Warren McCleskey Cross-Appellant v. Ralph Kemp, Warden, Cross-Appellee

RONEY, Circuit Judge, with whom Judges TJOFLAT, HILL, FAY, JAMES C. VANCE, ALBERT J. HENDERSON and R. LANIER ANDERSON, III, join *: This case was taken principally en banc argument arising to consider the in numer- capital ous cases that proof statistical Georgia capital shows the sentencing law being administered in an unconstitutionally discriminatory arbitrary and capricious and matter. After a lengthy evidentiary hear- ing which study by focused on a Professor Baldus, David C. the district court conclud- variety for ed of reasons that the statisti- support cal evidence was insufficient the unconstitutionality claim of in the death sentencing process Georgia. We affirm the district judgment court’s on point. this The en banc court has considered all the appeal. other claims involved on On appeal, the State’s we the reverse district grant corpus court’s habeas relief on the prosecutor claim that the failed to disclose promise of favorable treatment to a state Westmoreland, Mary Atty. Beth Asst. Giglio witness violation of v. United Atlanta, Gen., Ga., respondent-appel- for States, 150, 763, 405 U.S. lant, cross-appellee. L.Ed.2d 104 judg- We affirm the Ga., Atlanta, Stroup, denying H. John ment relief points Robert on all other Boger, Amsterdam, Anthony by defendant, (1) G. Charles raised the that is: Law, University-School of New York New defendant received ineffective assistance Judges Judge judg- *All Chief of the of the Court concur Godbold dissents from the judgment jury as to the death-oriented claim Giglio joins ment of the Court on the issue but and the ineffective assistance of counsel claim. opinion in the on all other issues. Johnson, Judges Tjoflat, join Judges Vance and Anderson in the Hatchett and Clark dissent opinion separately but each has written on the judgment from of the on the constitu- Georgia application of the constitutional application tional of the death sentence sentence. Judge Giglio and the issues each Sandstrom separately to concur Kravitch has written separate dissenting opinion. has written a portion opin- in the harmless error joins Giglio opinion ion but issue on all other issues. penalty hearing, counsel; (2) contra- neither side called wit- instructions process clause in violation of due documentary vened the nesses. The State introduced Montana, 442 McCleskey’s Sandstrom prior three evidence of convic- (1979); (3) robbery. tions for armed jurors death-scrupled exclusion of MeCleskey to death jury sentenced unbi- right impartial and to an violated police officer and to for murder representative from a jury drawn ased life sentences for the two consecutive community. cross-section robbery. These counts of armed convic- Thus, concluding that the district court by were affirmed tions sentences petition for writ of have denied the should Georgia Supreme McClesky Court. de- corpus, on all claims we affirm habeas State, 263 S.E.2d cert. 245 Ga. court, grant reverse the by nied but denied, 449 U.S. corpus Giglio on the claims. relief habeas (1980). MeCleskey peti- L.Ed.2d then corpus relief in state tioned for habeas FACTS petition denied court. This was after an MeCleskey was arrested and Warren hearing. Georgia Supreme evidentiary police of a officer charged with murder McCleskey’s application for a Court denied Fur- robbery of Dixie during an armed probable appeal. certificate of cause The store robbed niture Store. Court denied a United States through four men. Three entered band of MeCleskey petition for a writ certiorari. through the front. door and one back Zant, rear men in the of the store While the L.Ed.2d 631 cash, man entered searched who the show- through the front door secured MeCleskey petition then his filed for ha- forcing everyone there to lie face room corpus beas relief federal district court floor., Responding to a silent down on asserting, among things, other con- five alarm, by store police a officer entered the challenges ap- on stitutional at issue this the door. Two shots were fired. front peal. evidentiary hearing an and After in police struck One shot officer the filed consideration of extensive memoranda causing glanced his death. other head by parties, the court district entered pocket. in his cigarette lighter off chest a judgment from lengthy and detailed of MeCleskey by was identified two appeals MeCleskey which these are taken. personnel as the robber came store who Zant, (N.D.Ga.1984). F.Supp. arrest, Shortly door. after his front opinion This addresses each issue assert- MeCleskey participating confessed to (1) appeal following on ed order: that robbery but maintained he was not the claim, (2) constitutionality Giglio of eye- MeCleskey confirmed the triggerman. Georgia’s (3) application penalty, of en- that it was he who witness’ accounts counsel, (4) effective of death- assistance through the front door. One his tered (5) qualification jurors, and the Sand- Wright, accomplices, Ben testified that strom issue. MeCleskey shooting the officer. admitted MeCleskey jailA inmate near testi- housed MeCleskey “jail made fied that a house GIGLIO CLAIM in which he claimed he was the confession” granted district court habeas police

triggerman. The officer was killed MeCleskey corpus relief to because it deter a .38 bullet from caliber Rossi a fired prosecutor mined the state failed to that handgun. MeCleskey had stolen a .38 cali- of its reveal that one witnesses had been previous holdup. a ber Rossi as promised treatment a reward favorable PRIOR PROCEEDINGS due testimony. for his The State violates it process obtains conviction MeCleskey murder when convicted through At use of false evidence or on the robbery. and counts armed two testimony of a basis witness’s when that The Testimony Trial promise witness has failed to disclose trial, At the brought the State out on prosecution. treatment from favorable direct examination that Evans was incar- States, Giglio v. United charge cerated escape on the from a L.Ed.2d halfway federal house. Evans denied re- ceiving any promises prosecutor from the (1) promise hold We that there was no downplayed and seriousness es- case, contemplated by Giglio; cape charge. (2) event, Giglio had there been a Q: [by prosecutor]: Evans, Mr. have I violation, Thus, it would we be harmless. promised you anything for testifying grant corpus reverse the of habeas relief today? ground. on this sir, No, A: you ain’t. Evans, prisoner Offie Gene incarcerat- Q: escape You do charge have still McCleskey, ed with was called the State pending, is that correct? strengthen proof rebuttal on its Yes, one, A: got sir. really I’ve but triggerman McCleskey was the at the hold- escape, ain’t no what peoples out up. McCleskey Evans testified that admit- me, there tell something because went jail policeman to him in he ted shot the wrong just out so I there went home. said McCleskey he had worn I stayed at home and Iwhen called the makeup disguise his appearance during him man and told I that would be a robbery. in, coming placed little late he me on ” The “Promise escape charge and told me there back, wasn’t no coming use me corpus McCleskey’s At state habeas just stayed I home and he come hearing, gave following Evans account picked up. me conversations state offi- certain with Q: you hoping Are perhaps you cials. prosecuted won’t be escape? for that Evans, you THE Mr. me COURT: let ask Yeah, don’t, A: I I hope but I don’t— question. you At the time that testi- they me, what tell they going ain’t trial, McCleskey’s you fied in Mr. had charge escape me way. with promised anything exchange been Q: you try Have asked me to to fix it so your testimony? you get wouldn’t charged escape? No, THE I I WITNESS: wasn’t. wasn’t No, A: sir. nothing promised about—I wasn’t Q: I you try Have told I fix would to it promised nothing by the D.A. but the you? for told me that he would—he Detective No, A: sir. himself, going said was he to do it speak me. a word for That was what Corpus The State Habeas Decision me. Detective told rejected McCleskey’s state court Q: (by attorney): McCleskey’s The De- Giglio following reasoning: claim on the speak tective said he a for would word Mr. hearing Evans at the denied habeas you? promised that anything he was for his A: Yeah. testimony. He did state that he was told deposition prosecutor A of McCleskey’s by Dorsey Dorsey Detective would corpus that was taken for the state habeas ‘speak a word’ for him. The detective’s proceeding prosecutor con- reveals that parte ex communication recommendation federal tacted authorities after McCles- trigger ap- alone sufficient to trial key’s States, to advise them of Evans’ cooper- plicability of Giglio United escape charges and that ation were U.S. 150 S.Ct. 31 L.Ed.2d [92 104] dropped. trial, case, prom prosecutor petitioner’s Rus- In this the detective’s at speak ise to a falls far short of the Parker, that he was un- word stated

sel J. Giglio understandings and Na reached in understandings between any aware of Court, pue. As stated this thrust Depart- Police any Atlanta “[t]he Evans Giglio progeny to en its been regarding a favorable ment detectives the facts that sure that know on Evans’ to be made recommendation might giving motivate a witness testimo charge. Mr. escape Parker ad- federal Kemp, Smith ny.” 715 F.2d opportunity for there was mitted — denied, (11th U.S.—, Cir.), cert. good to put detectives word Atlanta (1983). The 78 L.Ed.2d de- with federal authorities. for Evans marginal tective’s statement offered such a However, he further stated that when benefit, Evans, as indicated it is any been killed and police officer has doubtful it would motivate reluctant wit- State, up testifying someone ends for ness, or that disclosure of statement danger, his it is not sur- putting life any would have had effect on his credibili- charges, against like prising that those ty. The nondisclosure therefore State’s Evans, dropped. will be infringe McCleskey’s process failed to due evidence, any other In the absence of rights. agreement ex- cannot conclude merely subsequent isted because Any Violation Harmless? Was charges disposition against a of criminal event, In there no “reason witness for the State. able likelihood” that the State’s failure conclude Although is reasonable to cryptic disclose the detective’s statement or state found that there was court escape Evans’ different scenario affected agreement pros- no between Evans and Giglio, See judgment jury. ecutor, specific finding was made as to no at Evans’ credi U.S. promised to Evans’ claim that a detective bility exposed impeach substantial merely him.” The “speak a word for court ment even without the detective’s state assuming held as a matter law that description ment and the inconsistent his truth, Giglio telling vio- Evans was prosecutor began escape. The his direct lation had occurred. by having examination Evans recite lita It a Promise? Was past ny of convictions. Evans admitted to forgery, burglaries, lar convictions two impos- rationale for Court’s ceny, carrying weapon, concealed ing jury’s rule is that estimate “[t]he mail. theft from United States On reliability giv- of a the truthfulness and examination, McCleskey’s attorney cross determinative of en witness well be attempted portray “profes Evans as a Illinois, Napue guilt innocence.” criminal”. Evans also admitted that sional testifying protect he was himself and provided The Court has never McCleskey’s light codefendants. one guidance definitive when the Govern- evidence, impeachment this substantial dealings prospective witness ment’s unlikely find the it that undisclosed we they credibility that so affect witness’ jury’s would have affected the information Giglio, must at trial. In be disclosed See credibility. Evans’ assessment prosecutor promised the defendant’s al- Anderson, 1347, United v. States F.2d leged co-conspirator charges no would that (5th Cir.1978). if he testified brought against be him Napue, prose- against McCleskey the defendant. In Evans’ testimo claims promised exchange only cutor a witness that in other testi ny was crucial because for his prosecutor pulled trigger rec- testimony mony would which indicated he codefendant, Wright. his Ben that was came from ommend the sentence witness testimony, McCleskey urges, Wright’s presently serving be Ben reduced. would have been insufficient under Evans’ testimony McCleskey that had law to him without corrobora- up made his face convict corroborated the identifi- provided by Georgia, tion Evans. testimony cation eyewitness- one accomplice’s testimony alone in felony Nevertheless, es. this evidence was not cases is insufficient establish a fact. crucial to the State’s case. That McCles- § Wright’s O.C.G.A. testimony, 24-4-8. key wearing makeup helps to establish however, by McCleskey’s was corroborated he was the robber entered who the furni- own McCleskey confession which admit- through ture store the front door. This participation robbery. ted Ar- See already fact had directly been testified to State, nold v. S.E.2d Ga. by McCleskey’s accomplice eyewit- and two Corroboration need not extend nesses as well by as corroborated McCles- every State, material detail. Blalock v. key’s own That confession. Evans’ testi- (1983); 250 Ga. 298 S.E.2d 479-80 mony eyewitnesses’ buttresses one of State, Ga.App. 304 S.E.2d Cofer relatively unimportant. identifications is Thus, although testimony might Evans’ thought The district court Evans’ testi- regarded important well in certain mony of the critical because information he respects, the corroboration of that testimo- supplied makeup McCleskey’s about and ny was such that the revelation of the shooting police intent in officer. Al- Giglio promise reasonably would not affect though agree testimony we his added jury’s credibility assessment of his case, weight prosecution’s to the we do not therefore would have had effect on the find that it could “in reasonable likeli- jury’s decision. grant' The district court’s judgment hood have affected the of the corpus of habeas relief on this issue must jury.” Giglio, 405 U.S. 92 S.Ct. at be reversed. Illinois, (quoting Napue 360 U.S. at 1178). Evans, 79 S.Ct. at who was rebuttal, called testified that CONSTITUTIONAL APPLICATION OF McCleskey him that had told he knew he GEORGIA’S DEATH PENALTY out, way had to shoot his and that even if In challenging the constitutionality of policemen there had been twelve he would application capital statute, of Georgia’s thing. statement, have done the same This McCleskey alleged grounds related two for prosecutor argued, showed malice. In (1) relief: that the “death is admin- however, closing argument, prose- his arbitrarily, capriciously, istered and whim- presented cutor to the three reasons sically (2) Georgia,” the State of supporting a conviction malice murder. imposed “is pursuant pattern ... to a First, argued physical he that the evidence practice grounds ... on the discriminate showed malicious intent because it indi- race,” Eighth both in violation McCleskey police cated that shot the offi- Fourteenth Amendments the Constitu- cer once in the head and second time tion. dying lay the chest as he on the floor. petitioner’s The district granted court Second, prosecutor asserted that evidentiary hearing motion for an choice, on his McCleskey either had a to surren- system-wide claim of racial discrimination der or to kill the That he officer. chose to Equal under the Clause Third, Protection prosecutor kill indicated malice. Fourteenth Amendment. court noted McCleskey’s contended that statement appears petitioner’s “it ... have way Evans he still would shot his Eighth argument Amendment has been re police out if there had been twelve officers jected Spinkellink showed malice. Circuit This statement McCles- (5th Wainwright, 578 612-14 key developed length during was not at F.2d Cir. 1978) testimony only petitioner’s Evans’ was ... Fourteenth and mentioned [but] passing by prosecutor closing appropriate in in claim for the ar- Amendment gument. in the of statistical consideration context stages, petitioner proposes manslaughter aggravated to are which the less

evidence mitigated more than the of October at and white-victim present.” Order disposed in cases similar fashion. hearing held in Au- evidentiary An was petition- in chief was The district court concluded that case gust, 1983. Petitioner's carry testimony two er failed to his ultimate burden of through the presented witnesses, persuasion, there David Bal- because is consistent expert Professor C. Woodworth, statistically significant as as evidence that George Dr. well the dus and witnesses, imposed penalty being is principal lay Edward Gates death the basis two particular Warr, by of race of In employed an official the defendant. and L.G. statistically significant Georgia of Pardons and Paroles. there was no evi- Board prosecutors testimony produced two to show that offered the dence state witnesses, seeking juries Joseph penalty Katz and Dr. the death or are expert Dr. rebuttal, imposing penalty re- Roger petitioner In the death because the Burford. Dr. defendant is or the victim white. called Professor Baldus and Wood- black is worth, presented expert study incap- conceded that and further testi- Petitioner is mony demonstrating singled that he from Dr. Richard Berk. able of was out for because of the opinion, reported comprehensive In a victim, and, race of either himself or his F.Supp. court conclud- district therefore, petitioner failed to demonstrate petitioner make ed that failed to out that racial considerations caused him to in sen- prima case discrimination facie penalty. receive the death the race of tencing based on either victims following approach dis- in adopt or the race of defendants. We ad- dressing argument disparities shown the Bal- that the district counted study ground refusing erred in hold dus on the research court to (1) unconstitutionally ap- in data substantial flaws statute showed base, revealing coding plied light as shown tests the statistical evidence. First, briefly errors and between items on we mismatches describe statistical study Baldus done this case. Study (PRS) Reform Procedural and (CSS) Second, Comprehensive Sentencing Study we evidentiary discuss the value (2) accuracy establishing and questionnaires; lacked such studies have the ulti- models, primarily be- mate facts that showed flaws control constitutional de- Third, do not cause the models measure decisions cision. we discuss the constitutional knowledge proved to law in based on available decision- terms of what must be predicts petitioner prevail argu- maker and outcomes 50 order for to on cases; (3) percent capital punishment and demonstrated that a ment state law is variables, multi-collinearity among unconstitutionally applied model because of race Fourth, interrelationship among showing the varia- discrimination. we discuss wheth- consequently distorting study generalized relation- er a statistical such bles as making interpretation prove ships, difficult. this could ever be sufficient allegations necessary of ultimate fact The district court further held that even sustain a successful constitutional attack established, prima if a case had been facie Fifth, on a sentence. defendant’s we dis- successfully the state had rebutted whether, study prove cuss this is valid to (1) showing because: the results were Sixth, purports prove. what it we de- product good statistical methodolo- particular study, assuming cide that its (2) gy, explanations other for the it validity proves what claims to demonstrated, as, results could be such prove, require insufficient either acting proxies white victims were for support petitioner. decision aggravated cases and that black-victim cases, summary, left we the district cases, (3) being affirm court black-victim cases, (3) that, assuming the being ground validity cases left on the black-victim research, voluntary support it would not behind at the life sentence and

«87 being law was victim decision white cases in circumstances, all would or unconstitutionally applied, much less a black defendant in all cases. compel finding, such which level object of study The the Baldus in Fulton petitioner to reach in to would have order County, where McCleskey convicted, was prevail appeal. on this was to determine the sentencing whether Study

