Determinants of the Amount of Time Taken by the Vinson Court To Process its Full-Opinion Cases

Jan Palmer and Saul Brenner

Is it possible to identify and to measure some of the factors that determine the amount of time needed by the Supreme Court to process a full-opinion case? Or, are the cases so heterogeneous that it is impossible to find any consistent predictors of how much time will pass between the conference vote on-the-merits and the announcement of the official opinion? To answer these questions, we investigate the Vinson Court (the October 1946 Term through the October 1952 Term). This Court is chosen because of the availability of a complete data set containing information necessary to test a number of hypotheses: the individual Justice's conference votes, the dates of these votes, and the names of the Justices who were assigned or reassigned the responsibility for writing the majority opinion.[1] Although the number of cases docketed has increased, the Supreme Court's procedures have not changed substantially since Vinson's tenure. Thus, this study of the Vinson Court not only provides insight into the decision making of that Court but also generates hypotheses that can be retested when complete and reliable conference vote and other data from later periods become available.

Observers of the Supreme Court have long wondered how Justices spend their time. This question is related to many others: Is the Court overworked? Should its jurisdiction be altered? Is it deciding too many or too few cases? Has the steady growth in the number of cases docketed each term reduced the Court's productivity? The Justices do not publish time charts and most of their activities are safely hidden from public view. Thus, scholars are forced to engage in educated guessing to answer these questions.

Hart[2] provided the first appraisal of how Justices of the modern Court spend their time. He estimated the amounts of time needed to complete various activities and inferred that the Court was overworked. Hart's study was criticized by Justice Douglas who claimed that the Court's caseload was not a burden.[3] But the Report of the Study Group on the Caseload of the Supreme Court"[4] (frequently known as the "Freund Report"), agreed with Hart's finding that the Court was overworked. Casper & Posner[5] extended Hart's analysis and concluded that the amount of time spent screening cases had increased but that the Court was not overworked. All these studies examined the typical Justice rather than the Court as a whole.

Rathjen[6] took a very different approach in his analysis of the time needed to process full-opinion cases. Rather than studying individual Justices, he examined the Court as a whole for the 1964-1973 Terms and concluded that processing time was positively related to importance, level of dissension, and whether the case dealt with "Freedom and Equality" issues. He also found that business issues and cases in which a liberal Justice wrote the opinion were processed more quickly.

Like Rathjen, we examine the Court as a whole. We retest some of his findings and are able to test additional hypotheses because we have the advantage of using data derived from the Justices' private papers to which Rathjen did not have access.

Almost all cases arrived at the Vinson Court either by a writ of appeal or as a petition for a writ of certiorari. Noting probable jurisdiction for an appeal or granting a certiorari petition required positive votes from four of the nine Justices. Most cases were denied review, which meant that the decision of the lower court remained the law in the case.

If a case was accepted for review, the litigants submitted written briefs after which the case was argued orally before the Court. A few days later, there was a secret conference vote at which time the Justices voted either to affirm or reverse the decision of the lower court. The Court sometimes had additional conference votes on-the-merits if the original conference vote resulted in an equally divided Court or if the opinion writing process uncovered new issues or changed several minds.

After the conference vote, the Chief Justice, if he was in the majority, assigned the writing of the Court's opinion either to himself or to another member of the majority. When the Chief Justice was in the minority, the senior Associate Justice in the majority assigned the opinion. The other Justices were free to write concurring or dissenting opinions. Drafts of all opinions were circulated to the Justices who frequently returned them with written comments and suggestions which served as a basis for further negotiations about the majority opinion's content. When a Justice was satisfied with an opinion, he joined it by sending a written memorandum to the author. Because Justices were free to change sides while the opinion was being written, the original majority opinion assignee at times lost the assignment to a colleague.

Some cases were so complex or contentious that the Court ordered reargument following a conference vote or after all the written opinions failed to attract a majority. Occasionally, reargument was required because the Court could not complete a case during one Term, requiring that it be held over until the next.

Once the majority, concurring, and dissenting opinions were completed, they were announced by the Court and printed in U.S. Reports in a format that allows scholars to infer each Justice's position. Conference votes, however, are not published and can only be obtained from the Justices' private papers. Similarly, opinions in US. Reports identify the author of the majority opinion but do not indicate whether he obtained the opinion through assignment or through reassignment. Information on assignments and reassignments is available only in private papers.

