¹ See the discussion in Goldman, Sheldon and Jahnige, Thomas P., The Federal Courts as a Political System (New York: Harper & Row, 1971), pp. 154–171Google Scholar and Murphy, Walter F. and Tanenhaus, Joseph, The Study of Public Law (New York: Random House, 1972), pp. 116–149Google Scholar. Also see Schubert, Glendon, “Judicial Process and Behavior, 1963–1971,” in Political Science Annual, Vol. III, ed. Robinson, James A. (Indianapolis: Bobbs-Merrill, 1972), pp. 73–280Google Scholar.

² See the citations contained in Goldman, Sheldon, “Voting Behavior on the United States Courts of Appeals, 1961–1964,” American Political Science Review, 60 (06, 1966), 375CrossRefGoogle Scholar, n. 6 and Goldman, Sheldon, “Conflict on the U.S. Courts of Appeals 1965–1971: A Quantitative Analysis,” University of Cincinnati Law Review, 42 (No. 4, 1973), 636, n. 3Google Scholar. Also see, Morris, Jeffrey B., “The Second Most Important Court: The United States Court of Appeals for the District of Columbia Circuit” (Ph.D. Dissertation, Columbia University, 1972)Google Scholar and Moseley, William H., “Personal Attitudes and Judicial Role in Judicial Decision-Making: A Study of the United States Courts of Appeals” (Ph.D. Dissertation, University of Hawaii, 1972)Google Scholar as cited in Comprehensive Dissertation Index 1861–1972, Vol. 27 (Ann Arbor: Xerox University Microfilms, 1973), pp. 187, 343Google Scholar; Howard, J. Woodford Jr., “Litigation Flow in Three United States Courts of Appeals,” Law and Society Review, 8 (Fall, 1973), 33–53Google Scholar; J. Woodford Howard, Jr., “Role Perceptions and Behavior in Three U.S. Courts of Appeals,” paper presented at the 1974 Annual Meeting of the American Political Science Association; Burton M. Atkins and Justin Green, “Problems in the Measurement of Conflict on the United States Courts of Appeals,” paper presented at the 1974 Annual Meeting of the American Political Science Association; Atkins, Burton M., “Opinion Assignments on the United States Courts of Appeals: The Question of Issue Specialization,” Western Political Quarterly, 27 (09, 1974), 409–428CrossRefGoogle Scholar; Atkins, Burton M. and Zavoina, William, “Judicial Leadership on the Court of Appeals: A Probability Analysis of Panel Assignments in Race Relations Cases on the Fifth Circuit,” American Journal of Political Science, 18 (11, 1974), 701–711CrossRefGoogle Scholar; Lamb, Charles M., “Warren Burger and the Insanity Defense-Judicial Philosophy and Voting Behavior on a U.S. Court of Appeals,” American University Law Review, 24 (Fall, 1974), 91–128Google Scholar.

³ Goldman, , “Voting Behavior,” pp. 374–383Google Scholar. To aid comparability with the earlier study, much of the original research design and format of the presentation of the findings are employed here.

⁴ In the earlier study, the government regulation of business issue was utilized, and 38 out of the total study population of 78 judges had decided at least five such cases and therefore had scores calculated for that issue. But now only 18 judges out of the total study population of 130 judges decided the minimum number of cases. The separate government regulation issue was therefore dropped but the cases were incorporated in the economic liberalism measure.

⁵ In the earlier study both unanimous reversals of district courts and nonunanimous decisions made up the case population. Because of the relatively small number of nonunanimous decisions in that case population, only one liberalism category (which consisted of only split decisions) was constructed.

⁶ A judge had to decide a minimum of five cases on a particular issue in order for a score to be calculated and utilized. In practice, for most issues, at least three or four times the minimum number of cases were utilized for most judges. There were some judges, however, who did not decide a sufficient number of cases on all issues; thus the N for each issue varied. The private economic issue had the smallest number of judges, 51, for whom scores were calculated. The second, third, fifth, ninth, and District of Columbia circuits accounted for 41 judges on that issue. The fourth, sixth, and seventh circuits contributed the balance. No judge on the first, eighth, and tenth circuits was included on this issue because none decided the minimum number of cases.