The Baldus pattern disparities were observed respect with statewide to race of the victim analyzed imposi- study Baldus and of pertinent race defendant were tion of sentence in cases deter- homicide County, Fulton and whether the evidence disparities mine the level of attributable to concerning Fulton County any shed light imposition race the rate of on McCleskey’s Warren death sentence the first study, death sentence. Proce- sentence, aberrant or whether (PRS), dural Reform re- Study results racial considerations have played a effects vealed race-of-defendant whatso- disposition role in the his of case. ever, and the results were unclear at that only there were ten cases in- stage as .Because effects. race-of-victim volving police officer victims in Fulton study, Charging The second and Sen- County, statistical analysis could not be tencing Study (CSS),consisted of a random effectively. utilized Baldus conceded that sample persons stratified of all indicted for it was difficult to draw inference con- through from murder 1979. The cerning the overall race effect in these study examined cases from indictment cases only because there had been one through sentencing. purpose of death sentence. He concluded that based racial was to estimate effects that the data there possibility product were combined effects that a racial factor existed McCleskey’s point decisions from the of all of indictment case. point final death-sentencing Social Science Research Evidence decision, strength and to include To some extent a broad issue before this evidence cases. Court concerns the role that social science study attempted to control for all of judicial is to decisionmaking. have So- play capital the factors which into a crime field is a broad-based consist- cial science system, aggravating such as circumstanc- ing many specialized discipline areas, es, circumstances, mitigating strength of psychology, anthropology, such as econom- evidence, period imposition time sen- ics, science, political history and sociology. tence, (urban/rural), geographical areas Sperlich, Social Science Evidence and Cf. race of and defendant and victim. The Reaching Beyond Courts: the Adviso- collection ex- data for these studies was Process, ry 63 Judicature 283 n. 14 ceedingly complex, involving cumbersome consisting parametric Research instruments, collection data extensive field nonparametric measures conducted by multiple work data collectors and so- laboratory under both controlled situations phisticated computer coding, entry and conditions, and uncontrolled such as real cleaning processes. data situations, life throughout observational completed Baldus and mul- Woodworth objectives the disciplines. The broad titude statistical tests on the data con- social science research are to better under- sisting regression analysis, indexing fac- stand mankind and its institutions in order tabulation, analysis, triangu- effectively plan, predict, tor cross modify to more society’s lation. The results showed a racial enhance 6% individual’s victim, systemwide effect for white black circumstances. Social science as a nonex- always an increase defendant cases with to 20% act science is mindful that its re- sug- mid-range dealing highly complex of cases. There was no search beha- uniform, gestion patterns that a institutional bias vioral and institutions that exist in adversely best, highly society. affected defendants technical At existed 888 court____ society materials to the of the “models” and “reflects” attention

research society trends and infor- did not the data provides argue with Brandéis were and existed____ generalizations. valid, for broad-based only they mation main the conclu- intent is to use The researcher’s extra-legal contribution ... to make plan, de- predict, research to from sions readily the court.” data available to Id. scribe, modify. To explain, or understand position This Court has taken a so- to research conclusions from such utilize play a role in cial science research does specific be- specific intent of a explain the decisionmaking in certain judicial situa- legiti- goes beyond the situation havioral tions, light even the limitations of Even when uses for such research. mate such research. Statistics have been used exactness, high at a research is level addressing primarily cases discrimina- results, read- social scientists design and tion. con- steadfast hesitancies to ily admit their explain specific be- results can clude such analysis Statistical is useful in a certain situation. havioral actions terms, In evidentiary facts. statisti show on are cal studies based correlation circum potential is judiciary aware They evidence. not direct stantial evi (1) inherent in such research: limitations States, dence. Teamsters v. United 431 (2) discipline; nature imprecise 1843, 1856, U.S. 97 S.Ct. 52 presented potential inaccuracies (1977). 396 studies L.Ed.2d Statistical do data; (3) potential research- bias of the purport to state the law is in what er; (4) problems inherent with the applied to (5) given situation. The law is training methodology; specialized as revealed research. and com- facts to assess utilize the data needed (6) debatability petently, and examined, In this case realities based empirical for courts to use appropriateness data, set on a certain of facts reduced decisionmaking. Henry, In- evidence Cf descriptive characteristics were Journey A into the Future— troduction: persons being numbers of sentenced to Empirical Evidence in Devel- The Role of reveal, Georgia. death in studies Such Law, 1, 4; 1981 U.Ill.L.Rev. oping Labor through circumstantial evidence their at 283 n. 14. Sperlich, 63 Judicature results, analyses possible, probable, Historically, beginning “Louis Bran- relationships that may exist in the realities empirical use of evidence before déis’ studied. persuasive ... social sci- The usefulness of statistics obvious to the presented evidence has been ence ly depends upon attempted is what Forst, Wellford, courts.” Rhodes & Sen- proved by disparate impact them. If tencing and Social Science: Research for proved, sought to be statistics are more Guidelines, 7 the Formulation Federal impact than if of that useful the causes (1979). L.Rev. 355 Muller v. Hofstra See proved. intent and must be Where motiva 412, 324, 52 Oregon, proved, must be the statistics have tion (1908); L.Ed. 551 v. Board Edu- Brown utility. less even This Court said 483, cation, U.S. 98 L.Ed. 347 cases, however, “that discrimination while Brandéis, (1954). presented 873 brief usually alone cannot in statistics establish judicial social facts as corroborative in the discrimination, under certain lim tentional O’Brien, decisionmaking process. Judi- Of they might.” Spencer ited circumstances Myths, cial Motivations and Justifica- Zant, (11th 715 1581 F.2d Cir. Postscript tions: A on Social Science 1983), pet. reh’g reh’g en Law, 64 Judicature for Cir.1984). banc, (11th 729 F.2d 1293 See The Brandéis brief “is a well-known tech- Valley v. Tennessee also Eastland Au asking judicial nique for court to take (11th Cir.1983); 704 F.2d thority, Sperlich, facts.” Judica- notice of social Inc., Ben’s, F.2d n. v. Uncle ture at 31. “It does not solve Johnson denied, (5th Cir.1980), problem bring of how scientific 421 valid cert. sion in University Houston, Wilkins v. (5th These limited 1981). circumstances are where the 654 F.2d 388 Unit A Cir. racially dispropor statistical evidence of Wilkins this Court held that “[mjultiple impact strong permit regression tionate is so as to analysis a relatively sophisti- other inference than that the results are cated means determining the effects in product racially discriminatory the of a number of different factors have Balkcom, purpose. particular tent or v. on a See Smith variable.” Id. at 402-03. (5th B), denied, 671 This F.2d 858 Cir. Unit cert. Court noted that the methodology “is 882, 181, subject 459 to U.S. 103 S.Ct. 74 L.Ed.2d 148 misuse and thus em- must be (1982). ployed with great care.” Id. at 403. Pro- cedurally, regression when multiple is used Statistical evidence has been re “it will be the subject expert testimony ways. ceived in two The United States knowledgeable and cross-examination from Supreme recognized Court the simply manner, both sides. In validity the studies existence statistical and social the and significance model the of its results making science research in certain deci fully trial, developed allowing will be at sions, being subject without such studies judge trial to make an informed decision as rigors evidentiary hearing. of an Mul probative value the analysis.” Id. 412, 324, 28 52 Oregon, ler v. 208 U.S. S.Ct. this, Having Court, done the Wilkins in an (1908); L.Ed. v. 551 Fowler North Caroli employment case, discrimination held “the na, 904, 3212, 428 U.S. 96 S.Ct. 49 L.Ed.2d statistical evidence associated with mul- (1976); Carolina, 1212 v. North Woodson tiple regression inconclusive, analysis is 280, 2978, 944 428 U.S. 49 L.Ed.2d raising questions more than it answers.” (1976); Texas, 262, 428 96 Jurek v. U.S. Id. 2950, (1976); S.Ct. Proffitt if Even strong statistical evidence is Florida, 428 U.S. 49 generally there is a need for additional (1976); Gregg Georgia, L.Ed.2d 913 428 In evidence. Mississippi Cooper- Wade v. S.Ct. L.Ed.2d 859 Serv., (5th ative Extension 528 F.2d 508 (1976). Court, “Supreme example, for Cir.1976),the results drawn from the multi- opposition encountered severe and criticism regression analysis variate were supported rulings on desegregation public to its by additional evidence. Id. at 517. schools, rule, exclusionary and the Wade the statistics did “stand alone” decisions, retroactivity precisely of its be proof as the sole of discrimination. empirical court general cause the relied on O’Brien, ization.” The Seduction the Much has been written about the rela- Courts, tionship of Social and law and Judiciary: Science social science. “If 8,19 (1980). In each Judicature of these social produce required science cannot answers, cannot, beyond probably situations Court “focused” and it use its is specifics likely the case it to the “insti disjointed before to continue to lead to a represented Daniels, through specific tutions” incrementalism.” Social Science ruling changes Cases, Penalty Pol’y effected in the institutions. And Death Law & hand, may Q. 336, (1979). On the other statistical evidence “Social science can presented through court di probably greatest be trial make its contribution testimony theory legal by investigating rect and cross-examination on the causal legislative judicial, statistical information that bears on an is forces and ad- behind carefully decisionmaking by probing sue. Such evidence is examined ministrative relevancy, subjected general to the tests of effects of such decisions.” Na- authenticity, probativeness credibility. gel, And The What Law Social Sciences: Contribute?, A.B. Henry, 1981 U.Ill.L.Rev. at 8. Science Can Social Cf. A.J. 357-58 difficulty One with statistical evidence observations, questions it ac- raise more than With these this Court cepts for answers. This Court reached that conclu- social science research what (Marshall, J., concurring); claim for it. As in 2790-91 id. at should scientist social cases, n. the infer- 389-90 92 S.Ct. at n. 12 evidence 2803-04 all circumstantial C.J., dissenting); (Burger, from drawn the statistics id. ences J., factfinder, (Powell, dissenting). but the statistics are at 2833 accepted the circumstances. to show later, years ap- Four Discrimination, Death Penal- Racial pursu- proved the redrawn statute ty, and the Constitution McCleskey to which and sen- ant tried 153, 96 Gregg Georgia, tenced. his McCleskey contends death sentence is At the Georgia’s death unconstitutional because approved time same the Court statutes discriminatorily on the applied which, and Texas like Geor- from Florida and the of the race defendant basis gia, guided approach, followed a discretion constitutional victim. Several different sentencing mandatory invalidated the but asserted. for this claim have been bases procedure of North Louisiana. Carolina and McCleskey arbitrary, capri- relies on the 242, Florida, v. 428 96 S.Ct. U.S. components pro- Proffitt and irrational cious 2960, (1976); 49 L.Ed.2d 913 v. Tex- Jurek punishment in of cruel and unusual hibition as, 262, 2950, 428 U.S. 96 S.Ct. 49 L.Ed.2d pro- Eighth equal Amendment and the (1976); Carolina, v. 929 North Woodson Amend- tection clause of Fourteenth 2978, 280, 428 U.S. 944 96 S.Ct. 49 L.Ed.2d with thought ment. The district court that 325, (1976); Louisiana, Roberts v. 428 U.S. respect to discrimination race-of-the-victim 3001, 96 S.Ct. 49 L.Ed.2d 974 petitioner properly more stated a claim process clause of the Four- under due consistently Since we have held Gregg, teenth Amendment. that to state claim of racial discrimination application capital of a constitutional kind are asserted Claims seldom statute, alleged. and motive intent must be they degree particularity, 316, v. 317 Wainwright, 721 F.2d Sullivan pre- generally several constitutional assert (statistical (11th Cir.1983) impact studies however, cepts. analysis, On there seems system insufficient to show state “inten- proof in to be little difference that tionally against petitioner”), discriminated prevail any might required to under be denied, 464 execution stay petition three theories. 109, 450, L.Ed.2d 210 78 104 S.Ct. U.S. 238, Georgia, 92 v. 408 U.S. Furman (1983); 709 F.2d Wainwright, Adams v. 2726, (1972), the 33 L.Ed.2d 346 Su- S.Ct. 1443, (11th Cir.1983) “a (requiring 1449 Georgia preme Court struck down the showing of an intent to discriminate” Amend- penalty system Eighth death impact strong of disparate so ... “evidence grounds, ment with several of the concur- only permissible is one that inference' holding oper- ring justices system that the discrimination”), of intentional cert. de- arbitrary capricious in an manner ated —nied, —104 79 S.Ct. way no to dis- because there was rational Balkcom, (1984); L.Ed.2d Smith tinguish the few cases death which was (5th B) (requiring F.2d Cir.Unit many in imposed from the it was not. which “circumstantial or statistical evidence (White, J., at at 92 S.Ct. Id. racially impact so disproportionate ... 309-10, concurring); id. at 92 S.Ct. at strong permit that other the results (Stewart, concurring). Al- 2762-63 J. of a they product inference are the but though imposi- race discrimination racially discriminatory purpose”), intent or penalty of the death was not the basis tion denied, cert. 459 U.S. 103 S.Ct. decision, it con- was one of several L.Ed.2d 148 concurring in both the cerns addressed 249-52, Initially dissenting Spinkellink Wainwright, opinions. See id. at (5th Cir.1978), denied, concurring); F.2d 582 (Douglas, at cert. 2731-33 J. 309-10, (Stewart, at 440 U.S. id. at 2762-63 364-65, (1979), rejected Eighth the Court and Four- concurring); id. 92 S.Ct. at J. opposite teenth Amendment claims that would Florida lead to conclusions on being was in a applied death dis- system’s constitutionality depending on criminatory fashion on the basis of the which theory a claimant asserts. Spinkellink victim’s race. Court read companion

Gregg holding Eighth and its cases “as A successful Amendment if a state follows a drawn properly challenge require proof would race in'imposing statute penalty, the death then factor operating system was such capriciousness the arbitrariness and pervasive fairly manner that it could be —and therefore the racial discrimination con- irrational, system said that the arbi demned in Furman —have conclusive- been trary capricious. For the same rea ly Spinkellink removed.” Id. at 613-14. sons that the Baldus would be insuf can automatically read to foreclose be ficient to discriminatory demonstrate intent challenges Eighth capital all Amendment or unconstitutional discrimination sentencing facially conducted under a con- context, Fourteenth Amendment would In Godfrey Georgia, stitutional statute. irrationality, insufficient show arbi 420,100 capriciousness trariness and under (1980), sustained Eighth kind of analysis. Amendment Eighth Georgia to a challenge Amendment The district court stated that were it because sentence court’s writing slate, on a clean would character- portion facially of a that construction ize claim McCleskey’s process as a due way left no to principled valid statute dis- claim. The took position court that tinguish death penalty the cases where the McCleskey’s argument, while couched imposed was from those in which it was “arbitrary capricious,” terms of and funda- Wainwright, not. v. 685 F.2d See Proffitt mentally that contended death Cir.1982). 1227, (11th 52 Neverthe- 1261 n. applied on the basis of a mor- less, Godfrey neither nor under- Proffitt prior mines ally impermissible this Court’s subsequent and criterion: the of the race pronouncements in Spinkellink, Smith, victim.

