dissenting.
The Court today seals the fate of this capital defendant. It determines that his death sentence is not disproportionate, and may be carried out. The looseness of the proportionality review that has led to a determination of such awesome finality demands the continued exposure of the fundamental flaws that beset our current system of death-sentence validation.
One of the most prominent defects in the Court’s proportionality review is the continued use of reversed-death sentences in the universe of comparable cases. The reversal of a death sentence robs it of reliability, disabling it as a measure of the deathworthiness of the sentenced defendant. Another major deficiency is the inconsistency and inherent subjectivity of the Court’s techniques for determining proportionality — frequency analysis, and precedent-seeking analysis. The Court persists in applying those methods without any clear standards or intelligible guidelines. As a result, its proportionality review lacks the analytical integrity necessary in the context of a life and death decision. Beyond those problems peculiar to proportionality review, there remains the great systemic flaw of our capital punishment regime, a regime founded on conflicting and contradictory principles and administered without any degree of consistency.
The deficiencies of design and application deprive proportionality review of whatever faint chance it might otherwise have had to provide constitutional legitimacy to the imposition of a death *82sentence. In short, as measured by notions of proportionality, the project of fairly and justly sentencing a defendant to death is doomed to failure. The Court’s proportionality review has, despite good intentions and prodigious work, become so unprincipled that its result — affirming a sentence of death — is intolerable in a society committed to procedural fairness and due process.
This case accentuates all the fundamental difficulties of proportionality review. Defendant was indicted for the kidnapping and murder of a New Jersey businessman. The State prosecuted the case as a capital cause by serving notice that it intended to prove two aggravating factors. At the close of trial, the jury found defendant guilty on all counts of the indictment. The trial court sentenced defendant to death following the jury determination that both aggravating factors existed and outweighed any mitigating factors as found by six jurors. On appeal, this Court affirmed defendant’s conviction and sentence. 131 N.J. 176, 619 A.2d 1208 (1993). Defendant then requested proportionality review pursuant to N.J.S.A. 2C:11-3(e).
The Court acknowledges that this case, involving as it does a kidnapping for ransom, is highly unusual. The Court’s method of proportionality review, problematic in other instances, breaks down in its application here.
I
The purpose of proportionality review is to ensure that the death penalty is “ ‘imposed fairly, and with reasonable consistency.’ ” State v. Marshall, 130 N.J. 109, 130, 613 A.2d 1059 (1992) (Marshall II) (quoting Pulley v. Harris, 465 U.S. 37, 42-43, 104 S.Ct. 871, 875-76, 79 L.Ed.2d 29, 35-36 (1984)). Ultimately, proportionality review aims to guarantee that a cognizable distinction exists between capitally-sentenced and life-sentenced defendants; to limit capital sentencing to the most aggravated cases; and to promote a rational, consistent, and fair application of the death sentence. Id. at 131, 613 A.2d 1059; David C. Baldus, Death Penalty Proportionality Review Project Final Report to *83the New Jersey Supreme Court 24-25 (Sept. 24,1991) (hereinafter Final Report). Proportionality review seeks to determine whether a defendant’s death sentence fails to meet those goals and is, therefore, disproportionate. See N.J.S.A 2C:11-3e. A capital sentence is disproportionate if other defendants with characteristics similar to those of the defendant under review generally receive sentences other than death for committing factually-similar crimes. State v. Bey, 137 N.J. 334, 343, 645 A.2d 685 (1994) (Bey IV).
Our proportionality review evaluates the “deathworthiness” of a defendant, as measured by that defendant’s “blameworthiness.” “Blameworthiness” is an elusive concept, but we nevertheless persist in using it because it encompasses both jury and prosecutorial decisions about the propriety of a death penalty in a particular case. Because the decisions of prosecutors, as well as juries, bear on deathworthiness, we use all death-eligible homicides, including those that prosecutors have chosen not to prosecute as capital crimes, in the comparison calculus with which we assess a defendant’s deathworthiness and the proportionality of the death sentence.1 Marshall II, supra, 130 N.J. at 137, 613 A.2d 1059.
This Court has decided that reversed death sentences provide “sufficiently-reliable information concerning the characteristics that prosecutors and juries consider important to warrant the inclusion of those cases in proportionality analysis” — as death-sentenced cases. Id. at 219-20, 613 A.2d 1059. The Court’s use of reversed death sentences as measures of deathworthiness is, analytically, illogical and treacherous. After all, the procedural *84rules that govern capital sentencing do not exist for their own sake. They exist because, in various significant ways, they attempt to ensure the rationality of the sentence. The failure to abide by those rules, then, fundamentally undercuts the reliability of the sentence. Recognizing that, the Court “acknowledges,” as surely it must, that a reversed death sentence is “a less persuasive indicator of deathworthiness” than an affirmed death sentence. Bey IV, supra, 137 N.J. at 348, 645 A.2d 685. The Court fails, however, to provide any cogent justification for using such death sentences at all in proportionality review. Moreover, the Court continues to leave us totally unenlightened as to how it discounts or accounts for the fact that a reversed death sentence is a “less persuasive” indicator of deathworthiness than an affirmed death sentence.
The Court’s insistence on using reversed death sentences in the same way it uses validly-affirmed death sentences produces the anomaly that a reversed sentence, by definition too unreliable to carry out on the sentenced defendant, is yet reliable enough for the purpose of comparison with other capital sentences in proportionality review. The treachery in the Court’s construct lies in the allowance of a reversed death sentence, which is never reimposed on the principal defendant, to be used collaterally to justify the death sentence of some other defendant. Bey IV, supra, 137 N.J. at 404, 645 A.2d 685 (Handler, J., dissenting).
The wayward premise of the Court’s decision to use reversed death sentences in proportionality review consists in the notion that a defendant’s “deathworthiness” is somehow distinguishable from the final, legitimate verdict reached in his or her ease. However, the only objective indicator that can establish “death-worthiness” is the imposition of a death sentence in strict conformance with the procedures set out to ensure fairness and accuracy in capital sentencing. Hence, the Court’s assumption that a defendant’s deathworthiness can be distinguished from his or her lawful sentence, is, I submit, illogical and unprincipled. I, therefore, strongly reaffirm my view that all reversed death cases *85should be excluded from the universe of cases used to assess deathworthiness and sentencing proportionality.