⁷ Occasionally, cases are decided by the entire court sitting en banc. During the seven-year period under study, there were only 202 nonunanimously decided en banc decisions accounting for little more than one half of 1 per cent of all cases decided after hearing or submission.

⁸ See, for example, Salisbury, Robert H., “The United States Courts of Appeals for the Seventh Circuit, 1940–1950: A Study of Judicial Relationships” (Ph.D. Dissertation, University of Illinois, 1955)Google Scholar; Downing, Rondal G., “The Federal Courts and Labor Relations Policy, 1936–1954: A Study of Judicial Decision-Making” (Ph.D. Dissertation, University of Illinois, 1956)Google Scholar; Nagel, Stuart S., “Political Party Affiliation and Judges' Decisions,” American Political Science Review, 55 (12, 1961), 843–850CrossRefGoogle Scholar; Vines, Kenneth N., “The Role of Circuit Courts of Appeal in the Federal Judicial Process: A Case Study,” Midwest Journal of Political Science, 7 (11, 1963), 305–319CrossRefGoogle Scholar; Vines, Kenneth N., “Federal District Judges and Race Relations Cases in the South,” Journal of Politics, 26 (05, 1964), 337–357CrossRefGoogle Scholar; Bowen, Don R., “The Explanation of Judicial Voting Behavior from Sociological Characteristics of Judges” (Ph.D. Dissertation, Yale University, 1965)Google Scholar; Dolbeare, Kenneth M., “The Federal District Courts and Urban Public Policy: An Exploratory Study (1960–1967),” in Frontiers of Judicial Research, ed. Grossman, Joel B. and Tanenhaus, Joseph (New York: Wiley, 1969), pp. 373–404Google Scholar; Atkins, Burton M., “Decision-Making Rules and Judicial Strategy on the United States Courts of Appeals, Western Political Quarterly, 25 (12, 1972), 626–642CrossRefGoogle Scholar; Walker, Thomas G., “A Note Concerning Partisan Influences on Trial-Judge Decision Making,” Law and Society Review, 6 (05, 1972), 645–649CrossRefGoogle Scholar; Walker, Thomas G., “Behavioral Tendencies in the Three-Judge District Court,” American Journal of Political Science, 17 (05, 1973), 407–413CrossRefGoogle Scholar. Wildhorn, Sorrel and Greenwood, Robert, Prosecution of Adult Felony Defendants: A Policy Perception (Santa Monica, California: Rand Corporation, 1973)Google Scholar.

⁹ A positive correlation (.24) was found between circuit dissent rate and size of court (i.e., the number of judges serving on the appeals court), suggesting some tendency for the larger circuits to be more contentious than the smaller ones. A negative correlation (−.44) was found between circuit dissent rate and per judgeship terminated appeals rate, suggesting that circuits with more dissent tended to be those that were less efficient in the disposition of their business. It might be thought that circuits with the heaviest workload pressures might be those with a tendency to dissent less but the correlation analyses lent little support for this hypothesis (the correlation between dissent rate and per judgeship civil appeals rate was −.09, the correlation between dissent rate and total appeals filed per judgeship was −17). Per judgeship rates were taken from Management Statistics for United States Courts (Washington, D.C.: Administrative Office of the United States Courts, 1972)Google Scholar. Sources of dissent are explored in Goldman, “Conflict,” pp. 637–642. A major finding reported there is that there appears to be a diversity of sources of conflict and that this diversity is probably indicative of regional, social, and economic differences among the circuits. Yet there is no basis to assume that regions are so distinct that cross-circuit comparisons of the same issue are unwarranted.

¹⁰ For more on this point, see Goldman, , “Conflict,” p. 639, n. 10Google Scholar.

¹¹ Note that the correlations were higher than those found in the earlier study. It is likely that the current case population of exclusively nonunanimous decisions highlights the differences in voting tendencies while the earlier study's inclusion of certain unanimously decided cases (without controlling for panel composition) probably took the edge off those differences. Recall also that the political liberalism issue subsumes criminal procedures as well as civil liberties (see Table 1) so that correlations between political liberalism and criminal procedures as well as civil liberties are artificially high. This is also true for the correlations between criminal procedures and activism as criminal procedures cases involving state prisoners were also coded on the activism issue. Note the overlap with economic liberalism and the separate labor, injured persons, and private economic categories.