Our earlier analysis[7] of opinion assignment patterns for the Vinson Court shows an inverse relationship between the number of assignments given to a Justice and the amount of time he took to complete opinions. Table 1 shows the number and percent of opinions assigned to each Justice as well as the average number of days each took to complete opinions. The data is divided between the Vinson era's two natural Courts, i.e., periods of constant membership. There are substantial differences among the Justices in both the number of assignments and the average amount of time.

During the first natural Court, the number of assignments ranged between 71 (Douglas) and 22 (Burton). The average number of days taken to complete opinions varied between40 (Black) and 104 (Frankfurter). As a group, the liberals, Rutledge, Murphy, Douglas, and Black wrote more quickly than did their conservative brethren. The results for the second natural Court show that although the amount of inequality in the number of assignments declined, the faster writers were still favored with more assignments. The three slowest writers (Burton, Frankfurter, and Vinson) became about three weeks faster than they had been during the first Court.

Additional findings which are not shown in Table 1 include the following: The distribution of assignments made by Chief Justice Vinson is similar to the distribution of all assignments shown in Table 1. Vinson favored the two fastest writers, Black and Douglas, even though their ideological orientation was substantially different from his own. The distributions of assignments in minimum winning cases (e.g. five-to-four) are different from the distributions in nonminimum winning cases. The wide variation in the amount of time taken to complete opinions poses an additional question. Did the Justices agree on whether the Court was overworked? For example, Frankfurter, who took twice as long to write opinions as did Douglas, complained more or less continuously that the Court was accepting too many unimportant cases.[8] Douglas, however, wanted the Court to accept more.[9]

We assume that there are at least three underlying factors that increase a case's processing time: the importance of the case, divisiveness or disagreement among the Justices, and fluidity or vote changes. Fluidity can be measured directly by inspecting the available data. But there are no direct measures of importance and divisiveness. For these two variables we use sets of proxies or related measures, three for importance and two for divisiveness. The first proxy for importance is obtained from the Justices' earliest conference vote on-the-merits. Ulmer,[10] Provine,[11] Brenner,[12] and Palmer[13] show that Justices were more likely to vote for granting review if they expected to reverse the lower court's decision because more was gained from reversing an incorrect decision than from affirming a correct one. Because the Court rarely granted review to unimportant cases when it agreed with the decision of the lower court, cases in which the Court expected to affirm are likely to be more important. Obviously, there were many cases in which the Justices did not know how they would vote on the merits when they voted to accept the case for review. In addition, it is impossible to determine from the printed record whether the Court intended to reverse or to affirm when it selected a case for review. The closest approximation of the Court's intention is obtained from the Justices' first conference vote. Thus, our first hypothesis, hereafter Hi, is that the processing time for affirm cases (i.e., cases in which a majority voted to affirm at the first conference) will be longer than for reverse cases.

The second and third proxies for importance are obtained from Schubert's data set[14] which categorizes Supreme Court cases according to the type of dispute. Two of Schubert's categories, civil liberties cases and economic cases, are used to retest Rathjen's[15] findings that "Freedom and Equality" cases took "a week longer to adjudicate" than the "average decision duration," while "Business" cases took "approximately 11 days fewer to adjudicate." One reason for this expectation is that civil liberties cases were often more important and more controversial to the Justices. Civil liberties cases, in addition, were likely to be more divisive. Thus, our second hypothesis, H2, is that the processing time was longer for civil liberties cases than it was for all other cases. Our third hypothesis, H3, is that the processing time was shorter for economic cases than it was for all other cases.

The next factor underlying the disposition time for full-opinion cases is divisiveness, which we measure as the difference between the size of the majority and the size of the minority at the first conference vote. The larger the difference, the greater the probability that the issues were uncomplicated or did not involve conflicting legal principles or precedents. A larger difference also reduced the likelihood that the majority would break up and thereby prolong the process. In addition, a larger majority simplified the opinion writing process for two reasons: First, a larger majority decreased the probability that a dissenting opinion would be written. Second, a larger majority increased the bargaining power of the opinion author while reducing that of the assenting Justices because no single vote was essential to holding the majority. Therefore, our fourth hypothesis, H4, is that processing time became shorter as the difference between the sizes of the majority and minority at the first conference vote increased.

