Petitioner's Reply Brief

    IN THE UNITED STATES DISTRICT COURT 
FOR THE NORTHERN DISTRICT OF GEORGIA 

ATLANTA DIVISION 

WARREN McCLESKEY, 

CIVIL ACTION FILE 

Petitioner, 
No. C81-2434A 

vs. 

WALTER ZANT, Warden, 
Georgia Diagnostic and 
Classified Center, 

N
t
”
 

N
t
”
 

a
t
”
 

N
t
?
 

u
w
 

u
t
?
 

w
a
l
 

C
u
l
 

w
w
 

u
r
;
 

S
a
u
?
 

Respondent. 

PETITIONER'S REPLY BRIEF 
  

Introduction. 
  

Comes now the petitioner, Warren McCleskey, and files this 

response to the "Post Hearing Brief on Behalf of Respondent.” 

The major arguments raised by respondent, for each claim, are 

addressed in the same order as briefed by respondent therein. 

Claim A. The Undisclosed Understanding Claim 
  

Respondent asserts [at page 9 of his brief], that relief is 

inappropriate with respect to the understanding between Atlanta 

Police Officer Sidney Dorsey and witness Offie Evans (1) because 

other evidence presented at trial would have allowed the jury to 

conclude that Mr. Evans had been impeached; and (2) no false 

testimony was presented from the witness in any manner. These 

are both incorrect. 

As to the suggestion that the evidence presented at trial 

would have allowed the jury to conclude that Evans had already 

been impeached, it is noteworthy that the trial prosecutor 

certainly did not believe Evans had been shown unworthy of 
ee 

 



  

belief. He made critical arguments to the jury based directly 

upon Evans' testimony. [Petitioner's 9/20/83 Brief, at pp. 3-4]. 

There is no basis to conclude that, given the evidence which was 

presented to the jury, as well as the evidence which was not, 

that failure to present the evidence regarding Evans' understand- 

ing with Detective Dorsey was harmless. 

As to the argument that no false testimony was presented by 

the witness in any manner, it is clear from the record that Evans 

was evasive regarding his understanding with Dorsey. When asked 

if such an understanding was motivating or underlying his cooper- 

ation with the prosecution, he stated simply that he wasn't 

worrying about the escape charge, thereby evading testimony on 

how he thought his cooperation with authorities might prove 

helpful, given his understanding with Detective Dorsey. (Tr., 

992). The State left this evasive testimony by Evans uncorrect- 

ed. 

Finally, although the Attorney General would have the Court 

understand that Evans's testimony was "not critical" Brief, at 

p. 10], the trial prosecutor thought it critical enough to recite 

it to the jury on the issue of malice. It can hardly be said 

that it was not critical, when it was the centerpiece of the 

prosecutor's argument on malice. (Tr., 975). 

Claim B. The Denial of Funds Claim 
  

The State argues that no request was made by trial counsel 

for a ballistics expert. [Brief, at 12]. However, the motion, 

as fairly read, includes a request for funds for a criminal 

investigator, which, if granted, would have permitted investiga- 

 



  

tion as to the availability of varying expert opinion on the 

subject of the identity of the murder weapon. (R. 33). Because 

petitioner was denied opportunity even for a criminal investi- 

gator, he was effectively denied the opportunity to develop the 

factual basis which would have led to the varying expert testi- 

mony which petitioner has shown could have been presented to the 

jury, but was not. 

The State also argues that, given the tentative nature of 

their own expert's identification of the murder weapon based upon 

his examination of the bullets, another expert's opinion would 

not have been sufficient to induce a reasonable doubt in the 

minds of a sufficient number of jurors to avoid a conviction. 

[Brief, at 17-18], 

That misses the point of the State's expert's actual testi- 

mony at trial. While he testified in terms of probabilities, his 

State trial testimony identified the .38 Rossi as the murder 

weapon, in a fairly conclusory fashion. Given what Fite himself 

stated in deposition testimony--that other specific makes of guns 

could have been the murder weapon--petitioner has met the appro- 

priate legal test. The State's expert's testimony at trial 

included only the conclusory opinion that the .38 Rossi was the 

weapon. That he indicated this in terms of probabilities was 

likely a subtle point lost on the jury, but which would not 

likely have been lost had another witness testified along the 

lines that Fite testified in his deposition. Indeed, that Fite 

indicated such in terms of probabilities cuts in petitioner's 

favor--showing the tentativeness of the view. What is critical 

 



  

is whether the opinion is subject to varying expert opinion. 

Fite's deposition shows that it is; that testimony was not 

presented to the jury. 

Claim C. The Improper Instructions Claim 
  

Subsequent to petitioner's prior briefing on this subject, 

the Eleventh Circuit Court of Appeals handed down its decision in 

Franklin v. Francis, F.2d + Bo, 83-8022 (11th Cir., 
  

November 16, 1983). That decision discusses the Sandstrom issue 
  

with a jury charge indistinguishable from the one used herein. 

The Eleventh Circuit concluded, as is the case herein: 

The problem with the charge on intent here is that 

the jury was never enlightened as to the nature of the 
burden on Franklin to rebut the presumption that he 
intended the killing. If the jury was persuaded that 
Franklin had to produce more than some evidence that he 
did not intend to kill, the burden shifted impermiss- 
ibly on an element essential for a malice murder 
verdict. 

The trial court's additional charge to the jury, 
(1) that criminal intent is not presumed and (2) that 
the burden to show every element of the crime is on the 
State, did not cure the faulty intent instruction. 
This charge was present in Sandstrom, and the court 
commented: 

The potential for [the impermissible] 
interpretations of the presumption was not 
removed by the other instructions given at the 
trial. It is true that the jury was 
instructed generally that the accused was 
presumed innocent until proved guilty, and 
that the State had the burden of proving 
beyond a reasonable doubt that the defendant 
caused the death of the deceased purposely or 
knowingly. But this is not rhetorically 
inconsistent with a conclusive or burden- 
shifting presumption. The jury could have 
interpreted the two sets of instructions as 
indicating that the presumption was a means by 
which proof beyond a reasonable doubt as to 
intent could be satisfied. 