¹² It might well be that political attitudes of the judges toward the federal government in such matters may be changing in that the image of benevolent and generous Uncle Sam collecting taxes to, among other things, redistribute wealth and provide benefits to the less fortunate has faded in the wake of Vietnam and the Nixon presidency. The federal government perhaps now is seen as topdog in economic struggle with nongovernmental litigant underdogs.

¹³ This is a finding contrary to the earlier study in which negative correlations were found between dissent and voting on civil liberties and on labor issues. See Goldman, , “Voting Behavior,” pp. 379–380Google Scholar.

¹⁴ The question that surely comes to mind is what about the overwhelming proportion of appeals court cases decided unanimously. Do these decisions by virtue of their unanimity undermine the attitude explanation? One key to an answer is found in the use of shifting three-member appeals court panels. Panels consisting of attitudinally similar judges would not be expected to have much dissent absent fundamental personality clashes. Indeed, Burton Atkins has demonstrated for the District of Columbia Circuit that panels consisting of judges whose en banc behavior was liberal in criminal cases (i.e., pro criminal defendant) are more likely to produce liberal results considering only and all their panel decisions than those panels dominated by previously identified conservatives in criminal cases. See Atkins, “Decision-Making Rules.” Dissent rates of judges have also been shown to be considerably less when the judges on the panels are attitudinally similar than when they are on panels with a majority of attitudinally dissimilar (as identified by en banc behavior) colleagues. See Atkins, Burton M., “Judicial Behavior and Tendencies Towards Conformity in a Three Member Small Group: A Case Study of Dissent Behavior on the United States Court of Appeals,” Social Science Quarterly, 54 (06, 1973), 41–53Google Scholar. Also note that the earlier appeals courts study by Goldman utilized both unanimous and nonunanimous decisions. The unanimous decisions studied were district court reversals. The earlier study utilized from between approximately 20 per cent to 35 per cent of the cases decided by each circuit during the time period examined and the importance of attitudes was one of the prominent findings. Thus attitudes are clearly important not only for nonunanimous decisions but also for a portion of unanimous decisions. But there are, of course, other reasons for consensus. Consensus may result because the precedents, statutes, and facts of cases are reasonably clear-cut pushing attitud inally dissimilar judges towards the same results. See Goldman, Sheldon, “Backgrounds, Attitudes, and the Voting Behavior of Judges,” Journal of Politics, 31 (02, 1969), 214–222CrossRefGoogle Scholar, for an elaboration of this point and some evidence supporting the preceding argument. For an analysis of interview data with appeals courts judges concerning consensus on their courts see Goldman, Sheldon, “Conflict and Consensus in the United States Courts of Appeals,” Wisconsin Law Review, Vol. 1968 (No. 2, 1968), 461–482Google Scholar particularly 476–480.

¹⁵ For details of the findings and the methods employed see Goldman, , “Conflict on the U.S. Courts of Appeals,” pp. 645–652Google Scholar. Essentially, blocs were defined by McQuitty Elementary Linkage Analysis of matrices of agreement scores.

¹⁶ One could argue that “ideology” should be used instead of “interrelated political attitudes.” The concept of ideology, however, suggests a more formal, explicitly rationalized system of ideas than the phenomena here seem to warrant. But see Spaeth, Harold J. and Peterson, David J., “The Analysis and Interpretation of Dimensionality: The Case of Civil Liberties Decision-Making,” Midwest Journal of Political Science, 15 (08, 1971), 415–441CrossRefGoogle Scholar and Schubert, Glendon, The Judicial Mind Revisited: Psychometric Analysis of Supreme Court Ideology (New York: Oxford University Press, 1974)Google Scholar.

¹⁷ The other circuits for which C scales were constructed were the second, fourth, fifth, ninth, and District of Columbia circuits. Although the commonly used coefficients of reproducibility and scalability met the usually acceptable minimum criteria, there were variations which suggested that the C scales for the District of Columbia and fourth circuits were the most acceptable scales, and the C and E scales for the third circuit only of minimum acceptability (the problem of non-participation was greatest for the third circuit).