Divisiveness is also related to the Supreme Court's membership. The Vinson era can be divided into two periods of constant membership, e.g., two natural Courts. Schubert,[16]Provine,[17] and Pritchett,[18] identify two loose ideological blocs during the Vinson era. For the first natural Court; October 1946-1948 Terms, the liberal bloc consisted of Justices Rutledge, Murphy, Douglas, and Black, while the conservative bloc included Justices Burton, Jackson, Frankfurter, Reed, and Vinson. Rutledge and Murphy, both of whom were liberals, died during the summer of 1949 and were replaced by Clark and Minton, both of whom were conservatives. Because the liberal bloc was reduced from four to two members, the second natural Court was more homogeneous in its ideology and, therefore, less divisive. In addition, during the second natural Court, Chief Justice Vinson succeeded in his efforts to get opinions written more expeditiously.[19] Thus, we hypothesize, H5, that cases decided during the first natural Court had a longer processing time.

The next element underlying disposition time is fluidity or vote changes by individual Justices. We measure fluidity directly by determining the number of strong vote changes (i.e., affirm to reverse or the converse) between the first conference vote and the decision of the Court. We expect, H6, that processing time increased as the amount of fluidity increased. Reassignment of the responsibility for writing the Court's opinion is directly related to both fluidity and divisiveness. We hypothesize, H7, that processing time was greater if there was a reassignment because of the time needed to discover that the original majority opinion assignee could not hold the majority as well as the time required for the new author to write an opinion. Finally, we include one hypothesis that is unrelated to importance, divisiveness, or fluidity: We know that the liberal Justices (i.e., Rutledge, Murphy, Douglas, and Black) were able to write majority opinions more expeditiously than were their conservative colleagues. As a consequence, the processing time was shorter when a liberal wrote the majority opinion. This hypothesis, H8, is based on our earlier finding (summarized in Table 1) that the liberal Justices (especially Black and Douglas) wrote more quickly during the Vinson era and on Rathjen's[20] similar findings for the Warren and Burger eras. The reasons for the liberals' greater speed are uncertain. Perhaps the liberal ideology is simpler and, therefore, defending a liberal position is easier.

The data set contains 716 cases. Most of the information was obtained by Palmer[21] from the private papers of seven Vinson era Justices. Information regarding whether a case is included in the civil liberties or economic cases was obtained from Schubert's data.[22] Unlike Rathjen's analysis, our data set includes cases that were held over, i.e., cases in which the Court voted on the merits in one Term but did not announce its decision until the next.

We measure processing time as the number of days, including Saturdays and Sundays, between the date of the first conference vote on the merits and the date when the majority opinion was handed down. The mean is 82 days. The median is 56.

We test the eight hypotheses using two methodologies: one bivariate (i.e., separately examining each explanatory variable's impact on processing time), the other multivariate (i.e., simultaneously examining the impact of all the explanatory variables). The statistical results, which are presented in the appendix, can be summarized as following.

There are six relationships strong enough to be considered statistically significant, i.e., not chance occurences. (1) Cases took approximately eleven days longer if the conference vote was to affirm rather than to reverse. (2) Processing time was related to the difference between the sizes of the majority and minority at the conference vote. A one vote increase in the difference decreased time by approximately 3.3 days. (3) Liberal Justices finished opinions about a month quicker than did their conservative colleagues. (4) Cases were completed approximately one month faster during the second natural Court. (5) Processing time increased by about a week for each Justice who switched sides between the conference vote and the opinion announcement. (6) Processing time increased by approximately 139 days if the opinion was reassigned.

The relationship between processing time and whether the cases dealt with civil liberties is not statistically significant. Likewise, the relationship between processing time and whether the cases dealt with economic issues does not have a significant relationship. The model explains about one-third of the variation in processing time. The other two-thirds of the variation result from the many factors not included in the model, e.g., the complexity of the case, the amount of research time needed by the opinion author, etc.