  

422 U.8. al 518 n.7, 99.8.Ct. at 2456" n.2. <The same is 
true of the general burden allocation instructions 
here. . 

 



  

Neither did the instruction that criminal intent 
should not be presumed eliminate the vice Sandstrom 
condemns. This instruction at best conflicted with the 
challenged presumption; it did not explain it. At 
worst, the jury could have made the instructions 
consistent, interpreting the burden to be on the 
defendant to rebut the presumption that he intended to 
kill Mr. Collie, and on the State to show that the 
killing itself was criminal. Even if the jury believed 
that the two presumptions conflicted, it would be 
impossible for us to tell which one they decided to 
apply, or whether they applied something in between. 
The only way the charge as a whole could have cured the 
instruction would have been to explain by what quantum 
of evidence the defendant must rebut the presumption. 
If no instruction is given expressly to explain or 
limit an instruction that otherwise impermissibly 
shifts the burden of persuasion under Sandstrom, we 
inevitably face a situation where the reasonable juror 
could conclude from either one impermissible shift or 
several conflicting presumptions, some of which are 
impermissible, that the burden has shifted. 
Franklin v. Francis, slip opinion at 680-81. 

  

  

  

As to the harmless error analysis under Sandstrom, petition- 

er submits that the authority relied upon previously [4/8/82 

Brief, pp. 6-9] shows that it cannot be said that the improper 

instruction was harmless error. The jury was, under the State 

trial court's own view of the evidence, instructed on felony 

murder as well as malice murder. There was evidence before the 

jury, then, that would have supported a felony murder, rather 

than malice murder, conviction. The issue of malice, or intent, 

was dramatically contested in the dispute between prosecution and 

defense as to who were the believable witnesses--Ben Wright and 

Offie Evans, or Warren McCleskey. [See discussion and record 

cites at 8-9 of petitioner's 4/8/82 Brief]. 

The jury could have been left with the understanding that 

the burden was shifted to McCleskey to rebut the presumption that 

he intended the killing. If the jury was persuaded that 

 



  

McCleskey had to produce more than some evidence that he did not 

intend to kill, the burden shifted impermissibly on an element 

essential for a malice murder verdict. 

For the foregoing reasons, the writ should be granted on 

petitioner's Sandstrom claim. 
  

Claim E. The Improper Instructions & Evidence at Sentencing 
Phase 
  

The respondent submits that the sentencing phase instruction 

that the jury should consider "all of the evidence received in 

court, presented by the state and the defendant throughout the 

trial," [Tr., 1028], was not faulty because the jury would have 

understood that the same limitations applied as had applied 

during the guilt-innocence phase of the trial. (Brief, ay 28). 

This, of course, {andres the fact that at the outset, the 

trial court had given a near-all-inclusive instruction that the 

evidence of "other acts" could be considered, not simply to show 

identity of the triggerman, as was the purpose for which the 

prosecution had offered the evidence (Tr, 667-68), but also to 

show state of mind, motive, or intent. 

At a penalty phase using evidence of "other acts" to show 

"intent, state of mind, or motive" when deliberating on the 

defendant's character is tantamount to telling the jury they can 

use the evidence for any purpose whatsoever. 

It was therefore, error to instruct the jury that the 

evidence of other acts, for which petitioner had not been 

indicted or tried, could be used even for such "limited" purposes 

at the sentencing phase. Even if the evidence could properly be 

We 

 



  

used with appropriate instructions regarding such matters as 
  

standard of proof with respect to the other acts, no such 

instructions were given. 

As to the respondent's suggestion that petitioner has failed 

to show that the Douglas County convictions and life sentences 

are constitutionally infirm, the record now before the Court is 

unrebutted in that respect. Petitioner's Exhibit WM-1 and WM-2 

show that the sentences were set aside upon agreement of the 

district attorney, and that included in the state trial record 

was a motion made during the course of petitioner's trial that 

the search, upon which evidence in his trial was based, contra- 

vened the Fourth Amendment. 

Moreover, petitioner's testimony, that the prior life 

sentences were set aside because of the unconstitutional search, 

is unrebutted: 

OQ. What had developed during the course of the trial 

that had provided a basis for discussions with the 

District Attorney regarding a plea to a reduced charge 

other than the life sentences that had been imposed? 

A. When in the arresting stage, at that time I was 

living in Fulton County, and I believe Douglas County 

became, came to Fulton County on a warrant that was 

issued by Cobb County, I believe, and they came in 

another jurisdiction and made a search of all my 

apartment, and during the course of the trial, I 

believe my attorney was cross-examining one of the 

justices of the peace that issued the warrant and it 

was discovered that the warrant was illegal and was out 

of his jurisdiction to issue such a warrant. 
(Fed'l. Habeas Tr., 1822-23.) 

0. All right. Then is it your testimony that the 

basis for the, or the circumstances surrounding the 

District Attorney's agreement to a new trial as 

reflected in the transcript that's been labeled with 

WM-1, was the Motion to Suppress that had earlier been 

filed during the course of the trial? 

A... Yes. 

 



  

(Fed'l. Habeas Tr., 1824.) 

The Supreme Court has, itself recognized in Zant v. 

Stephens, UseS. r- 103 S.Ct. 2733: (1983), that use of evidence 
  

such as this, at the penalty phase, would be constitutionally 

impermissible. 

Claim G and H. The Arbitrary and Capricious and Racial 
Discrimination Claims 
  

Petitioner now turns to respondent's responses to his claims 

that the death penalty is being applied in an arbitrary and 

discriminatory manner in the State of Georgia. Respondent failed 

to respond directly either to petitioner's evidentiary presenta- 

tion or to the arguments he has presented in his prior memoranda 

to the Court. In short, respondent did not even begin to assume 

the burden it bears to rebut pe TNS prima facie case. Its 

brief consists of nothing more than speculative, piecemeal 

objections to petitioner's statistical evidence, nearly all of 

which were anticipated by petitioner, addressed and disposed of 

during the August, 1983 hearing. 

Yet, clear authority in this Circuit requires that once a 

petitioner has demonstrated a prima facie statistical case of 

discrimination, the burden shifts to respondent to rebut the case 

either (i) "by showing that plaintiff's statistics are misleading 

+ + « [11 or] by presenting legitimate nondiscriminatory reasons 

for the disparity," Eastland v. TVA, 704 F.2d 613, 618-19 (11th 
  

Cir. 1983). 1In challenging a party's statistical analysis, a 

respondent shoulders a "heavy burden," for it must come forward 

with affirmative evidence that the data are seriously deficient. 

