1. Schwartz, E. and Schwartz, W., Decisionmaking by Juries Under Unanimity and Supermajority Voting Rules, Georgetown Law Journal 80(3): 775–807 (1992).Google Scholar

2. This approach is formalized in Klevorik, A. and Rothschild, M., A Model of the Jury Decision Process, Journal of Legal Studies 8:141 (1979).CrossRefGoogle Scholar Here, the authors employ a probabilistic model of persuasion, The larger a coalition, the more likely that the next juror “converted” is converted to that side. The possibility of reaching a point, other than unanimous consent, where no additional jurors can be converted is not even considered. See Apodaca v. Oregon, 406 U.S. 404 (1972)Google Scholar, and Johnson v. Louisiana 406 U.S. 356 (1972)Google Scholar, for examples ofthis sort of reasoning by the Supreme Court.

3. The concern that relaxation of the unanimity rule may lead to excessively brief jury deliberations is manifested in the requirement that the jury deliberate for a specified minimum period of time before rendering a verdict in foreign jurisdictions using some other decision rule. For instance, see Criminal Code Act (NT) §368 (Austl.)–12 hours; Juries Act of 1927Google Scholar (SA) §57 (Austl.)–4 hours; Juries Act of 1899Google Scholar (Tas.) §48 (Austl.)–2 hours, 6 hours if death penalty; Juries Act of 1957Google Scholar (WA) §41 (Austl.)–3 hours; Criminal Justice Act of 1967Google Scholar, ch. 80, §13 (Eng.)–2 hours Criminal Justice Act of 1984Google Scholar, No, 22, §25 (Ir.)–2 hours; Criminal Procedure Act (Majority Verdicts Act) of 1971Google Scholar (NI)–2 hours.

4. For discussion of the extent to which jurors can be counted on to focus exclusively on legal factors, see Friedland, S., The Competency and Responsibility of Jurors in Deciding Cases, Northwestern University Law Review 85(1): 190–220 (1990).Google ScholarWhite, L., Juror Decision Making in the Capital Penalty Trial, Law and Human Behavior 11 (2): 113–130 (1987)CrossRefGoogle Scholar, and Horowitz, I., Jury Nullification: the Impact of Judicial Instructions, Arguments, and Challenges, on Jury Decision Making, Law and Human Behavior 12(4): 439–483 (1988).CrossRefGoogle Scholar

5. The Supreme Court, in Furman v. Georgia, 408 U.S. 238 (1972).Google Scholar and subsequently in Gregg v. Georgia, 428 U.S. 153 (1976)Google Scholar and its companion cases, attempted to limit jury discretion in an attempt to insure that capital verdicts would not be arbitrary and capricious. Many of the guidelines employed are discussed below. However, as we illustrate below, much discretion does, indeed must, remain in the hands of the jurors.

6. The statutory language quoted in this paragraph is found in Fla. Stat. Ann. §941.141 (5) (West 1976).Google Scholar The formulations in the various states differ to some extent. All, however, confer very substantial discretion on the jury. The provisions of the state death penalty laws are summarized in Gillers, S., Deciding Who Dies, Penn, Law Review 129(1): 102–119.Google ScholarTushnet, M.. The Death Penalty 1970–1979 (Steinberg, New York, 1994)Google Scholar, reviews the constitutional decisions which shaped the procedures adopted by the states.

7. In two very recent (and linked) decisions, Tuiliaepa v. California and Proctor v. California (Cases 93–5131 and 93–5161. 62 LW. 472–28. 1994).Google Scholar the Supreme Court held that the aggravating and mitigating factors of the California statute were not too vague to be Constitutionally applied. The Court explicitly discusses the necessary exercise of discretion by jurors.

8. Schwartz, and Schwartz, , supra note 1.Google Scholar In that paper, we demonstrate that the decision rule employed can have a significant impact on the expected number of trials before a verdict is reached, as well as on the expected verdict. There, we focus most heavily on the power of the prosecutor to exert influence over the expected outcome by choosing the charge for the defendant in response to the prevailing decision rule. Here, the alternatives available to the jury are assumed to be fixed.

9. This is true for all states allowing for capital punishment except for Alabama and Florida. We discuss these two below.

