dissenting.
In October 1986, on a retrial, Nathaniel Harvey was convicted by a jury for the murder of Irene Schnaps. Schnaps was found alone in her apartment, having been struck on the head several times with a blunt instrument — killed in an apparent burglary. There were no signs of forced entry and no signs of a struggle in the bedroom where she was discovered. The jury found defendant guilty of purposeful-or-knowing murder, felony murder, first-degree robbery, and second-degree burglary.
At the penalty phase, the jurors determined that the State had proven beyond a reasonable doubt two statutory aggravating factors: N.J.S.A 2C:11-3e(4)(f) (murder committed to escape apprehension for another offense) and N.J.S.A. 2C:11-3c(4)(g) (murder committed during the course of a robbery and burglary). The jury did not find as an aggravating factor the State’s submission that the murder involved aggravated assault of the victim, N.J.SA 2C:11-3c(4)(c). Several of the jury members found some of the ten non-statutory mitigating factors presented by defendant *352pursuant to N.J.S.A. 2C:11-3c(5)(h), the “catch-all” mitigating factor. See ante at 312, 731 A.2d at 1139. Finding that the aggravating factors outweighed the mitigating factors, the jury-sentenced defendant to death for the capital charges. The trial court sentenced defendant to an aggregate sentence of life plus sixty-five years with a fifty-seven and one-half-year parole disqualifier for the non-capital counts of first-degree robbery and second-degree burglary. The Court affirmed defendant’s convictions and death sentence. State v. Harvey, 151 N.J. 117, 233, 699 A.2d 596 (1997) (Harvey II).
This appeal is based on defendant’s request for a proportionality review. In State v. Loftin, 157 N.J. 253, 724 A.2d 129 (1999) (Loftin II), the Court appointed a Special Master to evaluate its proportionality review methodology and make recommendations for improvements.1 The Court then proceeded to apply the existing methodology to Loftin’s case, stating, “Until we have had the benefit of [the Special Master’s] report, ... we will continue ... to carry out proportionality review as before.” Id. at 266, 724 A.2d 129.
The Special Master released his report on April 28, 1999. The Honorable David S. Baime, Report to the New Jersey Supreme Court: Proportionality Review Project (Apr. 28, 1999) (Special Master Report). In that report, the Special Master determined that several aspects of our methodology are faulty and require revision. Id. at 6-7. Despite the Special Master’s recommendations, the Court goes ahead with defendant Harvey’s case, applying existing methodology. This course of action, in addition to its needless inefficiency and unfairness, can only further confuse and undermine the accuracy and integrity of our proportionality review.
*353The Court, in proceeding with this review, makes it pointless to delve into the Special Master Report in detail here. Suffice it to say that the proposed revisions would — I predict, will — have a significant impact on defendant’s proportionality review. The Court’s decision to proceed with defendant’s - review when oral arguments on a new methodology are scheduled to occur the week of this decision’s filing, see ante at 300, 731 A.2d at 1133, does a grave disservice to both defendant and this Court’s commitment to justice.
The Court holds that defendant’s sentence is not disproportionate. See ante at 283, 731 A.2d at 1124. First, the Court has previously decided not to apply the Legislature’s 1992 capital murder statute amendment, which severely limits the universe of cases constituting a basis for comparison among defendants, until the appointed Special Master reviewed the validity of such a limitation. See Loftin II, supra, 157 N.J. at 265-66, 724 A.2d 129. In this case, even though the Special Master has spoken on the issue, the majority again defers judgment on the constitutionality of the statutory amendment. See ante at 288 - 289, 731 A.2d at 1126 - 27. Further, the Court holds that defendant has not relentlessly documented with adequate evidence that racial discrimination influences this State’s imposition of the death penalty. See ante at 319 - 20, 731 A.2d at 1143 - 44. Finally, the Court affirms the proportionality of defendant’s death sentence, holding that when defendant is compared to other similarly situated death-eligible defendants, defendant’s sentence is not disproportionate. See ante at 319, 731 A.2d at 1143.
I disagree with the Court’s holdings regarding systemic issues in the application of the New Jersey capital murder statute. First, I reiterate that consideration of the constitutionality of the 1992 amendment should not be postponed. Accord Loftin II, supra, 157 N.J. at 373, 724 A.2d 129 (Handler, J., dissenting). This Court has expressed a strong commitment to meaningful proportionality review and has firmly rejected the possibility that *354a universe limited to cases in which the death penalty has been imposed could form the basis for such review. See State v. Marshall, 130 N.J. 109, 137, 613 A.2d 1059 (1992) (Marshall II). In light of the Special Master’s strong statement that “a universe limited to cases in which the death sentence was imposed cannot support a coherent proportionality review system,” Special Master Report, supra, at 10, the Court should act at this time by declaring the 1992 amendment unconstitutional.
Second, I believe the statistical evidence before the Court, already presented in Loftin II, supra, demonstrates a constitutionally impermissible risk that race discrimination infects our State’s imposition of the death penalty. This risk is especially great in transracial cases like this one, involving black defendants and white victims. Given the startling evidence of race discrimination before us, the Court should declare the death penalty statute unconstitutional or, at the very least, place a moratorium on the imposition of the death penalty until such time as the evidence demonstrates that race is not playing a role in capital prosecuting and sentencing. Accord Loftin II, supra, 157 N.J. at 446, 724 A.2d 129 (Handler, J., dissenting).
With regard to defendant’s individual proportionality review, I object to the majority’s novel decision to limit the class of cases to which defendant is compared in precedent-seeking review to his salient-factors subcategory (E-l), a step that renders the Court’s proportionality review incomplete. Further, I find the Court’s statistical analyses and its precedent-seeking review extremely subjective, arbitrary, and ultimately unreliable. I strongly disagree with the Court’s conclusion and find defendant’s sentence to be disproportionate.
I, therefore, dissent.
I
Proportionality review has always been an integral and indispensable part of this Court’s review of capital sentences. It is meant “to ensure that the death penalty is being administered in a *355rational, non-arbitrary, and evenhanded manner, fairly and with reasonable consistency.” Loftin II, supra, 157 N.J. at 275, 724 A.2d 129 (quoting Marshall II, supra, 130 N.J. at 131, 613 A.2d 1059); see State v. Ramseur, 106 N.J. 123, 327, 524 A.2d 188 (1987); Gregg v. Georgia, 428 U.S. 153, 96 S.Ct. 2909, 49 L.Ed.2d 859 (1976). Our emphasis on the dual tasks of ensuring uniformity and accounting for the individual circumstances of each defendant’s crime, see Ramseur, supra, 106 N.J. at 330-31, 524 A.2d 188, has been, in my view, irreparably hampered by our inability to reconcile these two aims.
Proportionality review consists of a two-pronged approach: statistical frequency analysis and precedent-seeking review. Both of these methods, however, suffer from deficiencies. See Loftin II, supra, 157 N.J. at 413-15, 724 A.2d 129 (Handler, J., dissenting) (citing State v. DiFrisco, 142 N.J. 148, 212, 245, 662 A.2d 442 (1995) (DiFrisco III) (Handler, J., dissenting)); State v. Martini 139 N.J. 3, 81-82, 651 A.2d 949 (1994) (Martini II) (Handler, J., dissenting) (discussing flaws in universe of cases used for comparison, failure to use workable standard by which to judge general imposition of death penalty, and inherently subjective nature of precedent-seeking review).
. I now reiterate the most problematic of these deficiencies, namely, (1) the lack of an objective, numerical standard for frequency analysis determinations of when a death sentence is “generally” imposed; (2) deficiencies in the data base of comparison cases; and (3) the subjective nature of precedent-seeking review.
A.
The Court persists in refusing to attempt to quantify its standard for determining what constitutes a “generally” imposed sentence when examining frequency analysis measurements of disproportionality. See Martini II, supra, 139 N.J. at 20, 651 A.2d 949 (“A capital sentence is excessive and thus disproportionate if other defendants with characteristics similar to those of the *356defendant under review generally receive sentences other than death for committing factually-similar crimes in the same jurisdiction.”). This failure to bring some objectivity and consistency to its determinations renders meaningful proportionality review an illusory goal. See Loftin II, supra, 157 N.J. at 415, 724 A.2d 129 (Handler, J., dissenting) (stating that Court’s standard of review is inherently subjective, as evidenced by the majority’s shifting principles for proportionality determinations).
Not only is the purely subjective standard of “general imposition” difficult to pinpoint for any particular ease, it is only one of a number of substantively different, yet equally imprecise standards this Court has attempted to apply, often within the same case. The Court thus attempts the impossible: to ascertain “proportionality” without a stable benchmark or measure. The Court has, in fact, since its first proportionality review, invoked fourteen different amorphous standards and substandards when discussing the test for determining proportionality: (1) “ ‘The principal inquiry [in frequency analysis] is whether the degree of blameworthiness in the present case reasonably supports an expectation that such a case will generally result in a death sentence.’ ” Ante at 299, 731 A.2d at 1132 (quoting DiFrisco III, supra, 142 N.J. at 171, 662 A.2d 442) (internal quotes and citation omitted); (2) “ ‘A capital sentence is excessive and thus disproportionate if other defendants with characteristics similar to those of the defendant under review generally receive sentences other than death for committing factually similar offenses in the same .jurisdiction.’ ” Ante at 289, 731 A.2d at 1127 (quoting Martini II, supra, 139 N.J. at 20, 651 A.2d 949 (citing State v. Bey, 137 N.J. 334, 351, 645 A.2d 685 (1994) (Bey IV) (other citation omitted)); (3) “[W]e seek to determine the existence of any aberration in defendant’s sentencing.” Ante at 308, 731 A.2d at 1137 (citing Bey IV, supra, 137 N.J. at 369, 645 A.2d 685); see also ante at 307, 731 A.2d at 1137 (“[W]e do not believe that evidences that [defendant’s] sentence is an aberration.”); (4) “ ‘Frequency analysis helps us to determine whether defendant is in a category that renders him or her more likely than other killers to receive the death penalty.’ ” Ante at 299, 731 *357A.2d at 1132 (quoting DiFrisco III, supra, 142 N.J. at 171, 662 A.2d 442); (5) “If the death sentence is imposed often enough in a category of comparable cases, then we feel confident that there exists a societal consensus that death is the appropriate punishment.” DiFrisco III, supra, 142 N.J. at 172, 662 A.2d 442; (6) “[T]he figures for E-l defendants are higher than the overall death-sentencing rates----” Ante at 301, 731 A.2d at 1133; (7) “[D]efendant’s numbers are not the lowest score of a defendant whose claim of disproportionality we have denied.” Ante at 306, 731 A.2d at 1136; (8) “[D]efendant’s ‘results produce no showing of randomness____Ante at 307, 731 A.2d at 1137 (quoting DiFrisco III, supra, 142 N.J. at 183, 662 A.2d 442) (citations omitted); (9) “[Proportionality] review ... is [ ] a vehicle to ensure that the penalty-phase jury’s decision is not insupportable.” Martini II, supra, 139 N.J. at 22, 651 A.2d 949, (10) “[0]ur goal is to determine whether the jury’s decision to sentence a defendant to death is comparable to decisions reached in the appropriate capital cases in our universe of cases.” Ibid.; (11) “[T]his salient-factors analysis reveals that a significant portion of defendants in the pecuniary-motive category ... have received the death penalty.” DiFrisco III, supra, 142 N.J. at 174, 662 A.2d 442, (12) “The index-of-outcomes results in defendant’s case' are not consistently higher or lower than results generated for prior eases----” Loftin II, supra, 157 N.J. at 334, 724 A.2d 129, (13) disproportionality is measured by looking at whether defendant’s culpability is “more like that of defendants who received death sentences or of those who received life terms.” Martini II, supra, 139 N.J. at 50, 651 A.2d 949; see DiFrisco III, supra, 142 N.J. at 186, 662 A.2d 442; and (14) “As in Marshall II and Bey IV, “we do not [here] find a pattern of life sentencing or the taint of an invidious factor that would require us to reverse [defendant’s] death sentence.’ ” Loftin II, supra, 157 N.J. at 345, 724 A.2d 129 (quoting Bey IV, supra, 137 N.J. at 386, 645 A.2d 685).
*358The Court must acknowledge that its “generally imposed” standard, already inherently subjective and unquantifiable, has evolved into a range of equally vague and immeasurable, but substantively different, standards, rendering proportionality review unworkable. The two standards employed by the Court in this case could not be more different: when one declares that a death sentence is disproportionate if similarly situated defendants “generally” receive a sentence other than death, one imposes a very different standard than when one seeks to determine if such a sentence is “random” or “aberrational.” The Court’s errant application of these inconsistent standards only highlights their susceptibility to manipulation.
I reiterate my contention that the Court should use a numerical preponderance standard to determine when a death sentence is proportionate. Accord Loftin II, supra, 157 N.J. at 419, 724 A.2d 129 (Handler, J., dissenting). ‘Without the doctrinal discipline and certainty of a defined numerical majority consistent with the concept of generality to demonstrate proportionate sentences, it is inevitable that arbitrary death sentences of individual defendants will be lost in the verbiage that rationalizes a death sentence as proportionate.” Ibid. Verbal thresholds may work in other areas of the law, but they cannot be determinative when death is the outcome.
B.
Second, there are deficiencies in the database used as a basis for comparison to defendant’s case that undermine the Court’s proportionality review. This case involves data presented previously by defendant Loftin and incorporated into the database of the Administrative Office of the Courts (AOC). See Chew, Cooper, Harvey Report (Dec.1997) (CCH Report). Previously indicated problems with the database still exist: “[it] remains too small to provide a reliable review,” Loftin II, supra, 157 N.J. at 414, 724 A.2d 129 (Handler, J., dissenting); the universe of cases still erroneously includes death sentences that were reversed, id. at *359416 n. 16, 724 A.2d 129 (Handler, J., dissenting) (citations omitted); and the Court continues to erroneously include in its analysis data including defendant’s case, see id. at 420 n. 19, 724 A.2d 129 (Handler, J., dissenting) (“[P]lacing defendant’s case on both sides of the comparison, gauging the proportionality of a defendant’s sentence by comparing it to a group of which he is a member, skews the analysis.”).
C.