This Court has held that some errors requiring the reversal of death sentences negate their legality and, derivatively, derogate from their usefulness as markers of deathworthiness.' These errors involve both procedural requirements and substantive principles. Both kinds of standards guide and inform the jury determination of deathworthiness. Errors implicating those standards include: (1) the failure to instruct the jury that the aggravating factors found must outweigh the mitigating factors beyond a reasonable doubt before a death sentence can be imposed; (2) an improper or misleading charge related to the c(4)(c) factor if that was the sole aggravating factor on which the State relied in its attempt to gain a death sentence; and (3) an erroneous charge to the jury that a mitigating factor must be found unanimously. See, e.g., State v. Clausell, 121 N.J. 298, 345-46, 580 A.2d 221 (1990) (requiring that juries be informed of legitimacy and acceptability of non-unanimous, non-death-deserving verdict at penalty trial); State v. Gerald, 113 N.J. 40, 85, 549 A.2d 792 (1988) (requiring that to be death-eligible, defendant must have intended to kill, not merely to have inflicted serious bodily injury); State v. Williams, 113 N.J. 393, 453-54, 550 A.2d 1172 (1988) (reversing death sentence because prosecution relied on victim-impact evidence); State v. Bey, 112 N.J. 123, 162-77, 548 A.2d 887 (1988) (Bey II) (requiring reversal if Court incorrectly instructed jury concerning finding and weighing of mitigating factors); Ramseur, supra, 106 N.J. at 211, 524 A.2d 188 (narrowing grounds for e(4)(c) aggravating factor to infliction of severe suffering).
If a jury returns a sentence of death infected by such an error, then it defies reason to suggest that the verdict accurately, or even approximately, reflects community values about the death-worthiness of the defendant. Put differently, when a sentence of death is imposed without an adequate showing that it was reached by applying the procedural requirements or substantive standards that are essential to a sound, fair, and informed determination, in *86effect, no death-sentence verdict exists, and it is reckless to speculate about the verdict’s reliability as a reflection of community values.
The Court’s response to that criticism is unpersuasive. Ante at 25-28, 651 A. 2d at 959-961. It observes, for example, that in Bey IV the majority rejected defendant Bey’s argument that certain reversed death sentence eases, Ramseur, Biegenwald, and Coyle, should be treated as life-sentenced eases because the errors that negated those death sentences related to burden of proof and the Gerald issue, which affected only the procedural fairness of the trial, not the substance of the crime, and, therefore, “ ‘d[id] not necessarily bear on the jury’s determination of deathworthiness.’ ” Ibid, (quoting Bey IV, supra, 137 N.J. at 347, 645 A.2d 685). It “continues to believe that the death sentences in cases overturned for procedural error are acceptably-reliable societal determinations of deathworthiness.” Ibid.
The Court undervalues the significance of procedural requirements in the context of a capital prosecution and fails to understand the relationship between the procedural structure that frames and directs the jury’s deliberative process and the substantive standards that govern its deliberations and inform its ultimate determination of deathworthiness. “Death sentences are reversed,” I submit, “because at some point the carefully constructed environment of a capital-punishment prosecution has been contaminated, whether by procedural or substantive fault, thereby undermining the soundness and impugning the reliability of the jury’s ultimate determination of deathworthiness.” Bey IV, 137 N.J. at 405, 645 A.2d 685 (Handler, J., dissenting); see also Sullivan v. Louisiana, — U.S.-, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993) (reasoning that when court gives erroneous instruction on burden of proof, no verdict can'be rendered consistent with the Sixth Amendment’s “right to trial by jury”).
In using a death sentence as the critical and definitive measure of deathworthiness, the Court also treats dismissively the fact that a reversed-death-sentenced case has not been retried as a capital *87case. Ante at 25-28, 651 A.2d at 959-961 (“[T]he State’s decision not to reprosecute a defendant capitally is not necessarily a reflection of that defendant’s lack of deathworthiness.”). It rationalizes that conclusion by observing that the State’s decision not to reprosecute a defendant capitally may itself be based on considerations such as the availability of witnesses or the amount of financial resources that the State is willing to commit to retrial, that, it claims, are unrelated to deathworthiness. Although this may be true in certain situations, this reasoning does not change the fact that a reversed death sentence not subsequently reimposed simply and legally does not exist as a valid death sentence. Cf. Special Master, Final Report at 61-62 (expressing opinion that some reversed death sentences might be salvageable for use in proportionality review especially if, on re-trial, another death sentence was imposed). The Court’s position on that vital point bespeaks unseemly intransigence. It “remain[s] convinced that ‘even reversed death sentences are sufficiently valid indicators’ of ‘the conscience of the community’ to be used as death-sentenced cases,” ante at 27, 651 A.2d at 960 (quoting Bey IV, supra, 137 N.J. at 348-49, 645 A.2d 685), but concedes that its judgments have “some degree of subjectivity,” ante at 27, 651 A.2d at 960 (citing Marshall II, supra, 130 N.J. at 120, 613 A.2d 1059), and that “our data are not scientifically infallible,” ante at 27, 651 A.2d at 960 (citing Bey IV, 137 N.J. at 348, 645 A.2d 685).
At the very least, the Court should impose a rebuttable presumption that reversed death sentences are invalid determinations of deathworthiness. The Special Master suggested that the Court “examine the reason for a reversal ... and evaluate, on a case by case basis, the likelihood that the error involved substantially influenced the jury’s exercise of discretion” or otherwise impugned the reliability of the original sentence. Final Report at 61. The Court should not create what amounts to a presumption that reversed death sentences validly indicate deathworthiness and should not require the defendant to overcome that presumption. Ante at 25-26, 651 A.2d at 959-960.
*88I must add, further, that the Court’s use of reversed-death sentences to validate a defendant’s death sentence in proportionality review is a matter of constitutional dimension. The Court professes to accept such sentences only when their “reliability” as indicators of deathworthiness is unquestioned. Both due-process considerations and the prohibition against cruel and unusual punishments, I think, interdict the use in our scheme of proportionality review of reversed death sentences that fail to meet a standard of “sufficient reliability.”
