise;

SCHOOL is a dummy variable set equal to one if both judges in the pairing attended a wealthy private school, or if both judges attended a Catholic or state school and is zero otherwise;

OCCUPATION is a dummy variable set equal to one if the occupation of each judge’s father in the pairing can be described as either ‘working class’ or ‘middle class or upper class’ and is set equal to zero otherwise; and

RELIGION is a dummy variable set equal to one if both judges in the pairing have the same religious affiliation and is set equal to zero otherwise.

The signs underneath each of the variables are the expected signs on the coefficient.

The manner in which the explanatory variables were constructed and the expected signs on the coefficients require some justification. The argument for using the proportion of dissenting judgments to denote complexity is that the number of cases in which one or both of the judges give a dissenting judgment is likely to be higher when the legal issues are more complex. Thus, a negative relationship is expected between complexity and the level of agreement. This argument seems logical and has fairly wide support from academic commentators and the extra-judicial statements of judges.[60] The personality variable is a dummy variable set equal to one if Starke is a member of the pairing of judges and zero otherwise. The rationale for using this variable to denote personality differences is that, as discussed above, Starke’s personal animosities were the main reason why the Latham Court was acrimonious. The raw numbers in Tables 1 and 2 suggest that, ceteris paribus, Starke’s explicit agreement ratio with each other member of the Court is very low. Thus we expect a negative sign on the personality variable.

Table 3: Ideological Position and Socioeconomic Background of Judges on the Latham Court 1935–50

	Appointing Party	Religion	School Attended	Father’s Occupation
Latham	Conservative	Methodist	Scotch College	Secretary, Victorian Society for the Protection of Animals
Rich	Labor	Anglican	Sydney Grammar School	Clergyman
Starke	Conservative	Agnostic/ No religion	Scotch College	Medical doctor
Dixon	Conservative	Agnostic/ No religion	Hawthorn College	Solicitor
Evatt	Labor	Anglican	Fort Street High School	Hotel licensee
McTiernan	Labor	Roman Catholic	Sydney Marist Brothers’ High School	Bookkeeper/Clerk
Williams	Conservative	Anglican	Shore Grammar School	Solicitor

Sources: Graham Fricke, Judges of the High Court (1986); Eddy Neumann, The High Court of Australia: A Collective Portrait 1903–1972 (2nd ed, 1973).

Table 3 provides information on the ideological position of each judge (conservative, Labor or neutral) and their social background. In allocating an ideological position to each judge there are two possible approaches. One approach is to label each judge as conservative or Labor according to the political persuasion of the party which appointed him. This is the approach followed in the scalogram studies of the Dixon High Court by Schubert.[61] A second approach is to use the political affiliation of the party which appointed the judge as a starting position, and then make adjustments if the judge’s political position seems to differ. The advantage of the second approach is that it introduces more flexibility into the classifications, particularly in the Australian context, where most appointments are apolitical. The disadvantage is that it introduces subjectivity into the classifications and is thus open to attack on grounds of arbitrariness.

Because of the potentially arbitrary nature of the second approach, the ideological position of each judge was identified with the political persuasion of the appointing government. This avoids the subjectiveness of the second approach, but has its own limitations. In the case of Latham, Evatt and McTiernan, there is no problem because they were all politicians. Classifying Dixon as conservative and Rich as Labor is more contentious. It is widely recognised that Dixon was apolitical. Rich’s views clearly changed over time and, certainly towards the end of his judicial career, were quite close to the conservative Starke. When Evatt and McTiernan were appointed to the High Court by Labor, in controversial circumstances, Crockett records that Rich and Starke ‘were most disdainful’.[62] Fricke suggests that neither Rich nor Starke were ‘fondly disposed towards the Labor governments of the 1940s’.[63] He points out that in cases such as the Airlines Case[64] and the Bank Nationalisation Case,[65] both judges interpreted s 92 of the Constitution in a restrictive fashion designed to curtail government power.[66] Sawer suggests that Rich and Starke delayed their retirement until a conservative government won office because they did not want the Chifley Labor government to replace them.[67]

