McCleskey v. Zant Brief for Petitioners Post-Hearing Memorandum of Law

Public Court Documents
September 26, 1983

McCleskey v. Zant Brief for Petitioners Post-Hearing Memorandum of Law preview

Memorandum of Law in support of petitioner's claims of arbitrariness and racial discrimination

Cite this item

  • Brief Collection, LDF Court Filings. McCleskey v. Zant Brief for Petitioners Post-Hearing Memorandum of Law, 1983. a4f01e66-bc9a-ee11-be36-6045bdeb8873. LDF Archives, Thurgood Marshall Institute. https://ldfrecollection.org/archives/archives-search/archives-item/5885c326-5fb3-49d3-8ae4-8d6d84f9cd71/mccleskey-v-zant-brief-for-petitioners-post-hearing-memorandum-of-law. Accessed July 20, 2025.

    Copied!

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

ATLANTA DIVISION

X

WARREN McCLESKEY, )
Petitioner, )

-against- )

WALTER D. ZANT, Superintendent, )
Georgia Diagnostic & Classification 
Center, )

CIVIL ACTION 
NO. C81-2434A

Respondent )

X

PETITIONER'S POST-HEARING MEMORANDUM OF LAW 
IN SUPPORT OF HIS CLAIMS OF ARBITRARINESS 

AND RACIAL DISCRIMINATION

ROBERT H. STROUP
1515 Healy Building 
Atlanta, Georgia 30303

JOHN CHARLES BOGER
10 Columbus Circle 
New York, New York 10019

TIMOTHY K. FORD
600 Pioneer Building 
Seattle, Washington 94305

ANTHONY G. AMSTERDAM
New York University Law School 
40 Washington Square South 
New York, New York 10012

ATTORNEYS FOR PETITIONER



TABLE OF CONTENTS

Page
INTRODUCTION ............................................  1

STATEMENT OF FACTS ................... .................. 3
I. Petitioner's Case-In-Chief .....................  3

A. Professor David Baldus .....................  3
1. Areas of Expertise .....................  3
2. Development of Research Objectives ..... 5
3. Procedural Reform Study ("PRS") ........  7

a. Design of PRS ......................  8
b. Data Collection for PRS ............ 11
c. Data Entry and Cleaning for PRS .... 12

4. Charging and Sentencing Study ("CSS") .. 13
a. Design of CSS ......................  14
b. Data Collection for CSS ............ 17

B. Edward Gates ................................  18

1. Data Collection for PRS ................  18
2. Data Collection for CSS ................  20

C . Professor David Baldus (resumed) ........... 21
1. Data Entry and Cleaning for CSS .......  21
2. Methods of Analysis ....................  23
3. Analysis of Racial Disparities ......... 24

a. Unadjusted Measures of Disparities . 24
b. Adjusted Measures of Disparities ... 25

4. Racial Disparities at Different
Procedural Stages ....................... 34

5. Analysis of Rival Hypotheses ........... 35
6. Fulton County Data .....................  36

a. Analysis of Statistical Dispari­
ties ................................  37

b. "Near Neighbors" Analysis .......... 39
c. Police Homicides ...................  40

7. Professor Baldus' Conclusions .......... 41
D. Dr. George Woodworth ........................  42

1. Area of Expertise ....................... 42
2. Responsibilities in the PRS .............  43
3. CSS Sampling Plan ....................... 44
4. Selection of Statistical Techniques .... 44
5. Diagnostic Tests ........................ 45
6. Models of the Observed Racial Dispari­

ties ..................................... 47

l



E. Lewis Slayton Deposition ...................  48

F. Other Evidence ..............................  48
II. Respondent's Case ................................ 49

A. Dr. Joseph Katz .............................. 49

1. Areas of Expertise .....................  49
2. Critiques of Petitioner's Studies .....  51

a. Use of Foil Method ................  51
b. Inconsistencies in the Data ......... 51
c. Treatment of Unknowns ..............  51

3. Dr. Katz's Conclusions ................  52
B. Dr. Robert Burford ..........................  52

1. Area of Expertise ......................  52
2. Pitfalls in the Use of Statistical

Analysis ................................  53
3. Dr. Burford's Conclusions ..............  54

III. Petitioner's Rebuttal Case .....................  54
A. Professor Baldus ............................  54

B. Dr. Woodworth ................................ 57

1. Statistical Issues .....................  57
2. Warren McClesky's Level of Aggravation . 58

C. Dr. Richard Berk ............................. 59

1. Areas of Expertise .....................  59
2. Quality of Petitioner's Studies .......  60
3. The Objections of Dr. Katz and Dr.

Burford .................................  61
D. The Lawyer's Model ..........................  62

ARGUMENT

Introduction: The Applicable Law ....................... 63
I. The Basic Equal Protection Principles ...........  69

A. The Nature of the Equal Protection
Violations ..................................  72

gage



Page

1. The Historical Purpose of the
Amendment ..............................  72

2. Traditional Equal Protection
Principles ........ *................... 77

3. Race as an Aggravating Circumstance ... 81A

B. The Issue of Standing .....................  84
II. The Standards for Evaluation of Petitioner's

Equal Protection Claim ......................... 86
A. The Issue of Discriminatory Intent ......... 86
B . The Legal Significance of the Statistical

Evidence ...................................  93
C . The Relevant Universe for Comparison of

Disparities ................................  104

1. Statewide Jurisdiction ................  104
2. The Relevant Decisionmaking Stages .... 109
3. Consideration of the Aggravation Level. 113

D. The State's Burden of Proof ............. 115

III. The Appropriate Relief ........................  124

CONCLUSION ............................................  126



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

ATLANTA DIVISION

WARREN McCLESKEY,

Petitioner,
-against-

-X

)

)
)

WALTER D. ZANT, Superintendent, )
Georgia Diagnostic & Classification 
Center, )

Respondent. )
-X

CIVIL ACTION 
NO. C81-2434A

PETITIONER'S POST-HEARING MEMORANDUM OF LAW 
IN SUPPORT OF HIS CLAIMS OF ARBITRARINESS 

AND RACIAL DISCRIMINATION

INTRODUCTION

Petitioner Warren McCleskey ("petitioner") has alleged in 

his petition for a writ of habeas corpus two related grounds 

for relief, both of which challenge the application of Georgia's 
capital statute: (i) that the "death penalty is administered

arbitrarily, capriciously, and whimsically in the State of 
Georgia (Habeas Petition, Claim G, 1(1[ 45-50); and, that 

(ii) it "is imposed ... pursuant to a pattern and practice ... 
to discriminate on the grounds of race" (Habeas Petition, Claim 
H, If 11 51-53), in violation of the Eighth Amendment and the 
Fourteenth Amendment of the Constitution.



This Court, in an order entered October 8, 1982, granted 
petitioner's motion for an evidentiary hearing on his claim of 
systemwide racial discrimination under the Equal ProtectionV
Clause of the Fourteenth Amendment. An evidentiary hearing

was held in Atlanta on August 8-19, 1983. Petitioner's case
in chief was presented through the testimony of two expert
witnesses, Professor David C. Baldus and Dr. George Woodworth,
as well as two principal lay witnesses, Edward Gates, and L.G.
Warr, an official employed by Georgia Board of Pardons and

2/
Paroles. Respondent Walter D. Zant ("respondent") offered 
the testimony of two expert witnesses, Dr. Joseph Katz and Dr. 
Roger Burford. In rebuttal, petitioner recalled Professor 
Baldus and Dr. Woodworth, and presented further expert testi­

mony from Dr. Richard Berk.
At the close of the hearing, the Court invited the parties 

to file memoranda of law setting forth their principal legal 

arguments. This memorandum is being submitted pursuant to that

1/ The Court noted in its order that "it appears ... that 
petitioner's Eighth Amendment argument has been rejected by 
this circuit in Spinkellink v. Wainwright, 578 F.2d 582, 612-14 
(5th Cir. 1978) ... [but] petitioner's Fourteenth Amendment claim 
may be appropriate for consideration in the context of statisti­
cal evidence which the petitioner proposes to present." Order 
of October 8, 1982, at 4.
2/ Petitioner also introduced the transcript of a deposition 
of Lewis Stayton, the District Attorney of the Atlanta Judicial 
Circuit, and offered brief testimony from petitioner's sister. 
Petitioner proffered a report by Professor Samuel Gross and 
Robert Mauro; the report was excluded from evidence by the Court.

2



invitation.

presented to 
tions of his

In it. petitioner will first outline the evidence
3/

the Court, and then state the legal founda- 

constitutional claims.

STATEMENT OF FACTS
I. Petitioner's Case-in-Chief 

A . Professor David Baldus 
1 . Areas of Expertise

Petitioner's first expert witness was Professor David C.
Baldus, currently Distinguished Professor of Law at the University
of Iowa. Professor Baldus testified that a principal focus of
his academic research and writing during the past decade has been

upon the use of empirical social scientific research in legal
contexts. During that time, Professor Baldus has co-authored a

4/
widely cited (see DB6) work on the law of discrimination, 
see D. BALDUS & J. COLE, STATISTICAL PROOF OF DISCRIMINATION 

(1980), as well as a number of significant articles analyzing the 
use of statistical techniques in the assessment of claims of

3/ Due to the length and complexity of the evidentiary hearing, 
and the fact that no transcript of the testimony has yet been 
completed, petitioner does not purport to set forth a comprehen­
sive statement of the evidence in this memorandum. Instead, the 
statement of facts will necessarily be confined to a review of 
the principal features of the evidence.
4/ Each reference to petitioner's exhibits will be indicated 
by a reference to the initials of the witness during whose 
testimony the exhibit was offered (e.g., David Baldus becomes 
"DB"), followed by the exhibit number.

3



discrimination. Professor Baldus has also authored

several important analytical articles on other death penalty
6/

issues. Professor Baldus served in 1975-1976 as the 
national Program Director for Law and Social Science of the 

National Science Foundation (DB1, at 1), and he has been re­
tained as a consultant to the Supreme Courts of Delaware and of 
South Dakota to propose empirical techniques for their appellate 
proportionality review of capital cases (DB1, at 4). Professor 
Baldus is currently the principal consultant to the Task 

Force of the National Center for State Courts on proportionality 
review of capital cases. He is the recipient of numerous grants 
and awards from the National Institute of Justice, the National 
Science Foundation, the Edna McConnell Clark Foundation, and 

other organizations for his professional research on discrimina­
tion in capital sentencing (jld. , 3-4). Professor Baldus has been
invited to serve on the Board of Editors of several distinguished

7/
journals concerning the issues of law and social science, and

5/

5/ See Baldus & Cole, "Quantitative Proof of Intentional Dis­
crimination," 1 EVAL. QUAR. 53 (1977); Cole & Baldus, "Statistical 
Modelling to Support a Claim of Intentional Discrimination," 
PROCEEDINGS, AM. STATIS. ASSN., SOC. SCI. SECTION.
6/ See Baldus & Cole, "A Comparison of the Work of Thorsten 
Sellin and Isaac Ehrlich on the Deterrent Effect of Capital 
Punishment," 85 YALE L.J. 170 (1976); Baldus, Pulaski, Wood- 
worth & Kyle, "Identifying Comparatively Excessive Sentences 
of Death," 33 STAN. L. REV. 601 (1980); Baldus, Pulaski & 
Woodworth, "Proportionality Review of Death Sentences: An 
Empirical Study of the Georgia Experience," J. CRIM. L. & 
CRIMINOLOGY (1983) (forthcoming).

7/ Evaluation Quarterly (1976-1979); Law and Policy Quarterly 
(1978-1979) (see DB1, at 3).

4



has served as a consultant to an eminent Special Committee on 

Empirical Data in Legal Decision-Making of the Association of the 
Bar of the City of New York.

After hearing his qualifications, the Court accepted 
Professor Baldus as an expert in "the empirical study of the 
legal system, with particular expertise in methods of analysis 
and proof of discrimination in a legal context."

2. Development of Research Objectives

Professor Baldus testified that he first became interested 
in empirical research on a state's application of its capital 

puhishment statutes shortly after Gregg v. Georgia, 428 U.S.
153 (1976) and related cases had been announced by the Supreme 

Court in mid-1976. Those cases, Baldus explained, explicitly 
rested upon certain assumptions about how the post-Furman 

capital statutes would operate: (i) that sentencing decisions 
would be guided and limited by the criteria set forth in 
capital statutes; (ii) that under such statutes, cases would 
receive evenhanded treatment; (iii) that appellate sentence 
review would guarantee statewide uniformity of treatment, by 
corrcting any significant disparities in local disposition of 

capital cases; and (iv) that the influenced of illegitimate 
factors such as race or sex, would be eliminated by these 

sentencing constraints on prosecutorial and jury discretion.

Professor Baldus testified that his own research and 
training led him to conclude that the Supreme Court's assump­

5



tions in Gregg were susceptible to rigorous empirical evalution

employing accepted statistical and social scientific methods.
Toward that end —  in collaboration with two colleagues, Dr.
George Woodworth, an Associate Professor of Statistics at the

University of Iowa, and Professor Charles Pulaski, a Professor
of Criminal Law now at Arizona State University Law School —
Baldus undertook in 1977 the preparation and planning of a major

research effort to evaluate the application of post-Furman
capital statutes. In the spring semester of 1977, Professor
Baldus began a review of previous professional literature on
capital sentencing research and related areas, which eventually
comprised examination of over one hundred books and articles.

8/
(See DB13.) Baldus and his colleagues also obtained access
to the most well-known prior data sets on the imposition of
capital sentences in the United States, including the Wolfgang

rape study which formed the empirical basis for the challenge
brought in Maxwell v. Bishop, 398 F.2d 138 (8th Cir. 1968),

rev'd on other grounds, 398 U.S. 262 (1970), and the Stanford
9/

Law School study. They examined the questionnaires em-

8/ Baldus testified that his research was particularly aided 
by other pioneering works on racial discrimination in the appli­
cation of capital statutes, see, e.g., Johnson, "The Negro and 
Crime," 217 ANNALS 93 (1941); Garfinkel, "Research Note on 
Inter- and Intra- Racial Homicide," 27 SOCIAL FORCES 369 (1949); 
Wolfgang & Riedel, "Race, Judicial Discretion, and the Death 
Penalty," 407 ANNALS 119 (1973); Wolfgang & Riedel, "Rape, Race, 
and the Death Penalty in Georgia," 45 AM. J. ORTHO PSYCHIAT.
658 (1975); Bowers & Pierce, "Arbitrariness and Discrimination 
under Post-Furman Capital Statutes," 26 CRIME & DELINQ. 563 (1980).
9/ See "A Study of the California Penalty Jury in First Degree 
Murder Cases," 21 STAN. L. REV. 1297 (1969).

6



researchers, and ran additional analyses to learn about factors

which might be important to the conduct of their own studies.
After these preliminary investigations, Baldus and his

colleagues began to formulate the general design of their own
research. They settled upon a retrospective non-experimental

±0/
study as the best available general method of investigation.
They then chose the State of Georgia as the jurisdiction

for study, based upon a consideration of such factors as the
widespread use in other jurisdictions of a Georgia-type capital

11/statute, the favorable accessibility of records in Georgia, 
and numbers of capital cases in that state sufficiently large 
to meet statistical requirements for analysis of data.

ployed in those studies, reran the analyses conducted by prior

3. Procedural Reform Study ("PRS”)

The first of the two Baldus studies, the Procedural 
Reform Study, was a multi-purpose effort designed not only to 
address the question of possible discrimination in the admin-

10/ Under such a design, researchers gather data from available 
records and other sources on plausible factors that might have 
affected an outcome of interest (here the imposition of sentence 
in a homicide case) in cases over a period of time. They then 
used statistical methods to analyze the relative incidence 
of those outcomes dependent upon the presence or absence of 
the other factors observed. Professor Baldus testified that this 
method was successfully employed in, among others, the National 
Halothane Study, which Baldus and his colleagues reviewed 
carefully for methodological assistance.

11/ Baldus testified that he made inquiry of the Georgia De­
partment of Offender Rehabilitation, the Georgia Department 
of Pardons and Paroles, and the Georgia Supreme Court, all of 
which eventually agreed to make their records on homicide 
cases available to him for research purposes. (See DB 24.)

7



istration of Georgia's capital statutes, but to examine appellate 
sentencing review, pre- and post-Furman sentencing, and other 

questions not directly relevant to the issues before this Court. 
Professor Baldus limited his testimony to those aspects and 

findings of the PRS germane to petitioner's claims.
The PRS, initially supported by a small grant from the Uni­

versity of Iowa Law Foundation, subsequently received major 
funding for data collection from the National Institute of 
Justice, as well as additional funds from Syracuse University 
Law School. Work in the final stages of data analysis was 
assisted by a grant from the Edna McConnell Clark Foundation 
distributed through the NAACP Legal Defense and Educational 
Fund, Inc. Research data collection and analysis for the PRS 
took place from 1977 through 1983.

a. Design of PRS
In formulating their research design for the PRS, Baldus 

and his colleagues first identified the legal decision-points 
within the Georgia charging and sentencing system which they 

would study and then settled upon the "universe" of cases on 

which they would seek information. After reviewing the various 
stages which characterize Georgia's procedure for the disposition 
of homicide cases (see DB21), Baldus decided to focus the PRS 
on two decision-points: the prosecutor's decision whether to

seek a death sentence once a murder conviction had been obtained

8



trial. Baldus defined the universe of cases to
include all persons arrested between the effective date of
Georgia's post-Furman capital statute, March 28, 1973, and
June 10, 1978 (i) who were convicted of murder after trial

and received either life or death sentences, or (ii) who
received death sentences after a plea of guilty, and who either
(i) appealed their cases to the Supreme Court of Georgia (ii)
or whose cases appeared in the files of both the Department
of Offender Rehabilitation ("DOR") and the Department of Pardons

12/
and Paroles ("DPP"). This universe comprised 594 defendants. 
(See DB 26.) Penalty trials had occurred in 193 of these 

cases, including 12 in which two or more penalty trials had 

taken place, for a total of 206 penalty trials. In all, 113 
death sentences had been imposed in these 206 trials.

