In State v. Marshall, 123 N.J. 1, 586 A.2d 85 (1991) (Marshall I), this Court affirmed defendant’s conviction for the murder of his wife and the resulting sentence of death. This appeal requires us to resolve the “proportionality” issue reserved in that case. Id. at 170, 586 A.2d 85. The Capital Punishment Act provides that on the “request of the defendant, the Supreme Court shall * * * determine whether the [death] sentence [imposed on a defendant convicted of capital murder] is disproportionate to the penalty imposed in similar cases, considering both the crime and the defendant.” N.J.S.A. 2C:11-3e. In the process of making that determination we also discuss the broader issue of the format of proportionality review.
In State v. Ramseur, 106 N.J. 123, 524 A.2d 188 (1987), we noted that “[t]he proportionality review provision in the Act is an important procedural mechanism to safeguard against the arbitrary and capricious imposition of the death penalty.” Id. at 330, 524 A.2d 188. Observing that proportionality review implicated “difficult and sensitive issues,” we forecasted that
our efforts to devise a procedure of review that will adequately protect defendants from the arbitrary and capricious imposition of the death penalty prohibited by Furman v. Georgia, supra, 408 U.S. 238, 92 S.Ct. 2726, 33 L.Ed.2d 346, will be an evolving process. In addition to involving criminal justice experts, these efforts may involve experts from disciplines outside the law. We shall seek the advice of such experts to assist us in this process.
[Id. at 328, 524 A.2d 188.]
Therefore, by Order dated July 29, 1988, this Court appointed Professor David C. Baldus of the University of Iowa Law School as Special Master to assist us in developing a system for proportionality review. We requested that the “Special Master * * * produce for the Court a data base and files sufficient to enable the Supreme Court to conduct proportionality reviews as required by statute.” We authorized consideration of “the data *118base that formed the basis of the report of the New Jersey Public Defender entitled ‘the Re-Imposition of Capital Punishment in New Jersey’ ” and the collection of such “additional data * * * as may be needed.” We directed that nothing in the Order shall “be construed by the Special Master or the parties to represent a position of the Court on any issue, nor shall the recommended findings and conclusions of law of the Special Master include any determination concerning the excessiveness or disproportionality of any death sentence imposed in any case.”
In an earlier proceeding, In re Proportionality Review Project, 122 N.J. 345, 585 A.2d 358 (1990), we declined to determine in advance the appropriate “universe” of cases against which to compare challenged death sentences in order to assure proportionality. The Attorney General had contended that the only appropriate universe is one comprised exclusively of cases in which a death sentence has been imposed under our Capital Punishment Act and that establishing such a universe would be consistent with the practice of a majority of other jurisdictions that have developed proportionality-review systems. In addition, we noted that defining the universe of cases available for proportionality review would not automatically determine which cases within the universe will be used in the review of any specific death sentence. We awaited the Master’s Final Report, which was received on September 24, 1991.
Following oral argument, the Legislature amended the Capital Punishment Act to provide that “ [p]rop ortionality review * * * shall be limited to a comparison of similar cases in which a sentence of death has been imposed.” L. 1992, c. 5 (eff. May 12, 1992). Although that amendment is to “take effect immediately,” the Legislature did not state whether it intended the amendment to apply to pending appeals. The Attorney General has filed a letter memorandum suggesting that the 1992 amendment be applied to this appeal. Were the amendment to be applied to pending appeals, we would undoubtedly be required to resolve whether, as applied to offenses committed before its *119effective date, the Act might constitute an ex post facto law. In Collins v. Youngblood, 497 U.S. 37, 110 S.Ct. 2715, 111 L.Ed.2d 30 (1990), the United States Supreme Court summarized the meaning of the ex post facto clauses:
“It is settled, by decisions of this Court so well known that their citation may be dispensed with, that any statute which punishes as a crime an act previously committed, which was innocent when done; which makes more burdensome the punishment for a crime, after its commission, or which deprives one charged with crime of any defense available according to law at the time when the act was committed, is prohibited as ex post facto.”
[Id. at-, 110 S.Ct at 2719, 111 L.Ed.Zd at 39 (quoting Beazell v. Ohio, 269 U.S. 167, 169-70, 46 S.Ct. 68, 68, 70 L.Ed. 216, 217 (1925)).]
The Attorney General argues that A. 1992, c. 5 “clarifies” the prior law and thus does not transgress those bounds. Because of the long pendency of this appeal, we decide this appeal under prior law. Furthermore, given our rejection of the disproportionality challenge, the new law, clearly strengthening such rejection, could not affect the outcome. We therefore have no occasion to consider either its applicability or validity. Hence, our references to a “universe” of cases required for statutory proportionality review are to be understood to refer to the statutory provisions in effect at the time of this offense. Because there are several capital appeals pending under the prior law, we address the issues in sufficient detail to deal with those appeals as well as this, depending on the ultimate effect of L.1992, c. 5.
Much of the discussion is designed to explain the process of record-gathering and the methods of analyses, both of science and law, that can be used to conduct proportionality review and to assess the relevance of the data to system-wide claims of unconstitutional infliction of the death penalty. Because we address both the individual death sentence and the format for proportionality review under our capital-sentencing scheme, we refer either to defendant, Robert O. Marshall, or the Public Defender where appropriate to the context.
*120Although we recognize that proportionality is not a scientific determination, we have attempted to make our determinations as precise in terms of their bases and reasoning and as objective as possible. We have used scientific and statistical measures, when helpful, although we recognize that a value judgment is built into practically every measurement. A life is at stake, and although some degree of subjective value judgment may be required, we have attempted to make those judgments explicit so that they can be analyzed and tested against whatever objective measurements are applicable.
Parts of this opinion, then, that deal with the computer-based process of record-gathering are technical in nature; the remainder is more traditional legal analysis. For convenience, we outline the different aspects of the appeal.
TABLE OF CONTENTS
Introduction.....................................117
I. Facts...........................................121
II. Proportionality Review Defined..................124
III. The “Universe” from which “Similar Cases” are Drawn....................................131
IV. Identifying the Comparison Group of “Similar Cases” in the Universe........................141
A. The Catalog of the Cases ...................141
B. The Methods of Selecting a Comparison Group of Similar Cases from the Catalog.....143
1. By Salient Factors.......................146
2. By Aggravating/Mitigating Factors.......146
3. By Common Characteristics ■ Relating to all Life/Death Outcomes............147
C. The Criteria for Comparing the Cases for
Disproportionality.........................148
1. Frequency Analysis — Will Statistics Demonstrate that a Sentence is Disproportionate? .........................152
2. Precedent-Seeking Analysis — What Specifics of Comparison Cases will Demonstrate that a Sentence is Disproportionate? .........................154
D. Should we Consider an Alternative Method of Analysis by Categories of Culpability *121Derived Solely from the Statutory Aggravating Factors?........................159
Y. Application of the Methods of Proportionality Review to Robert O. Marshall.................166
A. Frequency Analysis — Suggests a Relatively High Degree of Blameworthiness in Contract-Killing Cases.....................166
B. Precedent-Seeking Comparison with Specific Cases does not Demonstrate Disproportionate Exercise of Sentencing Power....................................174
VI. That Juries in New Jersey do not “Generally” Return Death Verdicts does not Undermine the Deterrent Value of the Death Penalty to Such an Extent as to Render it a “Cruel and Unusual Punishment.”.........................188
VII. Geographic Patterns of Charging and Prosecuting Capital Cases do not Demonstrate an Arbitrary Exercise of the Prosecutorial Function......................................195
VIII. Neither the Race of the Victim Nor the Race of the Defendant Has Been Shown to be an Impermissible Invidious Factor in the Imposition of the Death Penalty....................207
IX. Revisions That Can be Made Now and in the Future to Simplify the Data-Gathering Process ......................:...................216
X. Conclusion......................................218
I
Facts
The facts of the Marshall I case are more fully stated in the Court’s earlier opinion. 123 N.J. at 28-62, 586 A.2d 85. We repeat only a general outline of the facts that the jury could have found as drawn from the State’s brief.
Defendant, a Toms River insurance agent, began an extramarital affair with Sarann Kraushaar, a married woman, in June 1983. As early as December 1983, defendant mentioned *122to Kraushaar the idea of killing his wife, Maria. In May 1984, defendant met Robert Cumber of Louisiana and questioned him about hiring an “investigator.” Defendant later telephoned Cumber, who referred defendant to Billy Wayne McKinnon, a former sheriffs officer from Louisiana. Defendant agreed to pay McKinnon $5,000 to meet him in Atlantic City, New Jersey.
Defendant met McKinnon at Harrah’s Casino in Atlantic City on June 18,1984, and offered to pay him $65,000 to kill his wife. In addition to the $5,000 that McKinnon had already received, defendant agreed to pay him $10,000 up front and $50,000 from the expected insurance proceeds on his wife’s life. At that meeting defendant paid McKinnon $7,000 and gave him a picture of his wife. Defendant told McKinnon to kill her that evening, when defendant would be present. In preparation for the killing, defendant and McKinnon discussed various ways to kill Maria. Defendant believed that he would not be considered a suspect because he was considered an outstanding citizen with influence in the community.
McKinnon did not carry out the murder at that time, but instead returned to Louisiana. Defendant communicated with him on numerous occasions and sent him additional money. Under pressure from defendant to complete the job, McKinnon returned to Atlantic City on July 19, 1984, and met with defendant, who proposed a second plan for the killing to take place that evening. Defendant told McKinnon that he would leave his wife in their car to be executed while defendant went into a restaurant under the pretense of using the bathroom facilities. However, McKinnon did not commit the murder at that time either.
Defendant, persistent in his efforts to have his wife killed, offered McKinnon an “extra fifteen” ($15,000) if he would return to New Jersey a third time to do the “job” before Labor Day. McKinnon agreed, and, on September 6, 1984, he and defendant met at a service area parking lot located south of Toms River. Together they selected a spot on the Garden State *123Parkway to carry out Maria’s murder and made final plans for the slaying, which was to occur that evening. The plan was to make the murder look like a robbery.
Defendant took his wife to Harrah’s Casino in Atlantic City on the night of September 6, 1984, under the pretext of an evening of dining and gambling. He met McKinnon outside Harrah’s at approximately 9:30 p.m. and told him that he and Maria would be leaving the casino at about midnight. Defendant also asked McKinnon for the return of the photographs of Maria and of their home that he had given him in June.
As previously arranged with McKinnon, defendant pulled into the Oyster Creek picnic area at milepost seventy-one on the Garden State Parkway at about 12:30 a.m. on September 7. While his wife lay sleeping on the front seat, defendant got out of the car under the ruse of needing to repair a flat tire. Defendant squatted down to prepare himself for being hit on the head as part of the simulated robbery. Maria Marshall was shot in the back twice. She died immediately.
When the police arrived on the scene, defendant continued to make the murder look like a robbery. The State argues that defendant showed no remorse after the crime, but pretended to join his three sons in grieving over the loss of their mother. The State argued at the trial level that he even staged a suicide attempt. Defendant protested his innocence then and continues to do so now in explanation of his conduct.
Defendant’s claims of innocence soon unraveled. Telephone records traced him to McKinnon, who turned State’s evidence. In exchange for a plea to conspiracy to commit murder, McKinnon implicated Marshall and identified a Louisiana man, Larry Thompson, as the triggerman.
Investigation disclosed that during his planning, defendant had been increasing the insurance policies on his wife’s life. At the time of her death, Maria Marshall’s life was insured for about $1,400,000. Defendant had been paying his wife’s premiums while neglecting his own. Defendant hastened to complete *124an application for a policy for a home mortgage before the murder. On the last day of her life, Maria underwent a physical examination for that policy. The State offered proof that defendant could have been motivated to kill by rising debts incurred in his business, including a $128,000 home-equity loan and a short-term bank debt in excess of $40,000. While amassing those large insurance policies, defendant also continued his relationship with Sarann Kraushaar, with whom he had intended to live after the murder.
A jury acquitted Thompson of the murder but accepted McKinnon’s version of defendant’s role and found him guilty of conspiracy to commit his wife’s murder and of murder-by-hire. The only aggravating factor submitted to and found by the jury was that defendant had hired another to commit murder. N.J.S.A. 2C:11-3c(4)(e).1 The two mitigating factors submitted to and found by the jury were that defendant had no history of criminal activity, c(5)(f), and the catch-all mitigating factor, c(5)(h). At the time of the offense defendant was forty-four years of age, and had been involved in charitable and community activities. The jury unanimously found beyond a reasonable doubt that the aggravating factor outweighed the mitigating factors. The trial court sentenced defendant to death.
II
Proportionality Review Defined
A.
The best way to understand the concept of proportionality review is to understand its origin. In Furman v. Georgia, 408 U.S. 238, 92 S.Ct. 2726, 33 L.Ed.2D 346 (1972), the Supreme Court invalidated Georgia’s death-penalty statute as violating *125the Eighth Amendment’s prohibition against cruel and unusual punishment. The teaching of Furman was that a state may not leave the decision of whether a defendant lives or dies to the unfettered discretion of the jury because such a scheme inevitably results in death sentences that are “wantonly and * * * freakishly imposed” and “are cruel and unusual in the same way that being struck by lightning is cruel and unusual.” Id. at 309-10, 92 S.Ct. at 2762-63, 33 L.Ed.2d at 390 (Stewart, J., concurring). Prior to that decision, the capital-sentencing procedures in most states delegated to judges and juries plenary authority to decide when a death sentence should be imposed. The sentencer was given “practically untrammeled discretion to let an accused live or insist that he die.” Id. at 248, 92 S.Ct. at 2731, 33 L.Ed.2d at 355 (Douglas, J., concurring).
Following the Furman decision, many states revised their capital punishment acts. In a series of cases decided four years after Furman, the Court upheld the capital-sentencing statutes of Texas, Florida, and Georgia, concluding that those statutes contained safeguards that promised to eliminate the constitutional defects noted in Furman. See Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976); Proffitt v. Florida, 428 U.S. 242, 96 S.Ct. 2960, 49 L.Ed.2d 913 (1976); Gregg v. Georgia, 428 U.S. 153, 96 S.Ct. 2909, 49 L.Ed.2d 859 (1976). The Supreme Court based its conclusion on the premise that those statutes insured that sentencers would be “given guidance regarding the factors about the crime and the defendant that the State, representing organized society, deems particularly relevant to the sentencing decision.” Gregg, supra, 428 U.S. at 192, 96 S.Ct. at 2934, 49 L.Ed.2d at 885 (plurality opinion of Stewart, Powell, and Stevens, JJ.).
Justice Stewart’s plurality opinion in Gregg cited two features of Georgia’s scheme that would guide and channel the exercise of sentencing discretion. Georgia’s statute had a bifurcated procedure for deciding a defendant’s guilt first and sentence later, and also provided for “the further safeguard of *126meaningful appellate review” of every death sentence. Id. at 195, 96 S.Ct. at 2935, 49 L.Ed.2d at 887.
When New Jersey reintroduced its Capital Punishment Act, it modeled its statute on Georgia’s statute, which had generally followed the Model Penal Code’s provisions with respect to the enumeration of aggravating and mitigating factors and the provision of the bifurcated procedure. Ramseur, supra, 106 N.J. at 183, 202-09, 524 A.2d 188. Hence, as enacted, the New Jersey Capital Punishment Act required that the Supreme Court conduct proportionality review to determine whether the death sentence imposed on a defendant is “disproportionate to the penalty imposed in similar cases, considering both the crime and the defendant.” L. 1982, c. 111.
Following the Supreme Court’s decision in Pulley v. Harris, 465 U.S. 37, 104 S.Ct. 871, 79 L.Ed.2d 29 (1984), that proportionality review was not an essential constitutional requirement of a state capital-sentencing scheme, our Legislature amended the statute to make proportionality review an option for defendants. L.1985, c. 478. We assumed that most defendants who receive a death sentence would request proportionality review. Ramseur, supra, 106 N.J. at 327, 524 A.2d 188. As noted, the Legislature has since limited statutory proportionality review to a comparison of death-sentenced cases.
B.
What did our Legislature intend when it provided for proportionality review in the context of Gregg v. Georgia? To answer that question we must digress to distinguish between two aspects of proportionality review. The first has been referred to as “substantive” proportionality review, the second as “procedural” proportionality review. See Lisa G. Bradley, Proportionality in Capital and Non-Capital Sentencing: An Eighth Amendment Enigma, 23 Idaho L.Rev. 195, 206-08, 211-15 (1986-87). We may think of those two aspects of proportionality review as offense-oriented and offender-orient*127ed. Simply stated, the substantive or offense-oriented proportionality review looks to whether the punishment of death is excessive for a particular offense, while procedural or offender-oriented review examines whether, when compared to factually similar cases involving the same offense, a defendant’s death sentence is excessive. See David C. Baldus, Charles A. Pulaski & George Woodworth, Comparative Review of Death Sentences: An Empirical Study of the Georgia Experience, 74 J.Crim.L. & Criminology 661, 665-66 (1983) (hereinafter Baldus and Pulaski I).
Gregg is particularly instructive because it illustrates the two differing aspects of proportionality review and gives insight into the most probable meaning of our statutory provision. The Gregg Court spoke of the useful function of proportionality review and characterized it as assuring that “ ‘no death sentence is affirmed unless in similar cases throughout the state the death penalty has been imposed generally * * *.’” Gregg, supra, 428 U.S. at 205, 96 S.Ct. at 2939, 49 L.Ed.2d at 892 (quoting Moore v. State, 233 Ga. 861, 213 S.E.2d 829, 832 (1975)); see also Jarrell v. State, 234 Ga. 410, 216 S.E.2d 258, 270 (1975) (asking whether “juries generally throughout the state have imposed the death penalty”), cert. denied, 428 U.S. 910, 96 S.Ct. 3223, 49 L.Ed.2d 1218 (1976).
The kind of proportionality review that asks whether the death penalty is “generally” imposed is an Eighth Amendment inquiry into substantive proportionality — does the punishment fit the crime? (That analysis has also been used to analyze Eighth Amendment disproportionality for certain classes of offenders, e.g., minors, Stanford v. Kentucky, 492 U.S. 361, 109 S.Ct. 2969, 106 L.Ed.2d 306 (1989), or mildly retarded, Penry v. Lynaugh, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989).)
That type of substantive review is best perceived in the context of cases such as Coker v. Georgia, 433 U.S. 584, 97 S.Ct. 2861, 53 L.Ed.2d 982 (1977), and Enmund v. Florida, 458 *128U.S. 782, 102 S.Ct. 3368, 73 L.Ed.2d 1140 (1982). In Coker, for example, the Court held that the imposition of the death penalty for rape violated the Eighth Amendment’s prohibition against cruel and unusual punishment because it was “grossly disproportionate and excessive punishment” for the commission of that crime. 433 U.S. at 592, 97 S.Ct. at 2866, 53 L.Ed.2d at 989. In reaching that conclusion, the Court paid particular attention to such factors as “public attitudes concerning a particular sentence — history and precedent, legislative attitudes, and the response of juries reflected in their sentencing decisions.” Ibid.
Similarly, in Enmund, the Court found that the imposition of the death penalty is disproportionate for one who “aids and abets a felony in the course of which a murder is committed by others but who does not himself kill, attempt to kill, or intend that a killing take place or that lethal force will be employed.” 458 U.S. at 797, 102 S.Ct. at 3376, 73 L.Ed.2d at 1151. Tracking the analysis followed in Gregg and Coker, the Enmund Court held that when “[t]he evidence is overwhelming that American juries have repudiated imposition of the death penalty for [particular] crimes [such as rape or accomplice-murder],” id. at 794, 102 S.Ct. at 3374, 73 L.Ed.2d at 1150, death is an unconstitutional penalty absent a showing that the actor killed, attempted to kill, or intended to participate in or facilitate a murder. Id. at 798, 102 S.Ct. at 3377, 73 L.Ed.2d at 1152. The Court observed that “if prosecutors rarely sought the death penalty for accomplice felony murder * * * it would tend to indicate that prosecutors, who represent society’s interest in punishing crime, consider the death penalty excessive for accomplice felony murder.” Id. at 796, 102 S.Ct. at 3376, 73 L.Ed.2d at 1151. When “legislatures and juries [have] firmly rejected the penalty of death,” id. at 814, 102 S.Ct. at 3385, 73 L.Ed.2d at 1162 (O’Connor, J., dissenting), for a particular offense, then the imposition of the death penalty for that crime is a substantially disproportionate and excessive punishment.
*129In later applications of that doctrine, as in Tison v. Arizona, 481 U.S. 137, 107 S.Ct. 1676, 95 L.Ed.2d 127 (1987) (imposition of death penalty for reckless indifference murder not overwhelmingly repudiated), the Court has adhered to that basic premise of substantive or Eighth Amendment disproportionality. Only in that context must the type of near-unanimous generality be found, and it is not a case-by-case generality but rather a jurisdiction-by-jurisdiction generality of rejection.
When the Gregg court spoke favorably of the Georgia Supreme Court’s requirement that the offense be one in which the death penalty has been imposed generally, it was actually referring to the Eighth Amendment, substantive analysis, because Gregg involved two sentences of death: one for armed robbery and one for murder. (At that time armed robbery standing alone was a statutory basis for capital punishment in Georgia.) The Georgia court vacated the death sentence for armed robbery but affirmed the sentence of death for murder. Gregg v. State, 210 S.E.2d 659, 667 (1974). In the case of robbery, it simply noted that the imposition of death sentences for that crime were “unusual in that they are rarely imposed for [armed robbery].” Id. at 667. Recognizing that the magnitude of the punishment imposed must be related to the degree of harm inflicted on the victim, absent the victim’s murder, juries generally would not impose the sentence of death for armed robbery.