10. Current practice is to exclude jurors prior to the guilt determination stage of a trial.

11. 428 U.S. 153 (1976).

12. The following 35 states employ capital punishment; Ala. Code §§ 12-22-150 (1986)Google Scholar, 13A-5-39 to 59 (1982), 15-18-80 to 85 (1992 and 1993 Supp.); Ariz. Rev.Stat.Ann.§13-704 to 13-706, 13-1105 (1989)Google Scholar; Ark. Code Ann. § 5-1-601 to 617, 5-10-101 (1987)Google Scholar; Cal. Penal Code Ann. §187, 189, 190–190.9 (Deering 1985)Google Scholar; Colo. Rev. Stat. Ann. §§16-11-103, 16-11-401 to 405, 18-2-101 to 102 (West 1986)Google Scholar; Conn. Gen. Stat. Ann. §§53A-45, 53A-46A to 46C, 54-99 to 101 (West 1985)Google Scholar; Del. Code Ann. Trr. 11 §4209 (1974)Google Scholar; Fla. Stat. Ann. §§792.04, 921.141-142, 922.09-11, 922.111-112 (West 1976)Google Scholar; Ga. Code Ann. §§16-5-1 (1992).Google Scholar 17-1-30 to 38 (Michie, 1990)Google Scholar; Idaho Code §§18-1001 to 4004, 19-2515 to 2519A. 19-2802. 19-2827 (1987)Google Scholar; Ill. Ann. Stat., Ch. 38 Para. 9-1 to 4, 119-5 (Smith-Hurd 1993)Google Scholar; Ind. Code Ann. §35-38-6-1 to 6-4, 35-42-1-1, 35-50-2-9 (West 1986)Google Scholar; Ky. Rev. Stat. Ann. §§507.020, 532.025, .030, .075, .100 (Baldwin 1992)Google Scholar; La. Code Crim. Proc. Ann. Art. 905-905.9 (West 1984)Google Scholar; Md Code Ann., Crimes and Punishments §§407, 412–414 (1992)Google Scholar; Miss. Code Ann. §99-19-13, -49, -51, -101, -103, -105, -107 (1972)Google Scholar; Mo. Ann. Stat. §565.008, .012, .014, .016, .030, .035, .040 (Vernon 1979)Google Scholar; Mont. Code Ann. §46-18-220, -301 to -310 (1993)Google Scholar; Nev. Rev. Stat. §175.522 (1991)Google Scholar; N.H. Rev. Stat. Ann. §630.5 (1993)Google Scholar; N.J. Stat. Ann. §2C:11-3 (West 1982)Google Scholar; N.M. Stat. Ann. §31-20A-1 (Michie 1994)Google Scholar; N.C Gen. Stat. §15A-2000 (1988)Google Scholar; Ohio Rev. Code Ann. §2929.02 (Baldwin 1992)Google Scholar; Okla. Stat. Ann. Trr. 21, §701.10 (West 1983and Supp. 1994)Google Scholar; Or. Rev. Stat, §163.105 (1993)Google Scholar; Pa. Cons. Stat. Ann. §9711 (1982 and Supp. 1992)Google Scholar; S.C.Code Ann. §16-3-20 (1976 and Supp. 1993)Google Scholar; S.D. Codified Laws Ann. §23A-27A (1988 and Supp. 1994)Google Scholar; Tenn. Code Ann. §39-13-204 (1991 and Supp. 1993)Google Scholar; Tex. Code Crim. Proc. Ann. Art 37.017 (West 1981 and Sup. 1994)Google Scholar; Utah Code Ann. §76-3-206 (1953 and Supp. 1994)Google Scholar; Va. Code Ann. §19.2-264.2 (1990)Google Scholar; Wash. Rev. Code §10.95 (1992)Google Scholar; Wyo. Stat. §6-2-102 (1988)Google Scholar

13. Gallup Poll Monthly, 12 1993.Google Scholar Survey GO 422017, Q. 6. Of the 1,244 respondents asked what they believed should be the penalty for murder, 59 percent responded “death penalty,” 29 percent responded “life imprisonment without parole,” 10 percent responded that “it depends,” and 2 percent responded “neither” or had no opinion.