Adams, and regarding Sullivan theory district court’s derives disparate amount impact that must be support some from Court’s shown either Eighth under an Amendment Stephens, 862, decision in U.S. Zant v. 462 equal protection or analysis. (1983). 103 S.Ct. 77 L.Ed.2d 235 out, pointed court As the district here recognized there that a Court state analytical such a indicates an nex- standard “aggravating” attach the as an label Eighth us claims between Amendment capital sentencing element to factors protection equal a Fourteenth Amendment constitutionally impermissible are that Zant, F.Supp. claim. v. McCleskey totally process, to sentencing irrelevant (N.D.Ga.1984). an Eighth Where done, such race. If that were general- Amendment claim centers around said, process require “due would that the disparate impact ized showings racial impose jury’s decision to be set capital sentencing, such in- a connection is at —, aside.” Id. 462 U.S. 103 S.Ct. at escapable. Although conceivably the level 2747, 77 L.Ed.2d 255. From this lan- disparate impact or amount of racial that process it is clear that due would guage sentencing a state’s capital would render making a prevent state from explicitly arbitrary system capricious under the aggravating murder of a white victim an Eighth might slightly Amendment differ capital sentencing. circumstance in But disparate from the level or amount of racial neutral, facially a due where statute impact compel inference that would by process supported proof claim must-be equal pro- discriminatory intent under the state, through jur- prosecutors, its tection clause Fourteenth Amend- ors, ment, judges, implicitly not need has attached the we do to decide whether magnitude aggravating there to race. could be difference label context, intent, purpose, and motive had charac petitioner if Even necessary components. Washington due as one under the his claim terized 238-42, Davis, clause, altered 96 S.Ct. it would not have process showing 2046-49, showing A governing 48 L.Ed.2d 597 legal standard application impact alone is not prevail. disproportionate he must make un- prove discriminatory clause is “an uncertain intent process of the due sufficient ‘fun can must discover what reasonable inference be enterprise which less no other particu consists of 429 U.S. at Arlington Heights, fairness’ damental drawn. considering any rele 264-66, first This Circuit lar situation 97 S.Ct. at 562-64. by assessing the and then precedents consistently applied principles these vant are at stake.” Lassi F.2d Wainwright, several interests Adams v. law. — Services, 452 denied, Social Cir.1983), Department (11th ter v. cert. 2153, 2158-2159, 24-25, 101 S.Ct. —, U.S. U.S. process Due also L.Ed.2d 640 (1984); Wainwright, 721 F.2d Sullivan risk that the requires the assessment Cir.1983). (11th lead to errone being used will procedures We, therefore, proof of a hold that Eldridge, decisions. Mathews ous impact alone is insufficient disparate 893, 903, system, un capital sentencing invalidate a process re a due claim (1976). Where great that it disparate impact is so less that to determine whether quires a court system compels a conclusion impermissibly affected the victim race of irrational, arbitrary capri unprincipled, sentencing process, decisions capital purposeful discrimination— cious such clause, charac equal protection under the i.e., intentionally being used as a race is to the Fourteenth as “central terized sentencing presumed factor —can discriminatory prohibition Amendment’s system. permeate State,” Mitchell, 443 Rose v. action 2993, 2999-3000, 554-55, 545, 99 S.Ct. U.S. and the Statistical Studies Generalized (1979), certainly “rele are 61 L.Ed.2d 739 Standard Constitutional in the assessment of the precedents” vant Thus, as in decisions. risk of erroneous initially as to question arises context, the claimant equal protection the study suggesting a any statewide whether theory present process must under a due of a disparity application racial capi that in the establishes evidence which support a could ever state’s motivating sentencing process “is a tal race on a defendant’s sen constitutional attack Village Arling in the decision.” factor in the sta The answer lies whether tence. of De Metropolitan Housing v. Heights ton of the study is sufficient evidence tistical 252, 266, 97 Corp., 429 U.S. velopment which must be shown. ultimate fact L.Ed.2d 450 50 S.Ct. 858, Balkcom, 671 F.2d In v. Smith denied, 882, process (5th B), and cruel and unusual Due cert. 459 U.S. Cir.Unit normally 181, (1982), on cases do not focus this punishment L.Ed.2d 148 S.Ct. governmental actor. But intent of the the said: claimed, discrimination is not where racial instances, or sta- circumstantial In some procedural the faults or flaws on basis racially disproportion- tistical evidence of law, in but on the basis the structure the may strong that impact be so ate process, of' the made within that decisions but no other inference permit results intent and motive are a natu purpose, then racially product of a they are the that discrimina component proof ral purpose. discriminatory intent or actually tion occurred. some apparently caused This statement as a often cited it is clearly confusion because Court has not stand. it does proposition \vhich prove constitutional claim of held that to study argues that his statistical Petitioner equal protection in discrimination the racial (citations there strong inference that is a 709 omitted). shows F.2d at 1449 Here based on race. is the disparity That again, in commenting on strength the step, step first however. The second focus- evidence, the court was referring not to great disparity how Once the es on is. or quality amount of evidence which proven, whether- disparity question is disparate impact, showed a but the amount compel disparity is sufficient a con- to impact disparate strong would be so discriminatory that it clusion results from inevitably as to lead to finding of motiva- purpose. key prob- intent and intent, tion and absent some other explana- proof, in principle lem lies that the disparity. tion for the disparity strong, matter how of some commenting In proffer Bal- alone insufficient. case, study dus in another Justice Powell 578 F.2d Spinkellink Wainwright, v. wrote in from stay dissent of execution (5th 582, Cir.1978), denied, 440 cert. pending en banc consideration of this case: 976, 1548, U.S. S.Ct. 59 L.Ed.2d If the Baldus is similar (1979), petitioner Florida claimed the several studies filed with us in Sullivan being applied statute a discriminato- Wainwright, 464 U.S. ry against murdering fashion defendants 450,78 (1983), L.Ed.2d 210 statistics whites, blacks, opposed as violation kind, many studies of date which punishment cruel and unusual as far merely general back components equal protection of the Consti- surveys hardly statistical that are partic- Evidence was in- disparity tution. of this respect any alleged ularized with “in- through expert troduced witnesses. The Surely, tentional” racial discrimination. argument court assumed for sake of no contention can be made accuracy that the entire petitioner’s re- statistics but Georgia judicial system, levels, op- jected Eighth argument. all Amendment rejected ar- equal protection Argu- The court erates to discriminate all cases. gument disparity because the shown ments to this effect have di- been petitioner’s prove statistics could racial- type rected to the of statutes addressed ly discriminatory purpose intent or as re- v. Georgia, Furman 408 U.S. 238 [92 Davis, quired Washington 426 U.S. As 346] (1976), clear, our subsequent cases make such Village Arlington Heights v. Metro- arguments cannot be taken un- seriously politan Housing Development Corp., 429 approved Gregg. der statutes L.Ed.2d 450 1027,1030, Kemp, n. Stephens (1977). 578 F.2d at 614-16. 104 S.Ct. 564 n. L.Ed.2d F.2d Wainwright, In Adams v. (1984) (Powell, J., n. 2 dissenting). — Cir.1983), (11th denied, U.S.—, cert. from lesson these and other cases (1984), 104 S.Ct. generalized must be that statistical studies court, denying hearing, in evidentiary an in deciding par- of little use whether a accepted arguably statistics which tended ticular defendant has been unconstitution- support to the claim that the Florida ally sentenced to death. As to whether imposed disproportionately attack, system can survive constitutional involving white The court cases victims. probative statistical at most are studies then said: disparity present, how much is is a but it impact to Disparate alone is insufficient legal question to disparity as how much is a violation of fourteenth establish accept required a federal before court will showing There amendment. must be a it as evidence constitutional flaws discriminate____ Only if of an intent to system. disparate impact the evidence of is so point especially This critical to a

strong becomes only permissible infer- request is one of court faced for an evidentia- ence intentional discrimination with ry produce hearing will it alone suffice. to future studies which independent made. Needless to variables. We hold in this undoubtedly be will hearing evidentiary would be neces- case that if statistical say, even results are particular that a sary accepted valid, evidence to hear as the evidence fails chal- to against be- was discriminated lenge defendant successfully constitutionality general But cause race. statistical of his Georgia system. Because of this deci- here not even the kind offered do studies of sion, necessary it is not for us to determine fact. prove that Aside from purport to court was right whether the district or evidence, however, it would kind wrong faulting study. of the Baldus its evi- necessary to conduct a full not seem court undertook an district extensive hearing as to studies which do dentiary presented. review of the research It re- unexplainable show an nothing more than ceived, analyzed complex and dealt with the ap- studies would disparity. Generalized court is to statistics. The district be com- every hope excluding pear have little outstanding mended for its endeavor in the might possible make a differ- factor handling aspects of the detailed of this defendants, exclu- ence crimes between case, particularly light of the consistent the extent there is a sive race. To arguments being made several cases component subjective judgmental to the Any study. on the based Baldus decision a sentence is invest- discretion with which study justify that the the Baldus results of ed, no two defendants be seen will corpus have habeas relief would to deal sentencers, by the but no two identical findings district court’s as to the single precisely will see a case sentencers study itself. Inasmuch as social science recognized, court has the same. As the by appellate research has been used courts racially neutral varia- there are “countless decisionmaking, Oregon, Muller sentencing of capital in the cases. bles” 412, 419-21, 324, 325-26, Balkcom, F.2d at 859. Smith v. (1908), L.Ed. 551 and has been tested like general to recede from This is not trial, other kinds of evidence at see Spink- may re- proposition that statistical studies Wainwright, ellink v. 578 F.2d 612-13 disparity great inevitably flect so (5th Cir.1978), question there as to the disparity that the re- lead to a conclusion finding a trial court’s standard of review of sults from intent or motivation. As decid- complex study. on highly based statistical opinion, ed the Baldus studies dem- system that the does not onstrate Findings of fact are un reviewed disparity required contain level of clearly der the erroneous standard which meet that constitutional standard. has defined as: “[a] Study Validity the Baldus finding ‘clearly erroneous’ when al research The social science of Professor it, though support there is evidence to reveal, through statisti- purports Baldus reviewing court on the entire evidence is disparities sentencing cal analysis, left with and firm the definite conviction in white victim cases in black defendants that a committed.” mistake been if Georgia. A is valid it measures Gypsum United States v. United States purports what to measure. Different Co., 525, 542, validity. different levels of studies have L.Ed. 746 validity study the is level of the degree directly related to the to which the impact disparate re Whether rely findings can on the social scientist flects an is ulti intent to discriminate measuring what study the it claims to fact under mate which must be reviewed measure. clearly the erroneous standard. Pullman- Swint, 456 U.S. Standard v. study court held the to be The district Pullman, In perceived invalid because of errors models, base, the that in Court said Fed.R.Civ.P. data the deficiencies the 52(a) multi-collinearity existing and the between exceptions purport

does not make penalty. to the Because the pro- Constitution categories exclude certain of factual find- hibits the consideration of racial factors as ings obligation from the justification of a court of penalty, for the McCleskey as- appeals accept a district court’s find- serts that the discernible racial influence ings clearly unless erroneous. It sentencing does on operation renders the of the categories; partic- not divide facts into Georgia system infirm. ular, it does findings not divide of fact addition, McCleskey asserts that the into those that deal with ‘ultimate’ and race-of-the-victim influence on the system those that deal ‘subsidiary’ facts. is particularly significant in range the 456 U.S. at 102 S.Ct. at 1789. cases involving intermediate ag- levels of gravation (mid-range aggravation cases).

There would seem to be two levels of argues He findings that because his based on case fell statistical evidence within range, he must has established first, finding imper- be reviewed: the con- missible racial cerning validity itself, operated the considerations study of the his second, case. the finding of ultimate fact based

upon the circumstantial evidence revealed We assume without deciding that study, if valid. study Baldus sufficient to show what it purports to reveal as to application

The district study court here found the Georgia penalty. death invalid. The court Baldus found the statistics of con- cluded study that his study system- showed that particularly be troublesome atic disparities and substantial base, the areas of the data existed in models and penalties imposed upon the relationship homicide independent between de- fendants in Zant, Georgia based variables. on race McCleskey F.Supp. victim, homicide (N.D.Ga.1984). disparities that the pretermit We existed at a less review of this substantial rate death finding concerning the sentenc- validi- ing defendants, based on ty of race of study itself. The district court the factors of went on to race of the victim hold that even if the and de- statistics fendant were at work in validly County. did reflect Fulton Georgia system, ultimate fact of intent to discriminate was general A comment about the limitations proven. not We finding review this of fact on what study purports Baldus by assuming the validity study and show, although covered in the subsequent holding rest our on the decision that the discussion, may helpful. The Baldus valid, study, even if only supports not study statistical evidence does purport judge’s district decision clearly under the to show that McCleskey was sentenced to review, erroneous standard compels but because either his race or the race it. of his victim. It group shows that in a Sufficiency Study Baldus whites, involving blacks and all of whose same, cases are virtually the there would that, McCleskey argues although the receiving be more blacks penalty the death post-Furman statute in yields now than whites and more murderers of whites predictable results, more the race of the receiving the death than murderers significant, is a victim imper- but of course of blacks. The guess” statisticians’ “best missible, factor which accounts for the im- is that race was a factor those cases and position penalty many of the death cases. sentencing has a role in structure in Geor- supports argument He sophis- with the gia. general These statements about the that, ticated Baldus statistical after results are legal insufficient to make a controlling legitimate for the factors that determination. analysis An must be made might rationally explain imposition as to how much disparity actually shown penalty, purportedly reveals significant by the research. race-of-the-victim system; influence the i.e., things all other being equal, Accepting figures, white the Baldus not the but victim crimes likely general conclusion, are more to result in accurately reflecting the race on the of the statistics are victim’s overall likelihood Georgia experience,

the given McCleskey penalty on the death in a case. Never- to relief inadequate entitle theless, figures enlighten all the serve to claim. his constitutional system operates. us on how the somewhat retrospective study Georgia-based The average figure composite The is of all 6% sample random of of stratified consisted aggravation cases contains both low and 1,066 indicted for mur- cases of individuals cases, penalty the never where is almost der-death, voluntary man- murder-life and race, imposed regardless of the victim’s arrested were between slaughter who cases, aggravation and where both high December 1973 and March likely to white and black victim crimes are from compiled 41-page The data were penalty. figure result in the is When consisted of more than questionnaire and classify eases accord- related to tables 500,000 Through complex statisti- entries. ing aggravation, of the level the 6% analysis, relationships examined cal Baldus average figure properly is seen as an variable, dependent the death-sen- between containing in which aggregate both cases rate, variables, tencing independent and is a factor race of the victim discernible factors, aggravating mitigating nine and 75 is and in which it not. those background controlling while for factors. evidence, penalty McCleskey’s of the cases a trial was and the evidence 10% state, held, presented by of the cases the also showed that defendants 5% death. factor diminishes as were sentenced to race-of-the-victim more are the model. variables added to study Georgia subjects data to a figure example, For the bottom line was analyses, under multitude statistical models, very simple dropped in the 17% signifi- a statistically each method there is model, finally the 230-variable 6% effect operating cant race-of-the-victim fell to the final were when 20 variables 4% difficult, however, It is more statewide. added effect of Georgia and the magnitude dem- ascertain effect review considered. study. Baldus onstrated The sim- unadjusted figures enlightening show ple, that death statistics also imposed operation legitimate were fac- sentences the white overall 11% potentially eligible supporting cases tors victim for death sentence. penalty, an ration- eligible essentially death Baldus revealed 1% system, high After al cases controlling aggravation victim cases. which black legitimate likely death factors could ex- were more to result in the various differential, aggravation As plain Baldus still sentence than low cases. concluded significant expect system, one in a fac- that there was a race-of-the-vic- would rational multiple effect. The result of most tors such as torture and victims tim Baldus’ model, receiving McCleskey pri- increased the greatly conclusive on which likelihood of relies, marily .06, penalty. signi- showed an effect fying average that on a white victim crime important There are dimensions that the likely more to result in the sentence 6% cannot reveal. testified statistics Baldus comparable victim than a black crime. Bal- system penalty death provided dus also tables that showed extremely complicated process in which signifi- race-of-the-victim effect to be most group single factor or of factors deter- involving in cases cant intermediate levels given mines outcome of a case. No aggravation. cases, In these on aver- could, single petitioner on the basis age, white crimes were victim shown to be alone, these that he re- statistics establish in the likely pen- more to result 20% because, and ceived the death sentence alty equally aggravated than black victim because, his victim Even was white. crimes. cases, race- mid-range where the strong, influence is figures None of the mentioned is a of-the-victim said to be above quantification legitimate justifying factors definitive of the influence of are, by very grounds). the definition the mid- A the plurality of Court in Ford range, present case. definitively in each stated it that had held “in two prior cases that the statistical evidence re- The statistics there is a race-of-the- show upon by support lied Ford to his claim of imposition the relationship victim discrimination was not sufficient to raise a enough in the death sentence discernible ground upon substantial which relief might significant to in statistically cases be the granted.” at —, be Id. S.Ct. at system magnitude as a whole. The cannot (citing L.Ed.2d at 912 Sullivan and Ad- any given called in determinative case. be ams). Sullivan, petitioners Adams, The in study The evidence in the Baldus seems all the study by and Ford relied on Gross support penalty sys- the death Mauro and of the death sys- Florida a operating tem as one in rational manner. figure tem. bottom line in the Gross factor, Although single or combination study and Mauro indicated race-of-the-vic- factors, irrefutably will lead to the effect, quantified by tim a “death odds mul- case, every system opera- in the in sentence tiplier,” of 1. Using about 4.8 to a similar pattern legislature tion follows the the in- methodology, Baldus a death obtained odds tended, Supreme which the Court found multiplier Georgia. of 4.3 to 1 in constitutional in and sorts out Gregg, cases It is of course possible that the Supreme according aggravation, to levels of as rejecting Court was methodology of the gauged legitimate factors. The funda- study, Florida rather than its bottom line. Eighth concern mental Amendment of Fur- It is true methodology that the of the Bal- man, Gregg, as discussed in which states study superior. posture dus is meaningful that “there is no basis for dis- cases, however, persuades Florida this tinguishing in which few cases [the Supreme Court that Court was not imposed is from the many death sentence] relying inadequacies on in the methodology in which it is not accurately not” does study. Florida in issue Sulli- operation the Georgia describe the stat- van, Adams, and Ford whether the ute. U.S. at petitioner’s proffer had raised a substantial Taking the line bottom revealed 6% ground evidentiary sufficient to warrant true, figures figure in the Baldus as is hearing. context, In that is reasonable presumption not sufficient to overcome suppose Supreme that the Court looked in a operating statute is constitu at the bottom line indication of racial effect any discretionary manner. sys tional held simply and that it was insufficient to tem, imprecision tolerated, some must be contrary assumption, a claim. A state simply study and Baldus insufficient Supreme analyzed extremely Court support a ruling, context of a complicated study Gross and Mauro operating intended, that is much as statute rejected methodological grounds, it on racial playing factors are role much less reasonable. system outcome sufficient to render Thus, assuming Supreme Court arbitrary capricious. whole Sullivan, Adams and Ford found the supported, possi- This conclusion is line study bottom and Mauro Gross bly compelled, even recent claim, insufficient to raise a constitutional opinions in Wainwright, Sullivan would compelled we reach the same L.Ed.2d 210 analyzing sufficiency result (1983) (denying stay of execution to allow comparable bottom the Baldus line evidentiary Eighth hearing on Amendment McCleskey on which relies. statistics); supported by Wainwright claim — Adams, McCleskey’s U.S.—, argument height- about the (1984) (vacating ened influence of fac- stay); L.Ed.2d 809 the race-of-the-victim —Ford, —, requires Wainwright mid-range 104 tor of cases (1984) analysis. (denying McCleskey’s somewhat different application stay range state’s to vacate other case falls within the of cases involv- aggravation. “close cases” and therefore that the death levels ing intermediate penalty will in be set aside “close cases.” study tended to show statistical The Baldus relationship to race-of-the-victim the previously, As discussed the statis greater was in these sentencing outcome that the tics cannot show race-of-the-victim involving very in cases low or than cases case, operated given in factor even in the high aggravation. very levels Rather, mid-range. the statistics show that, average, on the race-of-the-victim fac effect increases The race-of-the-victim likely tor more to affect the outcome penalty by ap- likelihood of the death the mid-range in cases than in those cases at mid-range in the of cases. proximately 20% high spectrum the and low ends of the figure appro- this analysis of Some 20% in aggravation. The statistics alone are priate. McCleskey’s sufficient to sen show. figure The in this case is not analo- 20% tence was determined the race of his figure reflecting percentage gous to a the victim, the or even that race of his victim jury composition in a case. Such disparity imposition to the contributed represents disparity figure the actual in his case. minority persons the number of on between McCleskey’s petition does not surmount jury and the of such the venire number stating the threshold a claim on burden contrast, population. persons statistics, from this issue. Aside he disparity purport in case does not 20% presents literally might no evidence that disparity. Rather, fig- an actual be support a tend to conclusion that the race reflects that the variables included in ure McCleskey’s any way victim in motivat- adequately explain do not impose ed the the death sentence in disparity and that the statisticians can 20% his case. only by assuming the racial ef- explain Conclusion importantly, fect. More Baldus did not tes- Court has held that to be tify significance that he found statistical constitutional the sentencer in death sen- disparity figure mid-range for 20% tence cases must have some measure of cases, adequately explain and he did not Gregg Georgia, discretion. 428 U.S. rationale of his definition of the mid- (1976); 49 L.Ed.2d 859 testimony range of cases. His leaves this Prof- Florida, 428 U.S. unpersuaded that there is a rational- fitt mandatory 49 L.Ed.2d 913 classified, ly well-defined class of cases death sentence statutes were declared un- it can be demonstrated that a race- which constitutional. Woodson v. North Caroli- operating of-the-victim effect with a na, S.Ct. L.Ed.2d magnitude approximating 20%. (1976); Louisiana, Roberts v. however, Assuming arguendo, v disparity fig

that the is an accurate 20% very exercise of discretion means ure, apparent disparity it is that such a persons exercising discretion cases, mid-range only in the and not in the dupli- reach different results from exact whole, system provide as a cannot the basis Assuming cates. each result is within the systemwide challenge. previously for a As discretion, range of all are correct discussed, system operat as a whole is eyes law. It would not make sense manner, ing in a rational and not system require exercise of fairly arbitrary manner that can be labeled facially in order to discretion constitu- capricious. system challenge A tional, valid system time hold a same only against mid-range cannot be made application where that unconstitutional purport of cases. Baldus did not to define discretion achieved different results for cases; mid-range duplicates, a defi appear nor is such what to be exact absent possible. simply showing nition It is not satisfac the state the reasons for the dif- narrow, tory say operates that the in ference. The discretion is focused racial effect ways and one the most basic directed, there is measure of which but still through it achieves the task is criminal discretion. against laws murder. however, would approach, The Baldus 153, 226, Gregg Georgia, 428 U.S. different results take the cases with (1976) facts, duplicate to be contended what are J., (White, concurring). not other- could be where the differences dif- and conclude that the explained, opinion wise The of the plurality Gregg Court on was based race alone. ferent result noted: petitioner legal perspective, would From a petitioner’s argument is nothing difference is not ex- argue since the that more than veiled contention that a Fur- which the social scientist

plained by facts indirectly capital punish- man outlawed explain satisfactory to the differ- thinks by placing totally ment unrealistic condi- ences, that the prima is a case there repair tions on its use. In order to facie on unconstitutional difference was based alleged pointed petition- to defects factors, would shift to and burden er, necessary require to would be prove the difference results state to charge capital prosecuting authorities a This from considerations. constitutional arguably offense there whenever had the realities. It not approach ignores capital they a been murder and that re- differences in cases: ignores quantitative plea bargain fuse to with the defendant. education, looks, personality, profes- age, jury If a even though refused to convict demeanor, remorse, sion, clothes, charge, and job, supported the evidence its few, incapable it is of just name a but verdict would to be reversed to have and measuring qualitative guilty differences of such verdict of entered or a new trial ordered, discretionary and fac- act things aggravating mitigating since the of as are, fact, dupli- permitted. no nullification not There exact would be tors. capital Finally, clemency and defend- acts of executive would capital cates in crimes system, prohibited. have to be type of research submitted here Such ants. The course, totally alien to our of the factors would be no- tends show which directed to effective, justice. tions of criminal is of restricted use were but undirected factors control showing what at 199 at 2937 n. Id. n. constitutionally required the exercise Powell, Stewart, Stevens, (opinion of and discretion. JJ.). recognized Gregg decid-

