NJ Finds Death Penalty Statistical Models Unstable
Baime finds death penalty statistical models unstable,screening procedures flawed

Padraic Cassidy
New Jersey Law Journal
May 10, 1999 

New Jersey's test for deciding whether the death penalty is imposed uniformly is unworkable because of the small number of cases reviewed and the unwieldy model used, a Supreme Court special master reported on Tuesday.

The conclusion, reached after consultation with experts at the behest of the justices last Feb. 1 in State v. Loftin, 157 N.J. 253, is that the Court should repudiate the "index-of-outcomes" test devised by court-appointed statistician David Baldus a decade ago.

In place of the current system, Baime recommended a reduction of the 55 aggravating and mitigating factors used in the proportionality review and an emphasis on the most salient factors.

"The vision of a mechanized approach to proportionality review was seductive," Appellate Division Judge David Baime wrote in a 114-page report. "Its promise was to extract human judgment from human decision making. Unfortunately, our experience with the index-of-outcomes test discloses that this was a promise unkept."

He added, "We have attained only a bitter semblance of efficiency by attempting to rely on these statistics and the calculation of chance."

Baldus, a University of Iowa College of Law professor whom the state Supreme Court hired in 1988 at a cost of $160,000, designed the index-of-outcomes test to compare cases that are factually dissimilar but nevertheless comparable as to the defendant's blameworthiness. It seeks to sort cases, using factors that appear to influence prosecutorial and jury decision-making, according to statistically relevant measures of culpability. The factors included statutory aggravating and mitigating circumstances; variables that reflected race, gender and socioeconomic status and other nonstatutory factors.

Baldus himself had cautioned the Court about the general uncertainty of many of the predictions yielded by his models. Yet, when they were first applied in State v. Marshall, 130 N.J. 139 (1992), the Court expressed confidence in his methodology, finding death penalty for contract-killer Robert Marshall proportional despite the minuscule pool of cases involving contract killings.

But with each proportionality review has come more Supreme Court questioning of the process and its variables. The consultants hired by Baime for his study -- David Weisburd, a criminologist at Hebrew University of Jerusalem, and Joseph Naus, a professor of statistics at Rutgers University in Piscataway -- said so many variables were included that the statistical models Baldus developed were unreliable.

"My recommendation is not intended to denigrate Professor Baldus," said Baime. "Much of the conceptual framework for proportionality review devised by the professor in 1991 had, and continues to have, great efficacy," Baime concluded. "Having said this, continued tolerance of the index-of-outcomes approach with its inferior statistical technique would represent an unfortunate example of willingness to sacrifice methodological rigor to overcome limitations inherent in the data."

New Jersey's death penalty law, enacted in August 1982, calls for a proportionality review along with a direct appeal, part of a three-stage process designed to ensure that the death sentence is not arbitrary. Proportionality review is carried out in 21 states by statute or by judicial initiative.

At present, the pool of cases in a proportionality review includes "clearly death-eligible" homicides, regardless of whether they are prosecuted as capital cases.

New Jersey juries have imposed 55 death sentences since 1982; the universe of death-eligible cases now stands at 433. Since 1988, the death penalty has been imposed an average of twice a year. At that rate, it might take 40 to 80 years of such cases to have enough data to show that index-of-outcome models are reliable, said Baime.

At this point, Baime said, the state has enough experience to discard its current model or at least revise it in favor of a frequency analysis of "salient factors."

Baime rejected the state attorney general's argument that the pool be limited to those in which a capital prosecution is begun, regardless of the outcome. "A death sentence may be disproportionate, at least in part, because prosecutors frequently exercise their discretion not to seek the death penalty for a particular type of homicide," Baime wrote. "A death sentence may be 'aberrant' or 'freakish' because in all factually similar cases, prosecutors engage in plea negotiations, thereby obviating the potential for capital punishment in those matters."

Lawrence Lustberg, speaking for the National Association for the Advancement of Colored People and the Association of Criminal Defense Lawyers, says the retention of the universe of cases is the most significant part of Baime's report. "We've argued that proportionality review would be completely meaningless if you used the universe the state recommended," says Lustberg, a partner at Newark's Gibbons Del Deo Dolan Griffinger & Vecchione.

In addition to his general findings and the inadequacy of the current system, Baime offered specific recommendations for remedying it.

As part of his inquiry, Baime said, he interviewed Administrative Office of the Court personnel connected with the screening process used to assemble data for proportionality review; reviewed all written protocols setting forth AOC policies and procedures; and conducted his own review of 105 cases to determine whether his screening decisions matched those of the AOC.

Calling the AOC's coding of cases highly subjective, its protocols vague and its checks on human error few, he said the AOC should:

  • Appoint a retired judge to serve as a standing master to supervise AOC case screening and data coding.
  • Adopt a rule to keep proportionality hearings before such a master secret.
  • Require that evidence of capital murder elements be "overwhelming" to be included in the death-eligible universe.
  • Adopt a modified salient factors test that accounts for mitigation.
  • Adopt on an experimental basis a questionnaire for judges to complete in all death-eligible cases.

A MODEL CHARGE ON RACE Though the Loftin case that prompted his study was a challenge to race-based imposition of the death penalty, Baime avoided a detailed discussion of race, saying the discovery of discriminatory patterns might be too difficult to detect even with the best model. But he recommended adoption of a model jury charge that reminds jurors to consider each case without regard to race, nationality religion or gender.

Baime also suggested that a proportionality review be conducted simultaneously with a direct appeal to streamline the process. "Whatever the deterrent value of capital punishment -- an issue upon which reasonable persons can disagree, it is surely diminished with the passage of time caused by endless appeals," he wrote.

Criminal defense attorneys support Baime's decision that the available pool of cases for proportionality review include, at a minimum, all cases in which a prosecutor files a notice of aggravating factors.

The state's plan would have omitted plea bargains and would have confined the proportionality review to cases in which the prosecutor sought the death penalty at trial.

Baime said that was incompatible with proportionality review, in part because prosecutors may not seek the death penalty for a particular type of homicide.

Lustberg says he won't quibble with Baime's suggestion to screen out cases that appear to be death-eligible on paper but that an experienced prosecutor would be unlikely to try.

Deputy Public Defender Claudia Van Wyk expresses similar sentiments. "The area that we are the most gratified by is the endorsement of a broad universe for proportionality review."

Prosecutors, too, found reason to applaud Baime's work.

Passaic County First Assistant Prosecutor Boris Moczula, who handled the proportionality review in State v. Marshall, says the Court should have done away with the complex statistical calisthenics long before Baime's report.

"The special master has confimed serious flaws in the methodologies, mainly by suggesting the rejection of the index-of-outcomes and the regression analyses," says Moczula.

Baime's study awaits the Supreme Court's review. Chief Justice Deborah Poritz has scheduled a hearing on the report for June 7 at 10 a.m. The Attorney General's Office, the Public Defender's Office, the ACDL and the NAACP are expected to argue their sides.

Copyright ©1999 NLP IP Company -- American Lawyer Media. All rights reserved.

Death Penalty Issues
Truth in Justice