For each case within this universe, Baldus and his col­

leagues proposed to collect comprehensive data on the crime, 
the defendant, and the victim. Factors were selected for inclu­

sion in the study based upon the prior research of Baldus, a 
review of questionnaires employed by other researchers such as 
Wolfgang as well as upon the judgment of Baldus, Pulaski and 
others about what factors might possibly influence prosecutors

12/ The decision to limit the universe to cases in which a 
murder conviction or plea had been obtained minimized concern 
about difference in the strength of evidence of guilt. The 
decision to limit the universe to cases in which an appeal had 
been taken or in which DOR and DPP files appeared was a necessary 
restriction based upon availability of data.

at trial; and the jury's sentencing verdict following a penalty

9



and juries in their sentencing decisions. The initial PRS 

questionnaire, titled the "Supreme Court Questionnaire," was 
drafted by Baldus working in collaboration with a law school 

graduate with an advanced degree in political science, Frederick 
Kyle (see DB 27), and went through many revisions incorporating 
the suggestions of Pulaski, Woodworth, and others with whom it 
was shared. In final form, the Supreme Court Questionnaire 
was 120 pages in length and addressed over 480 factors or "vari­
ables." After preliminary field use suggested the unwieldiness 
of the Supreme Court Questionnaire, and after analysis revealed 
a number of variables which provided little useful information, 
a second, somewhat more abbreviated instrument, titled the 

Georgia Parole Board (or Procedural Reform Study) Questionnaire, 
was developed (see DB 35). Much of the reduction in size of 
this second questionnaire came from changes in its physical 

design to re-format the same items more compactly. Other varia­
bles meant to permit a coder to indicate whether actors in the 
sentencing process had been "aware" of a particular variable were 
dropped as almost impossible to determine from available records 

in most instances. A few items were added to the second question­
naire. Eventually, information on 330 cases was coded onto the 
Supreme Court Questionnaire, while information on 351 cases was 

coded onto the Georgia Parole Board Questionnaire. Eighty-seven 
cases were coded onto both questionnaires. (See DB 28, at 

2. )

10



b. Data Collection for PRS

Data collection efforts for the PRS began in Georgia during 
the summer of 1979. Baldus recruited Frederick Kyle, who had 

assisted in drafting the Supreme Court Questionnaire, and two other 
students carefully selected by Baldus for their intelligence and 
willingness to undertake meticulous detail work. Initially, the 
Supreme Court Questionnaires were filled out on site in Georgia; 
quickly, however, it became evident that because of the unwield­

iness of that questionnaire, a better procedure would be to gather 
information in Georgia which would later be coded onto the 
questionnaires at the University of Iowa. Several items were 
collected for this purpose, including: (i) a Georgia Supreme
Court opinion, if one had been rendered (see DB 29); (ii) a trial 

judge's report prepared pursuant to Ga. Code Ann. § 27-2537(a), 
if one was available in the Georgia Supreme Court (see DB 30);

(iii) a "card summary" prepared by the Assistant to the Supreme 
Court of Georgia, if available (see DB 31); a procedural record 
of the case (see DB 32); (iv) an abstract of the facts, dic­

tated or prepared by the coders in Georgia from the appellate 
briefs in the case, supplemented by transcript information (see 
DB 33); and a narrative summary of the case (see DB 3, at 3).

In addition to those data sources, Baldus and his colleagues 
relied upon basic information on the crime, the defendant and the 
victim obtained from the Department of Pardons and Paroles, 
information on the defendant obtained from the Department of 

Offender Rehabilitation, information on the sex, race and age



of the victim —  if otherwise unavailable —  obtained from 
Georgia's Bureau of Vital Statistics, as well as information on 

whether or not a penalty trial had occurred, obtained from 
counsel in the cases if necessary (see DB 28; DB 36).

The 1979 data collection effort continued in the fall of
1980 under the direction of Edward Gates, a Yale graduate
highly recommended for his care and precision by former employers

at a Yale medical research facility. Baldus trained Gates and
his co-workers during a four-day training session in August,
1980, in the office of Georgia's Board of Pardons and Paroles,
familiarizing them with the documents, conducting dry run
tests in questionnaire completion, and discussing at length
any problems that arose.- To maintain consistency in coding,

Baldus developed a set of rules or protocols governing
coding of the instruments, which were followed by all the
coders. These protocols were reduced to written form, and a

copy was provided to Gates and other coders in August of 1980.
Baldus, who returned to Iowa, remained in contact with
Gates daily by telephone, answering any questions that may11/have arisen during the day's coding.

C . Data Entry and Cleaning for PRS

To code the abstracts and other material forwarded

13/ While information on most of the cases in the PRS was 
gathered in 1979 and 1980, Edward Gates completed the 
collection effort in the final 80 cases during the summer 
of 1981. (See DB 28, at 2.)

12



from Georgia onto the Supreme Court and PRS questionnaires, 

University of Iowa law students with criminal law course exper­
ience, again chosen for intelligence, diligence, and care 

in detailed work. The students received thorough training 

from Professors Baldus and Pulaski, and they worked under the 

supervision of Ralph Allen, a supervisor who checked each 
questionnaire. The students held regular weekly meetings to 
discuss with Professor Baldus and their supervisor any 
problems they had encountered, and consistent protocols were 
developed to guide coding in all areas.

Following the manual coding of the questionnaires, 
Professor Baldus hired the Laboratory for Political Research 
at the University of Iowa to enter the data onto magnetic 
computer tape. Rigorous procedures were developed to ensure 

accurate transposal of the data, including a special program 
to signal the entry of any unauthorized codes by programmers.

A printout of the data entered was carefully read by profes­

sionals against the original questionnaires to spot any errors, 
and a worksheet recorded any such errors for correction on the 
magnetic tapes (see DB 50).

3. Charging and Sentencing Study ("CSS")

In 1980, Professor Baldus was contacted for advice by the 
NAACP Legal Defense Fund in connection with a grant application 
being submitted to the Edna McConnell Clark Foundation seeking 

funds to conduct social scientific research into the death

13



penalty. Several months later, the Legal Defense Fund informed 

Baldus that the grant had been approved and invited him to con­

duct the research. Under that arrangement, the Legal Defense Fund 
would provide the funds for the out-of-pocket expenses of a study, 

ceding complete control over all details of the research and 
analysis to Professor Baldus (apart from the jurisdiction to be 
studied, which would be a joint decision). Once the analysis 
had been completed, Baldus would be available to testify concerning 
his conclusions if the Legal Defense Fund requested, but Baldus 
would be free to publish without restriction whatever findings

14/the study might uncover. After some further discussions, 
the parties agreed in the fall of 1980 to focus this Charging 
and Sentencing Study ("CSS") on the State of Georgia.

a. Design of CSS
The CSS, by focusing once again on the State of Georgia, 

permitted Professor Baldus and his colleagues to enlarge their 

PRS inquiry in several important respects: first, they were 

able, by identification of a different universe, to examine 
decision-points in Georgia's procedural process stretching back 

to the point of indictment, thereby including information 
on prosecutorial plea-bargaining decisions as well as jury guilt 
determinations; secondly, they broadened their inquiry to include

14/ Baldus indeed expressly informed LDF at the outset that 
his prior analysis of the Stanford Study data left him skep­
tical that any racial discrimination would be uncovered by 
such research.

14



cases resulting in voluntary manslaughter convictions as well 
as murder convictions; and thirdly by development of a new ques­
tionnaire, they were able to take into account strength-of- 
evidence variables not directly considered in the PRS. Beyond 
these advances, the deliberate overlapping of the two related 

studies provided Professor Baldus with a number of important means 
to confirm the accuracy and reliability of each study.

To obtain these benefits, Baldus defined a universe including 
all offenders who were arrested before January 1, 1980 for a 
homicide committed under Georgia's post-Furman capital statutes, 
who were subsequently convicted of murder or of voluntary man­
slaughter. From this universe of 2484 cases, Professors Baldus

15/
and Woodworth drew two samples. The first, devised accord­

ing to statistically valid and acceptable sampling procedures 

(see the testimony of Dr. Woodworth, infra), comprised a sample

of 1066 cases, stratified to include 100% of all death-sentenced
± 6/

cases, 100% of all life-sentenced cases afer a penalty 

trial, and a random sample of 41% of all life-sentenced cases 

without a penalty trial, and 35% of all voluntary manslaughter 
cases. The stratification had a second dimension; Professors 
Baldus and Woodworth designed the sample to include a minimum 
25% representation of cases from each of Georgia's 42 judicial 
circuits to ensure full statewide coverage.

15/ As indicated above, the PRS did not involve any sampling 
procedures. All cases within the universe as defined were 
subject to study.

16/ Because of the unavailability of records on one capitally- 
sentenced inmate, the final sample includes only 99% (127 of 128) 
of the death-sentenced cases.

15



The second sample employed by Baldus and Woodworth in the 
CSS included all penalty trial decisions known to have occurred 
during the relevant time period, on which records were available, 
a total of 253 of 254. Among those 253, 237 also appeared in the 

larger CSS Stratified Sample of 1066; the remaining 16 cases com­
prised 13 successive penalty trials for defendants whose 
initial sentences had been vacated, as well as 3 cases included 
in Georgia Supreme Court files, but not in the file of the 
Department of Offender Rehabilitation. (This latter sample, of 

course, permitted Baldus to analyze all penalty decisions 
during the period. In his analyses involving prosecutorial 

decisions, Baldus explained that, since a prosecutor's treatment 
on the first occasion inevitably would affect his disposition 
of the second, it could be misleading to count two dispositions 

of a defendant by a single decisionmaker on successive prosecutions. 
When two separate sentencing juries evaluated a capital defendant, 
however, no such problems arose. The two samples permitted both 
analyses to be employed throughout the CSS, as appropriate.)

After a universe had been defined and a sample drawn,
Baldus began development of a new questionnaire. Since the CSS 
sought to examine or "model" decisions made much earlier in the 

charging and sentencing process than those examined in the PRS, 

additional questions had to be devised to gather information on 

such matters as the plea bargaining process and jury conviction 
trials. A second major area of expansion was the effort to 
obtain information on the strength of the evidence, an especially

16



important factor since this study included cases originally 
charged as murders which resulted in pleas or convictions for 

manslaughter. Professor Baldus devised these strength-of-evi- 
dence questions after a thorough review of the professional 

literature and consultation with other experts who had also 
worked in this area. The final CSS questionnaires (see DB 38) 
also included additional variables on a defendant's prior record 

and other aggravating and mitigating factors suggested by profes­

sional colleagues, by attorneys and by preliminary evaluation 
of the PRS questionnaires.

b. Data Collection for CSS
Data for the CSS were collected from essentially the same 

sources used for the PRS: the Department of Pardons and Paroles, 
the Deparment of Offender Rehabilitation (see DB 40), the Supreme 
Court of Georgia, the Bureau of Vital Statistics (see DB 47), 
supplemented by limited inquiries to individual attorneys to 
obtain information on whether plea bargains occurred, whether 
penalty trials occurred, and the status (retained or appointed) 
of defense counsel (see DB 45, at 3-6; DB 46) (see generally 
DB 39).

Physical coding of the CSS questionnaires was completed 
directly from the official records in Georgia by five law students 

working under the supervision of Edward Gates, who had been 

one of Baldus' two coders for the PRS in Georgia in 1980.
The five students were selected by Baldus after a nationwide 
recruitment effort at 30 law schools; once again, Baldus

17



or Gates contacted references of the strongest candidates before 
hiring decisions were made (see DB 42).

As in the PRS, an elaborate written protocol to govern data 
entries was written, explained to the coders, and updated as 

questions arose. (See DB 43.) After a week-long training session 
in Atlanta under the supervision of Professor Baldus, Gates and 
the law students remained in contact with Baldus throughout the 
summer to resolve issues and questions that arose.

B. Edward Gates

At this point during the evidentiary hearing, petitioner 
presented the testimony of Edward Gates who, as indicated above, 
was integrally involved in data collection efforts both in the 
PRS and in the CSS. Gates testified that he was a 1977 grad­
uate of Yale University, with a Bachelor of Science degree in 
biology. Following his undergraduate training, Gates worked as 

a research assistant in the Cancer Research Laboratory of Tufts 
Medical School, developing data sets on cellular manipulation 

experiments, recording his observations and making measurements 

to be used in this medical research. (See EG 1.)

1. Data Collection for PRS

Gates testified that he was hired by Professor Baldus in 

August of 1980 to collect data for the PRS. Prior to travelling 
to Georgia, he was sent coding instructions and practice ques­
tionnaires to permit him to begin his training. During mid-

18



September, 1980, he met with Baldus in Atlanta, reviewed the 
practice questionnaires, and met with records officials in the 
Georgia Archives (where Supreme Court records were stored) and 

in the Department of Pardons and Paroles. After several 

additional days of training and coding practice, he worked at 
the Archives each workday from mid-September until late October, 
1980, reviewing trial transcripts, appellate briefs, trial 
judges's reports, and Supreme Court opinions before preparing 

abstracts and a narrative summary.
Gates testified that he followed the written coding 

procedures throughout, and that problems or inconsistencies were 
discussed with Professor Baldus each day at 4:00 p.m. When 
changes in coding procedures were made, Gates testified that he 
checked previously coded questionnaires to ensure consistent 

application of the new protocols.
In late October, coding work moved from the Archives to the 

Pardons and Paroles offices. There, Gates had access to police 

report summaries completed by Pardons and Paroles investigators, 

Federal Bureau of Investigation "rap sheets," field investigator 
reports on each defendant, and sometimes actual police or witness 
statements. Gates pointed out an illustrative example of a case 

he had coded (see DB 34) and reviewed at length the coding 
decisions he made in that case, one of over 200 he coded 
employing the Procedural Reform Study questionnaire. In 
response to questioning from the court, Gates explained that his 

instructions in coding the PRS questionnaire were to draw

19



reasonable inferences from the file in completing the foils. 

(These instructions later were altered, Gates noted, for 

purposes of the coding of the CSS questionnaire.)

Gates left Georgia in mid-January of 1981; he completed the 
final PRS questionnaires during the summer of 1981, during his 
tenure as supervisor of the CSS data collection effort in 
Atlanta.

2. Data Collection for CSS

During early 1981, Gates was invited by Professor Baldus to 
serve as project supervisor of the CSS data collection effort.
In the spring of 1981, he worked extensively with Baldus on a 

draft of the CSS questionnaire, assisted in hiring the coders 
for the 1981 project, and drafted a set of written instructions 
for the coders (see DB 4).

Gates came to Georgia in late May of 1981, participated 
with Professor Baldus in a week-long training session with the 

five law student coders, and then supervised their performance 
throughout the summer. He reviewed personally the files and 
questionnaries in each of the first one hundred cases coded by 

the students, to ensure consistency, and thereafter he regularly 
reviewed at least one case each day for each coder. At least 

twice during the summer, Gates gave all coders the same file and 
asked them to code and cross-check the results with those 

completed by the other coders. Gates spoke frequently by 
telephone with Baldus and discussed problems that arose in 

interpretation on a daily basis. As in earlier collection

20



efforts, the protocols resolving questions of interpretation 
were reduced to written form, the final end-of-summer draft of 
which is incorporated in DB 43 (EG 5). Gates testified that he 
made great efforts to ensure that all questionnaires were coded 
consistently, revising all previous coded questionnaires when a 
disputed issue was subsequently resolved.

Gates noted that for the CSS questionnaire, coders were 
given far less leeway than in the PRS to draw inferences from the 
record. Moreover, in the event of unresolved conflicting statements, 
they were instructed to code in a manner that would support the 
legitimacy of the conviction and sentence imposed in the case.

In sum, Gates testified that while the data for the PRS was 
very carefully coded, the data effort for the CSS was even more 
thoroughly entered, checked and reviewed. Both data collection 

efforts followed high standards of data collection, with 
rigorous efforts made to insure accuracy and consistency.

C. Professor David Baldus (resumed)

1. Data Entry and Cleaning for CSS
Upon receipt of six boxes of completed CSS questionnaires 

at the end of August,' 1981, Professor Baldus testified that he 
faced five principal tasks before data analysis could begin.
The first was to complete collection of any missing data, 

especially concerning the race of the victim, the occurrence of 

a plea bargain, and the occurrence of a penalty trial in life- 

sentenced cases. As in the PRS study, he accomplished this

21



task through inquiries directed to the Bureau of Vital Statistics 
(see DB 47) and to counsel in the cases (see DB 45-46). His 
second task was the entry of the data onto magnetic computer 
tapes, a responsibility performed under contract by the Laboratory 
for Political Science. The program director subsequently reported 
to Professor Baldus that, as as result of the careful data entry 
procedures employed,-including a special program that immediately 
identified the entry of any unauthorized code, the error remaining 

in the data base as a result of the data entry process is estimated 
to be less than 1/6 of 1 percent, and that the procedures he had 

followed conform to accepted social science data entry practices.

Baldus' third task was to merge magnetic tapes created by 
the Political Science Laboratory, which contained the data 
collected by his coders in Georgia, with the magnetic tapes 

provided by the Department of Offender Rehabilitation, which 
contained personal data on each offender. This was accomplished 
through development of a computer program under the supervision 
of Professor Woodworth. Next, Professors Baldus and Woodworth 

engaged in an extensive data "cleaning'' process, attempting 
through various techniques —  crosschecking between the PRS 

and CSS files, manually comparing entries with the case sum­
maries, completing crosstabular computer runs for consistency 

between two logically related variables —  to identify any 
coding errors in the data. Of course, upon identification,

22



Baldus entered a program to correct the errors. (See DB 51).
The final step preceding analysis was the "recoding" of 

variables from the format in which they appeared on the CSS 

questionnaire into a binary form appropriate for machine analysis. 
Professor Baldus performed this recoding (see DB 54, DB 55), 
limiting the study to 230+ recoded variables considered relevant 
for an assessment of the question at issue: whether Georgia's

charging and sentencing system might be affected by racial 

factors.

11/

2. Methods of Analysis
As the data was being collected and entered, Professor 

Baldus testified that he developed a general strategy of 
analysis. First, he would determine the patterns of homicides in 

Georgia and any disparities in the rate of imposition of death 
sentence by race. Then he would examine a series of alternative 
hypotheses that might explain any apparent racial disparities. 
Among these hypotheses were that any apparent disparities could 

be accounted for: (i) by the presence or absence of one or
more statutory aggravating circumstances; (ii) by the presence 

or absence of mitigating circumstances; (iii) by the strength of 

the evidence in the different cases; (iv) by the particular time 

period during which the sentences were imposed; (v) by the 
geographical area (urban or rural) in which the sentences were 

imposed; (vi) by whether judges or juries imposed sentence;

17/ Among the approximately 500,000 total entries in the CSS 
study, Professor Baldus testified that he found and corrected 
a total of perhaps 200 errors.

23



(vii) by the stage of the charging and sentencing system at 

which different cases were disposed; (viii) by other, less 

clearly anticipated, but nevertheless influential factors or 
combinations of factors; or (ix) by chance.

Professor Baldus also reasoned that if any racial dispari­
ties survived analysis by a variety of statistical techniques, 
employing a variety of measurements, directed at a number of 
different decision-points, principles of "triangulation” would 
leave him with great confidence that such disparities were real, 
persistent features of the Georgia system, rather than statis­
tical artifacts conditioned by a narrow set of assumptions or 
conditions.

For these related reasons, Professor Baldus and his 
colleagues proposed to subject their data to a wide variety of 

analyses, attentive throughout to whether any racial disparities 
remained stable.

3. Analysis of Racial Disparities

a. Unadjusted Measures of Disparities

Before subjecting his data to rigorous statistical

analyses, Professor Baldus spent time developing a sense for the
basic, unadjusted parameters of his data which could thereby
inform his later analysis. He first examined the overall
homicide and death sentencing rates during the 1974-1979 period

18/
(see DB 57), the disposition of homicide cases at

18/ Unless otherwise indicated, the Baldus exhibits reflect 
data from the CSS.

24



successive stages of the charging and sentencing process (see 

DB 58; DB 59) and the frequency distraction of each of the 
CSS variables among his universe of cases (see DB 60).

Next, Baldus did unadjusted analyses to determine whether 
the race-of-victim and race-of-defendant disparities reported 
by earlier researchers in Georgia would be reflected in his data 
as well. In fact, marked disparities did appear: while death
sentences were imposed in 11 percent of white victim cases, 
death sentences were imposed in only 1 percent of black victim 
cases, a 10 point unadjusted disparity (see DB 62). While a 
slightly higher percentage of white defendants received death 
sentences than black defendants (.07 vs. .04) (ijd. ) , when the 
victim/offender racial combinations were separated out, the 

pattern consistently reported by earlier researchers appeared:

Black Def./ 
White Vic.