The standards circumscribed by the Court for frequency review are not particularly suited to its precedent-seeking review. The Court’s precedent-seeking analysis suffers from a different deficiency: by putting the defendant under a microscope the Court employs a review so individualized that it becomes impossible to gather a class of cases that can be compared with the defendant’s and against which defendant’s sentence can be measured. Id. at 419, 724 A.2d 129 (Handler, J., dissenting).
Defendant’s ease starkly demonstrates this problem: Harvey is the only defendant in his subcategory who currently faces the death penalty. Yet, the Court finds his sentence is proportionate, in spite of the fact that his crime is not among the most egregious in the group. The Court fails to recognize the necessity of drawing a connection between its quantitative frequency analysis and the more quality-driven precedent-seeking review in order to reduce the subjectivity in our application of the latter. If the salient-factors test results indicate that cases like defendant’s yield death sentences only ten percent of the time,2 the precedent-seeking review should then be used to determine if defendant is among the one or two most blameworthy in his category of 20 *360defendants. Without linking the two analyses, the salient-factors test fails to inform the Court at all: the test tells us that the death penalty is not generally imposed for defendant’s category; if we fail to conclude from this that defendant’s sentence is disproportionate, the test loses all value unless the results are then used to serve as a backdrop for our precedent-seeking review.
In addition, the AOC-designated comparison subcategories, used by the Court to assess defendant’s disproportionality claim, do not always include all relevant cases. The Court defers to the AOC’s judgment in defining the comparison groups, ante at 294, 731 A.2d at 1130, failing to recognize that these groupings can be overly limiting; it is crucial that we look beyond them in precedent-seeking review in order to gain as much insight as possible into a defendant’s sentence.
D.
Given these problems, and the availability of the recommendations for improvement by the Court’s appointed Special Master, I object to the application of our current methodology to defendant’s case. In light of the Court’s concession that the system of review requires evaluation and perhaps reconfiguration, one must assume that any death sentence reviewed and found proportionate by existing methods, including Harvey’s, will be reconsidered in the event that the Court adopts some or all of the Special Master’s suggestions. The only just alternative is to postpone all proportionality reviews until such time as the Court has had the opportunity to consider and evaluate the Special Master’s findings towards the creation of a more sound method by which to judge the proportionality of a defendant’s death sentence. It is grossly unfair, profoundly misleading, and patently prejudicial to determine, upon faulty methodology, that defendant’s death sentence is proportionate. By doing so, the Court moves the defendant further down the path toward execution; simultaneously it erects another high and unfair hurdle — the need to have the Court reconsider and then reverse this determination of proportionality. *361To put defendant through such an obstacle course to save his own life, and secure a just sentencing result offends fundamental fairness.
II
Because the Court has applied our system of proportionality review to defendant’s case, I have attempted to determine the proportionality of his sentence. See infra at 374 - 415, 731 A.2d at 1173 - 96. There is, however, an issue that should be addressed first that is even graver and more importunate than defendant’s individual proportionality review. That compelling issue is posed by recurring evidence of a persistent risk of systemic race discrimination in the administration of the death penalty, particularly in the prosecution of cases such as this one and Loftin II, transracial murders in which the defendant is black and the victim is white.
The majority finds that defendant has failed to prove that New Jersey’s death penalty statute is implemented in a racially discriminatory manner. See ante at 319 - 20, 731 A.2d at 1143 - 44. We are faced here with the same data compiled in the AOC’s CCH Report, supra, presented by defendant Donald Loftin. Given certain methodological concerns with the statistical models employed to assess whether or not racism plays a role in this State’s prosecution and sentencing of capital cases, the Court in Loftin II asked the appointed Special Master to evaluate and, if necessary, redevelop the statistical methodologies or models to more accurately accomplish this important intended purpose of proportionality review. Loftin II, supra, 157 N.J. at 456, 724 A.2d 129. At the same time, however, the Court applied the existing models to defendant Loftin’s case in spite of their identified flaws. In doing so, the majority discounted as unreliable the overwhelmingly consistent and persistent results derived from numerous models that have indicated a race effect in our implementation of capital punishment.
*362Here, again, in spite of the addition of the Special Master’s findings that the development of more parsimonious models must be pursued if we are to be resolved in our long-held commitment to eliminating racism in our justice system, Special Master Report, supra, at 108, the Court applies our existing system of proportionality review to the case at hand, finding, as in Loftin II, that the statistics do not indicate an unconstitutional risk of race discrimination. Ante at 319 - 20, 731 A.2d at 1143 - 44.
I deeply disagree. I oppose the application of our existing system of proportionality review to Harvey’s sentence given that the methodology is currently under review. Second, I find that the evidence of racism demonstrated by the data, when supplemented by what we know through human experience and common knowledge, to be so compelling that it demonstrates an unconstitutional risk that New Jersey’s sentencing scheme singles out minority defendants (particularly those guilty of killing white victims) for the death penalty. The presence of this risk requires us to invalidate the death penalty statute.
A.
Defendant presents statistics highlighting a race effect in New Jersey’s prosecution and sentencing of death-eligible defendants. The raw data alone reveal worrisome, indeed ominous, trends: of the death-eligible black defendants who killed nonblack victims, over forty-eight percent (3/te) were charged capitally, while none of the six nonblack defendants who killed black victims proceeded to a penalty trial. Loftin II, supra, 157 N.J. at 383, 724 A.2d 129 (Handler, J., dissenting). Over forty-eight percent (10hi) of cases involving nonblack victims were capitally tried, while only twenty-eight percent (4%u) of black-victim cases were. Discrepancies in jury sentencing are startling as well: eighteen percent of all death-eligible black-defendant, nonblack victim cases (%) resulted in death sentences, while none of the six nonblack-defendant, black-victim cases did. Ibid.
*363The logistic regressions demonstrate the same very real potential that race discrimination is at work in our capital sentencing scheme, revealing for the first time a statistically significant race effect. In the three AOC models used to measure prosecutorial and jury sentencing decisions from the CCH Report, supra, a statistically significant black-defendant effect is reflected in the first and second models (Schedules 2 and 5), and a statistically significant white-victim effect is reflected in the first and third (Schedules 2 and 8). Id. at 379-80, 724 A.2d 129 (Handler, J., dissenting). These statistics are more alarming in light of the fact that the AOC’s culpability ratings indicate that white defendants are, on average, more culpable than black defendants (compare 1.43 and 1.32, respectively). Id. at 383, 724 A.2d 129 (Handler, J., dissenting).
The AOC multiregression analyses once again confirm the indications that racism pervades the capital system. “Differences in percentages of black and nonblack defendants sentenced to death are as high as thirty-three percent when the culpability levels are divided into five categories with equal ranges ... and thirty-six percent when the culpability levels are divided into levels with equal numbers of cases.” Id. at 386, 724 A.2d 129 (Handler, J., dissenting) (citations omitted).
The statistical models created by Dr. John Tukey, the acknowledged expert retained as a consultant by Special Master Richard S. Cohen, attempted to correct for what the Special Master found to be methodological flaws in the AOC models.3 These new models also revealed a race effect: the third model demonstrates a statistically significant black-defendant effect on the likelihood that a defendant will be sentenced to death. The second model, when reconfigured by the Public Defender’s statistician to include *364the white victim variable, also reveals a statistically significant black-defendant effect. Id. at 389, 724 A.2d 129 (Handler, J., dissenting).
The results in model two after the Public Defender’s modification highlight the effect that the race of the victim seems to have on sentencing outcomes, in particular in cases involving minority defendants, or transracial, white-victim cases. The disparity in capital prosecutions between transracial cases involving black defendants and those involving white defendants is forty-nine percent, and the disparity in capital sentencing (as a percentage of all death-eligible cases) is eighteen percent. Id. at 383, 724 A.2d 129 (Handler, J., dissenting).
While the number of cases in defendant’s AOC-assigned offense subcategory, E-l, is an inadequate sample from which to draw conclusions (nor should this category alone be used to do so in a systemic race discrimination claim), the prosecutorial and sentencing results in this group of cases underscore what the numbers on a system-wide scale seem to indicate about the intensified attention paid to cases involving white victims and, more specifically, those with minority defendants. The E-l category consists of twenty-two cases (including both of defendant’s), eighteen of which involve minority defendants (black or Hispanic) and four which involve white defendants. Twelve of the cases involve minority victims and ten involve white victims. Over seventy-eight percent of the cases (%) that were either not capitally prosecuted or that resulted in plea agreements involve minority victims, in spite of the fact that minority-victim cases comprise only about fifty-four percent of the total number of cases. At the same time, almost eighty-eight percent of those cases that were capitally tried without resulting in a plea agreement © involve white .victims, in spite of the fact that white-victim eases comprise less than fifty-six percent of the total — a disturbing suggestion of some form of prosecutorial profiling. When the race of the defendant is factored in as well, we see that five of the six transracial cases (all *365involving minority defendants and white victims) were capitally prosecuted.4 Further, three of the four death sentences resulting from capital prosecutions (Walter Gerald’s and both of defendant Harvey’s) were for transracial cases in spite of the fact that these cases comprise only twenty-seven percent of the total number of cases. Finally, all four of the cases resulting in a death sentence involve minority defendants.
While, admittedly, there are methodological flaws in the models created by Dr. Tukey, the Public Defender and the AOC, the Court’s decision to apply them to defendants Loftin and Harvey implies, at least, a level of confidence in the methodologies employed without the benefit of the Special Master’s recommendations. Even without perfected models, the Court has compelling raw data findings recounted herein to which it can refer. Its conclusion, therefore, that the numbers do not make out an adequate case for a risk of race discrimination, given the startling consistency throughout the data and across the models indicating a race effect, belies and betrays the Court’s long-professed position that it “could not, consistent with our State’s policy, tolerate discrimination that threatened the foundation of our system of law.” Marshall II, supra, 130 N.J. at 209, 613 A.2d 1059 (rejecting Supreme Court’s statement in McCleskey v. Kemp, 481 U.S. 279, 312, 107 S.Ct. 1756, 1778, 95 L.Ed.2d 262, 291 (1987), that “disparities in sentencing are an inevitable part of our criminal justice system.”).
*366B.
This Court has always paid scrupulous attention to minority-defendant cases, fully acknowledging the need to provide specific protections to ensure that racism does not infect the sentencing process. In the death penalty arena, we must be especially careful not to ignore signs, both statistical and social, that racism may be at work. From the outset, in this State’s first death penalty ease under the new death penalty statute, the Court intoned against the dangers of racism:
We are sensitive to the reality of racial prejudice, and to the possibility that jurors may prejudge a defendant because of his or her race, even in the absence of an interracial crime. Racial prejudice may operate, for instance, when the defendant is black simply because the defendant is black and regardless of the victim’s color. We must be particularly sensitive to this possibility in a capital case. As the Supreme Court has recognized, “[bjecause of the range of discretion entrusted to a jury in a capital sentencing hearing, there is a unique opportunity for racial prejudice to operate but remain undetected."
[State v. Ramseur, supra, 106 N.J. at 247, 524 A.2d 188 (quoting Turner v. Murray, supra, 476 U.S. 28, 35, 106 S.Ct 1683, 1687, 90 L.E.2d, 27, 35 (1986)).]
Our remedy in Ramseur was to require enhanced protections in death penalty trials in the form of a thorough voir dire, examining the potential racial biases of the jury pool members in cases involving minority defendants or other issues of race. Id. at 246, 524 A.2d 188; see also State v. Williams, 113 N.J. 393, 428, 550 A.2d 1172 (1988) (Williams II) (“Racial prejudice may be either blatant and easy to detect or subtle and therefore more difficult to discern. A probing voir dire that elicits more than a “yes” or “no” response will aid the trial court in excusing prospective jurors for cause and will assist the defense in exercising its peremptory challenges. When the defendant is a member of a cognizable minority group, a more searching voir dire should be conducted, if requested.”).
The sharply disproportionate representation of transracial, white-victim cases among those that are capitally prosecuted and result in death sentences seems to reflect a particular societal value placed on white life and a concurrent degradation of minority life. The Court has specifically noted that when the crime is *367interracial, a more thorough voir dire should be conducted by the trial court. Ramseur, supra, 106 N.J. at 245-46, 524 A.2d 188; see also State v. Harris, 156 N.J. 122, 237, 716 A.2d 458 (1998) (Handler, J., dissenting) (stressing need to delve into both blatant and latent racial bias, particularly in interracial cases and, even more specifically, in one' involving a white suburban woman and inner-city, black, male defendant). To this remedy may be added a specific jury instruction admonishing against racial bias to be given in all such cases. See Special Master Report, supra, at 7.
The Court’s early acknowledgment that protections need to be established to guard against racism bespeaks an understanding that racism is virulent, though not always obvious, and, more crucially, that it is so likely to infect a jury pool that when requested, a potentially time-consuming voir dire must be conducted to try to uncover it. Given that when we decided Ramseur, we had no data from which to gather statistical evidence that race might be a factor in jurors’ sentencing of capital defendants, this solution seemed the only viable one for addressing the risk of racial bias that might occur in an individual case.' But we must now acknowledge that this remedy is not efficacious to weed out jurors who may be subject to inarticulable and subconscious racial biases. In addition, the remedy cannot, and was never intended to, address the biases of prosecutors. See Marshall II, supra, 130 N.J. at 144, 613 A.2d 1059 (“[W]e believe ... 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____”).
Because the stakes are so high in death penalty cases, we must be willing to supplement attempts to eradicate racism, such as vigorous voir dires and clear curative jury instructions. These should be combined with our continued efforts to create methodologically sound statistical models and simplified and realistic readings of the resultant data informed by what we know about human nature and have learned from history, see Loftin II, supra, 157 N.J. at 405-10, 724 A.2d 129 (Handler, J., dissenting) (recounting *368history of racially discriminatory laws in New Jersey and around the United States, as well as discriminatory enforcement of racially neutral laws). This comprehensive approach is vital if we are to avoid the risk that individuals will be singled out for death simply because of their race.