But that offense-oriented analysis is not the review that we exercise here. Rather, we ask whether the “punishment fits the criminal.” The procedural, offender-oriented proportionality review undertaken by the Georgia Supreme Court with respect to the murder count consisted simply of a recital of a series of cases considered by that court in making its proportionality analysis and a statement that “[a]fter considering both the crimes and the defendant and after comparing the evidence and the sentences in this case with those of previous murder cases, we are * * * of the opinion that these two sentences of *130death are not excessive or disproportionate to the penalties imposed in similar cases.” Ibid. That court did not suggest or require that the sentence of death be “generally imposed” in the sense of near unanimity in the comparison cases.
C.
When the Supreme Court later held in Pulley v. Harris that proportionality review was not constitutionally required, it made clear that it was not discarding the Eighth Amendment proportionality analysis. The Pulley Court emphasized:
At the outset, we should more clearly identify the issue before us. Traditionally, “proportionality” has been used with reference to an abstract evaluation, of the appropriateness of a sentence for a particular crime. Looking to the gravity of the offense and the severity of the penalty, to sentences imposed for other crimes, and to sentencing practices in other jurisdictions, this Court has occasionally struck down punishments as inherently disproportionate, and therefore cruel and unusual, when imposed for a particular crime or category of crime.
********
The proportionality review sought by Harris * * * and provided for in numerous state statutes [referring specifically to Georgia’s] is of a different sort. This sort of proportionality review presumes that the death sentence is not disproportionate to the crime in the traditional sense. It purports to inquire instead whether the penalty is nonetheless unacceptable in a particular case because disproportionate to the punishment imposed on others convicted of the same crime.
[465 U.S. at 42-43, 104 S.Ct. at 875-76, 79 L.Ed.2d at 35-36 (footnotes omitted).]
The dissenting members in Pulley suggested not that in any sense there be a requirement of generality or nearly unanimous death verdicts for those convicted of the same crime, but rather suggested only “ ‘that discretion must be suitably directed and limited so as to minimize the risk of wholly arbitrary and capricious action’,” id. at 63, 104 S.Ct. at 886, 79 L.Ed.2d at 49 (Brennan, J.) (quoting Gregg, supra, 428 U.S. at 189, 96 S.Ct. at 2932, 49 L.Ed.2d at 883), and have “ ‘insiste[d] that capital punishment be imposed fairly, and with reasonable consistency, *131or not at all.’ ” Ibid. (quoting Eddings v. Oklahoma, 455 U.S. 104, 112, 102 S.Ct. 869, 875, 71 L.Ed.2d 1, 9 (1982)). In their view, proportionality review, “[although clearly no panacea, * * * often serves to identify the most extreme examples of disproportionality among similarly situated defendants.” Id. 465 U.S. at 71, 104 S.Ct. at 890, 79 L.Ed.2d at 53.
That, we believe, is an acceptable understanding of the intentions of the framers of our Act — that statutory proportionality review should seek to ensure that the death penalty is being administered in a rational, non-arbitrary, and evenhanded manner, fairly and with reasonable consistency. That review serves as “a means through which to monitor the imposition of death sentences and thereby to prevent any impermissible discrimination in imposing the death penalty.” Ramseur, supra, 106 N.J. at 327, 524 A.2d 188.
In conducting such proportionality review, then, one need not search for the nearly unanimous degree of generality that attends the Eighth Amendment rejection of the death penalty for particular crimes or categories of crimes as being disproportionate punishment. The quest in some ways is for the antithesis of that argument. Maryland has expressed its search for proportionality thus: “[A] death sentence is comparatively excessive if other defendants with similar characteristics generally receive sentences other than death for committing factually similar offenses in the same jurisdiction.” Tichnell v. State, 297 Md. 432, 468 A.2d 1, 17 n. 18 (1983) (citing David C. Baldus, Charles A. Pulaski, George Woodworth & Frederick A. Kyle, Identifying Comparatively Excessive Sentences of Death: A Quantitative Approach, 33 Stan.L.Rev. 1 (1980) (hereinafter Baldus and Pulaski II)).
m
The “Universe” from which “Similar Cases” are Drawn
The first step in any proportionality review undertaking is to establish the “universe” of cases that the Court will consider.
*132In our earlier decision, In re Proportionality Review Project, supra, 122 N.J. 345, 585 A.2d 358, we declined to make a preliminary determination of the relevant universe of cases. The Attorney General had contended that the only appropriate universe is one comprised exclusively of those cases in which a death sentence has been imposed. The Legislature has since adopted the Attorney General’s view. A. 1992, c. 5.
In the Final Report, the Master expressed the view that “penalty-trial cases are the narrowest universe that could support a coherent proportionality review system,” noting that at least nine other state courts conduct proportionality review on the basis of penalty-trial cases. David C. Baldus, Death Penalty Proportionality Review Project: Final Report to the New Jersey Supreme Court, 1, 45 (Sept. 24, 1991) (hereinafter Final Report). Nevertheless, the Master recommended that the universe also include clearly death-eligible homicide cases that did not advance to a penalty-phase hearing because of prosecutorial decisions not to seek the death penalty.
Although the universe issue has been vigorously contested, we consider the controversy to be significantly overstated. How detailed a compilation of homicide cases is required to facilitate an adequate proportionality review of a given death sentence depends on the purposes to be served by that review. We assume that the basic difference in the respective positions of the parties about the breadth of the field of homicide cases to serve as a source for proportionality review stems from disagreement about the objectives to be achieved by proportionality review. By identifying those objectives we shall also determine the appropriate universe of cases.
A.
We offer this preliminary observation. The Attorney General, in briefs and at oral argument, objects, to the inclusion of non-penalty-phase homicide cases in the universe, contending that consideration of such cases questions the correctness of *133the prosecutor’s discretion to seek or not to seek a death penalty in a specific case. In our view, that objection misconceives the issue. Courts that conduct proportionality review by considering both death-sentenced cases and life-sentenced penalty-phase cases focus not on whether the jury decision was correct, but rather on whether the differences in the dispositions of comparable homicide cases are relevant to whether the death sentence under review may be disproportionate. Similarly, a universe that includes death-eligible homicides that prosecutors elect not to prosecute as capital cases may also be relevant to proportionality review for purposes of comparison with a specific death sentence, irrespective of the merits of the prosecutor’s discretionary decision not to treat the homicide as a capital murder case.
In Ramseur, we specifically acknowledged that the basic purpose of the proportionality review afforded by our Capital Punishment Act is to determine whether the death penalty in a specific case is “ ‘disproportionate to the punishment imposed on others convicted of the same crime.’ ” 106 N.J. at 326, 524 A.2d 188 (quoting Pulley v. Harris, supra, 465 U.S. at 43, 104 S.Ct. at 875, 79 L.Ed.2d at 36).
Were the Court to limit the focus of proportionality review to that inquiry — whether a specific death sentence is disproportionate to the punishment imposed on others convicted of the same crime — comparison with only cases in which a death sentence was imposed would be inadequate. A simple example illustrates the point. On the assumption that 100 robbery-felony-murder cases are prosecuted as capital crimes, all defendants are convicted and one defendant is sentenced to death, a comparison of the death-sentenced defendant’s punishment with the punishment imposed only on other death-sentenced defendants would exclude from the proportionality-review process the ninety-nine robbery-felony-murder defendants that juries did not sentence to death. Indisputably, the determination whether that single death sentence is disproportionate can be made only by comparing it with the life sentences imposed on *134the ninety-nine defendants convicted of the same crime. We therefore consider it self-evident that the universe for proportionality review must, at a minimum, include all penalty-trial cases.
Comparison of a specific death sentence with the punishments imposed on other defendants convicted of capital murder and sentenced either to death or life imprisonment addresses the possibility that a jury may impose a disproportionate death sentence on a specific defendant. A closely-rélated issue is raised by the argument that a death sentence may be disproportionate, at least in part, because prosecutors frequently exercise their discretion not to seek the death penalty for a specific type of homicide. We first adverted to that expanded function of proportionality review in Ramseur, where we raised the question whether
to expand the potential cases for comparison to include all those in which the death penalty could have been requested by the State. Here we may anticipate considering whether to address concerns about possible misuse of prosecutorial discretion * * * including in the review all cases in which a prosecutor had the discretion to seek the death penalty.
[106 N.J. at 329, 524 A.2d 188 (citation omitted).]
We reiterated that concern in State v. Koedatich, 112 N.J. 225, 548 A.2d 939 (1988), cert. denied, 488 U.S. 1017, 109 S.Ct. 813, 102 L.Ed.2d 803 (1989), noting that the development of prosecutorial guidelines for the selection of capital cases would “promote uniformity in the administration of the death penalty, which will be an additional safeguard against arbitrariness and an assistance to this Court in its developing proportionality review.” Id. 112 N.J. at 258, 548 A.2d 939. More recently, in State v. Kiett, 121 N.J. 483, 582 A.2d 630 (1990), we again acknowledged, that disproportionality can originate in both prosecutorial and jury decisions. Id. at 498, 582 A.2d 630.
The point may best be illustrated by the prior example of 100 robbery-felony-murder defendants, only one of whom is sentenced to death. Were we to assume that the remaining ninety-nine defendants were prosecuted and convicted of non-capital *135murder because of prosecutorial decisions not to seek the death penalty, the disproportionality of the single defendant’s death sentence would arise not because of a disproportionate jury determination but because the prosecutorial decision to seek the death penalty was unique. That type of disproportionate death sentence could not be identified by a proportionality-review process that was limited to capital cases tried to a penalty phase; it could be identified, however, by a universe that included clearly death-eligible homicides that were not prosecuted as capital cases.
The Master’s report emphasizes that prosecutorial decisions not to seek the death penalty for certain death-eligible homicides are influenced by the prosecutors’ predictions of death-worthiness, ie., the likelihood that a jury would impose the death penalty after a penalty-phase trial. Accordingly, one of the purposes served by a universe expanded to include such death-eligible homicides not prosecuted as capital crimes is that the proportionality-review process can then consider both jury and prosecutorial decisions about deathworthiness in determining whether a specific death sentence is disproportionate.
We identified yet a third potential purpose for proportionality review in Ramseur — the prevention of “any impermissible discrimination in imposing the death penalty.” 106 N.J. at 327, 524 A.2d 188. We stated that “[discrimination on the basis of race, sex, or other suspect characteristic cannot be tolerated,” ibid., and noted that “factors such as race, sex, and socioeconomic status might also be appropriate considerations for reviewing proportionality.” Id. at 330, 524 A.2d 188. In that connection, we considered preliminarily in Koedatich, supra, 112 N.J. at 255-58, 548 A.2d 939, the contention that county-by-county as well as race-of-victim related disparities had already been detected in the prosecution of death-penalty cases. Elsewhere in this opinion we address those portions of the Master’s report that consider the geographic distribution of capital-charging and sentencing decisions within the state, infra at 195-207, 613 A.2d at 1102-1108, as well as the report’s sugges *136tion of a possible discrepancy in capital-sentencing rates that may be correlated with the race of the victim or the race of the defendant. Infra at 207-215, 613 A.2d at 1108-1113. We note that our history and traditions “would never countenance racial disparity in capital sentencing,” infra at 207, 613 A.2d at 1108. We consider and evaluate the statistical data in the Master’s report suggesting the possibility that a black defendant has a substantially greater risk of being sentenced to death than do other defendants, infra at 208-215, 613 A.2d at 1109-1112, as well as data suggesting that certain white-victim homicides are substantially more likely to advance to penalty trial than homicide cases involving non-white victims. Infra at 210, 613 A.2d at 1110.
Although our evaluation of those data persuades us that the observed deviations do not “compel a conclusion of substantial discriminatory effect” in the administration of our death-penalty law, infra at 212, 613 A.2d at 1111, we note that the Master’s statistical data concerning race-of-defendant and race-of-victim disparity necessarily include death-eligible homicides that did not advance to penalty trial. Similarly, the report’s data concerning the geographic distribution of capital-charging and capital-sentencing decisions also rely on death-eligible homicides that were not charged as capital murder. The conclusion is inescapable that a universe restricted to penalty-phase cases would be inadequate to enable us to verify that our capital-sentencing procedure does not tolerate “discrimination on an impermissible basis, including, but not limited to, race and sex.” Ramseur, supra, 106 N.J. at 330, 524 A.2d 188.
We note that courts in a few other jurisdictions have acknowledged the appropriateness for proportionality-review purposes of considering death-eligible homicides that did not advance to a penalty-phase hearing. See Tichnell v. State, supra, 468 A.2d at 18 (concluding proportionality-review process may take into account non-capital murder cases); State v. Moore, 210 Neb. 457, 316 N.W.2d 33, 44 (conducting proportionality review by comparison with all other first-degree-murder convic *137tions), cert. denied, 456 U.S. 984, 102 S.Ct 2260, 72 L.Ed.26 864 (1982); State v. Williams, 205 Neb. 56, 287 N.W.2d 18, 28-29 (1979) (same), cert. denied, 449 U.S. 891, 101 S.Ct. 255, 66 L.Ed.26 120 (1980); Commonwealth v. Pursell, 508 Pa. 212, 495 A.2d 183, 198 (1985) (conducting proportionality review by comparison with other first-degree-murder cases in which evidence could support an aggravating circumstance); State v. Rupe, 108 Wash.2d 734, 743 P.26 210, 229 (1987) (concluding that for purposes of proportionality review, similar cases include cases in which defendant convicted of first-degree murder regardless of whether death penalty was sought), cert. denied, 486 U.S. 1061, 108 S.Ct. 2834, 100 L.Ed.2d 934 (1988); State v. Harris, 106 Wash.2d 784, 725 P.2d 975, 982-83 (1986) (conducting proportionality review of death sentence for contract killing court considered contract-murder cases in which death penalty was not sought by prosecutor), cert. denied, 480 U.S. 940, 107 S.Ct. 1592, 94 L.Ed.26 781 (1987).
Accordingly, we hold that the purposes to be achieved by proportionality review require that the universe include clearly death eligible homicides in which the prosecutor elected not to seek the death penalty. We are also satisfied with the process by which the Master identified those death eligible cases that did not proceed to a penalty trial. A short summary of that process will be sufficient to explain the methodology used.
B.
By a review of statistics compiled by the New Jersey State Police, the Master reviewed the 3200 homicides committed in New Jersey since August 6, 1982. He eliminated homicides that were clearly not death eligible, such as cases involving juveniles, death-by-auto, or acquittal in a murder trial. He did the same with other non-penalty trial homicide cases that involved indictments for less than murder. The only cases to survive the preliminary screen were (a) pleas to murder, felony murder, and aggravated manslaughter when the original *138charge was a form of murder, and (b) jury convictions for any form of murder. Approximately 1500 cases remained.
The second stage involved the evaluation of the death eligibility of the remaining cases in terms of the defendant’s conduct, the mental state involved, and the presence of statutory aggravating circumstances. The Master and' staff provided through the Administrative Office of the Courts (AOC) consulted presentence reports and court records, trial counsel, and/or appellate records. Based on information gained from those sources, the cases were provisionally coded as (1) clearly death eligible, (2) questionable concerning death eligibility, or (3) clearly not death eligible. As the cases were screened, progress reports were sent to the offices of the Attorney General, the County Prosecutor’s Association, and the Public Advocate. Although meetings were held, neither the Attorney General’s Office nor the County Prosecutor’s Association responded to requests for information on specific cases.
All penalty-trial cases were listed initially as “clearly death-eligible.” Some were later reclassified because of changes in the law or because no aggravating factors were found at the penalty trial. Non-penalty-trial cases were classified as death eligible if the prosecutor had declined to prosecute the case capitally but there was evidence to support conclusions (a) that the defendant had intended to kill the victim, (b) that the defendant had killed the victim by his own conduct or paid another to kill the victim, and (c) that at least one statutory aggravating factor had been present.
Those cases were further classified as clearly death eligible, questionable, and clearly not death eligible. On the basis of evidentiary standards, the case would remain in the clearly death-eligible category only if its evidentiary strength was strong or overwhelming.
The study started with 3,200 homicides. After the threshold screen that excluded cases involving death-by-auto, juvenile defendants, and acquittals, 1496 cases remained. Following the *139factual-case screen, the evidentiary-strength screen, and the screen of the penalty-trial cases, the Master identified 246 clearly death-eligible cases. Of the clearly death-eligible cases, 132 had resulted in capital-murder convictions and had advanced to a penalty trial and 114 were cases in which the prosecutor had not sought the death penalty. Two hundred and fifty questionable cases were excluded from the universe.
We realize that other courts throughout the country have used different measures of comparison. For example, the Georgia Supreme Court uses all capital-felony cases that have been appealed for comparison purposes. Ross v. State, 233 Ga. 361, 211 S.E.2d 356 (1974), cert. denied, 428 U.S. 910, 96 S.Ct. 3222, 49 L.Ed.2d 1217 (1976). That court found that preventing the imposition of an arbitrary sentence “is ably served by reference to appealed cases which represent a sufficient cross section of similar cases upon which an adequate comparative review can be made.” Id. 211 S.E.2d at 359. Maryland’s universe of cases is comprised solely of “those first degree murder cases in which the State sought the death penalty * * * whether it was imposed or not.” Tichnell, supra, 468 A.2d at 17. Similarly, Delaware’s universe consists of a comparison between the subject case and “the penalties in all first degree murder cases which have gone to trial and a penalty hearing.” Dawson v. State, 581 A.2d 1078, 1108 (1990) (citations omitted), judgment vacated, 503 U.S.-, 112 S.Ct. 1093, 117 L.Ed.2d 309 (1992). The Pennsylvania Supreme Court uses a universe that requires “an independent evaluation of all cases of murder of the first degree convictions which were prosecuted or could have been prosecuted [as capital cases].” Commonwealth v. Frey, 504 Pa. 428, 475 A.2d 700, 707, cert. denied, 469 U.S. 963, 105 S.Ct. 360, 83 L.Ed.2d 296 (1984).
We do not doubt that a pool of all cases that go to a penalty trial would form a reasonably-reliable data base or that a pool of cases in which aggravating factors have been served would also be a reliable pool. The State’s expert, Dr. Herbert I. Weisberg, suggests the use of a penalty-trial pool in which at *140least one aggravating factor has been found. We discuss Dr. Weisberg’s report further in Part IV, section D, infra at 159-166, 613 A.2d at 1085-1088. That too is obviously a reasonably-reliable pool of cases.
Had it appeared to be an insurmountable task to examine all “clearly death eligible cases,” we might have made a mid-course correction. In re Proportionality Review, supra, 122 N.J. 345, 585 A.2d 358. During the proceedings involved in that appeal we learned that approximately ten to fifteen additional cases per year might have to be added to the pool of death-noticed cases to establish a pool of clearly death-eligible cases. Id. at 347, 585 A.2d 358.
We routinely gather information concerning all aspects of the criminal-justice system. The Task Force on Minority Concerns has evaluated, on a system-wide basis, the effect of race in the criminal-justice system. See Interim Report of the New Jersey Supreme Court Task Force on Minority Concerns (Aug.1989) (hereinafter Interim Report). The New Jersey Code of Criminal Justice requires that the Criminal Disposition Commission review “all aspects of the criminal justice system relating to the disposition of criminal offenders,” and submit an annual report detailing its findings and recommendations concerning the disposition of criminal offenders. See N.J.S.A. 2C:48-1 to -4. The New Jersey Code of Juvenile Justice, N.J.S.A. 2A:4A-20 to -91, has likewise created a Juvenile Delinquency Commission, N.J.S.A. 2A:4A-49(b), “to study and review the provisions of the [Code of Juvenile Justice] and all aspects of the juvenile justice system with particular reference to delinquency trends and dispositions.”
Because of the use that we intend to make of the data, as we shall explain them throughout the course of this opinion, we see no need to address in detail the claimed deficiencies in the data-gathering process undertaken by the Master. Suffice it to observe that while the litigation was pending, his efforts were hampered by the adversarial interests of the parties for which *141no blame is to be assessed. We shall, in Part IX of this opinion, infra at 216, 613 A.2d at 1113, suggest a means to address those claimed deficiencies.
IV
Identifying the Comparison Group of “Similar Cases” in the Universe
A.
The Catalog of the Cases
Once the universe of cases has been identified for comparison, there are two further steps to the process of proportionality review. The first is the development of the set of characteristics that will identify a comparison group of similar cases and similar defendants; the second is the development of methods of analysis that will enable the Court to identify disproportionality.
The first step requires the creation of what is today called a data base. Another generation of lawyers and judges would have thought of that process in terms of a series of index cards with the details of cases on them. One wishing to sort the cases into groups of “similar cases” would read the cards for distinguishing characteristics, such as a rape-murder, and place them within one pile. Within that group of rape-murders, one might sort the cases further by noting the cards that showed torture or mutilation of the victim or a prior murder as aggravating factors or, in contrast, the cards that revealed recognized mitigating factors, such as remorse or cooperation with the authorities.