.22
(50/228)

White Def./ 
White Vic.

.08
(58/745)

Black Def./ 
Black Vic.

. 01
(18/1438)

White Def./ 
Black Vic.

.03
(2/64)

b. Adjusted Measures of Disparities 

Baldus testified, of course, that he was well aware that 
these unadjusted racial disparities alone could not decisively 
answer the question whether racial factors in fact play a real 

and persistent part in the Georgia capital sentencing system.
To answer that question, a variety of additional explanatory 
factors would have to be considered as well. Baldus illustrated 

this point by observing that although the unadjusted impact of 
the presence or absence of the "(b)(8)" aggravating

25



circumstance on the likelihood of a death sentence

appeared to be 23 points (see DB 61), simultaneous consideration
or "control" for both (b)(8) and a single additional factor

20/
—  the presence or absence of the "(b)(10)" statutory factor

—  reduced the disparities reported for the (b)(8) factor from 
.23 to .04 in cases with (b )(10) present, and to -.03 in cases 
without the (b )(10) factor. (See DB 64.)

Baldus explained that another way to measure the impact of
a factor such as (b)(8) was by its coefficient in a least
squares regression. That coefficient would reflect the average
of the disparities within each of the separate subcategories, or
cells (here two cells, one with the (b )(10) factor present, and

one with (b )(10) absent). (See DB 64; DB 65.) Still another

measure of the impact of the factor would be by the use of
logistic regression procedures, which would produce both a
difficult-to-interpret coefficient and a more simply understood

"death odds multiplier," derived directly from the logistic
coefficient, which would reflect the extent to which the presence
of a particular factor, here (b)(8), might multiply the odds that

21/
a case would receive a death sentence. Baldus testified that,

19/ O.C.G.A. § 17-10-30.(b)(8) denominates the murder of a 
peace officer in the performance of his duties as an aggravating 
circumstance.
20/ O.C.G.A. § 17-10-30.(b)(10) denominates murder committed 
to avoid arrest as an aggravated murder.
21/ DB 64 reflects that the least squares coefficient for the 
(b)(8) factor was .02, the logistic coefficient was -.03, and 
the "death odds" multiplier was .97.

±9/

26



by means of regular and widely-accepted statistical calculations, 
these measures could be employed so as to assess the independent 

impact of a particular variable while controlling simultaneously 
for a multitude of separate additional variables.

Armed with these tools to measure the impact of a variable 
after controlling simultaneously for the effects of other 
variables, Professor Baldus began a series of analyses involving 
the race of the victim and the race of the defendant —  first con­
trolling only for the presence or absence of the other racial factor 

(see DB 69; DB 70), then controlling for the presence or absence 
of a felony murder circumstance (see DB 71; DB 72; DB 73), then 
controlling for the presence or absence of a serious prior 

record (see DB 74), then controlling simultaneously for felony 
murder and prior record (see.DB 77), and finally controlling 
simultaneously for nine statutory aggravating circumstances as 
well as prior record (see DB 78). In all these analyses, Baldus 
found that the race of the victim continued to play a substantial,

independent role, and the race of the defendant played a lesser,
22 /

somewhat more marginal, but not insignificant role as well.

22/ Professor Baldus testified concerning another important 
measure which affected the evaluation of his findings —  the 
measure of statistical significance. Expressed in parentheses 
throughout his tables and figures in terms of "p" values, (with 
a p-value of.10 or less being conventionally accepted as "margin­
ally significant," a p-value of .05 accepted as "significant," 
and a p-value of .01 or less accepted as "highly statisticaly 
significant"), this measure p computes the likelihood that, if in 
the universe as a whole no real differences exist, the reported 
differences could have been derived purely by chance. Baldus 
explained that a p-value of .05 means that only one time in 
twenty could a reported disparity have been derived by chance if, 
in fact, in the universe of cases, no such disparity existed. A 
p-value of .01 would reflect a one-in-one hundred likelihood, a 
p-value of .10 a ten-in-one hundred likelihood, that chance alone 
could explain the reported disparity.

27



Having testified to these preliminary findings, Professor

Baldus turned then to a series of more rigorous analyses (which
petitioner expressly contended to the court were responsive to
the criteria set forth by the Circuit Court in Smith v. Balkcom,

671 F.2d 858 (5th Cir. Unit B 1982) (on rehearing.)* In the
first of these (DB 79), Baldus found that when he took into
account or controlled simultaneously for all of Georgia's
statutory aggravating circumstances, as well as for 75 additional

mitigating factors, both the race of the victim and the race of
the defendant played a significant independent role in the

determination of the likelihood of a death sentence. Measured
, 23/

in a weighted least squares regression analysis, race of victim
displays a .10 point coefficient, a result very highly statist­

ically significant at the 1-in-1000 level. The logistic 
coefficient and the death odds multiplier of 8.2 are also very 
highly statistically significant. The race of defendant effect 
measured by least squares regression was .07, highly statist­
ically significant at the 1-in-100 level; employing logistic 
measures, however, the race of defendant coefficient was not 

statistically significant, and the death odds multiplier was 
1.4.

23/ Because the stratified CSS sample required weighting under 
accepted statistical techniques, a weighted least squares regres­
sion result is reflected. As an alternative measurement, Pro­
fessor Baldus performed the logistic regression here on the 
unweighted data. Both measures show significant disparities.

28



defendant effects measured after adjustment or control for a
graduated series of other factors, from none at all, to over 230
factors —  related to the crime, the defendant, the victim,

co-perpetrators as well as the strength of the evidence —
24/

simultaneously. (See DB 80.) Professor Baldus-emphasized
that as controls were imposed for additional factors, although 

the measure of the race-of-victim effect diminished slightly 
from .10 to .06, it remained persistent and highly statistically 
significant in each analysis. The race of defendant impact, 
although more unstable, nevertheless reflected a .06 impact in 
the analysis which controlled for 230+ factors simultaneously, 
highly significant at the 1-in-100 level.

Professor Baldus attempted to clarify the significance of 
these numbers by comparing the coefficients of the race-of- 
victim and race-of-defendant factors with those of other im­

portant factors relevant to capital sentencing decisions.
Exhibit DB 81 reflects that the race of the victim factor, 

measured by weighted least squares regression methods, plays 

a role in capital sentencing decisions in Georgia as signif­
icant as the (i) presence or absence of a prior record of 

murder, armed robbery or rape (a statutory aggravating circum­
stance —  (b)(1)); (ii) whether the defendant was the prime 
mover in planning the homicide, and plays a role virtually as

24/ This latter analysis controls for every recoded variable 
used by Professor Baldus in the CSS analyses, all of which are 
identified at DB 60.

Professor Baldus next reported the race-of-victim and

29



significant as two other statutory aggravating circumstances (the 

murder was committed to avoid arrest —  (b )(10) —  and the 

defendant was a prisoner or an escapee —  (b)(9)). The race 

of defendant, though slightly less important, yet appears a more 
significant factor than whether the victim was a stranger or an 

acquaintance, whether the defendant was under 17 years of age, 
or whether the defendant had a history of alcohol or drug abuse. 
The comparable logistic regression measures reported in DB 82, 
while varying in detail, tell the same story: the race of the
victim, and to a lesser extent the race of the defendant, 
play a role in capital sentencing decisions in Georgia more 
significant than many widely recognized legitimate factors.
The race of the victim indeed plays a role as important as many 
of Georgia's ten statutory aggravating circumstances in 
determining which defendants will receive a death sentence.

With these important results at hand, Professor Baldus 
began a series of alternative analyses to determine whether 

the employment of other "models" or groupings of relevant 
factors might possibly diminish or eliminate the strong racial 
effects his data had revealed. Exhibit DB 83 reflects the 
results of these analyses. Whether Baldus employed his full 

file of recoded variables, a selection of 44 other variables most 
strongly associated with the likelihood of a death sentence, or 

selections of variables made according to other recognized

30



both the magnitude and the statist-
25/

statistical techniques, 

ical significance of the race of the victim factor remained 
remarkably stable and persistent. (The race of the defendant 
factor, as in earlier analyses, was more unstable; although 

strong in the least squares analyses, it virtually disappeared in 
the logistic analyses.)

Baldus next, in a series of analyses (see DB 85- DB 87) 
examined the race-of-victim and defendant effects within the 
subcategories of homicide accompanied by one of the two statutory 
aggravating factors, —  (b)(2), contemporaneous felony, or 
(b)(7), horrible or inhuman —  which are present in the vast 
majority of all homicides that received a death sentence (see DB 
84). These analyses confirmed that within the subcategories 
of homicide most represented on Georgia's Death Row, the same 

racial influences persist, irrespective of the other factors 

controlled for simultaneously (see DB 85). Among the various 

subgroups of (b)(2) cases, subdivided further according to 
the kind of accompanying felony, the racial factors continue to 

play a role. (See DB 86; DB 87.)

25/ Two of Professor Baldus' analyses involved the use of 
step-wise regressions, in which a model is constructed by 
mechanically selecting, in successive "steps," the single factor 
which has the most significant impact on the death-sentencing 
outcome, and then the most significant remaining factor with the 
first, most significant factor removed. Baldus performed this 
step-wise analysis using both least squares and logistic 
regressions. Baldus also performed a factor analysis, in which 
the information coded in his variables is recombined into 
different "mathematical factors" to reduce the possibility that 
multicolinearity among closely related variables may be distorting 
the true effect of the racial factors.

31



analysis of the racial factors —  this method directly responsive
to respondent's unsupported suggestion that the disproportionate
death-sentencing rates among white victim cases can be explained
by the fact that such cases are systematicaly more aggravated.
To examine this suggesstion, Baldus divided all of the CSS cases
into eight, roughly equally-sized groups, based upon their overall
levels of aggravation as measured by an aggravation-mitigation 

26/
index. Baldus observed that in the less-aggravated categories, 

no race-of-victim or defendant disparities were found, since virtually 
no one received a death sentence. Among the three most aggravated 
groups of homicides, however, where a death sentence became a 
possibility, strong race-of-victim disparities, and weaker, but 
marginally significant race-of-defendant disparities, emerged.
(See DB 89.)

Baldus refined this analysis by dividing the 500 most 

aggravated cases into 8 subgroups according to his aggravation/ 

mitigation index. Among these 500 cases, the race-of-victim 
disparities were most dramatic in the mid-range of cases, those 

neither highly aggravated nor least aggravated where the latitude 
for the exercise of sentencing discretion was the greatest.

(See DB 90.) While death sentencing rates climbed overall as 
the cases became more aggravated, especially victims within the 
groups of the cases involving black defendants, such as petitioner 
McCleskey, the race-of-victim disparities in the mid-range

26/ Baldus noted that a similar method of analysis was a prominent 
feature of the National Halothane Study.

Professor Baldus then described yet another method of

32



reflected substantial race-of-victim disparities:
Black Def.

Category White Vic. Black Vic

3 .30 . 11
(3/10) (2/18)

4 .23 .0
(3/13) (0/15)

5 .35 .17
(9/26) (2/12/)

6 .38 .05
(3/8) (1/20)

7 .64 .39
(9/14) (5/13)

(DB 90.)

Race of defendant disparities, at least in white victim cases 

were also substantial, with black defendants involved in homi 
cides of white victims substantially more likely than white 
defendants to receive a death sentence.

White Vic.
Category Black Def. White Def.

3 .30 .03
(3/10) (1/39)

4 .23 .04
(3/13) (1/29)

5 .35 .20
(9/26) (4/20)

6 .38 .16
(3/8) (5/32)

7 .64 .39
(9/14) (5/39)

(DB 91. )

33



the hypothesis that racial factors play a significant role in
Georgia's capital sentencing system, but they conform to the
"liberation hypothesis" set forth in Kalven & Zeisel's The 

27/
American Jury. That hypothesis proposes that illegitimate 

sentencing considerations are most likely to come into play 
where the discretion afforded the decisionmaker -is greatest, 

i.e., where the facts are neither so overwhelmingly strong nor 
so weak that the sentencing outcome is foreordained.

4. Racial Disparities at Different Procedural Stages 
Another central issue of Professor Baldus' analysis, one 

made possible by the comprehensive data obtained in the CSS, 
was his effort to follow indicted murder cases through the 

charging and sentencing system, to determine at what procedural 
points the racial disparities manifested themselves. Baldus 
observed at the outset that, as expected, the proportion of 

white victim cases rose sharply as the cases advanced through 

the system, from 39 percent at indictment to 84 percent at 
death-sentencing, while the black defendant/white victim 

proportion rose even faster, from 9 percent to 39 percent.
(See DB 93.) The two most significant points affecting 

these changes were the prosecutor's decision on whether or 

not to permit a plea to voluntary manslaughter, and the prose­

cutor's decision, among convicted cases, of who to take on to a 
sentencing trial. (See DB 94.)

27/ H. KALVEN & H. ZEISEL, THE AMERICAN JURY 164-67 (1966).

These results, Professor Baldus suggested, not only support

34



The race-of-victim disparities for the prosecutor's decision 

on whether to seek a penalty trial are particularly striking, 
consistently substantial and very highly statistically significant 

in both the PRS and the CSS, irrespective of the number of 
variables or the model used to analyze the decision (see DB 95). 
The race-of-defendant disparities at this procedural stage were 
substantial in the CSS, though relatively minor and not statist­
ically significant in the PRS. (Id.) Logistic regression 
analysis reflects a similar pattern of disparities in both the 

CSS and the PRS. (see DB 96. ).

5. Analysis of Other Rival Hypotheses 

Professor Baldus then reported seriatim on a number of 
different alternative hypotheses that might have been thought 
likely to reduce or eliminate Georgia's persistent racial dispar­
ities. All were analyzed? none had any significant effects.
Baldus first hypothesized that appellate sentence review by the 

Georgia Supreme Court might eliminate the disparities. Yet 

while the coefficients were slightly reduced and the statistical 
significance measures dropped somewhat after appellate review, 
most models (apart from the stepwise regression models) continued 
to reflect real and significant race-of-victim disparities and 

somewhat less consistent, but observable race-of-defendant 

effects as well.

35



Baldus next hypothesized that the disparities do not reflect 

substantial changes or improvements that may have occurred in the 

Georgia system between 1974 and 1979. Yet when the cases were 
subdivided by two-year periods, although some minor fluctuations 

were observable, the disparities in the 1978-1979 period were 
almost identical to those in 1974-1975. (See. DB 103.) An 

urban-rural breakdown, undertaken to see whether different 
sentencing rates in different regions might produce a false 
impression of disparities despite evenhanded treatment within 
each region, produced instead evidence of racial disparities in 
both areas, (although stronger racial effects appeared to be 
present in rural areas (See DB 104.)) Finally, no discernable 
difference developed when sentencing decisions by juries alone 
were compared with decisions from by sentencing judges and 
juries. (See DB 105.)

6. Fulton County Data

Professor Baldus testified that, at the request of peti­
tioner, he conducted a series of further analyses on data drawn 
from Fulton County, where petitioner was convicted and sentenced. 
The purpose of the analyses was to determine whether or not the 
racial factors so clearly a part of the statewide capital 

sentencing system played a part in sentencing patterns in Fulton 
County as well. Since the smaller universe of Fulton County 

cases placed some inherent limits upon the statistical operations 
that could be conducted, Professor Baldus supplemented these 
statistical analyses with two "qualitative" studies: (i) a "near

36



neighbors" analysis of the treatment of other cases at a level of 

aggravation similar to that of petitioner; and (recognizing that 
petitioner's victim has been a police officer) an analysis of the 
treatment of other police victim cases in Fulton County.

a* Analysis of Statistical Disparities
Professor Baldus began his statistical analysis by observing 

the unadjusted disparities in treatment by victim/defendant 
racial combinations at six separate decision points in 
Fulton County's charging and sentencing system. The results 

show an overall pattern roughly similar to the statewide pattern:
Black Def. White Def. Black Def. White Def.
White Vic. White Vic. Black Vic. Black Vic.
.06

(3/52)
.05 .005 .0

(5/108) (2/412) (0/8)

(DB 106.) The unadjusted figures also suggest (i) a greater 

willingness by prosecutors to permit defendants to plead to 
voluntary manslaughter in black victim cases, (ii) a greater 
likelihood of receiving a conviction for murder in white victim 
cases, and (iii) a sharply higher death sentencing rate for white 
victim cases among cases advancing to a penalty phase. (DB 106; 
DB 107.) When Professor Baldus controlled for the presence or 

absence of each of Georgia's statutory aggravating circumstances 
separately, he found very clear patterns of race-of-victim 

disparities among those case categories in which death sentences 

were most frequently imposed (DB 108). Among (b)(2) and (b)(8) 

cases —  two aggravating cirstances present in petitioner's own

37



case —  the race-of-victim disparities were .09 and .20 respec­
tively (although the number of (b)(8) cases was too small to 
support a broad inference of discrimination).

When Professor Baldus controlled simultaneously for a host 
of variables, including 9 statutory aggravating circumstances, 
a large number of mitigating circumstances, and factors related 
to both the crime and the defendant (see DB 114 n.1 and DB 
96A, Schedule 3), strong and highly statistically significant 

race-of-victim disparities were evident in both the decision of 

prosecutors to accept a plea (-.55, p=.0001) and the decision to 
advance a case to a penalty trial after conviction (.20, p=.01) 

(DB 114). Race-of-defendant disparities were also substantial 
and statistically significant at the plea stage (-.40, p=.01) and 
at the stage where the prosecutor must decide whether to advance 
a case to a penalty trial (.19, p=.02) (DB 114). These racial 
disparities in fact, were even stronger in Fulton County 

than they were statewide.
Although the combined affects of all decision-points 

in this analysis for Fulton County did not display significant 
racial effects, Professor Baldus suggested that this was likely 

explained by the very small number of death-sentenced cases in 
Fulton County, which made precise statistical judgments on 

overall impact more difficult.

38



b. Near Neighbors" Analysis

Aware of the limits that this small universe of cases would 
impose on a full statistical analysis of Fulton County data, 

Professor Baldus undertook a qualitative analysis of those cases 
in Fulton County with a similar level of aggravation to petitioner 

the "near neighbors." Baldus identified these neighboring 
cases by creating an index through a multiple regression analysis 

of those non-suspect factors most predictive of the likelihood of 
a death sentence statewide. Baldus then rank-ordered all Fulton 
County cases by means of this index, and identified the group 
of cases nearest to petitioner. He then broke these cases, 32 
in all, into three subgroups —  more aggravated, typical, and 
less aggravated —  based upon a qualitative analysis of the 

case summaries in these 32 cases. Among these three subgroups, 
he calculated the death-sentencing rates by race-of-victim. As 

in the statewide patterns, no disparities existed in the less 

aggravated subcategory, since no death sentences were imposed 
there at all. In the "typical" and "more aggravated" sub­

categories, however, race-of-victim disparities of .40 and 

.42 respectively, appeared. (See DB 109; DB 110.) Professor 
Baldus testified that this near neighbors analysis strongly 

reinforced the evidence from the unadjusted figures that racial 

disparities, especially by race-of-victim, are at work not only 
statewide, but in Fulton County as well.