In other contexts, the Court has rightly understood that factors derogating from the reliability of a death sentence implicate the fundamental constitutional concerns for due process and for prohibiting cruel and unusual punishments. When “reliability” is the operative test for the satisfaction of a constitutional standard, it must be met. In Ramseur, supra, 106 N.J. at 316, 524 A.2d 188, the Court stressed the importance of the use of reliable information in the capital-sentencing system stating that “the death penalty can be constitutionally imposed only if the procedure assures reliability in the determination that ‘death is the appropriate punishment in a specific case.’ ” (citations omitted). It has pointed out that when the c(4)(c) factor is the sole basis for the death sentence and is explained by an erroneous instruction, no reliable inference about deathworthiness in fight of community values can be drawn. Ibid. The Court, in reversing the death sentence in State v. Biegenwald, 106 N.J. 13, 65-66, 524 A.2d 130 (1987) (Biegenwald II), similarly concluded that the erroneous instruction about the weighing of aggravating and mitigating circumstances undercuts the fundamental fairness of the entire proceeding. See Bey II, supra, 112 N.J. at 161, 548 A.2d 887; see also Mills v. Maryland, 486 U.S. 367, 108 S.Ct. 1860, 100 L.Ed.2d 384 (1988) (ruling that death sentence must be reversed if jurors are not instructed that they must individually weigh mitigating factors even though they were not found unanimously); cf. Sullivan, supra, — U.S. at-, 113 S.Ct. at 2082, 124 L.Ed.2d at 189-90 (holding that in criminal cases in which verdict was rendered without adequate showing that it had been reached beyond *89reasonable doubt as required by Sixth Amendment’s “right to a jury trial,” no verdict existed on which to apply harmless-error analysis). In Caldwell v. Mississippi, 472 U.S. 320,105 S.Ct. 2633, 86 L.Ed.2d 231 (1985), the United States Supreme Court reversed a sentence of death because of a prosecutor’s inaccurate statement to the jury that the ultimate responsibility for determining the appropriateness of death lay not with the jury, but with an appellate court. Justice Marshall noted that under the Eighth Amendment “the limits that this Court has placed on the imposition of capital punishment are rooted in a concern that the sentencing process should facilitate the responsible and reliable exercise of sentencing discretion.” Id. at 329,108 S.Ct. at 2639, 86 L.Ed.2d at 239. See Woodson v. North Carolina, 428 U.S. 280, 305, 96 S.Ct. 2978, 2991, 49 L.Ed.2d 944, 961 (1976).
The constitutional requirement that a death sentence be “sufficiently reliable” applies as surely in proportionality review, in which a court determines whether the sentence can be validated and carried out, as when a jury makes the original determination of deathworthiness by imppsing the death penalty. Thus, in its proportionality review of defendant’s death sentence, the Court includes death sentences that were reversed for errors that significantly undermined the reliability of the verdicts as reflections of deathworthiness. Those errors, for example, related to aggravating and mitigating factors in the cases of Richard Biegenwald, Walter Gerald, Raymond Rise, James Koedatich, James Williams, James Zola, Bryan Coyle, and Walter Johnson, and to a reversed death sentence that was based on an erroneous c(4)(e) charge in the case of Walter Oglesby. The Court’s use of reversed death sentences as reliable measures of deathworthiness is not a merely theoretical or academic exercise; it has a telling impact on the Court’s assessment of this defendant’s deathworthiness and on its final determination that his death sentence is proportional. Defendant is condemned by a proportionality review that uses as its primary feature comparison to seven other cases in which verdicts of death have been .rejected by this Court as invalid and unreliable.
*90The inclusion of reversed death sentences in the comparison calculus falsely suggests that defendant’s death sentence is proportional. The opposite is the truth. The Court’s continued use of reversed death sentences dooms its proportionality review — and dooms this defendant — from the start. Because the Court persists in following a course that contradicts constitutional principles as well as basic precepts of capital-murder doctrine and common sense, I must strongly register my dissent.
II
The Court’s proportionality review begins with a “frequency analysis.” By that analysis, the Court determines the frequency of imposition of death sentences in cases similar to that of the defendant. Frequency analysis has three methods of considering the defendant’s sentence in light of other capital sentences: (1) general factual similarity; (2) similarity of aggravating and mitigating circumstances; and (3) general similarity in culpability. The thesis underlying the analysis is that “ ‘[a] death sentence is comparatively excessive if other defendants with similar characteristics generally receive sentences other than death for committing factually similar offenses.’ ” Marshall II, supra, 130 N.J. at 153-54, 613 A.2d 1059 (quoting Tichnell v. State, 297 Md. 432, 468 A.2d 1, 17 n. 18 (1983)).
Implicit in the thesis that underlies frequency analysis is the assumption that similar defendants committing similar crimes are' equally blameworthy. In proportionality review, blameworthiness is the marker of deathworthiness. Hence, the determination of comparative blameworthiness is the mission of frequency review. Ante at 29, 651 A.2d at 961.
The Court, however, undermines the ambition of frequency analysis before the analysis even begins, by declining to set any standard by which to distinguish a high- from a low-predicted frequency of death. Ante at 29-30, 651 A.2d at 961-962. In the resulting analytical vacuum, the Court inserts its own predisposition toward equalizing different levels or kinds of blameworthi*91ness. As a consequence, it inevitably engages in a form of reasoning that amounts to little more than a selective and convenient rationalization for proportionality. Its application of frequency analysis betrays a palpable bias in favor of the proportionality of a death sentence.
The Court defends its refusal to set numerical standards by commenting on the “inherent failure” of such rigid numerical standards to “distinguish between defendants.” Ante at 32, 651 A.2d at 963. I do not dispute that the establishment of a clear standard can, with regard to cases very near its threshold, produce arbitrary results. I am convinced, though, that the establishment of some numerical standard for the sake of guidance would substantially improve our proportionality review. With such a standard, we could confine arbitrariness to the cases at the boundary. Under proportionality review as now conducted, arbitrariness runs rampant, for the Court has given no hint as to what degree of infrequency of imposition of the death penalty might constitute disproportionality.
The central weakness of the first of the three methods of frequency analysis, general factual similarity, is exposed by the Court’s exposition. In this case, that method fails because the salient factors of defendant’s case are not similar to those of other cases, and so his ease does not fall in any of the existing categories. Defendant’s is the “only kidnap for ransom case” in the proportionality universe. Lacking any established category that captures the critical facts of defendant’s case, the Court chooses to place defendant in the category of “kidnapping/abduction with particular violence or terror.” Ante at 33, 651 A.2d at 963.2 It also compares defendant’s case with other pecuniary-motive cases involving kidnapping, most of which were categorized as robbery *92with particular violence or terror.3 Still, even accepting this ad hoc categorization as the best that can be done under the circumstances, the frequency with which the death penalty is imposed is very low.