It when was broadly, seem that the Viewed it would justice system not capital here, ed would presented statistical assum- evidence not perfect that it need perfect, be but be ing validity, its rather than confirms con- constitutional. Justice White past order be system. In state demns where documented, said: discrimination is well effect, no discrimination as argued, that no showed Petitioner has marginal dis- penalty race defendant. how effective death matter government, parity based the race of the victim tends punishment, as a be humans, support the state’s contention run as it must created and differently system working is far from the inevitably incompetent to administer it. pre- which condemned. accepted proposition one Furman cannot be This rhyme or Imposition days, no rea- the Furman there was law. constitutional penalty surely got as to the death an awesome re- son who penalty now, majority system of who did not. But vast sponsibility justice cases, the a difference are reasons for participate in it. Mistakes will who those they That so well documented. and discriminations will occur be made of the cases is explain. percentage clear in a small How- will be difficult which entire^system un- reason to ever, society’s most is no declare basic tasks one of constitutional. protecting the lives its citizens properly rejected The district court need not be examined because there was aspect McCleskey’s claim. obviously prejudice. question is whether it was un

INEFFECTIVE ASSISTANCE subpoena reasonable not to the robbery

OF COUNSEL McCleskey’s victims as defense witnesses. attorney primarily relied on an alibi de McCleskey his trial counsel ren- contends defense, fense at trial. To establish this dered assistance ineffective both attorney put McCleskey on the stand. guilt/innocence phases of his He also called several witnesses an at trial in violation Sixth Amendment. tempt to discredit a Dixie Furniture Store Although a defendant is consti employee’s McCleskey identification of tutionally reasonably entitled to effective McCleskey’s to show that confession was attorney, assistance from his we hold that involuntary. It would have undermined his McCleskey preju has not shown he was if attorney defense had called witness by the diced claimed defaults in his coun testify es to as to which robber did the performance. sel’s Ineffective assistance shooting. prejudice No by can be shown warrants reversal of a conviction failing subpoena witnesses as a reason probability when there is a reasonable strategy able decision. attorney’s errors altered the outcome McCleskey’s attorney could have proceeding. A court decide an reasonably prepared to cross-examine the ground ineffectiveness claim on the of lack expert by reading State’s ballistics the ex prejudice considering without the rea pert’s report. in-person No interview was attorney’s performance. sonableness necessary. Watkins, Washington See — U.S.—, v. Washington, Strickland 1346, (5th Cir.1981), 655 F.2d 1358 cert. 2052, (1984). 104 S.Ct. denied, 949, 2021, 456 U.S. 102 S.Ct. 72 trial, guilt phase As to the of his McCles- L.Ed.2d 474 report inwas key (1) attorney claims that his failed to: prosecutor’s file attorney which the re prisoner interview who testified that viewed and no contention has been made gave confession; McCleskey jail (2) house that he did not read it. subpoena interview and as defense witness- sentencing phase trial, As to the of his es the victims of the Dixie Furniture Store McCleskey attorney asserts his failed to (3) robbery; and interview the State’s bal- investigate and find character witnesses expert. listics object and did not to the State’s introduc- McCleskey preju demonstrates no prior tion of convictions which had been set dice caused his counsel’s failure to inter aside. view Offie We Evans. have held there was No character witnesses testified no reasonable likelihood that the disclosure McCleskey for at his trial. At the State of the detective’s statement to Offie Evans corpus hearing McCleskey’s habeas attor would have affected the verdict. There is ney testified he talked with both McCles then probability” no “reasonable that the key potential and his sister about character attorney’s failure to discover this evidence They suggested possibilities. witnesses. affected the verdict. The sister testify refused to and advised attorney that their mother was too sick victims, robbery As to the McCles McCleskey to travel to the site of the trial. key in-person does not contend that an and his sister took the stand the State something interview would have revealed corpus hearing conflicting their oppor statements did He habeas and told not. had an tunity to stories. It is clear from the state court’s cross-examine several of the rob bery opinion attorney: investigating victims and that it believed the officers at McCleskey’s preliminary hearing. The rea Despite conflicting evidence his attorney’s sonableness of investigation point, ... the Court is authorized its

901 jurors Because they as fact finder to conclude that Coun- both indicated would role not under inquiries necessary to circumstances consider im- sel made all posing penalty, they proper- the death during the were present adequate defense Indeed, ly Witherspoon Illinois, excluded under sentencing phase. Counsel could 510, 1770, 391 U.S. 20 L.Ed.2d not exist. 776 present evidence that did (1968) Holman, . See also Boulden v. Although “finding of fact” is stated U.S. 22 L.Ed.2d conclusion, im- legal ultimate terms of the (1969). Their peti- exclusion did not violate plicit in that conclusion is the historical rights tioner’s Sixth Amendment to an im- attorney’s testimony finding that the partial, community-representative jury. Jarvis, F.2d Paxton v. credible. See Balkcom, Smith v. 660 F.2d 582-83 (11th Cir.1984); 1306, 1308 Cox v. Mont- (5th 1981), denied, Unit B Cir. cert. (11th Cir.1983). F.2d 1036 gomery, 718 pre- finding of fact is entitled to a This (1982); Spinkellink Wainwright, Based on the sumption of correctness. (5th Cir.1978), F.2d 593-94 cert. de- attorney, he facts as testified nied, investigation a reasonable conducted L.Ed.2d 796 character witnesses. aggravating As evidence of an cir THE SANDSTROM ISSUE three prosecutor introduced cumstance rejected The district court McCleskey’s sentences, resulting in life all convictions claim that the trial court’s instructions to set aside on Fourth had been which jury deprived on the issue of'intent him grounds. This evidence could Amendment process by shifting prose- of due from the prejudice, result in undue because proving cution to the defense the burden overturned, convictions were although the beyond a reasonable doubt each essential dropped charges were not and McCles element of the crimes for which he was key pleaded guilty and received sentences burden-shifting tried. Such is unconstitu- years. 18 The reduction sentence was Montana, tional under 442 Sandstrom v. at trial. disclosed 39 properly court denied relief The district ineffectiveness of counsel claim. on the McCleskey objects following por- to the

tion of the trial court’s instruction to the jury: DEATH-ORIENTED JURY says One section of our law that the acts Petitioner claims the district court person of a of sound mind and discretion jurors improperly upheld the exclusion of presumed product opposed adamantly capital

who were will, person’s person and a mind sound punishment. According petitioner, this presumed discretion to intend the and by right his to be tried exclusion violated probable consequences of his natural and from impartial and unbiased drawn an acts, presumptions may but both of these representative cross-section of his com be rebutted. pe munity. support proposition, of this analysis instruction opinions district court In its of whether this titioner cites two Sandstrom, under Grigsby Mabry, circuits. v. was unconstitutional from outside pan- recent (E.D.Ark.1983), the district court examined two F.Supp. hearing 1273 569 Circuit, ordered, (8th opinions this Franklin v. No. 83-2113 E.A. el banc en Cir.1983), Francis, (11th (March 15, 1984) 1983), 720 F.2d 1206 argued Cir. Nov. — U.S.—, 104 Garrison, F.Supp. granted, S.Ct. v. 578 1164 cert. and Keeten (1984), rev’d, (4th and v. Fran- (W.D.N.C.1984), 742 129 Tucker F.2d (11th Cir.), cis, pet. Cir.1984). 723 F.2d 1504 on the merits of those Whatever for banc, 723 F.2d 1518 controlling authority reh’g reh’g and en opinions, they are not Cir.1984). jury in- (11th though Even for Court. 902 identical, evidence,” upon than two cases were rather a four-jus

structions created a language plurality that the held Franklin held tice that one two tests persumption viola- mandatory employed by rebuttable for harmless error this Cir held that tive of while Tucker guilt Sandstrom cuit—whether the evidence is so permissive than a infer- created no more overwhelming that erroneous instruc ence and not violate Not- did Sandstrom. tion could not have contributed to the ing challenged portion of the in- that the jury’s inappropriate. verdict—is Connecti trial McCleskey’s struction used Johnson, 73, 85-87, cut v. 103 corresponding “virtually to the identical” 969, 976-978, L.Ed.2d S.Ct. 823 charges in portions Franklin The fifth vote to affirm was by added Tucker, court to follow the district elected Stevens, jurisdic Justice who concurred on pro- Tucker Court’s most recent as this grounds. tional Id. at 103 S.Ct. at 978 issue, held that on and it nouncement (Stevens, J., judgment). concurring charge not violated Sandstrom was however, justices, Four other criticized the on intent. plurality adopting for an “automatic rever decision, court’s the en Since the district for sal” rule Sandstrom error.. Id. at argument banc court heard in several (Powell, J., at 983 dissenting). cases in an to resolve the constitu effort subsequently Court has re burden-shifting tionality potentially in another viewed case which harmless er structions identical to the one at issue here. applied ror doctrine was to a Sandstrom Zant, (11th 721 F.2d 1478 Cir. Davis split evenly violation. once 1983), banc, en pet. reh’g reh’g on for again affirming opinion without a Sixth (11th Cir.1984); 728 F.2d Drake holding prejudi Circuit decision that “the Francis, (11th Cir.), pet. F.2d 990 on of a cial effect Sandstrom instruction is banc, 727 reh’g reh’g en F.2d for for largely function of the defense asserted (11th Cir.1984); Francis, Tucker v. Koehler, Engle at trial.” 707 F.2d (11th Cir.), pet. reh’g 723 F.2d 1504 (6th Cir.1983), equally di aff’d banc, (11th reh’g en F.2d —court, U.S.—, vided Cir.1984). Supreme The United States (1984) curiam). L.Ed.2d 1 (per 80 In En- argument oral in Court has heard Frank distinguished Sixth gle, the Circuit between Francis, (U.S. lin v. 53 U.S.L.W. 3373 Nov. violations Sandstrom where defendant 20, 1984) However these [No. 83-1590]. in nonparticipation has claimed the crime decided, cases for purpose this where and those the defendant has claimed decision, here we assume that the intent rea, holding lack mens that instruction in this case violated Sandstrom prejudicial was so never latter to consti- proceed issue of whether that tute harmless error. Id. Until the Su- error was harmless. preme controlling Court makes a decision requires that “before on question, the harmless error we contin- a be federal constitutional error can harm- apply propounded ue to standards less, the must declare court be able to our earlier cases. beyond belief a rea- that was harmless Since was Sandstrom decided California, Chapman sonable doubt.” v. 1979, analyzed this Circuit has unconstitu 24, 828, 386 87 17 S.Ct. burden-shifting tional instructions to deter recently, L.Ed.2d 705 More Su- they mine whether constituted harmless er preme divided the issue of Court has over See, Balkcom, e.g., ror. Mason v. 669 F.2d whether the error is doctrine harmless (5th 1982). 222, 227 Cir. B Unit Lamb v. applicable burden-shifting presump- ever (11th Cir.1982), 683 F.2d Jernigan, 1332 Reasoning tions violative Sandstrom. denied, 103 on a cert. presumption erroneous dis- “[a]n (1983), puted element of the crime irrele- 75 L.Ed.2d 496 the Court identified renders vant the evidence on the issue two situations which an unconstitutional because jury may upon presumption burden-shifting might relied be have instruction harm

903 discussing First, may test. Before whether intent erroneous instruction less. trial, however, guilt McCleskey’s at issue in if evidence of was harmless have been error could we that intent is an essential element overwhelming that so note was charged. decision he jury’s to the of the crime with which have contributed was not 1342; Lamb, provides F.2d at Ma three ele- 683 law essential to convict. (1) before son, 227. In the case the offense of malice murder: 669 F.2d at ments to finding that homicide; (2) its aforethought; court based malice us, the district on (3) was harmless Jernigan, violation v. the Sandstrom unlawfulness. Lamb has decided on ground. This Circuit this F.2d at 1336. The “malice” element means overwhelming evi provoca- occasions kill several the intent to absence viola guilt renders Sandstrom dence The erroneous instruction on tion. Id. Balkcom, v. intent, tion harmless. See Jarrell therefore, an essential ele- involved 1242,1257 (11th Cir.1984); Brooks F.2d charged, criminal offense ment of the (11th Francis, 793-94 Cir. 716 F.2d v. required prove the exist- the state was reh’g en 1983), reh’g and pet. on beyond element a reasonable ence of that for for Cir.1984); Spenc (11th banc, F.2d 1358 358, 364, Winship, In re 397 U’S. doubt. (11th Zant, Cir. 715 F.2d v. er reh’g en 1983), reh’g and pet. on question therefore becomes whether for for Cir.1984). (11th banc, 729 F.2d McCleskey conceded the element of intent by presenting a defense that admits Second, instruction the erroneous alleged intentional. the act where the instruction may harmless be that is not on an element the burden shifts course, a defendant in a crimi Of Lamb, at 1342. 683 F.2d at trial. at issue rely entirely presump nal trial on the adopted this rationale This Circuit and the tion of innocence State’s burden harmless. See violation find a Sandstrom beyond the crime proving every element of (11th Francis, 727 F.2d Drake a reasonable doubt. Connecticut John reh’g en Cir.), reh’g and pet. 16,103 son, for at 978 n. 460 U.S. at 87 n. (11th Cir.1984); banc, F.2d 1003 Col case, determining whether a In such a Francis, F.2d 1330-31 lins intent had the issue of defendant conceded Cir.1984), de (11th pet. reh’g en banc impossible. for (11th The record re might well Cir.1984). There nied, 734 F.2d 1481 however, veals, McCleskey chose not plurality in even the indication that is some Rather, he took the to take that course. endorse v. Johnson Connecticut would trial and testified that he was stand at cir harmless error limited type Dixie Furniture Store participant cumstances: killing of robbery resulted in the which harmless error Sandstrom McCleskey’s The end of [A] Schlatt. Officer conceded the issue if intent____ defendant the summa testimony on direct examination such presenting In a defense defense: rizes his alibi self-defense, alibi, a de- insanity, or as Q. you at the Dixie Furniture Were cases admit that may in some fendant day? that Store was alleged by prosecution the the act A. No. reducing intentional, sufficiently thereby Q. anyone? you Did shoot jury applied that the the likelihood No, A. I didn’t. permit to as erroneous instruction you said the Q. everything have Is consider the error court appellate truth? harmless. (citations A. Positive. at 978

omitted). attorney McCleskey’s argument, closing In client’s alibi defense. again stressed his reveals that of the record Our review undermining the credi- ren- He concentrated in this case is violation Sandstrom that eyewitness identifications bility under this second error dered harmless McCleskey as the pinpointed triggerman CONCLUSION questioning and on motives the other judgment district court in who robbery participants had testified that granting petition for writ of habeas McCleskey the fatal had fired shots. corpus petition is reversed and the is here- attorney emphasized McCleskey’s that by denied. McCleskey if Mr. the front of the REVERSED and RENDERED. McCleskey store and Mr. had the silver gun gun police if killed the and the silver TJOFLAT, Judge, concurring: Circuit officer, guilty. would be But then he opinion, I though concur in the court’s I that is not the circumstances that have question approach would of the consti- proven. been application tutional of the death McCleskey’s attorney’s Although. argu- Georgia differently.' I somewhat would be- with the alibi testi- ments were consistent gin proposition with the established that McCleskey himself, mony offered Georgia’s capital sentencing model is facial- jury testimony chose to disbelieve and ly constitutional. It contains the safe- testimony eyewit- rely instead on the guards prevent necessary arbitrary participants , nesses other capricious making, including decision deci- robbery. sions motivated the race of the defend- hold that in the safeguards We therefore ant or the These victim. asserting his present every stage capital course of alibi defense of a murder effectively McCleskey prosecution Georgia, conceded the issue grand from the intent, rendering jury through thereby indictment the execution of .»Sand beyond the death strom violation harmless a reason sentence. Some of these safe- holding, guards repeating. imply able In we do not are worth doubt. so a defendant raises a whenever defense At stage, indictment accused can violation on of alibi a Sandstrom an intent impanel grand insist that the State automatically or malice instruction ren represents a fair cross section of the Nor do suggest dered harmless error. we community, required by as the sixth and specifically argue that defendant must amendments, fourteenth and that the State intent did not exist in order for the issue of deny group, racial violation of the jury. intent remain before But equal protection clause of the fourteenth presented where the State has overwhelm amendment, right participate jur- ing killing evidence of intentional ors. In repre- this means that a where the defendant raises a defense of portion sentative of blacks will be on the nonparticipation in the crime rather than grand jury. rea, cf. lack mens Sandstrom violation safeguards same come play into on an intent instruction such as the one at petit the selection jury. of the accused’s beyond issue here is harmless reasonable addition, challenge can the accused Francis, doubt. See 728 F.2d at Collins cause found venireman to harbor a 1331; Koehler, Engle 707 F.2d at 246. against racial bias the accused or his vic- In this case the entered officer and made tim. The peremptorily accused can excuse it almost to the middle store before and, jurors suspected such bias at the he a was shot twice with .38 caliber Rossi time, prevent prosecutor same from revolver. The circumstances this shoot- exercising peremptory challenges his in a ing, coupled McCleskey’s decision to way systematically particu- that excludes a rely defense, an elevate alibi to mere persons, blacks, on lar such class of from speculation any See, Zant, scenario would create jury e.g., service. Willis v. — (11th Cir.1983), on the denied, reasonable doubt issue intent. F.2d 1212 cert. —, properly district court denied habeas U.S. corpus relief on issue. VANCE, jury, Judge,