The problem, though, is that if an objective measure such as appointing party is not used, reasonable minds can differ as to the true ideological position of a judge. The different views on Rich’s political outlook are representative. On one hand, Fricke suggests that ‘Rich was plainly conservative in outlook’.[68] On the other hand, Sawer assesses Rich as ‘non-political’.[69] Fitzhardinge expresses a similar view of Rich as ‘a lawyer’s lawyer with no interest in politics’.[70] This difference of opinion among legal scholars reflects the fact that Rich’s views on the Constitution shifted over a long career spanning 37 years. Marr describes Rich as the ‘weather vane’ on s 92.[71] Fricke recognises that ‘at different times in the Court’s history, and for extended periods, Rich was able to sign joint judgments with both the centralist Isaacs and the States’-righter Gavan Duffy’.[72]

In terms of religious affiliation, there were three Anglicans (Evatt, Rich and Williams), one Methodist (Latham), two judges who were agnostics or had no professed religion (Dixon and Starke) and one Roman Catholic (McTiernan). Following Schubert, the three Anglicans were grouped with Latham as having the same religious affiliation,[73] and Dixon and Starke were paired together as having no religion. Five of the judges attended wealthy private schools. The exceptions were McTiernan, who attended a Catholic school, and Evatt, who attended a government school. One approach would be to group all of the judges except Evatt together as attending private schools. The shortcoming of doing this, however, is that there would be little variance in the schools comparison, which almost guarantees an insignificant variable.[74] For this reason Latham, Rich, Starke, Dixon and Williams were grouped together as attending wealthy private schools, leaving Evatt and McTiernan as the exceptions. For the reasons discussed earlier, the occupations of Dixon, Starke, Rich and Williams’ fathers were treated as middle or upper class. The occupations of Latham, Evatt and McTiernan’s fathers were treated as working class.

IV RESULTS

Equations (1) and (2) were estimated using ordinary least squares. Columns (1) and (2) of Table 4 contain the results for equation (1) using a log-linear and linear functional form respectively. The results in terms of the signs on the coefficients and statistical significance are similar. The adjusted coefficient of determination (Adjusted R2) is slightly higher using a log-linear functional form. As a measure of goodness of fit, the adjusted coefficient of determination (both above 0.60) is fairly impressive, particularly given the essentially ad hoc nature of the regressions. The F-statistic is also reported for each specification in Table 4 and it is significant at 1 per cent in each instance. As a result, we can reject the null hypothesis that the true slope coefficients are simultaneously zero. Table 4 also presents the results of diagnostic testing for multicollinearity. The tolerance levels and variance inflation factors are both around 1, suggesting the results are not affected by multicollinearity. The condition index is about 7.4. Gujarati suggests that if the condition index is less than 10 there is no evidence that multicollinearity is a problem.[75]

Table 4: Regression Analysis of the Effect of Ideology, Complexity and Personality Differences on Voting Behaviour on the Latham Court 1935–50

	(1)	(2)	Collinearity Statistics
Constant	4.497* (9.846)	63.480* (8.325)		Variance Inflation Factors
IDEOLOGY	0.100 (0.335)	0.141 (0.028)
COMPLEXITY	-7.293* (3.351)	-160.436* (4.414)	0.913	1.096
PERSONALITY	-2.230* (6.795)	-24.560* (4.482)	0.908	1.101
F-statistic	25.768+	18.723+
N		30
Adjusted R2	0.719
Condition Index			7.398

Notes: The first equation reports results using a log-linear functional form. In (1) the dependent variable is the natural log of the explicit agreement ratio. The second equation reports results using a linear functional form. In (2) the dependent variable is the explicit agreement ratio.

* Indicates that the coefficient is significant at the 0.01 level of significance.

+ Indicates that the F-statistic is significant at the 0.01 level of significance.