When a hybrid category — cases of premeditated robbery/kidnapping with pecuniary motive, deception/entrapment, and a defenseless victim — is used, the frequency increases to .66. However, that result is problematic. When the data base is corrected .and new cases are added, that frequency drops to .20 with defendant and, indeed, drops away almost entirely to .11, if defendant’s own case is removed. In all, the salient factors test indicates that, as a general matter, the frequency of death sentences for crimes with facts purportedly similar to defendant’s is extremely low.
The second of the three methods of frequency analysis, similarity of aggravating and mitigating circumstances, also supports a finding of disproportionality in this ease. If one begins the analysis by looking simply at cases in which two aggravating factors and two mitigating factors were found, the frequency is 13/48 or .27 (12/47 or .25 without defendant) of the broad death-eligible universe. In cases in which the c(4)(f) factor (avoiding detection) is found as one of two aggravating circumstances and two mitigating circumstances were also found, the frequency rate is 5/18 or .27 (4/17 or .23 without defendant). Also, when the e(4)(g) factor (contemporaneous felony) is found as one of two aggravating circumstances and two mitigating factors were also found, the rate is 8/41 or .19 (7/40 or .17 without defendant). Further, within only the restricted penalty-trial universe, the *93frequency for cases involving two aggravating and two mitigating factors is 13/23 or .57 (12/22 or .54 without defendant).4
In short, the frequency analysis of the number of aggravating and mitigating factors reveals that as a general matter, defendant has a low predicted frequency of receiving a death sentence. Neither of the relevant aggravating factors has a high frequency of death imposition, and the mitigating factor of age has, in the opinion of the Special Master, a pronounced mitigating effect.
The third and final statistical methodology employed by the Court to assess the relative frequency "with which a death sentence is imposed among a group of similar cases is the so-called “index of outcomes” test. This test purportedly groups eases according to their common degree of culpability. The Court, relying on the AOC’s analysis, draws on a broad set of factors, some statutory and some non-statutory, thought to be probative of culpability, to determine patterns of capital sentencing. The test operates on the common-sense assumption that more culpable defendants have a greater probability of receiving the death penalty. Ante at 42, 651 A.2d at 968.
Using the broadest possible test — all death eligible cases with statutory and non-statutory variables — defendant is assigned a rank of .05 with a margin of error having a lower limit of .01 and an upper limit of .30. That places defendant in culpability level 1, which has a predicted frequency of death-sentence imposition of .04 (10/248). Among the twenty cases nearest to defendant’s on the index, none resulted in a death sentence.
*94In the broad death-eligible universe using statutory factors only, defendant is assigned a rank of .08 with a margin of error having a lower limit of .02 and an upper limit of .27. Once again, defendant falls into culpability level 1 with a predicted frequency of death imposition of .05 (12/249). There is only one death sentence among the twenty eases nearest to defendant.
Further, restricting the universe of cases to penalty-trial cases and using only statutory factors as variables, contrary to expectation, one does not dramatically increase defendant’s rank. In that narrower universe, defendant has a culpability rank of .15 with a margin of error having a lower limit of .03 and an upper limit of .48. Defendant remains in culpability level 1, which has a predicted frequency of death-sentence imposition of .05 (3/58).
A significant increase in frequency occurs only by using statutory and non-statutory factors within the narrow universe of penalty-trial cases. In that scheme, defendant is assigned a culpability ranking of .88 with a margin of error having a lower limit of .25 and an upper limit of .99, creating a probability range of seventy-four percent. That places defendant in culpability level 5, which has a predicted frequency of death-sentence imposition of .88 (23/26). Thus, by changing the focus of analysis, the Court can reassign defendant to a ranking where the death penalty is almost uniformly imposed.
The Court does not flinch in the face of its legerdemain, reassigning defendant from culpability level 1 to culpability level 5 by the device of adjusting the parameters of the analysis. The Court explains its preference for these parameters by stating that the several tables that yield extremely low percentages do not include non-statutory factors and therefore do not account for the salient factors of defendant’s crime (ransom and extreme terrorizing of the victim’s family), and, also, that defendant is unique in our case universe. Ante at 45, 651 A.2d at 970. The assertion that the gravamen of defendant’s crime requires consideration of non-statutory aggravating factors would fit more appropriately in the Court’s precedent-seeking review, for that assertion espouses *95the kind of subjective evaluation not susceptible to the statistical techniques of frequency review. Even accepting the Court’s tactic, however, one finds that the index of outcome results are higher for Bey, see Bey IV, supra, 137 N.J. at 362-65, 645 A.2d 685, and slightly higher for Marshall, see Marshall II, supra, 130 N.J. at 172-74, 613 A.2d 1059. Yet the Court does not feel that those results indicate disproportionality or suggest any aberration in defendant’s case with respect to the index-of-outcomes test. Ante at 45, 651 A.2d at 970.
The Court’s analysis and reasoning under the test, using only penalty-trial cases, is expedient and result-driven. The disparity between the results in Table 12, which yield a very high predicted frequency of death sentence, and the results obtained in Tables 13, 15c, and 15b, namely, a very low predicted frequency of death sentence, demonstrates the basic instability in the statistical framework. I cannot overemphasize that Table 12 represents an analysis that is confined to the penalty-trial universe, the results of which, the Court has already conceded, should be used for “informational purposes only” and not as a basic prop for proportionality. Marshall II, supra, 130 N.J. at 168, 613 A.2d 1059. As was the situation in Bey IV and in Marshall II, the small sample size of cases with similar levels of blameworthiness prevents us from giving great weight to those results. See Bey IV, supra, 137 N.J. at 364, 645 A.2d 685; Marshall II, 130 N.J. at 173-74, 613 A.2d 1059. In addition, Table 12 represents the least-reliable model in that it runs the most variables against the fewest number of cases, and therefore has a huge margin of error associated with defendant’s ranking on the culpability index.