Is it possible to distinguish and measure variables that affect processing time? The answer is clearly "yes." The empirical results identify six variables which together explain about a third of the variation in processing time and provide insights into the Court's procedures. Five of the six significant variables relate to the underlying factors of importance, divisiveness, and fluidity. The only significant variable not related to one of these factors is whether the opinion writer was a liberal.

There is no a priori reason to believe that the variables we identify will not pertain to other eras of the Court's history. The hypotheses tested in this analysis can be re-tested when comparable data become available for other time periods.

We have three ancillary findings. First, the wide variation in processing time shows that examining the number of cases docketed each Term is not a useful measure of the Court's workload. Second, the large differences among Justices in both the numbers of opinions written and the amounts of time taken to complete opinions indicates that the "Court's workload" may not be a meaningful concept--at least in terms of opinion writing. Third, examining the Court as a whole, rather than individual Justices, is useful, especially given that information on how individual Justices spend their time may never be available.

The bivariate results, which are presented in Table 2, compare values at or above the median processing time with those below the median. We employ median rather than the more usual mean values because the distribution is skewed by a few outliers, all of which are above the mean, i.e., a few cases that were held over until the next Term and therefore took more than a year to process. A GAMMA statistic, a measure which ranges between -1 and 1, is used to measure and to compare the strength of each variable's relationship with processing time. GAMMA statistics with absolute values above .10 show the existence of relationships. Those with larger absolute values indicate stronger relationships.

Table 2 shows that there is a seven-day difference between the median processing times of cases in which the Court's original conference votes were to affirm, 58 days, and to reverse, 51 days. This relationship is consistent with Hi. The GAMMA of .19 indicates a low positive relationship.

There is a six-day difference between the median processing time of civil liberties cases, 58 days, and non-civil liberties cases, 52 days. This difference is consistent with H2 and with Rathjen's[23] conclusion that such cases took "a week longer to adjudicate." The GAMMA of .08, however, shows only a negligible association.

For economic cases, the median processing time, 51 days, is a week shorter than the median processing time for non-economic cases, 58 days. This result appears to support H3 and is also similar to Rathjen's finding. The GAMMA of -.14 indicates a low negative relationship.

For DIFFERENCE, there is a monotonically decreasing relationship between processing time and the difference between the sizes of the majority and minority at the original conference vote. When the difference was 0 or 1 vote, the median processing time was 65 days. In contrast, when the difference was 8 or 9 votes, the median processing time was only 30 days. The GAMMA of .28 indicates that there is a low positive relationship between DIFFERENCE and median processing time. Thus, H4 is supported.

The median processing time declined by two weeks between the first, 65 days, and second, 51 days, natural Courts. The relationship is low (GAMMA = .22), upholds H5, and is consistent with earlier research that indicates that opinions were written more expeditiously during the second natural Court.[24]

For FLUIDITY there is a monotonically increasing relationship between the number of vote changes and processing time. The median time with no changes, 51 days, is more than three weeks shorter than the median with 4 to 9 changes, 77 days. This relationship upholds H6 with a GAMMA of .18.

With REASSIGNMENT, median processing time was dramatically longer for cases in which the majority opinion was reassigned, 142 days, as compared to those in which the original assignee wrote the opinion, 51 days. The GAMMA of .84 shows a very strong relationship which supports H7.

Lastly, median processing time was substantially shorter when a liberal Justice wrote the majority opinion, 44 days, as compared to when a nonliberal Justice wrote, 65 days. The GAMMA of .40 indicates a moderately strong relationship which supports H8.

Linear regression analysis measures the relationships between the dependent variable, TIME, and all the explanatory variables simultaneously. Because the explanatory variables are themselves interrelated, regression analysis gives a more appropriate and accurate measure of these relationships because it attempts to separate the impact of each explanatory variable from the impacts of the others.

The linear regression results are presented in Table 3. The regression coefficients indicate the impacts (measured in days) of the individual explanatory variables. The r-squared statistic, .32, indicates that about a third of the variation in TIME is explained by the explanatory variables. The F-statistic, 40.84, shows that the overall model is statistically significant. The t-statistics indicate the statistical significance of the individual variables.