Our perception, beginning in Ra/mseur, that racism is likely to take hold without adequate voir dire, has now been augmented and sharpened in ways that strongly suggest that our curative solution for warding off the inevitable has not sufficed. Governor Whitman recently acknowledged that some State Troopers engage in racial profiling on the New Jersey Turnpike — that is, they single out black and Hispanic drivers based on ostensible traffic violations and subject them to criminal searches. See Iver Peterson, Whitman Says Troopers Used Racial Profiling, N.Y. Times, Apr. 21, 1999, at A1, B8. Are we to assume that racism begins and ends with the New Jersey State Troopers? We must acknowledge that the risk of prosecutorial and jury-based racism is supported by the numbers we have before us and its documentation in all sectors of our society, and that our attempts to keep racial biases out of our capital sentencing scheme may well have failed.
In prior cases, in areas of law bearing significantly less risk of injustice than in life and death decisions, we have been willing to look beyond the numbers when they are inadequate to give us conclusive proof of a causal relationship. Most recently, this Court in State v. Cromedy, 158 N.J. 112, 727 A.2d 457 (1999), allowed for a flexible standard of knowledge in determining the usefulness of social science evidence as a basis for requiring specific jury instructions in transracial crimes. The Court held that a jury instruction explaining the potential unreliability of transracial identifications was appropriate in a case in which a black defendant was tried for rape based solely on a white victim’s identification of him. In spite of the prosecutor’s presentation of evidence that some researchers do not subscribe to the idea that *369cross-racial impairment affects real-life identifications, the Court held that
consistent with [various cases]; the Task Force Report; and our review of the professional literature of the behavioral and social sciences, we hold that a cross-racial identification ... requires a special jury instruction in an appropriate case---- We conclude that the empirical data encapsulate much of the ordinary human experience and provide an appropriate frame of reference for requiring [such an] instruction.
[Id. at 131-32, 727 A.2d 457.]
We must not be reluctant to follow this path in an arena where the stakes are significantly higher — where the ultimate, irreversible penalty is implicated. See Loftin II, supra, 157 N.J. at 399, 724 A.2d 129 (Handler, J., dissenting) (“Nowhere in the law is more at stake or is there a greater need for positing a definitive resolution on a sound and understandable basis than in a capital case, even if that resolution errs by falling on the side of fairness rather than accuracy.”) (citing Woodson v. North Carolina, 428 U.S. 280, 303-04, 96 S.Ct. 2978, 2991, 49 L.Ed.2d 944, 961 (1976).)
The Court continues to insist that scientific certainty — statistical significance — should be required before it will accept the notion that racism plays a role in our capital sentencing scheme. Even as it demands, however, that we must be ninety-five percent sure that race plays a role in death sentencing before it will consider such death sentences to be tainted by racial bias, the Court allows liability to be imposed on tortfeasors in toxic-tort cases when the causal relationship between the alleged harmful conduct and the plaintiffs injury is much less clear. See Rubanick v. Witco Chemical Corp., 125 N.J. 421, 434, 593 A.2d 733 (1991) (Because “plaintiffs in toxic-tort litigation, despite strong and indeed compelling indicators that they have been previously harmed by toxic exposure, may never recover if required to await general acceptance by the scientific community óf a reasonable, but as of yet uncertain theory of causation,” strict scientific standards may be relaxed.); see also Loftin II, supra, 157 N.J. at 400, 724 A.2d 129 (Handler, J., dissenting) (citing Rubanick, supra). In juxtaposition, the Court’s diametrically opposing views leave it in an irrational and nonsensical posture. The widely accepted standard of ninety-five percent certainty may be appropriate in the scientific community of statisticians, but we should be willing to adjust *370and modify the norms of that community when our focus is a decidedly non-scientific one and when insistence on near-certainty opens the doors of injustice.
Other disciplines question the value of inflexible and arbitrary line-drawing and the use of science to the exclusion of commonsense observations in circumstances less threatening than death. They stress the need to look beyond the hard sciences to nonquantiflable factors when evaluating certain relationships.
Economists, for example, are often faced with the almost impossible task of evaluating the likely effects of policy on the economy. Nobel Laureate Robert M. Solow believes that when science is inadequate to make the necessary evaluations, we must broaden our tools of reference for identifying and explaining causal relationships:
[Solow] argue[s] against thinking of economics as science with a capital S. ‘That is perfectly consistent,’ he wr[ites], *with a strong belief that economics should try very hard to be scientific with a small s. By that I mean only that we should think logically and respect fact.’ Fact ... should be enlarged ‘to include, say, the opinions and casual generalizations of experts and market participants, attitudinal surveys, institutional regularities, even our judgments of plausibility. My preferred image is the vacuum cleaner, not the microscope.’
[Louis Uchitelle, A Challenge to Scientific Economics, N.Y. Times, Jan. 23, 1999, at B7, B9.]
Even in the hard sciences, blind adoption of the scientific certainty threshold has been challenged. Robert J. Levine, Ethics and Regulation of Clinical Research 200-01 (2d ed. 1986) (‘We have chosen arbitrarily to say that something is true when probability is less than 0.05 that it could have occurred by chance ____”) (emphasis added). For example, some have challenged the notion that in order to preserve the value of medical trials, researchers should not disclose the suspected potential benefits of one treatment over another to trial participants until there is statistical' certainty (ninety-five percent) that one treatment is superior to the others. See, e.g., ibid, (suggesting that participants be able to choose a less exacting level of certainty in deciding whether clinical trials can be concluded).
*371Here, the Court itself must decide whether the values of the scientific community ought to be employed when the failure to scientifically pinpoint a causal relationship that may be at work results in the unfair execution of an individual. Statistical significance cannot displace all knowledge, or override basic understanding, or be dispositive in all contexts. Mahesh K.B. Parman & David Machín, Survival Analysis, A Practical Approach 15 (1995) (stating that results may be “clinically” significant even though not statistically significant); Michael D. Maltz; Deviating From the Mean: The Declining Significance of Significance, 31 Journal of Research in Crime and Delinquency 434, 440 (Nov.1994) (“Statistical significance does not imply substantive significance, and most researchers know this — but this does not stop them from implying that it does.”).
The Court’s failure to look not only beyond statistical significance — which even scientists agree is an arbitrary cut-off point— but also beyond the statistical results themselves, has the capacity to result in the gravest of injustices.
Th[e] ideal of mechanical objectivity, knowledge based completely .on explicit rules, is never fully attainable. Even with regard to purely scientific matters, the importance of tacit knowledge is widely recognized. In efforts to solve problems posed from outside the scientific community, informed intuition is all the more crucial.
[Theodore M. Porter, Trust in the Numbers: The Pursuit of Objectivity in Science and Public Life 7 (1995).]
This Court should no longer wait for the optimum statistical model or the statistically ideal number of cases. It has, in fact, rejected such a rigid, mathematical approach to proportionality review in developing its methodology for analyzing frequency analysis results in a defendant’s individual claim:
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.
*372[.Marshall II, supra, 130 N.J. at 153, 613 A.2d 1059 (quoting David C. Baldus, Death Penalty Proportionality Review Project: Final RepoH to the New Jersey Supreme Court, 1,42-13 (Sept. 24,1991).)] 5
The crucial question here is not whether we are certain racism plays a role in capital sentencing, but whether there is “a constitutionally significant risk of racial bias affecting the ... capital sentencing process.” McCleskey v. Kemp, supra, 481 U.S. at 313, 107 S.Ct. at 1778, 95 L.Ed.2d at 292. The Special Master himself acknowledges that “[i]t is entirely possible that our efforts will come to naught because the problem [of identifying the role of race discrimination, if any,] may be beyond the reach of the social sciences.” Special Master Report, supra, at 108-09. Given that the various models indicate a serious risk of race discrimination, we are adjured to consider all sources of potentially relevant knowledge and information.
We should look to other jurisdictions to determine whether these indications have been duplicated elsewhere. We are justified in doing so because defendant is not charged with proving that he alone was the victim of race discrimination, or even that racism on a systemic level is definitely at work. To determine if the diverse and numerous statistical indications of race discrimination form the basis for an unconstitutional risk, it is perfectly appropriate to expand our inquiry to examine the findings of other states given the limitations on our universe. See Note, Easing the Fear of Too Much Justice: A Compromise Proposal to Rinse the *373Racial Justice Act, 30 Harv. C.R.-C.L. L.Rev. 543, 564 (1995) (“Because the [Racial Justice Act] requires statistical significance, it will have absolutely no effect on jurisdictions where the number of death sentences is so small as to preclude any statistically reliable conclusions.”).
That inquiry is instructive. A 1990 Government Accounting Office report, based on the examination of twenty-eight state-specific studies on the role of race in death penalty sentencing, reveals alarming consistency across states in racial disparities. In eighty-two percent of the studies, the race of the victim influenced charging and sentencing patterns in capital cases (i.e., white-victim cases were more likely to result in death sentences). Further, more than three-fourths of the studies that identified a race-of-the-defendant effect found that black defendants were more likely to receive the death penalty than white defendants. United. States General Accounting Office, Death Penalty Sentencing: Research Indicates Pattern of Racial Disparities (Feb.1990), reprinted in 136 Cong. Rec. S6889-90 (daily ed. May 24, 1990).
The Nebraska Legislature recently passed a bill imposing a two-year moratorium on executions until further study on the possible role that the race and/or economic status of defendants and victims is playing in the state’s capital sentencing process.6 Dirk Johnson, Legislature of Nebraska Votes Pause in Executions, N.Y. Times, May 21, 1999, at A14. Republican Senator Kermit A. Brashear, while still unwilling to say racism has taken hold in Nebraska, notes that “there is clearly a racial component and a socioeconomic component nationwide” in death sentencing. Ibid. Black defendants, who account for thirteen percent of the country’s population, occupy forty-two percent of its death row cells. Ibid.
*374We must finally acknowledge that the consistent results of our statistical models are not an aberration, nor can they be explained away by rejecting not only the models themselves, but the common experience of a significant majority of other states that impose capital punishment. We must act now before we execute our first capital defendant under a legislative scheme that has likely been infected with race discrimination.
Ill
The Court’s decision to review the proportionality of defendant’s death sentence and its determination that the sentence is not disproportionate impels, again, an assessment of the soundness of proportionality review and the validity of defendant’s death sentence. I believe that application of our current methodology, employing only a “generally imposed” standard (to the extent possible) coupled with precedent-seeking review, reveals that defendant’s sentence is indeed disproportionate and should be vacated.
A.
The first step in proportionality review is frequency analysis. Here, we attempt, through statistical representation, to examine how often defendants who have committed offenses similar to that of defendant are sentenced to death. The Court recognizes the limitations of a statistically-based approach to determining proportionality and thus emphasizes that the statistics are not determinative. ‘“[T]he lower the overall rates and the reliability of our frequency analysis, the greater the need for precedent-seeking review.’ ” Ante at 307, 731 A.2d at 1137 (quoting DiFrisco III, supra, 142 N.J. at 183-84, 662 A.2d 442). Results indicating disproportionality “require us to place greater emphasis on precedent-seeking review.” Ante at 308, 731 A.2d at 1137.
*3751. Salient-Factors Test
Since the Court’s rejection of the numerical-preponderance test, see Loftin II, supra, 157 N.J. at 295, 724 A.2d 129, only two tests remain, the salient-factors and index-of-outcomes tests.
The majority concludes that the salient-factors test does not establish disproportionality. See ante at 303, 731 A.2d at 1134 (“[T]he mere fact that a statistical disparity exists does -not establish disproportionality.”)(citing Bey IV, supra, 137 N.J. at 352, 645 A.2d 685). The Court’s methodology in doing so, however, is faulty in two significant ways.
a.
For the first time, the Court relies only on the defendant’s subcategory, E-l,7 for its analysis of salient factors. See ante at 295 - 96, 731 A.2d at 1130 - 31. The Court’s decision to so drastically limit the universe of comparison eases has significant consequences for the accuracy and fairness of our proportionality review. The Special Master, in fact, recognizes the problem in maintaining the subcategories at all for purposes of statistical analysis:
Our reading of the 433 death-eligible eases indicates that many of the subeategories contained in the current [salient-factors] model lack resonance in terms of factual comparability and sentencing outcome____ While the subcategories are descriptive in terms of the particular circumstances surrounding the crime, our statistics reveal that they have no relevance in determining death outcome____
[Special Master Report, supra, at 56-57.]
The Special Master recommends that the subcategories be eliminated due to their inability to predict deathworthiness, id. at 57, adding that specific changes to defendant’s E category should be made.8 These changes would clearly have a significant impact on the Court’s salient-factors analysis, and while I withhold judgment *376on whether the Special Master’s remedies are appropriate, his finding that the subcategories are not meaningful must be heeded.
In all the proportionality reviews conducted by the Court to date, the composite salient-factors category in which the defendant is placed has formed the basis for comparison cases (the broader E category in Harvey’s case). The Court set the standard for determining how large the universe of comparison cases should be in its first proportionality review, Marshall II, supra, stating that cases in which “there are no striking factual dissimilarities between [them] and [defendant’s]” should form the basis for the precedent-seeking review comparisons. 130 N.J. at 181, 613 A.2d 1059; see also State v. Cooper, 159 N.J. 55, 72-74, 731 A.2d 1000 (1999) (Cooper II) (comparing defendant to all C-category defendants); State v. Chew, 159 N.J. 183, 214, 731 A.2d 1070 (1999) (Chew II) (comparing defendant to all I-category defendants); Loftin II, supra, 157 N.J. at 326-27, 724 A.2d 129 (comparing defendant to all B category defendants); DiFrisco III, supra, 142 N.J. at 186, 662 A.2d 442 (“the relevant factor ... is the statutory factor that [defendant] committed the murder for a pecuniary motive,” the I category); Martini II, supra, 139 N.J. at 51, 651 A.2d 949 (comparing defendant to cases involving “kidnapping of non-strangers with particular violence or terror, kidnapping of strangers with particular violence or terror, contract-murder principals, contract killers, and other non-robbery pecuniary-advantage killers”); Bey IV, supra, 137 N.J. at 368, 645 A.2d 685 (comparing defendant to all other defendants with prior murder conviction). The Court’s drastic step to limit its salient-factors review to the E-l subcategory (and, as a result, its precedent-seeking review, see infra at 302 - 03, 731 A.2d at 1134-35), is insupportable in light of our established methodology for frequency review and the Special Master’s findings.
