Published online by Cambridge University Press: 02 September 2013
It is axiomatic that some supreme courts are more influential than others. A dictum by one judge may carry more weight than a decision by another. Anyone who has studied the opinions of our highest courts is constantly assigning values to them, and the combination of these impressions may determine the relative standing of these tribunals for that individual. That this process of appreciation or depreciation is usually unconscious, and frequently irrational, does not make the prestige which results from it any less real or less potent a factor.
1 “We understand prestige to mean the attribute of social appreciation which may attach to a person, a group, or an institution, and which may vary not only from high intensity to low, but also from positive to negative poles.” White, L. D., Further Contributions to the Prestige Value of Public Employment (Chicago, 1932), p. 78Google Scholar.
2 “A large proportion of the general opinions of mankind are derived merely from authority, and are entertained without any distinct understanding of the evidence on which they rest, or the argumentative grounds by which they are supported.” SirLewis, George Cornewall, Essay on the Influence of Authority in Matters of Opinion (London, 1875), p. 5Google Scholar.
3 It is possible, of course, to attach too great a significance to the ratings given. One professor was afraid that the highest court in his state might do that very thing and wrote: “I rather incline to the view that, human nature being what it is, the judges of the supreme court of … might take considerable exception to an adverse vote by the faculty of the University of … Law School, and I question whether the matter might stop at such irritation, and might not lead to the court's changing its views or its methods.”
4 There was a possibility that the ratings secured might not be representative of the entire group of law professors because the older, better informed, and more cautious instructors might hesitate to return the forms. The returns dispelled this fear, however. Although there was no request that the forms be signed, many were, and the names showed clearly that many of the forms were returned by leaders in the profession.
5 It is interesting to observe that the New York court of appeals was the most widely known notwithstanding that only 8 of the 259 raters were from that state. Likewise, New Jersey (with one rater) and Michigan (with eight raters) had courts more widely known than California (with 17 raters) or Ohio (with 20 raters). In general, however, there was some slight relation between the number of forms returned from a state and the number of times the court of that state was rated. The coëfficient of correlation between these two variables was .50 ± .08.
6 Thurstone, L. L. and Chane, E. J., The Measurement of Attitude (Chicago, 1929)Google Scholar. See especially Chap. 3.
7 In this study, the data were plotted free hand to give a smoothed cumulative frequency curve.
8 The measure used was the difference between the quartiles (Q3 – Q1). In cases where the first quartile fell in the same scale unit as the median, it was assumed that only part of the complete curve was available. In those cases, the position of the missing quartile was determined on the basis of the projected curve. The mean deviation of the ratings for each state was also computed, and it was found that there was substantial agreement between the two indexes.
9 The coëfficient of correlation between the index of agreement and the number of ratings for each state was –.25 ±.09. The coëfficients of correlation used in this article were computed by the product-moment formula, using the method described in Mills, F. D., Statistical Methods (New York, 1924), pp. 372–378, 402–403Google Scholar. To aid in the interpretation of the coëfficient, and as a check on the computations, the data were plotted on graph paper and the coëfficient estimated from the regression lines. Due to the specially high standing of the New York and Massachusetts courts, the coëfficients are somewhat higher than the data for the other courts would have given. On the other hand, the data in some instances were slightly curvilinear, and in those cases the coëfficient of correlation understates the amount of relationship between the variables. Usually these two factors tended to neutralize each other, and it is believed that the coëfficients given represent the general situation with reasonable fidelity.
10 The coëfficient of correlation between the number of times a court was rated and the average rating of esteem assigned to it by the raters was .72 ± .05.
11 The coëfficient of correlation between the average esteem of a court and the lack of agreement in the opinion concerning it was .70 ± .05
12 This observation is supported by the low correlation between the index of agreement and the number of raters who expressed an opinion. See note 9 above.
13 “Cases are great by what we make of them.” Cardozo, Benjamin, Law and Literature (New York, 1931), p. 39Google Scholar.
14 In order to ascertain whether the inclusion of additional volumes in the list would materially affect the result, the data were tested for reliability. The case-books were divided into two groups, similar volumes from each subject being placed in each group, and the number of cases listed for each of the states in both groups was found to be substantially the same. The coëfficient of correlation between the number of cases from each state in the first group of case-books and those in the second group of case-books was very high, i.e., .95 ± .01. Clearly, the inclusion of additional volumes would not materially affect the position of the various courts.
15 The coëfficient of correlation between the average esteem in which the court was held and the number of times its opinions were reprinted in the case-books was 70 ± .05.
16 The coëfficient of correlation between the number of times the opinions of each court were reprinted in the case-books and the number of ratings returned for each court on the rating forms was .86 ± .03.
17 It was found that there was little relationship between the number of opinions found in the case-books and the extent to which opinion regarding the standing of the court was in agreement. The coëfficient of correlation between the index of disagreement and the number of opinions reprinted was –.37 ± .08.