The results in Table 4 provide strong support for the second and third hypotheses. The coefficients on the COMPLEXITY and PERSONALITY variables are negative and significant at 1 per cent in each specification. The results for COMPLEXITY mean that, for the period of the study, there was a statistically significant inverse relationship between the level of agreement between the judges and the complexity of the legal issues in the case proxied by the proportion of dissents. The results for PERSONALITY confirm that Starke had a statistically significant negative effect on the levels of agreement on the Latham Court. Irrespective of whether this was due to Starke’s personal antipathy towards several of his colleagues or what Adam describes as his rugged individualism,[76] Starke’s reluctance to write joint judgments clearly impeded attempts to build consensual approaches to decision-making. On the basis of the results in Table 4, the first hypothesis is rejected. IDEOLOGY has the correct sign, but is statistically insignificant using both linear and log-linear functional forms.

Table 5: Regression Analysis of the Effect of Socioeconomic Background on Voting Behaviour on the Latham Court 1935–50

	(1)	(2)	Collinearity Statistics
Constant	2.616* (4.796)	29.106* (3.657)
RELIGION	-0.100 (0.139)		0.905	1.105
OCCUPATION	-0.138 (0.229)		0.950	1.052
SCHOOL	-0.361 (0.598)	-9.329 (1.060)	0.950	1.052
F-statistic	0.154	0.980
N	30	30
Adjusted R2	-0.096	0.102
Condition Index			3.659

* Indicates that the coefficient is significant at the 0.01 level of significance.

Columns (1) and (2) of Table 5 contain the results for equation (2) when the explicit agreement ratio and natural log of the explicit agreement ratio were regressed on the social background variables. The diagnostics show that multicollinearity is not a problem in either regression. On the basis of the results in Table 5, hypothesis four cannot be accepted. In each specification, each of the background variables is statistically insignificant. The adjusted coefficient of determination (Adjusted R2) is very low and the F-statistic is insignificant in each specification. These results are generally consistent with the majority of studies of courts in the United States, which have failed to establish a causal connection between judges’ socioeconomic background and voting behaviour.

The results for SCHOOL might reflect the fact that, while Evatt and McTiernan did not attend wealthy private schools, both still received an excellent education at Fort Street and the Marist Brothers’ School respectively. The results for OCCUPATION have to be seen in the context that, while Evatt, McTiernan and Latham have been grouped together in this study as having humble social origins, Latham, as discussed above, attended Scotch College and Evatt and McTiernan had highly supportive parents. McTiernan’s family moved to Sydney so that young Edward could receive a good education at the Marist Brothers’. Meanwhile, Evatt’s mother ‘displayed a consuming interest in [his] education’.[77] For instance, Crockett records that Evatt’s mother was instrumental in getting him elevated to a higher grade at Fort Street in recognition of his outstanding performance at school.[78]

V CONCLUSIONS

The results of this study represent a first attempt to explore which factors determine differences in the level of agreement between pairings of judges in Australian courts using multiple regression. For the reasons discussed earlier, multiple regression is a more sophisticated methodology than the scalogram analysis used in the pioneering Australian studies published by Blackshield, Douglas and Schubert in the 1960s and 1970s. The results for party affiliation differ from the United States studies, reflecting the essentially apolitical nature of High Court appointments. Nevertheless, as a first attempt to explore the applicability of hypotheses, which were developed to explain judicial decision-making in the United States, the results in this study provide some support for the more general application of the ‘American judicial model’. An important qualification to this statement is that the institutional features of decision-making in the High Court are rather different from most of the courts in the United States. The ‘American judicial model’ was developed for United States courts that have conferencing and allocation of majority and minority decision-making. The fact that the High Court does not operate in that fashion restricts the conclusions we can draw about the operation of the Court. Related to this, as discussed earlier, in the majority of cases during the period in which Latham was Chief Justice, all of the Justices delivered separate judgments; therefore this study focuses on a by-product of High Court practice.

There are other qualifications to the general conclusions relating to the preliminary nature of the research in this study and its various limitations. First, the study covers only one period in the High Court’s history: the period Latham was Chief Justice. More research needs to be done on other periods in the Court’s past, and for longer periods of time, to confirm or qualify the findings in this study. It is quite likely that some of the findings might be different for other periods. For example, personality differences are unlikely to have been as important as they were under Latham, certainly in the Court’s more recent past. Second, some of the proxies used in this study are quite crude; in particular, the use of dissenting judgments to proxy complexity. The problem with using dissenting rates is that it incorporates into the independent variable components of the dependent variable. Future studies, particularly if they employ more disaggregated data, could use better proxies for complexity, such as the number of legal issues per case.