14. See Enelowand, J.Hinich, M., The Spatial Theory of Voting (1984)Google Scholar, Chs. 1 and 2, for an excellent treatment of social choice along a single dimension.

15. See Ordeshook, P., Game Theory and Political Theory (1987)Google Scholar, and Riker, W., Liberalism Against Populism (1983)Google Scholar for an introduction to social choice.

16. Typically, this is life imprisonment without parole. Cal. Penal Code Ann. §190.05 (Deering 1985).Google Scholar

17. Recently, the Supreme Court reviewed a case in which the jury voting on capital punishment asked thejudge whether the alternative of imprisonment included any possibility of parole (Simmons v. South Carolina, Case 92-9059). Clearly, some of the jurors were contemplating whether they would vote for life imprisonment without parole over capital punishment but might vote for the death penalty if the alternative included the possibility of parole. In the case, the judge refused to provide them with an answer, instructing them to interpret life imprisonment with its “plain meaning.” The defendant was sentenced to death. The Supreme Court reversed and remanded. This result is quite consistent with our approach, in that the Court recognized that the set of available alternatives matters when the jury must exercise discretion. The fact that the Supreme Court has yet to espouse a consistent theory of jury sentencing is reflected in the inability of the seven justices who agreed in the judgment to converge on a majority opinion. In fact, the seven produced four opinions among them, none of which secured more than four signatures.

18. If, for instance, the only alternative to the death penalty were acquittal, the incidence of capital punishment would increase dramatically.

19. The probability that all twelve jurors are not opposed to capital punishment on principle is (0.70)¹² = 0.014. Therefore the probability of selecting at least one juror who is opposed to capital punishment is 1 – 0.014 = 0.986.

20. Even if only 10 percent of the population objects to the death penalty, the probability of randomly selecting twelve jurors without one being such a person is only 28 percent.

21. 391 U.S. 510 (1968).

22. Witherspoon, 391 U.S. at 523. n. 21.Google Scholar

23. Id. at 517–518.

24. 469 U.S. 412, 423 (1985).

25. Cal. Penal Code §190.4(D) (Deering 1985)Google Scholar; Nevada Rev. Stat. §175.556 (1991).Google Scholar The Indiana Statute provides that “If a jury is unable to agree on a sentence recommendation after reasonable deliberations, the court shall discharge the jury and proceed as if the hearing had been to the court alone. In. Code Ann. §35-50-2-9(F) (West 1986).Google Scholar It should be noted, however, that the jury verdict in Indiana is only advisory. In. Code Ann. §35-50-2-9 (D)

26. Cal. Penal Code §194(b), supra note 23.Google Scholar

27. Should the case be an “easy” one in favor of life improsonment (i.e., everyone in society believes this to be the appropriate outcome), thejury will sentence the defendant accordingly. There may, of course, be no “easy” cases for capital punishment. If only one supporter of this result is required to impose it, unanimous support for this result is trivially sufficient.

28. Of course, the likelihood of a case qualifying as “hard” depends critically on the distribution of opinion about capital punishment within the population. Were all citizens to think exactly alike, all cases would be “easy” cases. Suppose, for instance, that the case is “hard,” in that only 90 percent of the population believes that the death penalty is the appropriate sentence, while the remaining 10 percent prefers life imprisonment. For this example, 28 percent of randomly selected juries will return a death sentence. One jury in a trillion will unanimously agree to life imprisonment. And over 70 percent of juries will be hung. This is a very high number, especially in light of how little heterogeneity we have introduced into this example.

29. In our example with 90% of the population favoring the death penalty, a jury that manages to agree to a verdict will be one favoring the death penalty with essentially 100% probability. The likelihood that a unanimously lenient jury is convened first is negligible. If the population is spii 160% in favor of the death penalty in a particular case and -10% opposed, the likelihood that the first unanimous jury convened is one favoring capital punishment is 99%. If the population is split 52% to 48%. the likelihood of the defendant .receiving the death penalty is 72%.

30. Actually, the likelihood of one versus the other occurring depends on the distribution of preferences within the population. The point here is that during all but the “final” trial, the median member of society's preferences are reflected in expectation. Once the “final” trial arrives, the preference of one-twelfth of the population most reluctant to impose capital punishment is decisive.