39



c. Police Homicides
Professor Baldus' final Fulton County analysis looked 

at the disposition of 10 police-victim homicides, involving 

18 defendants, in Fulton County since 1973. (See DB 115.)
Among these 18 potential cases, petitioner alone received

28/
a death sentence. Professor Baldus divided 17 of the cases 
into two subgroups, one subgroup of ten designated as "less 
aggravated," the other subgroup of seven designated as "aggra­

vated." (See DB 116.) The "aggravated" cases were defined 
to include triggerpersons who had committed a serious contem­
poraneous offense during the homicide. Among the seven aggra­
vated cases, three were permitted to plead guilty and two were 

convicted, but the prosecutor decided not to advance the cases 

to a penalty trial. Two additional cases involved convictions 
advanced to a penalty trial. In one of the two, petitioner's 
case, involving a white officer, a death sentence was imposed; 

in the other case, involving a black officer, a life sentence 

was imposed.
Although Professor Baldus was reluctant to draw any broad in­

ference from this analysis of a handful of cases, he did note 
that this low death-sentencing rate for police-victim cases in 

Fulton County paralleled the statewide pattern. Moreover, 

the results of this analysis were clearly consistent with peti­

tioner's overall hypothesis.

28/ One defendant, treated as mentally deranged by the system, 
was not included in the analysis.

40



7. Professor Baldus' Conclusions

In response to questions posed by petitioner's counsel 
(see DB 12), Professor Baldus offered his expert opinion —  
in reliance upon his own extensive analyses of the PRS and CSS 
studies, as well as his extensive review of the data, research 
and conclusions of other researchers —  that sentencing dis­
parities do exist in the State of Georgia based upon the race of 
the victim, that these disparities persist even when Georgia 

statutory aggravating factors, non-statutory aggravating factors, 
mitigating factors, and measures of the strength of the evidence 
are simultaneously taken into account. Professor Baldus further 
testified that these race-of-victim factors are evident at 
crucial stages in the charging and sentencing process of Fulton 

County as well, and that he has concluded that these factors 
have a real and significant impact on the imposition of death 

sentences in Georgia.
Professor Baldus also addressed the significance of the 

race-of-defendant factor. While he testified that it was not 
nearly so strong and persistent as the race of the victim, he 

noted that it did display some marginal effects overall, and that 
the black defendant/white victim racial combination appeared to 

have some real impact on sentencing decisions as well.

41



D• Dr. George Woodworth
1. Area of Expertise

Petitioner's second expert witness was Dr. George Woodworth, 
Associate Professor of Statistics and Director of the Statistical 
Consulting Center at the University of Iowa. Dr. Woodworth 

testified that he received graduate training as a theoretical 
statistician under a nationally recognized faculty at the 

University of Minnesota. (See GW 1.) One principal focus of 
his academic research during his graduate training and thereafter 

has been the analysis of "nonparametric" or discrete outcome 
data, such as that collected and analyzed in petitioner's case. 
After receiving his Ph.D. degree in statistics, Dr. Woodworth 
was offered an academic position in the Department of Statistics 

at Stanford University, where he first became professionally 
interested in applied statistical research. While at Stanford, 
Dr. Woodworth taught nonparametric statistical analysis, multi­
variate analysis and other related courses. He was also selected 

to conduct a comprehensive review of the statistical methodology 
employed in the National Halothane Study, for presentation to 

the National Research Council. Thereafter, upon accepting an 

invitation to come to the University of Iowa, Dr. Woodworth 
agreed to become the director of Iowa's Statistical Consulting 
Center, in which capacity he has reviewed and consulted as a 

statistician in ten to twenty empirical studies a year during 

the past eight years.

42



Dr. Woodworth has published in a number of premier 
refereed professional journals of statistics on nonparametric 
scaling tests and other questions related to his expertise 

in this case. He has also taught courses in "the theory of 
probability, statistical computation, applied statistics, 

and experimental design and methodology. In his research 
and consulting work, Dr. Woodworth has had extensive 

experience in the use of computers for computer-assisted 
statistical analysis.

After hearing his credentials, the Court qualified Dr. 
Woodworth as an expert in the theory and application of sta­
tistics and in statistical computation, especially of discrete 
outcome data such as that analyzed in the studies before the 
Court.

2. Responsibilities in the PRS
Dr. Woodworth testified that he worked closely with Professor 

Baldus in devising statistically valid and acceptable procedures 
for the selection of a universe of cases for inclusion 

in the PRS. Dr. Woodworth also reviewed the procedures 

governing the selection of cases to be included in the three 
subgroups on which data were collected at different times and 
with different instruments to ensure that acceptable principles 
of random case selection were employed.

Dr. Woodworth next oversaw the conversion of the data 

received from the PRS coders into a form suitable for statistical 
analysis, and he merged the several separate data sets into one

43



comprehensive file, carefully following established statistical 
and computer procedures. Dr. Woodworth also assisted in the 
cleaning of the PRS data, using computer techniques to uncover 

possible errors in the coding of the data.

3. CSS Sampling Plan
Dr. Woodworth's next principal responsibility was the 

design of the sampling plan for the CSS, including the develop­
ment of appropriate weighting techniques for the stratified 
design. In designing the sample, Dr. Woodworth consulted with 
Dr. Leon Burmeister, a leading national specialist in sampling 
procedures. Dr. Burmeister approved the CSS design, which Dr. 
Woodworth found to have employed valid and statistically accept­
able procedures throughout. Dr. Woodworth explained in detail 

how the sample was drawn, and how the weights for analysis of the 
CSS data were calculated, referring to the Appendices to GW 2 

(see GW 2, pp. 5ff.)

4. Selection of Statistical Techniques

Dr. Woodworth testified that he employed accepted statist­
ical and computer techniques in merging the various data files 
collected for the CSS, and in assisting in the data cleaning 

efforts which followed.
Dr. Woodworth also made the final decision on the appro­

priate statistical methods to be employed in the analysis of 

the CSS and PRS data. He testified at length concerning the

44



statistical assumptions involved in the use of weighted and un­
weighted least squares regressions, logistic regressions and 

index methods, and gave his professional opinion that each 
of those methods was properly employed in these analyses 

according to accepted statistical conventions. In particular,
Dr. Woodworth observed that while certain assumptions of least 
squares analysis appeared inappropriate to the data in these 
studies —  especially the assumption that any racial effects 
would exercise a constant influence across the full range of 
cases —  the use of that method did not distort the effects 
reported in the analyses, and its use allowed consideration of 
helpful and unbiased information about the racial effects.
Moreover, Dr. Woodworth noted that the alternative analyses 

which employed logistic regressions —  a form of regression analysis 

dependent upon assumptions closely conforming to the patterns of 

data observed in these studies —  also found the persistence of 

racial effects and showed that the use of least squares analysis 
could not account for the significant racial disparities observed.

5. Diagnostic Tests

Dr. Woodworth conducted a series of diagnostic tests 
to determine whether the methods that had been selected might 

have been inappropriate to the data. Table 1 of GW 4 reflects 

the results of those diagnostic tests, performed on five models 
that were used throughout the CSS analysis. For both the race 
of the victim and race of the defendant, Dr. Woodworth compared

45



coefficients under a weighted least squares regression 

analysis, an ordinary least squares regression analysis, a 
"worst case" approach (in which cases with "missing" values 
were systematically coded to legitimize the system and run 
counter to the hypotheses being tested), a weighted least 
squares analysis removing the most influential cases, a weighted 
least squares analysis accounting for possible "interactions" 

among variables, a weighted logistic regression analysis, and an 

unweighted logistic regression analysis. (GW 4, at Table 1.)

Dr. Woodworth also employed a conservative technique to cal­
culate the statistical significance of his results (see GW 3, at 
6 n.1, and Schedule II, for a calculation of Cressie's safe 
method) and a "modified Mantel-Haenzel Procedure (see GW 3, 

Schedules 1 and 3) to test the logistic regressions. These 
various diagnostic tests did not eliminate, and in most cases 
did not even substantially diminish, the race-of-victim effects. 
The levels of statistical significance remained strong, in most 
instances between two and three standard deviations, even 

employing Cressie's conservative "safe" method to calculate 
significance.

Dr. Woodworth testified that, after this extensive diagnos­

tic evalution, he was confident that the statistical procedures 
selected and employed in the PRS and CSS analyses were valid, 

and that the racial disparities found by the two studies were 

not produced by the use of inappropriate statistical methods or 
by incorrect specification of the statistical model.

46



6. Models of the Observed Racial Disparities 

Dr. Woodworth then directed the Court's attention to two 
figures he had developed to summarize the overall racial 

disparities in death-sentencing rates identified by the CSS 

study, employing the "mid-range” model in which both Dr. Wood- 
worth and Professor Baldus had expressed particular confidence. 

(See GW 5A and 5B.) As Dr. Woodworth explained, these figures 
represented the likelihood of receiving a death sentence 

at different levels of aggravation. Among black defendants such 
as petitioner (see GW 5B, Fig. 2), Dr. Woodworth noted that the 
death-sentencing rate in Georgia rises far more precipitously 
for white victim cases as aggravation levels increase than does 

the rate for black victim cases. For example, Dr. Woodworth 
observed, at the .4 level of aggravation, those black defendants 
who had killed white victims were exposed to a .15 point higher 
likelihood of receiving a death sentence. A similar disparity, 
based upon race of the victim, obtained among white defendants. 
(See GW 5A, Fig. 1.)

From these figures, Dr. Woodworth concluded that although 
white victim cases as a group are more aggravated than black 

victim cases, strong racial disparities exist in Georgia even 
when only those cases at similar levels of aggravation are 
compared.

47



E. Lewis Slayton Deposition
Petitioner offered, and the Court admitted pursuant to 

Rule 7 of the Rules Governing Section 2254 Cases, a transcript 
of the deposition of Lewis Slayton, the District Attorney for 
the Atlanta Judicial Circuit. In his deposition, while 
District Attorney Slayton stated several times that race did 
not play a role in sentencing decisions (Dep., at 78), he ac­

knowledged that his office had no express written or unwritten 

policies or guidelines to govern the disposition of homicide 
cases at the indictment stage (Dep., 10-12), the plea stage, 
(Dep., at 26) or the penalty stage (Dep., 31, 41, 58-59). 

Moreover, murder cases in his office are assigned at different 
stages to one of a dozen or more assistant district attorneys 
(Dep., 15, 45-48), and there is no one person who invariably 
reviews all decisions on homicide dispositions (Dep., 12-14, 
20-22, 28, 34-38). Slayton also agreed that his office does not 

always seek a sentencing trial in a capital case, even when 
statutory aggravating circumstances are present (Dep., 38-39). 
Slayton testified further that the decisionmaking process in his 
office for seeking a death sentence is "probably ... the 
same" as it was in the pre-Furman period (Dep., 59-61), and that 

the jury's likely verdict influences whether or not a case will 

move from conviction to a penalty trial (Dep. 31, 38-39).

F. Other Evidence

Petitioner offered the testimony of L. G. Warr, a parole 
officer employed by the Georgia Board of Pardons and Paroles.

48



Officer Warr acknowledged that in preparing the Parole Board 

reports used by Professor Baldus in his study, parole investi­
gators were obligated by statute and by the Board Manual of 

Procedure in all murder cases to speak with the prosecuting 

attorney and police officers if possible, soliciting records, 
witness interviews and other sources of information, including 
comments from the prosecutor not reflected in any written 
document or file. The Manual instructs investigators that it 
is imperative in cases involving personal violence to obtain 
information on all aggravating and mitigating circumstances.
The portions of the Manual admitted as LW 1 confirm Officer 

Warr's testimony.
Petitioner also introduced testimony from petitioner's 

sister, Betty Myer, that petitioner's trial jury included 

eleven whites and one black.
Finally, petitioner proffered a written report by Samuel 

Gross and Robert Mauro on charging and sentencing patterns in 

Georgia which was refused by the Court in the absence of live 

testimony from either of the report's authors.

II. Respondent's Case
Respondent offered the testimony of two expert witnesses, 

Dr. Joseph Katz and Dr. Roger Burford.

A. Dr. Joseph Katz
1. Areas of Expertise

Dr. Katz testified that he had received bachelors degrees

49



in mathematics and computer science from Louisiana State Univer­
sity. Katz received a Master degree in Mathematics and a 
Ph.D. degree in Quantitative Methods from L.S.U. A major 

focus of his professional research has been on input-output multi­
plier models used in the projection of economic developments 
by experts interested in regional growth. Dr Katz has taught 
various courses in basic statistics, operations research and 
linear programming in the Department of Quantitative Methods at 

L.S.U., in the Department of Management Information Sciences at 
the University of Arizona, and in the Department of Quantitative 
Methods at Georgia State University, where he is currently an 
Assistant Professor. Dr. Katz has published a number of articles 

on input-output multipliers in several refereed journals of 
regional science.

Respondent offered Dr. Katz as an expert on statistics, 
statistical analysis, quantitative methods, analysis of data, 
and research design. On voir dire, Dr. Katz acknowledged that 

he had no expertise at all in criminal justice or in the appli­
cation of statistics to criminal justice issues. Dr. Katz 
was unfamiliar with any literature or research in the area. 

(Counsel for the State expressly conceded that the State was not 
offering Dr. Katz to shed light in the criminal justice area.)

Moreover, Dr. Katz has only one prior academic or profes­
sional experience in the design of empirical research or the 

collection of empirical data —  and that one experience involved 

the gathering of Census data from library sources. He acknowl­

edged having taken no academic course in multivariate analysis.
50



Upon completion of voir dire, the Court agreed to accept 
Dr. Katz as an expert in statistics. The Court declined to 
qualify him as an expert in criminal justice, research design, 
or empirical research.

2. Critiques of Petitioner's Studies
a. Use of Foil Method

Over petitioner's objection predicated on his lack of exper­
tise, Dr. Katz was permitted to testify that the use of the foil 

method of data entry for some of the PRS variables might have 
resulted in the loss of some information in those instances in 
which there were insufficient foils. The foil method also 
prevented a coder from reflecting completely certain data 
because of the arrangement of several of the foils.

Dr. Katz admitted that the CSS questionnaire, which 
largely avoided any foil entries, was an improvement over the 
PRS questionnaires, although Dr. Katz faulted the one or two 
items in the CSS which reverted to a foil approach.

b. Inconsistencies in the Data

Dr. Katz testified that he had run cross-checks of variables 
present in cases included in both the PRS and the CSS that 

appeared to be identical. These checks uncovered what seemed to 

Dr. Katz to be a number of "mismatches," suggesting that data 
may have been entered erroneously in one study, or the other, or 
both.

c. Treatment of Unknowns

Dr. Katz presented several tables showing what he described
51



In his judgment, deletion of all casesas "missing values." 

with such missing values was necessary, thereby rendering 
any regression analysis virtually impossible.

3. Dr. Katz1 Conclusion

Dr. Katz hypothesized that the apparent racial disparities 
reflected in the PRS and CSS research might be explained if 

it were shown that white victim cases generally were more 

aggravated than black victim cases. Dr. Katz introduced a 
number of tables to establish that, as a whole, white victim 
cases in Georgia are more aggravated than black victim cases.
Dr. Katz admitted, however, that he had performed no 

analysis of similarly-situated black and white victim cases, 
controlling for the level of aggravation, nor had he performed 
any other analyses controlling for any variables that eliminated, 
or even diminished, the racial effects reported by Baldus and 
Woodworth.

B. Dr. Roger Burford

1. Area of Expertise
Dr. Burford testified that he was a Professor of Quanti­

tative Methods at Louisiana State University. He was also 

vice-president of a private research and consulting firm 
that conducts economic, market and public opinion research 
requiring extensive use of empirical methods. In his capacity 
as a consultant, Dr. Burford has testified as an expert 
witness between 100 and 150 times.

Dr. Burford has taught courses in sampling theory, 
research methods, multivariate analysis, computer simulation

52



modelling, and linear programming. He has published three 
textbooks on statistics and a wide range of articles on regional 

economic growth, computer simulation methods, and other topics.
Petitioner stipulated to Dr. Burford's expertise in the 

area of statistical analysis. On voir dire, Dr. Burford admitted 
that apart from his participation in the statistical analysis of 
one jury pool, he has had virtually no professional exposure to 
the criminal justice system and was not qualified as an expert 

in this area.

2. Pitfalls in the Use of Statistical Analysis
Dr. Burford testified that his involvement in the review 

of the PRS and CSS studies was largely as a consultant to 
Dr. Katz. Dr. Burford conducted almost no independent analysis 

of these studies, but rather reviewed materials generated 
by Dr. Katz. Dr. Burford believed that Dr. Katz' approach 

to the PRS and CSS studies was reasonable, and testified 
that it "could be useful" in evaluating these studies.

The remainder of Dr. Burford's testimony focused upon the 
general limitations of statistical analysis. He suggested 

that statistics can provide evidence, but cannot constitute 
"proof in a strict sense." Dr. Burford warned that regres­

sion analysis can be misused, especially if the underlying 

data are invalid. Data sets rarely meet all of the assump­
tions ideally required for the use of regression analysis. 
Possible multicolinearity, he warned, could confound regression 

results, although use of factor analysis admittedly reduces

53



the problems of multicolinearity. Dr. Burford also cautioned 

that step-wise regressions can result in an overfitted model 
and can thus be misleading.

3. Dr. Burford's Conclusions 

Dr. Burford did not offer any ultimate conclusions on the 
validity of the statistical methods used in the PRS and CSS 

studies. He did acknowledge on cross-examination that the 
regressions run by Baldus and Woodworth were "pretty conclusive."

III. Petitioner's Rebuttal Case

A. Professor Baldus
On rebuttal, Professor Baldus disposed of several issues 

raised by respondent. He first addressed the questions raised 
by Dr. Katz concerning certain of his coding conventions, 

especially the failure to distinguish in his machine analysis 
between items coded 1 ("expressly stated in the file") and items 
coded 2 ("suggested by the file") on the questionnaires. Baldus 
testified that to examine the effect of this challenged practice, 
he had completed additional analyses in which, for 26 aggravating 

and mitigating variables, he recoded to make distinctions 

between items coded 1 and 2, rather than collapsing the two 

categories into one. He found that the distinctions had no 
effect on the racial coefficients, and only marginally affected 

the level of statistical significance.

Turning to a criticism that, in multiple victim cases, 
information had not been coded concerning the characteristics 
of the second and successive victims, Professor Baldus again

W54



testified that he had conducted supplemental analyses to 

consider the problem. For the eight principal victim variables 

on which the questionnaires or case summaries contained sufficent 
information, he recoded the computer for each of the 50-60 

multiple victim cases, and then reran his analyses. The 

race-of-victim effects dropped by one-half of one percent,

Baldus reported, and the race-of-defendant effects remained 
unchanged.

Baldus next discussed Dr. Katz' table identifying "missing 
values." He explained that, in his 230+ variable models, the 
table would reflect approximately 30 missing values per 230- 
variable case. Baldus noted that much of the data that truly 
was missing was absent, not from Baldus' own data-gathering 

effort, but from the magnetic tape provided by the Department 
of Offender Rehabilitation. Moreover, most of such missing data 

related to characteristics of the defendants which had not been 
used in Professor Baldus' analyses in any event. Other data 

"missing" from one variable was in fact suppied by data present 
somewhere else in the questionnaire in another variable.

More centrally, Professor Baldus testifed that his entire 
philosphy in the coding of unknown values, fully consistent 
with most of the relevant professional literature, was to 
assume that wherever an item was coded "unknown" or missing 

because of an absence of information in the files, the decision­
maker, prosecutor or jury, necessarily had been forced to treat 

that factor as nonexistent. The basis for that assumption, he 
explained, is that rational judgments normally are made upon

55



what is known; information not available cannot normally affect 

a decision. Moreover, Baldus testified that he knew of nothing 
to suggest any systematic bias created by missing values or 
unknowns that might possibly affect the racial disparities 
observed.