Third, the study is restricted to testing four hypotheses formulated around a basic conceptual model of decision-making. Future research could test other more sophisticated hypotheses which could not be tested in this study because of the way the dependent variable was constructed. For instance, some recent studies in the United States have used a game-theoretic approach to test whether judges’ voting patterns are a result of strategic behaviour.[79] This tests the view that, as judges are engaged in long-term interactions with their colleagues, they are likely to adopt tit for tat strategies. In other words, judges are more likely ‘to reward colleagues who have cooperated with them in the past and punish those who have not.’[80] With pooled cross-sectional and time series data over several periods, strategic considerations could be examined by introducing an independent variable measuring explicit agreement with a one-period lag.[81]

There are several more general directions that future research on the political economy of decision-making in Australian courts could take. One approach would be to examine what influences a judge’s decision to write a dissenting judgment. There are a number of studies of this sort for courts in the United States.[82] A second approach might be to examine so-called ‘freshman effects’. A freshman effect occurs where a judge who has prior judicial experience in a lower court exhibits unstable voting behaviour during the first term in which he or she is in office in a higher court. It is hypothesised that this is due to initial bewilderment or disorientation upon assuming the new appointment.[83] There are a range of studies which explore freshman effects on the United States Supreme Court. The studies which have examined the first terms of O’Connor,[84] Scalia,[85] Kennedy[86] and Thomas JJ,[87] have produced mixed results. Australian studies could compare the voting behaviour of High Court judges with prior judicial experience before and after their appointment to the Court. These are just a few of the avenues that future research in this area could pursue. What is clear is that without further studies of courts outside the United States it is difficult to know whether the findings of studies of voting behaviour in the United States can be generalised or whether they are unique to their specific institutional context.

[*] BEc (Hons), LLB (Hons), MEc (Monash), PhD (London); Senior Lecturer, Faculty of Business and Economics, Monash University. This paper is part of an ongoing project investigating citation patterns, voting behaviour and the socio-politics behind decision-making in courts in Australia and New Zealand. The paper has benefited from the helpful suggestions of two referees; all errors are, of course, my own.

[2] Some of the references exploring these issues using data on courts in the United States are: ibid; S Sidney Ulmer, ‘The Analysis of Behavior Patterns on the United States Supreme Court’ (1960) 22 Journal of Politics 629; C Neal Tate, ‘Personal Attribute Models of the Voting Behavior of US Supreme Court Justices: Liberalism in Civil Liberties and Economics Decisions,

[3] See Burton Atkins, ‘Party Capability Theory as an Explanation for Intervention Behavior in the English Court of Appeal’ (1991) 35 American Journal of Political Science 881, 881–2; Burton Atkins, ‘A Cross-National Perspective on the Structuring of Trial Court Outputs: The Case of the English High Court’ in John Schmidhauser (ed), Comparative Judicial Systems: Challenging Frontiers in Conceptual and Empirical Analysis (1987) 143, 143–4.

[4] C Neal Tate, ‘The Methodology of Judicial Behavior Research: A Review and Critique’ (1983) 5 Political Behavior 51; C Neal Tate, ‘Judicial Institutions in Cross-National Perspective: Toward Integrating Courts into the Comparative Study of Politics’ in John Schmidhauser (ed), Comparative Judicial Systems: Challenging Frontiers in Conceptual and Empirical Analysis (1987) 7, 7–8.

[5] See A R Blackshield, ‘Quantitative Analysis: The High Court of Australia, 1964–1969’ (1972) 3 Lawasia 1; A R Blackshield, ‘Judges and the Court System’ in Gareth Evans (ed), Labor and the Constitution 1972–1975 (1977) 105; A R Blackshield, ‘X/Y/Z/N Scales: The High Court of Australia 1972–1976’ in Roman Tomasic (ed), Understanding Lawyers (1978) 133.