If the sentencer is the as it is in concurring: Circuit (the judge being Georgia by trial bound Although Judge I concur in Roney’s opin recommendation), it can jury’s be instruct- ion, I am troubled its assertion that put aside racial considerations in ed proof there is “little difference in the reaching sentencing its recommendation. might required prevail” be under either sentence, eighth If the recommends the death amendment or fourteenth amend equal protection accused, ment claims of the Georgia kind appeal on direct to the presented According Furman, here 1. Court, Supreme challenge can his sentence eighth inquiry amendment centers on independent assign- grounds on racial as an general capital results of sentencing in propor- ment of error or the context of systems, and governed by condemns those And, tionality if the court affirms review. unpredictable chance, such factors as ca sentence, he can his chal- his death renew price equal protection or whim. An inquiry lenge petition rehearing by way in or very different. It system centers not on attack. of collateral irrationality, ic independent but rather the assessing validity In the constitutional intentional, evil of invidious discrimination scheme, sentencing Georgia’s capital one against given individuals. argue could that the role of the federal I am conscious of the dicta in the various on courts—the certiorari opinions disap- Furman which note with Georgia Supreme from the Court and proval possibility that racial discrimina- judicial system in habeas cor- entire federal application tion was a factor in the pus considered. For review—should be penalty death under the and Texas layer they provide still another of safe- mind, in my statutes then effect. To how- guards against arbitrary capricious ever, merely possi- such dicta indicate the imposition penalty. bility system permits of the death that a the exer- cise of standardless discretion not Petitioner, attacking in his conviction and capricious, may give play but to discrim- sentence, death makes no claim that either inatory equal protec- motives which violate any stage was motivated a racial bias given tion standards as well. Whether a prosecution. of his criminal His claim eighth set of facts make out an amendment solely transpired in stems from what has systemic irrationality claim of under Fur- prosecutions. homicide To the extent other therefore, is, question entirely man inde- data of cases in which the that his consists pendent of whether facts those establish and sentence— defendant’s conviction discrimination violative of the deliberate imprisonment to life whether sentence equal protection clause. unassailable, constitutionally death—is I am able to concur because neither data, hold, I indicates no invidious would us the others case before nor racial discrimination as a matter of law. presently pending would the difference in- To the extent the data consists of Judge Roney fluence the outcome. As and/or sentences that are con- convictions out, points petitioner’s insuffi- statistics are infirm, stitutionally the data is irrelevant. cient to establish intentional discrimination data, petitioner’s summary, which shows capital imposed sentence in his case. nothing more disproportionate than sen- amendment, eighth As to the I doubt that a results, tencing probative is not racially of a caprice claim of arbitrariness or is even discriminatory part any motive on the presented, petitioner’s entirely since case is Georgia’s participants proving penalty is devoted to that the death sentencing petitioner’s or being applied altogether explicable— model—either in an impermissible any other case. albeit —fashion. analysis rely process it in his brief. addressed the due tioner did not

1. I have not by peti- employed the district court because petitioner sponding Claims such as are now difference in the need for reliabil- presented regularity such ity with we in the determination that death is the guidance may reasonably hope for from appropriate punishment.”). There is no Supreme my expressed the time ease, however, need this to reach out and concerns are outcome determinative in a try to define precisely more what evidentia- given case. ry showing required. Judge would be Ro- ney’s opinion clarity demonstrates with

KRAVITCH, Judge, concurring: Circuit why the evidentiary showing in this case is insufficient. I majority opinion except concur in the as view, my to the Giglio issue. for rea- GODBOLD, Judge, Chief dissenting in Judge sons stated Chief Godbold’s dis- part, concurring part, with whom sent, surrounding the facts Evans’ testimo- JOHNSON, CLARK, HATCHETT and Cir ny did constitute a Giglio violation. I Judges, cuit join as to the dissent on the agree however, the majority, with Giglio issue *: beyond error was harmless a reasonable doubt. Evans, At the merits trial who had been McCleskey, incarcerated with testified that join Judge I also special Anderson’s con- McCleskey admitted to him that he shot the currence Application on the “Constitutional policeman acknowledged that he wore Georgia Penalty.” Death makeup disguise to during himself the rob- bery. Evans also testified that he had ANDERSON, III, LANIER Circuit R. pending against him a escape [federal] KRAVITCH, Judge, concurring with whom charge, that he had not prosecu- asked the Judge, joins to the Circuit constitutional tor “fix” charge, to prose- and that the application Death Statute: promised cutor had not anything him testify. I join Judge Roney’s opinion for the ma- At the state hearing habeas the follow-

jority, and separately only empha- write ing transpired: size, respect to the Part entitled “Con- Application Evans, stitutional Georgia’s Death The Court: Mr. you let me ask Penalty,” question. that death is different At you kind the time that testified sanctions, from all other criminal McCleskey’s trial, Mr. you Woodson had been Carolina, 280, 305, promised v. North anything in exchange your testimony? Thus, proof of racial motivation re- No, The witness: I wasn’t. I wasn’t case, quired in a pursuant whether promised nothing prom- about—I wasn’t Eighth an theory Amendment equal nothing by ised the D.A. But the Detec- protection theory, presumably would be tive told me that he would—he said he less required strict than that in civil cases going himself, to do speak a word or in justice system the criminal generally. for me. That was what the Detective adjudication Constitutional would tolerate told me. less risk that a death sentence was influ- By Stroup: Mr. enced race. The Court’s Q: The you Detective told that he would Eighth jurisprudence Amendment has es- speak you? a word for supervision tablished a constitutional over A: Yeah. the conduct of state death penalty systems Q: Dorsey? That was Detective exacting which more than that with re- A: Yeah. spect justice to the system gener- criminal Transcript State Habeas ally. Carolina, Woodson v. North id. at (“Because 96 S.Ct. at 2991 granted The district court habeas relief qualitative difference, U.S., there is a corre- McCleskey under Giglio v. * Giglio I Judge Roney's dissent on opinion issue. I concur in on all other issues. of law no violation occurred. A Giglio enough. subject the district court At the threshold trivial offer is not applies only Giglio out that of the offer to Evans was pointed matter substan- by prosecutor tial, made jury “traditional deals” or at least a was entitled to con- “any testimony also to exchange McCleskey but sider it so. After was tried and understandings by any made promises convicted, charge dropped. the federal was team, which prosecutorial of the member Third, the concludes there was no court investigators.” F.Supp. police includes reasonable likelihoodthat Evans’s testimony The court then made these subsidi at 380. judgment jury. affected the Co-de- (1) testimony ary findings: that Evans’s was Wright only eyewitness. fendant was the (2) jury that “the was highly damaging; accomplice, testimony, He was an thus his impression that Evans clearly left with corroborated, unless was insufficient to es- any charges which unconcerned about was McCleskey trigger- tablish that was the him against and that pending were recognizes en banc court man. The this made which would af promises had been problem by holding avoids it that but 381; (3) credibility,” id. at fect his that at Wright’s testimony by was corroborated hearings Evans petitioner’s state habeas “McCleskey’s own confession.” This could investi testified “that one of detectives guilt refer to either of two admissions of promised speak to gating the case had McCleskey. He by Wright, “confessed” to behalf,” id.; (4) authorities on his federal Wright’s testimony subject but escape charges pending against Wright’s could not used to corroborate subsequent dropped were to McCles Evans accomplice otherwise insufficient testimo- key’s trial. ny. Testimony accomplice anof cannot be seems to me to err on The en banc court by accomplice’s corroborated own testi- proper appli- grounds. It blurs the several mony. The other “confession” was made by focusing sharply on the Giglio cation of by to Evans and testified to Evans. Thus proper inquiry “promise.” The is not word a minor Evans is not or incidental witness. contracts, limited to formal unilateral or testimony, describing Evans’ what McCles- law, bilateral, or of contract but “to words him, key “confessed” to is the corrobora- jury that the knew the facts ensure only eyewit- testimony tion for the might giving motivate a witness in testimo- ness, Wright. eyewitness gave And that Kemp, F.2d ny.” Smith McCleskey direct evidence that Cir.1983). (11th the infor- Giglio reaches killed officer. understanding as the formal. mal as well properly granted The district court is, dealings infor- point if the even grounds. judgment Giglio writ on Its mal, reasonably the witness can view be affirmed. should offering him government’s undertaking as juror knowing of it and can benefit JOHNSON, Judge, dissenting Circuit reasonably motivating it as the wit- view part concurring part, with whom testimony? The verbal un- giving ness in CLARK, Judges Circuit HATCHETT and in- dertaking made in this instance an join: officer, vestigating state who is member team, “put he prosecution will McCleskey presented con- Warren pending him” on his federal in a word for vincing evidence to substantiate his claim charge undertaking that a was Georgia has administered its death to know entitled about. way that discriminates on the charac- Study, race. The Baldus basis of Second, the en banc court finds bene- complete away “far and the most course, terized as possible marginal. fit too Of sentencing” thorough analysis of ever so and potential to a can be benefit witness out,1 that in demonstrates could find as a matter carried minimal that a court Berk, panel the Nation- description given member of a at trial Dr. Richard 1. This was the penalty kills a victim a death and McCleskey who white does not have person receiving higher penalty risk of the death prove to intent to discriminate order to person than a who kills black victim. penalty being ap- show that the death explain part higher alone can of this Race plied arbitrarily capriciously. and majority concludes that the evi- risk. The “confirms rather than condemns the

dence Viability Eighth A. The of an Amend- system” support and that it fails to a con- Challenge ment my challenge. disagree. I In stitutional majority recognizes, As the the fact that disturbing evidence can and opinion, this penalty facially a death statute valid under support a constitutional claim does Eighth holding does not foreclose an Amendment Eighth other- Amendment. wise, challenge systemwide majority applica- commits two critical er- based on the requires McCleskey prove rors: tion of that statute. The district court against to discriminate the State intended certainly Apply- most erred on this issue. it underestimates what personally him and ing penalty racially the death in a discrimi- prove. I actually did will ad- his evidence natory Eighth violates the Amend- manner concerns com- dress each of these before majority ment. Several members of the validity Bal- menting briefly on the of the 238, 245-57, Georgia, Furman v. 408 U.S. Study addressing the other issues dus and 310, 364-65, 2726, 2729-36, 2762, 92 S.Ct. in this case. 2790-91, (1972)(concurring 33 L.Ed.2d 346 Stewart, Marshall, opinions Douglas, 1. THE AMENDMENT AND EIGHTH JJ.), part disproportionate relied in on the IN THE RACIAL DISCRIMINATION impact penalty of the death on racial minor- THE DEATH ADMINISTRATION OF concluding penalty ities in that the death as PENALTY arbitrary then administered constituted McCleskey claims that adminis- capricious punishment. penalty way in a that dis- ters the death When decisionmakers look to the race of criminates on the basis race. The dis- victim, completely a factor unrelated to opinion argument trict court treated this proper sentencing pro- concerns arising under the Fourteenth Amend- one determining cess enters into the sentence. explicitly rejected petition- ment2 and Reliance on the race the victim means argument er’s claim that he could raise the part that the sentence is founded on a Amendment, Eighth under the as well. morally constitutionally repugnant majority possibil- each of these reviews judgment regarding the relative low value ities and concludes that there is little dif- of the lives of black victims. Zant v. proof necessary prevail ference Cf Stephens, 462 U.S. under theories: whatever the (1983) (listing L.Ed.2d 235 race of defend- challenge, constitutional source of the “constitutionally impermis- ant as a factor petitioner disparate impact must show a totally sentencing sible or irrelevant to great enough compel the conclusion that process”). legitimate There is no basis in purposeful permeates discrimination relying sentencing reason for on race in the system. positions These reflect a misun- process. of race derstanding Eighth Because use is both of the nature of an sentencing impermissible, Amendment irrelevant to claim death con- prohibits sentencing part by determined in race is Eighth text: Amendment racially capricious discriminatory application arbitrary and therefore a clause, Academy charged reviewing expressed opinion might al that it Sciences but previous sentencing process all research on criminal claim. It best be understood as a due issues in order to set standards for the conduct appear does not that a different constitutional such research. basis for the claim would have affected district court's conclusions. by precedent 2. The district court felt bound analyze protection equal the claim under the Amendment. See meaningful way of Eighth distinguish- vide some violation 238, 256, ing those who receive the Georgia, between Furman Godfrey (1972) sentence and those who do not. Georgia, 446 U.S. 420, 433, J., (“the high concurring) service (Douglas, (1980); Furman v. punish- and unusual’ 64 L.Ed.2d 398 by the ‘cruel rendered Georgia, is to Eighth Amendment ment clause of (1972) (White, J., general con- see to it that