As a further safeguard on this point, however, Baldus 

testified about a table reporting regression results, controlling 
for the racial factors as well as nine statutory aggravating 
circumstances and prior record, in which he had deleted all 
cases with missing values, a method recommended by Dr. Katz.
(See DB 120). The only effect of the deletions was to increase 

the race-of-victim coefficient by .02. The race-of-defendant 

coefficient remained the same, although somewhat less statisti­
cally significant (compare DB 78 with DB 120). A similar re­
sult occurred after reanalysis of the table reported in DB 121.

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 DB 78 under those extreme "worst case" 

assumptions, Baldus found that 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 also DB 123).

To counter Dr. Katz' further suggestion that the lack of 

information on the race of the victim in a small number of 
cases might be important, Professor Baldus recoded those cases,

56



assigning black victim variables in death cases and white victim
variables in life cases. Once 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) .

Finally, in addressing Dr. Katz' "mismatch" tables 

for the PRS and CSS files, Professor Baldus observed that some 

of the "mismatches" simply reflected Dr. Katz' misunderstanding 
of differences in variable definition between the two files.
Other "mismatches" occurred because Dr. Katz identified as 
errors certain discrepancies between the cases of co-defendants, 

unmindful that cases of co-defendants often reflect different or 
inconsistent factual versions of a single crime. In those 
mismatches where genuine discrepancies existed, Baldus noted, an 
analysis of the case summaries revealed that the error rate was 

higher in the PRS and lower in the CSS (on which most of the 
analyses relied.) Finally, Baldus noted that Dr. Katz had made 

no assertion that any systematic bias had been introduced by these 
few random errors.

B. Dr. Woodworth

1. Statistical Issues

Dr. Woodworth on rebuttal spoke to several additional 
minor points raised by the State. He first addressed the 
observation of Dr. Katz that an estimated eleven cases existed 

in the CSS in which penalty trials had occurred but had not been 
identified by Baldus' coders. Katz speculated that these 
eleven omissions might have adversely affected the weighting

57



scheme for the CSS sample. Dr. Woodworth acknowledged that 
eleven missing penalty trial cases would have affected the 
weighting scheme; however, he calculated the degree of likely 

impact as affecting the third decimal place of the racial 
coefficients (e.g., .071 vs. .074.)

Dr. Woodworth confirmed Professor Baldus' testimony that, 

from a statistical standpoint, the few inevitable, but insignifi­
cant errors that may have been identified by Dr. Katz' cross­
matching procedures could only have affected the racial coeffi­
cient if they had been systematic, rather than random, errors.

Dr. Woodworth next addressed an implication by Dr. Katz 
that since the level of statistical significance of the CSS 

racial disparities had dropped upon the introduction of 
additional variables to the model, the introduction of still 
further variables would eliminate statistical significance 

entirely. Through the use of a simple figure (see GW 6), Dr. 
Woodworth demonstrated the fallacy in Dr. Katz' reasoning, 

explaining that there was no statistically valid way to predict 

the effect of the addition of additional variables to a model.
2. Warren McClesky's Level of Aggragation

Finally, in response to a a question posed to him by the 
Court on petitioner's case-in-chief, Dr. Woodworth reported 
that, on the aggravation scale reported at GW 5A and 5B, Warren 
McClesky's case fell at the .52 level (see GW8). At that 
level, Dr. Woodworth explained, the disparities in black 
defendant cases dependent upon whether the victim was white or 
black was approximately 22 points.

58



Dr. Woodworth testified that, to arrive at the best overall 

figure measuring the likely impact of Georgia's racial dispari­
ties on a case at petitioner's level of aggravation, he had 
employed a triangulation approach, using three separate measures. 

From GW8, he drew a measure of 22 points; from DB 90, at level 
5 where petitioner's case is located, the disparity was 18
points; from Dr. Woodworth's recalculation of logistic proba-

29/
bilities, the disparity in the midrange model was 23 
points. Dr. Woodworth noted this "almost complete convergence" 

suggested a measure of the racial impact in a case at petitioner's 
level of over 20+ percentage points.

C. Dr. Richard Berk
1. Areas of Expertise

Petitioner's final rebuttal witness was Dr. Richard Berk, 

Professor of Sociology at the University of California at 
Santa Barbara. Dr. Berk has an undergraduate degree from Yale 
and a Ph.D from John Hopkins University. (See RB 1.) Dr. Berk 

has taught courses in econometrics, statistics, and research 
design, and has published extensively in the areas of criminal 

justice statistics and sentencing issues. Dr. Berk has served 
as a consultant to the National Institute of Justice, to the

29/ Both Baldus and Woodworth, as well as Dr. Burford testified 
that this or a similar model, which did not contain the hundreds 
of variables that might raise problems of multicolinearity, was 
probably the best model for measuring possible racial effects.

59



the counties of Baltimore and Santa Barbara, for which he has
designed jury selection systems. Dr. Katz has also served on a
select panel of the National Academy of Science which, during
the past two years, has examined virtually every major empirical

sentencing study ever conducted and formulated criteria for
30/

the conduct of such research. After hearing his testimony, 
the Court accepted Dr. Berk as an expert in statistics and in 

sociology.

California Attorney General's Committee on Statistics, and to

2. Quality of Petitioner's Studies 
Dr. Berk testified that he had received a copy of the 

magnetic tape containing the PRS and CSS studies some ten months 
prior to his testimony. During the intervening period, he 
had conducted some preliminary analyses on the data and had 
reviewed the Baldus and Woodworth preliminary report, as well as 
Dr. Katz' written evaluation of that report. Dr. Berk found both 
the PRS and CSS to be studies of "high credibility." He testified 
that among the hundreds of sentencing research efforts he had 

reviewed for the National Academy of Sciences, the Baldus and 

Woodworth studies were "far and away the most complete," that 
they employed "state of the art diagnostics," that the data 
quality was "very salient" —  in sum that he knew of no better 

published studies anywhere on any sentencing issue. Dr. Berk also 

commented favorably on such features of the studies as the

30/ The report of the Special Committee has been published as 
RESEARCH ON SENTENCING: THE SEARCH FOR REFORM (1983).

60



comprehensive use of alternative statistical analyses, the 

computer system employed, and Baldus' assumptions about the 

proper treatment of "unknowns" or "missing values." Moreover,

Dr. Berk testified that after reading the Katz report and 
hearing the testimony of Dr. Katz and Dr. Burford, he came 
away even more persuaded by the strength and reliability of 

petitioner's studies.

3. The Objections of Dr. Katz and Dr. Burford
Dr. Berk testified that he concurred with Dr. Burford's 

testimony listing possible pitfalls in the use of statistical 
analysis; however, Berk saw no evidence that the Baldus and 
Woodworth studies had fallen victim to any of these errors, 
and he did not understand Dr. Burford to have identified any 

serious weaknesses in either of the studies.
Turning to Dr. Katz' testimony, Dr. Berk first addressed 

the possible effects of multicolinearity on the racial dispari­
ties observed by Baldus. He noted that the diagnostics that had 

been performed by Dr. Woodworth failed to reveal serious multico­
linearity in the studies, but that such effects, even if serious, 
could have only dampened or diminished the racial effects.

Dr. Berk faulted the logic of Dr. Katz' suggestion that the 
more aggravated general level of white victim cases was a 
plausible hypothesis to explain the racial disparities observed. 

He noted that the important question was how white and black 
victim cases were treated at similar levels of aggravation; while

61



Dr. Katz had not even attempted to address this latter question, 
petitioner's experts had done so, and he found convincing Dr. 
Woodworth's proof that at similar levels of aggravation, marked 
differences were clear in the treatment of cases by race 
of the victim.

Addressing Professor Baldus' coding of "unknowns," Dr. Berk 
observed that the National Academy of Sciences committee had 
discussed this very question, concluding as did Professor Baldus 

that the proper course was to treat unknown data as having no 
influence on the decisionmaker. Berk further observed, respect­

ing the "missing data" problem, that missing data levels no 
greater than 10 to 15 percent of the total (the PRS and CSS 
figures were 6 percent or less) "almost never makes a difference" 
in the outcome of statistical analysis. Moreover, were such 
missing data having a serious effect on the studies, a predic­
table symptom would be a skewing or inverting of other anticipated 
effects, such as those of powerful determinants of sentence such 
as the statutory aggravating circumstances. In Baldus' studies, 
however, no such symptons appeared, leading Dr. Berk to discount 

missing data as a serious problem.

D. The Lawyer's Model

Several weeks after the August, 1983 evidentiary hearing, 

Professor Baldus submitted an affidavit describing in detail 
the results of an analysis employing a model developed by the 
Court, including factors selected as likely to predict whether a 
homicide case would receive a capital sentence. The race-of-

62



victim disparities reported by Professor Baldus upon completion 

of extensive analyses using the Lawyer's Model were fully 

consistent with the results presented during the evidentiary 
hearing:

"There are persistent race of victim effects 
and when the analysis focuses on the more 
aggravated cases, where there is a substan­
tial risk of a death sentence, those effects 
increase substantially.

Baldus Aff., at 10. See id., at 19.

63



ARGUMENT

Introduction: The Applicable Law

Petitioner's central claim is that his death sentence has 
been imposed under a statutory scheme which permits, and has in 
fact resulted in, the unequal imposition of capital punishment 
based upon the race of the defendant and the race of the victim. 
This persistent disparity in the valuation of white life over 
black life in Georgia's death sentencing system means there is 
a substantial likelihood that petitioner was sentenced to die 
for a crime that would not have incurred the death penalty had 

he been white or his alleged victim black, and this fact inval­

idates his death sentence.
Petitioner maintains that the consideration of race by 

various decisionmakers in Georgia's capital sentencing process 
implicates both the Eighth Amendment's prohibition against cruel 
and unusual punishment and the Fourteenth Amendment's guarantee 
of equal protection of the law. He recognizes, however, that 

this Court views the Eighth Amendment claim as foreclosed by
the Fifth Circuit's decision in Spinkellink v. Wainwright, 57811/F.2d 582 (5th Cir.), cert. denied, 440 U.S. 976 (1978). In

31/ In Spinkel]ink, the Circuit Court recognized that the 
Eighth Amendment concept of "arbitrariness" developed in Furman 
v. Georgia, 408 U.S. 238 (1972), encompasses racial discrim­
ination as well, and that Furman prohibits discrimination by 
either the race of the victim or by the race of the defendant, 
id. at 613 n.38, 614 n.40. Yet Spinkellink interpreted the 
Supreme Court's 1976 death penalty decisions "as holding that 
if a state follows a properly drawn statute in imposing the 
death penalty, then the arbitrariness and capriciousness —

-64-



to reserve his Eighth Amendment claim and will not separately
32/

set forth the arguments underlying it in this brief. His
argument here, as a consequence, is limited to a development of 
the equal protection claim, which has been recognized as a 
legitimate ground for constitutional challenge in the context of 

capital cases under the law of this Circuit. See Spinkellink v.

light of that Court's position, petitioner respectfully proposes

31/ continued
and therefore the racial discrimination —  condemned in Furman 
have been conclusively removed." Id. at 613-614 (footnotes 
omitted). The Supreme Court's 1982 per curiam decision in 
Zant v. Stephens, 456 U.S. 410 (1982) best states petitioner's 
disagreement with this disposition of his Eighth Amendment 
claim: the Court in Zant explicitly stated that the statutes
it considered in 1976 were upheld because they "promised to 
alleviate the arbitrariness condemned in Furman," but that 
their "constitutionality ultimately would depend" on the 
fulfillment of that promise. 456 U.S. at 413. See also 
Proffitt v. Wainwright, 685 F.2d 1227, 1261, n.52 (11th Cir.
1982)(questioning the continuing validity of this holding in 
Spinkellink).
32/ Although we will not elaborate on the Eighth Amendment 
argument here, we would point out that in many respects the 
prohibitions of the Eighth Amendment track those of the Equal 
Protection Clause. Furman surely holds that the Eighth Amend­
ment imposes more rigorous restraints on state action than the 
Fourteenth, at least where capital punishment is involved. The 
major difference between the standards applicable to a showing 
of unequal enforcement under the two constitutional provisions 
seems to be that under the Eighth Amendment such a pattern may 
be found unconstitutional regardless of any showing of invid­
ious intent. See Furman v. Georgia, 408 U.S. 238 (1972); Woodson 
v. North Carolina, 428 U.S. 280 (1976); Godfrey v. Georgia, 446 
U.S. 420 (1980). In capital sentencing, under the Eighth Amend­
ment, "[i]t is of vital importance to the defendant and to the 
community that any decision to impose the death sentence be, 
and appear to be, based on reason rather than on caprice and 
emotion." Gardner v. Florida, 430 U.S. 349, 358 (1977).
(emphasis added.)

65



Wainwright, supra; Smith v. Balkcom, 660 F.2d 573 (5th Cir. 1981)

modified, 671 F.2d 858 (1982), cert, denied, 103 S.Ct. 181 (1982).
Part I of the argument below discusses the general Fourteenth 

Amendment principles applicable here.
First, it shows that the language and history of the Equal 

Protection Clause establish that the Framers of the Fourteenth 
Amendment intended it to prohibit the administration of criminal 
justice to punish crimes against whites by penalties that 
were not employed to punish similar crimes against blacks.

Second, Part I argues that, because the inequalities in 

the Georgia system involve prejudice against a racial minor­
ity and impinge on the right to life, the state must not only- 
demons trate the rationality of any disparate or unequal state 

action, but also justify it by some compelling state interest.
The State, however, has not attempted to argue that the dispar­

ate racial treatment demonstrated here serves any legitimate 
penological purpose —  nor could it, in view of a constitutional 
history that makes racial discrimination the quintessential 

legislative irrationality.
Third, Part I demonstrates that because the Georgia sta­

tute delegates open-ended sentencing authority to the jury, 

and because considerations of race have been shown to affect 

the determination of who shall die more powerfully than some of 

the statutory aggravating factors, the Georgia system in fact 

permits use of the aggravating factor of race specifically held

66



impermissible in Zant v. Stephens, ___ U.S. ___, 77 L.Ed 235,

255 (1983).
Part II of the argument discusses in more detail the stand­

ards by which the evidence of discrimination here should be eval 
uated, and the controlling caselaw on that question. Throughout 

it refers to a recent discrimination decision petitioner submits 
provides a model for much of its analysis: the
Eleventh Circuit panel opinion in Jean v. Nelson, 711 F.2d 1455

33/
(11th Cir. 1983).

The first section of that discussion deals with the law 
governing the finding of discriminatory intent in a case such 
as this. It points out the special importance of statistical 
evidence of discrimination in a system like this one: full of
decisionmaking discretion, delegated to a multitude of decision­
makers .

The second section reviews the caselaw regarding the 
evaluation of statistical evidence, showing that discrimin­

atory intent has regularly been found from statistical pat­
terns comparable to those here, and noting that Jean and other 

courts have looked to additional evidence similar to that peti­

tioner has attempted to gather and submit, but the court has not

33/ As this brief was being completed, Petitioner's counsel 
learned that, pursuant to an unpublished order, the panel 
decision in Jean is being reheard en banc. Under the Circuit 
rules, it is thus not binding precedent at this time. Peti­
tioner believes, however, that —  because of the similarity of 
the issue and evidence there, and the established caselaw it 
gathers and relies on —  the Jean panel opinion remains useful 
and instructive as a model for the analysis appropriate here.

-67-



irrelevant.

The third section discusses three issues, raised by this 
Court, regarding the evaluation of petitioner's evidence: 
whether the data should be evaluated on a statewide or on a 

local basis; whether the focus should be on certain particular 
stages of the decisionmaking process; and whether each case 
should be considered in comparison with all others, or only with 
those at its particular aggravation level. On all these points 
we argue that these separate narrower evaluations may be relevant, 
if at all, in rebuttal of Petitioner's general prima facie case 
—  and the State has produced nothing to show they can be used 

as such. The final section of Part II of the argument discusses 

the State's general burden of proof once a prima facie case 
of discrimination has been made, and shows that, under Jean, 

the State has failed to meet that burden as a matter of law.

The third Part of the argument addresses the question of 
relief. It points out that the only relief legally possible 
upon a finding of discrimination in this case is a grant of the 

writ, releasing petitioner unless the state resentences him to 
life imprisonment within a specified time. It notes that any 

broader questions —  regarding the relevance of the Court's 

finding to other, differently situated cases, or to the con­
stitutionality of the state system generally —  necessarily must 
be reserved for decision in any later cases as they may arise.

-68-



I. t h e b as ic e qu al p r o t e c t i o n p r i n c i p l e s

There is no question that a statute adjusting the sever­
ity of punishment for a crime according to the race of the 
defendant or the victim would be in direct violation of the 
Equal Protection Clause of the Fourteenth Amendment. Such a 
statute would not "rest upon some ground of difference having 
a fair and substantial relation to the object of legislation, 
so that all persons similarly circumstanced [would] be treated 
alike," F.S. Royster Guano Company v. Virginia, 253 U.S. 412,
415 (1920); see also Reed v. Reed, 404 U.S. 71 (1971), and 
would have "no legitimate overriding purpose independent of 
invidious racial discrimination . . . [justifying the] classi­

fication," Loving v. Virginia, 388 U.S. 1, 11 (1967).

Though neutral on its face, petitioner's evidence shows 
that, in its administration, the Georgia capital statute has 
produced the same unconstitutional result: punishing with
death in significantly greater proportions those defendants, 

especially those black defendants, convicted of killing victims 
who are white.

The distinction between an attack on the facial constitu­
tionality of a statute and a challenge to its administration 
has no bearing on the scope of the Equal Protection guarantee.

The Fourteenth Amendment prohibits not only discriminatory legis­

lation, but the discriminatory administration of a law as well. 
The Supreme Court in Yick Wo v. Hopkins, 118 U.S. 356 (1886),

-69-



confirmed the availability of habeas corpus relief or discrim­

inatory enforcement of a facially valid statute:

. . Though the law itself be fair on its face 
and impartial in appearance, yet if it is applied 
and administered by public authority with an evil 
eye and an unequal hand, so as practically to make 
unjust and illegal discriminations between per­
sons in similar circumstances, material to their 
rights, the denial of equal justice is still 
within the prohibition of the Constitution."

118 U.S at 373-74. See Norris v. Alabama, 294 U.S. 587 (1935);
Turner v. Fouche, 396 U.S. 346 (1970); Allee v. Medrano, 417
U.S. 802 (1974). This case presents a challenge to such state
action within the context of capital sentencing. The Supreme
Court has expressly recognized that such a challenge can be

brought against the administration of a capital statute:

If a statute that authorizes the discretionary imposi­
tion of a particular penalty for a particular crime is 
primarily used against defendants of a certain race, and 
if the pattern of use can be fairly explained only by 
references to the race of the defendant, the Equal Pro­
tection Clause of the Fourteenth Amendment forbids con­
tinued enforcement of that statute in its existing form.
Cf. Yick Wo v. Hopkins, 118 U.S. 356 (1886).