The Court proposes to overcome the index-of-outcome test’s low culpability finding by insisting that defendant’s case is “unique in our ease universe.” Ante at 45, 651 A.2d at 970. If so, the project of frequency review becomes a sham, because frequency review assumes the existence of similar cases, and proceeds to consider whether, in light of the sentences imposed in those similar cases, a defendant’s sentence is disproportionate. If, however, the similar*96ity of other cases is ignored or discounted and defendant’s case is deemed unique, other cases are rendered automatically dissimilar and become useless for comparison, and frequency review falters. By declaring the ease itself unique, the Court deserts the basic thesis of frequency review and should not pretend that a death sentence’s proportionality can be established by such a review.
The Court, through a convoluted analysis, finds that compared to other penalty-trial cases, defendant’s case shows predicted probabilities of receiving the death sentence of thirty-three percent under the salient-factors test; fifty-seven percent under the numerical-preponderance test; and, under the index-of-outcomes test, five percent considering only statutory factors and eighty-eight percent considering both statutory and non-statutory factors.5 Ante at 45, 651 A.2d at 970, referring to the Martini Report tbl. 19). The Court concludes that “those results produce no showing of randomness or aberration” and that defendant has not produced “reliable evidence of disproportionality,” and hence does not find that for cases such as his a sentence other “than death is generally imposed.” Ibid.
In actuality, compared to all death-eligible cases, defendant has a predicted salient-factors-test probability of seventeen percent; a numerical-preponderance-test probability of twenty-seven percent; and index-of-outcomes-test probabilities of five percent, considering statutory factors only, and four percent, considering both statutory and non-statutory factors. Martini Report tbl. 20. Thus, as a general matter, defendant’s predicted frequency of *97death sentence imposition hovers between .05 and .27. The conclusion that must be drawn is that under all three statistical methodologies that constitute frequency analysis, defendant falls at the low to extremely-low end of the frequency scale. The Court sidesteps that damning result by observing that death sentencing is reserved for only “the worst murderers.” Ante at 39, 651 A.2d at 967. Obviously, the Court sincerely but subjectively believes that defendant is one of those “worst murderers,” and, embracing that belief, effectively abandons principled proportionality review.
The Court’s opinion only deepens and perpetuates the fundamental flaws in its proportionality review methodology. Although it has lately equivocated on the matter, the Court at first regarded the frequency approach as primary and the precedent-seeking approach as supplementary. Compare Marshall II, supra, 130 N.J. at 159, 613 A.2d 1059 (making the intensity of the precedent-seeking review depend on the outcome of the frequency analysis) with Bey IV, supra, 137 N.J. at 350, 645 A.2d 685 (relying “more heavily” on precedent-seeking as long as the pool of cases remains small). In this opinion, though, we see the Court thoroughly deconstruct frequency review, characterizing its conclusions throughout as statistically unreliable. See ante at 30-31, 33, 35, 36, 651 A.2d at 962, 963, 964, 965, 966. The Court does not explain how statistics so thoroughly discredited in the body of the analysis can, at its end, be characterized, particularly with respect to the salient-factors measure, as “the most persuasive of the statistical measures [that] supports a finding of no disproportionality.” Ante at 37, 651 A.2d at 965.
Inevitably, a frequency analysis bereft of clear standards produces, in the hands of a Court determined to put the burden of proof on the defendant, a finding of proportionality. Bearing the burden of proof, and arguing to a Court that introduces extraneous, precedent-seeking-type considerations into its frequency review, no defendant could establish disproportionality. Proportionality is prescripted.
*98The Court insists that proportionality review does not validate a death sentence, but rather is merely “a vehicle to ensure that the penalty-phase jury’s decision is not insupportable.” Ante at 22, 651 A.2d at 958. To the extent that there is a distinction between ensuring that the sentence is supportable and ensuring that it is not insupportable, the Court identifies a genuine point of disagreement. More significant, though, is the fact that the Court does not even make proportionality review into “a vehicle to ensure that the ... decision is not insupportable.” Rather, the Court allows a sentence that bears compelling indicators of disproportion to be carried out because, owing to the inadequacy of the data and of the techniques for conducting proportionality review, neither the State nor any defendant can positively prove or disprove disproportionality to the satisfaction of the Court. This combination of an insistence that defendant shoulder a rather formidable burden to prove disproportionality, with the circumstances of inadequate data and techniques, makes it bitterly ironic that the Court should say that “[b]eing the first murderer in a category does not support a conclusion of disproportionality.” Ante at 34, 651 A.2d at 964. Indeed, it does not; on the contrary, by the Court’s method, being the first apparently guarantees the proportionality of the sentence.
Ill
The statistical aspect of proportionality review exemplified by frequency analysis is intended to serve as a point of departure from which the Court turns to a more traditional form of judicial review — a case-by-case assessment of the relative deathworthiness of the defendant. Marshall II, supra, 130 N.J. at 153, 613 A.2d 1059. In contrast to the frequency approach, in which we look to groups of cases, the precedent-seeking approach considers each case individually. The Court explains:
Through this method we determine whether a defendant’s criminal culpability exceeds that of similar life-sentenced defendants and whether it is equal to or greater than that of other death-sentenced defendants, such that the defendant’s culpability justifies the capital sentence; or whether a defendant’s culpability is *99more lilte that of similar life-sentenced defendants and less than that of death-sentenced defendants, such that the defendant’s culpability requires a reduction of sentence to a life term.
[Ante at 47, 651 A.2d at 971.]
An appropriately critical view of the whole enterprise of precedent-seeking review begins by pointing out a certain conceptual vagueness in the method itself. In introducing its review, the Court writes:
We note, as we did in Bey IV, supra, that even closely-similar cases do not require identical verdicts to be proportionate, in light of the different defendants, juries, facts, and legal issues involved.
[Ante at 51, 651 A.2d at 973.]
The Court’s statement exemplifies both the irrationality of proportionality review and the means by which it conceals that irrational- • ity.
First, one notices how the statement conceals. The notion that closely similar cases do not require identical verdicts at first hardly offends, because although “closely similar,” the cases are of course different and so can hardly claim to warrant “identical” treatment. The problem, though, is that the cases that the Court dismisses as merely not “identical” fill the chasm between life and death. Recasting the Court’s statement in plain terms, one discovers that the Court allows that from cases that do not differ much or are “closely-similar” one defendant can get death and another life.