require judges to selectively, curring). Appellate proportionality sparsely, review applied are not laws every consistency groups”). is not needed case but spottily unpopular and indispensable is still to a constitutional sen- and Proof Eighth Amendment B. The import tencing system.3 single of . Discriminatory Intent sentencing depends decision less on the in- tent of the decisionmaker than on the out- Eighth The central concerns comparable come in cases. Effects evi- decisionmaking Amendment deal more type dence is well suited to this of review. groups of cases than with processes and phrase In or cases. individual decisions emphasis produced This on the outcomes cases, throughout its later repeated system springs the entire from the Gregg Georgia, v. 428 Supreme special duty State’s to insure fairness with 2909, 46, 153, 2935 n. 195 n. 96 S.Ct. U.S. regard something as serious as a death (1976) opinion), 46, (plurality Stephens, Zant v. 49 L.Ed.2d 859 See sentence. 462 U.S. arbitrary and “pattern 862, 2733, 2741, that a stated 77 103 S.Ct. L.Ed.2d 235 Ohio, sentencing” 586, violate the Lockett v. capricious (1983); 605, would 438 U.S. fact, 2954, 2965, (1978); In the Court has Eighth Amendment. 57 L.Ed.2d 973 S.Ct. Carolina, v. North adopted systemic perspective Woodson consistently U.S. looking opera- penalty, to the 49 L.Ed.2d 944 on the (1976) sentencing (plurality opinion). Monitoring pat- entire structure tion of a state’s especially it inflicted sen- terns of sentences offers an ef- determining whether Eighth way duty. Amend- fective to detect breaches of that tences in violation of Oklahoma, Indeed, See, v. the death retains e.g., Eddings because ment. 104,112,102 discretion to make individual- the need for U.S. S.Ct. (1982) punishment judgments must im- ized while at same time (capital heightening and con- reasonable consist- the need for fairness posed “fairly, and with Oklahoma, sistency, Eddings supra, v. all”); Godfrey Georgia, v. ency, or not at 874-75, 110-12, pat- 420,100 at at 64 L.Ed.2d 398 U.S. S.Ct. often be the (1980) (“[I]f to authorize terns of decisions a State wishes hin- acceptable of review. Discretion it has a constitutional basis capital punishment if unfairness and apply inquiry its ders into intent: to tailor and law responsibility arbitrary inconsistency are to be detected even when that avoids the a manner obvious, overwhelming ef- penalty.”). they are not capricious infliction of the death upon. relied fects evidence must be perspective, re- systemic this Without objective stan extremely systemwide lim- on sentencing would be Insistence view of sentencing reliably prevents ited, arbitrary guide dards to very for the idea having probe its fullest decisions without sentencing takes on aberrant capricious juries A or other decision- comparative context. non- the intentions meaning in a Georgia, supra, Gregg sentencing pro- must makers. arbitrary structure — Harris, Pulley importance evidence of arbitrariness in indi- Supreme Court in 3. The —, (1984), exclusively "systemic" and looks U.S. emphasized vidual cases importance of factors other The case further underscores arbitrariness. claims, appellate proportionality review that than responsibility such to be alert to court’s jury discretion and assure makes, would control allege McCleskey more as the one arbitrary pat- not fall into an sentences would single disproportionality in a sentence. than deemphasizes Pulley tern. The decision 2936; presence 96 S.Ct. at statutory aggravating U.S. at Woodson fac- Carolina, supra, 428 North upon. tors that a could relied If have (objective necessary standards at 2990 present the factors are record does rationally process to “make reviewable jury may not matter that the have misun- penalty”). imposing the death aggravating derstood the role of circum- to constrain the discre need for the State If unintentionally stances. the State can penalty area is juries tion of in the death in preventing arbitrary capri- succeed by comparison unusual to other areas of sentencing, cious it would seem that rely the need to the law. -It demonstrates duty can also fail in though State its even way systemic controls as a to reconcile none of the relevant decisionmakers intend consistency; the same com discretion and such a failure.5 argue for the use ef objectives bined sum, systemic Court’s waiting than fects evidence rather for evi objective perspective in the review and specific improper dence of motives cases. sentencing control of death indicates that a Objective control and review cf. sentenc- pattern by of death sentences skewed race ing structures is carried so far that a support arbitrary alone will a claim of presumed or other decisionmaker be capricious sentencing in violation of the non-arbitrary to have intended a result Eighth Amendment. See Furman v. Geor- non-arbitrary by when outcome is an gia, 408 U.S. law, standard; short, objective in looks J., (Douglas, concur- to the result rather than the actual mo- (“We ring) say that cannot these defend- Zant, tives.4 In v. Westbrook 704 F.2d ants were sentenced to death they because (11th Cir.1983), this held 1504 Court were Yet our task black. is not restricted that, though judge might prop- even not to an effort to impelled divine what motives sentencing erly jury regarding instruct a penalties.”). these death majority’s proper aggravating cir- definition holding every on this issue conflicts with cumstances, the “uncontrolled discretion of limit pen- other constitutional on the death jury” by an uninstructed can be cured re- Georgia Supreme alty. today, After in view the Court. The this Circuit arbitrari- state court find that the record must shows ness on race will more based be difficult to Ohio, 4. portion Lockett v. 57 3. The Smith court cites to a (1978), L.Ed.2d 973 and other cases demonstrate Spinkellink opinion dealing equal protec with the actual deliberations of the sentencer are arguments. tion 578 F.2d at 614 n. 40. Neither Amendment, Eighth relevant under for miti- pertinent of the cases took note of the most gating proper place factors have their must Eighth precedents Amendment decided sufficiency all deliberations. But the of intent Supreme Court. proving Eighth Amendment violation does Other Eleventh Circuit cases mention that ha necessity imply of intent such all corpus petitioners prove beas must intent claims. racially against personally discriminate them only application 5. The Fifth or Eleventh of the death sentence. But these Circuit cases touch ing discriminatory though the issue of intent under cases all either treat the claim as it arose Eighth appear Amendment to be inconsist rely under the Fourteenth Amendment alone or Supreme approach ent with Court’s on Smith or one of its successors. See Sullivan wrongly therefore decided. The court in Smith (11th Cir.1983); Wainwright, F.2d Balkcom, (5th 660 F.2d Cir. Unit B (11th Wainwright, Adams v. F.2d Cir. 1981), modified, (5th Cir.1982), 671' F.2d 858 course, 1983). Of to the extent these at cases Eighth challenges stated that Amendment based Eighth tempt to foreclose Amendment chal intent, require showing on race but lenges require proof particular of this sort or wrong court reached this conclusion because it discriminate, they are ized intent to inconsistent ly Spinkellink Wainwright, believed that Supreme interpretation with the Court’s (5th 1978), compelled F.2d 582 Cir. such a result. Collier, Eighth Amendment. Gates v. Cf. Spinkellink ques court never reached (5th Cir.1974) (prohibition F.2d 1300-01 intent, holding tion of against punishment cruel and unusual "is not precedent Eighth foreclosed all Amendment specific limited to acts directed at selected indi challenges except for extreme cases where the viduals”). shockingly disproportionate sentence is to the supra F.2d at crime. 578 606 & n. See note type than other sort of fers a eradicate arbitrari- of effects ap- evidence that sentencing system. proaches intent, ness evidence of no matter what level of disparity is shown. For ex- EF- II. PROVING DISCRIMINATORY ample, the statistics this case show that THE FECT AND INTENT WITH penalties a certain number were STUDY BALDUS probably imposed race, because of without inquiring ever directly into the motives of conducted Dr. The statistical jurors prosecutors. provides possible Baldus the best evidence impact. began It racially disparate Regression analysis becoming a com fact: single unexplained killers of white proving mon method of discriminatory ef Georgia victims in over the last decade in employment fect discrimination suits. penalty have received the death eleven fact, Study the Baldus shows effects at times more often than killers of black vic- least as dramatic convincing as those employed tims.6 It then several statistical in statistical studies past. offered in the techniques, including regression analysis, Smith, (D.C.Cir. Segar v. 738 F.2d 1249 Cf. disparity the amount of that at- to isolate 1984); Mississippi Cooperative Wade v. racial and non-racial fac- tributable both Service, (5th Extension F.2d 508 Cir. techniques yielded Each of the tors. 1976). Nothing more should necessary be statistically significant racial influence of prove applying its death words, they percent; at least six other way statute in a arbitrarily pattern sentencing all showed that the capriciously illegitimate relies on an explained by assuming that could factor—race.8 made all defendants the race of the victim proof Even if discriminatory of intent killing of victims at least convicted white necessary were to make out a constitution- likely to receive the death percent six more challenge, any al under reasonable defini- such the num- penalty. Other factors7 tion of intent the Study provides Baldus aggravating of circumstances or the ber procf. majority ignores sufficient The for occupation of the victim could account McCleskey fact that has shown discrimina- differential, of the eleven-to-one some but tory sentencing sys- intent at work in the the race of the victim remained one of the though pointed any tem even he has not strongest influences. specific responsible act or actor for discrim- Assuming study actually proves that the inating against particular.9 him in prove, assumption an what it claims to recognizes though The law in- even make, un- majority claims to the evidence tentional discrimination will be difficult to doubtedly disparate impact. Re- shows situations, workings detect in some its great advantage gression analysis has the Mitchell, pernicious still and real. Rose v. showing perceived that a racial effect is of 2993, 3001, 443 U.S. 99 actual racial effect because it controls S.Ct. 61 an By L.Ed.2d 739 for the influence of non-racial factors. Under some circum- stances, therefore, screening explanations proof discriminatory out non-racial for of outcomes, regression analysis important step of- effect will certain first I, course, Among eligible part proof supra. 6. those who were for the death 8. See Of penalty, percent significant enough eleven of the killers of white racial effects is under penalty, Amendment, received the death while one victims Eighth requirement prov- percent of black received of the killers victims ing large pervasive or effects is tantamount it. proof of intent. models, largest of these the one In one leading 9. Thesame factors to the conclusion majori- district court and the focused on Eighth claim does not re that an Amendment indepen- ty, the statisticians used 230 different strong quire proof of intent militate even more pat- (possible dent variables influences on the using understanding ly against too restrictive an including sentencing), several different tern of intent. aggravating many possible mitigating fac- tors. 912 v. intent, Board Edu- inherent

proving process: the selection “The Crawford cation, 527, 3211, challenged application 102 458 U.S. S.Ct. 73 statute often (1982), subjective involves discretion L.Ed.2d 948 be the best criteria uti point lized at a crucial in the Washington v. intent. decision-mak proof available Davis, 229, 241-42, 2040, ing process.” 96 426 U.S. S.Ct. 2048-49, (1976); United 597 jury same concerns work in the Agency,

States v. Texas Educational operate equal discrimination context (5th 910, Cir.1978), nn.& 5-7 F.2d 913-14 force in the death context. The denied, 443 U.S. 915, 3106, cert. prosecutor has considerable discretion and L.Ed.2d 879 jury but bounded irreducible discre- tion. realistically hope Defendants cannot instance, proof of intentional For discrim- to find direct of discriminatory evidence jurors ination in the selection has tradi- precisely intent. This is the situation envi- showing on tionally depended racial ef- sioned in Arlington Heights, where the Partida, See Castaneda fects. U.S. pointed “[sjometimes out that a clear 1272, 482, (1977); 51 L.Ed.2d 498 pattern, unexplainable grounds Fouche, other 346, Turner v. 396 U.S. S.Ct. race, emerges than from the effect of the Zant, (1970); Gibson v. 24 L.Ed.2d 532 state action the governing legis- even when (11th Cir.1983). 705 F.2d 1543 This is be- face____ appears lation neutral on its allowed cause the discretion com- evidentiary inquiry relatively is then easy.” missioners, legitimate, although could easi- 429 U.S. at at 564. S.Ct. ly be used to mask conscious or uncon- scious racial discrimination. result, As a of discriminatory evidence recognized presence Court has presented effects Study, Baldus like this sort of discretion calls for indirect disparities evidence of racial compo- Davis, procf. Washington methods of sition jury pools10 and in other con- 241-42, 2048- texts,11 every excludes reasonable infer- (1976); Arlington discriminatory ence other than intent at Heights Metropolitan Housing Corp., system. work in the This Circuit has ac- n. 564 n. knowledged on several occasions that evi- 13, (1977). 50 L.Ed.2d 450 support dence of this sort could a constitu- v. Wainwright, tional challenge. Adams prob This Court has confronted the same 1443, (11th Cir.1983); Smith 709 F.2d 1449 In Searcy v. analogous setting. lem in an Balkcom, (5th 660 573 F.2d Cir. Unit B Williams, 1003, (5th 656 F.2d 1008-09 Cir. 1981), part, 671 F.2d 858, in cert. modified 1981), Hightower sub nom. v. Searcy, aff’d denied, U.S. S.Ct. 984,102 U.S. L.Ed.2d 844 (1982); Spinkellink, supra, L.Ed.2d at (1982), facially court overturned valid 614. selecting procedure for school board mem bers because selections fell petitioner into an A not exclude need all infer- overwhelming pattern of racial imbalance. ences other discriminatory than intent part The on particular decision rested his or her major- discretion case.12 Yet the distinguishes majority jury 10. The Agency, discrimi- United States v. Texas Educational grounds, stating (5th Cir.1978), denied, nation cases on tenuous 579 F.2d 910 cert. disparity minority between number of (1979), 61 L.Ed.2d 879 persons jury venire and on the the number of involving segregated system, provides school persons population such is an "actual example applied another of effects evidence as disparity,” while the racial influence in this case decisionmaking system. to an entire disparities is not. sidered, If actual are to be con- employ then the should court actual particularity requirement appeared 12. (and overwhelming) eleven-to-one differential sporadically prior in this Court’s decisions between white victim cases and black victim time, although part this was not a figures percentage presented by cases. The original observation about sort of statistical really Study Baldus are more reliable than "ac- Balkcom, supra. evidence in Smith v. disparities they possi- tual” because control for ble non-racial factors. ity particularity improperly stresses this victim as a factor of considerable influence. requirement interprets it so as to close Where racial discrimination contributes to open by a door left Court.13 decision, an official the decision is unconsti- nearly prove impossible It would be though tutional even discrimination was system’s through evidence of a usual ef- primary not the motive. Personnel Ad- that intent must have been a factor fects Feeney, 256, 279, ministrator v. 442 U.S. evidence, case; effects in this con- one 60 L.Ed.2d 870 text, necessarily many deals with cases at Neither can the impact racial be ex- Every jury charge once. discrimination plained away by the need for discretion in stillborn if the defendant had to would be the administration of the penalty prove by jury direct evidence that the com- by any “presumption the statute is deprive missioners intended to him or her operating in a constitutional manner.” The right composed to a of a fair necessary discretion to the administration community. Requiring cross-section of the of the death does not include the particular proof of discrimination case discretion to consider jury may race: especially inappropriate regard any proper consider aggravating factors, claim, Eighth Amendment for even un- but it not consider the race of the majority’s description proof der the victim aggravating as an factor. Zant v. necessary Eighth to sustain an Amendment Stephens, challenge, operating pervasive race in a And a statute system” manner “in the will suffice. presumption deserves a of constitutionality majority, sowing after doubts of this only where there is uncertainty real as to sort, despite nevertheless concedes that whether race influences application. its particularity requirement, evidence of the Evidence such Study, as the Baldus show- system’s strong enough effects could be ing pattern that the of sentences can only purpose.14 demonstrate intent and Its sub- explained by assuming significant ra- sequent implications efforts to weaken the influence,15 cial overcomes pre- whatever Study to be drawn from the Baldus sumption exists. uniformly unsuccessful. majority’s effort to im discount the example, majority For takes comfort portance of the hypothesis” “liberation also aggravation in the fact that the level of support fails. of his contention that powerfully sentencing influences the deci- juries rely were more inclined to Georgia. race sion Yet this fact alone does when other factors did not militate system not reveal a “rational” toward work. another, one The statistics not outcome or Dr. show the num- Baldus noted aggravating significant pronounced ber of factors is a that a more racial influence influence; they point appeared also to the race of the aggravation cases of medium dissenting opinion acceptance The majority 13. of Justice Powell in ent with his of the out- — U.S. —, 562, Kemp, Stephens "exacting” supervi- 104 S.Ct. come. The constitutional 370, (1984), 78 L.Ed.2d 372 does not undermine sion over the death established the import Supreme compels the clear of cases such as the dis- Court the conclusion that dis- thing, criminatory support Eighth crimination cases. For one a dissent effects can an Furthermore, summary prece- challenge. from a order does not have the Amendment the ma- is, weight fully opinion jority’s dential of a considered of evaluation of the evidence in this case another, Stephens anything, the Court. For dissent if more strict than in other contexts. 10, Study equal protec- supra. considered the Baldus as an See note argument only tion and did not address what operates average 15. The racial influence might necessary prove Eighth be an Amend- probably therefore case at work in ment violation. single majority case. The the na- misconstrues agree Judge analysis regression says IWhile Anderson's observa- ture of it that the when proof repre- "the of racial motivation re- tion that coefficient of the race-of-the-victim factor quired percentage in a death case ... would be less strict sents the of cases in which race required rep- than or in the crimi- a civil cases could have been factor. That coefficient justice system generally,” it nal I find inconsist- resents the influence of race across all the cases.

914 (6 (20 Sullivan, than all cases combined than their percent) sufficiency. majority Supreme percent). stay The states that racial Court refused to the execu- simply tion it impact provide agreed in a subset of cases cannot because with the deci- Court, sion systemwide challenge. of this a decision based on the the basis for However, validity of the alone.18 absolutely justifica there is Sullivan v. (11th Cir.1983) Wainwright, 721 F.2d 316 a claim. fact tion for such The (citing prior rejecting cases statistical evi- system mishandles sizeable subset of dence because it did not account for non-ra- persuasive cases is evidence that the entire effects). explanations cial of the As the operates improperly. system Connecti Cf. mentions, majority methodology 440, Teal, 2525, 457 U.S. cut v. S.Ct. Study easily surpasses Baldus that of the (1984) (written test discrimi L.Ed.2d earlier studies involved those cases. against employees); some nates Lewis v. Orleans, City 415 U.S. New S.Ct. Thus, Study Baldus offers a convinc- (1974)(statute infring 39 L.Ed.2d ing explanation disproportionate ef- ing First Amendment in some interests Georgia’s fects of penalty system. cases). system applied arbitrarily A can be pattern It shows a clear sentencing capriciously even if it resolves the obvi explained race, can be in terms of in a rational manner. ous eases Admitted it does so in a context where direct evi- precise ly, the lack of a definition of medi practically impossible dence of intent is aggravation um cases could lead to either imagination obtain. It strains the to be- overstatement or understatement significant lieve that the influence on sen- however, Accepting, racial influence. tencing unexplained by left 230 alternative racial factor is accentuated to some racial, espe- factors is random rather than cases,16 degree range the middle cially in history a state with an established impact of racial must taken all evidence racial discrimination. Turner seriously. more Fouche, supra; Chapman King, (5th Cir.), denied, F.2d 460 cert. Finally, majority places undue re- 90 L.Ed. 1025 on several recent liance petitioner certainly presented has evi- Strickland, argues cases. It that Ford v. — dence of intentional racial discrimination at —, U.S. — Georgia system. Georgia work in the (1984), Wainwright, Adams v. meaning within the Eighth Amend- —, (1984), 80 L.Ed.2d 809 applied arbitrarily ment its statute and ca- and Sullivan v. Wainwright, 464 U.S. priciously. (1983), support its conclusion that the Baldus III. THE VALIDITY OP THE BALDUS Study strong enough does not make a STUDY showing justify of effects to an inference majority of intent. But to the extent that these purport does not to reach any guidance cases offer regarding Study all the issue of whether the Baldus reli- legal applicable ably proves standards to these it prove. what claims to How- studies,17 ever, majority is clear that the Court con- does state that the dis- validity sidered the of the studies rather findings regarding validity trict court’s majority apparently ignores Study: ruling 16. its commit- nounced statistic in the Baldus accept validity Study ment to insufficiency Baldus based on that one indicator questions it the when aggravation definition of "medium controlling would not be here. by cases” used Dr. Baldus. Indeed, the Court indicated it would opinion 17. The in Ford mentioned this issue in a have reached different conclusion if the dis- sentence; single the order in Adams was not given trict court and this court had not been the accompanied by any opinion written at all. opportunity analyze adequately. to the statistics argument None of the three treated this possible aas —U.S.—, 104 S.Ct. at L.Ed.2d n. Eighth Finally, Amendment claim. the n. 3. multiplier” pro- "death odds is not the most study might foreclose habeas of the relief too much discretion. But most of the al- Moreover, majority opin- leged the on this issue. “mismatches” resulted from inten- questions improvements ion in several instances the validi- tional in coding the tech- ty study claiming niques the inter- remaining of while and the errors were not in I sufficiency large enough ested its alone. therefore to affect the results. of the reasons that will summarize some missing The data in some cases was also clearly court was erroneous in the district a matter of concern for the district court. Study finding the Baldus invalid. The missing small effects of the data leave to a The district court fell victim miscon- much of that concern unfounded. The race ception findings. that distorted its factual of the in victim was uncertain 6% most20; pointed goodly out a number of cases at penalty trial information study imperfections rarely in the but went was in percentage unavailable the same significance relatively ahead to determine the cases.21 The small amount of data, imperfections. missing A court not sim- large those combined with the num ply point analysis to flaws in a statistical ber of used in variables several of the models, completely and conclude that it is unrelia- should have led the rely court to prove study. analyses ble or fails to what was intended Statistical have never Rather, explain prove. perfection the Court must been held to a standard of imperfection study perfection makes the less why the near order for courts to treat proving proposition competent that it them capable as evidence. Trout v. Lehman, support. (D.C.Cir. was meant to Eastland Ten- 702 F.2d 1101-02 1983). Valley Authority, problems 704 F.2d 613 nessee Minor inevitable — U.S.—, (11th Cir.1983), denied, study scope cert. of this and complexity: the stringent L.Ed.2d 741 standards used the district spell court would the loss of most statisti imperfections noted Several cal evidence. legally significant district court were not Many imperfections of their minimal effect. Other study because were significant the errors data base match this de- not because there was no reason instance, scription. problem For the “mismatches” to believe that would work systematically expand in data entered once for cases the Proce- the size of the Study again Reform for the same dural race-of-the-victim factor rather than to con- Charging Sentencing unchanged. cases in the tract it or leave it The multi- negligible. Study scientifically collinearity problem problem were is a of notable proportions district court relied on the data that that nonetheless did not in- changed from one to the next in crease the size of the race-of-the-victim fac- concluding Ideally independent the coders were tor.22 allowed variables in remaining ing many 19. errors affected little more how of the cases led to trials. percent prediction the data in than one Since the was based on discernible cases, Data errors of less than 10 or models. 12% trends the rest of the the district court generally validity give clearly weight do not threaten of mod- was erroneous to to the prediction. el. method," "imputation coding pro- 20. Dr. Baldus used an 22. The treatment of the conventions whereby example. the race of victim assumed to vides The district court another criti- (indicat- treating as the be the same race of defendant. Given Dr. Baldus for "U” codes .cized predominance ing uncertainty of murders where the victim as to whether a factor was race, case) present being beyond and defendant were of the same this meth- in a the knowl- estimating way (“absent”) was a reasonable the num- edge prosecutor od rath- It ber of victims of each race. further reduced assuming er than that the decisionmakers knew missing significance of this data. ("present"). contended about factor Baldus that, if the extensive records available on each factor, court, assessing weight presence of a 21. The district case did not disclose the omission, good did be accorded this assumed that Dr. chances were that the decisionmakers predict- presence, completely either. Dr. Berk testi- Baldus was unsuccessful in not know of its proper understanding A of statis analysis not be related isms.25 regression should a independent shows, however, If one varia- that these to one another. tical methods another, merely proxy as a shortcomings ble serves in the Baldus are not serious “multicollinearity.” from the model suffers Study. reduce the sta-