 



  

Vuyanich v. Republic Nat'l Bank of Dallas, supra; accord, Trout 
  

  

v. Lehman, 702 F.2d 1094, 1101 (D.C. Cir. 1983); Detroit Police 
    

Officer's Ass'n v. Young, 608 F.24 671,687 {6th Cir.. 1979), 
  

cert. denied, 452 U.S. 938 (1981). See also Pet. Mem. 115-118, 
  

Pet. Supp. 6-7.1/ 

To meet his prima facie burden of proof, petitioner has 

offered the Court a wide range of statistical analyses which 

demonstrate that the race of the victim and, to a less pervasive 

degree, the race of the defendant, wields an independent effect 

on the likelihood of receiving a death sentence, as powerful as 

such legitimate factors as that the defendant had a prior capital 

record, that the murder was vile, horrible or inhuman, or that 

the victim was a policeman. 

While respondent's brief is replete with references to the 

"severe" flaws in petitioner's analyses, nowhere in its eviden- 

tiary presentation and nowhere in its brief does it demonstrate 

how these alleged imperfections could have biased petitioner's 

evidence so as to render statistically insignificant, or even 

marginally less significant, the demonstrated racial disparities. 

Though many of respondent's quarrels with petitioner's 

statistical case issue from a misunderstanding of quantitative 

proof, those that do voice reasonable general concerns were all 

addressed and resolved during the August hearing by petitioner. 

  

1/ Each reference to Petitioner's Post-Hearing Memorandum of 
Law, dated September 26, 1983, will be indicated by the 
abbreviation "Pet. Mem.". Each reference to Petitioner's 
Supplemental Memorandum of Law, dated November 1, 1983, will be 
indicated by "Pet. Supp." "Resp. Mem." refers to the 
Respondent's Memorandum of Law, dated November T4, 1983. 

-9- 

 



  

Indeed, at that time petitioner himself assumed the validity of 

respondent's gloomiest diagnoses and demonstrated that! none of 

them altered the racial disparities consistently found by peti- 

tioner. [see infra at 18-19]. 

As a last refuge, respondent has repeatedly suggested that 

no statistical analyses of racial disparities in capital sentenc- 

ing could ever satisfy its as yet unarticulated standards. This 

suggestion declines petitioner's good faith invitation to come 

forward with a serious scientific challenge to his study resort- 

ing instead to skepticism about the power of any scientific 

evidence to prove any proposition. Such general agnosticism 

provides respondent with a defense impervious to rational re- 

sponse; yet it would require this Court to reject the clear 

teachings of the Supreme Court and of this Circuit that statisti- 

cal evidence can and often does suffice to establish a constitu- 

tional claim. 

Respondent's one statistical attempt to dispel the appear- 

ance of disparate impact -- is obviously unavailing. Respondent 

and his experts, perhaps wisely, never tested their own hypothe- 

sis by any statistical procedures. Petitioner, however, demon- 

strated on rebuttal by rigorous statistical proof that the 

hypothesis is false: where cases at similar levels of aggrava- 

tion are compared, white victim cases are systematically more 

likely to receive death sentences. [see infra at 23-24}. Thus, 

racial disparities persist even when varying levels of aggrava- 

tion have been fully accounted for. 

-30~ 

 



  

The State's failure to present any serious response to 

petitioner's statistical case is underscored by the fact that its 

experts were either not qualified or not able to challenge the 

persistent racial disparities which petitioner has shown. Lack- 

ing any research experience outside the area of statistical 

computation, the State's principal expert witness, Dr. Joseph 

Katz, was not qualified to give his opinion on any aspect of the 

criminal justice system, or on questionnaire design, sample 

design, or other areas of research design. Tr.“ 1375. While Dr. 

Burford, the State's only other witness, was qualified in statis- 

tical analysis and research design, he merely testified generally 

about problems that might attend statistical analyses. Dr. 

Burford made no attempt to demonstrate that those problems did 

attend petitioner's statistical study.2/ In fact he acknowledged 

that assuming the correctness of petitioner's methods, petition- 

er's results show that the race of victim does have an effect on 

death sentencing outcome. Tr. 1671. 

In sum, then, even after receiving the response of respon- 

dent, petitioner can stand on his previous assertion that: 

In response to petitioner's prima facie case, the State 

of fered nothing more than 'unquantified, speculative, 

and theoretical objections to the proffered statis- 

tics,' Trout v. Lehman [702 F.2d 1094 1102 (D.C. Cir. 

1983], ignoring judicial warnings that ‘'the most 
  

  

2/ As petitioner's rebuttal expert, Dr. Richard Berk testified, 
Dr. Burford did not establish any "clear relationship between 

[the possible pitfalls in the use of multiple regression 

analyses] and the particular data analyses that Baldus and his 

colleagues undertook. Tr. 1779. Dr. Berk further testified 

that he "came away more convinced [about the validity of 

petitioner's results] because I thought Dr. Burford had 

particularly raised a variety of things that can go wrong in 

prigeipie and in this particular case didn't seem tO go wrong in 

fact.” Tr. 1785. 
i 

-11- 

 



  

effective way to rebut a statistically based prima 

facie case is to present more accurate statistics.' Id. 

The State, in short, presented no affirmative 

statistical case on rebuttal at all. 

Supp. Pet. at 21-22. 

Petitioner will now address the respondent's claims indivi- 

dually, providing references to the post-hearing transcript for 

the Court's convenience. 

A. The Role of Factors "Unique to" Individual Cases in 

Petitioner's Statistical Case 
  

The crux of respondent's challenge to petitioner's case is 

its speculation that "there are likely to be some unique factors 

present in individual cases which account for the imposition of 

the sentence." Resp. Mem. at 70. This claim is variously 

presented as an aspect of the problem with the "other" designa- 

tion on the Baldus questionnaires, with "unknowns" and "missing" 

variables, and with "subjective factors" affecting prosecutor or 

jury decisions. In a breathtaking leap of illogic, respondent 

postulates the existence of these additional factors -- variables 

so elusive as to escape identification by either the petitioner 

or the respondent -- and then concludes that "[such factors] 

could discount any possible race of victim or race of defendant 

effects that might otherwise appear." Resp. Mem. at 71. Because 

no statistical study can account for these "unique" factors, 

respondent's ultimate point amounts to an ultimate denial of the 

possibility of any statistical case. 