If it believes that literally, the Court comes close to accepting arbitrariness, insofar as it would seem the very mark of arbitrariness if two closely-similar eases can produce two such starkly different verdicts. Because the Court does not want to abide arbitrariness, it adds the qualification quoted above: “in light of the different defendants, juries, facts, and legal issues involved.” Ibid. The Court explains that “because juries must consider each case individually, we cannot expect the same result even in similar cases,” ante at 77, 651 A.2d at 986 (citing Bey IV, supra, 137 N.J. at 369, 645 A.2d 685). Accordingly, even if those factually-dissimilar cases have similar levels of culpability, that circum*100stance does not require, according to the Court, the conclusion that the jury in defendant’s case acted aberrantly in sentencing him to death. Id. at 78, 651 A.2d at 986.
The rub is that once we admit the uniqueness of every case, precedent-seeking review becomes unmanageable. Contending with the unmanageable, the Court in this case is forced to make subjective and arbitrary decisions at each step of its analysis. For example, the Court rejects without good reason defendant’s two suggested generalized features that, help identify similar culpability. The first of these involves the “number of victims.” Can one seriously question that killers of more than one person are more culpable, other things being equal, than killers of only one? The Court evades the obvious, direct answer to that question by answering a different question:
We cannot say that a murderer who MUs one person through torture is necessarily less culpable than a murderer who kills more than one person quickly.
[Ante at 53, 651 A.2d at 973-974]
The Court is correct, but misses the point. In effect, the Court confuses two distinct steps in the analysis. First, there is a relevance step, in which the Court should distinguish those factors that bear on culpability from those factors that do not. Second, after identifying the factors that bear on culpability, there follows a significance step, in which the Court considers the significance or probative weight of each factor in the particular ease. In this case, however, the Court condenses the two steps into one. The result of this condensation is a test for relevance that no factor can ever meet, because no factor can, by its mere absence, render any defendant “necessarily less culpable” than a defendant who has that factor. The point becomes clear if one substitutes the factor of pecuniary gain for the multiple victim factor. Applying the Court’s test, we must exclude as irrelevant the pecuniary gain factor because one cannot say that “a murderer who kills [without pecuniary motive but] through torture is necessarily less culpable than a murderer who kills [for pecuniary gain] quickly.”
*101I do not suggest that the Court in fact reject as irrelevant a comparison based on pecuniary gain. Rather, I find deficient the Court’s test of the relevance of factors. In order to decide whether some suggested factor warrants consideration, we must first consider whether that factor as such bears on culpability. That analysis is' undertaken by temporarily holding other things equal, enabling the Court to distinguish those factors, such as pecuniary motive, number of victims, and torture, that do bear on culpability, from those, such as for example the race of the victim, that do not bear on culpability. The Court, however, merges these steps without taking the trouble first to identify the multiple-victim factor as relevant to culpability, and then to explain what weight or significance should be accorded that factor in assessing culpability. Thus, the Court recognizes that “[t]he number of victims obviously affects blameworthiness,” and, at the same time, concludes that it “is an extremely important component of it.” Ante at 47, 651 A.2d at 971. The Court fails, however, separately to consider or elaborate on the weight or significance of the “obvious [ejffect” that this factor has in assessing culpability, and almost glibly, “reject[s] ... defendant’s arguments that one who purposefully and knowingly murders a single victim is less deathworthy than one who murders more than one victim.” Ante at 52-53, 651 A.2d at 973. Hence, without any acknowledgement that a factor, such as number of victims, is directly relevant to culpability and without an explanation of its significance, the Court’s conclusion as to defendant’s deathworthiness remains imprinted with its own subjective and intuitive assessment of comparative culpability.
It is possible, but hardly edifying or productive, to engage the Court in a debate about the significance of the particular comparison cases in assessing the proportionality of defendant’s sentence. Thus, against the Court’s selection of and emphasis on the features surrounding those with sentences of death, one could oppose the various heinous features of the crimes of those defendants in the comparison group who were not sentenced to death. Because defendant is the first ransom killer in the universe, one who *102regards that crime as uniquely terrible can easily declare that defendant is the worst of his comparison group, and hence deserving of death. But how does the Court know that defendant’s crime was worse than Mayron’s heinous murder, or those of many others in the group?
One can find horrible features in many of the cases, and this fact should require the Court to explain how it knows that the brand of suffering inflicted by defendant is indeed worse. Objective standards by which to measure the gravity of different varieties of suffering are hard to find. Perhaps we ought to look to the Legislature for guidance. When we do so, we find that the Legislature has decided which cases are worst, by turning their particular features into aggravating factors. Until the Legislature makes the infliction of grave suffering on another besides the victim an aggravating factor, the Court ought to assume that that circumstance, although obviously relevant to culpability, is not uniquely worse than circumstances present in other eases. So long as the Court allows its unexplained intuitive moral response to the crime to dictate the outcome of its precedent-seeking review, our system of proportionality review does not live up even to the promise of Marshall II, supra. Instead, our review comes to resemble that species of proportionality review that entails no more than a recital of past cases and a bald conclusion that the particular death sentence is proportionate. See, e.g., State v. Harris, 870 S.W.2d 798, 819 (Mo.1994) (en banc).
Recognizing the special authority of objective factors, the Court professes to evaluate the comparison cases based only on objective criteria that were presented to the jury, citing Bey IV, supra, 137 N.J. at 368, 645 A.2d 685. Ante at 50, 651 A.2d at 972. It attempts to consider non-statutory factors that are objective, rooted in traditional sentencing guidelines, clearly submitted to the jury, and likely to influence a jury’s decision. Ibid.
According to the Court, the relevant factors for determining the proportionality of defendant’s death sentence are, first, the statutory factors of the contemporaneous felony of kidnapping and the *103murder to escape detection, and, second, the demand for ransom, which likens defendant’s case to other murders that involve a pecuniary motive. Ante at 50, 651 A.2d at 972. The Court recognizes that defendant’s is the only kidnapping case involving a demand for ransom. Ante at 77, 651 A.2d at 985-986. It chooses to consider those cases in the Martini Report in which the defendants committed kidnapping with particular violence or terror, were contract principals, or were contract killers. It then undertakes to decide whether defendant’s culpability is more like that of defendants who received death sentences or of those who received life terms.
The Court further relies on the fact of the extreme victimization of defendant’s family, a victimization made all the more reprehensible in the eyes of the Court for having been necessary for completion of the criminal scheme. The Court concludes that those elements create a different and greater kind of culpability than is present in the other non-stranger-kidnapping cases, and accordingly, it finds no disproportionality in defendant’s death sentence when compared to the life sentences imposed in the other non-stranger-kidnapping cases. Ibid.