31. Ala. Code §13A-5–16(1) (1982)Google Scholar

32. Fla. Stat. Ann. §921.14(3) (West 1976)Google Scholar

33. This is almost certainly the result of a conscious decision by these states. The standards set in Wilhmpoon and Wainwright attempted to impose uniform rules for acceptable exclusions. Since different states face different crime problems and no doubt experience different distributions of opinion about the appropriate use of capital punishment, some will be more satisfied with the Supreme Court's standards than others. Those states that wish to affect the set of outcomes in capital cases must then resort to means other than exclusionary rules to do so (decision rule, participation of judges, etc). While it is beyond the scope of this article, one might fruitfully investigate whether the rules employed by the individual states vary systematically with measures of crime rates and public opinion about capital punishment.

34. In Spaziano v. Florida 468 U.S. 447Google Scholar, the Supreme Court upheld Florida's procedure for capital sentencing. In particular, the Court upheld the practice of a judge overruling a jury's recommendation by referring to the state's own interpretation of that practice, as espoused in Tedder v. State 322 So. 2d 908. FL (1975).Google Scholar In Tedder, the state argued that the judge should only impose the death penalty, over the recommendation of ajury to the contrary, if “the facts suggesting a sentence of death are so clear and convincing that virtually no reasonable person could differ.” However, the Supreme Court has since refused to grant certiorarifor any case in which a judge imposed the death penalty contrary to the jury's recommendation. Justices Brennan and Marshall have written numerous dissents to these certiorari decisions, claiming that the Tedder practice has been virtually abandoned in favor of allowing judges to impose whichever sentence they desire. See Heiney v. Florida 469 U.S. 920Google Scholar, Eutzi v. Florida 471 U.S. 1045Google Scholar, Adams v. Florida 475 U.S. 1103Google Scholar, and Engle v. Florida, 485 U.S. 924Google Scholar, for examples of certiorari denials.

In Harris v. Alabama, Case 93-7659, 63 L.W. 30008, (1994).Google Scholar the Supreme Court has agreed to review whether a judge has the right to impose the death penalty in Alabama in spite of a recommendation by the jury to give a sentence of life imprisonment without the possibility of parole (AP, June 27, 1994).

35. See Alabama Statute, supra note 29.Google Scholar

36. However, to the extent that the common legal training of judges reduces the variance in their views, this effect will be mitigated. A discussion of this possibility (and its desirability) is beyond the scope of this paper.

37. Nev. Rev. Stat. §175.556 (1991).Google Scholar The three-judge panel must be unanimous for the death penalty to be imposed.

38. The version that will not be passed might be in the form of a “killer amendment.” See Denzau, A., Riker, W., and Shepsle, K.. Farquharson and Fenno: Sophisticated Voting and Home Style, Apsr 79: 117–34 (1985)Google Scholar, for a nice treatment of one example of sophisticated voting in the U.S. House of Representatives. The example is the Powell amendment to the education aid bill of 1956.

39. See Kaplan, M. and Drupa, S.. Severe Penalties Under the Control of Others Can Reduce Guilt Verdicts, Law and Psychology Review 10:1–18 (1986).Google Scholar for some empirical evidence of nullification by jurors.

40. Actually, Juror 1 can only Torce the jury to be hung because a two-way unanimity rule applies at the guilt stage. However not all hungjurics result in retrials. In any event, it is almost certainly the case that all twelve of the jurors would prefer to give the defendant life imprisonment rather than be hung on the determination of guilt

41. Tex. Code Grim. Proc. Ann. Art. 37.071 (c) (2), (f)(2) (West Supp. 1994)Google Scholar

42. This assumes that there is a correlation between the propensity of a juror to believe the defendant to be guilty and the desire of a juror to impose a more severe sentence. Empirical research shows there to be a strong relationship. See Luginbuhl, J. and Middendorf, K., Death Penally Beliefs and Jurors' Responses to Aggravating and Mitigating Circumstances in Capital Trials, Law and Human Behavior 12(3): 263–81.Google Scholar Jurors who would qualify as Witherspoon-excludables were more sympathetic to evidence in favor of the defendant and more skeptical of aggravating evidence. To the extent that jurors view the guilt determination stage as instrumental, they will exhibit such a correlation even beyond their natural tendencies.