The majority explains its decision to limit its comparison of cases in two ways. First, it contends that in prior proportionality reviews our comparison class has only been extended to a defen*377dant’s composite category in order to make up for small sample sizes in the subcategories. See ante at 295 - 97, 731 A.2d at 1130 - 31 (stating that in Chew II, Court’s comparison extended beyond Chew’s 1-3 subcategory because group consisted of only one other defendant; that in DiFrisco III defendant was compared to entire I category to provide for “a productive statistical analysis;” and that in Loftin II defendant was compared to all B-category defendants because of “ ‘exceedingly small number of eases’ ” in his B-l subcategory).
The Court misconstrues our method for selecting a comparison group, as is demonstrated by its glaring failure to note the approach taken in the companion case Cooper, supra, 159 N.J. 55, 731 A.2d 1000. In Cooper II, the Court selected the entire composite C category for comparison not on the basis of the size of defendant’s C-l subcategory, which contained an ample forty cases, but because of the essential similarities between the defendant’s crime and those of other defendants. In spite of the fact that Cooper did not even specifically request that the C-3 cases be compared to his, the Court stated, ‘We previously have performed the salient-factors test using both the assigned subcategory as well as the composite category ... and we will do so in this appeal.” Id. at 76, 731 A.2d 1000 (citation omitted). The Court proceeded to compare defendant Cooper not only to all C category defendants, but also to two cases proposed by the defendant from other categories. See id. at 92, 731 A.2d 1000. The Court’s focus was on the fact that Cooper’s murder was accompanied by a sexual assault, the defining feature of the C category.
Further, the Court here mischaracterizes the approach to defining the universe of comparison eases taken in DiFrisco III, supra, where the rationale for defining the group of comparison cases was similar to that in Cooper II. Although it is indeed correct that DiFrisco’s 1-1 subcategory contained too few eases (nine) to provide for a meaningful proportionality review alone, see ante at 297, 731 A.2d at 1131, the Court did not expand the universe of *378comparison cases to the composite category solely because of the deficiency in the size of the subcategory. In DiFrisco III, the Court examined not only the I-category defendants, but also the possibility of comparing defendant to cases beyond the composite I category, despite the adequate sample size contained in DiFrisco’s composite group (fourteen). Although the Court ultimately rejected defendant’s recommendations for expanding the comparison, it did so not because it had reached an adequate number of cases to provide for a “statistically productive review,” but rather because the Court found that the proposed cases were not factually similar enough to defendant’s to warrant their inclusion. See DiFrisco III, supra, 142 N.J. at 168, 662 A.2d 442 (excluding proposed defendants from pecuniary gain category because “there appears to be no basis for an allegation that any of those defendants were either paid to commit murder or that they paid another to do so....”).
The Court’s second reason for excluding E-2 and E-3 cases from its proportionality review is that because of the differences that distinguish the defendants in the E-2, E-3 and G-3 cases from Harvey, “such cases provide little insight into the propriety of the jury’s decision in this case, and are inapplicable to our proportionality review.” Ante at 299, 731 A.2d at 1132. The Court’s examination of the cases, however, is both misguided and incomplete.
First, the Court bases its restriction of the comparison cases to defendant’s subcategory on the fact that the other E-subcategory cases are distinguishable by the number of aggravating factors in their case, the mental problems of the defendants, the reduced age of the defendants, the prior records of the defendants, the level of remorse they demonstrated or their level of intoxication when they committed their crimes. See ante at 297 - 99, 731 A.2d at 1131-32. These kinds of comparisons, however, are the very heart of our precedent-seeking review, and should be made in the context of examining the actual crimes of the defendants. The initial basis for determining the class of comparison cases should *379be premised solely on the categorical groupings made by the AOC, which are defined not by any of the above characteristics, but rather, as previously stated, by the essential elements of the defendant’s offense. In this case, the essential characteristic of the E category is that the defendant murdered in the course of a robbery. The cases are then broken down into the E-1, E-2, and E-3 subcategories based on the particular violence or terror employed in the robbery and/or whether or not the robbery involved a forced entry. The Court, therefore, can justify its decision to distinguish Harvey from the E-2 and E-3 subcategory defendants for purposes of selecting the comparison group only by determining first that these defendants have been correctly placed in their subcategories, see Cooper II, supra, 159 N.J. at 76-77, 731 A.2d 1000 (examining four defendants whom the State contended were improperly included in defendant’s category and excluding two of them based on State’s arguments); and second, that the distinctions between the subcategories are meaningful for purposes of our proportionality review, see Chew II, supra, 159 N.J. at 253-54, 731 A.2d 1070 (Handler, J., dissenting) (arguing that I-3, “other pecuniary motive” defendants are more similar to some robbery-category defendants than to 1-2 and 1-3 contract killer defendants). The Court has clearly failed to meet this burden.
Second, the Court discusses only five of the seven E-2 and E-3 defendants and only one of the four proposed A- and G-category defendants, see ante at 297 - 99, 731 A.2d at 1131-32. Notably, the Court fails to distinguish Jesus DeJesus (E-2) and Larry Durden (E-3) from defendant Harvey. Not only are these defendants’ crimes factually similar to Harvey’s, the measure by which we ought to be determining our universe of comparison cases, see supra at 313, 731 A.2d at 1140, the defendants themselves are in many respects indistinguishable from Harvey. Neither DeJesus nor Durden were young enough to claim their age diminished their culpability (DeJesus was thirty-one when he committed his murder and Durden was forty-four); neither had a drug problem; neither suffered from emotional or mental problems; and both *380had prior criminal histories. The Court’s failure to address why these defendants should be excluded from the comparison group, given that they pass even the Court’s misplaced test for determining the level of similarity between proposed cases and defendant’s, makes one question the Court’s basis for exclusion, and, indeed, if one exists.
The Court’s reasons for limiting the comparison class here are based on selective analysis of precedent and utter disregard for the AOC’s method of defining the categories, greatly compromising the completeness of our proportionality review.
b.
Next, the Court’s salient-factors methodology is suspect due to the reliance on a comparison of the death-sentencing rate for the defendant’s E-1 subcategory to the average rate for death-eligible cases, see ante at 301 - 03, 731 A.2d at 1133 - 35; see also DiFrisco III, supra, 142 N.J. at 173, 662 A.2d 442 (stating that because death sentencing rates of contract killers were significantly higher than the average rates, contract killers are “viewed by society as ‘significantly blameworthy.’”) (citation omitted). This approach renders the salient-factors test superfluous.
I believe the Court misinterprets the logical purpose of the salient-factors test. It is designed to classify defendants in subcategories purporting to measure similar characteristics of the crimes committed. Using the average death-sentencing rate as the benchmark of proportionality would defeat the purpose of dividing the death-eligible cases into subcategories at all. By comparing death rates in each category to the average death rates, the Court might as well be comparing defendant to all other death-eligible defendants who were not sentenced to death. Instead, the Court ought simply to be using the test results to assess defendant’s culpability according to the actual death-sentencing rates in his assigned categories. Accord Cooper II, 159 N.J. at 169, 731 A.2d 1000 (Handler, J., dissenting).
*381Even if one accepts the Court’s use of average rates as á basis of comparison to examine the disproportionality of a sentence, the majority’s analysis is problematic. The Court chooses to apply selectively a relatively quantifiable standard, established in Di-Frisco III, supra, when comparing the death sentencing rates within categories to the average rate: if the death sentencing rate of defendant’s category is higher than the overall average, this leads “to the conclusion that society views those, who commit particular violence or terror in a residential forced entry as significantly blameworthy.” Ante at 301, 731 A.2d at 1133-34 (citing DiFrisco III, supra, 142 N.J. at 173, 662 A.2d 442). From this, one can also conclude that if the sentencing rate of defendant’s category is lower than the overall average, society views these defendants as not particularly blameworthy when compared to other defendants. In spite of the finding, however, that E-l defendants are sentenced to death only ten percent of the time, two percent lower than the average death sentencing rate for all death-eligible defendants, the Court concludes, “Such a deviation [does not show] that robbery killings are viewed by society as less blameworthy,” ante at 303, 731 A.2d at 1134, thereby implying that the salient-factors test does not indicate a disproportionate sentence.
This duplicitous application of the one standard that the Court has developed that is even remotely concrete calls into question the Court’s desire to craft any meaningful, consistent standard in proportionality review. The statistics show that only ten percent of death-eligible defendants in the E-l sub-category and thirty-three percent of those advancing to the penalty trial are sentenced to death. CCH Report, supra, tbl 7. Even more compelling, when the E category as a whole is examined, the death-sentencing rate for death-eligible defendants is only six percent, and for those proceeding to a penalty trial only twenty-four percent. Ibid. Both of these figures are below the average rates of sentencing for all death-eligible defendants. These numbers, in my opinion, are adequate for a finding that such cases will “‘generally receive sentences other than death____’” See ante at 289, 731 A.2d at *3821127 (quoting Martini II, supra, 139 N.J. at 20, 651 A.2d 949) (other citations omitted).
2. Index-of-Outcomes Test
The index-of-outcomes test is a different attempt to examine the blameworthiness of the defendant: the categories are created not according to similarities between the crimes, but rather similarities between the defendants. Characteristics meant to measure the defendant’s culpability, both statutory and non-statutory, are examined regardless of whether the crimes themselves are similar in nature. Each factor is weighted and assigned a coefficient" according to how often a death sentence is imposed when the factor is present. Then each case is assigned, a culpability score based on the factors present and their weighted coefficients.
As the majority correctly points out, there is a persistently wide range in culpability scores assigned to defendants across the four regression models. See ante at 303 - 04, 731 A.2d at 1135 (noting defendant’s culpability scores here range from thirteen to forty-three percent; DiFrisco’s culpability scores ranged from eleven to seventy-four percent; Martini’s culpability scores ranged from five to eighty-eight percent; Bey’s culpability scores ranged from twenty-five to seventy-six percent; and Marshall’s culpability scores ranged from seventeen to fifty-two percent). This wide range renders the accuracy of the models highly questionable.9 See Loftin II, supra, 157 N.J. at 422 n. 22, 724 A.2d 129 (Handler, J., dissenting).
In addition, the wide range of confidence intervals among the cases detracts significantly from the reliability of the test.10 See ante at 305 - 06, 731 A.2d at 1135 (detailing results of defendant’s four regression models with confidence intervals displaying, at *383their greatest, fifty-seven percent range between the upper and lower levels for predicted probability of death); accord Loftin II, supra, 157 N.J. at 422 n. 21, 724 A.2d 129 (Handler, J., dissenting).
Beyond the flaws in the test itself, the majority misinterprets its results by relying not on death sentencing rates for the various culpability levels as indicators of proportionality, but on defendant’s predicted probability of receiving a death sentence. See ante at 305 - 06, 731 A.2d at 1135 - 36. Predicted probabilities are used to place the defendant in a culpability level and are not meant to provide a basis for a finding of proportionality.
[T]he Constitutional mandate of individualized consideration in capital sentencing requires that we rely on actual decisions by sentencing juries, and the characteristics of actual felons' and their crimes, to determine when a death sentence is generally imposed upon similarly situated defendants. Certainly, we could not point to our surprise at defendant’s death sentence as grounds for reversal; nor should our expectation that a certain defendant will receive the death penalty serve as adequate support for a finding of proportionality.
[Loftin II, supra, 157 N.J. at 423, 724 A.2d 129 (Handler, J., dissenting) (emphasis added).]
Our focus, therefore, should be on sentencing rates for the levels of culpability in which the defendant has been placed for each model (as it is in the salient-factors test for each crime category).11
The death sentencing rates for the culpability levels to which defendant was assigned in the four models are thirty-five percent (culpability level 2), thirteen percent (culpability level 1), forty-. three percent (culpability level 3), and nineteen percent (culpability level 1). Not one of these rates indicates general imposition of *384the death penalty and two of the four strongly indicate that, by any standard, defendant’s sentence is disproportionate. This conclusion is supported by the fact that defendant’s confidence intervals are much lower than those of other defendants. His upper limit for all four models never exceeds sixty-nine percent, while all other defendants examined by the Court for proportionality reviews have at least one confidence interval that reaches ninety-four percent. This suggests that defendant’s low numbers are more stable predictions than those of the other defendants.
Under the old methodology, both the salient-factors and the index-of-outcomes tests, in my view, suggest that defendant’s sentence might be disproportionate. Accordingly, the Court should apply precedent-seeking review in only one way: unless defendant is one. of the most culpable defendants in his category, see supra at 291, 731 A.2d at 1128, unless we can point to some justification for the break in the pattern of life sentences, see Marshall II, supra, 130 N.J. at 181, 613 A.2d 1059, we must find Harvey’s sentence to be disproportionate. Although these statistical tests have their limitations, they can serve as a useful kaleidoscope through which to examine precedent-seeking review, which is designed to help us identify possible justifications for a defendant’s seemingly-arbitrary sentence.
B.
Precedent-seeking review is the Court’s attempt to engage in an individualized approach to proportionality review by making factual comparisons between death-eligible defendants and their crimes. Because precedent-seeking review is not statistical in nature, it is inevitably more qualitative and less quantitative than the salient-factors and index-of-outcomes determinations. Loftin II, supra, 157 N.J. at 425, 724 A.2d 129 (Handler, J., dissenting). Precedent-seeking review is meant to complement frequency analysis, and the Court has defined this functional relationship as follows: “[T]he lower the overall rates and the reliability of our frequency analysis, the greater the need for precedent-seeking *385review.” DiFrisco III, supra, 142 N.J. at 183, 662 A.2d 442 (citation omitted). Because defendant’s frequency analysis rates are quite low in comparison to those of other defendants, precedent-seeking review must take on added significance in this case. See Marshall II, supra, 130 N.J. at 154, 613 A.2d 1059 (“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.”).12
In addition to statutory aggravating and mitigating factors, precedent-seeking review takes into account non-statutory factors “rooted in traditional sentencing guidelines,” Marshall II, supra, 130 N.J. at 159, 613 A.2d 1059 (citation omitted), which fall within three categories: defendant’s moral blameworthiness, the degree of victimization resulting from defendant’s crime, and defendant’s character. See ante at 309, 731 A.2d at 1138 (citing Marshall II, supra, 130 N.J. at 155, 613 A.2d 1059).