Furman v. Georgia, 408 U.S. 238 (1972) (Burger, Ch. J., dissent­
ing). The Eleventh Circuit also has recognized the cognizability 
of such a challenge in the specific context of the administra­
tion of a state capital sentencing statute. Smith v. Balkcom, 
671 F.2d 858, 859 (5th Cir. 1982 Unit B), cert, denied 103 

S.Ct. 181 (1982) (applying this principle to an equal protec­
tion claim of discrimination in capital sentencing by race-of- 

victim). This remains the law governing this case. Adams v.

-70-



Wainwright, 709 F.2d 1443, 1449 (11th Cir. 1983; Smith v. Kemp,

___ F.2d ___ (11th Cir. No. 83-861, September 9, 1983), slip
op. at 19.

Petitioner asserts that the evidence in this case estab­

lishes such a violation, demonstrating the persistent and 
intentional effects of race in every sense such intent can be 
revealed within the framework of a systemwide administration 
of a law which involves the exercise of discretionary judgment 
by a multitude of decisionmakers. Before examining this evi­

dence, however, we will first discuss the basic nature of the 
Fourteenth Amendment guarantee the evidence implicates and 

petitioner's standing to challenge the racially discriminatory 
decisionmaking evidenced in this case.

-71-



A . The Mature of the Equal protection violations 
The unequal enforcement of criminal statutes based on 

the race of the defendant constitutes a clear violation of 

the Fourteenth Amendment. Loving v.. Vir£inia, suera; YickW° 
t  Hopkins, supra. In a series of analyses of Georgia's capital 

charging and sentencing system, petitioner has shown that 
Georgia law has produced that kind of inequality in its treat­
ment of black offenders. Even stronger, though, and more per­
sistent in its effects, is petitioner's proof of race-conscious 

decisionmaking with a different focus: the race of the
Discrimination based upon the victim's race also violates the 

Equal Protection Clause, under three different lines of Four-

teenth Amendment authority.
The Historical Purpose of the Amendment

The language and legislative history of the Equal Protec­

tion Clause show that its framers specifically intended it to 

prohibit the unequal administration of state criminal laws 

based upon the race of the victim. The Clause clearly says 
"nor shall any state deprive any person within its junsdicion 

of the Equal Protection of the Law." On its face, that lan­

guage speaks more directly to discrimination by the race of 
the victim in the enforcement of criminal sanctions than to

any other form of discrimination.
This original focus upon discrimination directed against

• i.' purified early in the history of judicialblack victims was clantiea eaixy
interpretation of the Amendment:

-72-



The 14th Amendment makes no attempt to enumerate 
the rights it designed to protect. It speaks in 
general terms, and those are as comprehensive as 
possible. Its language is prohibitory; but every 
prohibition implies the existence of rights and 
immunities, preeminent among which is an immunity 
from unequality of legal protection, either for 
life, liberty, or property. Any state action that 
denies this immunity to a colored man is in con­
flict with the Constitution.

Strauder v. West Virginia, 100 U.S. 330, 310 (1886). Similarly,
the history underlying the Fourteenth Amendment, and the history
of its enactment, also speak to this primary purpose with unusual

clarity. That history was summarized by the Court in Strauder:

This is one of a series of constitutional 
provisions having a common purpose, namely: 
securing to a race recently emancipated, a race 
that through many generations had been held 
in slavery, all the civil rights that the superior 
race enjoy. The true spirit and meaning of the 
Amendments, as we said in the Slaughter-House 
Cases, 16 Wall .36, 21 L.Ed. 394, cannot be under­
stood without keeping in view the history of the 
time when they were adopted, and the general 
objects they plainly sought to accomplish. At 
the time when they were incorporated into the 
Constitution, it required little knowledge of 
human nature to anticipate, that those who had 
long been regarded as an inferior and subject 
race would, when suddenly raised to the rank 
of citizens, be looked upon with jealousy and 
positive dislike, and that state laws might be 
enacted or enforced to perpetuate the distinc­
tions that had before existed. Discriminations 
against them had been habitual. It was well 
known that, in some States, laws making such 
discriminations then existed, and others might 
well be expected. . . . [The Black community] 
needed the protection which a wise government 
extends to those who are unable to protect them­
selves. They especially needed protections 
against unfriendly action in the States where 
they were resident. It was in view of these 
considerations the 14th Amendment was framed 
and adopted. It was designed to assure to the 
colored race the enjoyment of all the civil

-73-



rights that under the law are enjoyed by white 
persons, and to give to that race the protection 
of the General Government, in that enjoyment 
whenever it should be denied by the States. It 
not only gave citizenship and the privileges of 
citizenship to persons of color, but it denied to 
any State power to withhold from them the equal 
protection of the laws, and authorized Congress 
to enforce its provision by appropriate legis­
lation. To quote the language used by us in the 
Slaughter-House Cases, "No one can fail to be 
impressed with the one pervading purpose found 
in all the Amendments, lying at the foundation 
of each, and without which none of them would 
have been suggested; we mean the freedom 
of the slave race, the security and firm estab­
lishment of that freedom, and the protection of 
the newly made freeman and citizen from the 
oppressions of those who had formerly exercised 
unlimited dominion over them." So again: "The
existence of laws in the States, where the newly 
emancipated negroes resided, which discriminated 
with gross injustice and hardship against them 
as a class, was the evil to be remedied, and 
by it [the 14th Amendment] such laws were for­
bidden. . . .

If this is the spirit and meaning of the 
Amendment, whether it means more or not, it is 
to be construed liberally, to carry out the 
purposes of its framers. It ordains that no 
State shall make or enforce any laws which shall 
abridge the privileges or immunities of citizens 
of the United States (evidently referring to the 
newly made citizens, who, being citizens of the 
United States, are declared to be also citizens 
of the State in which they reside). It ordains 
that no State shall deprive any persons of life, 
liberty or property, without due process of law, 
or deny to any person within its jurisdiction the 
equal protection of the laws. What is this but 
declaring that the law in the States shall be the 
same for the black as for the white: that all 
persons whether colored or white, shall stand 
equal before the laws of the States and, in regard 
to the colored race, for whose protection the 
Amendment was primarily designed, that no dis­
crimination shall be made against them by law 
because of their color? The words of the Amend­
ment, it is true, are prohibitory, but the con­
tain a necessary implication of a positive immunity, 
or right, most valuable to the colored race —

-74-



the right to exemption from unfriendly legislation 
against them distinctively as colored; exemption 
from legal discrimination, implying inferiority 
in civil society, lessening the security of their 
enjoyment of the rights which others enjoy, and 
discrimination which are steps toward reducing 
them to the condition of a subject race.

Id. at 100 U.S. 306-07 (emphasis added).
The framers of the Fourteenth Amendment knew that differ­

ential punishment based on the race of the victim of a crime 
was one of the "steps toward reducing them to the condition of 

a subject race." In pre-Civil War times, statutes regularly 
punished crimes less severely when the victim of their commis­
sion was a black person or a slave. During the post-war era 
immediately preceding the enactment of the Fourteenth Amendment, 

Southern state authorities frequently declined to administer
their statutes to prosecute persons who committed criminal acts 

34/against blacks.— ' In those cases that were prosecuted,

34/ See, e .g ., Report of the Joint Committee on Reconstruction, 
at the First Session, Thirty-Ninth Congress, Part II, at 25 
(1866) (testimony of George Tucker, commonwealth attorney 
(The southern people "have not any idea of prosecuting white 
men for offences against colored people; they do not 
appreciate the idea."); id. at 209 (testimony of Lt. Col.
Dexter Clapp) ("Of the thousand cases of murder, robbery, 
and maltreatment of freedmen that have come before me, . . .
I have never yet known a single case in which the local 
authorities or police or citizens made any attempt or 
exhibited any inclination to redress any of these wrongs or 
to protect such persons.") id. at 213 (testimony of Lt.
Col. J. Campbell) ("There was a case reported in Pitt County 
of a man named Carson who murdered a negro. There was also 
a case reported to me of a man named Cooley who murdered a 
negro near Golasborough. Neither of these men has been tried 
or arrested.") .

75



authorities acquitted or accorded disproportionately light

sentences to persons who were guilty of crimes against blacks.

The congressional hearings and debates which led to the 

enactment of the Fourteenth Amendment are replete with refer­

ences to this pervasive problem. [The congressional record 
of these proceedings is summarized in Appendix A to this Memo­
randum. ] The Amendment and the statutes enforcing it were 

intended in part to declare that this kind of discrimination 
was invalid.

The Supreme Court has recently reaffirmed this: "[i]t is
clear from the legislative debates that, in the view of the ... 
sponsors [of the Ku Klux Klan Act of 1871], the victims of 

Klan outrages were deprived of 'equal protection of the laws' 

if the perpetrators systematically went unpunished." Briscoe

v. Lahue, ___ U.S. ___, 75 L.Ed.2d 96, 109 (1983). The
evidence in this case plainly demonstrates a violation of these 

Equal Protection Clause objectives.

35/

35/ See, e .g ., id., Part III, at 141 (testimony of Brevet 
M.J. Gen. Wagner Swayne) ("I have not known, after six months' 
residence at the capital of the State, a single instance of a 
white man being convicted and hung or sent to the penitentiary 
for crime against a negro, while many cases of crime warranting 
such punishment have been reported to me."); Id., Part IV, at 
75-76 (testimony of Maj. Gen. George Custer) ("I believe a white 
man has never been hung for murder in Texas, although it is the 
law. Cases have occurred of white men meeting freedmen they 
never saw before, and murdering them merely from this feeling 
of hostility to them as a class.").

-76-



2 . Traditional Equal Protection Principles

Even without reference to the history summarized above, 
race-of-victim sentencing disparities violate basic equal 
protection principles long recognized and applied to all areas 
of state action. It is fundamental equal protection law that 
any disparate treatment at the hands of the state renders the 
operation of a law unconstitutional under the Fourteenth 
Amendment, absent a rational explanation for subjecting one 
individual to harsher treatment than another. See Dep't. of 
Agriculture v. Moreno, 413 U.S. 528 (1973); F.S. Royster Guano 
Co. v. Virginia, supra.

Moreover, under well-established equal protection doctrine,

even a "rational" explanation for the pattern would not suffice
to protect the state action proven here: this claim involves
racial discrimination which impinges upon the right to life, a
right explicitly guaranteed by the Fourteenth Amendment and

36/
inherent in the constitutional framework.

36/ See, e.g., Yick Wo v. Hopkins, 118 U.S. 356, 370 (1886) 
("the fundamental rights to life, liberty and the pursuit of 
happiness"); Johnson v. Zerbst, 304 U.S. 458, 462 (1938) 
("fundamental human rights of life and liberty"); Screws v. 
United States, 325 U.S. 91, 131-32 (1945) (Rutledge, J., con­
curring); _id. at 134-35 (Murphy, J. , dissenting) ("He has 
been deprived of the right to life itself . . . that right 
was his because he was an American citizen, because he was 
a human being. As such, he was entitled to all the respect 
and fair treatment that befits the dignity of man, a dignity 
that is recognized and guaranteed by the Constitution."); May 
v. Anderson, 345 U.S. 528, 533 (1953) (a right "far more pre­
cious than . . . property rights").

77



The Supreme Court has made clear that where "fundamental 

rights" or "suspect classifications" are involved, discrimin­
atory state action "may be justified only by a 'compelling 
state interest' . . . and . . . legislative enactments must 
be narrowly drawn to express only the legitimate state inter­

ests at stake." Roe v. Wade, 410 U.S. 113, 155 (1973). See 
also Cleveland Board of Education v. LaFleur, 414 U.S. 632 

(1974); Stanley v. Illinois, 405 U.S. 645 (1972).
The "fundamental rights" branch of equal protection law 

orginated in Skinner v. Oklahoma, 316 U.S. 536 (1942), in which 

the Oklahoma Legislature had imposed a punishment of steriliza­

tion upon those convicted of certain crimes no more heinous than 
other crimes for which a term of years was imposed. The Supreme 
Court held that this unjustifiable discrimination in the 
choice of defendants sentenced to be deprived of their "fundamental 

right" to procreate met no compelling state need and thus 

violated the Fourteenth Amendment.
Certainly a principle that protects the right to procreate, 

absent a compelling state interest, is at work when the stakes 
are life and death, and the impact of that state action destroys 

not just one right, but all rights. "[B]ecause there is a 
qualitative difference between death and any other permissible 
form of punishment, ’there is a corresponding difference in the 

need for reliability in the determination that death is the 

appropriate punishment in a specific case.'" Zant v. Stephens,

78



supra, 51 U.S.L.W. at 4897 (quoting Woodson v. North Carolina,
37/

428 U.S. at 305). That statement, secured in the Eighth
Amendment context, applies as powerfully to this Fourteenth 

Amendment claim: when life is at stake, the Court should
review the operation of a statutory scheme with the more 

exacting level of scrutiny.
Skinner also establishes that more exacting judicial review 

must be given to severe criminal sanctions having their harshest 
impact upon relatively powerless minorities. Id. at 536-37, 

541-42.
There are limits to the extent to which a legislatively 
represented majority may conduct biological experiments 
at the expense of the dignity and personality and natural 
powers of a minority —  even those who are guilty of what 
the majority define as crimes."

Id. at 546 (concurring opinion of Justice Jackson). Because
petitioner has shown that the Georgia capital sentencing sta-

37/ See, e.g., Reid v. Covert, 354 U.S. 1, 77 (1957) (capital 
cases "stand on quite a different footing than other offenses. 
In such such cases the law is especially sensitive to demands 
for . . . procedural fairness . . ."). Williams v. Georgia,
349 U.S. 375, 391 (1955) ("that life is at stake is of course 
another important factor in creating the extraordinary situa­
tion. The difference between capital and non-capital offenses 
is the basis of differentiation in law in diverse ways in which 
the distribution becomes relevant"). See also Andres v. U.S., 
333 U.S. 740, 752 (1958); Griffin v. Illinois, 351 U.S. 12,
28 (1956); McGautha v. California, 402 U.S. 183, 311 (1971) 
(Brennan, J., dissenting).

79



tute discriminates along far more suspect lines of race, more­

over, it is even clearer that the state should be held to the 

highest standard of explanation. Racial classifications are 
"subjected to the strictest scrutiny and are justifiable only 

by the weightiest of considerations." Washington v Davis, 426 

U.S. at 242 (1976), citing McLaughlin v. Florida, 379 U.S. 184 

(1964).
No state action is more suspect than racial discrimina­

tion in the administration of justice. These inequalities "not 

only violate our Constitution and the laws enacted under it, 
but [are] at war with our basic concepts of a democratic soci­
ety and a representative government." Smith v. Texas, 311 U.S. 
128 (1940) (footnote omitted). See also Ballard v. United 
States, 329 U.S. 187, 195 (1946) ("The injury is not limited 

to the defendant —  there is injury . . .  to the law as an 

institution, to the community at large, and to the democratic 
ideal reflected in the processes of our courts.") For that 
reason, "discrimination on the basis of race, odious in all 

aspects, is especially pernicious in the administration of 

justice." Rose v. Mitchell, 443 U.S. 545, 555 (1979). Such 

discrimination destroys "the appearance of justice" and

80



casts doubt on "the integrity of the judicial process.”

Id. at 535. Again, this concern is heightened when the penalty 
is death:

From the point of view of society, the action of the 
sovereign in taking the life of one of its citizens 
. . . differs dramatically from any other legitimate 
state action. It is of vital importance to the 
defendant and to the community that any decision to 
impose the death sentence be, and appear to be, 
based on reason rather than caprice or emotion.

Gardner v. Florida, 430 U.S. 349, 357-58 (1977).
Under any level of Fourteenth Amendment scrutiny, the 

discrimination shown here cannot be justified. No rational 
justification to sanction racial discrimination in capital 
sentencing has been suggested by the State, for none is pos­
sible: race-based decisionmaking is the quintessential irra­

tionality which offends the mandate of equal treatment embodied 
in the Fourteenth Amendment. McLaughlin v. Florida, supra. 

Certainly, the theoretical goals of retribution and deter­
rence which justify the death penalty, see Gregg v. Georgia,

428 U.S. at 183-84 (plurality opinion), cannot be served by 

any discriminations based upon race. To the contrary, such 
disparities can hardly help maintain "respect for law." Ibid. 

Any legitimate goal served by this sanction can only be under­
mined by its selective reservation of the death penalty for an 

invidiously selected group of offenders.

81



Race-conscious death sentencing thus violates the Equal 

Protection Clause for reasons wholly separate from, and in addi­

tion to, the fact that such state action was a specific target 
of the Fourteenth Amendment. It is an irrational exercise of 
governmental power, in its most extreme form, unrelated to any 

legitimate state purpose.
3. Race As An Aggravating Circumstance 

In the context of Georgia law, a showing of race-of-victim 
discrimination implicates a third Fourteenth Amendment principle 
as well: the prohibition of explicitly race-conscious legisla­
tion. See, e.g. , Loving v. Virginia, supra; Strauder v. West 
Virginia, supra. The Supreme Court in Zant v. Stephens, supra, 
held that it would be unconstitutional, in an otherwise valid 
sentencing system, to:

attac[h] the aggravating label to factors that 
are constitutionally impermissible or totally irrel­
evant to the sentencing process, such as for example 
the race, religion, or political affiliation of the 
defendant. . . .  If the aggravating circumstance at 
issue in this case had been invalid for reasons such 
as these, due process of law would require that the 
jury's decision to impose death be set aside.

77 L.Ed.2d at 255. Yet, in a real sense, that is precisely what
the State of Georgia has authorized and what the evidence here
shows Georgia juries and prosecutors have done: "attached the

aggravating label" to race.

The Georgia Supreme Court, in its response to the question 

certified to it in Zant, made clear that consideration of race 
is not expressly proscribed by Georgia law.

81A



In making a decision as to the penalty, the factfinder 
takes into consideration all circumstances before it 
from both the guilt-innocence and the sentence phases 
of the trial. These circumstances relate both to the 
offense and the defendant.

A case may not pass the second plane into the area in 
which the death penalty is authorized unless at least 
one statutory aggravating circumstance is found. How­
ever, this plane is passed regardless of the number of 
statutory aggravating circumstances found, so long as 
there is at least one. Once beyond this plane, the case 
enters the area of the factfinder's discretion, in which 
all the fact and circumstances of the case determine, 
in terms of our metaphor, whether or not the case passes 
into the third plane and in the area in which the death 
penalty is imposed.

Zant v. Stephens, supra, 77 L.Ed.2d at 247, quoting 250 Ga.
38/

97, 99-100, 297 S.E.2d 1, 3-4 (1982) (emphasis added).

The jury in this case was so instructed. T. 1028. Absent 
a limiting instruction, "all the circumstances" includes—  or 

at least, it certainly does not exclude —  the race of both 
the defendant and the victim. As the trial judge noted on 
the questionnaire submitted to the Supreme Court in this case,

38/ The only place in Georgia law which embodies any pro- 
phylatic measure against the influence of racial prejudice 
on sentencing decisions is the Georgia Supreme Court's sen­
tencing review power. As the Georgia Court noted in the 
remand in Zant, Georgia Code Annotated § 27-2537 authorizes 
it to determine "whether the penalty of death was imposed 
under the influence of passion, prejudice, or any other 
arbitrary factor. . . . "  250 Ga. at 99, 297 S.E.2d at 3.
Yet, as this Court can judicially notice, the Georgia 
Supreme Court has never reversed a single death sentence 
on the grounds that it was imposed under the influence of 
passion, prejudice or any arbitrary factor, or on the basis of 
the race of the defendant or the victim, and petitioner has 
proven that the racial effects identified at the trial level in 
Georgia's charging and sentencing system are not eliminated by 
appellate sentence review.