$d 

 



  

Petitioner and his experts have readily acknowledged that 

some "unique factors" may be predominant in explaining the 

sentencing decision in any individual case. However, such 

factors are simply irrelevant in discounting the persistent 

significance of the racial disparities which petitioner has 

demonstrated. Unless respondent can show that such "unique" or 

atypical variables are systematically related to race, occurring 

for example, only in white victim cases or in black victim cases, 

the impact of such factors cannot diminish the patterns of 

disparities based upon race that petitioner has reported. If 

such factors are randomly or evenly distributed, they can have no 

effect. Professor George Woodworth, petitioner's expert on 

statistical computation, succinctly stated this point: 

Errors which are randomly distributed with respect to 
the race of the defendant, or the race of the victim, 
are not going to create a race of victim disparity. In 
other words, they're not going to bias the effects. 

They, in effect, are simply additional terms that 
are added to the residual variation. 

Tr. 1727-28. Once again, respondent presented no expert testi- 

mony or other evidence to the contrary. 

"Unique" or subjective factors peculiar to individual cases 

thus cannot alter the racial effects petitioner has demonstrated 

unless they constitute systematic factors omitted from peti- 

tioner's statistical analyses. If this is the meaning of respon- 

dent's contention, respondent never named or otherwise identified 

-- much less tested the probable impact of -- any other system- 

atic variables. 

-13- 

 



  

Petitioner's experts have testified that no important 

variable was missing from his statistical case. Professor Baldus 

conducted analyses employing more than 230 variables, each of 

which was chosen on the basis of his extensive professional 

experience, after exhaustive analysis and research into Georgia's 

capital sentencing scheme. (See, e.g., DB 80) Tr. 801-805. 

Professor Richard Berk, another expert in the criminal justice 

system, confirmed Baldus' opinion that no important variable had 

been omitted from the analyses. Tr. 1754-57. Further, Baldus 

concluded numerous analyses to determine whether any combinations 

or "interaction" of variables might eliminate or diminish the 

race of victim effects. None 4id. (see, e.g., DB.83), Tr. 819, 

{DB 96), Tr. 917; {DB 98), Tr. 938; See also Supp. P's Mem. at 

16-20. In addition, Professor Woodworth conducted a comprehen- 

sive series of diagnostic tests to determine, for example, 

whether the existence of "missing" data, the influence of the 48 

most important cases, or the presence of possible "interaction" 

effects among the variables might explain the demonstrated racial 

disparities. Tr. 1250-1303.3/ His results confirmed the Baldus 

  

3/ In addition to the numerous models employing different 
variables, and different constellations of variables constructed 
by Professor Baldus, the Court also has before it "The Lawyer's 
Model," a model developed by the Court which included factors 
selected as likely to predict whether a homicide defendant would 
receive a capital sentence. Upon completion of extensive 
analyses, Professor Baldus submitted an affidavit which stated 
that, even employing the lawyer's model, he found 

"persistent race of victim effects and when the 
analysis focuses on the more aggravated cases, where 
there is a substantial risk of a death sentence, those 
effects increased substantially." 

Baldus ‘Aff, , at 10... See i4., at 19. - 

-}4= 

 



  

findings. 

By thus anticipating respondent's contention that other 

variables are at work in the capital sentencing process, peti- 

tioner went far beyond the rebuttal case that respondent himself 

actually presented. Indeed, respondent offered no analysis in 

which it had controlled for any variable. He did not propose, 

much less test the effect of, any plausible explanatory variable 

that Professor Baldus had not included. He proposed no alterna- 

tive model employing a different combination of variables that 

might plausibly reduce the racial factors. He pointed to no 

statistical analysis in which the racial effects disappeared or 

ran counter to petitioner's claims. Instead, at the August 

hearing, respondent blindly conjured up, and recalls in his 

brief, the apparition of unknowable "other" variables, any one of 

which "could" explain away petitioner's racial effects. This is 

not acceptable rebuttal evidence, for "unquantified, speculative, 

and theoretical objections to the profiered statistics are 

properly given little weight by the trial court,” Trout v. 
  

Lehman, supra, 702 F.2d at 1102, see also, Castaneda v. Partida,   

  

supra, 430 U.S. at 499, n.19; Eastland v. TVA, supra, 704 r.24 at 
  

22-3, n.14. With no more than this sole insubstantial foundation 

for its rebuttal case, it is not wonder that respondent retreats 

behind a general denial of the reliability of any statistical 

analysis challenging the administration of Georgia's capital 

statute. Yet this position too constitutes nothing more than 

"mere protestation and arguments of counsel," equally meritless 

and equally unbinding on this Court. 

-15- 

 



  

B. The Effect of "Unknown" and "Missing" Items on the 

Demonstrated Racial Disparities 
  

Turning from its speculation about unknown variables on 

which data was not sought, respondent also contends that peti- 

tioner's statistical evidence is flawed because of the occasioned 

missing data on which information was sought, but which was 

"unknown." At the August hearing, petitioner's expert responded 

to the problem of unknowns with statistically accepted methods, 

Resp. Mem. at 46; indeed, the unrebutted testimony of petition- 

er's other experts is that those methods were "state-of-the-art." 

Yet respondent now asserts in its brief that these sound tech- 

niques simply will not suffice for "this type of study." Resp. 

Br. at 46.4/ However, respondent never offered an alternative 

method or analysis which undermined petitioner's findings, and 

never demonstrated that the supposed absence of this information 

could have biased petitioner's results. This is a particularly 

  

4/ Respondent also challenges the presence on the question- 
naires of both studies of the "other" designation. Professor 
Baldus explained that he had created this designation as a 

cautionary measure to catch any unforseen, but recurring, 
variables likely to have an effect on the analyses. Without 
disputing the fact that this category generated only fragmentary 
information, respondent nevertheless suggested at the hearing 
that "there could be" "other" variables biasing petitioner's 
results. Resp. Supp. at 40. Petitioner's experts tested this 
speculative claim. Professor Baldus testified on rebuttal that 
he recoded all the "other" information, recomputed the analyses 
and found that the inclusion of the added information merely 
intensified the racial effects. Tr, 1710. . This laborious 
process thus merely confirmed Professor Baldus' prior expert 

testimony that information which occurs as rarely and randomly as 
that which appeared in the "other" category could not cause or 
explain petitioner's racial effects. Tr. 1709. = 

-16- 

 



  

telling failure in light of the fact that respondent has within 

its possession all the information in the files upon which 

decisionmakers relied. 