Defendant’s case demonstrates as much or more than past cases presented to this Court that precedent-seeking review is an inherently subjective exercise that “invokes culpability assessments by the Court, which are, essentially, moral judgments.” Loftin II, supra, 157 N.J. at 426, 724 A.2d 129 (Handler, J., dissenting). As some features of defendant’s case demonstrate, a death sentence can always be justified given the level of individuality afforded the analysis. Accord Chew II, supra, 159 N.J. at 268, 731 A.2d 1070 (Handler, J., dissenting) (stating that proportionality review has been reduced to “whatever works” to justify a finding of proportionality).
*3861.
The Court first examines defendant’s case independent of other cases, taking into account Harvey’s moral blameworthiness and character, as well as the level of victimization resulting from his crime. The majority concludes that “[w]ith respect to defendant’s moral blameworthiness and character, defendant is highly culpable.”’ In contrast, defendant’s culpability is merely moderate with respect to the degree of victimization. Ante at 315, 731 A.2d at 1141. I disagree with the Court’s assessment of defendant’s moral blameworthiness.
In determining a defendant’s blameworthiness, the Court is to examine the following: motive; premeditation; justification or excuse; evidence of mental disease, defect or disturbance; knowledge of victim’s helplessness; knowledge of effects on nondecedent victims; defendant’s age; and defendant’s involvement in planning the murder. See ante at 300, 731 A.2d at 1133 (citing Marshall II, supra, 130 N.J. at 155, 613 A.2d 1059) (other citation omitted).
Defendant’s motive for committing the murder was, according to the jury, to escape detection. Although this aggravating factor (c(4)(f)) is considered to increase defendant’s moral blameworthiness, its widespread, almost universal application, regardless of the lack of evidence presented to establish it in various cases, destroys its efficacy as an appropriate aggravating factor. Its unbounded, amoebic application is inherently expansive, making it impossible to narrow the class of death-eligible defendants adequately to allow for meaningful distinctions regarding defendants’ blameworthiness. See Loftin II, supra, 157 N.J. at 428, 724 A.2d 129 (Handler, J., dissenting).
Harvey’s case demonstrates the problem: the prosecution presented no evidence at trial that would indicate defendant murdered to escape detection. The only evidence that defendant tried to escape detection came after the murder (he tried to clean up the blood). The jury most likely found the aggravating factor by mistakenly or inappropriately projecting defendant’s post-murder *387actions onto the murder itself. The evidence actually proves only that defendant was concerned about escaping detection for the murder, not that he murdered the victim to escape detection for the robbery. The conclusion that defendant murdered to escape detection is wholly speculative; its application, given the lack of evidentiary support for it, demonstrates that the factor is a highly suspect measure of culpability. Similarly, in cases where the evidence that a defendant killed to escape detection is overwhelming, the jury may fail to find the factor. See State v. Mejia, 141 N.J. 475, 662 A.2d 308 (1995) (defendant shot fleeing victim in back of head and jury did not find c(4)(f) factor); State v. Alexander, CCH Report, supra, Narratives for “No Penalty Trial/Life” Cases at 1 (defendant shot fleeing victim in back and AOC did not classify ease with aggravating factor c(4)(f)). The record in this case points strongly to the conclusion that the motive of defendant is unknown and cannot be used to increase his blameworthiness. Accord Loftin II, supra, 157 N.J. at 428, 724 A.2d 129 (Handler, J., dissenting).
Next, defendant’s murder involved no premeditation. He entered the house to rob it and killed Irene Schnaps in the process of doing so when he found her. No evidence was presented to suggest that the murder was planned.
Third, defendant had no apparent justification or excuse for the murder, thus this factor fails to decrease his moral blameworthiness.
Although the jury did not find the mitigating factors indicating that defendant suffered from mental disease, defect or disturbance, when ten non-statutory mitigating factors were presented to the jury as part of the catch-all c(5)(h) factor, some members found evidence that defendant suffered from emotional trauma for various reasons: six jurors found that defendant was traumatized when he witnessed his sister die as a result of a kerosene accident in which she caught on fire; one juror found that defendant was physically and verbally abused when he was sent to live with his grandparents; and four jurors found that defendant suffered *388feelings of abandonment when his parents promised to but did not take him with them when they moved from their home. See ante at 312, 731 A.2d at 1139. These factors decrease the moral blameworthiness of defendant.
Fifth, we have no evidence indicating the victim was helpless when defendant attacked her. In fact, the State presented evidence at trial that DNA tests conducted on blood found in Schnaps’s apartment indicated that the blood was consistent with defendant’s DNA. State v. Harvey, 151 N.J. 117, 143-44, 699 A.2d 596 (1997) (Harvey II). We have no way of knowing, therefore, if defendant struck first, or if he responded' to an attack from the victim with disproportionate force.13 Without more information, this factor cannot be used to heighten defendant’s blameworthiness.
Defendant had no knowledge of the effects of his actions on nondecedent victims. He did not know the victim, and therefore could not have known how her death would affect those left behind.
Defendant’s age does not decrease his blameworthiness because he was forty when he committed the offense. See ante at 313, 731 A.2d at 1140.
Finally, because defendant’s murder involved no premeditation, the factor regarding defendant’s involvement in planning the murder is irrelevant and does not increase his level of blameworthiness.
In sum, a extremely low level of blameworthiness: He did not plan the murder, his motive (to escape detection) is a motive that *389indiscriminately accompanies many other defendants’ crimes when a burglary/robbery is involved, and defendant had no particular knowledge of the effect his crime would have on nondecedent victims. Although defendant’s murder was not a random or senseless act, he had no excuse or justification for the murder; still some jury members found that defendant had suffered and does suffer from severe emotional trauma as a result of incidents throughout his childhood. Defendant’s age neither increases nor decreases his blameworthiness.14 Finally, we have no evidence indicating that the victim was entirely helpless.
The majority reaches a different conclusion. It relies heavily on Harvey’s motive, to escape detection, a factor that fails to lend much insight into the overall blameworthiness of defendants, see supra at 312, 313, 731 A.2d at 1139, 1140. By way of explaining the alleged helplessness of the victim, the Court provides only a detailed description of the murder itself, not the condition or position of the victim before the homicidal attack: “She was attacked with blunt instruments and struck so hard that her skull was fractured, her brain lacerated, and her jaw broken. She was beaten about the face and sustained many bruises in a brutal murder.” Ante at 313, 731 A.2d at 1140. The Court is not correct in concluding without evidence that the victim was helpless when she was attacked, supra at 388 n. 13, 731 A.2d at 1180 n. 13 —the detailed description of injuries should be considered only when evaluating the degree of victimization. If the savagery of the homicide can without more, serve to characterize the victim “helpless,” this, like the aggravating factor c(4)(f) to escape detection, will lose any utility as a measure of blameworthiness. All victims are, to some degree, ultimately helpless if they have been murdered. What we should be looking for here is a particular *390helplessness, such as a victim’s handicap, young age or unconsciousness, of which the defendant knowingly takes advantage.
The Court finds that knowledge of the effects of the decedent’s death on nondecedent victims weighs against defendant, stating, “Although defendant may not have specifically known that Irene had family and friends, we have previously recognized that ‘[w]hile a defendant might be unaware of the specific characteristics of his victims or of the particular survivors that the victim will leave behind, it is completely foreseeable that the killing will eliminate a unique person and destroy a web of familial relationships.’ ” Ante at 313, 731 A.2d at 1140 (quoting State v. Muhammad, 145 N.J. 23, 46, 678 A.2d 164 (1996)). The fact that defendant “must have realized that Irene had family and friends because there were personal photographs in her apartment and he stole a man’s Seiko LaSalle watch,” ante at 313, 731 A.2d at 1140, only adds to the transparency of this Court’s fictionalized assessment. Defendant is to be compared to others who committed robberies. The majority, in seriously proffering this rationale, implies that only by entering a bare house containing no personal effects indicating that the victim has relationships with other people, and perhaps then only if the person actually does not have such relationships, would the Court decline to find that a defendant had knowledge that his crime would impact and victimize other unknown or unknowable persons. The Court’s emphasis on a factor of such universality which requires no evidence fails to narrow the field of comparable cases in any meaningful way.
Next, the Court fails to address any evidence of defendant’s mental state when considering his blameworthiness, even though prior to its analysis it fills almost four pages with details regarding defendant’s presentation of relevant mitigating evidence beginning in his early childhood, as well as the jury’s findings on that evidence, ante at 310 -12, 731 A.2d at 1138 - 40. Similar factors have been considered by the Court in prior cases even when the mitigating factors indicating mental disease or defect are not *391found by the jury. Indeed, in this very ease, the Court, in concluding that E-l defendant Walter Gerald is less culpable than Harvey, notes that “Gerald’s sister testified about their family life, how their father’s death affected Gerald and Gerald’s use of alcohol and drugs.” Ante at 316, 731 A.2d at 1142. Surely, then, it would be worth mentioning in the Court’s examination of Harvey’s blameworthiness that defendant was deeply affected by the abuse he suffered in childhood and the death of his sister in a kerosene fire which he accidentally caused.15
Finally, the majority ignores the premeditation factor completely. There was none here.
The Court’s conclusion that defendant is “quite blameworthy,” see ante at 312, 731 A.2d at 1139, on this record is, in my view, an inaccurate assessment of this factor and, on a comparative scale, an unjust conclusion.
I agree with the Court’s assessment of the two other factors used in precedent-seeking review — level of victimization and defendant’s character. The victimization resulting from the murder is accurately described as “moderate.” See ante at 316, 731 A.2d at 1142.16 Defendant’s character is certainly highly culpable: he *392has a serious criminal history, including guilty pleas to rape, sexual assault, assault, second-degree attempted kidnapping, second- and third-degree burglary, and receiving stolen property. See Ante at 313-15, 731 A.2d at 1140-41.17 He showed no remorse, there was no evidence that he cooperated with the authorities, and there was no evidence presented that he was susceptible to rehabilitation, although his allocution statement and his positive relationships with his children could perhaps speak to this factor.
I conclude that defendant has a low level of culpability. His bad character is counteracted by his relatively low moral blameworthiness and the moderate degree of victimization.
2.
The second part of precedent-seeking review is a comparison of defendants case to those of similarly situated defendants. As in its salient-factors analysis, the Court here limits the universe of comparison cases for precedent-seeking review to only those placed in the E-l salient-factors subcategory. See ante at 296, 731 A.2d at 1131. By fading to compare defendant to other factually similar cases in the broader E category, the Court conducts only a partial proportionality review. The Court’s decision to so drastically limit the universe of comparison cases in this case, one in which the frequency analysis results are so low as to require a greater emphasis on precedent-seeking review, see ante at 307, 731 A.2d at 1137, is a grave mistake.
*393The only difference between the E-l and E-2 subcategories is that E-l cases involve forced entry into a residence. The questionable wisdom of adhering to such a seemingly inconsequential distinction for purposes of precedent-seeking review, in light of the crimes committed, is clear on its face; but when the cases are examined more closely, we see a marked inconsistency in the placement of eases within these categories, further calling into question the Court’s decision to limit precedent-seeking review here to the E-l group. Harvey’s case provides the perfect example. The Court states in its presentation of the facts that “[djefendant broke into the apartment of Irene Schnaps.” Ante at 309, 731 A.2d at 1138. There is, however, no evidence to support this statement. Harvey is placed in the forced entry category in spite of the fact that the “investigating police detected no signs of a forced entry.” Ante at 284, 731 A.2d at 1124. In fact, the person who discovered Schnaps’s body “entered through an unlocked doorway,” Harvey II, supra, 151 N.J. at 138, 699 A.2d 596, and defendant confessed to the police that he had entered through that same unlocked doorway. See Harvey I, supra, 121 N.J. at 412, 581 A.2d 483. The AOC, with full knowledge of all of these details, nonetheless placed defendant in the E-l, forced entry category.
Gerald Williams’s case provides another apt example. Williams, an E-l defendant, entered the apartment he robbed through an ajar door, finding his victim asleep inside. Why, then, is he classified as a defendant committing a robbery that involved forced entry? Another E-l defendant, Charles Ploppert, and his accomplice gained entrance to the home they robbed by identifying themselves to the inhabitant, whom defendant knew, and getting him to open the door. They then even sat talking with the man at his kitchen table before they eventually attacked him. There is no evidence, therefore, that the co-defendants forced their way into the house. Ploppert, however was also placed in the forced entry category.
*394These clear categorization errors, combined with the meaninglessness of the distinction between the E-l and E-2 categories to begin with, render the Court’s decision not to include E-2 defendants in its proportionality review indefensible.
The distinction between the E-l and E-3 categories, while not as inconsequential as that between the E-l and E-2 categories (it is based on whether or not the murder was committed with particular violence or terror), still lends itself to rather subjective line-drawing that is not always wholly understandable. Many of the cases placed in the E-l category seem to belong there due to the extremely violent components of their crimes. What is more difficult to discern are the justifications for placing many of the E-3 defendants in the less blameworthy category. Defendant Aaron Huff is a good example. The jury found the c(4)(c) aggravating factor, indicating that Huff caused the murder by extreme suffering, having struck his victim’s head on a coffee table and beaten him until he died. Yet Huff was placed in the E-3 subcategory, indicating that the robbery was not committed with particular violence or terror. The fact that prosecutors have sought the death penalty against E-3 killers at more than twice the rate they have sought the death penalty for E-l killers (comparing sixty-seven percent and thirty-percent, excluding defendant), CCH Report, supra, tbl. 7, should be indication enough that E-3 eases are not necessarily less blameworthy than E-l cases. Many of the AOC narratives describing these crimes indicate much more terror and violence used against the victims than in Harvey’s case. Comparison of defendant’s case to these E-3 defendants is warranted given the subjectivity of the line-drawing between the two subcategories.
The Special Master agrees that the distinctions between the salient-factors subcategories should be eliminated given their failure to adequately measure blameworthiness. Supra at 309 -10, 731 A.2d at 1138; Special Master Report, supra, at 57, 62 (“I recommend that we abandon the current hierarchical structure which is based primarily on intuition.”). The Court’s, decision to reduce the universe of comparison cases of the general E category *395to the far more restrictive and even, at times, arbitrarily drawn subeategory of E-l defendants is a serious departure from past proportionality reviews. Further, this drastic step is taken in a case in which precedent-seeking review plays a particularly crucial role in our overall proportionality review given the results of the frequency analysis.