[6] See R N Douglas, ‘Judges and Policy on the Latham Court’ (1969) 4 Politics 20.

[7] See Glendon Schubert, ‘Opinion Agreement among High Court Justices in Australia’ (1968) 4 Australia and New Zealand Journal of Sociology 2; Glendon Schubert, ‘Political Ideology and the High Court’ (1968) 3 Politics 21; Glendon Schubert, ‘Judicial Attitudes and Policy-Making in the Dixon Court’ (1969) 7 Osgoode Hall Law Journal 1; Glendon Schubert, ‘The Dimensions of Decisional Response: Opinion and Voting Behavior of the Australian High Court’ in Joel Grossman and Joseph Tanenhaus (eds), Frontiers of Judicial Research (1969) 163.

[8] Important general exceptions are Brian Galligan, Politics of the High Court: A Study of the Judicial Branch of Government in Australia (1987); Geoffrey Sawer, Australian Federalism in the Courts (1967); Eddy Neumann, The High Court of Australia: A Collective Portrait

1903–1972 (2nd ed, 1973); David Solomon, The Political High Court: How the High Court Shapes Politics (1999). More recent specific exceptions include Mita Bhattacharya and Russell Smyth, ‘The Determinants of Judicial Prestige and Influence: Some Empirical Evidence from the High Court of Australia’ (2001) 30 Journal of Legal Studies 223; Russell Smyth, ‘The “Haves” and the “Have Nots”: An Empirical Study of the Rational Actor and Party Capability Hypotheses in the High Court 1948–99’ (2000) 35 Australian Journal of Political Science 255; Russell Smyth, ‘“Some Are More Equal than Others” — An Empirical Investigation into the Voting Behaviour of the Mason Court’ (1999) 6 Canberra Law Review 193.

[9] The published literature specifically on the Latham Court consists of Clem Lloyd, ‘Not Peace but a Sword! — The High Court under J G Latham’ [1987] AdelLawRw 9; (1987) 11 Adelaide Law Review 175; Zelman Cowen, Sir John Latham and Other Papers (1965) ch 2; Douglas, ‘Judges and Policy on the Latham Court’, above n 6.

[10] Important cases decided toward the end of Sir John Latham’s period as Chief Justice were Bank of NSW v Commonwealth [1948] HCA 7; (1948) 76 CLR 1 (‘Bank Nationalisation Case’), British Medical Association v Commonwealth [1949] HCA 44; (1949) 79 CLR 201 (‘Pharmaceutical Benefits Case’) and Communist Party of Australia v Commonwealth [1951] HCA 5; (1951) 83 CLR 1 (‘Communist Party Case’). For an overview of the personality clashes between the Justices of the Latham Court, see Lloyd, above n 9, 178–87.

[14] The actual relationship between social background and propensity to dissent might be quite complicated. The evidence on whether there is a causal relationship is mixed. In a seminal contribution in this area, Schmidhauser has argued that judges of the United States Supreme Court from working class backgrounds are more likely to be frequent dissenters than judges from middle class backgrounds, and that regional background is also correlated with a propensity to dissent: see John Schmidhauser, ‘Stare Decisis, Dissent, and the Background of the Justices of the Supreme Court of the United States’ (1962) 14 University of Toronto Law Journal 194. This has been challenged in more recent research: see below n 40 and accompanying text. The relationship between social background and voting patterns in general is further discussed later in the text: see below Part II(E).

[15] An example of a fairly recent study which adopts a broad theoretical approach to explaining consensual and non-consensual behaviour on the Burger United States Supreme Court

[17] See, eg, Glendon Schubert, Quantitative Analysis of Judicial Behavior (1959); Stuart Nagel, ‘Political Party Affiliations and Judges’ Decisions’ (1961) 55 American Political Science Review 843; Sheldon Goldman, ‘Voting Behavior on the United States Courts of Appeals,

[34] Letter from Starke to Latham, 2 December 1938, cited in Lloyd, above n 9, 181.

[35] Letter from Starke to Latham, 8 January 1936, cited in Lloyd, above n 9, 181.