In contrast, petitioner's experts did undertake extensive 

analyses to test the possible biasing effects of "unknowns," a 

code designation which primarily encompassed information not 

present before the decisionmaker, as well as some information not 

known to the coder of the questionnaire. Professor Baldus 

testified that his method in the coding of unknown values was to 

code an item "unknown" or missing because of an absence of 

information in the files. Tr. 1684-86. The basis for that 

coding decision, he explained, was the assumption decisionmakers, 

including prosecutors and juries, normally act upon what is known 

to them; information not available cannot normally affect their 

decisions. Petitioner's rebuttal expert, Professor Berk, con- 

firmed that Baldus' coding assumptions were fully consistent with 

professional literature and practices. Tr. 1761-63. 

To secure this theoretical justification for "unknown" 

values, Professor Baldus conducted a battery of traditional 

statistical tests to determine the effects of the missing data on 

petitioner's results.5/ First, he performed regression analyses 

controlling for the racial factors, as well as nine statutory 

  

5/ Some apparently missing data was in fact not "missing" at 
all: many Of the items present in Dr. Katz' table identifying 
"missing values" were absent, not from Baldus' own data-gathering 
efforts, but from the magnetic tape provided by the Department of 
Offender Rehabilitation; other data related to characteristics of 
defendants which Baldus never used; still other "missing values" 
appeared as a result of Dr. Katz' misreading of code 
designations. Tr 1681-83. - 

7 

 



  

aggravating circumstances and the factor of prior record, in 

which he deleted all cases in which there were unknowns, or 

missing values (a method recommended by respondent's expert, Dr. 

Katz (See DB 120)). Tr. 1694. The only effect of the deletions 

was to increase the race of victim coefficient by .02. The 
  

race-of-defendant coefficient remained the same, though somewhat 

less statistically significant. (Compare DB 78 with DB 120). 

Tr. 1695, 

Baldus conducted yet another alternative analysis in which 

he assumed that every missing value would, if identified, run 

counter to his hypothesis, diminishing the racial effects. 

Recalculating his analysis in DB 78, Tr. 782-87, under those 

extreme, "worst case" assumptions, Baldus found the race-of- 

victim coefficient did drop from .07 to .05, but it remained 

highly statistically significant at the 1-in-100 level. (See DB 

122). The race-of-defendant coefficient dropped from .04 to .03, 

and remained non-significant. (See DB 123). Tr. 1703-1705. 

Because respondent also suggested that the absence of 

information on the race of the victim in a small number of cases 

might be important, Professor Baldus recorded those cases indulg- 

ing the extremely implausible assumption that all life sentence 

cases involved white victims and all death sentence cases, black 

victims. Again, the result of this "worst-case analysis” 

revealed persistent race-of-victim effects, with a very high 

degree of statistical significance. (See DB 124). Tr. 1706- 

1708. 

=38- 

 



  

Finally, Professor Berk, petitioner's rebuttal expert, 

testified that the amount of missing data in petitioner's statis- 

tical evidence did not even approach the "danger level" for 

social statistics in this area, and that the coherent character 

of petitioner's results and, in particular, the persistence of 

the racial effects supported the conclusion that missing vari- 

ables did not bias the analysis. Tr. 1764-66. Indeed, Professor 

Berk testified that the unusually small number of missing values, 

and the evidence that those values in no way flawed petitioner's 

results, was one powerful basis for his observation that 

We [members of the National Academy of Science] 
reviewed hundreds of studies on sentencing over this 
two-year period, and there's no doubt at this moment 
this is far and away the most complete and thorough 
analysis of sentencing that's been done. I mean 
there's nothing even close. There are several studies 
underway which are comparable, but they're not on death 
penalty and they're not that far along. 

C. Sample Design in the Charging and Sentencing Study 
  

Respondent has also called into question the accuracy of the 

sample design used in the Charging and Sentencing Study. Once 

again, petitioner's experts addressed these concerns with care, 

only to conclude that the impact of any inaccuracies in the 

sample design on petitioner's results is so slight as to be 

trivial. Respondent's claims that the absence of eleven missing 

penalty trials may have affected the weighting scheme of the 

sample design, for example, is its only specific challenge to the 

design. It was countered by Dr. Woodworth's unrebutted testimony 

of his calculations revealing that the likely impact of that 

“10 

 



  

missing information would be at most to affect the third decimal 

place of the racial coefficients. (E.G. .071 vs. .074)}. Ty, 

1727. 

Respondent's vaguer challenges to the accuracy of the sample 

design simply cannot withstand the professional judgment of 

petitioner's experts 1in statistical research and design, Tr. 

1209-1210; Tr. 1767, buttressed by the confirmatory testimony of 

a leading national specialist in research design, Tr. 1210, that 

the Baldus studies employed valid and statistically sound proce- 

dures. Indeed, Dr. Katz, the respondent's "expert" in this area 

was not even qualified to give his opinion on research design. 

Tr. 1368, 1374, 1377. Professors Baldus and Woodworth, on the 

other hand, testified that they had carefully devised two differ- 

ent samples cnly after consultation with a leading expert on 

sampling design. 

Though Professors Baldus and Woodworth were exacting in 

their development of these designs, professor Baldus undertook 

analyses to test the effect of the weighting procedures they had 

chosen on the demonstrated racial disparities. After drawing a 

25% random sample stratified only on sentencing outcome and 

weighted equally in each of the categories, Professor Baldus 

concluded that the racial effects were persistent, appearing with 

substantially the same magnitude in the weighted and unweighted 

samples. Tr 1712. . Dr. Berk, petitioner's rebuttal expert, 

confirmed the accuracy of the sample designs based on his 

scrutiny of the larger weighted sample and the 25 percent random 

-20- 

  

  

 



  

sample, adding only that technical literature supports the view 

that weighted samples tend to understate the effects of the 

variable of interest. Tr. 1767-68. 