No index of agreement among the compilers of the case-books was constructed. An inspection of the data, however, showed that certain courts seemed to have developed higher standing in some subjects than in others. This may have been due to local conditions, or to the special interests of certain judges. Examples toward judicial specialization may be observed in the frequent use of opinions of the Illinois supreme court in future interests cases, of the New Jersey court of chancery in equity cases, of the California supreme court in cases involving water rights, of the Oklahoma supreme court in oil cases, and of the Pennsylvania supreme court in cases involving damages.
20 The coëfficient of correlation between the number of times a state was cited before 1910 and the number of times it was cited in 1919–1921 was .88 ± .03 (28 states).
21 An effort was made to secure a more recent table of citations than that published in 1922. It seems, however, that none has been compiled, although it is reported that the West Publishing Company has, from time to time, checked the citations in isolated volumes of the state reports to detect any change in the trend shown by the 1922 count. No substantial change was found. This same fact was ascertained independently by a check of the citations in one state—Connecticut. The volumes of the Connecticut Reports contain a list of the cases cited in them, and tables were made from these lists showing the number of states cited in these reports between 1900 and 1910 and also the number cited between 1923 and 1929. The correlation between the citations in these two periods was found to be very high, the coëfficient of correlation being .98 ± .004 (47 states). The high standing of the Connecticut supreme court makes this result all the more significant.
22 A table was made of the states in the 1922 compilation which had been cited fewer than 1,200 times. It was found that seven of these states were also reported in the 1910 table. The average number of citations reported for these seven states in 1910 was one-half of the number reported in 1920. Accordingly, the number of citations for the states not listed in 1900 was estimated at one-half of the 1922 figure. This estimate was included in the total.
23 The Docket, April–May, 1910, pp. 233–237Google Scholar, contains data on the number of times certain courts cited courts in other states.
24 The coëfficient of correlation between the average esteem in which a court was held and the number of times it was cited by other courts was .65 ± .06. A higher correlation was found between the total citations by other state courts and the number of opinions quoted in the case-books, the coëfficient being .89 ± .02.
25 The coëfficient of correlation between the frequency of citation by other state courts and the frequency of citation by the United States Supreme Court was .91 ± .02.
26 The coëfficient of correlation between the frequency of citation and the number of ratings returned on each court was .95 ± .01.
27 It is true that only a small proportion of the cases appealed to the state supreme court are again appealed to the United States Supreme Court, and that these appeals involve only limited classes of cases. Nevertheless, the count covered appeals from lower federal courts as well as from the state courts, and although not all the fields of law were covered in the same proportion as they are covered by state court cases, the decisions of the United States Supreme Court cover a wide territory.
28 The coëfficient of correlation between the gross number of cases cited by the United States Supreme Court and the number of cases cited with approval was .83 ± .03.
29 The coëfficient of correlation between the cases cited with approval by the United States Supreme Court and the cases cited by other state courts was .88 ± .02
30 The coëfficient of correlation between the cases cited with approval by the United States Supreme Court and the average estimate of esteem by law professors was .71 ± .05.
31 The coëfficient of correlation between the cases cited with approval by the United States Supreme Court and the number of opinions quoted in case-books was .92 ± .02.
32 An attempt also was made to secure an index of the opinion of the courts held by various publishers of law-books. It was found, however, that no reliable gauge of their attitudes could be secured. The only feasible index of opinion of this group appeared to be the number of cases from each state selected for inclusion in the two series of reports, Lawyers' Reports Annotated and American Law Reports. A comparison of the index constructed from this data with the other gauges of prestige, however, disclosed that the relation between them, although perhaps somewhat significant, was not close. This is probably due to the variety of considerations which the editors of these series have in mind in selecting cases for inclusion in them, such as the importance of the decision, the uniqueness of the facts, or the popularity of the series among the lawyers of the state. It may be noted that the coëfficient of correlation between the number of cases from each state included in LRA and ALR and the index of esteem, of the law professors was .44 ± .08. A considerabty higher correlation was found between the cases in LRA and ALR and the total number of citations of state courts by each other (r = .71 ± .05). This indicates, perhaps, that these series of volumes have more influence with judges than with law faculties. The coëfficients of correlation between the cases included in LRA and ALR and the other indices of prestige developed in this paper ranged between the two figures just given.
33 These various factors were reduced to an equal weight in the construction of the composite rating by dividing the index for each state by the standard deviation of all the states for that index.
34 The coëfficient of correlation between the composite personnel index and the composite index of prestige was .53 ± .08 (32 cases). The composite personnel index is given in the article entitled “Judicial Personnel,” Annals of the American Academy of Political and Social Science, May, 1933, pp. 143–155Google Scholar.