Despite its effort at objectivity, the Court’s reasoning and conclusion expose the pervasive subjectivity of its precedent-seeking analysis. An equally convincing case can be made with respect to many of the other defendants in the comparison group that their crimes were uniquely terrible. Consider, for example, the case of Jamie Barone, the sole defendant in the stranger-kidnapping category. One could readily conclude that defendant is less culpable than Barone because Barone, although protesting his innocence, failed to show any remorse. Further, Barone could be considered more blameworthy because of the greater level of victimization present in his case. Although the State tried Bar-one’s ease as a capital prosecution, the jury was unable to reach a unanimous decision in the penalty phase.
Similarly plausible, if not just as strong, arguments can be made for the view that other defendants in the comparison group exceed *104defendant in blameworthiness. For example, a murder that occurs when the victim’s death is itself essential and central to the successful completion of a criminal scheme could readily be viewed as most culpable. Indeed, examining the comparison group using the distinction between such preplanned murders, referred to by defendant as “fatal-precondition murders,” and other murders, one finds a strong correlation between death sentences and such fatal precondition murders. Thus, among the fatal-precondition murderers are Robert O. Marshall (killed wife for insurance proceeds and received death sentence), Patrick Lanzel (killed two people to ensure receipt of inheritance and did not face penalty trial), Anthony DiFriseo (contract killer who received death sentence), and James Clausell (killed to impress organized crime figures and received death sentence, which was reversed, on remand received life sentence). Furthermore, if we compare defendant’s case, in which the pecuniary motive was a central part of the crime, with approximately thirty-four others6 involving a robbery or kidnapping for pecuniary gain and a premeditated, deliberate, execution-style killing, only two, Bobby Lee Brown and Jacinto Hightower, received death sentences. Moreover, five of those cases involved the murder of two people (John Allen, Bobby Lee Brown, Frank Masini, John Marsieno, and Ray Watson), which, as defendant argues, implies greater culpability than the murder of only one person. Additionally, as defendant points out, at least twelve of the non-death-sentenced defendants in the remaining group of twenty-nine cases inflicted extreme victimization sufficient to ren*105der them more culpable and deathworthy than he is.7
Without explanation, the Court rejects the second of defendant’s suggested generalized features: the notion that a murder essentially necessary to the accomplishment of another criminal act exceeds in culpability other murders. Without saying more, it simply does “not accept the proposition that one who, with intent to kill, commits a crime such as robbery or rape is always less culpable than one who commits a ‘fatal precondition’ crime.” Ante at 52, 651 A.2d at 973. Once again, the Court evades the direct question, which asks only whether, other things being equal, a murder necessary to complete a criminal scheme is worse than other murders. There may not be any objectively justified answer to that question. However, the intuition that would find such a murder worse than others cannot be gainsaid, and the Court gives no objective reason for rejecting it out of hand.
These examples illustrate the kind of “insoluble moral conundrums” that precedent-seeking analysis presents. Marshall II, supra, 130 N.J. at 274, 613 A.2d 1059 (Handler, J., dissenting). Indeed, whether one is more culpable when engaging in a crime in which death is a necessary pre-condition, as argued by defendant, or in which death is only a highly-probable outcome calls for a moral judgment. Such moral judgments are endlessly debatable. Thus, it is reasonable to believe that the purposeful killing of two persons is more heinous than the killing of just one. It is also fair to consider that defendant’s crime did not include the kind of sadistic torture and depravity that surrounds so many murders in the death-eligible universe. Indeed, the Court’s own moral compass has shifted in these cases, seeming most impressed in Marshall II by the culpability attached to killing for money, 130 N.J. at 166-167, 613 A.2d 1059, and in Bey IV by the combination of sexual assault and murder. 137 N.J. at 367, 645 A.2d 685.
*106Marshall II requires that when subjective value judgments are made, the Court make those judgments “explicit so they can be analyzed and tested against whatever objective measurements are applicable.” Id. at 120, 613 A.2d 1059. The Court must have a coherent theory of what makes one case more deathworthy than another. It is not enough simply to say that the Court merely assesses the reasonability of a jury’s determination when in fact it only speculates about a possible reason to support the jury’s determination and makes its own subjective judgment of the moral significance of that reason. In proportionality review, the Court professes to review the determination of deathworthiness made by a jury to see if that determination is aberrant. Concededly, that process of review will involve some level of second-guessing. As Marshall II noted, a “value judgment is built into practically every measurement of proportionality.” Id. at 119, 613 A.2d 1059. If, in reality, the Court is making its own judgment of deathworthiness, that value-laden judgment must be made explicit and must be given the force of precedent if this aspect of proportionality review is ever to be fair, consistent, and intelligible. The Court’s precedent-seeking analysis in this case falls woefully short.
IV
The Court, I firmly believe and emphatically urge, must recast its proportionality review of death sentences. As I noted with respect to the defendant in Bey IV, falling, as this defendant does, in the low range of predicted frequencies of death sentence imposition makes him peculiarly vulnerable to the ad hoc character of this Court’s frequency analysis. 137 N.J. at 409, 645 A.2d 685 (Handler, J., dissenting). The Court’s analytical failure with respect to frequency analysis serves only further to devalue the already scant protections defendant is afforded by frequency analysis. Although originally designed as the more objective of the two methods that make up proportionality review, frequency analysis as applied by the Court does little more than set the *107stage for whatever subjective determinations or moral judgments might be made under the precedent-seeking approach. Ibid.
One cannot escape the vagueness of the Court’s explication of the interaction of frequency analysis and precedent-seeking review. The Court says simply that “We * * * compare the results of the two analyses to ensure that our proportionality review is reliable.” Ante at 27, 651 A.2d at 960. In past cases, it has said that a finding of borderline proportionality under the frequency approach will require a heightened scrutiny of the results of the other approach. Marshall II further suggests that the value determinations made at the precedent-seeking stage should be “analyzed and tested” by the “objective” criteria that are available. 130 N.J. at 120, 613 A.2d 1059. That implies that some degree of commensurability should exist between the two approaches. With both Marshall and Bey, at least arguably, some commensurability was present between the statistical analysis and the precedent-seeking analysis. That analytical luxury is not available in this case. Here, the ambiguity of the relationship of the approaches confounds the Court’s deteimination of proportionality.