D. Questionnaire Design 
  

The State's quarrels with the questionnaire design in the 

Charging and Sentencing Study and the Procedural Reform Study are 

easily answered.6/ 

Respondent challenged, for example, the Procedural Reform 

Study Questionnaire, which provided for "foils" for answering a 

number of questions, as well as the Charging and Sentencing Study 

Questionnaire, which used the foil method for coding two ques- 

tions relating to special aggravating features. Respondent 

contended that this method possibly prevented coders from enter- 

ing additional aggravating circumstances, thereby masking the 

level of aggravation in some cases. Professor Baldus, however, 

laid any doubts about the foil method in either study to rest by 

recording all the cases in which foils had been employed, adding 

otherwise unincluded variables. Baldus then recomputed his 

analyses only to find that the change in method merely enhanced 

the racial disparities slightly. Tr. 1100-1101. 

Though respondent found the instructions with respect to the 

coding of co-perpetrators "confused", its objections to that 

designation are certainly even more so: Edward Gates, the 

supervisor of much of the physical coding in the Charging and 

Sentencing Study, clearly testified that the questionnaires of 

  

6/ As a preliminary matter, we note that Dr. Joseph Katz, 
respondent's principal expert, was not qualified to give an 
opinion on questionnaire design. Tr. 1368, 1374 

-3 Y= 
r= 

 



  

co-perpetrators were coded together, either by one coder, or by 

two working together. Tr. 482-83, Tr. 581. In this way, the 

coders attempted to insure that the data gathered on co-perpetra- 

tors accurately reflected degrees of culpability and participa- 

tion in the crime. Further, the questionnaires themselves were 

designed to reflect subtle distinctions in a particular defen- 

dant's degree of involvement in a crime. Tr. 380.7/ 

E. Inconsistencies in Coding 
  

Respondent has also pointed to purported inconsistencies 

between the data entries for the two Baldus studies. Yet in 

addition to thoroughly training and supervising the coders, and 

guarding against errors of judgment in coding through the draft- 

ing of a comprehensive coding instruction book, Tr. 310-312, 395, 

in the Charging .and Sentencing Study, Professor Baldus also 

carefully collected information missing from thetauestiodna res 

and, with Professor Woodworth, used various techniques -- cross- 

checking between the PRS and CSS files, manually comparing 

computer entries with the case summaries, computing cross tabular 

computer runs for consistency between two logically related 

variables -- to identify any coding errors in the data. Tr. 

602-616. Among the approximately 500,000 total entries in the 

CSS study, Professor Baldus testified that he found only about 

  

77 Since the Charging and Sentencing Study examined decisions 
made much earlier in the charging and sentencing process than 
those examined in the prior Procedural Reform Study, additional 
questions were devised to obtain informtion on the strength of 
the evidence, especially important in cases originally charged as 
murders which resulted in pleas or convictions for manslaughter. 
Thus, the fate of coperpetrators in light of this enhanced 

information source was also reflected in the few coding 
variations reported. - 

“De 

 



  

200 errors. Tr. 616. These were entered on a computer program 

for correction, and could not have produced, or even affected 

significantly, the racial disparities that Baldus reported. 

F. Respondent's "Alternative Explanation" for the Demonstrated 

Racial Disparities 
  

Respondent's one theoretical attempt to dispel the appear- 

ance of the discriminatory application of Georgia's capital 

statute resorts to the assertion that white victim cases are 

generally more aggravated than black victim cases. This asser- 

tion is simply irrelevant: petitioner's observed racial dispari- 

ties in this case remain strong even when white victim and black 

victim cases at similar levels of aggravation, the critical basis 
  

for comparison, are analyzed. 

While respondent's brief lays out the analyses undertaken by 

its expert in reaching the conclusion that white victim cases are 

generally more aggravated than black victim cases, it never once 

coherently addresses petitioner's point that this admittedly 

greater average level of aggravation in white victim cases does 

nothing to explain why white victim cases are systematically 

sentenced in a harsher fashion when cases at similar levels of 

aggravation are compared. 

Not only is respondent's hypothesis beside the point; 

respondent never even tested the hypothesis by readily available 

statistical methods. Petitioner's expert witness, however, Dr. 

George Woodworth, conducted analyses that destroyed that specula- 

tive connection by comparing similarly situated cases. (See GW 

6, G¥ 7}. Tr. 1730-33. 

«dw 

 



  

Based as it is on factors which unrebutted evidence shows 

cannot account for the racial disparities evidenced in petition- 

er's statistical case, respondent's meager hypothesis amounts to 

nothing more than the "pretext" of an alternative explanation for 

those results. McDonnell Douglas Corp. v. Green, 411 U.S. 792, 
  

804-06 (1973). A rebuttal case predicated upon allegedly "legit- 

imate non-discriminatory reasons for the disparity" cannot 

succeed merely by challenging petitioner's prima facie case "in 

general terms," Wade v. Mississippi Cooperative Extension 
  

Service, 528 F.2d 508, 517 (5th Cir. 1976). See also Castenada 
  

V. Partida, supra, 430 U.5. at 499 n. 19. 
  

G. The Accuracy and Reliability of the Methods of 

Proof Petitioner Has Adopted 
  

Respondent's final point is that quantitative proof is 

complex and can be improperly applied or interpreted. Resp. Mem. 

at 67. Professor Baldus, as co-author of a comprehensive and 

highly influential text of the statistical proof of discrimina- 

tion, never disagreed. See D. Baldus & J. Cole, Statistical 

Proof of Discrimination (1980). Yet the scrupulous care which 

petitioner's experts demonstrated throughout their statistical 

analyses renders indefensible respondent's further suggestion 

that, because of the possible pitfalls in the use of statistical 

evidence, this Court should not place confidence in petitioner's 

results, 

The quibbles which underlie this point are easily addressed. 

Respondent's concern with possible "multicolinearity" was square- 

ly met by petitioner's rebuttal expert: first, petitioner's 

- 

Wed I 

 



  

analyses were prepared so as to reduce a multicolinear model; 

second, such models might underestimate, but do not overestimate 

the extent of possible discrimination. Tr. 1780-82. Third, 

Professor Baldus performed a "factor analysis" where the 

variables were recombined to avoid multicolinear results. In 

addition, one of the diagnostic tests which Professor Woodworth 

conducted to determine whether the statistical methods used were 

appropriate for the data was analyses of the presence of possible 

"interaction" effects among the variables. This analysis showed 

that such effects had no impact upon the racial disparities 

reported (see GW 4) Tr 1261-65. 