82



the jury was never instructed to disregard this single aspect 

of the case. See Trial Judge's Questionnaire at 2.
Georgia prosecutors similarly have, under State law, unfet­

tered discretion in the bringing, charging and pressing of capital 

charges. See Gregg v. Georgia, supra, 428 U.S. at 199. In 
such a system, they can be expected to exercise that discre­
tion based largely upon the likelihood a jury will impose a 

sentence of death. See id. at 225 (concurring opinion) (see 

also Dep., 31, 38-39). Where the statute specifically allows 
race to be considerd by the jury, it can hardly be expected 
that it will be ignored by prosecutors.

A showing, such as that here —  that racial factors have 

been determinative of outcome —  at a level equal to or in 
excess of the influence of some statutory aggravating factors 
(see DB 81), translates into a showing that this legally- 
delegated authority to discriminate has been exercised by these 
decisionmakers in fact. This means that race has been used as 
an aggravating factor, in direct violation of the proscription 

set forth in Zant v. Stephens, supra.

83



B. The Issue of Standing

There can be no question that Warren McCleskey has standing 

to raise the challenge he does here, on all these Fourteenth 

Amendment grounds. He is a member of the class of persons 
most seriously disadvantaged by the discrimination he alleges: 
he is black, like the defendants in Georgia most harshly pun­
ished and the victims whose cases the State treats least ser­
iously. His alleged victim was white, making his case, he 
claims, much more likely to be singled out for this uniquely 
harsh treatment. The prosecutor and the jurors who imposed 
it on him —  only one of whom was a member of his race —  acted 

under the same legal standards that have applied, and produced 

these discriminatory results, throughout the State of Georgia 
generally. He has been "aggrieved" by this system and this 

practice in the ultimate sense as as result of it: he stands
to lose his life.

Under the law of this Circuit, this last fact alone is

sufficient to establish standing here. Spinkellink v. Wainwright,
supra, a case in which the defendant was white, made this clear:

[the petitioner] has standing to raise the equal pro­
tection issue, even though he is not a member of the 
class allegedly discriminated against, because such 
discrimination, if proven, impinges on his constitu­
tional right under the Eighth and Fourteenth Amend­
ments not to be subject to cruel and unusual punishment.
See Taylor v. Louisiana, 419 U.S. at 526.

Spinkellink v. Wainwright, supra, at 612 n.36.

84



In Taylor v. Louisiana, 419 U.S. 522 (1975), the Supreme
Court held that a male defendant had standing to object to the 

exclusion of women from his jury even though he was not a member 
of the excluded class. Similarly, in Rose v. Mitchell, 443 
U.S. 545 (1979), a defendant was permitted to raise the claim 

that his right to equal protection had been violated when he 

was indicted by a grand jury from which blacks had been systemat­
ically excluded. The Court explicitly conferred standing on 
him, asserting that discrimination in the administration of 
criminal justice "strikes at the fundamental values of our 
judicial system and our society as a whole." 443 U.S. at 556.
The unequal administration of Georgia's death penalty statute 
based on race of the victim, no less "cast[s] doubt on the 

integrity of the entire judicial process." Peters v. Kiff,
407 U.S. 493, 502 (1972). Having been condemned to die by 

that process, petitioner clearly has the right to question its 
fairness.

85



II. THE STANDARDS FOR EVALUATION OF PETITIONER'S 
EQUAL PROTECTION CLAIM_______________________

A. The Issue of Discriminatory Intent
Under the law of this Circuit, an equal protection chal­

lenge to the racially discriminatory administration of a cap­

ital sentencing statute may be supported by statistical evidence 
of disproportionate impact which gives rise to the inference of 
discriminatory intent on the part of decisionmakers. See Smith 

v. Balkcom, supra, 671 F.2d at 868. The Court in Smith set 
forth that standard as controlling, in an opinion which clearly 

omitted the former requirement that a petitioner need show 
specific acts of intentional discrimination against him.

This is consistent with equal protection law in other 
areas.

[D]iscriminatory intent need not be proven 
by direct evidence. "Necessarily an invidious 
discriminatory purpose may often be inferred from 
the totality of the relevant facts, including the 
fact, if it is true, that the law bears more 
heavily on one race than another." [Citing 
Washington v. Davis, 426 U.S. 229 (1976)]. Thus, 
determining the existence of discriminatory pur­
pose "demands a sensitive inquiry into such cir­
cumstantial and direct evidence of intent as may 
be available." [Citing Arlington Heights v . 
Metropolitan Housing Authority, 429 U.S. 252
(1977).]

Rogers v. Lodge, ___ U.S. ___, 50 U.S.L.W. 5041, 5042 (1982).
Cases in which no overt use is made of race as a criter­

ion, but in which the record leaves race as the only plausible 

explanatory factor for the demonstrated inequities, nearly

86



always turn on circumstantial proof. This has been recognized 

since Yick Wo v. Hopkins, where the Court found statistical 
proof that a statutory scheme "in actual operation . . . war­
rant [ed] and require[d] the conclusion that whatever may have 
been the intent of the ordinances as adopted, they are applied 

by the public authorities charged with their administration, 
and thus representing the State itself, with a mind so unequal 

and oppressive as to amount to a practical denial by the State 
of equal protection of the laws." 118 U.S. at 373-74. That 

principle remains the law. For example, the Supreme Court has 
applied this analysis in the jury cases, another area in which 
a showing of discriminatory impact is the only practical —  
indeed the only possible —  evidentiary means of demonstrating 
an equal protection violation:

It is also clear from the cases dealing with racial 
discrimination in the selection of juries that the 
systematic exclusion of Negroes is itself such an 
"unequal application of the law . . .  as to show 
intentional discrimination" [citation omitted]. . . .
It is also not infrequently true that the discrimin­
atory impact —  in the jury cases for example, the 
total or seriously disproportionate exclusion of 
Negroes from jury venires —  may for all practical 
purposes demonstrate unconstitutionality because in 
various circumstances the discrimination is very 
difficult to explain on nonracial grounds.

Washington v. Davis, 426 U.S. 229, 241-42 (1976).

Inquiry into the results arising from a statute's operation

is critical in cases of governmental action that involve complex

processes in which numerous influences are at work. Thus

87



[fjrequently the most probative evidence of intent 
will be objective evidence of what actually happened 
rather than evidence describing the subjective state 
of the mind of the actor. For normally the actor is 
presumed to have intended the consequences of his 
deeds. This is particularly true in the case of gov- #  
ernmental action which is frequently the product of 
compromise, of collective decisionmaking, and of 
mixed emotion.

Washington v. Davis, supra, 426 U.S. at 253 (Stevens, J. con­
curring). As the Court of Appeals for the Fifth Circuit asserted 

in assessing an equal protection challenge to school board 
procedures analogous to petitioner's challenge here: "The
most effective way to determine whether a body intended to 
discriminate is to look at what it has done." United States 
v. Texas Ed. Agency, 579 F.2d 910 (1978).

The role of circumstantial evidence is particularly crucial 

in cases involving a multitude of decisionmakers. The Court 

in Jean v. Nelson pointed out:
The very nature of legislative and administra­
tive action makes it difficult to ascertain the 
"intent" of the acting body. For that reason, 
in Arlington Heights the Supreme Court provided 
some examples of "circumstantial and direct 
evidence" that courts might properly consider 
in judging whether invidious discrimination 
permeated offical action.

Jean v. Nelson, supra, 711 F.2d at 1485-6 (footnote omitted).

Thus, in such cases the evidence "need not identify an 
intentional discriminatory act or malevolent actor in the defend­
ants's particular case." See United States v. Texas Educ. Agency, 

579 F .2d 910, 913-14, nn.5-7 (5th Cir. 1978)." Jurek v. Estelle,

88



593 F.2d 672, 685 n.26 (5th Cir. 1979), vacated and reaffirmed

on other grounds, 623 F.2d 929 (5th Cir. 1980), (en banc), cert.
39/

denied, 450 U.S. 1001 (1981).

The presence of various decisionmakers in the jury cases 

and the school board cases triggered judicial reliance upon 

circumstancial evidence as the most appropriate proof of 
discriminatory intent. Castaneda v. Partida, supra (jury 
selection); United States v. Texas Ed. Agency, supra (school 

desegregation); Penick v. Columbus Board of Education, 583 F.2d 
787 (6th Cir. 1978), aff'd 443 U.S. 449 (1979) (school 
desegregation).

The courts have been most willing to find discrimination 
from statistics alone in systems involving numerous decision­
makers who are permitted relatively uncontrolled discretion 

in the administration of the law.

39/ Though it is clear, after Smith v. Balkcom and in light of 
this other Fourteenth Amendment authority, that it is unneces­
sary to show a particular racially-motivated act in the individ­
ual case to establish a claim of sentencing discrimination, it 
is noteworthy there is evidence of a significant racial incident 
in this case. Petitioner has shown that the racial composition 
of his jury panel —  11 whites and one black —  differed marked­
ly from the makeup of the Fulton County adult population as a 
whole, which is roughly 50% black. Though the cause of his 
result is not evident from the record here, Dr. Berk's calcula­
tions establish it is highly unlikely that final result oc­
curred by chance —  sufficiently unlikely to support an infer­
ence of discrimination in related jury contexts. Castaneda v. 
Partida, 430 U.S. 475, 512 n.17 (1980); Alexander v. Louisiana, 
405 U.S. 625, 630 (1972). Though petitioner has not raised 
a separate jury challenge, the Court can and should still con­
sider this fact insofar as it finds events in the individual 
case relevant to the overall discrimination issue here.

89



Cases are legion [within our constitutional tradi­
tion] that discern the dangers of unguided discre­
tion, preeminent among them the risk of selective 
and discriminatory enforcement. E.g., Papachristou 
v. City of Jacksonville, 408 U.S. 156, 170-71 (1972);
Yick Wo v. Hopkins, 118 U.S. 356, 369-70, 373 81885); 
Environmental Defense Fund, Inc, v. Ruckelshaus, 439 
F.2d 584, 598 (D.C. Cir. 1971) Where, as here, 
there are no standards governing the existence of 
discretion . . . the scheme permits and encourages 
an arbitrary and discriminatory enforcement of the 
law. Papachristou, 405 U.S. at 170-171.

Jean v. Nelson, supra, 711 F.2d at 1502 (emphasis added).
In Yick Wo, the court similarly emphasized that the ordinance
struck down "confer[s], not a discretion to be exercised upon
a consideration of the circumstances of each case, but a naked
and arbitrary power to give or withhold consent . . .  as to
persons. . . . The power given [to the decisionmakers] is not

confided to their discretion in the legal sense of that term,

but is granted to their mere will. It is purely arbitrary,

and acknowledges neither guidance nor restraint." Yick Wo v .

Hopkins, supra, 118 U.S. at 366-67.

In the jury cases, too, the reliance upon statistical
evidence for a showing of "discriminatory intent" turns on
the discretionary nature of the selection procedures. Thus,

equal protection violations based on statistical showings,
which fall short of the extreme pattern demonstrated in Yick
Wo, were condemned in the jury cases precisely "because of
the nature of the jury-selection task." Arlington Heights v .
Metropolitan Housing Authority, 429 U.S. at 266 n.13 (1977).
That task rests on a subjective process which at every junc­

90



ture presents "the opportunity to discriminate" such that " . . .  

whether or not it was the conscious decision on the part of 

any individual jury commissioner," the Courts have been con­
fident, when presented with a showing of disparate impact, 
in concluding that "[t]he result bespeaks discrimination." 

Alexander v. Louisiana, 405 U.S. 625, 632 (1972); see also 
Hernandez v. Texas, 347 U.S. 475, 482 (1954); Norris v. Alabama, 
294 U.S. 587, 591 (1935). "[A] selection procedure that is 
susceptible of abuse or is not racially neutral supports the 
presumption of discrimination raised by the statistical showing," 

Castaneda v. Partida, supra, 430 U.S. 482, 494 (1977), citing 

Washington v. Davis, supra, 426 U.S. at 241.
Thus, because the sentencing system here involves a multi-

40/
tude of decisionmakers, each with substantial discretion 
and each involved in a governmental process which has the sever­
est impact on individual life and liberty, the analyses of 
Jean and the jury cases are directly on point in this case.
The required prima facie showing of discriminatory intent thus

40/ Petitioner recognizes that Zant v. Stephens, supra, held 
that the discretion allowed by the Georgia death penalty 
statute did not render it unconstitutional on its face, under 
Furman v. Georgia, supra. But that clearly does not mean that 
discretionary aspect is irrelvant to the discrimination claim 
here: the Court in Castaneda v. Partida, supra, similarly
noted that "[t]he facial constitutionality of the key-man sys­
tem, of course, has been accepted by this Court. . . . Never­
theless, the Court . . . noted that the system is susceptible 
to abuse as applied," 430 U.S. at 497 —  and held its discre­
tionary nature supported a finding of discrimination, id. at 
494. See also Yick Wo v. Hopkins, supra; Jean v. Nelson, supra.

91



can, and must be, made out by a demonstration of significant

racial disparities resulting from the 
This is precisely what petitioner has

discretionary process, 
demonstrated.

92



B. The Legal Significance of the Statistical Evidence

In Yick Wo v. Hopkins, supra, the Court did not address
the question of the magnitude of statistical disparity needed

to establish the presence of race discrimination, in part
because the statistical disparities there were virtually
total. 118 U.S. at 374. In this case the facts are more

complex: while death sentences have been imposed in capital
cases of all racial compositions, death sentences have been

imposed significantly more frequently where the defendant is
black and the victim white. Though the evidence does not show
every capital defendant who kills a white victim is sentenced
to death, nor that every one whose victim is black is not,
the pattern no less bespeaks discrimination.

Plaintiffs need not prove a discriminatory purpose was 
the primary, or dominant purpose, Arlington Heights, 429 
U.S. at 266, 97 S.Ct. at 564, 50 L.Ed.2d at 465, but must 
show that the action taken was, at least in part "because 
of," and not merely "in spite of" its adverse effects 
upon an identifiable group. Feeney, supra, 442 U.S. at 
279 & N.24, 99 S.Ct. at 2296 & n.24.

Jean v. Nelson, supra, 711 F.2d at 1495.

Courts have frequently measured and found discrimination
from statistical patterns of differential treatment like this

in other contexts, and in so doing have developed standards

for their consideration. Those standards have evolved in the

line of Supreme Court cases reviewing a Fourteenth Amendment

issue closely analogous to the issue here: challenges by
criminal defendants to the racial makeup of their juries. In

93



those cases, a prima facie case of discrimination can be

proven by statistical disparities no stronger than those shown
here, and under statistical analyses far less sophisticated

41/
and controlled.

From this line of cases has emerged a general rule regard­
ing the strength of a statistical showing required to support 

an inference of discriminatory intent under the Equal Protec­
tion Clause. That rule was comprehensively reviewed 
and applied in Jean v. Nelson, supra. In Jean, as here, the 
plaintiffs' analyses compared the discretionary treatment 
afforded similarly situated individuals from different racial 
groups. 711 F.2d at 1488. The disparities between the average 
outcome of cases involving individuals in these different 
groups were measured in terms of statistical significance 
expressed by the use of standard deviations:

The standard deviation, a final expression of binomial 
analysis, is nothing more than a measure of the differ­
ence between the predicted number and the actual number.
"As a general rule for [] large samples, if the differ­
ence between the expected value and the observed number 
is greater than two or three standard deviations," then 
we suspect an unexplained factor, in this case discrimi­
nation, is responsible for the difference. See Castaneda 
v. Partida, 430 U.S. at 496-97 n. 17, 97 S.Ct. at 1281 n.17, 
51 L.Ed.2d at 512 n.17.

Ibid.

41/ See, e.g., Whitus v. Georgia, 385 U.S. 545 (1967) (7.8% 
and 9.1% black jurors in county 27.1% black); Turner v. 
Fouche, 396 U.S. 346 (1970) (37% black jurors in county 
60% black); Alexander v. Louisiana, 405 U.S. 625 (1972)
(5%a black jurors in county 21% black).

94



Since the time of the Supreme Court's decision in Castaneda,
this rule of statistical significance has been widely applied.
See BALDUS & COLE, STATISTICAL PROOF OF DISCRIMINATION, 295

42/
n.13 (1980) and cases there cited. That baseline test has

been recognized not only in equal protection cases, but also 
in class action Title VII challenges, to undercut the hypothesis 

that decisions were being made randomly with respect to race." 
Hazelwood School District v. United States, 433 U.S. 299, 311 

n.17 (1977).

42/ The evidence in this case has expressed the same kind of 
information in terms of "P" values, rather than standard devia­
tions. The P-value measure is a more rigorous one, especially 
applied to samples of small size. Both measure the statistical 
significance of the discrepancies found, and generally, the 
two types of expressions can be interchanged:

The "standard error" or "standard deviation" for a dis­
tribution is a number that describes the degree to which 
disparities spread out above and below the mean of the 
distribution. When the sampling distribution in a case 
has a bell-shaped curve ... about 68 percent of the dis­
parities will be not more than 1 standard error on either 
side of the central point. ... Similarly, about 95 or 
96 percent of the cases would involve disparities smaller 
than 2 standard errors ... and somewhat more than 99 
percent of the cases would have disparities less than 3 
standard errors .... Thus, in a case where a disparity 
is larger than 2 standard errors, fewer than 5 percent 
of the disparities in a random system wuld be that 
large, and if the disparity exceeded 3 standard errors 
we would know that fewer than 1 percent of random 
selections would result in larger disparities. The 
underlying idea is that when the case involves a large 
sample, a disparity in treatment two times its standard 
error is usually significant at the 0.05 level. When 
the disparity is three times its standard error, it is 
usually significant at the 0.01 level. Thus, for large 
samples, the "2 or 3 standard error" rule is essentially 
equivalent to a rule requiring significance at a level in 
the range below 0.05 or 0.01. Id̂ . at 295-96 (footnote 
omitted); see also Equal Employment Opportunity Commis-

95



The court in Jean made it clear that this kind of binomial
analysis applies in discriminatory enforcement proceedings 

closely analogous to the challenged processes in this case.
This is true even though discriminatory enforcement proceedings 
theoretically do not involve "random process[es]" as do the 
selection procedures at issue in Castaneda. 711 F.2d at 1494- 
1499. The court

... emphasize[d] that plaintiffs may prove a prima 
facie case without accountng for qualifications when 
the statistical impact is sufficiently glaring. This 
we established in Fisher v. Procter & Gamble 
Manufacturing Co., 613 F.2d 527 (5th Cir. 1980), 
cert, denied, 449 U.S. 1115, 101 S.Ct. 929, 66 L.Ed.2d 
845 (1981):

Mindful of [the admonition in] Hazelwood [con­
cerning proof of special qualification], we never­
theless find the plaintiff's statistics adequate 
to establish a prima facie case. First, a prima 
facie case may be shown without evidence of 
qualifications where the inference of discrimina­
tion is supported by a compelling level of racial 
underrepresentation in a sizeable work force.
In Hayes International Corp., [456 F.2d 112 (5th 
Cir. 1972)] we noted that when substantial under­
representation is shown as compared with general 
population figures, the burden of proving lack of 
qualifications is on the Company.