That type of sorting process is described as an a priori or clinical approach to case comparisons. Here, the lawyer’s basic skills developed from the case-method analysis taught in law school are used. The reviewing court uses intuition or experience to select features that it determines probably influenced the life/death decision.
*142If there is a sufficient number of cards with a definable pattern of similar characteristics, a court can evaluate whether a particular death sentence fits the pattern established by the case characteristics. A good illustration of that process is found in the dissenting opinion of Judge Davidson of the Maryland Court of Appeals in which he considered a series of cases involving the shooting deaths of public officials or other victims in the course of a robbery and the various aggravating and mitigating factors in those cases. Tichnell, supra, 468 A.2d at 27. The National Center for State Courts recommends that a court begin its process of review similarly by examining cases on “all fours” with the subject case. David C. Baldus, Charles A. Pulaski, Jr. & George Woodworth, Arbitrariness and Discrimination in the Administration of the Death Penalty: A Challenge to State Supreme Courts, 15 Stetson L.Rev. 133, 176 & n. 89 (1986) (hereinafter Baldus and Pulaski III).
If there is an insufficient number of cases fitting that pattern, a court might wish to consider developing a pattern of similar characteristics by matching features in other classes of cases. That matching process would require some form of cross-index from different piles of cards and would produce yet another set of cases with a broader cross-section of converging characteristics — for example, all felony-factor cases, whether rape, robbery, or kidnapping, that displayed similar characteristics such as no prior record or the influence of extreme mental or emotional disturbance.
A second type of sorting might start with no preconceived notions of what might bring about either a life verdict or a death verdict. Then one would look at all the cards and attempt to choose from them the features that are common to life outcomes and those that are common to death outcomes. That method of case comparison is described as an actuarial or empirical analysis of the cases. That method recalls that of the epidemiologist. For instance, if a heavy smoking habit is found *143in 90 out of 100 cases of lung cancer, smoking may be thought to be a cause of the cancer.
The empirical approach to classifying capital cases as similar or dissimilar seeks to employ “those characteristics of the cases that best explain the sentences actually imposed.” Baldus and Pulaski III, supra, 15 Stetson L.Rev. at 181. Not all jurisdictions have or maintain the data necessary to use the empirical approach. In combining aspects of the two approaches, for example, the case method with the statistical method, the Master has taken advantage of the available data to sort out the cases on the basis of the characteristics that both prosecutors in the charging process and juries in the deliberative process deem most relevant.
Without minimizing the effort required to collate the data in that way, we may describe the process engaged in by the Master as a form of electronic-index-card sorting. Under the Master’s direction, our AOC took the information on the index cards and fed it into a computer using a series of codes that will allow the information to be retrieved from the computer.
B.
The Methods of Selecting a Comparison Group of Similar Cases from the Catalog
The next step in the process is to develop the methods of analysis, that is, the manner in which the Court must look at the electronic-index cards to find the similarities. The mathematical and statistical concepts that underlie the process are set forth in the Final Report and the Weisberg Report submitted by the Attorney General. Further background may be found in the Report of the National Center for State Courts or the works of Professor Baldus and his colleagues. See Baldus and Pulaski II, supra, 33 Stan.L.Rev. 1.
We need give only a broad outline of the concepts and have enough familiarity with them ourselves so that the product of *144the analysis will enable us to accomplish our function of reviewing similarities to determine whether there has been a meaningful distinction in the imposition of the death penalty.
As noted, there are two ways of approaching the data. Under the first method, using the index-card illustration (for example, the robbery-murders of police officers), the reviewing court will establish a priori (that is, before the fact) what it believes are the similar characteristics that influence the life/ death decision. Under the second method, the reviewer assembles all the life/death verdicts and attempts to extract from those the distinguishing characteristics that appear to present a pattern of life/death decisions. One familiar with the scientific method of analysis will recognize that a pattern of decision will develop. The Court will not rely on either method exclusively. Rather, we will use each as a check on the other in reviewing the proportionality of Robert Marshall’s sentence. As we have explained in Part II, supra at 124, 613 A.2d at 1066, we believe, as does the United States Supreme Court (albeit in the more limited context of Eighth Amendment proportionality, see Coker, supra, 433 U.S. 584, 97 S.Ct. 2861, 53 L.Ed.2d 982), that the charging decisions of prosecutors, as well as the sentencing decisions of juries, both representing society’s interest in punishing crime, will demonstrate when a death sentence is excessive and thus should be factored into any analysis of relative disproportionality.
Recognizing, for example, that rarely, if ever, will comparison crimes of offenders be on “all fours” (e.g., no two convenience-store-holdup murders are exactly the same), a broader search for similarity is required. One method of evaluating substantive disproportionality, which allows the Court to broaden its base of similar cases, is that of “blameworthiness,” mentioned by Justice O’Connor in her dissenting opinion in Enmund. 458 U.S. at 823-25, 102 S.Ct. at 3390-91, 73 L.Ed.2d at 1168-70. The concept of “blameworthiness” can serve in *145procedural disproportionality review as well. Building on that concept, the Master has sorted the index cards to measure similar cases in terms of objectively-verified measures of blameworthiness derived from comparing assumed and empiric characteristics with actual results in the implementation of New Jersey’s death penalty. A thumbnail sketch of that process is:
[[Image here]]
*1461. The Salient-Factors Measure
That measure defines “similar cases in terms of factual comparability.” Final Report, supra, at 80. It is based on both a priori or assumed considerations (e.g., that a murder by a defendant with a prior murder conviction will be more blameworthy than not), and by empirical considerations (e.g., the data collected showed that factor is statistically and practically important in explaining the decisions of prosecutors and juries). Take, for example, the convenience-store-robbery murder by a previously-convicted murderer. Although in each case the crime is the same, experience in the form of actions of prosecutors and juries confirms the intuitive judgment that the first-time offender is less blameworthy. By sorting out those factual patterns (typologies) of cases in terms of statutory aggravating and mitigating factors, see N.J.S.A. 2C:11-3c(4) & (5), as well as other non-statutory aggravating and mitigating circumstances that are “important conceptually or statistically,” Final Report, supra, at 84, the data cards are catalogued to rank cases in order of the predicted probability of receiving a death sentence. Those electronic-index cards arranged by weighted sorting yield a sample of “similar cases” in terms of factually-related blameworthiness.
2. Numerical Preponderance of Aggravating and Mitigating Factors
Aside from the type of case, e.g., a rape-murder or robbery-murder, intuitively we are convinced that a case with more aggravating factors and fewer mitigating factors would be more likely to result in the imposition of a death sentence. A convenience-store robbery murder committed by a repeat killer to escape detection is more likely to produce a death sentence. Hence, these raw numbers may be used as a measure of blameworthiness. The Final Report has confirmed that intuition. For instance, the results of the Master’s analysis reveal that “in cases with a single aggravating factor, the death-*147sentencing rate declines sharply in the presence of one or more mitigating factors (the average rate among those cases is .10 (5/50)).” Final Report, supra, at 90. Thus, the index of cards will yield samples of similar cases in terms of the numbers of factors. That is one method used by Pennsylvania. See Commonwealth v. Pirela, 510 Pa. 43, 507 A.2d 23, 32 (1986).
The problem with that analysis is that it assumes that prosecutors and juries weigh aggravating and mitigating factors equally. For example, the “grave risk of death to another” factor, c(4)(b), carries much less weight than the “cop-killer” factor, c(4)(h). Hence, the Master weighted the statutory factors to reflect what the data showed. Finally, he added non-statutory aggravating and mitigating factors, for example, victimization that falls short of the torture/depravity factor, c(4)(c), but which is characteristically found to influence prosecutors and juries. That produces the final method of sorting cases to determine the class of similar cases. )
3. Index-of-Outcomes Test
The Master refers to that as a “logistic multiple-regression” analysis. Final Report, supra, at 92. The index produced “reflects the differential weights placed by jurors on the different statutory and nonstatutory aggravating and mitigating circumstances.” Id. at 92-93. It adds to our a priori assumptions what experience has taught us — for example, that
there is a penalty-trial death-sentencing rate of only .12 in the grave risk [to others] (4b) cases versus a rate of .67 in the pecuniary gain killer (4d) and police-victim (4h) cases. The impacts of the individual mitigating factors also vary. The defendant’s age (5c) has the greatest mitigating effect, while in contrast, the (5b) factor, victim contribution to the homicide, may have an aggravating effect.
[Id. at 92.]
Testing the weighted statutory and non-statutory aggravating and mitigating factors present in both penalty-trial cases and in all clearly death-eligible cases yielded a scale of overall defendant culpability “as measured by the presence or absence in the cases of factors that appear to influence prosecutorial *148and jury decision-making.” Id. at 93. The cluster of cases ranked near the subject case on the index yields a sample of cases that are similar in culpability. The end-product of the electronic sorting is a series of cases not unlike the series of index cards that we started with in our analogy.
C.
The Criteria for Comparing the Cases for Disproportionality
The data having been assembled and the means having been established to sort them to identify the comparison group of “similar cases,” the final step in the process is to develop criteria for evaluating when a particular sentence is disproportionate in relation to the other “similar cases” identified.
Jurisdictions that conduct proportionality review use varied criteria to evaluate when a death sentence is disproportionate. Missouri classifies similar cases as those “in which both death and life imprisonment were submitted to the jury, and which have had been affirmed on appeal.” State v. Mercer, 618 S.W.2d 1, 11, cert. denied, 454 U.S. 933, 102 S.Ct. 432, 70 L.Ed.2d 240 (1981). Without elaboration, the Mercer court concluded that those cases supported affirmance of the death penalty in the subject case, stating “defendant’s sentence to death for the murder * * * is not excessive or disproportionate to the penalty imposed in similar cases considering the crime and the defendant.” Id. at 11.
Following prompting by Justice Exum in his dissenting opinion in State v. Pinch, 306 N.C. 1, 292 S.E.2d 203, 230, cert. denied, 459 U.S. 1056, 103 S.Ct. 474, 74 L.Ed.2d 622 (1982), the North Carolina Supreme Court articulated its method for comparing similar cases. That court uses all cases arising since the effective date of its capital punishment statute that have been “tried as capital cases and reviewed on direct appeal by this Court and in which the jury recommended death or life imprisonment or in which the trial court imposed life imprisonment *149after a jury’s failure to agree upon a sentencing recommendation within a reasonable period of time” as a pool for comparison purposes. State v. Williams, 308 N.C. 47, 301 S.E.2d 335, 355, cert. denied, 464 U.S. 865, 104 S.Ct. 202, 78 L.Ed.2d 177 (1983). North Carolina does not “propose to attempt to employ mathematical or statistical models involving multiple regression analysis or other scientific techniques, currently in vogue among social scientists, which have been described as having ‘the seductive appeal of science and mathematics.’” Ibid. (quoting Blake v. Zant, 513 F.Supp. 772, 827 (S.D.Ga.1981)). It believes that a reviewing court might tend to “disregard the experienced judgments of its own members in favor of the ‘scientific’ evidence resulting from quantitative analysis.” Id. at 356. Thus, the court concluded that it would rely on its own case reports in the “similar cases” forming the pool that would be used for comparison purposes. Ibid.
The methodology of comparison used in Williams consisted of an identification of those cases (without description), a recital of the facts of the case being reviewed (a bestial torture-rape of an one-hundred-year-old woman), and a conclusion that the murder was “so brutal and so utterly senseless as to lead us to conclude that the sentence of death imposed in this case is not disproportionate or excessive considering both the crime and the defendant.” Id. at 357.
As stated previously, Georgia’s universe of cases includes all capital-felony cases that are appealed. In order to determine proportionality, the Georgia court selects for comparison cases those that are factually similar to the subject case. See Godfrey v. State, 248 Ga. 616, 284 S.E.2d 422, 430 (1981). It lists those cases in an appendix to its proportionality review. Without reviewing the facts of the cases in its opinion, it noted that juries had returned seven death verdicts in “domestic murder” cases. In “multiple murder” cases (Godfrey killed his wife and mother-in-law) seven death verdicts were returned. In three of those seven cases, juries returned death sentences despite the *150fact that the defendants had good records and a history of psychiatric disorders. Ibid.
Maryland establishes a pool of similar cases for comparison purposes by examining those with similar facts. As a second factor, Maryland considers the aggravating and mitigating circumstances. Tichnell, supra, 468 A.2d at 20-22. It recites the facts of the comparison cases in its opinion. For instance, in Tichnell it reviewed five killings committed during a robbery that involved either a police-officer or non-police-officer victim. Although four life sentences had been imposed, the court found sufficient differences in the mitigating circumstances or the actors’ role in the crime to sustain Tichnell’s sentence despite the pattern of life sentences found in similar cases. Ibid.
In Delaware, the court looks to those cases in which a capital-sentencing proceeding has been conducted to select its pool of similar cases, Flamer v. State, 490 A.2d 104, 139 (Del.1984). From that group, the court draws cases with similar objective factors to determine the proportionality of the death sentence in the subject case. Dawson, supra, 581 A.2d at 1108. Delaware is quite explicit in its analyses. Although it recognizes that “definitive comparison of cases is almost impossible and necessarily touches upon the realm of speculation,” Flamer, supra, 490 A.2d at 144, it does search for “a pattern of death sentences,” e.g., multiple murders of helpless elderly, ibid., and other objective distinctions, e.g., perpetrators were surprised during burglary, or were juvenile “show-offs.” Ibid.
Finally, as noted, Pennsylvania uses the measure of aggravating factors in a case. See Pirela, supra, 507 A.2d at 32; Commonwealth v. Maxwell, 505 Pa. 152, 477 A.2d 1309, 1318, cert. denied, 469 U.S. 971, 105 S.Ct. 370, 83 L.Ed.2d 306 (1984). Pennsylvania also looks to the salient factors that may bear on a defendant’s character. Commonwealth v. Travaglia, 502 Pa. 474, 467 A.2d 288, 304 (1983), cert. denied, 467 U.S. 1256, 104 S.Ct. 3547, 82 L.Ed.2d 850 (1984).
*151Justice Utter, in his separate dissent in State v. Jeffries, 105 Wash.2d 398, 717 P.2d 722, cert. denied, 479 U.S. 922, 107 S.Ct. 328, 93 L.Ed.2d 301 (1986), suggested the use of a “balancing approach superimposed upon a ‘salient factors’ approach” to identify the pool of similar cases.2 Id. 717 P.2d 722 at 744. That approach involves two steps. The first step requires a court to select similar cases from the statutorily-defined pool by choosing “three or four of the most important factors of the subject case.” Ibid. The second step would “compute the frequency of death sentences within the pool of similar cases.” Ibid. That would assure that “no death sentence is affirmed unless in similar cases throughout the state the death penalty has been imposed generally and not ‘wantonly and freakishly imposed.’ ” Id. at 743 (citing Moore v. State, 233 Ga. 861, 213 S.E.2d 829, 832 (1975), cert. denied, 428 U.S. 910, 96 S.Ct. 3222, 49 L.Ed.2d 1218 (1976) (quoting Furman v. Georgia, supra, 408 U.S. at 310, 92 S.Ct. at 2763, 33 L.Ed.2d at 390 (Stewart, J., concurring))).
The National Center summarizes the various approaches as follows:
The first is to rely on “generalized notions of reasonableness,” based on the court’s “own values, experience, and general familiarity with prior cases.” * * * The second is the “precedent-seeking approach.” Using this approach, the court: (a) identifies the relevant aggravating and mitigating factors; (b) makes a judgment regarding the proportionality or excessiveness of the sentence based upon those factors; and (c) identifies one or more comparable cases that support its decision. The third is the “frequency approach.” This method involves: (a) specifying which features of the review case should be used to find comparable cases; (b) identifying the other cases that share the selected characteristics; (c) determining the percentage of defendants in the similar cases who were sentenced to die; and (d) deciding whether death sentences *152were imposed with sufficient frequency within this “class of similar cases [so] as to * '* * serve as an effective deterrent * * * or to constitute a justifiable form of retribution in light of contemporary community standards.”
[National Center for State Courts, Proportionality Review Project 2-3 (1984) (footnotes omitted) (quoting Baldus and Pulaski I, supra, 74 J.Crim.L. & Criminology at 668-69).]
We discuss the frequency method first.
1. Frequency Approach
As noted, Justice Utter believed that the frequency analysis is the most acceptable approach to proportionality review. Use of the word “generally” suggested to him that the threshold frequency at which a death sentence becomes appropriate is “significantly greater than 50 percent.” Jeffries, supra, 717 P.2d at 744. He viewed that approach as simple, maintaining a reasonable amount of objectivity “by defining a genuine ‘threshold frequency,’ ” and identifying the “specific factors by which to select ‘similar cases.’ ” Ibid. Nonetheless, he had to concede that to employ that test “in some cases where a limited selection of salient factors will make the case unique” is not feasible. Ibid. In such situations, the court, in his view, probably would have “to turn to a more subjective comparison of the ‘severity’ of dissimilar cases.” Id. at 745.
The Public Defender, we believe misperceiving the Gregg reference to whether death sentences have been imposed “generally in similar cases,” has suggested that we are bound to disapprove a sentence of death whenever there is a predictable frequency rate of less than eighty percent. As we noted in Part II, these concepts of “generally imposed” are not transferrable from the Eighth Amendment context. In Ramseur, supra, we predicted the difficulty of reconciling the concepts of predictability of sentencing with individualized sentencing. 106 N.J. at 330-31, 524 A.2d 188. We believe that in the individualized-sentencing process mandated by Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978), and by our own *153Capital Punishment Act, that death be neither normal nor general in capital sentencing is inevitable.
If we are correct that death need not be normal or general to be a licit sentence, the frequency approach will not provide a sole-source, fail-safe method of proportionality review. For even were we to establish a threshold level of acceptability (let us say the fifty percent suggested by Justice Utter), to ignore striking disproportion between death-sentenced cases and those in the life-sentenced pool could be unfair to those in the death-sentenced pool. Given the inevitable role of mercy in the death-sentencing process, there will be those whose life sentences will not yield to statistical analysis.
Although the Master endorses the frequency approach, he has not suggested that we substitute a numerical process- for a reasoned process of decision.
[The Master’s] endorsement of the frequency approach does not carry with it, however, a recommendation that the Court quantify mathematically its judgments of the death-sentencing frequency among similar cases. Several courts have expressed concern that the application of a strictly quantitative approach to the subject could lead to arbitrary line drawing and limit the legitimate exercise of judicial discretion. More importantly, such an approach may inappropriately suggest that the complex judgments involved in proportionality determinations can be expressed with mathematical precision.
[Final Report, supra, at 42-43.]
Rather than employ the frequency method as a cutoff, we believe that it will serve as a coefficient of consistency. The higher the frequency of a death sentence among the comparison group of “similar cases,” the more certain the determination that the sentence is proportionate. The lower the frequency, the more strictly the Court must scrutinize the case for the possible influence of impermissible factors.
We believe that the frequency approach will help us to review cases in terms of the substantive principle that we believe should be controlling in these cases, namely, “[a] death sentence is comparatively excessive if other defendants with similar characteristics generally receive sentences other than death *154for committing factually similar offenses in the same jurisdiction.” Tichnell, supra, 468 A.2d at 17 n. 18. We need not go so far as to suggest that the ratios tip very strongly in the opposite direction to justify the invalidation of a sentence on the basis of the frequency approach. Within its limits, the scientific method is an aid to the Court.
A frequency approach with a broad universe can cull a group of comparable cases from a larger group of death-eligible cases. In this way, the proportion of life and death sentences within this similar group can be estimated and an empirical assessment of comparative excessiveness can be made.
[Raymond Paternoster & AnnMarie Kazyaka, An Examination of Comparatively Excessive Death Sentences in South Carolina 1979-1987, 17 N.Y.U.Rev.L. & Soc.Change 475, 486 (1989-1990).]
The greater the statistical frequency of life sentencing in the comparison group of similar cases, the greater will be the need for the Court to focus on the “real people” involved in the defendant’s and other similar cases. We have a data base system that can define similar cases in three ways: salient factors, the numerical preponderance of aggravating and mitigating factors, and the index of outcomes. Examining each of those pools in terms of relative frequency will help us to identify comparable cases and to determine whether offenders like Marshall have received, or usually will receive, a life sentence. If so, then those comparable pools will allow us to consider what characteristics in Marshall’s case might sustain the imposition of a death sentence.
2.
The Precedent-Seeking Approach of Comparative Proportionality Review
The final step, then, is the familiar judicial process of case-by-case comparison of life-sentenced and death-sentenced similar cases. (Life-sentenced cases include, under some systems of analyses, plea bargains and other non-capital disposi*155tions.) This is the second of the methods outlined in the National Center for State Courts’ Report. As noted, we see it operating in functional relationship to the frequency approach.
To return to our analogy, having taken the cards containing “similar cases” from the file, we must determine what information on the cards we should consider in evaluating whether the sentence is disproportionate. The Master suggests a comparison of similar cases according to the defendant’s criminal culpability. His model uses non-statutory case characteristics defined in terms of three basic elements: (1) a defendant’s moral blameworthiness, (2) the degree of victimization, and (3) the character of the defendant. A sketch of the Master’s precedent-seeking model is:
[[Image here]]
The Public Defender believes that the Court should not undertake a “subjective moralistic judgment[] based on non-statutory factors.” Rather, he argues in favor of what he describes as an “objective analysis focusing on the facts underlying the statutory aggravating and mitigating factors.” Recall that in sorting out the cases, the Master had taken note of *156and recorded on the index cards fed into the computer both statutory and non-statutory factors that appeared to reflect the deathworthiness judgments of juries.