That condition could either mistrusted smaller district court or dis- significance of the variables tistical placed weight models because it too much relationships to one another. Of tort their complementary goals one of the several on course, multicollinearity to the extent that analysis. statistical Dr. Baldus testified significance suggests it reduces statistical opinion in his model that 39-variable be even the racial influence would that among many was the best models he multicollinearity if had not more certain the produced. The district court assumed artificially depressed the variable’s statisti- mechanistically that the more somewhat significance. As for the distortions cal independent encompassed by variables variables, relationships between the ex- model, the better able it was to estimate mul- petitioner explained perts for proper influence of non-racial factors. ticollinearity dampen the racial tends to models, bigger inBut statistical is not al- than enhance it.23 effect rather point, After a ways better. certain addi- every fail in The district court did not independent tional variables become correl- analyze significance instance already being con- ated with variables issue, Yet it did reach this problems. when suppress and distort or their influ- sidered appeared to misunder the court at times accurate strike an ence. The most models regres study the nature of this or of stand appropriate balance between risk of analysis generally. In several related sion omitting significant factor and the risk of criticisms, any it found that of the models Hence, multicollinearity. the district court accounting independent for less than 230 rejecting largest erred all but the mod- (580 completely worthless variables were els. 361), complete F.Supp. at the most earli The other two criticisms mentioned capture every models were unable to spring single er from a source—the misin (580 every F.Supp. case at nuance terpretation of the R² measurement.26 371), were not suffi and that models capture every The failure of the models ciently predictive upon light to be relied every nuance of case was an inevitable but (580 361).24 F.Supp. their low R² value Regression analysis ac harmless failure. majority implicitly questions the validi counts for this limitation with an R² meas ty Study the Baldus on several occasions result, it matter adopts it the first two of these critic urement. As a does not when Academy theory multicollinearity er’s other affects fied that the National of Sciences had significance. F.Supp. at this same issue and had recom- statistical 364. considered inconsistent, Baldus. Dr. The two theories are not for nei- mended the course taken Katz, Dr. state, suggested expert ther Dr. Baldus nor Dr. Woodworth denied that witness for the multicollinearity might multiple removing have the cases with the U codes from the effects. analyze altogether. sugges- possible The two theories each one ef- The district court’s then, Therefore, tion, rejected fect. the district court this the U codes be treated as testimony grounds. appears groundless clearly improper present, to be erroneous. represents 24. The R² measurement the influ Baldus later demonstrated that the U codes unique ence of random factors to each case that did not affect the race-of-the-victim factor captured by could not be addition of another recoding treat- all the items coded with U and independent approaches variable. As R² val ing present. them as Each of the tests showed 1.0, ue can be more sure that the inde one significant reduction in the racial variable. already pendent identified are accu variables rejected court this demonstration The district significant are mas rate and that no influences using largest it carried out because was not querading as random influences. available model. rejected expert The district court testimo- See, 23. e.g., pp. 25. testimony, ny, rebuttal but because allegedly petition- 26. See footnote 24. conflicted with the because every regard validity to the fails to consider nuance with the Baldus study that a (fac Study. I random factors would also reverse that court’s every case because sporad Eighth the outcome determination that an that influence Amendment tors way) impugn petitioner. do not unsystematic claim is not available He ic and to the systemwide factors reliability of the is entitled to relief on this claim. the identified, including race of the vic already IV. OTHER ISSUES extra factors be Failure to consider

tim. they operate problem where comes opinion I concur in the of the court with is, R² is throughout system, that where regard jury to the claim death-oriented inappropriately low. by in the result reached the court on the ineffective assistance of counsel claim. I the R² of court did find that The district dissent, however, on remain- must the two near study, which was the 230-variable disagree I ing issues the case. with the .48, But an R² of that ly too low.27 was issue, holding Giglio on the on the basis every inappropriately low size is not findings and of the conclusions district The R² measures random fac context.28 dissenting opinion court and the of Chief case: in areas where unique to each tors Judge Godbold. As for the occur, Sandstrom likely to especially such factors claim, I hold that the would instruction was ex expect, a low R². As the one would erroneous and that error was not harm- majority and the perts, the district court less. out, penalty death pointed no two have alike, exactly and it said to be cases can be by It is no means certain that an error of study unlikely for a statistical especially this sort can harmless. be See Connecti on a sentence. capture every influence Johnson, cut v. U.S. random factors at work light if Even an error context, the district court penalty death harmless, McCleskey could the fact that be finding the R² of all the Baldus erred relied on an alibi defense does not mean too low.29 Study models “not at issue” in the that intent was case. crime can Any element of a issue appear elsewhere in Errors of this sort presents whether or not defendant evi opinion and leave me with district court prosecution’s disputes dence that case that the and firm conviction the definite point. could find on that ruling court’s on the for the district

basis prosecution dispel failed to all reason had clearly errone- invalidity of the regard with to intent even able doubts Co., 333 Gypsum ous. United States though specifically did not the defendant L.Ed. argument. make such an Intent is at issue analysis, while This statistical support there is evidence to wherever sufficiently complete and reli- imperfect, is mind a reason reasonable doubt competent evidence to to serve as able existence of criminal juror I re- able as to the Accordingly, would guide the court. 683 F.2d Jernigan, court intent. Lamb v. of the district See judgment verse the Houston, University 654 F.2d finding that a model 28. Wilkins v. on the fact 27. It based Cir.1981), (5th contrary. predict is not to the "does not an R² less than .5 only that it could not know This is an That court stated in half of the cases.” inac outcome ² statement, percent actually represents R of or .53 would be whether an .52 for an R² curate context since the original inappropriately low in that percentage 11-to-l differen any argument parties independent not made is explained variables had all the tial ² sue. with an R of less than .5 combined. A model necessarily predict out fail to would not Furthermore, petitioner im expert the cases because the model come in half for the of way correctly proves upon pure unchallenged opinion as a chance the R² fered out For dichotomous out predicting an outcome. of dichotomous measurements in studies 50%, (i.e. imposed it is comes as much are understated comes ² not), Study predictions succeed half the could Baldus mod placing random the R values of the time. .7 and .9. somewhere between els (11th Cir.1982) (“no Montana, 1342-43 reason have concluded. Sandstrom v. juror could have determined ... able U.S. 61 L.Ed.2d 39 provocation appellant (1979); Bell, acted out self- United States v. 678 F.2d defense,” harmless). error was therefore (5th 1982) (en banc), B Cir. Unit aff’d grounds, on other majority raising that the an states (1983). Therefore, the automatically does not alibi render defense judgment of the district court should be violation harmless. It con- a Sandstrom ground, on reversed as well. cludes, however, raising that the of a non- coupled participation defense with “over- an kill-

whelming evidence of intentional HATCHETT, Judge, dissenting Circuit finding a of harmless ing” will lead to part, concurring part:1 and majority’s position is indistin- error. The case, In this the Georgia system of im- guishable finding from a of harmless error posing penalty the death is shown to be solely overwhelming evidence.30 based unconstitutional. Although Georgia the normally may not relieve Since a defendant penalty statutory death scheme was held responsibility to make factu- jury the of its constitutional “on in Gregg its face” regarding every element findings al of an Georgia, offense, only way for intent to “not the (1976), application L.Ed.2d 859 of the trial if at issue” in a murder is the evidence produces scheme explaina- death sentences by provides possi- side presented either only ble on the basis of the race of the regard fact with issue of intent. ble defendant and the race of the victim. Thus, McCleskey’s chosen defense in this I clearly write to state simply, and with- obscure the sole case should not basis jargon statisticians, out the of the the re- disagreement majority between the the produced by sults application the myself: inferences that reasonable scheme, statutory death from the circumstances of could be drawn by shown Study. Baldus killing. agree majori- I cannot with the ty juror, that no based on reasonable Study The Baldus is valid. The facts, interpretation of the could have had designed questions was to answer regarding a reasonable doubt intent. when, ever, much, all, if and how if at race - is a factor in impose the decision to factors in this case Several bear on the penalty Georgia. death study gives shooting of intent. The did not issue occur In Georgia, answers: Furthermore, when the defend- point-blank range. the of- ant is black and the victim of murder is moving ficer at the time of the shoot- white, percent greater a 6 chance exists ing. the basis of these facts and On other that the defendant will receive the shooting, juror circumstances of the death penalty solely could have had a reasonable because the victim is white. doubt as to percent person firing disparity present This 6 weapon through- whether in- is out majority range tended to kill. While the the total dismiss- of death-sentenced possibility speculation,” es this as “mere in Georgia. black defendants While requires appellate troublesome, spec- percent the law court to disparity it is the juror mid-range ulate about what reasonable could on which I focus. When Indeed, analysis the entire contrary improperly infringes harmless error evidence to the by may employed the court be based jury's duty on false on the to consider all relevant evi- dichotomy "overwhelming between evidence” dence. and elements “not at issue.” Wherever intent is crime, only an element of a it can be removed 1. Although majority opinion I concur with the by overwhelming as an issue evidence. The on the ineffective assistance of counsel plurality observation in Connecticut v. issues, separately death-oriented I write Johnson, supra, defendant in some express my thoughts Study. on the Baldus issue, only apply cases “admit” an should where dissent, join Judge I also Chief Godbold's as to the evidence allows one conclusion. To issue, Giglio Judge Johnson’s dissent. place allow an admission to take the face fall in sight considered which the mid- We must not lose cases are fact that very less serious and seri- the 39-factor model range, between considers information circumstances, impact to the aggravating where the relevant of the decisions ous be- white, ing officers, defendant has a made law enforcement victim black receiving prosecutors, judges, juries percent greater chance the deci- white, impose penalty. death penalty because the victim is sion to The ma- intolerable; jority suggests disparity This is it is that if such a rather than black. re- range agency cases that the deci- sulted from an actor or in this middle identifiable prosecution sentencing process, most diffi- in the and proper sion on the sentence is penalty present percent imposition disparity of the death the racial cult great enough could be to declare the questionable. most Geor- gia system unconstitutional under the study the arises disparity The shown eighth disparity amendment. Because this variety analyses made from a of statistical great enough is not considered to satisfy First, colleagues. by Dr. Baldus and his majority, the or because identification of tried to determine the effect of race Baldus agency preci- an actor or can be made with (PRS study) in cases of the victim 594 sion, majority the holds that the statutory persons of murder comprising all convicted approved by scheme is the Constitution. during particular period. To obtain bet- unidentified, Identified or the result of the results, techniques ap- with ter consistent race, ingredient unconstitutional at a Academy proved by the National of Sci- significant in system, level the is the same 2,500 ences, in Baldus identified cases inability on the defendant. The black indicted for murder persons which were identify agency the actor or has little to do during particular period and studied constitutionality system. of the 1,066 closely of those cases. He identified information, factors, percent greater The 20 bits about chance in the crime, defendant, (because mid-range and other circum- cases the defendant is surrounding white), produces stances each case which he black and the victim is impact disparity high. on a sen- thought study had some death that is too dem- Additionally, percent disparity, he fo- onstrates that the 20 tence determination. in world, of these factors which he the real means that one-third cused on 230 (with victims) thought most reflected the relevant consid- black defendants white in the penalty mid-range in erations a death decision. cases will be affected model, study receiving Through penalty. this 230-factor race factor the death signif- proved that black defendants indicted Race should not be allowed to take a impose convicted for murder of a white victim be- icant role the decision to stage signifi- penalty. trial gin penalty with a death greater probability receiving cantly Court has reminded us on is penalty, solely because the victim death “if more than one occasion that a state white. capital punishment wishes to authorize factors, thirty-nine responsibility has a Baldus also observed constitutional tailor aggravating apply cir- and in a that avoids including information on its law manner cumstances, arbitrary capricious match the circumstanc- infliction of which study penalty.” Godfrey Georgia, the death in this case. This focused es in the mid- 446 aggravating circumstances (1980). A severity that “white vic- L.Ed.2d 398 statute that inten-

range of indicated percent tionally unintentionally allows for such shown to be 20 tim crimes were unconstitutional under the likely penalty in a death racial effects is more to result majority eighth Because the equally aggravated than black vic- amendment. sentence otherwise, I dissent.2 holds Majority tim crimes.” at 896. application regarding Nothing majority opinion on discrimination in of studies penalty significance scheme should be validity, impact, the Florida death or constitutional

CLARK, dearer, tem Judge, dissenting part the life of a white is Circuit the life *: concurring part cheaper. a black to determine how challenged We are Before at a looking figures, few of the discrimination, any, if is tolera- much racial perspective necessary. Race is a factor imposition penalty. ble in the of the death system only where there is room Although join Judge I also Johnson’s discretion, is, where the decision mak- dissent, is directed ma- this dissent to the er has large a viable choice. number jority’s that the evi- erroneous conclusion cases, race has no effect. are These dence in this does not establish a case mitigated cases where the facts are so prima facie Fourteenth Amendment viola- even as a considered tion. possible punishment. At the other end of spectrum the tremendously aggra- Study vated murder cases where the defendant study, peri- The Baldus which covers the very will probably penal- receive the death od detailed 1974 to is a of over ty, regardless his race or the race of 2,400 From homicide cases. these homi- victim. In mid-range between is the cides, persons penal- the death received cases is an approximately where there 20% ty. types disparity Two of racial estab- racial disparity. lished—one based on race of the victim study designed Baldus was to deter- race of the and one based on the defendant. mine whether situated like cases are treat- white, If the victim is a defendant is more similarly. ed a starting point, As an unan- likely penalty. death to receive the If the alyzed comparison arithmetic of all black, he likely defendant is more to cases following: reflected the penalty. only receive the death One can operation sys- conclude that in by Sentencing Rates Defendant/

Death Combination1 Racial Victim figures gross disparate These show a racial Similarly, only 8% the victim was black. impact where the victim compared was white white defendants to of black 22% —that sentences, there were compared death 11% penalty defendants received the death percent to 1.3 death sentences when when the victim was white. The Supreme issues, imply separately construed to the United oriented I write ex- States Su- to preme passed squarely press my Court has on the thoughts Study. Florida on the Baldus also I studies. Neither nor join Judge Judge Chief Godbold’s dissent and passed Eleventh Circuit on the Florida stud- Johnson's dissent. ies, (as case), fully developed on a record in this eighth under fourteenth and amendment chal- 1. DB Exhibit 63. lenges. * Although majority opinion I concur the ineffective assistance counsel and death any aggravating one of the as much as disparities gross similar