Defendant also seeks to compare himself to several defendants outside the E category. Some of the cases presented by defendant are factually similar to his case and, therefore, I disagree with the Court’s decision not to include them in its precedent-seeking review.
First, Daniel Hart, who was classified as a G-category defendant (murder while committing a burglary) was also convicted for robbery. Because robbery is more serious than burglary, and the cases are meant to be classified in the most serious category that applies, see CCH Report, supra, at App. E-3 (classifying G category as “Burglary murder not involving a robbery or sexual assault”), Hart should have been placed in the E category and should therefore be among the cases to which defendant is compared.
Next, defendant argues that William Godette, a defendant from the B category, should be in the comparison group. I agree. Godette is in the prior murder conviction category because in addition to his New Jersey murder, he was convicted of killing his step-father in North Carolina. Defendant suggests that Godette should not be classified as a prior murderer, however, because Godette had not yet been convicted of his step-father’s murder at the time of his New Jersey murder. In fact, Godette was arrested for the North Carolina murder a full ten months after the New Jersey murder. It is unclear why the AOC placed Godette in the B category, given that the prosecutor and jury for his New Jersey trial would not have been able to consider a prior murder conviction at the time the defendant was tried and convicted.18 Because *396Godette’s crime was strikingly similar to Harvey’s, and because the essential element of his crime that the AOC should have used to categorize him was his robbery, he should be included in defendant’s comparison universe.
Finally, defendant suggests that some A-category defendants (involving multiple victims) would have been in the E category but for the fact that they committed multiple murders. Defendant’s assertion should again alert the Court to some potential problems with simply deferring to the AOC’s judgment on its category groupings. Walter Gerald and his accomplices attacked three people, killing two of them; and yet he is placed in the E-l category and not the multiple victim category. Gerald plays a significant role in the Court’s finding of proportionality because he is one of only two defendants in the E-l category other than Harvey to have received a death sentence (though his sentence was later reversed and the prosecution declined to capitally prosecute at retrial). For the Court to place so much emphasis on a multiple-victim defendant, placed in the E-l category for no identifiable reason, and then to decline to compare defendant to similar cases in the multiple-victim A category makes no sense. Further, although I find that the number of victims is an important distinction between these categories, some of the A-category eases are factually similar enough in all other ways to warrant comparison. The Court must agree, or it would remove Gerald’s case from its analysis.
In three separate instances, the Court has conducted comparisons with defendants that fall outside the AOC-designated salient-factors grouping for the defendant subjected to proportionality review. See Cooper II, supra, 159 N.J. at 93, 731 A.2d 1000 (including two cases relied on by defendant outside defendant’s *397composite C category); Martini II, supra, 139 N.J. at 50-51, 651 A.2d 949 (including cases outside defendant’s I category involving pecuniary motives); Marshall II, supra, 130 N.J. at 175, 613 A.2d 1059 (comparing eases proposed by defendant). In DiFrisco III, Loftin II, Chew II, and now Harvey III, the Court has opted to “ ‘defer generally to the AOC’s expertise, and particularly to its unique assignment of defendants to only one comparison category____’ ” Ante at 295, 731 A.2d at 1130 (quoting Loftin II, supra, 157 N.J. at 327, 724 A.2d 129 (citing DiFrisco III, supra, 142 N.J. at 167, 662 A.2d 442) (citations omitted)). This deference is ironic in light of the Court’s past acknowledgement that there is “danger inherent in ‘any attempt to define in advance all characteristics of a murder case [to] capture the critical facts of a defendant’s case’ because that ‘would fail to distinguish between individual defendants.’ ” DiFrisco III, supra, 142 N.J. at 164, 662 A.2d 442 (quoting Martini II, supra, 139 N.J. at 24, 651 A.2d 949) (other citation omitted). Indeed, in the past, even when stating that it will “defer generally to the AOC’s expertise,” see id. at 167, 662 A.2d 442, the Court has allowed for adjustments to the AOC’s categories when necessary, see id. at 170, 662 A.2d 442 (‘Williams’s exclusion [from defendant’s comparison group] is justified on the grounds that the jury rejected the pecuniary motive aggravating factor at the penalty trial.”). Although defendant’s ease cannot realistically be compared to all other death-eligible defendants, see Loftin II, supra, 157 N.J. at 436, 724 A.2d 129 (Handler, J., dissenting), limiting the comparison group solely to the E-l subcategory is too restrictive. Such a limitation fails to recognize the factual similarities between cases in the same composite category, and indeed, in others.
Adding the factually similar eases proposed by defendant from the E-2, E-3, G, B and A categories,19 highlights the dispropor*398tionality of defendant’s sentence. See supra at 297 - 99, 731 A.2d at 1131 - 32. A close examination of the facts of each case renders problematic the Court’s conclusion that because of the differences that distinguish defendants in the E-2, E-3 and G-3 cases from Harvey, “such cases provide little insight into the propriety of the jury’s decision in this case, and are inapplicable to our proportionality review.” Ante at 299, 731 A.2d at 1132.
3.
The Court concludes that the E-l category defendants are distinguishable from defendant’s and therefore defendant has not made a case for disproportionality. See ante at 385, 731 A.2d at 1179. I strongly disagree and explain the basis for that disagreement by a detailed analysis of the compared cases.
Defendant is compared first to Walter Gerald and Rigoberto Mejia, the only other two defendants in the E-l subcategory who were sentenced to death, both of whom are currently serving life sentences. First, the majority concludes that defendant is more blameworthy than both Mejia and Gerald because the jury found a rational basis that they intended only to inflict serious bodily injury on the victims, while Harvey intended to kill Irene Schnaps. Ante at 316, 731 A.2d at 1141. In addition, the Court points out that the jury found that Gerald suffered from emotional disturbance and mental disease or defect, that he had expressed remorse, that his father’s death affected Gerald, that defendant used *399alcohol and drugs, and that Gerald had no significant prior record. See ibid.
The Court’s comparison of Harvey to Gerald is suspect given that Gerald’s death sentence was reversed and that the State did not seek the death penalty again on retrial.20 If anything, the ease highlights the disproportionality of Harvey’s sentence because the facts demonstrate, in many ways, that Gerald is more blameworthy than defendant, and yet the prosecutor concluded the case was not worthy of a second capital prosecution.
Gerald and his two co-defendants broke into a house to steal a television set. They had been observing the household for some time, so they knew the number of inhabitants and entered according to a formulated plan. There were three intruders and three victims, all of whom were helpless (the two men were not self-sufficient, one eighty-five years old and one fifty-five years old; and the woman who took care of them had no weapon). Gerald knew of all of the victims’ vulnerabilities before breaking in. The woman was punched in the face, stomped on and threatened with a knife, suffering numerous serious injuries, including contusion's of the face, neck and chest. As a result, she was hospitalized for twelve days and her jaw was wired shut for six weeks; The younger of the men was killed after having been beaten by all three perpetrators and then, finally, struck in the face with a television set. He drowned in the blood from his broken nose and suffered contusions and swelling to the brain. It was unclear whether the nose was broken by a foot stomp to the head (evident by a' sneaker print across the victim’s face) or by the television set. The third victim, the eighty-five-year-old, was beaten and dragged from his bed, suffering bruises and lacerations to his face caused *400by blunt-force blows with a lamp and his own cane. He died shortly after the incident.
While Gerald’s prior criminal history is minimal in comparison to Harvey’s and he suffered from drug addiction, the jury made no findings regarding a troubled childhood as in Harvey’s case.21 In addition, the jury that sentenced Gerald to death believed that his crime was outrageously or wantonly vile, finding the c(4)(c) aggravating factor. This finding surely counteracts the Court’s contention that because Gerald only intended serious bodily injury, he is less culpable than defendant. See ante at 382, 731 A.2d at 1177. Though the State submitted the c(4)(c) aggravating factor to the jury in Harvey’s case, the jury did not find his crime to be outrageously or wantonly vile.
Further, there were three victims (two of whom died) resulting from Gerald’s crime, whereas Harvey had one victim. Although the Court describes Harvey as a “cold and calculating murderer,” ante at 319, 731 A.2d at 1143, there is more evidence that Gerald’s crime involved calculated premeditation and the exploitation of the helplessness of his three victims. Confrontation was part of Gerald’s plan because he knew the inhabitants were home. Harvey, by contrast, seems to have come upon his victim merely by chance.
The only other E-l defendant sentenced to death is Rigoberto Mejia. Inclusion of Mejia’s sentence as a death sentence is also not appropriate, given that on retrial he received only a life sentence because his crime was deemed not death-eligible. The AOC chose to exclude Mejia’s non-death-eligible, life sentence from the database, and yet the AOC included his death sentence. Either Mejia’s case is death-eligible and therefore both his life and death sentences should be included; or his case is non-death-eligible and both sentences should be excluded.
*401Mejia and an accomplice attacked Balbino Garcia in a hotel basement. When Garcia ran, Mejia and his accomplice chased him into a bedroom, where Mejia pointed the gun at Garcia and one of his relatives. When Garcia tried to take the pistol, Mejia struck him in the face with it, fracturing his skull. He then chased Garcia and shot him from within inches of the victim’s back.
Mejia had a similarly troubled childhood to that of defendant, resulting in the jury’s finding of the catch-all mitigating factor, but his crime was more aggravated than defendant’s. Mejia armed himself with a pistol and sought to confront a specific victim. He did not simply kill the person unfortunate enough to catch him in the middle of a robbery; he threatened an individual and shot him when he ran away. The jury’s failure to find the aggravating factor that defendant shot Garcia to escape detection is important to note: either this finding demonstrates that the factor is indiscriminately applied (to defendant in spite of the lack of evidence that he killed to escape detection but not to Mejia in spite of the seemingly overwhelming evidence that he did, since he shot Garcia as he was running away); or it demonstrates the cruelty with which Mejia carried out his crime — to shoot Garcia as he ran away, if not to escape detection, could only have been a conscious choice to make Garcia suffer, to kill for killing’s sake. Harvey’s case does not demonstrate this kind of calculated choice.
Finally, when all is said and done, neither Mejia nor Gerald is on death row. Even, therefore, if their crimes, characters and blameworthiness indicate less culpability than defendant’s, for the Court to base its finding of proportionality on these cases is suspect. The fact that less culpable defendants received life sentences does not mean that a more culpable defendant deserves death.
The majority reduces its comparison of defendant’s case to other E-l cases to a simple discussion of the characteristics of the defendants, omitting any examination of the crimes themselves. *402Although these factors are certainly an important part of our analysis, the Court’s failure to examine the crimes renders its proportionality review startlingly incomplete.22 When, as here, the salient-factors results indicate a low incidence of death, we must do everything we can to ensure defendant has not been unfairly singled out. Marshall II, swpra, 130 N.J. at 154, 613 A.2d 1059. It is especially important, therefore, that we examine and compare the facts of the cases, along with the defendants, however lengthy and gruesome that endeavor might be. The crimes committed by these E-l defendants are, without question, more heinous than that committed by defendant, and all of the E-1 defendants are currently serving life sentences or less.
*403Will Alexander and his co-defendants gained entry to an apartment in which the victim and his girlfriend were home with two small children. Upon entry, one of the defendants pushed the girlfriend to the floor. When the man attempted to flee, defendant shot him in the lower back, killing him. The defendants then forced the woman and her children into another room and ordered them to stay on the floor while they raided other apartments in the building. Alexander was not capitally prosecuted and was sentenced to life in prison.
Jerry Britton, against whom the State did not seek the death penalty, climbed through the window of a young woman’s apartment and stabbed her sixteen times with two knives when she began to call the police. The stab wounds were in the head, neck, back and shoulder, and one of the knives was broken off in the victim’s neck. The victim also appeared to have been beaten. Britton told a friend that he hoped that he had killed her so that she could not be a witness. Britton had a heroin habit, but no mental health problems. He was sentenced to life in prison.
David Brown planned a robbery with two other accomplices for drug and beer money. They went to a drug dealer’s apartment to confront and rob him. When an argument ensued, Brown pulled out a knife and stabbed the victim multiple times all over his body. The State did not seek the death penalty in the case, and Brown was sentenced to thirty-five years in prison.
Alphonso Brunson broke into an eighty-two-year-old woman’s home for the third time on the day she was killed. The woman was found two days later, having received several blows to the head which had caused her death. Brunson was sentenced to two concurrent terms of life and fifty years.
Duane Vance Caviness, accompanied by two co-defendants, entered an apartment of a fífty-four-year-old man, tied him up and beat him with a baseball bat. The man was later found dead on his apartment floor with severe head wounds, his hands and feet *404bound. Caviness was permitted to plead guilty to felony-murder and two counts of burglary/robbery. He was sentenced to life.23
Albert Carrow Fains murdered his wheelchair-bound neighbor, Arthur Williams, by striking him in the head thirteen times with a claw hammer. Williams had sent Fains to buy him cigarettes, sandwiches and marijuana, but the following morning the victim was found on the floor with a knife in his back and blood everywhere, including on the chairs and walls. A plastic bag had been pulled over Williams’s head. Williams suffered three fractures on the right side of the skull, a wound on the bridge of his *405nose, and eight other wounds on the side of the head, the combination of which were determined to be the cause of death. Fains was sentenced to life.
Carlton.Felder rang the bell at his seventy-five-year-old neighbor’s apartment. When she opened the door, Felton pushed her inside and started stabbing her repeatedly in the left side of her chest. He then grabbed the gold chains from h'er neck and proceeded upstairs to look for money. At the time of the murder, the woman was babysitting three small children. The State did not prosecute Felder capitally and he was permitted to plead guilty to aggravated manslaughter, robbery and burglary. He was sentenced to fifty years.
Franklin Flowers Hudson entered a home through the basement "window and tied up and gagged at knife-point the homeowner who found him. When the owner’s sixty-five-year-old boarder returned home, Hudson confronted him, leaving the owner tied up. Even after the boarder gave Hudson his money and keys, Hudson stabbed him multiple times. Not yet dead, the boarder tried to run upstairs, and Hudson chased him and knocked him down by kicking him and repeatedly hitting him over the head with a baseball bat. The boarder did not die from his injuries until over a month later. Hudson was permitted to plead guilty to felony murder and he was sentenced to life.