43. 476 U.S. 162.

44. Lockhart v. McCree, 476 U.S. at 167.Google Scholar

45. Id. at 178

46. Id. at 174.

47. John Wasleff wrestles with these difficulties in his article, Lockhart v. McCree: Death Qualification as a Determinant of the Impartiality and Representativeness of a Jury in Death Penalty Cases, Cornell Law Review 72:1075–1113.Google Scholar The author, however, concludes that the Supreme Court decided the case correctly. He recognizes the inconsistencies in Lockhart, but insists that there was no other suitable conclusion to which the Supreme Court could have arrived, given the long-standing practices of voir dire, unanimity, and peremptory challenges.

One referee for this article suggested that symmetric censoring or providing the defendant one extra precmptory challenge for each Witherspoon-excludable might overcome the fairness problem here. In some sense this is posing the question of whether two wrongs could make a right We are not generally optimistic about the use of challenges as a method of generating an increased likelihood of jury consensus. We arc currently working on an article comparing peremptory challengers and relaxed decision rules as substitute instruments for producing decisive juries. We plan to use the opinions in J.E.B. v. Alabama, , 114 S. CL 1419Google Scholar, as motivation for the analysis.

48. We are not the first to question the desirability of the American system of voir dire and peremptory challenges. See Alschuler, A., The Supreme Court and the Jury: Voir Dire, Peremptory Challenge, and the Review of Jury Verdicts, U. Chicago Law Review 56(1): 153–233 (1989)CrossRefGoogle Scholar, for a thorough treatment of the subject. In particular. Alschuler focuses on the ineffectiveness of the current system In remedying problems of racial pejudice.

49. SirDevlin, P., Trial by Jury (Stevens, London, 1988)Google Scholar, in addressing the question of why the unanimity rule was adopted, says, “The answer is that no one ever planned that it should be that way; the rule is simply an antique.” (p. 48).Google Scholar In Cheatle, v. Queen, , 177 C.L.R. 541, 550 (1993)Google Scholar (Austl.), the Court said “The origin of that requirement would seem to lie not in any reasoned development of principle, but in a requirement of the concurrence of twelve jurors in the verdict in the early dap when jurors performed the function of local witnesses in trial by compurgation.”

50. Criminal Justice Act of 1967Google Scholar, ch. 80. §13 (Eng.); Criminal Code Act (NT) §368 (Austl.) (not capital cases); Juries Act of 1927Google Scholar (SA) §57 (Austl.) (not capital cases); Juries Act of 1899Google Scholar (Tas.) §48 (Austl.); Juries Act of 1957Google Scholar (WA) §41 (Austl.) (not death penalty); Criminal Procedure (Majority Verdicts) Act (NI) of 1971 (not death penalty); Criminal Justice Act of 1984, No. 22 §25(Ir.).Google Scholar

51. Scotland uses a fifteen-person jury. There are also three possible verdicts: guilty, not guilty, and not proven. See Affeckv.H.M. Advocate 1987 S.C.C.R. 150 (1987)Google Scholar and Criminal Procedure in Scotland, Second Report. 1975 CMND 6218, at 194–208.Google Scholar

52. Apodaca, v. Oregon, (406 U.S. 404, 1972).Google Scholar

53. Grofman, Bernard, in Tht Case for Majority Verdicts, Trial 15: 23–29 (1979).Google Scholar demonstrates that majority rule has many normatively desirable qualities. Using the data of Kalven and Zeisel, he shows that a switch from unanimity to majority rule has a negligible effect on conviction rates. Second, by applying a Gelfand and Soloman statistical model to Dalven and Zeisel's data, Grofman shows that switching from unanimity to simple majority rule reduces the probability of wrongly convicting a defendant from 0.033 to 0.006.

54. Gillers, S., supra note 5, at 96Google Scholar, concludes that, although the issue is far from clear, it would be constitutionally permissible to require only ten of twelve votes to impose capital punishment.

55. See Aschuler, A., supra note 45Google Scholar, for discussion of this question.

56. Cite Baldus Study.