[38] John Schmidhauser, ‘The Justices of the Supreme Court: A Collective Portrait’ (1959) 3 Midwest Journal of Political Science 1.

[39] Many of the pioneering studies (prior to 1970) which found social background to be an important predictor in explaining voting patterns are discussed in Ulmer, ‘Dissent Behavior’, above n 1, 585–7. The most comprehensive study of the relevance of social background, which found many aspects of social background to be important in explaining voting behaviour, is Tate, ‘Personal Attribute Models’, above n 2.

[40] Sheldon Goldman and Austin Sarat, American Court Systems: Readings in Judicial Process and Behavior (1978) 374.

[41] The data set contains 30 observations. From a statistical viewpoint, this severely limits the available degrees of freedom in the multiple regression analysis and thus the reliability of the results if the number of explanatory variables is made too large.

[43] Tate, ‘Personal Attribute Models’, above n 2, 358, citing S Sidney Ulmer, ‘Social Background as an Indicator to the Votes of Supreme Court Justices in Criminal Cases: 1947–1956 Terms’ (1973) 17 American Journal of Political Science 622; Kenneth Vines, ‘Federal District Court Judges and Race Relations Cases in the South’ (1964) 26 Journal of Politics 337; Nagel, ‘Multiple Correlation’, above n 17; Goldman, ‘Voting Behavior Revisited’, above n 17.

[44] R N Douglas, ‘Courts in the Political System’ [1968] Melbourne Journal of Politics 36, 47.

[46] See Graham Fricke, Judges of the High Court (1986) 123–42; Neumann, above n 8, 22–34.

[54] A ‘natural court’ is a term used in the literature in the United States to refer to the situation where the same Justices interact for the whole research period. See, eg, the discussion in Paul Edelman and Jim Chen, ‘The Most Dangerous Justice: The Supreme Court at the Bar of Mathematics’ (1997) 70 Southern California Law Review 63.

[55] For studies of voting patterns in Australia and Canada which make this assumption, see Smyth, ‘Some Are More Equal than Others’, above n 8; Peter McCormick, ‘Birds of a Feather: Alliances and Influences on the Lamer Court 1990–1997’ (1998) 36 Osgoode Hall Law Journal 339; Peter McCormick, ‘Follow the Leader: Judicial Power and Judicial Leadership on the Laskin Court: 1973–1984’ (1998) 24 Queen’s Law Journal 237; Peter McCormick, ‘The Most Dangerous Justice: Measuring Judicial Power on the Lamer Court 1991–97’ (1999) 22 Dalhousie Law Journal 93.

[59] Tables 1 and 2 and the ensuing analysis exclude the few cases in which Webb J was involved. This ensures two stable ‘natural courts’ of six Justices, which makes the results easier to interpret.

[73] Schubert, ‘Opinion Agreement among High Court Justices’, above n 7, 12.

[74] In an earlier version of this paper, McTiernan was grouped with the judges who attended private schools. In all specifications the SCHOOL variable was statistically insignificant.

[81] Wahlbeck, Spriggs and Maltzman construct an independent variable along these lines: ibid.

[82] See, eg, Dean Jaros and Bradley Canon, ‘Dissent on State Supreme Courts: The Differential Significance of Characteristics of Judges’ (1971) 15 Midwest Journal of Political Science 322; Paul Brace and Melinda Hall, ‘Neo-Institutionalism and Dissent in State Supreme Courts’ (1990) 52 Journal of Politics 54; Saul Brenner and Harold Spaeth, ‘Ideological Position as a Variable in the Authoring of Dissenting Opinions on the Warren and Burger Courts’ (1988) 16 American Politics Quarterly 317.

[84] John Scheb and Lee Ailshie, ‘Justice Sandra Day O’Connor and the “Freshman Effect”’ (1985) 69 Judicature 9.

[85] Thea Rubin and Albert Melone, ‘Justice Antonin Scalia: A First Year Freshman Effect?’ (1988) 72 Judicature 98.

[86] Albert Melone, ‘Revisiting the Freshman Effect Hypothesis: The First Two Terms of Justice Anthony Kennedy’ (1990) 74 Judicature 6.

Melbourne University Law Review