Respondent's additional suggestion that multiple regressions 

are not a reliable statistical tool is simply without any basis 

in the law. Clear precedent establishes that multiple regres- 

sions "may be the best, if not the only, means of proving class- 

wide discrimination . . . . in a case where a number of factors 

operate simultaneously to influence" the outcome of interest. 

Wilkins v. University of Houston, 1054 F.2d 388, 402-03 (5th Cir. 
  

1981) (See generally Supp. Mem. 10-15). The Eleventh Circuit has 

embraced multiple regression analyses as an appropriate tool for 

the proof of discrimination claims, relying in part on the 

discussion of that method in Baldus' and Cole's Statistical Proof 
  

of Discrimination. See, e.q,., Eastland v., TVA, supra, 707 F.24 
    

at 621-23. Respondent's quotation of Baldus' written reminders 

of the rigors of proper statistical proof is undoubtedly intended 

to suggest that no quantitative analyses can be reliable, yet 

courts nave repeatadly relied on the methods outlined in that 
- 

-25- 

 



  

text as accurate and precise in the statistical proof of discrim- 

ination. See, e.g., Vuyanich v. Republic Nat'l Bank of Dallas, 
    

supra, 505 F. Supp. 224 (N.D. Texas 1980). Moreover, it is 

precisely Professor Baldus' knowledge of the possible pitfalls of 

statistical analysis, that compels confidence in the care with 

which he avoided those pitfalls in petitioner's case. 

In sum, petitioner rests his claim of arbitrariness and 

racial discrimination on a statistical study so thorough-going 

that the strong racial effects, which persist throughout a long 

series of analyses employing every reliable method, and dozens of 

plausible alternative variables and hypotheses, can only reflect 

the existence in the Georgia system of racial factors systematic- 

ally influencing the decision to impose sentences of life or 

death. Not only has petitioner mapped the presence of those 

racial effects through unadjusted analyses, Tr. 672-82, as well 

as adjusted analyses, logistic regressions and weighted least 

squares regressions, which unrebutted evidence shows were proper- 

ly employed in this case. See, e.g., Tr. 755-56, 800-10, 905-16.   

He has also analyzed those racial effects at different procedural 

stages, (DB 93-96), Tr. 894-922, at the statewide level, the 

countywide level, (DB 106-108), Tr. 977-84, and at the level of 

petitioner's individual case. (DB 109-116), Tr. 989-994. (See, 

generally, Pet. Mem. 3-48; Supp. Mem. 16-23.) 

Petitioner thus has himself anticipated, and satisfactorily 

answered, challenges to his statistical case far more sophisti- 

cated and probing than the attacks brought by the respondent: 

Professor Baldus conducted an extensive series of alternative 
KK 

 



  

analyses to determine whether the employment of other "models" or 

groupings of relevant factors might possibly diminish or elimi- 

nate the strong racial effects his data had revealed. (See DB 

83) Tr. 819-31. Wie ghar Baldus employed his full file of 

recorded variables, a selection of 44 other variables most 

strongly associated with the likelihood of a death sentence, or 

selection of variables made according to other recognized statis- 

tical techniques, both the magnitude and the statistical signifi- 

cance of the race of victim factor remained persistent. (The 

race of defendant factor, as in earlier analyses, was more 

unstable.) Baldus' analyses of other rival explanations likely 

to reduce the demonstrated racial disparities =-- including the 

hypothesis that appellate review by the Georgia Supreme Court 

could eliminate the disparities and that the disparities may not 

reflect improvements in the Georgia system between 1974 and 1973 

-- were no better able to undermine the evidence in this case. 

(DB 103-105), Tr. 961-74. 

These results were secured by means of Professor Woodworth's 

rigorous scrutiny of the technical framework of the study. In 

addition to confronting diagnostic tests to establish the sound- 

ness of the statistical methods used, (GW 4), Tr. 1250-1287, in 

the Procedural Reform Study, Professor Woodworth carefully 

supervised the conversion of the data into a form suitable for 

statistical analyses, Tr. 1207-1209, and in the Charging and 

Sentencing Study, designed the sampling plan, including weighting 

techniques, according to "state-of-the-art" procedures. (GW 2), 

Tr. 1210-20. These judgments were strongly confirmed by peti- 
»- 

ly i A 

 



  

tioner's rebuttal expert. Tr. 1763, 1766, 1779-80. Indeed, the 

State's only expert qualified to give an opinion of petitioner's 

case, apart from its statistical framework, acknowledged that, 

assuming the soundness of petitioner's methods, his analyses did 

show the presence of persistent racial effects. Tr. 1671, 

Because petitioner's results compel the conclusion that the 

Georgia statutory scheme permits, and has in fact resulted in, 

the unequal imposition of capital punishment based upon distinc- 

tions of race, this Court should apply the Fourteenth Amendment's 

guarantee of equal protection of the laws to vacate petitioner's 

death sentence. 

Claim M. The Suggestive Lineup Claim 
  

Respondent asserts (at p. 77) that "no evidence has been 

shown which would establish that anyone was pointed out to any of 

the witnesses or that they were specifically asked to view these 

persons in the nature of a lineup." 

This is incorrect. The transcript of Classie Barnwell's 

testimony expressly shows that someone pointed out to her, at the 

very least, which of the prisoners in the box was co-defendant 

Ben Wright. (T. Tr. 230). When further efforts were made to 

establish whether other persons had also been pointed out to her; 

or who it was who had identified Ben Wright for her, she was 

unable to recollect. (H. Tr., 110-12). The very suggestive 

nature of the display is reflected in this testimony of Classie 

Barnwell, as well as the fact that petitioner was the only 

light-skinned defendant. (See 9/20/83 Brief, Pp... 17, notes 6 & 

7). 