We agree that courts are ill-fitted to make moralistic judgments about who should live or who should die under a capital-sentencing scheme. We have no intention of translating concepts that fall outside the Code of Criminal Justice into the capital-sentencing structure. However, we believe that we can examine the data on the cards to see whether there is evidence of objective factors (beyond the listed c(4) statutory aggravating factors) that will assist us in determining why a jury has regarded one defendant as more deathworthy than another.
We understand the basic premise of the counter-argument. The Legislature has given us a list of factors (the statutory aggravating factors that made the murder death eligible, for example, c(4)(a) (a prior murder) or c(4)(c) (aggravated assault/torture)) that it deems relevant to the death-sentencing process. See Trimble v. State, 300 Md. 387, 478 A.2d 1143, 1167 (1984) (“We see no bright line by which this Court can say when death shall be imposed. Moreover, we believe the guidelines established by the legislature represent the clearest course of action in attempting to resolve this problem.”), cert. denied, 469 U.S. 1230, 105 S.Ct. 1231, 84 L.Ed.2d 368 (1985). If factors beyond the statutory factors are given weight, the Capital Punishment Act will lose all structure and revert to the wholly-random pre-Gregg process under which sentences were imposed randomly and freakishly.
We think that argument misperceives the function of statutory aggravating factors. Those factors perform the function of “categorical narrowing,” Zant v. Stephens, 462 U.S. 862, 879, 103 S.Ct. 2733, 2744, 77 L.Ed.2d 235, 251 (1983), to channel the discretion of juries. Once channeled, deathworthiness requires that an “individualized assessment of the defendant” be undertaken. Ramseur, supra, 106 N.J. at 331, 524 A.2d 188 (citing *157Lockett v. Ohio, supra, 438 U.S. at 605, 98 S.Ct. at 2965, 57 L.Ed.2d at 990).
If we did not consider the circumstances of the case beyond the c(4) aggravating factors in a search for disproportionality, we would be ignoring the reality of the situation and perhaps disadvantaging defendants. For example, if we had a defendant with a previously-blameless life who had impulsively killed a police officer with a single shot, we would be unable to say that his crime was less deathworthy than that of one with a similar past life who had premeditated a taunting, execution-style revenge killing of the officer. None of those is a statutory aggravating factor, yet it seems that they inevitably reflect on deathworthiness.
Other examples of factors that would not make a murderer death eligible but that help to explain the jury judgments are the type of victimization that might not constitute a statutory aggravating factor — for example, the particularly vulnerable victims found in State v. Gerald, 113 N.J. 40, 549 A.2d 792 (1988) , and State v. Zola, 112 N.J. 384, 548 A.2d 1022 (1988), cert. denied, 489 U.S. 1022, 109 S.Ct. 1146, 103 L.Ed.2d 205 (1989) . In noting that such factors bear on relative deathworthiness, we impart no judge-made concepts of ethics or morality. We often have to give definition to the Capital Punishment Act. See Ramseur, supra, 106 N.J. at 210, 524 A.2d 188 (defining the c(4)(c) depravity factor by the absence of recognized human emotions).
Another example is the extent of mutilation. That can fall outside of the c(4)(e) factor because to prove that a defendant had an intent to inflict any more pain than that necessary to kill may be difficult. See State v. Harvey, 121 N.J. 407, 413-14, 581 A.2d 483 (1990), cert. denied, — U.S.-, 111 S.Ct. 1336, 113 L.Ed.2d 268 (1991). Those are not subjective moralistic considerations. Many of the so-called “non-statutory” aggravating factors in the “card index” in fact appear to be recognized sentencing factors. For example, N.J.S.A. 2C:44-1a(2), which *158prescribes the authority of a court in sentencing, specifically requires courts to consider the
gravity and seriousness of harm inflicted on the victim, including whether or not the defendant knew or reasonably should have known that the victim of the offense was particularly vulnerable or incapable of resistance due to advanced age, ill-health, or extreme youth, or was for any other reason substantially incapable of exercising normal physical or mental power of resistance.
[Ibid.]
Surely, the other so-called non-statutory aggravating factors present in the data-collection system — such as the scene of the crime, that is, whether the crime involved an intrusion in the victim’s home or a kidnapping, or whether the killing was with an axe, was a particularly brutal stomping or beating, or was a planned homicide — do not appear to us to present a subjective, value-laden approach. Yet, although we agree that one who protests one’s innocence should not be faulted for a lack of remorse, the factor need not be discarded because its presence may be appropriate to consider in certain cases. Hence, we think that a proportionality comparison that limits itself to the presence of statutory aggravating and mitigating factors fails fully to explain the sense of proportion in the jury verdicts. The statutory aggravating factors do not encompass all of the characteristics that affect the blameworthiness or deathworthiness of persons who commit murders.
At least the Public Defender appears to recognize the use of the factors mentioned by Justice Utter in his dissenting opinion in Jeffries, supra, namely, “(1) the number of victims; (2) the conscious amount of suffering imposed on the victim; (3) the degree of premeditation; (4) the aggravating circumstances found; and (5) the personal background of the accused.” 717 P.2d at 745. Although those criteria are helpful in any attempt to explain a jury’s verdict, they do not exhaust the logical objective criteria by which the deathworthiness of cases may be compared.
*159Hence, in conducting a precedent-seeking comparative-culpability review we shall identify the relevant aggravating and mitigating factors in the comparison cases (including the so-called non-statutory factors). Examples of such an exercise in precedent-seeking are found in the dissenting opinions of Judge Davidson in Tichnell, supra, 468 A.2d at 27, and Justice Utter in Jeffries, supra, 717 P.2d at 742. We shall attempt to confine ourselves to objective criteria rooted in traditional sentencing guidelines. See N.J.S.A. 2C:44-1. By reference to those specifics in the comparison cases, we shall suggest which factors might reasonably explain the difference between a death-sentenced and a life-sentenced case and why the cases may be “similar.” As noted, supra at 153, 613 A.2d at 1081, the higher the frequency of life sentences in the pool of similar cases, the more searching will be the inquiry to test whether comparison with the life-sentenced cases (or more culpable death-sentenced cases) suggests that Marshall’s sentence was disproportionate in the sense of his having been singled out unfairly for capital punishment.
D.
Should we Consider an Alternative Method of Analysis by Categories of Culpability Derived Solely from the Statutory Aggravating Factors?
Before applying the system we must evaluate the alternatives that have been suggested to us. The State advocates matching cases by selecting those with the same configurations of statutory aggravating factors. The State points out that the advantage of such an approach is that it is objective and simpler. Relying solely on aggravating factors would, in the State’s expert’s view, give the Court a consistent definition of similar cases.
In support of that position, the State relies on a report, prepared and submitted by its consultant, Herbert I. Weisberg, Ph.D., entitled Proportionality Review of Death Sentences in *160New Jersey: An Independent Analysis of Data on Capital Charging and Sentencing (Nov. 26, 1991) (the Weisberg Report ). At the request of the New Jersey Attorney General’s Office, Dr. Weisberg conducted an independent analysis of available data on the administration of the death penalty. Although he pointed out a list of apparent errors discovered by staff of the Attorney General, his report relied primarily on the AOC data base, supplemented by limited information provided by the Attorney General.
1.
Dr. Weisberg began his analysis by questioning what he regarded as an inconsistent use of data by the Master, derived from cases decided under legal interpretations that were later held to be incorrect. In some instances, he asserts, the Master found such sentences to be a “valuable source” of information; in others (including the geographic analysis) the Master excluded cases that are not death eligible under current law. Weisberg Report, supra, at 9-10. In addition, Dr. Weisberg questioned the Master’s method of determining death eligibility by a set of aggravating and mitigating factors that are objectively present in a case, while limiting his evaluation in penalty trials to factors actually found by the penalty-phase decisionmaker.
Dr. Weisberg questioned further the Master’s treatment of the jury decision as a simple dichotomy — death versus life. He believed that the analysis failed to distinguish between a unanimous life-sentence verdict and a hung jury. In Dr. Weisberg’s view, a deadlock falls somewhere between a unanimous life sentence and a unanimous death sentence as an indicator of culpability. He attempted, then, to analyze jury deadlocks on the weighing of aggravating and mitigating factors. Id. at 11.
Hence, he attempted to “separate cleanly” the prosecutorial component from the jury component and to identify the relevant data and cases for each. For prosecutorial-charging decisions, he would examine only whether a notice of factors had been served, regardless of any subsequent guilt-trial decision. *161Finally, he challenged what he called the “highly elaborate statistical manipulations involved in the ‘logistic regression’ models.” Id. at 13. In his view, the models included “far too many potentially predictive independent variables” that might distort the association, and errors associated with predictions would thus increase. Ibid.
In Dr. Weisberg’s view, to accept and work within what he calls “existing realities” is preferable. He does not regard the goal of estimating future death-sentencing frequencies as feasible. He envisions a more modest role for quantitative analysis of past decisions. In his view, “Statistical analysis is potentially capable of providing the Court with valuable insight into the main factors that influence prosecutors and juries. Such general information can serve as a useful backdrop against which particular circumstances of a case can be arrayed and evaluated.” Id. at 15.
2.
We shall try to illustrate Dr. Weisberg’s method in terms of our concept of the electronic-index cards. As we understand it, Dr. Weisberg recommends the use of a “universe” of penalty trials in which at least one aggravating factor was found. He examined the outcomes for each possible combination of aggravating factors found and also for every combination of factors served by the prosecutor. He focused primarily on the aggravating factors found at the penalty trial. Ibid.
In other words, Dr. Weisberg sorted the electronic-index cards on the basis of the aggravating factors actually found at penalty trials. He found twenty-one distinct combinations of factors appearing among the 108 cases in which at least one factor was found. Ibid. He attempted to group cases on the basis of the proportion of death sentences and deadlocks. Dr. Weisberg reported that cases involving factors c(4)(a) (prior murder), c(4)(d) (hired gun), and c(4)(h) (cop-killer) always resulted in at least a deadlock, as did cases involving a combina*162tion of c(4)(c) (torture/depravity), c(4)(f) (escaping detection), and c(4)(g) (in commission of felony). Ibid.
Conversely, cases involving only one or more factors, c(4)(b) (grave risk to another), c(4)(f) (escaping detection), and c(4)(g) (in commission of felony) received mostly life-sentence verdicts, with the exception of those involving both factors c(4)(f) and c(4)(g). He reported further that there was only one death sentence among those cases. Ibid. His index-card sorting indicated to him that cases fall into a rough hierarchy in terms of deathworthiness. Here we use his style of referring to the factors in capital letters.
Category I: Any of factors A, D, E, H or CFG combination
Category II: None of above, but CF or CG combinations
Category III: None of above, but C or FG
Category IV: None of above, but (B, F, G, BF, or BG)
Dr. Weisberg declined to incorporate analysis of mitigating factors because of the multiplicity of combinations and the uncertainty of the relationship with outcomes. He concluded that further sorting was made complex by the fact that mitigating factors need not be found unanimously, and especially without data on the number of jurors who actually voted for the factor. Id. at 16.
Dr. Weisberg thus sorted the cases into four culpability categories. Category I, the highest culpability category, included those cases “always resulting in at least a deadlock.” He placed every configuration of aggravating factors3 with a deadlock or death-sentencing frequency of 1.0 in Category I (i.e., any of the factors A, D, E, H, and CFG combinations). Id. *163at 15-16. Weisberg made two exceptions. Finding that the three E cases (contract killer) were “similar” to D cases already in Category I, Weisberg added those cases to Category I. Id. at 15. Weisberg also placed the CF combination in Category II.
Weisberg next sorted the cases into Category IV, the lowest culpability category. Category IV cases included those aggravating-factor combinations with a zero death-sentencing frequency. Weisberg found that cases involving only one or more of factors B, F, and G received mostly life sentences, with the exception of FG combinations. Id. at 15-16.
Weisberg sorted the remaining cases at the fifty-percent mark. That is, when the majority of cases with certain combinations of aggravating factors resulted in death, Dr. Weisberg classified them as Category II cases. If the majority of cases received life sentences, Weisberg placed them in Category III. Id. at 15.
The results of that three-outcome, four-level approach are presented in Table 9 of the Weisberg Report, reprinted below.
[[Image here]]
If the Court were to adopt a system of case categorization similar to that developed in Dr. Weisberg’s research, then all penalty-trial cases in the same category would form a class of roughly “similar” cases.
The proportion of cases in this category that resulted in a death sentence would provide useful insight, as would also the proportion of hung juries. However, *164this quantitative data would represent only a starting point. Detailed qualitative comparisons among the cases within the category would be necessary to refine the Court’s culpability assessment.
The Special Master’s vision of a more mechanized approach to proportionality review is seductive. However, the attempt to push statistical analysis beyond a more limited role may simply open the door to methodological controversy that will ultimately frustrate and confuse the Court.
[Weisberg Report, supra, at 38-39.]
The defense characterizes Weisberg’s approach , as inconsistent and incomplete. In general, the defense notes that because the model is based solely on aggravating factors, the character of the defendant is ignored. In addition, Weisberg’s approach does not remedy the alleged flaws in Baldus’s study, specifically, the small sample size, which reduces the confidence one has in the conclusions and the factual dissimilarity between the cases in the culpability categories.
The Public Defender makes several specific challenges to Weisberg’s approach. He charges that Weisberg does not treat deadlock cases consistently. In creating Category I, Weisberg equates a deadlock with a death sentence; however, he equates the deadlock with life sentences in forming the other culpability categories. Defendant claims that treating deadlock cases the same as a death sentence is insupportable. Important to proportionality review is the outcome of the jury’s decision. Deadlocked cases cannot be properly characterized as reflecting deathworthiness judgments.
The Public Defender raises a variety of other concerns related to the model’s statistical consistency. Most importantly, that after establishing numerical rules for Category I, which place factor c(4)(d) (killed for money) cases in Category I, Weisberg deems factor c(4)(e) (hired a killer) cases similar, because both involve contract killings. The Public Defender claims that that manipulation was designed to place Marshall, the sole E case to receive a death sentence, in the highest culpability category. In fact, the three E cases have a frequency rating of .33, which would place Marshall in culpability *165Category III, not I. Dr. Weisberg was candid to recognize that he felt “least certain” about the assignment of Marshall’s E case to Category I because of the “paucity of available data.” Weisberg Report, supra, at 15.
Obviously, any analytical method can be found to be subject to improvements. The exclusion of mitigating factors from the State’s approach is especially problematic for us. Merely using the statutory aggravating factors skews the process because New Jersey’s aggravating factors focus primarily on the crime. Only the c(4)(a) factor (prior murder conviction) relates to characteristics of the defendant. N.J.S.A. 2C:11-3e instructs the Court to assess similarity in terms of “the crime and the defendant.” Accordingly, the mitigating factors must be used because they focus on the defendant’s characteristics. Ramseur, supra, 106 N.J. at 330, 524 A.2d 188 (emphasizing that both components are important in identifying similar cases).
The issue of treatment of mitigating factors arose during the preparation of the Final Report. Mitigating factors need not be found unanimously. State v. Bey, 112 N.J. 123, 159-61, 548 A.2d 887 (1988) (Bey II). If a mitigating factor received one or more votes, the Master coded the factor as being present in that case. Final Report, supra, at 13 n. 10. The State argues that statutory mitigating factors should be considered, but the actual vote should be recorded. The State’s position reflects the concern that a case in which only two jurors found the c(5)(a) factor (emotional or mental disturbance) should be weighted differently from a case receiving ten votes on that factor. Because the number of jurors voting for each mitigating factor is not always available, Baldus’s statistics indicate the average effect of the particular mitigating factor having been found or not found in the case. Ibid.
Although the lack of complete information on mitigating factors is a concern, the State’s approach creates a greater concern because it does not use any mitigating factors in defining a comparison group of cases. The greater overall *166measure of completeness in the Master’s data base better serves the Court’s need to assess disproportionality in terms of “the crime and the defendant.”
V
Application of the Methods of Proportionality Review to Robert O. Marshall
Having developed both the means and methodology for analyzing whether a subject case is disproportionate to the sentences imposed in “similar cases,” we undertake now the proportionality review of Robert O. Marshall’s sentence. We have not included the names of all cases used in the analysis. Those cases have been included in the Master’s reports. The cases that we have selected under the “Precedent-Seeking” approach will all be identified in this opinion.
A.
The Frequency Analysis
At the most transparent level, there appears to be an appalling disproportionality in sentencing Robert O. Marshall to death. Alone among the 246 death-eligible cases, his conviction of capital murder stands affirmed.4 Sooner or later, however, that day had to come. But simply because Marshall may be the first does not mean that his death will be disproportionate under our statute. There may be more paradox than disproportion in his sentence. For no matter how upstanding or prominent his former life may have been, the data show that among those for whom death is a fitting punishment, contract killers, *167whether principal or agent, are among the more frequent recipients of the death sentence.
As noted, we do not believe that an eighty-percent predictability rate is a prerequisite to a sentence of death. Nor do we believe that a fifty-fifty chance of life or death (as exemplified in this case by the stark contrast between Robert Marshall’s death sentence and William Engel’s life sentence, which is discussed infra at 179, 613 A.2d at 1095), is a necessary prerequisite. We do find, however, evidence of the reliability of Robert Marshall’s sentence in the frequency analyses performed under the three evaluative methods for assessing criminal culpability: the salient factors, the numerical preponderance of aggravating and mitigating factors, and the index of outcomes.
Although acknowledging “the difficulty of estimating reliable death-sentencing frequencies on the basis of a small sample of similar cases,” David C. Baldus, State v. Robert Marshall, Death Penalty Proportionality Review Project: A Report to the New Jersey Supreme Court 1, 10 (Sept. 24, 1991) (hereinafter Marshall Report), the Master nevertheless concluded that the frequency data are relevant and can be validated “through close comparisons of the cases involved.” Id. at 16. That procedure will “provide a valuable supplement to the court’s prior experience and knowledge.” Id. at 16-17.
There is a preliminary issue of whether Marshall’s case should be considered in the count of death cases among similar cases for frequency-analysis purposes. Because the purpose of frequency calculations is to estimate death-sentencing rates among cases that are similar to the defendant’s, it would seem that the subject case should not be included. On the other hand, because the case before the Court is a partial reflection of community values, even if aberrational, it should be considered. Because we do not attach a conclusive life/death determinant to a statistical analysis, we shall consider the frequency data under both theories. Finally, although we have chosen all *168death-eligible cases as our universe, we furnish the penalty-trial statistics for informational purposes.
1. The Salient-Factors Measure
Because this “facts-of-the-case” type analysis has a presumptive objectivity (the criteria for inclusion are almost intuitive), it seems most persuasive here. The “typologies” of the data-gathering process were not an ad hoc construct for the Marshall case. These were, and are, the a priori, that is, not after-the-fact assumptions as to which cases are similar. Contract killings are the type of killings that prosecutors and juries have regarded as highly blameworthy. Thirty percent of contract murderers (principals and hitmen) have initially received a death sentence. That figure is significant considering only thirty-two cases of all death-eligible cases have received the death penalty.
Contract-Murder Principals
The most factually-comparable cases to advance to a penalty trial, the c(4)(e) contract-murder principals (W. Engel, H. Engel, and Brand) all received life sentences. Tabulations are as follows:
Penalty-Trial Eligible
Universe Universe
Including Marshall .33 (1/3) .25 (1/4)
Excluding Marshall .00 (0/2) .00 (0/3)
Contract-Murder Hitmen
N.J.S.A. 2C:11-3c(4)(d) makes one who “commit[s] the murder as consideration for the receipt, or in expectation of the receipt of anything of pecuniary value” death eligible. In State v. DiFrisco, 118 N.J. 253, 571 A.2d 914 (1990), we explored the question of whether society regards as more reprehensible the principal or the agent in a contract killing. Because of the *169close relationship between c(4)(d) and c(4)(e), one who procures the commission of the offense “by payment or promise of payment of anything of pecuniary value,” the Master includes hitman cases as factually comparable, and we agree.
This contract-murder hitmen pool includes six cases.5 The death-sentencing rate among the penalty-trial cases is .50 (2/4) and .33 (2/6) for all death-eligible hitman cases.
Thus, the figures for the entire contract-murder pool are as follows:
Penalty-Trial Universe Eligible Universe
Including Marshall .43 (3/7) .30(3/10)
Excluding Marshall .33 (2/6) .22(2/9)
Among the five triggerman c(4)(d) cases in the proposed universe, four advanced to a penalty trial with the c(4)(d) factor charged: Melendez, Michael Rose, Clausell (death sentenced), and DiFrisco. Among those four cases the jury found the c(4)(d) factor present in three cases and imposed a death sentence in two. The Court vacated Clausell’s death sentence on appeal and he was later sentenced on the non-capital murder charge. Clausell, supra, 121 N.J. at 307, 580 A.2d 221. We *170vacated DiFrisco’s sentence and he is awaiting disposition on the remand. DiFrisco, supra, 118 N.J. at 283, 571 A.2d 914. The Marshall Report concludes that “[t]his very high death-sentencing rate, albeit in a small sample of cases, produced in our principal penalty-trial analysis a large and statistically significant multiple regression coefficient for the 4d [hired killer] factor.” Marshall Report, supra, at 23.
