In the period from 1947 to 1976, the United States Supreme Court has denied certiorari in more than half the cases involving conflict with Supreme Court precedent or intercircuit conflict. In both instances, the denial rate has been higher in the Burger Court than in the Vinson and Warren Courts and denial has been greater for intercircuit conflict cases than for cases in which the ruling in the lower court was in conflict with one or more Supreme Court precedents. When conflict was conceptualized as a predictor of decision and examined along with federal government as petitioning party, economic issues, and civil liberty issues, it was found to have 4 to 7 times the predictive power of the other variables combined for the Vinson and Warren Courts. For the Burger Court, the petitioning party variable was found to be a better predictor than conflict, but conflict was a much better predictor than the subject variables. Discriminant function models using the four predictor variables were able to account for up to 36.9% of the variance in the Supreme Court's certiorari decisions, almost all of which was the result of the contributions made by the conflict and party factors.

No abstract

8 The phrase is Llewellyn's. 'A recent example of a nonquantitative and nonreplicable predictive technique can be found in Fred Rodell's discussion of Baker v. Carr, For Every Justice, Judicial Deference is a Sometime Thing, 5o GEo. L. J. 70o (1962). Apparently believing that judicial votes are more accurately predictable in terms of the personal predilections of the judges than on the basis of impersonal, or objective, reasoned rules, Rodell predicted a 5-4 Supreme Court vote in Baker v. Carr eleven days before the Supreme Court dccision. Moreover, he identified the Justices he expected to find on each side and predicted that the Court would "order the requested redistricting." Id. at 707. While the Court took action likely to promote redistricting, it did not "order redistricting," as Justice Stewart makes clear in his concurring opinion. Rodell correctly predicted seven of the eight votes cast-an impressive result indeed. But in a larger perspective, it may be meaningless. For who knows how Rodell reached his result? Can any lawyer replicate Rodell's experiment and results? The important question is not whether one can predict judicial votes in one case by intuition or sheer guess, or through personal contact with judges or their clerks, but what replicable procedures are significantly successful over a long run of cases. If Rodell has discovered any successful predictive device based on "human factors," he has yet to make it available to the profession at large. 'Materials on quantitative methods in the social sciences are abundant. Especially useful are J. P. GUILFORD, FUNDAMENTAL STATISTICS IN PSYCHOLOGY AND EDUCATION (x956), and SIDNEY SIEGEL, NONPARAMETRIC STATISTICS FOR THE BEsAvsORAL SCIENCES (1956). For a more philosophical treatment see LEoN FESTINGER & DANIEL KArz (EDs.), RESEARCH METHODS IN THE BEHAVIORAL SCIENCES (r953). A widely used compendium of articles dealing with the methodology of social research is PAUL LAZARSFELD & MORRIS ROSENBERG (EDs.), THE LANGUAGE OF SOCIAL RESEARCH (1955). ' Chapters on probability theory can be found in most textbooks in statistics. A classical statement is PIERRE SIMION, MARQUIS DE LA PLACE, A PmsiosomcAL EssAY ON PROBARBiLsEs (translated from the French by F. W. Truscott & F. L. Emory, 195). A good introduction to this subject is HAROLD CRAMER, THE ELEMENTS OF PROBABILITY THEORY (1955). 'See Karl N. Llewellyn's works generally, but particularly his JURISPRUDENCE (1962). On page 357 of that work, he declares: "The rule part of the law-institution is thus one tool-part of the institution; one hugely developed part; one part vital to communication over space and time and variant personnel; but one vital tool-part and no more than that, and a tool-part nowhere nearly so effective as it purports to be." The quotations that follow are taken from KARL N. LLEwELLYN, THE COMMON LAW TRADITIoN-DECIDING APPEALS 15 (196o). ' CLEMENT VOSE, CAUCASIANS ONLY: THE SUPREME COURT, THE NAACP, AND THE RESTRiCTIvE COVENANT CASES (1959); GLENDON A. SCHUBERT, CONSTITUTIONAL POLITICS: THI...

Political scientists now agree that the Supreme Court is a major institutional policy maker in the American political system. But they have done little to relate that conclusion to democratic theory. Nor have they bothered to draw the implications for structural reform of the Court that seem to flow from such a relationship. Specifically, we have been less concerned with holding Supreme Court justices accountable to those they govern than seeing to it that legislators and executives provide acceptable policies. Such a discrepancy becomes increasingly diffiult to justify as we dilute the traditional role distinctions between judicial and other policy making institutions of government.Generally speaking, accountability is promoted through two devices. The first is access to the policy maker—i.e., the opportunity to be heard, to express oneself, to communicate one's preferences effectively to those whose policy actions are directly relevant. If these preferences are not sufficiently heeded, then we reserve the right to replace the incumbent with one who will provide greater satisfaction in that regard. For executives and legislators, free expression and frequent elections are appropriate means to insure accountability. But since Supreme Court justices are appointed for life (formally “good behavior,” but in fact for life), access to Court review becomes crucial.

In making review decisions, Supreme Court justices are predisposed to support underdogs and upperdogs. disproportionately but, also, are motivated to hide any "bias" that may be at work in determining votes.In balancing these two values, justices may be expected to vote their "bias" more frequently (1) when that vote will determine outcome, and (2) when the "bias" will be harder to detect. The latter goal may be served by voting the "bias" more frequently in close cases and less frequently otherwise.In an analysis of the voting patterns of five justices in the decade 1947–56, I found that two liberal and two conservative justices conformed to these expectations. A fifth, or control justice, defined as neither liberal nor conservative, did not pattern his votes in the manner predicted for liberals and conservatives. This relationship held when four projected intervening variables were controlled individually and collectively.