Timothy Paul Lee took a knife and kicked in the back door to the home of a sixty-five-year-old man. When the man woke from the noise, defendant stabbed him in the chest, killing him. Lee was permitted to plead guilty to felony murder and was sentenced to life.24
*406Dwayne Mann and two co-defendants broke into the home of a man with the intention of robbing him. When the man woke up, defendant shot him in the head, killing him. Mann was sentenced to two consecutive terms of life and five years in prison.
Incenzio Mendez, attempting a robbery, lay in wait for the approaching ninety-five-year-old woman who owned the farm on which he worked. Coming up behind her, defendant used a stick to knock down the woman with three hits to the head. The victim tried to get up, at which point defendant kneed her in the side and struck her in the neck. The woman died from the injuries. Defendant was capitally prosecuted and was sentenced to consecutive terms of life, twenty years and ten years.
Lance Phillips and his accomplices stormed a house armed with guns hoping to steal a kilogram of cocaine that Phillips had seen at the house earlier. During the course of the raid, Phillips and his co-defendant shot everyone in the house (a twenty-year-old man, his girlfriend, a seventeen-year-old-girl, and an eleven-year-old girl). Phillips shot the man five times, killing him. He also shot the seventeen-year-old in the arm. One of the co-defendants shot the eleven-year-old in the chest. The prosecutor did not seek the death penalty against Phillips and he was sentenced to consecutive terms of life and twenty years.
.Charles Ploppert and a co-defendant knocked on the door of a blind man, whom Ploppert knew, with the intention of immediately hitting him on the head with a baseball bat in order to be able to *407rob the house. The man forced Ploppert to identify himself before he would open the door. When Ploppert did so, the man let him and his co-defendant in. After chatting amicably with the man, Ploppert attacked him, beating him unconscious by hitting him with his fists and kicking him. Before leaving, Ploppert piled wood on the unconscious victim, spread lighter fluid over him and around the house, and then set a fire. Ploppert was capitally tried and sentenced by the jury to life.
Thomas Reigle broke into the apartment of his uncle and aunt to steal money. Hearing his aunt stir in her bed as he was looking through her purse, Reigle beat her with a pipe. She' survived. Reigle then went into his uncle’s room and beat him to death with the same pipe. The prosecutor sought the death penalty and the jury sentenced Reigle to life.
Anthony Szadorski and a co-defendant broke into the home of a seventy-six-year-old woman whom Szadorski had met at an Alcoholics Anonymous meeting. When the woman jumped out of bed upon Szadorski’s entering her bedroom, he stabbed her several times. He continued to do so as she tried to crawl away. Szadorski then asked his co-defendant for a BB gun that he used to beat the woman over the head. She eventually died from her injuries. The prosecutor did not seek the death penalty against Szadorski and he was sentenced to life.
Gerald Williams, a co-defendant and Williams’s eight-year-old daughter entered an apartment they happened to be passing through an ajar door. Inside, they found a fifty-one-year-old man who awakened when Williams turned off the television set to steal it. The co-defendant punched the man and Williams threw a cover over the man’s head and started beating him against the window sill. The man broke free and yelled for help, at which time Williams picked up the television set and hit the man over the head with it. He then put down the television set and threw the man out of the window. The man fell three stories and died. Unlike the other defendants, Williams had a significant criminal history, having served twelve terms of incarceration as an adult *408and a juvenile. The prosecutor did not seek the death penalty and Williams was sentenced to life.
Herman Williams, armed with a handgun, entered a home planning to rob one of the family members whom he had observed. Upon entering, Williams found six residents there. He hit one in the face with his handgun and then got into a struggle with an older, handicapped man who had an artificial arm. After the man successfully knocked the gun out of Williams’s hand, the defendant picked up the gun and shot the man in the chest. The man died seventeen days later in the hospital. The prosecutor did not seek the death penalty against Williams. The defendant was sentenced to life.25
In sum, many of these cases involve multiple victims, and those that do not, involve more extensive victimization than Harvey’s. With the exception of Franklin Flowers Hudson and Gerald Williams, all of the defendants either knew their victims personally or had observed them enough to know of their vulnerabilities. Many of the victims were even selected, presumably, because of •their vulnerabilities: one was in a wheelchair, one was blind, one was handicapped, two were unable to be self-sufficient (one fifty-five and one eighty-five-years old), two were women in their seventies, one was an eighty-two-year-old man, one was a ninety-five-year-old woman, and one was an eleven-year-old girl.
The fact that all of the defendants in these cases are serving life sentences and Harvey faces execution boggles the mind. Defendant is most definitely not among the one or two most culpable defendants in his category, as the salient-factors test indicates he *409should be in order to be sentenced to death. Is the only difference here that Harvey is black and his victim was a white woman?26 See data, supra at 297, 731 A.2d at 1131 (noting that five of six transracial eases were capitally prosecuted and three of those five resulted in death sentences even though transracial cases account for only twenty-seven percent of the E-l subeategory). I can only hope that this is not the explanation. The Court provides no other.
When the universe of comparison cases is extended to similar eases in the E-2, E-3, G, B and A categories, the Court’s holding becomes even more implausible.
Jesus DeJesus, an E-2 defendant, entered the apartment of a forty-nine-year-old woman living below him. He stabbed her to death and then set her bed on fire with her body on it. The remains were identified with her dental records. DeJesus stole some jewelry. He was not capitally tried and received consecutive life and fifteen-year sentences.
Wayne Busby, an E-3 defendant, should certainly have been classified as an E-l defendant given the facts of the case. Busby broke into the apartment of a seventy-four-year-old woman who lived behind his residence. He had been watching her prior to the break-in to try to discern when she was home. When the woman came downstairs unexpectedly, half-dressed, the defendant hit her in the face, broke her ribs, and used a broom to strangle her to death. He used enough force to break the broom handle. While being strangulated, the woman tried to fight Busby off by scratching him about the neck. Busby took money, a camera, film and perhaps other items. This case clearly involves particular violence or terror. Busby was sentenced to life in prison.
Thomas Dollard and two co-defendants entered an. apartment building in search of someone to rob. They encountered two people on a stairwell and forced them to take their pants down so *410the defendants could search them for drugs. When they found none, defendants made the couple take them around the building to help them gain access to apartments. When a woman behind the door of the first apartment refused to let them in, defendants kicked in the door and brought the couple inside with them. They were confronted by a man in the bedroom, who asked ‘Why are you doing this? There is nothing here.” Dollard shot the man immediately in the chest. The man, still alive and expressing incredulity, asked, “Why did you do that? You didn’t have to do that.” He died from the wound shortly afterward. The man from the couple encountered in the hall then jumped out the window from fear of being shot. Dollard was not capitally charged and received a life sentence.
■ Larry Durden ate dinner at the home of a seventy-two-year-old woman who had offered to have him over if he changed the locks on her doors. At some time during the evening, Durden stabbed the woman with a small ax-type object. She died of wounds to the forehead and abdomen. Dollard took the woman’s groceries, a television and a radio. Durden was sentenced to life.
The gruesome list goes on: Aaron Huff struck his victim’s head on a coffee table and then beat him until he stopped moving. He was sentenced to life. Michael Suarez stabbed a man in the neck, back and chest. The victim was found in the fetal position lying between the wall and bed on top of blood-soaked clothes and wearing boxer shorts. The prosecutor did not seek the death penalty and Suarez was sentenced to life. Thomas Wolfe slashed a woman’s throat three times and she also suffered from numerous puncture wounds. He was sentenced to life.
All of these cases involved a robbery, forced entry and murder of one victim, yet none of them was classified in the E-l category requiring particular violence or terror.
Daniel Hart, who was convicted of both robbery and burglary, but was placed in the G category, is surely more blameworthy than defendant. Hart and a co-defendant formulated a plan to kill a twenty-three-year-old woman they thought was a snitch. When *411the woman confronted them in the main entrance to her apartment, Hart tried to smother her with a pillow. He then killed her by slashing her throat and stabbing her thirty times in the neck, head and back. He took $25 from her. The prosecutor did not seek the death penalty and Hart was sentenced to fifty years in prison.
William Godette, a B-category defendant who was convicted of felony murder and robbery, went to the home of a seventy-nine-year-old man to demand payment for work done earlier in the day. When the man refused to pay, Godette pushed his way inside and pounced on the man. The defendant then tried to strangle his victim and finally killed him by striking him several times in the head with a hammer. Godette was permitted to plead guilty and was sentenced to life.
Several robbery cases classified as A-category, multiple murder eases highlight the disproportionality of defendant’s sentence as well. Significantly more blameworthy defendants committing murder in the course of a forced-entry robbery were not sentenced to death. Felix Diaz and his co-defendant planned the robbery and murder of an older man who lived with a younger man and an eight-year-old girl. The defendants shot all three victims and burned their bodies. Diaz was given three consecutive life terms.
Peter Regan broke into his girlfriend’s mother’s house and, upon being found by a fifteen-year-old girl, hit her five times over the head with a baseball bat until she stopped screaming. Regan then killed his girlfriend’s twelve-year-old sister by hitting her six times over the head with the same bat. He removed the dead girl’s clothing from the waist down in order to make it appear as if there had been a sexual assault. He was convicted of robbery and the prosecutor did not seek the death penalty. Regau was sentenced to two concurrent terms of life.-
Roy Watson broke into the home of an eighty-four and seventy-nine-year-old couple on his street. He went into the bedroom and beat the man to death. The attack was so severe that Watson *412knocked out the lens of one of the man’s eyes. The man’s wife woke up during the attack and Watson then beat her to death. The penalty-trial jury sentenced Watson to two consecutive terms of life.
Harvey’s crime, while certainly brutal, pales in comparison to other similar crimes in terms of victimization, and often in terms of moral blameworthiness given the number of victims who are elderly or handicapped in the aforementioned scenarios and the level of premeditation and knowledge of the helplessness of the victims involved. When other life-sentenced defendants committing robbery in a theoretically more blameworthy category are added to the universe of comparison cases, the Court’s finding that defendant’s sentence is proportionate seems even more unjustifiable, by any standard. While the Court is correct in pointing out that we do not require identical verdicts in all similar cases, see ante at 319, 731 A.2d at 1143, Harvey was placed in one of the least culpable categories and no defendant in that category but he now sits on death row. It is time for the Court to articulate what it means by disproportionate, rather than to continue to insist that it knows only when a sentence is not disproportionate. Only then will it be unable to keep moving the line by adding the very specific facts of each new defendant’s case to justify including him or her to the increasingly growing list of proportionately-sentenced defendants.
IV
I believe the Court’s principal findings in this case are not only in error, but have grave implications for the defendant before us and those to come. In finding, once again, that the statistical models presented do not reflect an unconstitutional risk of race discrimination in our capital sentencing scheme, the Court ignores not only hard evidence that race plays an impermissible role in capital charging and sentencing procedures, but the general evidence all around us that racism is ubiquitous, and poses a clear, significant and present risk. In particular, the evidence that *413prosecutors and juries are significantly more likely to charge with and sentence to death black defendants killing white victims, is overwhelming. We must no longer ignore the serious risk that these defendants are singled out because of the premium our society places on white life and its simultaneous degradation of minority life. No system of justice can operate fairly as long as this risk exists.
I find, also, that this case starkly underscores the fact that our individual proportionality review process is rife with subjectivity. See Cooper II, supra, 159 N.J. at 92, 731 A.2d 1000 (“[W]e acknowledge that the .process of conducting precedent-seeking review is inherently subjective____”). Our methodology is devoid of the kind of quantifiable standards that are necessary to provide a real basis for comparison against which to measure all defendants claiming disproportionality. The Court has simply raised the bar for disproportionality with each new case, narrowing the class of comparison cases and reducing the discussion of the cases to cursory statements that lump together numerous defendants with complete disregard for the facts of their crimes. This blatantly incomplete review obscures the arbitrariness of Harvey’s death sentence. While relying on nothing less than scientific certainty of race discrimination in response to defendant’s systemic challenge, the Court has declined to define any numerical cut-off for disproportionality in its frequency analysis in defendant’s individual proportionality review.
Proportionality review is the very last means available to us for correcting arbitrariness in capital sentencing. Yet, the Court’s review seems almost to be guided by an effort to find proportionality at all costs, rather than to root out disproportionality. Our very first proportionality review portended this inevitable result:
The majority depreciates and excuses the occasional unequal sentencing treatment of similar defendants by attributing that to sentencer “mercy” ... But that rationalization explains aberrational life sentences, not aberrational death sentences.
[Marshall II, supra, 130 N.J. at 241, 613 A.2d 1059 (Handler, J., dissenting).]
*414Here, in spite of the salient-factors groupings indicating that defendant is in a category of defendants who are almost never sentenced to death, the Court has effectively begun with the contention that we would expect defendant to be sentenced to death. It requires defendant to show unusual circumstances warranting a life sentence in order to achieve such a sentence.27 While the burden does indeed rest on the defendant’s shoulders to prove disproportionality, a presumption in favor of proportionality was never intended to be integral to our method of proportionality review.
The Court in Marshall II stated that in recognizing disproportionate sentences, we are looking to see if an identifiable pattern of life sentences has been broken with no explanation. See Marshall II, supra, 130 N.J. at 181, 613 A.2d 1059 (“That [two other defendants] were spared their lives does not establish a pattern of life-sentencing for such killings.”). Here, there is without a doubt a pattern of life sentences that has been broken with Harvey’s death verdict. Because there are no defendants in Harvey’s salient-factors category who await execution, we would expect that Harvey himself would also receive a life sentence. Although the fact that he is the only one in his category awaiting death cannot alone denote proportionality, see id. at 166, 613 A.2d 1059 (“[S]imply because Marshall may be the first does not mean that his death will be disproportionate under our statute.”), we must demand some defensible reason to distinguish him from the others — to single him out for death — in order to avoid an arbitrary sentence. The Court fails to do so, trampling on the very heart of our Constitution’s equal protection clause:
A capital sentencing system which results in differential treatment of similarly situated capital felons has effectively classified similar felons differently with respect to their rights to life ... Where capital sentences cannot be rationally distinguished from a significant number of cases where the result was a life sentence, more is present than the irremediable failure of an imperfect human *415system. When this occurs, the capital sentencing system has become constitutionally arbitrary.
[Id. at 241, 613 A.2d 1059 (Handler, J., dissenting) (quoting Gary Goodpaster, Judicial Review of Death Sentences, 74 J.Crim. Law & Criminology 786, 788, 802-03 (1983).)]