Spousal Murders Involving High Levels of Blameworthiness and a Defenseless Victim.
The Master has also analyzed the frequency of spousal murders that involved a high level of blameworthiness and a defenseless victim. The Marshall Report compares Marshall with three cases involving highly-premeditated, cold-blooded murders of a defenseless wife. Those comparison cases are Collins, Dreher, and Williams. Only Collins and Williams advanced to a penalty trial at which both defendants received a life sentence. Thus, all three defendants received life sentences. The figures are as follows:
Penalty-Trial Eligible
Universe Universe
Including Marshall .33 (1/3) .25 (1/4)
Excluding Marshall .00 (0/2) .00 (0/3)
Premeditated Robbery/Kidnap Murder Cases Involving Extensive Premeditation, a Pecuniary Motive, Deception/Entrapment of the Victim, and a Defenseless Victim
The Master also compared this fourth group of cases with Marshall. Of the five cases considered, three advanced to a penalty trial and one, Martini, resulted in a death sentence. In terms of criminal culpability, the Master ranked Martini higher than Marshall, finding that Martini’s case involved two aggravating circumstances, extreme blameworthiness, substantial victimization, and a defendant with a poor character.
*171A review of all of the categories of factually comparable cases shows a death-sentencing frequency of .22 (4/18) when Marshall is included. Excluding Marshall, the death-sentencing rate is .18 (3/17).
2. The Numerical Preponderance of Aggravating and Mitigating Factors.
This test is more problematic than either the salient-factors or the index-of-outcomes analyses. Abstractly, we sense that a quantitative rather than a qualitative analysis will be rather unproductive. No matter how many aggravating factors and how few mitigating factors are presented, the jury’s decision is intensely qualitative. We realize that the Master’s data do not reflect a numerical analysis but rather reflect the weight that prosecutors and juries give to a specific factor; for example, a rape-murder will be viewed as more blameworthy than a robbery-murder. Still, a degree of abstraction exists that may not be present in the other frequency-analysis measures. We have repeatedly emphasized that juries are to make a qualitative, and not a quantitative, analysis of aggravating and mitigating factors.
Hence, when we measure Robert Marshall’s case against other cases with one aggravating factor and two mitigating factors, we are concerned about the infrequency.
Cases with One Aggravating Factor and Two Mitigating Factors
Penalty-Trial Eligible
Universe Universe
Including Marshall .20 (3/15) .07 (3/44)
Excluding Marshall .14 (2/14) .05 (2/43)
Cases With One Aggravating Factor
Penalty-Trial Eligible
Universe Universe
Including Marshall .10 (5/50) .04 (5/123)
Excluding Marshall .08 (4/49) .03 (4/122)
*172As shown by the data, the overall death-sentencing frequency among capital cases involving a single aggravating factor is .08 (4/49). Among penalty-trial cases that, like Marshall, include one aggravating factor and two mitigating factors, the rate is .14 (2/14). However, death-sentencing frequencies among cases involving the c(4)(a) (prior murder) factor, c(4)(d) (hired gun) factor, c(4)(e) (payment for murder) factor, and c(4)(h) (cop-killer) factor have “above-average death-sentencing rates.” Marshall Report, supra, at 31 and Appendix H, Table 9 to the Final Report.
3. Index-of-Outcomes Test
This method of analysis also indicates no significant disproportionality in Robert Marshall’s sentence.
Recall that this measure looks not only at the “salient factors,” or the “number of factors,” but at all of the facts and all of the cases to determine a pattern of capital sentencing. In this process we have spread all of the index cards on the table and have sought to identify the characteristics common to the cases in terms of their degree of blameworthiness as perceived by prosecutors and juries. These cases may be as factually dissimilar as a killing by a child molester or killing by a bank robber. All that they have in common is a roughly-equivalent measure of blameworthiness in the eyes of prosecutors and juries.
Even under this largely empirical standard (i.e., derived from the actual experience within the system), the measure of Robert Marshall’s blameworthiness is significant. His support of family, community involvement, and prior good record do not counterbalance the other measures of blameworthiness found in similarly-blameworthy case records involving premeditation of the act, helplessness of the victim, and baseness of the motive.
*173Under the index-of-outcomes analysis, using the expanded indices that include non-statutory factors, the predicted probability of a death sentence in Robert Marshall’s case is .50 among all penalty-trial cases. The Master cautions, however, that that estimate is quite unstable because of the small number of c(4)(e) penalty-trial cases. The predicted probability of death for Marshall’s case among all death-eligible cases is .17. Here again, the Master cautions that that estimate is uncertain because of the small sample number.
After reranking the cases to reach a qualitative culpability level, based on blameworthiness, victimization, and character of the defendant,6 the overall death-sentencing rate for the ten cases most comparable to Marshall’s in the penalty-trial model is .60 (6/10). Including Marshall, the rate is .64 (7/11). Among all death-eligible cases the overall death-sentencing rate among the ten cases nearest Marshall’s is .20 (2/10), excluding Marshall. With Marshall, the death-sentencing rate is .27 (3/11).
When the indices are limited to statutory aggravating and mitigating circumstances, the predicted likelihood of a death sentence in the penalty-trial model for Marshall is .52 and among all death-eligible cases the probability of death is .27. Reranking the cases as above, among the cases nearest Marshall’s in the penalty-trial model the death-sentencing rate is .69 (11/16), excluding Marshall. With Marshall, the rate is .71 (12/17). Among all death-eligible cases the death-sentencing rate among the ten cases nearest Marshall’s is .50 (11/22), excluding Marshall. With Marshall, the rate is .52 (12/23).
Summarizing all of the data, the Master concludes:
[B]ecause of the small number of 4e eases, and no 4e cases that match Marshall on both the blameworthiness and victimization dimensions, we have a much less solid basis for saying that cases like his either will or will not be associated with frequent death sentencing over the long run. Because of this small sample *174problem, there is no way to resolve with confidence the uncertainty associated with predicting the future for defendants like Marshall.
[Marshall Report, supra, at 41.]
On balance, by using the composite of these measures of frequency we conclude that capital death sentencing for contract murderers is not random or aberrational. See Tyler v. State, 247 Ga. 119, 274 S.E.2d 549, 555 (1981) (reasoning that “[although lesser sentences than death are frequently imposed in domestic murder cases, it does not follow that the death penalty would not be authorized for the murder of one spouse by another under any circumstances”).
B.
Precedent-Seeking Approach
We turn now to the precedent-seeking comparison of defendant’s death sentence with those imposed “in similar cases, considering both the crime and the defendant.” N.J.S.A. 2C:11-3e.
In his well-reasoned brief, the Public Defender has challenged the Master’s conclusion that “in terms of moral blameworthiness Marshall exceeds that found in the three other 4e cases, and as far as we can determine, all other cases in our universe of New Jersey cases.” Marshall Report, supra, at 12. In particular, the Public Defender questions the Master’s reliance on what he perceives to be a subjective evaluation that because adultery is so morally blameworthy, Marshall’s crime is viewed as more reprehensible than, for example, the killing of a wife and a child solely for insurance proceeds when another woman is not involved. He challenges, as well, the Master’s reliance on Robert Marshall’s double life — “plotting [his wife’s] death and sleeping with his paramour while living and sleeping with his wife and pretending that all was well” as giving “an additional aggravating edge to the case.” Id. at 13-14 & n. 13. In addition, he questions the reliance on a “total *175lack of remorse for his crime,” id. at 14, as inconsistent with defendant’s right to assert his innocence.
We share some of the Public Defender’s concerns. In this section we evaluate Marshall’s case on the basis of objective characteristics by comparing cases that appear to be found within familiar sentencing patterns. For purposes of analysis, we proceed along the lines set forth in the Public Defender’s brief and identify the cases and characteristics that bear on Marshall’s relative blameworthiness. We use primarily the Public Defender’s factual description of the cases; some of those are more fully set forth in the appendices to the Final Report. At least one of the cases is on appeal before us so we give no stamp of approval to those factual versions but give to defendant the benefit of his view of the facts.
1. First-time murderers of spouses.
The Public Defender asserts that among comparable first-time offenders who murdered their wives and who cannot plead excuse for lack of education and intelligence, only Robert Marshall received a death sentence. We consider each of those cases.
Walter Williams
Williams, a twenty-seven-year-old police officer, murdered his wife by poisoning her with potassium cyanide. Williams allegedly had a two-fold motive for killing his wife: to hide his bigamous marriage to another woman and to obtain the proceeds of his wife’s estate.
Williams’s web of deceit began when he met his second wife, then a high-school junior, while on duty at a high-school function in 1979. Their relationship intensified and, in November 1984, Williams and she married. Williams deceived both his second wife and the minister who performed the ceremony by providing a falsified judgment of divorce, complaint of divorce, and a birth certificate.
*176He lied to his wife to whom he was legally married by telling her that he had to be treated for Agent Orange in the evenings at the Veterans’ Hospital. During that time Williams had been living with his “new wife” in her parents’ home. Williams’s wife became suspicious of his bigamous relationship and confronted him on different occasions. Williams obtained potassium cyanide and hydrochloric acid in July 1984, using his official position as a police officer to assist him in purchasing the chemicals.
Williams’s wife died of cyanide poisoning six months later, on January 31, 1985, after again having confronted Williams earlier that day. The State served notice of aggravating factors c(4)(d), pecuniary motive, and c(4)(f), killing to escape detection. The defense offered mitigating factors c(5)(c), age (thirty-four years of age at time of the offense); c(5)(f), no significant criminal history (no prior convictions, no alcohol or drug abuse); and c(5)(h), any other factor (served in the military during the Vietnam war, trustee at church).
The jury found aggravating factor c(4)(f) and mitigating factors c(5)(f) and c(5)(h), but found the aggravating factor in equal balance with the two mitigating factors. (Had the jury imposed a death sentence on the basis of the c(4)(f) (killing to escape detection) factor in that case, we might not have sustained the imposition of the death sentence. See State v. Hightower, 120 N.J. 378, 420-21, 577 A.2d 99 (1990) (defining the c(4)(f) factor).) Williams received a life sentence with thirty-years parole ineligibility for the crime of purposeful and knowing murder.
John Dreher
Because the Appellate Division has reversed John Dreher’s conviction and remanded his case for retrial, State v. Dreher, 251 N.J.Super. 300, 598 A.2d 216 (1991), we repeat again our caveat that the facts below are as alleged by the State and here by Marshall. We evaluate the Dreher case on those assumptions.
*177Dreher brutally murdered his wife in the basement of their home. His was an aggravated, cold-blooded murder. He contemplated the crime for months, attempting to obtain a gun two to three months before the murder. Like Marshall, he too had been having an extra-marital affair for two years before his wife’s murder and had asked his paramour what she would think of him if he killed his wife.
On the day of the murder, Dreher had asked his paramour to meet him at the family home at 7:30 a.m. for a “confrontation” with his wife. After his sons left for school, Dreher dragged his wife unwillingly into the basement, as she begged him not to hurt her. The murder was brutal. He tied her hands behind her back and tied a rope around her neck, which was then tied to a column in the basement. Dreher then ordered his paramour to bring him something sharp; she complied. As Dreher tightened the rope around his wife’s neck with one hand, he stabbed her in the neck with the other. His paramour struck her on the head several times with a heavy tool and stabbed her in the back.
The medical examiner determined that Dreher’s wife had died of strangulation, and found that she had been stabbed eight times in the back and once in the throat. In his opinion, the back wounds were inflicted while she lay dying. Following the murder, Dreher tried to make the murder appear as if it had occurred during a robbery. He callously disposed of his wife’s jewelry, and lacked remorse.
In the prosecutor’s view, however, despite all of its horror, the case lacked the c(4)(c) statutory aggravating factor that the defendant had intended to inflict torture on his wife beyond the pain of killing. The prosecutor thus declined to prosecute the case capitally. A jury found Dreher guilty of murder and the judge sentenced him to life imprisonment with a minimum parole-ineligibility term of thirty years. Significantly, a principal ground of defense both at trial and on appeal was that the *178perpetrator of this offense was not Dreher but the paramour. See State v. Dreher, supra, 251 N.J.Super. 300, 598 A.2d 216.
Darrell Collins
Collins plotted the murders of his wife and eighteen-month-old son to obtain insurance proceeds on their lives. This, too, was a brutal murder. Collins, a martial-arts enthusiast, slashed his wife in the throat, face, and breasts with a sharp weapon, and then beat and suffocated his eighteen-month-old son as he lay in his crib.
Collins was twenty-six years of age and worked as a chef at the time of the murders. He had no prior criminal record and had completed one year of college. The Public Defender argues that although the case is not technically in the Master’s universe, it should be considered because Collins murdered a defenseless family member for pecuniary gain.
The jury found Collins guilty of the non-capital murder of his wife and of the capital murder of his son. The jury, however, did not find either the c(4)(d) aggravating factor of killing for pecuniary gain or the c(4)(f) factor of killing to escape detection. The trial court sentenced him to life with a thirty-year period of parole ineligibility for the murder of his wife, and to thirty years with a thirty-year period of parole ineligibility for the murder of his son, consecutive to the first term.
Thomas Johnston
Although it was not included in the Master’s universe, we have considered Johnston’s case because it involved the planned murder of a defenseless victim, his wife. The Master found insufficient evidence to substantiate either the c(4)(c) (torture/assault) or c(4)(d) (pecuniary motive) aggravating factors, but the Public Defender strongly suggests that both factors were present.
Johnston killed his wife by hitting her in the head with a hammer over twenty-five times. The severity of the beating was evidenced by the injuries she sustained — a fractured jaw, a *179depressed skull, five skull fractures, and lacerations to her left cheek. After the murder, Johnston attempted to cover up his crime. „ Johnston dragged her body into the woods near their home, leaving one shoe and her purse beside her car to make it appear as if she had been abducted. He arranged her clothes to give the appearance that she had been sexually assaulted, and then covered her body with a rototiller cover.
The murder was apparently precipitated by one of many arguments that the couple had had over financial matters during what for many years had been a stormy, abusive relationship. In 1980 Johnston had become financially dependent on his wife, and five years later she filed for divorce on the grounds of extreme cruelty. The State argued that the impending divorce had provided Johnston with a pecuniary motive for killing his wife. The defense asserted that Johnston had suffered from severe mental problems and that on the night of his wife’s murder, he had consumed a large amount of alcohol.
William and Herbert Engel
The murder committed by William and Herbert Engel is, for us, the most difficult because that case bears striking similarities to the subject case. William and Herbert Engel are brothers who were tried capitally for the murder of William’s ex-wife. The facts and background of this case are familiar to the Court because it participated in earlier bail-release applications, see State v. Engel, 99 N.J. 453, 493 A.2d 1217 (1985), and has reviewed defendants’ petitions for certification from the convictions of non-capital murder. 130 N.J. 393, 614 A.2d 616 (1991).
William procured his ex-wife’s murder with the assistance of his brother, Herbert, apparently because of jealousy and an obsession with his ex-wife. William delegated to Herbert the task of hiring a contract killer, much like McKinnon in Marshall’s case, one month before the murder, and had told a private investigator the year before that he wanted to get rid of his ex-wife. Herbert pressured an employee, McFadden, into *180agreeing to commit the murder-for-hire of his ex-sister-in-law by plying him with liquor and promising him money.
The murder was committed in a warehouse owned by William Engel. Engel lured his ex-wife to the warehouse on the false pretext that they were going shopping for birthday and Christmas gifts for their five-year-old daughter. Once inside, Engel pretended that the warehouse lights were not working when he escorted her past the bathroom door where McFadden waited, ready to strangle her.
William stood over his ex-wife and watched, smoking a cigarette, while McFadden strangled her to death. At one point during the four-minute ordeal, William called her a “bitch.” The two men then loaded her body into a waiting car, and, as planned, McFadden and a cohort, Pee Wee Wright, transported it to South Carolina where it was burned beyond recognition. (Herbert later ordered McFadden to execute the person who had helped dispose of the body because he feared that the executioner would tell the police of the murder.)
William Engel’s deception continued after the murder, when he twice telephoned his ex-wife’s house and reported first to her grandfather and then to her daughter that the victim had not arrived to meet him as planned. He told her mother that she had never arrived at their first meeting.
Like Marshall, William Engel was a well-educated, successful businessman, who had a good reputation in the community, was, involved in charitable work, and had no prior criminal record. Unlike Marshall, however, William Engel’s murder of his ex-wife lacked a pecuniary motive: he killed solely out of a jealous anger towards his ex-wife. The jury spared the brothers’ lives, finding the presence of mitigating factors c(5)(a) (mental/emotional disturbance) and c(5)(e) (duress), in addition to c(5)(f) (no prior criminal record) and c(5)(h) (any other factor). The jury found that those mitigating factors outweighed the sole aggravating factor, c(4)(e) (hired a killer). The trial court sentenced *181the brothers to life imprisonment with a thirty-year period of parole ineligibility.
The Public Defender suggests that one factor that may have saved the two brothers was the testimony of their elderly father, who told the jury of his own experiences in Austria during World War I and how his own father had been killed in a concentration camp. Defense counsel managed to suggest to the jury that Herbert felt beholden to William for his job. Overall, the criminal culpability of William Engel seems no different from that of Robert Marshall. The two husbands came from the same economic stratum. The victims were not strikingly dissimilar. The shattering impact on the families was the same.
The ultimate question concerns whether the fact that a jury spared the Engel brothers requires the invalidation of Robert Marshall’s death sentence. We do not believe that statutory disproportionality ever contemplated that two New Jersey juries must reach identical verdicts even in closely-similar circumstances. Our search should be for some impermissible or invidious factor or pattern that has been broken. That the Engel brothers were spared their lives does not establish a pattern of life-sentencing for such killings. We do not sense that some invidious factor tainted Marshall’s sentencing process.
The remaining spousal murders are distinguishable. Both Walter Williams and Darrell Collins escaped death when a jury found that they had not killed for pecuniary gain. John Dreher was thought by the prosecutor not to have met the criteria for a c(4)(c) aggravating factor.
2. Contract killers (principals).
This is another category of similar cases that we must consider. Because the Engel brothers were essentially contract principals, we need not discuss their cases further. We agree that there are no striking factual dissimilarities between their case and Robert Marshall’s, other than the fact that William *182Engel did not murder his wife for money. One cannot say with any degree of confidence that there is a significant difference in blameworthiness in either situation.
We now address the remaining contract principals and then discuss the hitmen and the principals associated with their cases.
Francis Brand
The Public Defender describes Francis Brand as the “remaining contract principal” who is compared to defendant in the Marshall Report. Arthur Brand, Francis’s older brother, had physically and mentally abused his family for years and had sold drugs from the family home. Francis, living in fear of his brother for years, procured the assistance of Randy Burroughs, a long-time friend, to kill his brother. Burroughs, who had a limited educational background, had tried to resist Brand’s attempts to convince him to kill, but after Francis Brand had urged him for about eighteen months and with the promise of money, Burroughs agreed. Burroughs shot Arthur Brand twice with a shotgun and killed him as he awoke.
The prosecutor did not seek to prosecute Francis capitally, perhaps because of the victim’s violent assaults on other family members and history of drug abuse. Those factors suggest less blameworthiness in Brand’s case.7
*1833. Contract killers (hitmen).
Anthony DiFrisco
The Court is familiar with the facts of that cold-blooded, execution-style killing. Anthony DiFrisco agreed, for $2,500 in cash and cancellation of a $500 drug debt, to kill a pizzeria owner because he was going to inform the police about DiFrisco’s employer’s drug dealings. DiFrisco, a drug addict, carried out the “hit” by shooting the owner four times in the head and once in the body as he reached to fill a soda order.
DiFrisco’s moral blameworthiness is at least equivalent to Robert Marshall’s. By his own account, he committed the contract murder to earn himself a professional reputation with a crime syndicate. The trial court sentenced DiFrisco to death after finding that the two aggravating factors, c(4)(d) (pecuniary motive) and c(4)(f) (killing to escape detection), outweighed the only mitigating factor, c(5)(g) (substantial assistance to the State in the prosecution of another for murder).
On appeal, we affirmed DiFrisco’s conviction of murder, but reversed his death sentence because of the absence of any extrinsic corroboration of his connection with the contract principal. DiFrisco, supra, 118 N.J. 253, 571 A.2d 914. We take it that the police are investigating the responsibility of the contract principal for that senseless killing.
James Clausell
We are familiar with this case, too, having heard the capital appeal. Clausell, supra, 121 N.J. 298, 580 A.2d 221. Clausell appears to have an overall culpability greater than Robert Marshall because of the absolute inanity of the killing. Like DiFrisco, Clausell’s motive for killing the victim was to receive $2,000 in cash and to develop a reputation as a hitman. The principal, a known drug dealer, hired Clausell to seek revenge *184on his neighbor who had filed a municipal-court complaint against him concerning his dog.
Clausell and a cohort carried out the “hit” by firing two shots through the front door of the victim’s house. One bullet fatally struck the victim in the chest, the other narrowly missed his daughter. A jury sentenced Clausell to death after finding the presence of c(4)(b) (grave risk) and c(4)(d) (pecuniary motive) aggravating factors. All that saved him from the imposition of the death penalty was some evidence that showed that he might have intended only to intimidate the victim and not to kill him, entitling him to a Gerald retrial. The contract principal in his case was given a life sentence, but again there was a residual doubt with respect to the mission on which he had dispatched Clausell.