The regression coefficient for AFF, 11.4, is statistically significant, i.e., Hi is supported by the analysis. This means that, ceteris paribus, cases took approximately eleven days longer to process if the conference vote was to affirm rather than to reverse.

For civil liberties cases, the regression coefficient is negative and statistically insignificant. Thus, H2 must be rejected. The difference between the bivariate analysis, which show a positive relationship between processing time and civil liberties cases, and the multivariate analysis, which shows a negative relationship, result from the dissimilarity of the statistical methods.

The regression coefficient for economic cases has the correct (negative) sign but is not statistically significant. H3, therefore, is rejected.

For DIFFERENCE, the regression coefficient is statistically significant, i.e., H4 is supported. A one vote increase in the difference between the sizes of the majority and the minority, everything else the same, decreased the processing time by approximately 3.3 days.

Cases were processed about a month faster, ceteris paribus, during the second natural Court. The regression coefficient for FIRST, 33.5, is statistically significant. H5 is, therefore, supported by the analysis.

For FLUIDITY the regression coefficient, 6.8, is statistically significant and supports H6. For each Justice who switched sides between the conference vote and the opinion announcement, processing time increased by approximately one week. The regression coefficient for LIBERAL, -32.0, is statistically significant. Thus, H6 is supported. Everything else being the same, a case in which a liberal wrote the opinion was processed about a month faster.

Finally, for REASSIGNMENT, the regression coefficient is quite large, 139, and is statistically significant. Thus, H6 is supported. If the opinion was reassigned, ceteris paribus, the processing time was increased by four and a half months. This very large increase in processing time resulted because reassigned cases were often the most difficult and divisive and because the reassignment frequently required reargument during the next term.

1. Palmer, Jan. The Vinson Court Era: The Supreme Court's Conference Votes: Data and Analysis, New York: AMS Press (1990).
2. Hart, Henry. "The Time Chart of the Justices," 73 Harv. L. Rev. 84 (1959).
3. Douglas, William O., "The Supreme Court and Its Case Load," 45 Cornell L. Rev. 401 (1960).
4. Federal Judicial Center, Report of the Study Group on the Caseload of the Supreme Court (1972).
5. Casper, Gerhard and Richard Posner, "A Study of the Supreme Court's Caseload," 3 Journal of Legal Studies 339 (1974).
6. Rathjen, Gregory "Time and Dissension on the United States Supreme Court," 7 Ohio N.U. L. Rev. 227 (1980).
7. Brenner, Saul and Jan Palmer, "The Time Taken to Write Opinions as a Determinant of Opinion Assignments," 72 Judicature 179 (1988).
8. Felix Frankfurter Papers, Harvard Law School Library, 176-4; Provine, Doris. Case Selection in the United States Supreme Court, Chicago: University of Chicago Press (1980), p. 189.
9. For example, Douglas's grant rate for certiorari petitions was the highest on the second natural Court and was second only to Murphy on the first. Palmer, Chapter 5.
10. Ulmer, S. Sidney, "The Decision to Grant Certiorari as an Indicator to Decision on the Merits," 4 Polity 429 (1979).
11. Provine, pp. 105-113.
12. Brenner, Saul. "The New Certiorari Game," 41 Journal of Politics 649 (1979).
13. Palmer, Jan. "An Economic Analysis of the U.S. Supreme Court's Certiorari Decisions," 39 Public Choice 387 (1982).
14. Schubert, Glendon. The Judicial Mind: The Attitudes and Ideologies of Supreme Court Justices 1946-1963.
15. Rathjen, p. 243.
16. Schubert, p. 103-116.
17. Provine, pp. 135-139.
18. Pritchett, C. Herman. Civil Liberties and the Vinson Court. Chicago: University of Chicago Press (1954), chapter IX.
19. Brenner and Palmer, p. 183.
20. Rathjen, p. 245.
21. Palmer, ch. 4.
22. Dataset available from the Inter-University Consortium on Political and Social Research in Ann Arbor.
23. Rathjen, p. 243.
24. Brenner and Palmer, p. 183. Slotnick, Elliot. "Who Speaks for the Court? Majority Opinion Assignment from Taft to Burger," 23 American Journal of Political Science 60 (1979).