-28- 

 



  

As to the identification made by Paul Ross subsequent to the 

viewing of petitioner in the suggestive display on the morning 

trial began, respondent simply relies upon the discussion of the 

Georgia Supreme Court, which fails to consider the factors 

established in United States v. Wade, 388 U.S. 218, 87 S.Ct. 1926 
  

(1967) as necessary inquiries. When those factors are consi- 

dered, it cannot be said that clear and convincing evidence 

demonstrates an independent origin for Ross's identification. 

[See Petitioner's 9/20/83 Brief, at 20-29]. 

Claim P. The Ineffective Assistance of Counsel Claim 
  

Respondent argues on this claim that the fundamental facts 

found by the state habeas court are entitled to a presumption of 

Cortech hess under 28 U.S.C. §2254 (d). This, like the State 

court's findings, misses the point. The State court's findings 

entirely ignore the undisputed facts upon which petitioner relies 

for relief in this case. 

There 1s, for example, no dispute but that trial counsel 

failed to conduct any independent investigation of any of the 

defenses recognized by trial counsel as possible prior to trial. 

The only "investigation" was a look at the prosecutor's files 

four days prior to the start of trial. None of these facts are 

found by the State court, yet are undisputed and at the core of 

the petitioner's ineffective assistance claim. (For additional 

facts, not in dispute, and not "found" by the State Court, see 

Petitioner's 9/20/83 Brief, 30-82). 

-29- 

 



  

Simply to argue that the State Court findings are entitled 

to a presumption of correctness fails to address the petitioner's 

claims in any substantive manner. Even respondent recognizes 

that "an examination of the effectiveness of trial counsel . . . 

requires an examination of the totality of the circumstances." 

[Brief, at 87]. This the respondent has never done; nor do the 

State Court findings reflect comprehensive findings directly 

meeting the substance of petitioner's claims. 

As to petitioner's specific claims that counsel was in- 

effective in his failure to investigate Offie Gene Evans's 

appearance on the witness list, respondent argues that defense 

counsel was "justified in relying on this statement by his client 

[that he had not made any incriminating statements or discussed 

the case with anyone] ." {Brief, at 92.}. 

Of course, defense counsel was justified in relying on this 

statement by his client--but, Turner himself recognized the risk 

that someone from Fulton County Jail might come and testify in a 

manner inconsistent with his client on this point. .(H. Tr., 75« 

76, 86). In that situation, it was incumbent upon Turner to do 

all he could to support his client. If the jury was to decide 

who was telling the truth on this point (i.e., McCleskey or 

Evans), on a critical issue of malice, as was the case as it 

developed at trial, McCleskey was entitled to counsel that did 

not leave him helpless in this regard. 

Had Turner conducted an investigation, there was substantial 

evidence he could have turned up--evidence of Evans' understand- 

ing with Atlanta detective Sidney Dorsey; Evans' recent drug 
- 

30 

 



  

abuse problems; Evans' history of drug abuse; Evans' recent claim 

to have worked for the State in a drug operation--all of which 

would have substantially bolstered McCleskey's claim that he was 

the one who was telling the truth on the issue of malice. [See 

Petitioner's 9/20/83 Brief, pp. 39-43]. 

Turner's abdication of his role as pre-trial investigator 

left his client helpless. The prosecution was, as a result, able 

to argue to the jury the malice issue, relying upon Evans' 

testimony regarding an alleged jailhouse conversation with 

McCleskey--a conversation McCleskey insists did not occur. 

As to petitioner's argument that it was incumbent upon trial 

counsel to contact the State's ballistics expert prior to trial, 

the respondent insists that had trial counsel done "a slightly 

better job in highlighting" the probability the murder weapon was 

other than the Rossi the State hypothesized, such would not be "a 

basis for a finding of substantial prejudice." [Brief, at 92]. 

But, Fite's tentative testimony in his deposition, when 

contrasted with his conclusory trial testimony, is more than a 

matter of highlighting. His specific recitation of examples of 

other makes of weapons that could have been the murder weapon is 

substantial evidence which the jury never heard, in part Pocauss 

of trial counsel's failure to investigate. 

Finally, respondent suggests that perhaps defense counsel 

looked at the file more than on the occasion reflected on the 

district attorney's file, and that, even if he did look at the 

3 

 



  

file for the first time only four days prior to trial, that*is 

not sufficient to conclude counsel was ineffective. (Brief, 93- 

94). 

The first suggestion is frivolous. Defense counsel himself 

testified that he only examined the prosecutor's file once (H. 

Tr., 38); and the file shows that to have occurred on October 5, 

1978. (Parker deposition, Exhibit A, attached to Petitioner's 

9/20/83 Motion for Supplement the Record). 

Further, while the failure to review the file until four 

days prior to trial may not, in itself, be enough to constitute 

ineffectiveness per se, it surely comes close. Essentially, it 

was the only pretrial investigation conducted independent of 

family members or petitioner, and leaving that to the eve of 

trial prevented any follow-up investigation which the prose- 

cutor's office itself had not done. So long as the prosecutor's 

office had done the investigation, petitioner would, perhaps, not 

have suffered. But, relying upon the prosecutor's office to do 

the investigation for the defense is ineffective assistance. 

That is particularly so here, in light of the concrete examples 

of favorable and mitigating evidence which could have been 

developed, but were not. The failure to review the file until 

“33 

 



  

four days prior to trial in light of this resultant harm, did 

constitute ineffective assistance of counsel. 

Respectfully submitted, 

[Cobent 3]. Dewey — 
  

Robert H. Stroup y 
1515 Healey Building 
Atlanta, Georgia 30303 

Jack Greenberg 
John Charles Boger 
99 Hudson St. 
New York, NY. 10013 

Timothy K. Ford 
600 Pioneer Bldg. 
Seattle, Washington 98136 

Anthony G. Amsterdam 
New York University Law School 
40 Washington Square South 
New York, NY 10012 

Attorneys for Petitioner 

33 

  

  

 



  

CERTIFICATE OF SERVICE 
  

I hereby certify that I have this day prior to filing served 

a copy of the within "Petitioner's Reply Brief" upon: 

Mary Beth Westmoreland, Esq. 
Assistant Attorney General 
132 State Judicial Building 
40 Capitol Square, S. W. 
Atlanta, Georgia 30334 

counsel of record for Respondent, by depositing a copy of same in 

the United States Mail, adequate postage affixed thereto. 

<S to 
This = day of January, 1984. 
  

  ROBERT H. STROUP