Miguel Melendez
Miguel Melendez shot his victim twice in the head, killing him in the presence of the victim’s daughter after being hired by Lazaro Trimino for $2,500 to do so. Melendez had no role in planning the murder; even the handgun was supplied to him. Evidence suggested that he was at least mildly-mentally retarded and suffering from organic brain damage, had been abandoned by his father as a child, and had been institutionalized in state hospitals for most of his childhood. In addition, he cooperated with the police and confessed that Trimino had hired him to commit the murder, and at the trial expressed remorse for his part in the murder. For those reasons, his criminal culpability appears less than Robert Marshall’s.
Michael Rose
Michael Rose brutally murdered Kathryn Cveticanin for the payment of $540 in cash. The principal, Zoran Cveticanin, hired Rose to kill his step-mother to prevent her from inheriting his father’s estate. Rose, who also is mildly-mentally retarded, carried out that task in a vicious fashion — he stabbed his pregnant victim eighty-three times and beat her over the head with a blunt object approximately twenty times. When the victim became too weak to defend herself, Rose smoked a *185cigarette and then proceeded to beat her head and neck with a sump pump.
A jury found aggravating factor c(4)(c) (torture, depravity of mind), and mitigating factors c(5)(e) (duress), c(5)(f) (no prior record), c(5)(g) (substantial assistance to the State), and c(5)(h) (any other relevant factor). Because the jury could not reach a decision on the weighing of the factors, Rose received a sentence of life imprisonment with thirty-years parole ineligibility. His case is close; the jury was hung. The factors of duress and assistance to the State distinguish it.
Randy Burroughs
Burroughs killed Arthur Brand with the promise of $2,000 from the contract-principal, Francis Brand, the victim’s younger brother. After eighteen months of constant pressure to kill from Francis, Burroughs, without any planning, opened Arthur’s bedroom door and fired twice with a shotgun, killing him instantly. The evidence showed that Burroughs, who was mildly-mentally retarded, had killed not only for the money but to help his friend rid the family of his abusive brother.
Burroughs confessed to the killing and implicated Francis Brand in the conspiracy, later testifying for the State at Brand’s trial. He received a sentence of life imprisonment, with thirty-years parole ineligibility. The Public Defender recognizes that Burroughs’s blameworthiness is less than Marshall’s but believes that Burroughs’s lower blameworthiness would increase the contract-principal’s overall criminal culpability.
In sum, contract killers, like contract principals, appear to be recognized as significantly blameworthy. DiFrisco and Clausell were sentenced to death; Melendez, Rose, and Burroughs appear to have been somewhat less blameworthy because of their mild mental retardation; Burroughs showed the highest degree of remorse.
Although there is little point in debating whether society regards contract principals, who are less likely to repeat the *186crimes, as more blameworthy than contract hitmen (who appear to have no motives other than money), there is no clear line of demarcation that would indicate to us that to sentence Robert Marshall to death is disproportional. Many of the spousal or companion murders that we review, e.g., State v. Moore, 122 N.J. 420, 585 A.2d 864 (1991) (S. Moore) (killing of wife and child with hammer during fit of rage), and State v. Erazo, 126 N.J. 112, 594 A.2d 232 (1991) (killing of wife after an evening of drinking and arguing), result from violent outbursts of emotion. Marshall’s killing did not arise from a fit of rage. His involved a long period of premeditation without the slightest showing of any controlling emotion overcoming societal norms. Delaware finds particularly deathworthy those few who are “guilty of committing an unprovoked, cold-blooded murder of a person who lacked the ability to defend herself, solely for pecuniary gain.” Dawson, supra, 581 A.2d at 1108 (citing Riley v. State, 496 A.2d 997, 1027 (1985)).
4. Premeditated robbery or kidnapping murder cases.
These cases have some factual comparability with Marshall.
John Martini
John Martini received a death sentence for the kidnap-murder of a businessman.8 After receiving $25,000 ransom money from the victim’s wife, Martini completed his well-devised plan by shooting his victim, execution-style, three times to the back of the head.
Martini, fifty-eight years of age at the time of the killing, raised a diminished-capacity defense, claiming that a severe cocaine-abuse problem impaired his capacity to commit the crimes purposefully or knowingly. Marshall argues that he is-far less culpable than Martini and thus deserves a life sentence.
David Mark Russo
David Mark Russo planned the robbery of a gas station and the murder of its employees. Russo carried out the robbery by *187first attempting to execute his three victims as they lay on the floor. He shot each person at point-blank range, killing one and seriously injuring the other two.
Russo was thirty-two years of age at the time he committed his crimes and had a history of alcoholism, cocaine abuse, and depression. The penalty-trial jury found the c(5)(a) (emotional disturbance) and c(5)(d) (mental impairment) mitigating factors, and sentenced him to life imprisonment. Because of the high level of blameworthiness and victimization, the Master found Russo as criminally culpable as Robert Marshall.
Terrence Scales and Howard Thompson
Terrence Scales and Howard Thompson murdered their victim, whom they had met in a tavern, in order to steal his new car. After luring him back to their apartment, Scales and Thompson strangled their victim with a clothesline, and when he resisted, they beat him. After the victim’s death, the two men wrapped his body in a blanket and callously disposed of it on a river embankment.
Scales and Thompson each received a sentence of life imprisonment. Scales’s case proceeded to a penalty phase in which the jury deadlocked on the issue of death after it had found aggravating factor c(4)(f) (escape detection) and mitigating factors c(5)(d) (mental impairment) and c(5)(h) (catch all). Thompson’s case did not advance to a penalty trial.
Vernon Mclver
Vernon McIver, a nineteen-year-old male prostitute, murdered his homosexual client by plunging a large butcher knife into his neck as he lay on the bed. After the killing, Mclver stole the victim’s wallet and car. He pleaded guilty to felony murder and received an aggregate forty-year prison term with a thirty-year period of parole ineligibility.
We are unable to agree that those cases demonstrate the disproportionality of Robert Marshall’s death sentence. That *188John Martini may be more deathworthy than Robert Marshall does not establish that Marshall’s sentence is disproportionate. In the Russo case the jury found mitigating factors not present in Marshall’s. The Scales/Thompson and McIver cases are far too dissimilar to aid in the proportionality review of Marshall's death sentence.
5. Other cases.
We have considered as well, although we do not discuss in detail, each of the other cases mentioned in the Public Defender’s brief. They concern Eugene Berta, who committed a possibly pecuniary-motive killing of a girlfriend; Salvatore Ferrari, who killed his mother by his own hand, allegedly in order to obtain over $30,000; William Todd Lewis, who killed his twin brother to inherit a share of his estate; Nelson Jalil, who murdered his pregnant wife to relieve himself of the impending financial strain of a child; Jose Machado (a case with which we are familiar), who brutally killed his pregnant girlfriend; Edward Freeman, who poisoned his wife to gain his freedom; Bert Rindner, who killed a business partner; Joseph Presher, who brutally killed a close woman friend.
Although each of those cases demonstrates deathworthiness to some degree, each contains sufficient factual differences, i.e., not contract killings, and either the absence of specific findings of the statutory aggravating factors or evidentiary problems concerning the aggravating or mitigating factors that do not require us to invalidate Marshall’s death sentence.
VI
That Juries in New Jersey do not “Generally ” Return Death Verdicts does not Undermine the Deterrent Value of the Death Penalty to Such an Extent as to Render it a “Cruel and Unusual Punishment.’’
In Ramseur, supra, 106 N.J. at 166-82, 524 A.2d 188, this Court upheld the death-penalty statute on both federal and *189state grounds against a facial claim that the statute constituted cruel and unusual punishment. The Court agreed with the plurality in Gregg, supra, 428 U.S. 153, 96 S.Ct. 2909, 49 L.Ed.2d 859, that although the death penalty is severe and irrevocable, it is not an excessive or disproportionate penalty for the crime of murder.
The Court stated in Ramseur that “although the view is not unanimous, it is a widely held belief, and a strongly held one in our society, that the appropriate penalty for murder may be death.” 106 N.J. at 172, 524 A.2d 188. Citing information supplied by the Public Defender, the Court noted that “[i]f the actions of jurors are to be taken as a true reflection of society’s morality, the most recent evidence strongly supports the view that the death penalty does not violate contemporary standards of decency. Since the restoration of capital punishment in 1982, juries in this state have imposed twenty-six sentences of death.” Id. at 173, 524 A.2d 188.
The Public Defender asks us to reconsider that conclusion, arguing that as a result of the work of the Proportionality Review Project, this Court now has before it complete evidence that strongly supports the conclusion that the decisions rendered by New Jersey’s juries demonstrate that the death penalty violates contemporary standards of decency.
A.
The Public Defender’s argument is drawn from the premise of Justice White’s concurring opinion in Furman, supra, 408 U.S. at 310, 92 S.Ct. at 2763, 33 L.Ed.2d at 390. He wrote:
I begin with what I consider a near truism: that the death penalty could so seldom be imposed that it would cease to be a credible deterrent or measurably to contribute to any other end of punishment in the criminal justice system. * * * But when imposition of the death penalty reaches a certain degree of infrequency, it would be very doubtful that any existing general need for retribution would be measurably satisfied. Nor could it be said with confidence that society’s need for specific deterrence justifies death for so few when for so many in like circumstances life imprisonment or shorter prison terms are *190judged sufficient, or that community values are measurably reinforced by authorizing a penalty so rarely invoked.
[Id. at 311-12, 92 S.Ct. at 2763, 33 L.Ed.2d at 391.]
Justice White reasoned that in such a circumstance death would then be “the pointless and needless extinction of life with only marginal contributions to any discernible social or public purposes.” Id. at 312, 92 S.Ct. at 2764, 33 L.Ed.2d at 391. Justice White concluded that “[a] penalty with such negligible returns to the State would be patently excessive and cruel and unusual punishment violative of the Eighth Amendment.” Id. at 312, 92 S.Ct. at 2764, 33 L.Ed.2d at 391-92. In Gregg, the opinion announcing the judgment observed that “[t]he death penalty is said to serve two principal social purposes: retribution and deterrence of capital crimes by prospective offenders.” 428 U.S. at 183, 96 S.Ct. at 2929-30, 49 L.Ed.2d at 880. Unless the death penalty measurably contributes to one or both of those goals, it “is nothing more than the purposeless and needless imposition of pain and suffering,” and hence is an unconstitutional punishment. Coker, supra, 433 U.S. at 592, 97 S.Ct. at 2866, 53 L.Ed.2d at 989.
That thread of Supreme Court jurisprudence has remained constant. Thus, the Court has embraced the view that “when legislatures and juries [have] firmly rejected the penalty of death,” Enmund, supra, 458 U.S. at 814, 102 S.Ct. at 3385, 73 L.Ed.2d at 1162 (O’Connor, J., dissenting), the penalty of death is substantially disproportionate to the offense.
B.
The Public Defender argues that that point has come in the administration of New Jersey’s Capital Punishment Act. He relies on the Master’s Report to this Court:
The principal trend in New Jersey’s capital charging and sentencing system between 1983 and 1991 has been a marked decline in the frequency with which death sentences are imposed among death-eligible eases. * * * The overall rate before 1988 was .21 (29/140), while the overall rate after 1987 has been .06 *191(5/87). Among the death-eligible murders committed since January 1, 1987, a death verdict has been returned in only two cases.
[Final Report, supra, at 15.]
The Public Defender argues that the average level of frequency for all death-eligible cases over the past nine years is .15. Id. at Table 1. The data in Table 2 of the Final Report indicate that only during the first calendar year of the statute’s operation did the frequency level, in all penalty-trial cases, rise to .50, and in only one other calendar year, 1987, did the level of frequency in all penalty-trial cases rise to .43. Overall, the average level of frequency is .30. Id. at Table 2. The Public Defender suggests that when recent penalty trials are added, the average rates are even lower.
After discussing various suppositions in the Master’s Report with respect to the causes for such a decline in death-sentencing rates, the Public Defender argues that whether those prosecutorial judgments are a function of attempts to predict jury-sentencing behavior or resource allocations, the trend in prosecutorial decisionmaking lends further support to the proposition that the death penalty offends current standards of decency in New Jersey.
The Public Defender thus argues from the Master’s writings that to satisfy Justice White’s constitutional concern over “[a] penalty with such negligible returns to the State,” Furman, supra, 408 U.S. at 312, 92 S.Ct. at 2764, 33 L.Ed.2d at 391, the frequency of death sentencing must be “substantial,” noting that “imposing the death penalty in only one out of every two factually similar cases does not satisfy conventional notions of evenhandedness,” and that a death-sentencing frequency rate less than .80 is “too low to comply with notions of evenhandedness or regularity.” David C. Baldus, George Woodworth & Charles A. Pulaski, Equal Justice and the Death Penalty: A Legal and Empirical Analysis, 76-77 n. 50 (1990).
*192c.
We disagree that the declining rate of the imposition of the death penalty by New Jersey juries makes it a cruel and unusual punishment. In Ramseur, we emphasized that there was no way in which a capital-sentencing system could be produced that could mechanically predict the outcome of capital trials. We explained in the face of a dissent questioning the procedural protections afforded under the Act: “If the suggestion is that capital defendants are entitled to perfection, to totally consistent, accurate and reliable procedures, obviously not only this Act but any death penalty act would be unconstitutional.” 106 N.J. at 192, 524 A.2d 188.
In the face of claims that the statute could be arbitrarily and capriciously applied, we have attempted to set standards to narrow the class of death-eligible murderers. In Ramseur, we undertook to “narrow imprecise [c(4)(c) ] statutory language in order to render it constitutional.” 106 N.J. at 200, 524 A.2d 188. In Gerald, supra, 113 N.J. at 89, 549 A.2d 792, we interpreted the Act to limit the sentence of death to those who intended to kill, not just injure. In Bey II, supra, 112 N.J. at 165, 548 A.2d 887, we rejected any idea of a mechanical or numerical balancing of factors in the death-sentencing process. We made it clear that jurors must be instructed that it is they who must make the qualitative judgment about who shall live or who shall die. Ibid. In Zola, supra, 112 N.J. at 428-31, 548 A.2d 1022, we recognized the request that defendants be permitted to plead for life at the hands of a jury. In State v. Davis, 96 N.J. 611, 618-21, 477 A.2d 308 (1984), we allowed any relevant evidence bearing on the defendant’s potential for rehabilitation to be presented to a jury in a capital-sentencing phase.
That individualized assessments would be made of defendants was inevitable in those rulings. In Ramseur, we noted that “the Supreme Court has categorically rejected blind unifor*193mity in the sentencing of capital defendants. Under constitutionally approved sentencing schemes, the process must guarantee an individualized assessment of the defendant.” 106 N.J. at 331, 524 A.2d 188 (citation omitted). The strangest anomaly would result if those guarantees, intended to insure recognition of common humanity in even the most shocking of crimes, were thought to make punishment under the Act “cruel and unusual.” On the contrary, the very humanity of the sentencing process, channeled as best the Legislature can channel it, is what makes the Act constitutional. In Woodson v. North Carolina, 428 U.S. 280, 293, 96 S.Ct. 2978, 2985-86, 49 L.Ed.2d 944, 954 (1976), the Court concluded that a mandatory death sentence that would surely produce statistical uniformity would just as surely violate the Eighth Amendment.
The premise of the Coker/Enmund analysis, as explained by Justice O’Connor in her dissenting opinion in Enmund9 is fourfold. 458 U.S. at 801, 102 S.Ct. at 3379, 73 L.Ed.2d at 1154. That analysis involves (1) the evaluation of “contemporary standards” of decency drawn from two sources, (a) legislatures and (b) juries; and (2) evaluation of punishment in relation to (a) the degree of harm (the question of inflicting death for a nonfatal rape), as well as (b) the degree of blameworthiness. Id. at 825-26, 102 S.Ct. at 3391, 73 L.Ed.2d at 1169-70.
In this case that the magnitude of harm inflicted on the victim (death) and the degree of blameworthiness (intentional contract killing) surpass any measure of constitutional restraint is clear. In the context of mandatory sentencing, the Eighth Amendment is violated only when “[t]he two crucial indicators of evolving standards of decency respecting the *194imposition of punishment in our society — jury determinations and legislative enactments — both point conclusively to the repudiation” of the death penalty. Woodson, supra, 428 U.S. at 293, 96 S.Ct. at 2986, 49 L.Ed.2d at 954 (opinion of Stewart, J.). Although one could argue that twenty-six10 death sentences out of 246 death-eligible cases is a “credible threat essential to influence the conduct of others,” Furman, supra, 408 U.S. at 311, 92 S.Ct. at 2763, 33 L.Ed.2d at 391 (White, J., concurring), we need not debate the point because we are convinced that a low level of jury death verdicts would not render the Act unconstitutional.
That New Jersey juries have reserved the death penalty for a small segment of the death eligible does not make a sentence arbitrary and capricious so long as adequate standards are in place and there is no evidence of impermissible factors influencing the jury. As the Public Defender has noted in his race-of-victim, race-of-defendant analysis, infra at 210, 613 A.2d at 1110, jury behavior is quite predictable at the high and low ranges of aggravating and mitigating factors. In that respect, New Jersey juries may be seen to be fulfilling the Legislature’s wish that the death penalty be but rarely invoked:
[T]he bill has some rather rigorous provisions in it for the protection of the defendant, the theory being that ultimate penalty is paid, as difficult as that will be on all of us who have a part in it, when that day comes, we want to at least feel we have tried to cover every possible contingency for the protection of the defendant and hopefully it will be utilized only in the most extreme cases.
*195[Public Hearing, N.J.Senate Judiciary Committee, Senate No. 112, at 1 (Feb. 26, 1982) (statement of the bill’s sponsor, Senator John Russo).]
VII
Geographic Patterns of Charging and Prosecuting Capital Cases do not Demonstrate an Arbitrary Exercise of the Prosecutorial Function.
Next, we consider defendant’s claim that New Jersey’s Capital Punishment Act is unconstitutional because the geographic distribution of capital-charging and sentencing decisions within the state demonstrate that capital punishment is arbitrarily imposed. Defendant claims that decisional disparities between the counties render imposition of death sentences inconsistent, unfair, and arbitrary, contrary to the Eighth Amendment to the United States Constitution and Article I, Paragraph 12 of the New Jersey Constitution.
In addition, defendant advances a more particularized argument regarding his death sentence. Defendant asserts that his sentence should be vacated because he was twice adversely affected by geographic variations in the pursuit and implementation of capital sentences. Defendant was indicted for capital murder in Ocean County at a time when, according to his calculations, such prosecutions were undertaken without exception. Afterwards, because of a change of venue, defendant was tried and sentenced in Atlantic County, which he asserts has a higher-than-average conviction rate in capital cases. Defendant argues that this unique geographic combination meant that his chance of receiving a death sentence was greater than that of any other defendant in the universe of capital cases.
After a careful review of the statistical evidence submitted by the Master and the State’s expert, we conclude that defendant has not shown that any variations in capital-prosecution and sentencing practices in the state amount to a constitutional deficiency in the application of the death penalty. We remain mindful of the potential for abuse of prosecutorial discretion in *196capital decisionmaking and the threat it poses to the desired uniformity in pursuit of such sentences. However, the data presented in this appeal do not establish the existence of such arbitrariness on the part of prosecutors.
A.
We recognized the potential for arbitrariness in prosecutorial decisionmaking with respect to capital cases early in our experience with the death-penalty statute. In State v. McCrary, 97 N.J. 132, 141, 478 A.2d 339 (1984), we acknowledged the significant consequences that flow from a decision to seek a death sentence and found ourselves persuaded “that some judicial scrutiny of prosecutorial charging [was] necessary.” Thus, we held that defendants who had been served with a notice of an aggravating factor could, through summary proceedings before the trial court, challenge the sufficiency of the evidence to support that factor when evidence is clearly lacking to support the charge. Id. at 142, 478 A.2d 339. Our stated goal was “to effect only a minimal intrusion into this area of prosecutorial discretion” in light of the “broad discretionary powers” historically exercised by prosecutors in determining charges. Ibid.
When undertaking our first examination of a death sentence, we upheld the Act against a broad-based constitutional attack but recognized that we would eventually be asked to review “concerns about possible misuse of prosecutorial discretion” in seeking the death penalty. Ramseur, supra, 106 N.J. at 329, 524 A.2d 188. Our prediction in Ramseur was fulfilled the next year in Koedatich, supra, 112 N.J. 225, 548 A.2d 939, when the defendant asserted that the death-penalty statute provided prosecutors with unfettered discretion to determine which defendants would be exposed to the possibility of a capital sentence, in violation of the Constitution. Koedatich claimed that “[w]hile limiting jury sentencing discretion helps to assure that those sentenced to death are a rational subset of those actually charged of capital crimes, it does not at all *197assure that those convicted are a rational subset of those who could be charged with a death-eligible offense.” Id. at 251-52, 548 A.2d 939. Koedatich did not make a claim of individualized abuse of discretion in the selection of his case for capital prosecution. Instead, he relied, as defendant in this appeal does, on statistical data that suggested that cases were prosecuted as capital matters in a manner that was inconsistent from county to county.