¹⁸ Goldman, , “Conflict on the U.S. Courts of Appeals,” p. 652–656Google Scholar.

¹⁹ See Nagel, “Political Party Affiliation and Judges' Decisions”; Nagel, Stuart S., “Multiple Correlation of Judicial Backgrounds and Decisions,” Florida State University Law Review, 2 (Spring, 1974), 258–280Google Scholar; Grossman, Joel B., “Social Backgrounds and Judicial Decisions: Notes for a Theory,” Journal of Politics, 29 (05, 1967), 334–351CrossRefGoogle Scholar; Adamany, David W., “The Party Variable in Judges' Voting: Conceptual Notes and a Case Study,” American Political Science Review, 63 (03, 1969), 57–73CrossRefGoogle Scholar; Dolbeare, “The Federal District Courts”; Ulmer, S. Sidney, “Dissent Behavior and the Social Background of Supreme Court Justices,” Journal of Politics, 32 (08, 1970), 580–598CrossRefGoogle Scholar; Ulmer, S. Sidney, “Social Background as an Indicator to the Votes of Supreme Court Justices in Criminal Cases: 1947–1956 Terms,” American Journal of Political Science, 17 (08, 1973), 622–630CrossRefGoogle Scholar; Jaros, Dean and Canon, Bradley C., “Dissent on State Supreme Courts'. The Differential Significance of Characteristics of Judges,” Midwest Journal of Political Science, 15 (05, 1971), 322–346CrossRefGoogle Scholar; Beiser, Edward N. and Silberman, Jonathan J., “The Political Party Variable: Workmen's Compensation Cases in the New York Court of Appeals,” Polity, 3 (Summer, 1971), 521–531CrossRefGoogle Scholar; Feeley, Malcolm M., “Another Look at the ‘Party Variable’ in Judicial Decision-Making: An Analysis of the Michigan Supreme Court,” Polity, 4 (Autumn, 1971), 91–104CrossRefGoogle Scholar; Walker, “Partisan Influences”; Donald Leavitt, “Political Party and Class Influences on the Attitudes of Justices of the Supreme Court in the Twentieth Century,” paper presented at the 1972 Annual Meeting of the Midwest Political Science Association. For a more general discussion, see Goldman, and Jahnige, , The Federal Courts, pp. 167–171Google Scholar and Murphy, and Tanenhaus, , The Study of Public Law, pp. 103–112Google Scholar.

²⁰ See the discussion and citations in Adamany, , “The Party Variable in Judges' Voting,” pp. 58–62Google Scholar.

²¹ Tests of statistical significance are rneaningful here only if one were to assume that the judges and cases were a sample of a larger judge and case universe.

²² The Mann-Whitney U Test was used to determine whether there was a statistically significant difference in medians and distribution of the scores by issue by political party. The U Test is described in Siegel, Sidney, Nonparametric Statistics for the Behavioral Sciences (New York: McGraw-Hill, 1956), pp. 116–127Google Scholar.

²³ See, in general, Goldman, “Backgrounds, Attitudes, and the Voting Behavior of Judges.”

²⁴ Note that when the judges were categorized by appointing administration, the median score of the Nixon appointees on the criminal procedures issue was the lowest of all groups, 0.53. In contrast, the median score of the Johnson appointees on the criminal procedures issue was the highest of all groups, 1.20. Appointees of other Republican administrations had a median of 0.67 and appointees of other Democratic administrations had a median of 0.82. On the civil liberties issue, the Nixon appointees again had the lowest median score (0.57) and the Johnson appointees the highest (1.50). Appointees of other Republican administrations had a median of 0.61 and appointees of other Democratic administrations had a median of 0.86. For this analysis, a judge's political liberalism score was used when no separate criminal or civil liberties scores could be constructed because of participation in too few cases. For an in-depth study of the Nixon appointees, see, Gottschall, Jon, “The Nixon Appointments to the United States Courts of Appeals: The Impact of the Law and Order Issue on the Rights of the Accused” (Ph.D. Dissertation, University of Massachusetts, in preparation)Google Scholar.