The Court’s opinion serves as a call for the re-examination and reconstruction of proportionality review. It demonstrates that precedent-seeking review and frequency analysis coexist in a muddled relationship characterized by the worst kind of circular reasoning. If these analytical tools are individually imprecise, it is unclear how the Court expects two such tools in combination to produce a mechanism that serves reliably to prevent arbitrariness in death sentencing.
The errors that infect the Court’s proportionality review are symptoms of the fundamental incoherence of our capital murder jurisprudence. The Court’s continued uncritical use of reversed death sentences as a measure of deathworthiness is an irrationality of the first order. The Court’s application of frequency analysis is driven by no more than intellectual convenience and institutional expedience, with little care given to distinguish a high from a low frequency and with lax attention to its functional relationship to *108precedent-seeking review. Lacking a workable methodology, the Court over relies on a precedent-seeking analysis that is riddled with subjectivity and moral judgments.
Today’s decision serves as further confirmation of the failure of our experiment with capital punishment. The Court’s early belief that it could fashion a eonstitutionally-legitimate process for imposing the death penalty, see Ramseur, supra, 106 N.J. at 831, 524 A.2d 188 (“How we will resolve this paradox remains as yet fully unrevealed to us. We shall continue to labor on the process.”), has foundered on yet another rock — proportionality review. The inconsistency, subjectivity, and moralizing evident in today’s decision are the inevitable products of a futile endeavor: the quest to devise and to apply a standard of due-process protection commensurate with the gravity of a death sentence. We noted in Ramseur that proportionality review was to assist the Court in ensuring that “ ‘we have designed procedures [that] are appropriate to the decision between life and death.’ ” 106 N.J. at 326, 524 A.2d 188 (quoting Pulley, supra, 465 U.S. at 67-68, 104 S.Ct. at 888-89, 79 L.Ed.2d at 52). The signal failure of proportionality review as designed and applied by this Court is now manifest. I think it also evident that the Court must either reject its effort to carry out capital punishment or accommodate itself to the juridical brutality of imposing death without due-process protections commensurate to its awesome finality. See Callins v. Collins, — U.S.-,-, 114 S.Ct. 1127,1129,127 L.Ed.2 d 435, 438 (1994) (Blackmun, J., dissenting).
We are constrained in capital cases to concentrate unremittant attention and expend enormous public resources on persons who deserve no sympathy whatsoever. Sympathy has nothing to do with our judicial duty; our common humanity, however, has everything to do with it. If we allow death to be imposed without the full measure of constitutional protection and defend endlessly the legitimacy of what we do, we invite only disrespect for the law. Bey IV, supra, 137 N.J. at 429, 645 A.2d 685 (Handler, J., dissenting) (citing John C. Jeffries, Jr., Justice Lewis F. Powell, *109Jr., 452 (1994)). Surely the time has come for the Court to concede that no death sentence can be validated under a process of review that is extraordinarily vague, rife with contradictions, wildly inconsistent, and inextricably mired in subjective valuations and intuitive moral judgments.
For affmnance — Chief Justice WILENTZ and Justices CLIFFORD, POLLOCK, O’HERN, GARIBALDI and STEIN — 6
For reversal — Justice HANDLER — 1
The death-sentence frequency in category H(2) (Kidnapping/abduction with particular violence), including the defendant, is 1/7 or .17 If the defendant is removed, the frequency is zero, 0/7 or .00 Of the three cases out of seven that went to penalty trial, 1/3 or .33 received death. Yet again, if defendant is removed the frequency is again zero, 0/3 or .00.
In the E(2) category of robbery with particular violence, only two of twenty-one cases in the death-eligible universe received the death penalty, 2/21 or .10 (not including defendant). If Raymond Rise's case is excluded, see, supra at 22, 651 A.2d at 958, then the frequency drops to 1/20 or .05. Of those cases that went to penalty trial, 2/7 or .29 received the death penalty.
If two reversed death sentences are changed to life (Lodato and Ramseur) and two death sentences are eliminated because based on "deliberative errors" (Biegenwald 1A and James Zola 1A), the overall frequency reported in Table 9 drops to 9/46 or .19 (8/45 or .17 without defendant). Applying the alternative assumptions in relation to the analysis of the c(4)(g) factor — changing one death sentence to life (Lodato) and deleting one case (Zola) — results in a new frequency of 6/40 or .15 (5/39 or .13 without defendant). Applying the alternative assumption to the c(4)(f) factor does not require any changes in cases, and thus the numbers remain the same.
As the Court concedes, the rate of penalty trials in death-eligible cases and the rate of death sentencing for cases that advanced to a penalty trial are both fairly low — forty-two percent (125/298) and thirty percent (38/125) respectively. Martini Report tbls. 2, 3. Ante at 30, 651 A.2d at 962. That yields a total death-sentencing rate of thirteen percent (38/298). It masks this problem by stating "we are looking for a potential aberration, not a perfect comparison to all other cases,” observing further, " ‘Not every statistical disparity establishes disproportionality.’ ” (quoting Bey IV, supra, 137 N.J. at 352, 645 A.2d 685. Ante at 30, 651 A.2d at 962).
The 34 cases are: (1) John Allen; (2) Jamie Barone; (3) Shawn Jackson; (4) Richard Reddon; (5) David Mark Russo; (6) Terence Robert Scales; (7) Howard Thompson; (8) Joseph Armstrong; (9) Anthony Carrozza; (10) Dwayne Caviness; (11) Frank Masini; (12) John Marsieno; (13 Ira Musgrove; (14) Corey Washington; (15) George Lazorisak; (16) Matthew Ploppert; (17) Rafael Slaughter; (18) Roy Watson; (19) Bobby Lee Brown; (20) Jacinto Hightower; (21) Benjamin Balisonomo; (22) Craig Hart; (23) Hashona Clark; (24) Carl Culley; (25) Michael Darby; (26) Marvin James; (27) Bruce King; (28) William McCray; (29) Daniel Nicini; (30) Carl Norman; (31) Kevin Smith; (32) Quincey Spruell; (33) Stanley Tucker; and (34) Richard Cain.
The twelve are: (1) Jackson; (2) Scales; (3) Thompson; (4) Carrozza; (5) Caviness; (6) Musgrove; (7) Ploppert; (8) Darby; (9) Norman; (10) Smith; (11) Tucker; and (12) Cain.