We began our analysis in Koedatich by noting that Gregg, supra, 428 U.S. 153, 96 S.Ct. 2909, 49 L.Ed.2d 859, holds that the federal Constitution “does not require limits on prosecutorial discretion beyond the aggravating factors outlined in the statute.” 112 N.J. at 252, 548 A.2d 939. We looked beyond that requirement, however, and found that “the New Jersey Constitution * * * mandates consistency and reliability in the administration of capital punishment.” Id. at 251, 548 A.2d 939 (citing Ramseur, supra, 106 N.J. at 190, 524 A.2d 188). Of course, “[t]o ensure a complete absence of discretion at each stage of decisionmaking would be an impossible task for the Legislature or for this Court. Indeed, to restrict discretion completely at the prosecutorial stage would be unconstitutional.” Id. 112 N.J. at 252, 548 A.2d 939. “[C]apital defendants are not ‘entitled to perfection, to totally consistent, accurate and reliable procedures.’ ” Id. at 251, 548 A.2d 939 (quoting Ramseur, supra, 106 N.J. at 192, 524 A.2d 188).
Instead, we held that
[t]he critical question in assessing prosecutorial discretion is what standards are applied to move a case from death-possible to death-eligible status. To assist in answering this question, this Court cannot rely solely on county-by-county statistical discrepancies or on findings that are developed exclusively by defense counsels’ evaluation of the case included in the data base.
That there are differences among the counties in the likelihood that a prosecutor will pursue — or the coincidence that the prosecutor has more often pursued — a capital prosecution does not, standing alone, demonstrate that the death penalty is being arbitrarily imposed. Surely, there are a myriad of reasons why a prosecutor handles different cases differently, such as the willingness of a defendant to plead guilty, the strength of the State’s case, a defendant’s cooperation in the State’s case against a co-defendant, the relative *198weight of the statutory aggravating and mitigating factors, the availability and relative credibility and persuasiveness of witnesses, and the resources of the county prosecutor’s office, to list only a few.
[Id. 112 N.J. at 256, 548 A.2d 939.]
Although we were reluctant to draw conclusions from the preliminary statistical data submitted in Koedatich, we nevertheless “recognize[d] the need for greater guidance for prosecutors as they attempt to perform their constitutional duty of enforcing this statute.” Id. at 258, 548 A.2d 939. Therefore, we “strongly recommend[ed] that the Attorney General, and the various County Prosecutors, in consultation with the Public Defender, adopt guidelines for use throughout the state by prosecutors in determining the selection of capita! cases.” Ibid. We predicted that such guidelines would “promote uniformity in the administration of the death penalty, [and would] be an additional safeguard against arbitrariness and an assistance to this Court in its developing proportionality review.” Ibid.
In considering defendant’s challenge, we operate in a more informed atmosphere than that which was available at the time we decided Koedatich. First, we now have before us highly-developed data compiled by our Special Master as well as a statistical report submitted by an independent expert hired by the State. In addition, the Guidelines for Designation of Homicide Cases for Capital Prosecution were adopted in February 1989, and prosecutors throughout the state have been operating pursuant to them since that time.
B.
The Master’s Final Report examines both the prosecutorialselection process and jury determinations in the various counties. The Report categorizes geographic-sentencing disparities in three ways: (1) the. state’s twenty-one counties are divided into three regions: “north,” “northwest,” and “southern;” (2) they are classified as “urban” or “non-urban;” and (3) the state is examined as a whole. The Report concludes that the overall *199death-sentencing rate among death-eligible offenses is more than twice as high in non-urban as in urban areas. In addition, the Report determines that a substantially-higher death-sentencing rate in the southern part of the state makes the overall rate there approximately two times higher than it is in the north and northwest.
. The Attorney General challenges the Report’s method of classifying counties as “urban” or “non-urban.” For example, the Report labels as “urban” those counties that Professor Baldus determined to embrace a major urban center: Camden, Essex, Hudson, Mercer, Middlesex, Passaic, and Union. But absent from that list is Bergen County, which, like all remaining counties, was classified as “non-urban.” However, Bergen County, with its substantial population, sits adjacent to three large cities: Paterson, Passaic, and Clifton, and shares a border with Manhattan and the Bronx in New York. Additionally, the county is home to major industrial development, highly-travelled highways, and concentrated retail establishments. Although some of the county is decidedly suburban in character, it should have been included in the urban counties for purposes of the Report.
Furthermore, the urban/non-urban classifications group substantially different counties together. For example, suburban Morris County in the northern part of the state shares non-urban status with rural Salem County far to the south. In addition, the Report does not account for counties that have both significant urban and significant rural characteristics. Atlantic County embraces Atlantic City, a municipality that could hardly be more urban. At the same time, however, the western end of the county is quite rural. Classification of the county solely as one or the other cannot accommodate those characteristics.
The Attorney General also questions the classification of the regions. Instead of dividing the state into the familiar north, central, and south, the Report uses regions of north, northwest, *200and south. For example, Monmouth and Ocean Counties, which are clearly within the ring of counties comprising the New York metropolitan area, see Metropolitan Regional Report, are classified as part of South Jersey.
We note that the foregoing issues underscore the difficulty of packaging the various counties into neat, absolute categories for the purpose of statistical comparison and to explain why we do not accept the Final Report’s findings in this area as conclusive. Although the Final Report is not perfect in its statistical analysis (such a standard cannot reasonably be expected), nevertheless, its overall findings are instructive on the general trends in capital-charging and sentencing decisions and serve as a useful basis from which to examine the geographic application of the death-penalty statute.
C.
The Final Report, using the above classifications, determines that prosecutors in non-urban counties seek the death penalty at a rate 1.6 times more frequent than their urban counterparts. In addition, regional disparities demonstrate that prosecutors in the northwest region seek the death penalty at a rate 1.6 times more frequent than do prosecutors in the north and 1.3 times more frequent than do prosecutors in the south. When individual counties are considered, the Final Report shows about a sixty-eight-percentage-point spread, from a low penalty-trial rate of .32 to a high in two counties where all cases determined by Baldus to be death eligible advanced to a penalty trial. Although those figures are worthy of concern, they come nowhere near the aberrations noted in Maryland. In one Maryland county prosecutors seek the death penalty in “100% of the death-eligible cases, whereas in Baltimore City, they seek that penalty in only 1.8%.” Tichnell, supra, 468 A.2d at 31-32 (Davidson, J., dissenting).
With respect to jury behavior, the Report indicates that the overall death-sentencing rate is more than twice as high in non-*201urban than in urban counties. In addition, the death-sentencing rate in the southern part of the state is approximately two times higher than it is in the north and northwest. Notably, almost one-fourth of the counties have a death-sentencing rate of zero, compared to three counties that have a rate of fifty percent.
In response, the Weisberg Report, submitted by the Attorney General, attempts to explain the unreliability of Professor Baldus’s geographic determinations. Dr. Weisberg first observes that the sample sizes for most counties are simply too small for any significant statistical analysis. Overlooking that flaw, however, Dr. Weisberg concludes that the Final Report discloses only raw numerical disparities, which demonstrate neither statistical significance nor constitutional deficiency.
Most significantly, the Weisberg Report argues that the Final Report fails to take into consideration the case-mix differences across the state, which could account for some of the variations in prosecution rates between the counties. In other words, the Special Master’s figures are unadjusted because they do not consider the aggravating or mitigating circumstances attendant to each case. To illuminate this point, Dr. Weisberg divides the universe of cases into four culpability categories beginning with Category I (the most culpable) and ending with Category IV (the least culpable).
Without reciting the data, we note that the Weisberg Report’s ultimate conclusion is that
the variation in penalty-trial rates * * * across areas of the state is primarily attributable to the different case mixes in these areas. After adjustment for culpability level, these differences are greatly reduced and fail to achieve statistical significance. Thus, the notion that prosecutors differ substantially in terms of their personal propensities to seek the death penalty appears implausible.
[Weisberg Report, supra, at 37.]
Defendant challenges the validity of Dr. Weisberg’s conclusions. Defendant first asserts that the Weisberg Report does *202not address county-by-county disparities in prosecution and sentencing practices because its culpability categories are applied only by region and urban/non-urban classifications. Second, defendant argues that the Weisberg Report supports a conclusion that the entire capital-punishment system is flawed because it demonstrates that some variation in prosecution and sentencing exists even after application of Dr. Weisberg’s culpability categories.
Defendant argues that because prosecutors operate with a county-wide jurisdiction and not on a regional or urban/non-urban basis, Weisberg’s report does nothing to address discretion of individual prosecutors. Again, we note that Dr. Weisberg’s report, like any other statistical study of capital-punishment practices, does not address every possible combination of death-penalty practices in the state. However, Dr. Weisberg’s statistics are helpful because they shed light on capital-prosecution and sentencing practices throughout the state, and we view his findings as informative in our review of defendant’s arguments.
D.
Defendant argues that the Final Report portrays a “crazy-quilt” of prosecutorial decisionmaking practices that have rendered the Prosecutorial Guidelines nugatory and that result in an unpredictable and unconstitutional application of capital sentences. The Master’s Report and Dr. Weisberg’s rejoinder together demonstrate that capital-prosecution decisions are not carried out with complete uniformity. Yet, we do not accept the proposition that the exercise of any discretion on the part of the prosecutor renders our system of capital punishment unconstitutional. Prosecutors must make determinations regarding the likelihood of successfully securing a death sentence when deciding whether to commit resources to the effort to do so. That process appropriately involves evaluation of the evidence and predictions about the ability of the State to prove beyond a *203reasonable doubt that the aggravating factor or factors outweigh those in mitigation.
Inevitably, in some cases that could result in a capital conviction a prosecutor will determine that accepting a plea to murder would be a prudent course of action based on an evaluation of the evidence. Another prosecutor presented with the same case will have no slide rule to tell him or her when the expenditure of resources in pursuit of a capital sentence can wisely be undertaken. Such a variation does not offend the State Constitution. Although the statistics suggest the possibility that different norms and cultures (and societal conditions) may be producing different outcomes not desirable under our notions of equal justice, and although we are disturbed by that possibility, nevertheless ensuring that all twenty-one prosecutors would make the same capital-charging decision in every case is simply unattainable and constitutionally unnecessary.
Of course, we would find unacceptable any suggestion that a particular prosecutor or group of prosecutors had engaged in a charging decision that amounted to an aberration and related not to evidentiary strength but to some other impermissible consideration. The guidelines adopted by the New Jersey County Prosecutors Association require prosecutors to hew closely to the statutory requirements of the Act and to evaluate the weight of the evidence to sustain any aggravating factor. The guidelines also require avoidance of any extraneous influences of race, sex, status of defendant or victim, or of notoriety of any case or the resources to prosecute it. The guidelines explicitly negate “any substantive or procedural rights.” The guidelines provide no system of administrative review at the county or state level.
Aside from the concerns about the race of the victim and race of the defendant that we shall discuss in Part VIII, infra at 207, 613 A.2d at 1108, the Final Report presents no other evidence of impermissible bias in the prosecutory function. Although capital-charging decisions vary to some extent from *204county to county, we do not find the deviations constitutionally significant. Nor does the statistical evidence suggest that jurors in a particular county or counties are acting arbitrarily. Jury deliberations will, by their nature, vary to some degree. Long before capital jurisprudence, defense counsel resisted transfers of venue from metropolitan areas to less-urban ones. That was not done here. Defendant sought a transfer from non-urban Ocean County because of adverse pretrial publicity. Atlantic County, except for Atlantic City, shares much of the flavor of Ocean County.
If all death-penalty cases resulted in a death sentence in South Jersey and none of the death-penalty cases in North Jersey resulted in a death sentence, then our system might be unconstitutional. However, the community for death-penalty purposes is the entire State of New Jersey. Thus we do not find that the system as a whole is unconstitutional simply because death-sentencing rates may vary somewhat from county to county. The statutory aggravating factors as well as limits on prosecutorial decisionmaking ensure that only death-eligible cases arrive before a penalty-phase jury. Jurors must unanimously find aggravating factors and are required to - determine beyond a reasonable doubt that such factors outweigh those in mitigation. Absent evidence of impermissible systemic bias, some variation among juries with respect to sentencing is inevitable; it does not equate with unconstitutional application of our Capital Punishment Act.
E.
Defendant also challenges his specific sentence as arbitrary and capricious because of the unique circumstances of having been charged with capital murder in a county with a higher-than-average indictment rate and sentenced to death in a county with a higher-than-average death-sentencing rate.
Defendant argues that at the time of his indictment the Ocean County Prosecutor exercised no discretion with regard to *205seeking the death penally. According to defendant, such a rate of prosecution was twice the state-wide average. Thus, defendant contends that had he murdered his wife a little further south on the Parkway in Atlantic County, the likelihood of his being subject to a capital prosecution would have gone from 100% to 43%. For his argument, defendant relies on the statistics contained in the Final Report.
After defendant’s indictment, his trial was moved to Atlantic County, which he contends has the highest penalty-trial death-sentencing rate in the state. According to the Final Report, Atlantic County juries impose a death sentence in one out of every two penalty trials. That sentencing rate is twenty percentage points higher than the state average and is consistent with the rate in only one other county.
The State counters that defendant applies an oversimplistic reading to the Report’s findings, which do not account for the case-mix differences across various areas. All of the cases identified by the Final Report as death eligible arising from Ocean County were prosecuted to the penalty trial as capital cases. Thus, the Report concludes that the county initiates capital prosecutions at a rate of 100%. However, a review of the aggravating factors presented by specific cases indicates that all cases identified as “death eligible” in Ocean County fell into either Category I or Category II of Dr. Weisberg’s classifications. Thus, that county was presented with an unusually-culpable group of defendants, which could justify its high prosecution rate. In contrast, the Atlantic County cases represented a wider range, and one-half of the total Atlantic County cases were classified as Category IV, the only category for which capital prosecution is uncommon. Hence, a less-frequent-prosecution rate for that county would be reasonable.
Defendant makes a related argument with respect to his sentencing in Atlantic County. Because of a change of venue, defendant was tried in Atlantic County. Atlantic County is one of two counties with the highest penalty-trial death-sentencing *206rates in the state. Juries in Atlantic County impose death in one out of every two penalty trials, a rate twenty percent higher than the state average.
The State argues that defendant’s Atlantic County argument relies on flawed data that present no reliable evidence of arbitrariness. The State points out several mistakes in defendant’s sentencing data. The Public Defender erroneously counted Craig Hart and Alfonso Timpson as non-penalty-trial Atlantic County cases. Recalculating defendant’s figures, the State concludes that there were eight penalty-trial cases in Atlantic County (Gerald, Hart, Highlander, Huff, Long, Marshall, Perry, and Timpson) of which three (Gerald, Long, and Marshall) resulted in a death sentence. Such calculations arrive at a .375 frequency rate, a death-sentencing rate lower than Ocean County.
Apart from the statistics, the State contends that there is a basic element of defendant’s case that precludes any legitimate argument regarding county-charging differences: the murder-for-hire setting and resulting aggravating factor c(4)(e). In three of the four cases identified by the Master as having aggravating factor c(4)(e) the prosecutor instituted a capital prosecution, disclosing that even with that small sample the circumstances of Robert Marshall’s case cut across county lines and received uniform treatment.
We do not find defendant’s argument on these issues to be persuasive. The prosecution and sentencing rates in Ocean and Atlantic Counties do not vary significantly from the rates of other counties. Moreover, any county-by-county disparities are the result of case mixes rather than of geography. Similarly, the fortuitous case mix presented to Ocean County, rather than geography, explains its high rate of capital prosecutions. Defendant has not convinced us that the prosecution or sentencing rates of either county amount to an aberration that resulted in an unjustified sentence.
*207VIII
Neither the Race of the Victim Nor the Race of the Defendant Has Been Shown to be an Impermissible Invidious Factor in the Imposition of the Death Penalty.
A.
A disturbing aspect of the Master’s Report is the suggestion that there may be a discrepancy in capital-sentencing rates that correlate to the race of the defendant or the race of the victim.
The United States Supreme Court seems resigned to accept that such “[a]pparent disparities in sentencing are an inevitable part of our criminal justice system.” McCleskey v. Kemp, 481 U.S. 279, 312, 107 S.Ct. 1756, 1778, 95 L.Ed.2d 262, 291 (1987). Absent purposeful discrimination, the Supreme Court apparently will not invalidate a death sentence on the basis of racial disparity, no matter how profound.
New Jersey’s history and traditions would never countenance racial disparity in capital sentencing. As a people, we are uniquely committed to the elimination of racial discrimination. All of our institutions reflect that commitment. We were among the first of the states that enacted a civil rights law. L. 1945, c. 169, codified at N.J.S.A. 10:5-1 to -42. “[Racial] discrimination threatens not only the rights and proper privileges of the inhabitants of the State but menaces the institutions and foundation of a free democratic State * * N.J.S.A. 10:5-3. Our decisional law has always reflected the “strength of the State’s policy” in this area. Jackson v. Concord Co., 54 N.J. 113, 123, 253 A.2d 793 (1969). To countenance racial discrimination in capital sentencing would mock that tradition and our own constitutional guarantee of equal protection of the laws under New Jersey Constitution Article I, paragraph 1.
As a Court, we have repeatedly emphasized our special commitment to equality in the administration of justice. As we have stated, to exclude from jury service qualified groups “ ‘not *208only violates our Constitution and the laws enacted under it but is at war with our basic concepts of a democratic society and a representative government.’ ” State v. Gilmore, 103 N.J. 508, 526, 511 A.2d 1150 (1986) (quoting Smith v. Texas, 311 U.S. 128, 130, 61 S.Ct. 164, 165, 85 L.Ed. 84, 86 (1940) (footnote omitted)). The imposition of capital-death sentences based solely on the race of the defendant or the victim would be equally at war with the basic concepts of a democratic society and a representative government.
B.
We have no doubt that a white defendant may argue that our capital-sentencing system is constitutionally invalid. In Spinkellink v. Wainwright, 578 F.2d 582 (5th Cir.1978), cert. denied, 440 U.S. 976, 99 S.Ct. 1548, 59 L.Ed.2d 796 (1979), a convicted murderer raised Eighth and Fourteenth Amendment due-process and equal-protection challenges on the basis of a statistical study purporting to demonstrate a race-based disparate impact in sentencing. The court ruled: “Spenkelink11 ha[d] standing to raise the equal protection issue, even though he is not a member of the class allegedly discriminated against, because such discrimination, if proven, impinges on his constitutional right under the Eighth and Fourteenth Amendments not to be subjected to cruel and unusual punishment.” Id. at 612 n. 36. That court relied on the Supreme Court’s reasoning in Taylor v. Louisiana, 419 U.S. 522, 95 S.Ct. 692, 42 L.Ed.2d 690 (1975). Taylor involved a male defendant who had objected to the exclusion of women from his jury. Although “Taylor was not a member of the [injured] class * * * there is no rule that claims such as Taylor presents may be made only by those defendants who are members of the [injured class].” 419 U.S. at 526, 95 S.Ct. at 695, 42 L.Ed.2d at 695.
*209Defendant surely has a right to raise a structural challenge to the constitutional fairness of the New Jersey Capital Punishment Act. A death-penalty statute that systematically discriminates on the basis of race of the victim or race of the defendant “menaces the institutions and foundation of a free democratic State.” N.J.S.A. 10:5-3.
The McCleskey. Court reasoned that although the statistical data showed a discrepancy that may correlate with race, “disparities in sentencing are an inevitable part of our criminal justice system.” 481 U.S. at 312, 107 S.Ct. at 1778, 95 L.Ed.2d at 291. The Court held that McCleskey’s equal-protection claim must fail because there was no showing of purposeful discriminatory intent. It shrank from recognizing McCleskey’s claim because “taken to its logical conclusion, [it] throws into serious question the principles that underlie our entire criminal justice system.” Id. at 314-15, 107 S.Ct. at 1779, 95 L.Ed.2d at 293. The Court feared that if it accepted McCleskey’s claim that racial bias impermissibly tainted the capital-sentencing decision, it “could soon be faced with similar claims as to other types of penalty.” Id. at 315, 107 S.Ct. at 1779, 95 L.Ed.2d at 293. Carrying the parade of horribles to its extreme, the Court posited that it would have to study any set of arbitrary variables such as the defendant’s physical characteristics or those of the victim that “some statistical study indicates may be influential in jury decisionmaking.” Id. at 317-18, 107 S.Ct. at 1780-81, 95 L.Ed.2d at 295.
This Court cannot refuse to confront those terrible realities. We have committed ourselves to determining whether racial and ethnic bias exist in our judicial system and to “recommend ways of eliminating it wherever it is found.” Interim Report, supra, at i. Hence, were we to believe that the race of the victim and race of the defendant played a significant part in capital-sentencing decisions in New Jersey, we would seek corrective measures, and if that failed we could not, consistent with our State’s policy, tolerate discrimination that threatened the foundation of our system of law.
*210We do not find, however, in this case, evidence of constitutionally-significant race-based disparities in sentencing. By way of contrast, the statistics presented to the Court in McCleskey were startlingly disturbing. Justice Brennan’s dissenting opinion recited the statistical data that the majority accepted as sound for purposes of its analysis. The unadjusted data (for mitigating or aggravating effect of other factors) showed that the capital-sentencing rate for all (at all culpability levels) white-victim cases was almost “11 times greater than the rate for black-victim cases. * * * [B]lacks who kill[ed] whites [were] sentenced to death at nearly 22 times the rate of blacks who kill[ed] blacks, and more than 7 times the rate of whites who kill[ed] blacks.” 481 U.S. at 326-27, 107 S.Ct. at 1785, 95 L.Ed.2d at 300-01. In addition, prosecutors were shown to “seek the death penalty for 70% of black defendants with white victims, but for only 15% of black defendants with black victims, and only 19% of white defendants with black victims.” Id. at 327, 107 S.Ct. at 1785, 95 L.Ed.2d at 301.