²⁵ Congressional Quarterly, 12 16, 1972, p. 3160Google Scholar.

²⁶ Circuit-by-circuit analysis of the other issues revealed that only in the fifth circuit was there a higher proportion of Republicans than Democrats above the circuit median for the criminal procedures, civil liberties, and labor issues. For the injured persons category, the fifth circuit was joined by the eighth circuit. For the activism issue the fifth circuit was joined by the third and fourth circuits. On the private economic issue only on the fourth and sixth circuits was there a higher proportion of Republicans above the circuit median than Democrats.

²⁷ Note that because there was only a small number of Jewish judges (13) and they were concentrated on the second and third circuits (8 of the 13), they were not included in the analysis of the religion variable. Their median scores on the issues were: criminal procedures 1.60 (N = 13); civil liberties 1.50 (N = 10); labor 1.73 (N = 7); private economic 1.17 (N = 7); government fiscal 1.17 (N = 5); injured persons 1.39 (N = 8); political liberalism 1.47 (N = 13). economic liberalism 1.19 (N = 13); activism 1.62 (N = 8); dissents 0.56 (N = 13).

²⁸ Goldman, , “Voting Behavior,” p. 382Google Scholar.

²⁹ Bowen, “The Explanation of Judicial Voting Behavior.”

³⁰ In so doing, one could go beyond the permissive position some methodologists hold about the appropriateness of making assumptions about the data that are technically not warranted in order to utilize more sophisticated statistical methods. See, for example, Kerlinger, Fred N., Foundations of Behavioral Research (New York: Holt, Rinehart and Winston, 1964), p. 428Google Scholar. On the other hand, Multiple Classification Analysis could have been employed as “the MCA technique can be considered the equivalent of a multiple regression using dummy variables.” Andrews, Frank M., Morgan, James N., Sonquist, John A., and Klem, Laura, Multiple Classification Analysis: A Report on a Computer Program for Multiple Regression Using Categorical Predictors, second edition (Ann Arbor: Institute for Social Research, University of Michigan, 1973), p. 47Google Scholar.

³¹ Percentage of variance explained is derived from the partial correlation. See Nie, Norman H., Bent, Dale H., Hull, C. Hadlai, Statistical Package for the Social Sciences (New York: McGraw-Hill, 1970), p. 161Google Scholar. This follows Bowen, “The Explanation of Judicial Voting Behavior,” and Ulmer, S. Sidney, “Revising the Jurisdiction of the Supreme Court: More Administrative Reform or Substantive Policy Change?” Minnesota Law Review, 58 (11, 1973), pp. 149–151Google Scholar. But see Ulmer, , “Social Background as an Indicator,” Table 3, p. 626Google Scholar. Andrews et al., Multiple Classification Analysis; note that “the partial correlation (when squared) assesses the importance of a predictor in terms of the variance in the dependent variable marginally explainable by the predictor relative to the as-yet-unexplained variance” (italics in the original, p. 35). Nagel, Stuart has argued in “Multiple Correlation of Judicial Backgrounds and Decisions,” p. 274Google Scholar, that the proportion of variance of each independent variable should be calculated by multiplying the standardized regression coefficient by the zero order correlation coefficient. But see Guilford, J. P., Fundamental Statistics in Psychology and Education, fourth edition (New York: McGraw-Hill, 1965), pp. 399–400, 414Google Scholar.

³² The proportion of explained variance of the remaining categories follows: criminal procedures, 30 per cent; civil liberties, 42 per cent; private economic, 23 per cent; government fiscal, 13 per cent; injured persons, 47 per cent; political liberalism, 30 per cent; economic liberalism, 29 per cent; and activism, 26 per cent.

³³ It may be that the background variables tested here are simply too crude and that more refined vari-ables would yield more impressive results. It is the view of this researcher, however, that while back-grounds of course represent life experiences which nurture and shape political attitudes and values, this is such a dynamic and complex process subject to biological, environmental, political system, and judicial system variables among others that we ought not to be surprised that relatively little association of background variables to the voting behavior of aggregates of judges has been found. But see Ulmer, “Social Background as an Indicator.” In general, see Schubert, Glendon, “Justice and Reasoning: A Political Science Perspective,” Rivista Internationale Di Filosofia Del Diritto, 46 (10, 1969), 474–496Google Scholar and Fred Kort, “A Special and a General Multivariate Theory of Judicial Decisions,” paper presented at the 1974 Annual Meeting of the American Political Science Association.