Court has found in Georgia’s listed death sup- circumstances proof of discrimination be sufficient Amendment violation.2 port penalty a Fourteenth Therefore, statute.4 in the applica

tion of the statute in Georgia, race of the to deter study undertook The Baldus defendant victim, and of the when it is sentencing disparity was mine if this racial black/white, functions as if it were an ag or of race because caused considerations gravating circumstance in a discernible order to or find of other factors both. number of cases. See Zant Stephens, analyze out, necessary and com 462 potential penalty (1983)(race death pare L.Ed.2d 235 each of as an aggravating factors circumstance what relevant would be cases and ascertain constitutionally im permissible). for consideration the de were available many factors There were cision makers.3 part Another of the study compared the record, contemporane capital prior such as disparities in death penalty sentencing ac- motive, offense, killing avoid arrest ous cording to race the defendant and race hire, as well as race. or for the victim and reflected the differences race had as much or more sentencing showed that depending upon pre- factor. single death, other See dicted impact i.e., than chance of whether the type 76-78, T-776-87. Stated an case DB was was not Exhibits one where the just penalty death given. influences the verdict would be way, other race Page below at 9. 2. See discussion cases in which not be imposed." Stephens, Zant v. sentencing makes method of An individualized 3. particular case differentiate each possible to evenhanded, substantially objective, “in an 4. o.C.G.A. 17-10-30. § many murder way from the rational 43, DB, B Ex. 91.5 step pro- gravated Columns A and reflect cases. Table aggravated ag- compare sentencing D rates gression of least to most Columns C and eight sub-groups eight sub-groups in 5. The were derived from the then sub-divided into the group penalty ascending giving of cases where death was to more order consideration likely analysis predictably upon most based aggravating larger combina- serious factors and factors that resulted in the vast relevant steps progress. pages Tr. tions of them as the pen- majority receiving of defendants the death 877-83. alty group was out of the total 128. This —116 to defendants white defendants of minorities of black who jury served to in when the victim white and reflect that if a disparity determine amounted to a sub- penalty given 1 and 2 no death Steps stantial underrepresentation minority Step In penal- 41 cases. 45 death those jurors.7 simple Because this method did cases, only given ties were two blacks not many equation, consider variables its escaping penal- and three whites the death it was as not accurate complex as the sta- ty group obviously representing —this equations widely tistical today.8 used aggravated By comparing cases. most disparities mathematical that have through Steps 3 one can see that in each accepted by been as adequate pen- group black defendants death received prima establish facie of purposeful case disproportionately alties to white defend- range discrimination approximately from .19, .27, .15, .22, ants differences greater “Whether or 14% 40%.9 and .25. This indicates that unless the disparities prima constitute facie evidence certainly murder is vile as to almost so depends upon of discrimination the facts of 8), penalty (Step evoke death blacks are each case.”10 approximately likely more get 20% penalty. disparities Statistical jury selection cases are not sufficiently comparable to right side of the chart reflects how provide complete analogy. defendant, There are unlikely no any it is that but more guidelines in defendants, decided so particularly cases in this white will case receive rely we have to the death on reason. We when victim is start with a black. sentencing procedure that ap- has been as Statistics proved by the Supreme object Proof Court.11 The system, of this well as constitution- selection cases dif- have utilized ally permissible capital system, sentencing analysis ferent methods of statistical provide is to individualized treatment of determining disparity whether is suffi- eligible those for the death prima penalty to in- cient to establish a facie case of factors, sure that non-relevant purposeful Early discrimination.6 i.e. factors jury se- cases, that do Alabama, particular lection such not relate as Swain v. individ- committed, very simple ual or equations primarily play part used the crime which analyzed eligi- deciding the difference of minorities who does and who does not re- jury duty ble for the actual from number ceive the death penalty.12 The facts dis- Mattson, F.Supp. inability significance its because of to assess the Villafane (D.Conn.1980), tools.”). noted court that four forms of statistical data without mathematical analysis (1) have been used: the absolute Alabama, Partida, 495-96, difference test used in Swain v. 9. Castaneda v. U.S. (2) (1965); 40%); (disparity S.Ct. at 1280-82 Turner v. (3) away Fouche, approach; ratio test that moves percentages (1970) 23%); (disparity from examination of and focus- Whitus v. Geor- by underrepresen- 545, 643, gia, es on the differences caused 385 U.S. 17 87 S.Ct. L.Ed.2d 599 (4) jury; (1967) 18%); (disparity tation in each the statistical deci- Georgia, of v. Sims 389 theory fully 404, 523, which (1967) sion embraced in Cas- U.S. 88 S.Ct. 19 L.Ed.2d 634 Partida, 17, 19.7%); taneda v. U.S. at n. (disparity Georgia, 430 496 97 S.Ct. of Jones v. 389 U.S. Finkelstein, 24, 4, Applica- (1967) n. See at 1281 17. also The (disparity 88 S.Ct. 19 L.Ed.2d 25 of Theory Jury 14.7%). figures tion Statistical Decision to the computa- These result from the of Cases, Discrimination 80 Harv.L.Rev. 338 tion used Swain. (1966). Blackburn, 10. United States rel ex Barksdale v. Alabama, 202, 1115, (5th Cir.1981) (en banc). 7. See Swain v. 380 639 F.2d U.S. 85 1122 824, (1965); 13 L.Ed.2d 759 v. Man- Villafane son, F.Supp. 153, 2909, 504 Gregg Georgia, 83. v. 11. 428 U.S. 96 S.Ct. (1976). 49 L.Ed.2d 859 Finkelstein, Application 8. See The Statistical of Theory Jury sentencing body’s Decision to the Discrimination 12. The be decision must fo- Cases, 338, (“The (1966) "particularized 80 Harv.L.Rev. 363 cused on the nature of the crime problems particularized Court did not reach these and the the in- Swain characteristics of 924 legal cpntext study, vary of which which is some by the Baldus

closed discussed, demon- variety Because of previously applied. of the situa- been have disparate occur, there sufficient that is tions in which discrimination can strate prima to establish of blacks treatment proving of is the critical method intent fo- discrimination. of facie case failing recognize majority, by cus. this, meaning of intent in misconceives the discrimination, coupled with when This jurispru- equal protection of the context facts, prima demonstrate the historical dence. Amendment violation Fourteenth facie It is that Protection Clause. Equal proven circumstantially by Intent Equal Pro- against which discrimination utilizing variety objective factors and ma- protect. The stands to Clause tection totality of inferred from the can be however, give full reach to fails to

jority, appro- facts.16 factors most relevant to ac- one has While our Constitution. (1) presence case priate in this are: knowledge prejudice in our existence discrimination; (2) the im- historical accept cannot its society, one and does study, by Baldus pact, as shown no- This is application certain contexts. suspect on a capital sentencing law has administration more true than in the where has class.17 indicated justice capital cases. of criminal that: discrimination Evidence of historical The Fourteenth Amendment pur- inference of drawing relevant to Equal Protection discrimination, particularly poseful ... long re showing of has been “A intent that discrimi- where the evidence shows types equal protection in all quired commonly utilized, natory practices were charging discrimination.”13 cases racial they enjoined when were abandoned required proof of intent The Court rights illegal by courts or civil made strictly it will scrutinize actions before replaced legislation, they were entity.14 legislature or of a official which, though practices laws and neu- respect, the rule is a tool of intent face, their to maintain tral on serve purpose of that serves the self-restraint quo.18 status judicial policymaking.15 limiting review and impact may disparate Evidence of dem- monolithic struc- The intent test not a tests, purpose its focus will that an unconstitutional legal As with all onstrate ture. 206, countervailing rule U.S. at 96 S.Ct. at 15. The intent "serves a con- dividual defendant.” 428 586, Ohio, limiting judicial policy making. U.S. See also Lockett v. 438 98 2940. cern of Wash- 2954, (1978) ("the need 57 L.Ed.2d ington S.Ct. treating 973 v. Davis ... as a can be understood capital each in a case with defendant of the own of institu- reflection Court’s sense respect uniqueness degree due the power limitation on the tional self-restraint —a important is far more than non- individual having judicial review avoids the Court 605, 2965); capital S.Ct. at cases.” U.S. at Note, super legislature____” sit as a Section 104, Oklahoma, Eddings v. U.S. 102 S.Ct. Discriminatory Purpose Disproportion- 1981: 869, focus on a characteristic does 137, (1980); Impact, 160-61 ate 80 Colum.L.R. defendant, particular impermis- albeit an 229, Davis, Washington 247- see also infra, p. one. See 3. sible (1976). L.Ed.2d 597 Lodge, Rogers U.S. 13. Metropoli- Village Arlington Heights 16. See Housing Development Corp., tan ("Purposeful Id. at n. 5 racial discrimination L.Ed.2d 450 scrutiny strictest of adverse differ- invokes the purpose, differ- treatment. Absent such ential Lodge, Rogers also 102 S.Ct. at 17. Id. See v. subject impact is to the test of ration- ential ality.”); 3280. Sellers, Impact see also Intent on of Jurisprudence, Equal Protection 84 Dick.L.Rev. Lodge, Rogers at 102 S.Ct. 3280. 18. 363, (1979) ("the profoundly intent 377 rule of posture toward the Court’s affects equal claims.’’). protection

925 work, especially to at may group, prima continue fendant’s facie case of dis- explainable purpose not criminatory the discrimination is intent or where is established 43, acquires, grounds.19 supra non-racial and state of on Table burden rebut- 4, ting the accompanying and the case.23 p. the table evi- unexplained the racial dis- dence leave 20% many respects imposition In black parity the defendant is and the where penalty is similar to the selection of white and the murders victim is occurred jurors processes in that both are discretion- very under similar circumstances. nature, in ary vulnerable to bias maker, susceptible decision rigor- and to a Although rarely the Court found the analysis.24 ous statistical disproportionate of existence intent where has, impact only proof, is the for exam- The Court has refrained relaxing from proof jury in ple, relaxed the standard of proof the standard of where case does of selection cases because the “nature” of jurors not involve selection of because in jur- the task involved the selection of (1) policy of: deferring its to the reason Thus, equal protection ors.20 show executives; able acts of administrators and cases, in jury a de- violation selection (2) preventing tax, questioning and prove procedure must that “the fendant welfare, service, public regulatory, li employed resulted a substantial under- censing disparate statutes where impact is representation of his race or of the identifi- However, only proof.25 utilizing the group belongs.”21 he able which proof jury standards selection simple. behind this method is As idea cases to establish intent this case will out, pointed disparity is suffi- (1) “[i]f this policy contravene because: def ciently large, unlikely then it is that it is erence is not warranted where accident, and, solely due chance or grave is and less severe alternatives are contrary, to the available; (2) absence evidence one the court did not contem must conclude that racial or other class-re- plate capital sentencing statutes when it pro- Thus, lated factors entered into the selection policy. established this statistics 22 showing cess.” Once there is a of a prove alone could be utilized to intent underrepresentation background substantial of the de- this case. But historical Davis, 242, Washington prima U.S. 19. 426 at 96 lish a facie case of racial discrimination 2049, cases.”). S.Ct. at the Court stated: "It is also not selection infrequently discriminatory impact true that the practical purposes for all ... demonstrate 494, Partida, 482, 21. Castaneda v. 430 U.S. 97 unconstitutionality because in various circum 1272, 1280, L.Ed.2d 51 498 very stances the discrimination is difficult to explain grounds." nonracial See also Person 22. at 13. Id. n. Feeny, Administrator Mass. v. nel 442 U.S. 256, 2282, 24, n. 60 99 S.Ct. L.Ed.2d 870 495, at 23. Id. 97 S.Ct. at 1280. (1979) ( Washington Arlington recognize disparate impact upon a neutral law has a when Joyner, Legal Attacking 24. Theories Racial historically group a discrimination, that has been a victim of 101, Disparity Sentencing, Crim.L.Rep. purpose an unconstitutional (“In (1982) many respects sentencing 110-11 work). may still be at jury panels similar to selections of as in Castaneda."). majority opinion 926 made, Supreme as the Court supports the statistical been but also and relevant acknowledged, in 1979 conclusions. deny that, years 114 cannot after we also race, of odi- the basis “Discrimination the close of the War between the States pernicious in especially aspects, in is ous all Strauder nearly years 100 and after [100 It administration Justice.”26 is of 303, L.Ed. racial and other U.S. 25 664] through- see to it that duty courts to a of still remain forms discrimination bringing person a to for procedure out the life, jus- in the administration of fact of protection “the enjoy which

justice, society he shall in as a whole. Per- tice as our today discrimination takes a haps In an im- that guarantees.”27 the Constitution more than But it is form subtle before. to it is society, has admit perfect one pernicious.30 no less real or adminis- guarantee impossible pernicious jurors, especially is judges and If discrimination justice, both trators of no- justice, in administration of it is successfully racial in ev- wear blinders will than where more sinister and abhorrent However, prejudice of ery risk case.28 plays part in the decision when clearly and must minimized where sanction, pen- impose society’s ultimate present eradicated. tragic alty death.31 It is also a fact of against minorities in the Discrimination very part is much of this discrimination document- justice system criminal well country’s experience with the death say progress is not to penalty.32 Again majority ed.29 This and as Mitchell, jury handpicked by 99 were v. 443 U.S. S.Ct. number blacks 26. Rose 2993, for service. 61 L.Ed.2d commissioners 739 557, Rose, supra, at at 27. 443 U.S. 99 S.Ct. 3000. 558-59, Rose, supra, at 30. 443 U.S. at S.Ct. 99 3001. years Robespierre 200 contended almost 28. As ago: 238, See, e.g., Georgia, v. 408 92 31. Furman U.S. judicial you imagine perfect Even if most 2726, (see (1972) especially 33 L.Ed.2d 346 S.Ct. upright system, you if and even find the most J., Douglas, opinions concurring, at id. the 249-252, you enlightened judges, will still the most 2731-2733; Stewart, J., at con- 92 S.Ct. prejudice. place allow or have to error 309-310, 2762; curring, at Mar- id. at 92 S.Ct. (G. 1967). Robespierre ed. Rude 364-365, shall, J., concurring, id. at at 2790; C.J., Burger, dissenting, id. at n. See, 389-390 Virginia, e.g., U.S. Johnson v. 373 83 12; Powell, J., n. (1963) 92 S.Ct. at 2803-2804 (invalidating S.Ct. 2833). dissenting, at courtrooms); id. at segregated seating in Hamilton v. Alabama, 650, 84 (conviction (1964) in 32. This the death reversed when black de- historical discrimination 979 racially pointed out in on cross-exami- Justice Marshall fendant was demeaned Furman, nation); Mississippi, concurring opinion supra. 408 U.S. in Davis v. 89 his 364-65, (1969) (mass finger- at S.Ct. at a look ‘‘[i]ndeed L.Ed.2d rape regarding printing young in sus- the bare statistics executions blacks search Mitchell, overturned). betray pect enough also much of the discrimination.” See Rose opinions (racial grand jury supra Id. See also footnote 32 for other discrimination in selec- Britton, tion); (E.D. discussing Rogers F.Supp. and Furman racial discrimination Ark.1979). example, example penalty. very poignant and the death For between A recent 1980, 3,863 justice persons were executed discrimination the criminal racial States, Bailey or system 54% of those were blacks can be in the case of United found order, minority groups. men Vining, unpublished Of the 455 civ. act. no. 76-199 members (M.D.Ga.1978). Bailey, rape, black minori- declared executed for 89.5% were the court Dike, County, Capital system Punishment in the jury ties. Sarah T. selection in Putnam 2,307 States, p. people Of the The Office of United to be unconstitutional. during period, that time executed the South the Solicitor sent commissioners fifty-year During demonstrating they underrep- the same how could 1659 were black. memo executed, people period Georgia, grand of the 366 resent blacks and women traverse Fifty-eight exe- blacks were juries prima were black. but avoid a facie case discrimi- rape opposed three whites. percentage disparity cuted for because would nation robbery armed parameters blacks were executed for still Six be within Bedau, ed., Hugh A. supra 7-8 while no whites were. Circuit case law. See notes Fifth 1982). (3rd Penalty America ed. Death text. The result was that limited relevant out, points post-Furman new statutes factor in the decision impose whether to improved the have situation but the Baldus death penalty. Equal Protection very shows that race is still a real Clause the Fourteenth Amendment does capital in Georgia. factor in cases Some of not allow this result. The decision of the discrimination, this is conscious of it some district court the Baldus issue should be unconscious, but it is real nonetheless and reversed required state to submit important is that we least admit that evidence, available, if is disprove present. discrimination prima case facie plaintiff. made Finally, Georgia state of also

compelling justify penal- interest to a death

ty system that discriminates on the basis of Hypothetically,

race. if a racial bias re-

flected randomly itself of the convic- 20%

tions, one would abolish not the criminal

justice Ways of system. ridding sys- sought

tem of bias would be but absent a case,

showing bias in given a little else done. imperative

could be The societal

maintaining justice a criminal system to

apprehend, punish, perpetra- and confine tors of serious violations the law would America, UNITED STATES of outweigh mandate that race other Plaintiff-Appellee, prejudice process. the legal infiltrate words, accept In other we would have to doing that we are best can be done CRUZ-VALDEZ, Pedro Ruben Martin- system that must be administered Gonzalez, and Manuel Fortunado

people, all their conscious and uncon- Ariza-Fuentes, Defendants-Appellants. scious biases. No. 82-5310.

However, reasoning such sensibly cannot be invoked and bias cannot be tolerated Appeals, United States Court considering pun- when penalty, the death Eleventh Circuit. unique finality.33 ishment that is its prima evidence in this makes a case facie Jan. 1985. case penalty death being applied disproportionately because of P.A., Forman, McMaster & D. James percentage race. The differentials not McMaster, Miami, Fla., appointed, for court de minimis. To allow Cruz-Valdez. approve

under such circumstances is to Carroll, Miami, Fla., Linda ap- L. court preference in racial the most serious deci- pointed, for Martin-Gonzalez. system justice sion our criminal must Retter, Margaret make. This is a our E. Federal Public result Constitution Asst. Defender, Hermann, Robyn Deputy cannot J. Fed- tolerate. Defender, Miami, Fla., eral Public for Ari- majority in this case does not za-Fuentes. up this squarely face to' choice and its consequences. prejudice/preference Racial Bondi, Miami, Atty., Robert J. Asst. U.S. still both conscious and unconscious is a Fla., plaintiff-appellee. part capital process making decision Georgia. system To allow stand is to concede that in certain number

cases, the consideration race will be a See, Carolina, e.g., Woodson v. North notes Arlington Heights Metropolitan Village study ignores quantitative the Baldus difference Housing Development Corp., U.S. n. "looks, education, age, personality, in cases: 13, ("Because 97 S.Ct. at 564 n. of the nature demeanor, clothes, job, profession, morse____” re- task, however, jury-selection we have However, Majority opinion at 62. finding permitted a of constitutional violation it' is these differences that often are used to ap- pattern even when the statistical does not mask, intentionally unintentionally, either or proach the extremes of Yick Wo U.S. [118 prejudice. racial 30 L.Ed. or Gomillion 220] [364 110]"); see also Davis, States, Washington v. 25. See 426 U.S. at Bro. Teamsters v. United International 2051; Note, S.Ct. at Section 1981: Discriminato- (1977) (“We Impact, repeatedly ap- ry Purpose Disproportionate 80 Co- L.Ed.2d have proof proved use estab- 146-47 of statistical ... lum.L.R.

Case Details

Case Name: Warren McCleskey Cross-Appellant v. Ralph Kemp, Warden, Cross-Appellee

Court Name: Court of Appeals for the Eleventh Circuit

Date Published: Jan 29, 1985

Citation: 753 F.2d 877

Docket Number: 84-8176

Court Abbreviation: 11th Cir.