A more extensive set of relationships between the statistical variables was presented to the McCleskey Court. Here, the only statistics presented by the Master were the rate at which black defendants are sentenced to death and the rate at which cases with white victims proceed to a penalty trial.
With respect to the race of defendant, the most critical assertion made by the Public Defender, derived from an extrapolation of the Master’s figures, is that
[a]t the mid range of aggravation level, culpability Level 4, there is a 64-percentage-point higher risk that a black defendant will be sentenced to death than any other defendant * * * [and] at culpability Level 3, no non-black defendant has been sentenced to death, although 3 of 10 black defendants received death sentences.
The Public Defender emphasizes that point because racial effects seem undetectable at the highest culpability levels and at the lowest culpability levels. Hence, he asserts that when juries are faced with real choices, racial effects are striking.
*211Although the race-of-defendant disparities appeared in the pr e-Gerald decisions, “[i]n the post-Gerald data, there were too few penalty-trial death-verdict cases involving black and non-black defendants with comparable levels of culpability to support any finding at all.” Final Report, supra, at 103.
With respect to the race-of-victim analysis, the principal concern is the way in which cases advance to penalty trial. Again, extrapolating from the various culpability levels, the Public Defender suggests that “at the mid-range culpability level[ ] [,] * * * Level 4, white victim cases are 1.4 (.78/.56) times more likely to advance to a penalty trial than other cases [and] [a]t culpability Level 3, white victim cases are 3.4 (.67/.20) times more likely to advance to penalty trial than other cases.” The Master reported that “on average, cases with a white victim may have a 14 percentage point or higher risk of advancing to a penalty trial than do other cases.” Ibid. In this instance, his post-Gerald data actually demonstrated a slightly stronger race-of-victim effect than the pre-Gerald period data. Id. at 113-14.
The Master recognizes, however, that those results are not conclusive. He reports that
[b]ecause discrimination was not the primary mandate in this project, we consider these results to be strictly preliminary. More work will be required to determine if they persist under closer scrutiny and alternative analyses, to determine, for example, whether they are statistical artifacts or flukes, and to assess their legal and practical significance.
[Final Report, supra, at 101.]
We have but little knowledge of the science of statistical probabilities. We have had occasion to study that science in connection with the jury selection process in Ramseur, supra, 106 N.J. at 221, 524 A.2d 188. There we explained how statisticians determine the statistical significance of data to measure the likelihood that “aspects of [a] selection process do not operate randomly” and how courts may use the techniques of *212statistical analysis to correct inequality in the process. Ibid. We gave the example of how the
jury selection process in which two groups are being compared can be likened to filling a box with a population of 1,000 slips of paper of which 600 are pink and 400 gray, and having someone randomly select a sample of 100 slips. The expected number of pink slips would be 60 and the expected number of gray slips would be 40. * * * However, a statistician would not be surprised if the number of pink slips “deviated” from the expected. Statisticians measure this deviation by a formula that enables them to tell whether the result is so far from the expected as to demonstrate that the result was not random [or by chance].
[Ibid.]
A complex mathematical formula determines that standard deviation. In the case of the 100 slip sample a statistician expects that the standard deviation would be plus or minus 4.8 slips, and if a result is “more than 2 or 3 standard deviations from the expected,” a statistician assumes it is suspect. Id. at 222, 524 A.2d 188.
If a court concludes that the statistical evidence is so deviant as to compel a conclusion of substantial significance, the court must then look to the circumstances surrounding that statistical showing to determine its full constitutional import. The constitutional importance of the statistical showing depends in part on the degree of subjectivity involved in the selection mechanism. The more discretionary the selection process, the more concern for bias. In addition, courts consider the time period over which violations are alleged to have occurred, and, finally, courts will look at the State’s efforts to deal with the problem of potential bias. Id. at 224-26, 524 A.2d 188.
Although those principles are not clearly applicable to the circumstances of this case, they do give us guidance. We do not have a definitive report that the deviations are sufficiently alarming to compel a conclusion of substantial discriminatory effect in the application of the New Jersey Capital Punishment Act. Despite the apparent race-of-defendant disparities at the mid-range culpability level, those same data show no race-of-*213victim effects in the penalty-trial decisions.12 Final Report, supra, at 101 n. 109. And, most importantly, the Master concluded that “the race of victim effects we have observed in these data for the * * * cases advancing to penalty trial are less stable than the effects we observe for the race of defendant variable in the penalty trial decisions.” Final Report, supra, at 104 n. 114.. In short, we do not yet confront a record in which “[t]he statistical evidence * * * relentlessly documents the risk that [Marshall’s] sentence was influenced by racial considerations.” McCleskey, supra, 481 U.S. at 328, 107 S.Ct. at 1786, 95 L.Ed.2d at 301 (Brennan, J., dissenting). If that were so, we would not hesitate to invalidate the sentence of death.
Before oral argument, we denied the Public Defender’s motion to remand these issues for further factual findings. There has been no showing that would warrant further consideration of this issue.
Although, then, we must be concerned by the disparate rate at which white-victim cases have moved toward capital disposition, we can balance that concern with the fact that the system discloses no significant race-of-victim effects in penalty decisions. Concerning race-of-defendant disparities, we need to examine over time whether those effects are systemic. We balance that concern with the observation drawn from the same data that since 1987 there “were too few penalty-trial death-verdict cases involving black and non-black defendants with comparable levels of culpability to support any finding at all.” Final Report, supra, at 103. Whether due to the more-clearly-defined requirements for the death penalty, e.g., an objective *214standard for the c(4)(c) torture/depravity factor, Ramseur, supra, 106 N.J. at 211, 524 A.2d 188; the requirement of a showing of an intent to kill, Gerald, supra, 113 N.J. at 69, 549 A.2d 792; the requirement that aggravating factors be found unanimously beyond a reasonable doubt to outweigh non-unanimous findings of mitigating factors, State v. Biegenwald, 106 N.J. 13, 53, 524 A.2d 130; or to the State’s own efforts to channel and guide discretion, see, e.g., Koedatich, supra, 112 N.J. 225, 548 A.2d 939, “[t]he principal trend in New Jersey’s capital charging and sentencing system between 1983 and 1991 has been a marked decline in the frequency with which death sentences are imposed among death-eligible cases. * * * 1987 marks the dividing line.” Final Report, supra, at 15. Although the Master does not attribute the decline to the interpretative changes, the data suggest that the legislative and judicial efforts to achieve the required “categorical narrowing” of death-eligible cases apparently have not limited the effectiveness of the State’s Capital Punishment Act, but rather, have strengthened the constitutionality of the Act by providing critically-needed guidance at all decisional stages. Decreasing discretion within the system decreases the likelihood that the determination of who shall live and who shall die will be arbitrary and capricious and the likelihood that death would be considered a cruel and unusual punishment in violation of the Eighth Amendment.
We have no doubt that the people of New Jersey would not tolerate a system that condones disparate treatment for black and white defendants or a system that would debase the value of a black victim’s life. Whether in the exercise of statutory proportionality review or our constitutional duty to assure the equal protection and due process of law, we cannot escape the responsibility to review any effects of race in capital sentencing. See N.J. Const. art. VI, § 5, par. 1(c); State v. Koedatich, 98 N.J. 553, 489 A.2d 659 (1984) (finding that court *215must review death sentence even when defendant wishes not to appeal).
We have encouraged the Attorney General, as the chief law-enforcement officer of the State of New Jersey, to exercise his undoubted authority to instill uniformity in charging and prosecuting practices throughout the state. We realize the differing attitudes in counties and the jurisdictional concerns. If the system fails to eliminate unconstitutional disparities, that failure should not be because of hesitancy to invoke authority currently conferred on the Attorney General. A statewide review panel, appointed by the Attorney General in consultation with prosecutors, would be able to screen out any possible effects of race or socioeconomic status in the charging and selection process. Certain of the statutory factors, e.g., prior murder, contract killing, or multiple factors, obviously make a case more deathworthy. It does not seem burdensome to articulate such reasoning. In part IX, infra at 216, 613 A.2d at 1113, of this opinion we outline a process by which to resolve any differences between the Attorney General and Public Defender concerning the reliability of either the data-gathering process or the specifics of a particular case in the universe.
Although this generation ought not bear the burden of the past, we seem unable to rid ourselves of its vestiges of racial oppression. As Justice Brennan so eloquently stated in his dissent in McCleskey:
In more recent times, we have sought to free ourselves from the burden of this history. Yet it has been scarcely a generation since this Court’s first decision striking down racial segregation, and barely two decades since the legislative prohibition of racial discrimination in major domains of national life. These have been honorable steps, but we cannot pretend that in three decades we have completely escaped the grip of a historical legacy spanning centuries.
[T]he Court[] fear[s] that McCleskey’s claim is an invitation to descend a slippery slope * * * [believing] that any humanly imposed system of penalties will exhibit some imperfection. Yet to reject * * * evidence [of racial bias] on this basis is to ignore both the qualitatively different character of the death penalty and the particular repugnance of racial discrimination * * *.
[481 U.S. at 339, 344, 107 S.Ct. at 1792, 1794, 95 L.Ed.2d at 309, 312.]
*216IX
Revisions That Can be Made Now and in the Future to Simplify the Data-Gathering Process
The Master has compiled the data base to assist the Court in its constitutional and statutory functions of death-sentence review. We must take steps to assure that it is kept current and accessible to the parties and the Court. Obviously, the extent of future maintenance of the data base must await the adjudication of parties’ rights under L.1992, c. 5 (limiting statutory proportionality review to death-sentenced cases). Still, the data base is relevant to evaluate the system-wide claims of constitutional dimension, as well as potentially the claims of defendants whose offenses were committed before the Act’s effective date.
In his Final Report, supra, at 107, the Master reflected on the project’s first years of experience. The original proposal contemplated a collaborative effort with full participation by defense and prosecution. Initially, he expected that the studies of cases would be enhanced or corrected by input from the parties, particularly trial counsel. Input from trial counsel has varied greatly, however. In particular, the prosecutors have offered no assistance, even with death cases. The Attorney General provided appellate briefs, when available and requested, as well as jury verdicts for both death-sentenced cases and a residual group of life-sentenced penalty-trial cases. That the process can benefit greatly from a continuing collaborative process is self-evident.
The Master has made certain recommendations to the Court with respect to data-collection entry and retrieval and a possible continuing role for the AOC. We need not outline in detail those recommendations at this time.
*217We have no way of knowing how long the Master will continue to assist the Court in the data-gathering process. Any system we maintain should be able to operate efficiently even after his departure, and, indeed, after our departure. Hence, because the Attorney General and his consultant have questioned certain of the data-gathering methods employed by the Master, we shall appoint a Superior Court judge familiar with the relevant statistical and data-assembly concepts to oversee development of a recommended plan for future maintenance of the Court’s records. That judge will entertain applications from any of the parties to correct any perceived inaccuracies or deficiencies in the record-gathering system. For example, questions about the quantum of premeditation (is five minutes enough?) and even clerical errors should be readily resolved. In addition, the data base may need revisions to accommodate our belief that the culpability models should be as objective as possible, i.e., rooted in traditional sentencing considerations.
The pool of clearly death-eligible cases is the pool of cases that the Court must consider for constitutional review of impermissible bias. See Ramseur, supra, 106 N.J. at 186 & n. 18, 524 A.2d 188 (one feature that sustains constitutionality of Capital Punishment Act is mandatory “appellate review” that is “meaningful”). We expect that once the parties’ adversarial positions have been resolved, the Attorney General and prosecutors will cooperate with the Court and its representatives in maintaining the data. Given society’s many needs and the demands placed on all branches of government, to resolve many of the questions about accuracies in the data base by a telephone call rather than by an after-the-fact analysis of the transcripts and available documents will be far simpler.
In a collaborative setting, the parties may be able to develop a more efficient and cost-effective system as well. As noted in our discussion of geographic and racial disparity, much of the unease about those issues could be readily resolved if there were some type of inter-agency review to provide the most rudimentary monitoring of the capital-charging decisions. In *218addition, the parties may wish to ask us to implement the type of reporting (or some variation of it) submitted to the Supreme Court of Delaware. See Flamer, supra, 490 A.2d 104. We gather that presentence reports with respect to the non-capital charges are regularly made available now to the proportionality-review office. With any necessary mid-course corrections and refinements suggested to us by the monitoring judge or the parties, we are confident that to implement the record-keeping process that is necessary to assure the constitutional and judicial integrity of capital sentencing on review will be both fair and feasible.
X
CONCLUSION
The extent to which the parties’ arguments will have to be reconsidered under L.1992, c. 5 must await the eventuality of a case arising under that law. We have attempted, however, to make our judgment under the law that applies to this case as explicit as possible so that it could be tested against whatever objective measurements are available and applicable. Once defined, however, the process of proportionality review will not “frustrate and confuse the Court,” as the State believes. Weisberg Report, supra, at 39. Although the process is thorough, the concepts are not overly difficult: use the information at our disposal to identify similar cases in terms of blameworthiness; examine the prosecutorial and jury patterns for frequency or of frequency of death verdicts in the similar cases; and, finally, look at the comparison cases to identify disproportionality.
The debate between the parties under the prior law has focused on the size of the envelope containing the comparison cases, not what the envelope contains or how to withdraw the information. We share some of the concerns of our concurring and dissenting member, Justice Garibaldi, about the reliability of the data base. Post at 224, 613 A.2d at 1117. For that reason, we have established in Part IX of this opinion, supra, at *219216-218, 613 A.2d at 1113-1114, a procedure to test and resolve any doubts about the specifics of any case categorization as death eligible or deathworthy. Regardless of which method of cataloguing the cases we use, whether index card or computerized data base, we can never dispense with the obligation to assure that the burden of the past does not create a genuine risk that defendants will be sentenced to death either because of their race or the race of the victim. To repeat the words of our Legislature: “[Racial] discrimination threatens not only the rights and proper privileges of the inhabitants of the State but menaces the institutions and foundation of a free democratic State.” N.J.S.A. 10:5-3.
And, finally, we cannot await a more certain state of proof before we render judgment. We must confront in this case the problem of the small sample of comparison cases. We do not know what will happen in the next ten similar cases, but we must nevertheless make our judgment now. Were the next ten similar cases to result in death and were we to declare this penalty disproportionate, we would have made a mistake; and if the next ten similar cases were to result in life and we were to affirm the death penalty, we would have made an even greater mistake.
Proportionality review recognizes in that sense that society’s standards may change and that which is proportionate in one era may be disproportionate in another. There are no absolutes. Ours is a finite role defined by our obligation to see that justice is done at a given time. We have had to make some assumptions. For example, we have included for statistical purposes in our analysis death verdicts such as Clausell and DiFrisco that have been reversed. We recognize the argument made in Justice Handler’s dissent, post at 253-258, 613 A.2d at 1132-1134, that there is a logical step in that process that may be scientifically unjustified. We continue, nonetheless, to believe that such jury verdicts provide sufficiently-reliable information concerning the characteristics that prosecutors and juries consider important to warrant the inclusion of those cases *220in proportionality analysis. Although we have had a fairly large sampling in our universe (246 death-eligible cases), the system itself is new. Society is entitled to a longer experience before one can say that it cannot apply proportionally the punishment that it has legislated.
Our dissenting member, Justice Handler, sees in those uncertainties “the disquieting truth that capital punishment cannot really be made to work in a civilized society.” Marshall I, supra, 123 N.J. at 266, 586 A.2d 85. He sees the death of one so sentenced as diminishing our common humanity. Yet, both proponents and opponents of the death penalty invoke the same ideal of justice. The opponents believe that we diminish ourselves when the State, representing organized society, takes a life without having established a system of certain predictability. Proponents believe that we diminish our common humanity as well when we do not respect the principle of just retribution for the taking of an innocent murder victim’s life.
Even the most profound search for an essence of law must confront the reality of human experience. “Hence justice is traditionally thought of as maintaining or restoring a balance or proportion, and its leading precept is often formulated as ‘Treat like cases alike’; though we need to add to the latter ‘and treat different cases differently.’ ” H.L.A. Hart, The Concept of Law 155 (1961).
In the best of all worlds there would be no doubt about like cases being treated alike. From defendant’s point of view, all that he knows is that he is the only one whose .death sentence we have affirmed. But when our measuring system counts two cases as the same because both have the same mitigating factor — c(5)(h), “[a]ny other factor which is relevant to the defendant’s character or record,” it allows for a characterization of sameness that may conceal the greatest differences between the two cases.
The United States Constitution requires that juries be permitted to consider any aspect of the defendant’s charac*221ter; therefore, to some extent, that each case be considered as a different case is not only inevitable but almost constitutionally required. Statutory proportionality review could not have been intended to invalidate that which the United States Constitution requires.
The sentence of death is affirmed.
N.J.S.A. 2C:11-3 contains the New Jersey Criminal Code’s murder provisions. The death-penalty provisions are found in subsections c to e of that section. When referring to those subsections we shall, for instance, use c(4)(e) to refer to N.J.S.A. 2C:11-3c(4)(e).
A Washington statute defines the universe of cases to include all " 'cases reported in the Washington Reports or Washington Appellate Reports since January 1, 1965, in which the judge or jury considered the imposition of capital punishment regardless of whether it was imposed or executed,’ ” and cases that had been reported under a statutory proceeding of "aggravated first-degree murder.” Jeffries, supra, 717 P.2d at 742 (Utter, J., dissenting) (quoting RCW 10.95.130(2)(b)).
For convenience, we have supplied the code to the aggravating factors:
A. prior murder
B. grave risk of death to another
C. torture, depravity
D. killed for money
E. hired a killer
F. escaping detection
G. in commission of rape, robbery, etc.
H. murdered a public servant
Although the penalty-retrial conviction and sentence of Marko Bey have been affirmed in a judgment filed this day, 129 N.J. 557, 610 A.2d 814 (1992), during the pendency of this case Marshall was the only defendant whose sentence had been affirmed. In addition to Bey, an appeal by John Martini awaits decision by this Court. Several other penalty-trial cases are pending before the Law Division.
The Master includes State v. Clausell, 121 N.J. 298, 580 A.2d 221 twice: death on the first trial and life on the second. Rather than rework the figures, we present the Master’s figures. We recognize that the death sentences of DiFrisco and Clausell have been reversed; however, we will not disregard those cases. In Clausell, the State elected to accept the life sentence because of the Gerald issue, rather than retry the case. In DiFrisco, there was an absence of a finding of corroboration by the sentencing judge. The Court does not believe that those procedural infirmities undermine the findings of deathworthiness. We acknowledge that there is no certainty that either case would or will produce a death verdict on a retrial. We have been candid to acknowledge that there is no scientific infallibility in the frequency data that we cite. We know only what we know — that of the 246 death-eligible cases there were thirty-two death verdicts. Three of the thirty-two (including Marshall) involved contract killings.
As noted, we have attempted to limit our analysis as much as possible to objective criteria. Hence, we state these reranked results not to justify our decision but only to report them.
We have considered the other contract-principal cases set forth in the Public Defender’s brief. Lazaro Trimino hired Miguel Melendez to murder the victim. Trimino claimed to have served as the intermediary between the actual principal, Pedro Gerome, who is still a fugitive, presumably in Nicaragua. Although the reports do not show a clear motive for the killing, the relationship between the absent procurer of the murder and the victim undoubtedly would establish the motive. Trimino pled guilty to conspiracy.
We have also considered the remaining two principals suggested by the Public Defender: Zoran Cveticanin, the principal in State v. Rose, and Anthony Franciotti, the principal in State v. DiFrisco. The latter hired Anthony DiFrisco to murder the victim, a pizzeria owner. We do not know if the police have apprehended him. Zoran Cveticanin hired Michael Rose to murder Cveticanin’s step-mother, who was eight-months pregnant, to prevent her from inheriting his father’s money. Cveticanin fled to Yugoslavia, where he was tried and *183convicted of murder and sentenced to thirty-years hard labor. We find sufficient distinction in those cases to rule out a finding of disproportion.
John Martini’s death-penalty appeal is currently pending before this Court.
We make reference to that dissent because it is based on a differing view of the evidence of rejection of the death penalty, not on a differing view of the concepts of Eighth Amendment jurisprudence.
The Public Defender argues that even those figures are suspect. He suggests that after removing from the thirty-two death verdicts those cases that do not meet the statutory criteria, e.g., Marie Moore, Bey I, we should remove, as well, all reversals that have not resulted in retrial-death verdicts. Of those thirty-two vacated death verdicts, only four have gone to a remand-penalty trial, and in only two were the defendants sentenced to death: We believe, however, as does the Master, that the original penalty trials, although reversed for various reasons, most often for the burden-of-proof and Gerald issues, have reflected juror values of deathworthiness in terms of deterrent effect.
The court points out that the correct spelling of the defendant's name is ‘Spenkelink.” 578 F.2d at 582 n. 1.
At the intermediate range in McCleskey, death was imposed in "34% of the white-victim crimes and 14% of the black-victim crimes." 481 U.S. at 325, 107 S.Ct. at 1784-85, 95 L.Ed.2d at 300 (Brennan, J., dissenting). This represents approximately a 2.4 times greater imposition of the death penalty in white-victim cases.