³⁴ Perhaps it would be more accurate to conclude that judges behave as if they held attitudes which we have inferred from their voting patterns. It is clear nonetheless that numerous studies over the years which have uncovered voting patterns from which attitudes could be plausibly inferred (see, for example, the discussion and citations in Schubert, “Judicial Process and Behavior, 1963–1971”), as well as the replication in this revisit of earlier attitudinal findings, provide support for emphasizing the attitudinal variable in the explanation of voting behavior. There is still other evidence for inferring attitudes from the voting patterns of appeals judges. In interviews with 27 appeals judges the author found that 21 judges (i.e., 78 per cent) attributed dissension on their courts to attitudinal cleavages in the broader political sense. See Goldman, , “Conflict and Consensus,” pp. 474–476Google Scholar. More recently, Judge Henry J. Friendly of the second circuit conceded in a law review symposium that it is “in the closest cases that such attitudes may tip the balance.” Friendly, Henry J., “Of Voting Blocs, and Cabbages and Kings,” University of Cincinnati Law Review, 42 (No. 4, 1973), 677Google Scholar.

³⁵ Also see Atkins, “Judicial Behavior and Tendencies Towards Conformity,” and Richardson, Richard J. and Vines, Kenneth N., “Review, Dissent and the Appellate Process: A Political Interpretation,” Journal of Politics, 29 (08, 1967), 597–616CrossRefGoogle Scholar.

³⁶ In a sense, the backgrounds explanation of the voting behavior of appeals judges is in competition with the attitudinal explanation, yet it is only the background variables that are tested directly in terms of their contribution in explaining the variance of the voting behavior on the various issues. It should be pointed out, however, that one can interpret the correlation coefficients in Table 3 in terms of two-variable relationships to suggest that attitudes explain a larger proportion of the variance than do background variables. When one examines the zero order correlations of the background variables to each issue as reported in Table 8, we find they are generally considerably lower than those of Table 3.

³⁷ Cf. Beiser and Silberman, “The Political Party Variable.”

³⁸ See the excellent summary of the literature concerning the age hypothesis in Glenn, Norval D., “Aging and Conservatism” in Political Consequences of Aging, Eisele, Frederick R., ed., The Annals (Philadelphia: Academy of Political and Social Science, 09, 1974), pp. 176–186Google Scholar.

³⁹ Schubert, Glendon, The Constitutional Polity (Boston: Boston University Press, 1970), pp. 118–129Google Scholar.

⁴⁰ Ulmer, S. Sidney, “The Longitudinal Behavior of Hugo Lafayette Black: Parabolic Support for Civil Liberties, 1937–1971,” Florida State University Law Review, 1 (Winter, 1973), 131–158Google Scholar. Also see Ulmer, S. Sidney, “Dimensionality and Change in Judicial Behavior,” in Mathematical Applications in Political Science, VII, ed. Herndon, James F. and Bernd, Joseph L. (Charlottesville: University Press of Virginia, 1974), pp. 40–67Google Scholar.

⁴¹ Indeed, as Norval Glenn in “Aging and Conservatism” suggests with reference to the sociological literature, “Empirical evidence on the topic is not definitive; moreover, in view of intransigent methodological problems which plague the study of aging effects, the evidence may never be definitive” (p. 176).

⁴² But see Spaeth, Harold J., Meltz, David B., Rathjen, Gregory J., and Haselswerdt, Michael V., “Is Justice Blind: An Empirical Investigation of a Normative Ideal,” Law and Society Review, 7 (Fall, 1972), 119–137CrossRefGoogle Scholar.

⁴³ With regard to the Supreme Court see Ulmer, S. Sidney, “Supreme Court Justices as Strict and Not-So-Strict Constructionists: Some Implications,” Law and Society Review, 8 (Fall, 1973), 13–32CrossRefGoogle Scholar.