This case demonstrates more compellingly than any to date the errant standard by which the Court has chosen to implement our final defense against arbitrariness: “a death sentence is disproportionate if other defendants with similar characteristics generally receive sentences other than death for committing factually similar offenses unless the Court concludes through its subjective intuitive examination of precedent that the sentence is fair.” Id. at 250, 613 A.2d 1059 (Handler, J., dissenting). This “standard,” of course, defies the Court’s oft-stated goal of maintaining consistency: “[Tjhere can be no justice without a predictable degree of uniformity in sentencing.” Id. at 239, 613 A.2d 1059 (Handler, J., dissenting) (quoting State v. Hodge, 95 N.J. 369, 379, 471 A.2d 389 (1984)). In employing its faulty and subjective method of review, the Court has managed to find a clearly disproportionate sentence proportionate.
I, therefore, dissent. Justice STEIN joins in the conclusions reached in Point III, 2B, and 3 of this opinion, and also dissents.
STEIN, J., concurs in part and dissents in part.
For affirmance — Chief Justice PORITZ and Justices POLLOCK, O’HERN, GARIBALDI and COLEMAN — 5.
For reversal — Justices HANDLER and STEIN — 2.
The Court directed the Special Master to examine eight substantive areas. See ante at 299, 731 A.2d at 1132 (citing Loftin II, supra, 157 N.J. at 454-56, 724 A.2d 129).
I disagree with the AOC’s inclusion in the database of two defendants whose death sentences were reversed. Without inclusion of these cases, the death sentencing rate for E~1 defendants would be zero percent, lower still than the ten percent cited by the majority. However, even under the Court's use of the AOC's coding decisions, I contend that the Court misapplies and underestimates the evidence of disproportionality.
The primary concern with the AOC models is that they suffer from a problem known as overfitting — that is, the small number of cases in the database and the large number of variables in each model serves to undermine the results. Dr. Tukey employed fewer variables by pooling various factors together, attempting to solve the overfitting problem with more parsimonious models.
One transracial case not capitally tried is Walter Gerald’s on retrial. Gerald was originally tried capitally and sentenced to death. After this Court’s reversal of his sentence, the prosecutor declined to conduct another penalty trial, so Gerald’s life sentence was not challenged. The fact that the State attempted to capitally prosecute Gerald the first time, in spite of the lack of evidence at its disposal to prove intent to murder, only heightens the argument that race might be playing a role in prosecutorial decisions. The State's decision not to try to obtain a capital sentence on retrial after the Court’s reversal reflects, perhaps, simply the prosecutor’s appropriate realization that the case was never death-worthy in the first place.
While, unlike the Court, I believe a numerical preponderance cut-off is a necessary means for quantifying and fairly applying our "general imposition" standard, this is because we are not attempting to actually prove a causal relationship with numbers in individual proportionality review, we are merely defining a standard using a numerical base. In our attempt to quantify systemic racism, however, such a cut-off point is unrealistic. The numbers will never be able to prove or disprove the presence of racism. The Court, therefore, should be at least as adamant about its reluctance to rely on statistical models in its examination of race discrimination as it is in its individual proportionality review analysis. I do not suggest that we do away with the models or give up on our attempts to perfect them, but rather that we supplement them with common sense and experience.
Republican Governor Mike Johanns vetoed the bill five days later, stating that "it would allow death row inmates to 'advance further unnecessary appeals.’ ” Nebraska Leader Vetoes Suspension of Executions, N.Y. Times, May 27, 1999, at A21.
The E-l category is comprised of defendants who commit robbery murders with forced entry and particular violence or terror.
Specifically, the Special Master recommends that the robbery salient factor be divided into "residential, business and other” subcategories. Special Master Report, supra, at 59.
The Special Master agrees, recommending that for this reason, and various others, the index-of-outcomes test be eliminated from the Court’s proportionality review. Special Master Report, supra, at 90-107.
Again, the Special Master agrees. Special Master Report, supra, at 95.
Another problem with the Court's analysis is its comparison of Harvey to other defendants who have not been classified within the same culpability level. By comparing defendants across categories this way, the entire purpose of the test — to measure defendant’s culpability and establish the sentencing rate of similarly culpable defendants — is eviscerated. See Loftin II, supra, 157 N.J. at 423-24, 724 A.2d 129 (Handler, J., dissenting). ”[T]he singular fact that a defendant was the subject of a prior proportionality review does not justify his comparison with the present defendant.” Id. at 424, 724 A.2d 129 (Handler, J., dissenting).
In the face of the Court’s acknowledgement that precedent-seeking review has heightened significance in this case, see ante at 307, 731 A.2d at 1137, the majority’s decision to drastically limit the universe of comparison cases to the E-1 subcategoiy is especially puzzling. We should make very effort to afford the defendant a comprehensive precedent-seeking review when the individual comparisons of cases plays such an important role in that review.
The fact that defendant's blood at the scene of the crime might be an indication he was attacked first is consistent with defendant’s confession, in which he told the police that Schnaps woke up when she heard him and then punched defendant in the nose, causing him to bleed. See State v. Harvey, 121 N.J. 407, 412, 581 A.2d 483 (1990) {Harvey I). While the confession was suppressed in defendant’s second trial, this background information should make the Court especially wary of assuming the victim was helpless when there is no evidence to support that conclusion.
Defendant's age and lack of justification are not factors that increase defendant's blameworthiness, such as knowledge of the effect on nondecedents or helplessness of the victim — they simply fail to distinguish defendant as less blameworthy than others of average maturity who murder without justification.
When examining moral blameworthiness, the Court has cited evidence of a defendant's childhood abuse or emotional trauma. See, e.g., Bey IV, supra, 137 N.J. at 384, 645 A.2d 685 (observing that despite jury's failure to find catch-all mitigating factor and no submission by defendant Koedatich of any other mitigating factors, "Bey suffered an abusive childhood. His violent childhood, however, does not differ materially from that of ... Koedatich, ... who also had suffered from child abuse or other violence.”).
Again, however, the subjectivity of precedent-seeking review is revealed in the Court's analysis. Having acknowledged that the victim was struck from behind and rendered unconscious by the blows, and was therefore not aware of much of the brutality forced upon her, the Court states, “Even when the victim is not aware of impending death, as was the case in DiFrisco III, this Court has observed that ’at the end of the day there is still a victim, a [woman] who was [brutally] murdered____' " Ante at 314, 731 A.2d at 1140 (quoting DiFrisco III, supra, 142 N.J. at 205, 662 A.2d 442). What, then, is to distinguish defendant’s *392blameworthiness from other capital defendants? There will, at the end of the day, always be a victim who was brutally murdered, unless the Court means to assert that not all murders are brutal. The Court implies that it could never find a low level of victimization and, indeed, it never has.
Although I agree with the Court that defendant's record of prior convictions is both violent and lengthy, I oppose the Court's inclusion of the two robberies to which he confessed on the day of his arrest for Schnaps's murder. See ante at 314, 731 A.2d at 1141. Because defendant’s confession was not admitted into evidence in Harvey's retrial, the jurors had no knowledge of these alleged robberies and therefore could not have factored them into their sentencing decision. See Harvey II, supra, 151 N.J. 117, 699 A.2d 596.
The AOC may have based its placement of Godette in the prior murder category on the prosecution’s submission to the trial court of notice of the prior *396murder aggravating factor (c(4)(a)), but such reasoning would be misplaced. In order to make a c(4)(a) finding, the jury must find that "[t]he defendant has been convicted, at any time, of another murder ... [A] conviction shall be deemed final when sentence is imposed____" N.J.S.A. 2C:11 — 3(c)(4)(a) (emphasis added).
I agree with the majority that to compare defendant to all 126 cases classified as E cases would be impractical, see ante at 297, 731 A.2d at 1131; but the Court should, at a minimum, examine all the cases proposed by the State and *398the defendant to determine which are factually similar enough to defendant’s case to warrant inclusion in the analysis. Following that approach, I do not incorporate into my analysis several of the cases proposed by defendant, including one B-categoiy defendant (prior murder convictions) and all of his proposed C-categoiy defendants (sexual assault). Barring any misplacement of defendants in these categories, see supra at 396, 731 A.2d at 1185 ( discussing why Bcategoiy defendant Godette should be included in defendant's comparison class), I agree with the Court’s general finding that these types of cases are too dissimilar to defendant’s to contribute meaningfully to our proportionality review. See ante at 295, 731 A.2d at 1130.
The comparison is also problematic because Gerald’s crime involved multiple victims. The Court’s willingness to compare Gerald's case to Harvey’s in spite of its denial of defendant's request that other, A-category, multiple-victim defendants be included in the comparison, is suspect. See supra at 396, 731 A.2d at 1185. '
The Court's emphasis on Gerald’s childhood and its concurrent disregard for defendant's, in this context, is therefore all the more problematic. See supra at 390-98, 731 A.2d at 1182-86.
The Court perhaps tries to set the stage for its truncated analysis when it describes the nature of our proportionality review: “[Procedural or offender-oriented review presumes that the death penalty is proportional to the offense and focuses on the defendant, not the crime committed.” In such review, the question is "whether the punishment fits the criminal." Ante at 290, 731 A.2d at 1127 (citing Marshall II, supra, 130 N.J. at 129, 613 A.2d 1059) (additional internal quotations and other citation omitted).
Although certainly it is true that our review here is not a substantive one, that is, it is not offense-oriented as described by the Court in Marshall II, see 130 N.J. at 127, 613 A.2d 1059 ("[T]he substantive or offense-oriented proportionality review looks to whether the punishment of death is excessive for a particular offense."), our procedural review specifically incorporates not only the characteristics of the defendants, but also the circumstances of the crime. Such factors as the particular violence or terror with which the crime was carried out, the premeditation involved, and the defendant's knowledge of non-decedent victims are critical when we are examining the defendants' culpability. See ante at 376, 731 A.2d at 1173-74. The Court haips on these factors in its assessment of Harvey, see ante at 379-81, 731 A.2d at 1175-76, and then fails to discuss them at all with regard to most of the other defendants.
Does the Court really mean to suggest that if a murder were carried out with particular torture and depravity against a helpless victim, something so insignificant as the defendant's immaturity would justify a life sentence if the death-sentenced defendant under review carried out his murder with no torture or premeditation but was older? By failing to examine the crimes of the comparison cases here, the majority does exactly that. See ante at 318, 731 A.2d at 1143 (citing age as the only difference between Harvey’s murder and those of two defendants, Lance Phillips, age 20, and Charles Ploppert, age 24, whose crimes were significantly more brutal than defendant’s).
The Court minimizes Caviness’s culpability by asserting that the defendant’s own confession indicated that he did not have the baseball bat, his co-defendant did. Because the Court effectively asserts here that Caviness’s case should perhaps not be in the death-eligible universe at all because of an own-conduct problem, not only does the majority fail to "defer to the AOC’s judgment" in the categorization of comparison cases, see ante at 317-18, 731 A.2d at 1142, as it does in all other instances in this opinion, it calls into question Caviness’s sentence itself. If the defendant pled guilty under the assumption that he would face the death penalty if he went to trial, and the Court now questions his death eligibility, it must also question the voluntariness of Caviness’s plea. If the Court is to make this kind of challenge to Caviness's plea, it should remove Caviness’s case from the death-eligible universe. Having chosen to include him, it cannot then successfully argue that Caviness's lack of death-eligibility distinguishes his level of culpability from Harvey’s. Either Caviness is death-eligible for purposes of our review or he is not.
Here, the Court has clearly redefined the scope of proportionality review, rendering it a re-examination of the judge or jury’s sentence rather than a mere vehicle for quality-control. Accord Loftin II, supra, 157 N.J. at 426, 724 A.2d 129 (Handler, J., dissenting); see also Marshall II, supra, 130 N.J. at 223, 613 A.2d 1059 (Garibaldi, J., concurring in part and dissenting in part) ("Proportionality review is not a second appellate review nor a broad review of due process concerns---- It is a narrow concept directed to whether the defendant received a sentence disproportionate to that imposed on other defendants____").
Although the Court and/or the AOC may want to engage in this kind of detailed analysis for the purpose of defining the class of comparison cases, see, e.g., Chew II, supra, 159 N.J. at 254, 731 A.2d 1070 (Handler, J., dissenting) (arguing that Walter Williams should not be included in group of contract killer defendants because improper trial jury instruction led to jury’s finding of contract killer aggravating factor), it is entirely inappropriate to do so as part of its precedent-seeking review once the case has been included in the comparison group.
Although the Court suggests that Lee’s mental capacity might have been impaired due to a drug addiction when he committed his murder, see ante at 317-18, 731 A.2d at 1142 ("There were facts in ... Lee ... that ... would allow a jury to conclude that defendant’s capacity to appreciate the wrongfulness of his conduct was impaired by mental disease or defect or intoxication.”), the Court misconstrues the evidence. At trial, evidence was indeed presented that Lee was addicted to heroin; the defendant, however, broke into his victim's house with *406the intention of stealing money from her so he could buy heroin. The AOC narrative states, specifically, "Timothy Lee ... woke up feeling the need for drugs.” Therefore, the defendant was not on drugs when he committed the murder. There was no other evidence presented at trial of mental .disease or defect. In fact, the AOC narrative states, "D was in good health and has no mental health problems.” Significantly, - the defendant did not present any expert evidence at trial indicating that he was unable to control his actions because of his drug addiction, as, for example, defendant Walter Gerald had. The Court's conclusion, then, that Lee's mental state at the time of the murder distinguishes him from Harvey is highly flawed and affirms the application of disparate sentences for two similarly situated defendants.
The Court notes that "Herman Williams viciously shot his victims to death. However, Williams was characterized as "culturally retarded.” This characterization alone should not have the capacity to distinguish Harvey from Williams. It does not provide an excuse or justification in the way that a finding of mental disease or defect would. The testifying doctor stated clearly that Williams is "not mentally retarded but culturally retarded." Without further explanation of the meaning of this term, I fail to see that it alone should justify a 45-year term of imprisonment for Williams and a death sentence for Harvey.
Alphonso Bronson is the only minority defendant with a white victim who was not capitally prosecuted.
I believe defendant has in fact met this high burden here, but contend, in any case, that such a hurdle is not the appropriate one to raise.