Id. at 544 (citation omitted). And, there is good 
reason for this. Requiring too much proof initially 
from a plaintiff may defeat a valid claim of discrim-

continued
sion v. American National Bank, 652 F.2d 1176, 1192 (4th 
Cir. 1982), cert. denied 103 S.Ct. 235 (1983), citing 
HAYES AND WINKLER, STATISTICS: PROBABILITY, INFERENCE
AND DECISION, 218-219, 381-82 (1971).
Professor Woodworth's diagnostic tests on the CSS studies 

(See GW 4, Table 1) revealed standard deviations, measured 
according to Cressie's "safe" or conservative method, consis­
tently at the level of two to three standard deviations.

96



ination before its validity is discerned. Allowing 
proof of the prima facie case with glaring statistics, 
however, does not unduly burden a defendant. The 
defendant, in rebuttal, can prove the unaccounted-for 
qualifications, thereby undermining plaintiff's case.
Cf. Vuyanich v. Republic National Bank of Dallas, 505 
F.Supp. 224, 356 n.169 (N.D.Tex. 1980).

Id. at 711 F .2d 1495-96.
In that case, petitioner's statistical evidence does 

control for a host of "qualifications" or alternatively, legitimate 
possible explanations for the racial disparities, including 

statutory aggravating factors and potentially explanatory non- 

statutory aggravating and mitigating factors. Even petitioner's 
preliminary analyses limited to unadjusted race-of-defneant 
and race-of-victim figures demonstrate gross disparities of a 

statistical significance that extends far beyond the baseline 
rule of Castaneda and Jean. Though petitioner certainly does 

not rest his case on these unadjusted comparisons, it is instruc­
tive to note that these figures are analogous to those the court 
was examining in Castaneda and the other jury cases, where 
racial discrimination was found on far less substantial showings 

of racial disparity than the unadjusted figures here.
More importantly, petitioner's evidence reveals the 

presence of pervasive race discrimination, at levels consistent 

with those required by Castaneda and the cases following it, 
even when other potentially explanatory factors are controlled 

for. In this sense petition has, like the plaintiffs in Jean, 

"anticipated and adequately met the government's statistical 

challenge."

97



Jean v. Nelson, supra, 711 F.2d at 798. His evidence comprises 

analyses which control for every "racially neutral variable" 
that could thwart an inference of discriminatory intent, in the 

face of a statistical showing, under the standards set forth in 
Smith v. Balkcom, supra, 671 F.2d 858-89 n.83; and McCorquodale 
v. Balkcom, 705 F.2d 1553, 1556 (11th Cir. 1983).

Throughout Professor Baldus' tireless analyses, the dis­
crepancies along racial lines persist. Race of victim dispari­

ties remain at the level of statistical significance, many at 
or beyond the .01 level, roughly equivalent to three standard 
deviations. The fact that some do not —  usually due to the 
small size of the sample of cases involved —  in no way under­
mines the inference of discrimination supported by the many 
that do. The Fourth Circuit has recently made this clear, in 
reversing a district court finding that statistical disparities 
falling below significant levels rebutted other data which met 
them. The Court stated:

If a legal rule of analysis can properly be 
derived from the Castaneda footnote, it can 
only be that standard deviations greater 
that two or three necessarily exclude 
chance as a cause of underrepresentation.
The converse of this —  that standard devia­
tions of not "more than two or three" nec­
essarily exclude discriminatory design as 
the cause —  is nowhere to be found.

The Castaneda Court had no need to ex­
pore the levels of probability that exist in 
the range of "two or three" standard devia­
tions, being content to note that beyond this 
range social scientists would find "suspect" 
for scientific purposes the hypothesis of 
random choice. When this range is explored,

98



however, it appears that well short of three 
standard deviations the probability levels for 
chance as explanation have already dropped far 
below the point at which courts of law —  
concerned with proof by the "greater weight" 
or "preponderance" of the evidence —  would 
presumably have discarded the hypothesis of 
chance. Just short of two standard deviations
—  specifically at 1.96 —  the probability of 
chance is only 5 in 100; at just over two and 
one half, it is only 1 in 100; by three it is 
less than 1 in 100. W. Hays & R. Winkler, 
Statistics: Probability, Inference and Decision 
218-219, 381-382 (1971). For this reason, 
authority can be found for the proposition
that most social scientists, applying laboratory 
rigor to rule out chance as even a theoretical 
possibility rather than the law's rough gauge 
of the "preponderance of the evidence," are 
prepared to discard chance as an hypothesis 
when its probability level is no more than 5%, 
i.e. at approximately two standard deviations. 
Id. at 394.
From all this we conclude that courts of law 
should be extremely cautious in drawing any 
conclusions from standard deviations in 
the range of one to three. Above this range, 
with standard deviations of more than three, 
the analysis may perhaps safely be used 
absolutely to exclude chance as a hypothesis, 
hence absolutely to confirm the legitimacy of 
an inference of discrimination based upon 
judicial appraisals that disparities are, to 
the legallly trained eye, "gross." This we 
conclude is all that the Supreme Court has 
ever directly approved by its own use of the 
process. Within the range of one to three 
standard deviations, where the probability of 
chance as explanation for revealed underrepre­
sentation declines precipitately from only 5% 
at two standard deviations to less than 1% at 
three, we do not see how a court can properly 
find the only other hypothesis —  discrimination
—  dispelled by ... [the State's contrary] 
analysis alone. On this basis, we conclude 
that the district court's analysis revealing 
standard deviations which, as applied to the 
aggregate of years and work categories involved, 
lay preponderately within or above this range, 
could not properly be taken as dispelling the

99



inference of discrimination based upon the 
disparities in the static work force data. To 
the extent this was the basis of the district 
court's finding, we conclude that it was 
clearly erroneous.

Equal Employment Opportunity Com'n v. Am. Nat. Bank, 652 F .2d 

1176, 1192-3 (4th Cir. 1981), cert, denied 103 S.Ct. 235 
(1983). Petitioner's controlled analyses reveal persistent racial 

disparities which are of statistical significance sufficient, 
under this authority, to alone support a finding of discrimi­

natory violations of the equal protection guarantee.
But the strength of the evidence does not stop there.

The statistical proof offered here —  unlike that in Yick Wo, 
Castaneda and the other jury cases —  is not limited to a 
single form of analysis. Petitioner's analyses draw on a 
number of different statistical techniques: a variety of
cross-tabulations among similarly situated cases, least 

squares regression analyses, weighed and unweighted logistic 
regressions, factor analyses, stepwise regressions, "worst 

case" regressions, and cohort studies in Fulton County of 
cases similar to petitioner's own. Virtually all of these 
analyses produce similar results. None indicate, even remotely, 
that the consideration of the race of victim does not operate 

throughout the Georgia death sentencing system.
In Jean, the court acknowledged the significance of such 

consistency in a statistical showing utilizing different 
techniques. Though the Jean opinion does not discuss the

100



full scope of the research methodology used there, it is 

apparent that those analyses were not as varied, nor the 

factors for which they controlled as comprehensive as those 

presented in this case. See 711 F.2d at 1489. Yet the Circuit 
Panel plainly was impressed by the consistent —  though not 
uniform —  result of those tests, in its finding of a prima 
facie case of discrimination. 711 F.2d at 1488-89.

The statistical significance of disparities found in this 
case, in all these tests, is clearly sufficient to dispel any 
notion that they are the result of chance. They statistically 
"permit no other inference but that [the results] are the 
product of a racially discriminatory intent or purpose."

Smith v. Balkcom, supra, 671 F.2d at 859. The magnitude of 
the racial discrepancies found, moreover, reveals that 
racial factors have been at least as influential as several of 
the statutory aggravating bases for the death penalty, under 

the Georgia sentencing scheme. The statistical showing here 

thus translates into a demonstration that Georgia decision­
makers have discriminated in their decisionmaking on the basis 
of race to the extent that, in its operation, the Georgia statute 
makes race an unwritten aggravating circumstance in the decision 
to impose death in the state.

This statistical showing, standing alone, establishes a 
prima facie case of racial discrimination in the administration 
of this law sufficient to support an Equal Protection claim. 

Petitioner maintains, however, that the statistical evidence

101



need not stand alone. He has attempted to obtain and offered 

to present other evidence of racial discrimination in Georgia's 
criminal justice system, and in the State generally, as 
"historical background" constituting additional circumstan­
tial evidence of the discrimination in his case. But the 
Court has denied petitioner discovery of that evidence, holding 
it irrelevant. Order of June 3, 1983.

Petitioner respectfully maintains other circumstantial 
evidence is relevant and appropriate under Jean v. Nelson, 
supra, 711 F.2d at 1486, and cases there cited. Even adhering 
to its ruling, however, the Court remains free to judicially 
notice the general historical background in Georgia of discrim­

ination against blacks. The Court in Jean did just that, 
noting "numerous lawsuits initiated in the past to challenge 
disparate treatment of Haitian immigrants." 711 F.2d at 1490.

Certainly, the Court can and should at least take judicial 
notice of the findings of several Justices of the Supreme Court 

of the United States, in Furman v. Georgia, supra, recognizing 
the influence of race on capital sentencing under the discre­

tionary statute which preceded the one at issue here. Those 
findings give the statistical evidence here added significance: 
for what that evidence shows is that the history of racial 
discrimination in the imposition of the death penalty, sus­

pected in Furman, was real —  and has continued to be part 

of the fabric of capital sentencing in Georgia in

102



the period following that 1972 decision. That provides a 

backdrop against which the evidence here emerges with a coherence 
which simply cannot be attributed to chance or coincidence, or 
anything else but race discrimination.

103



c. The Relevant Universe for 
Comparison of Disparities

Petitioner's data reveals disparate racial results not 
merely in the State of Georgia as a whole, but in virtually 

every conceptual subcategory of cases into which the evidence 
can be divided. The Court has asked about three categorical 
distinctions that may be relevant to its evaluation of that 

evidence: the appropriate geographical unit —  state or county;
the relevant decisionmaker —  prosecutor or jury; and the rel­
evant cases for comparison —  all cases or only those at peti­

tioner's level of aggravation. Petitioner contends that the 
relevant universe is the overall pattern of decisionmaking, 

and discrete evaluation of various subcategories is relevant, 
if at all, only if they demonstrate that the pattern could not 
affect petitioner's case, or some subgroup of which it is a 
part. We will here discuss our position with reference to 
each of the Court's suggestions in turn.

1. Statewide Jurisdiction
The Fourteenth Amendment explicitly prohibits the states 

from denying to any person the equal protection of the laws. 

Although certainly a local official may, alone, violate this 
prohibition, see, e.g., Yick Wo v. Hopkins, supra, it is fin­

ally, "the State [which] must administer its capital sentenc­

ing procedures with an even hand." Gardner v. Florida, 430 
U.S. 349, 361 (1977). The Supreme Court has expressly said 
that it is the state itself which must prevent discrimination 
in the imposition of capital sentences: "if a state wishes

104



to authorize capital punishment it has a constitutional respon­
sibility to tailor and apply its laws in a manner that avoids 
the arbitrary and capricious infliction of the death penalty." 

Godfrey v. Georgia, supra, 446 U.S. at 428.
The relevance of statewide data has been underscored in 

every one of the Supreme Court cases which has examined or 
discussed claims of discriminatory capital sentencing. To 
the extent the decision in Furman v. Georgia, supra, was based 

on a finding of sentencing discrimination, the data cited and 
relied upon by the Justices there dealt with the death sentenc­
ing patterns throughout individual states, or all states nation- 

43/
wide. In Gregg v. Georgia, supra, and its companion cases,

the Court emphasized this aspect of Furman; referring to the
statutory duty of the Georgia Supreme Court to compare death
sentence imposed "throughout the state," 428 U.S. at 205, the

three concurring Justices noted that the statute
. . . gave the Georgia Supreme Court the power and 
the obligation to perform precisely the task which 
three Justices of this Court, whose opinions were 
necessary to the result, performed in Furman: namely, 
the task of deciding whether in fact the death penalty 
was being administered for any given class of crime 
in a discriminatory, standardless or rare fashion.

43/ See 408 U.S. at 250-51 (Justice Douglas); i(3. at 291-93 
(Justice Brennan); î d. at 364-66 (Justice Marshall); jui. at 
312— 13 (Justice White); _id. at 309-10 (Justice Stewart). Cf. 
Beck v. Alabama, 447 U.S. 625 (1980), examining questions con­
cerning the operation of the Alabama statute by looking to the 
results of cases statewide. 447 U.S. at 641 n.18, 645 n.22.

105



428 U.S. at 223. In holding this review function key to the 

constitutionality of statutes after Furman, in the Gregg cases, 
the Court consistently spoke of the importance of the "state­

wide jurisdiction . . . [to] assure consistency, fairness, 

and rationality in the evenhanded operation of the state law," 

Proffitt v. Florida, 428 U.S. 242, 260 (1976); see also Gregg 
v. Georgia, supra, 428 U.S. at 198; Jurek v. Texas, 428 U.S. - 
262, 276 (1976).

Similarly, the cases in this Circuit focus on statewide 
data in discussing and examining claims of racial discrim­
ination in the administration of a state capital sentencing 

scheme. In rejecting such a claim in Spinkellink v. Wainwright, 
supra, the Court of Appeals looked to statewide data which 

was both offered to support the petitioner's claim, and used 

by the State to refute it. 578 F.2d 612, 614-15; see also id. 
at 606-04.

Similarly, the cases in this Circuit focus on statewide 

data in discussing and examining claims of racial discrim­

ination in the administration of a state capital sentencing 
scheme. In rejecting such a claim in Spinkellink v. Wainwright, 
supra, the Court of Appeals looked to statewide data which 

was both offered to support the petitioner's claim, and used 
by the State to refute it. 578 F.2d 612, 614-15; see also id. 

at 606-04. While rejecting the discrimination claim because 
of inadequate evidence in Smith v. Balkcom, supra, the Court

106



acknowledged that the issue involved "the application of 
Georgia's death penalty," 660 F.2d at 585. In the modifica­

tion of its opinion discussing defects in the evidence in Smith, 

the Court stated further that it dealt with sentences "ultimately 
imposed, after trial, in the state," 671 F.2d at 860, and did 
not suggest that data for smaller subdivisions would be nec­
essary or relevant. Although the issue has never been directly 
addressed, the same assumption —  that statewide data is the 

principal touchstone —  appears unquestioned in the Circuit's 
other cases on the subject. See Corn v. Zant, 708 F.2d 549,
563 (11th Cir. 1983); Adams v. Wainwright, 709 F.2d 1443, 1449 

(11th Cir. 1983); Smith v. Kemp, supra at slip op. 19.
Any doubt on this issue remaining, in light of these author­

ities, should be dispelled by the fact that the State of 
Georgia itself had undertaken the responsibility of eliminating 
racial discrimination in death sentencing on a statewide level. 

The Georgia Supreme Court has the statutory duty to review 

capital sentences for "the influence of passion, prejudice, 
or any other arbitrary factor." O.C.G.A. § 17-10-35(c )(i)
(Mich. 1982). Although statewide review is an aspect of 

Georgia sentencing which the Supreme Court has repeatedly 
emphasized in finding the statute constitutional on its face, 
see Stephens v. Zant, supra, it is a protection which the 
Georgia Court has never exercised.

107



The Supreme Court decisions, from Furman to Gregg to

Godfrey to Zant, make clear that the facial validity and actual
operation of this state's statute, at least, must be examined as
a unified whole. Petitioner contends that this wide spectrum
of authority establishes that the pattern of death sentencing,
statewide, is clearly the proper, principal focus of the Court's 

44/
inquiry here.

Data examined more narrowly by judicial circuit may, how- 
everr be relevant. Petitioner has offered evidence here of 
the sentencing patterns in Fulton County, where his case was 
tried, which shows patterns of discrimination based on the race 
of the victim similar to, though necessarily not as sharp as, 

the evidence from the state as a whole. Had the State demon­

strated Fulton County results clearly contrary to those of the 
state as a whole, that fact might have affected the Court's

11/ Maxwell v. Bishop, 398 F.2d 138 (8th CIr. 1968), vacated 
on other grounds, 398 U.S. 262 (1970) does not hold to the 
contrary. In Maxwell there was no statewide data: the data
came from "19 other Arkansas counties [among them not the 
county of the petitioner's conviction] and ... counties 
in 11 other states." 398 F.2d at 146. The Eighth Circuit 
took notice that in Garland County, where Maxwell was convicted, 
there was "no death sentence at all ... until Maxwell's case," 
and of "two other Negroes charged, one was not prosecuted 
and the other was sentenced on a reduced charge." ]̂ 3. at 147 
n.6. It was for those reasons, certainly not present here, 
that the Court of Appeals in Maxwell held it was "not yet 
ready to nullify this Petitioner's Garland County trial 
on the basis of results generally, but elsewhere, throughout 
the South." Id. at 147.

108



inference of intentional discrimination. Yet, there is nothing 
in the record to rebut that inference.

Again, the decision in Jean v. Nelson, supra, is instruc­
tive on this point. In Jean, the Court of Appeals noted that 
parole decisions throughout the immigration system "were made, 

if at all, at a rather low level of authority." 711 F.2d at 
1497; Noting that "proof of non-discriminatory individual actions 

need not bind a court where the concern is proving a pattern 
of discrimination," _id. at n.45, and citing Teamsters v. United 
States, 431 U.S. 324 (1977), the Court in Jean held that, at 

the very least, it was the government's burden to come forward 

with evidence of non-discriminatory decisionmaking at the dis­
trict level, after the systemwide pattern of discrimination was 
shown. Ibid.

Here, as in Jean, the State has offered no such evidence. 
Instead, petitioner anticipated this contention and demonstrated 
that the overall statistical pattern statewide was reflected in 

the county where he was tried. The critical relevance of the 
statewide data to the individual jurisdiction, and to this case, 
thus is not dispelled by an analysis in Fulton County.

2. The Relevant Decisionmaking Stages

Although the evidence includes breakdowns of the racial 

disparities at each stage in the charging and sentencing process, 

petitioner maintains that the results of the system operating as

109



a whole serves as the appropriate framework for assessing 

discrimination. Again, the principal authority on this point 
is Furman v. Georgia, supra.

All of the Justices in Furman who discussed patterns of 

imposition of death sentences did so in terms of overall 
outcome; none focused on the influence of any particular stage 
of the decisionmaking process. Neither have the Circuit 

opinions following Furman, which have discussed the Fourteenth 
Amendment claim made here. See Spinkellink v. Wainwright, 
supra; Smith v. Balkcom, supra. In one early reference to this 
issue, the former Fifth Circuit expressly said that the evidence 
"need not identify an intentional discriminatory act or malevo­
lent actor in the defendants's particular case. See United 

States v. Texas Educ. Agency, 579 F.2d 910, 913-14, nn.5-7 (5th 
Cir. 1978)." Jurek v. Estelle, 593 F.2d 672, 685 n.26 (5th 
Cir. 1979), vacated and affirmed on other grounds, 623 F.2d 

929 (5th Cir. 1980) (en banc).
Although the Jurek decision is no longer controlling author­

ity, the principle it sets forth remains firm. To attempt to 
measure discriminatory impact by focussing on a particular level 
of decisionmaking would be consistent with neither the law of 

capital punishment nor the law of equal protection elsewhere.

In Gregg v. Georgia, while rejecting an argument "that 
prosecutors behave in a standardless fashion in deciding which

110



the three concurring Justices said:

Absent facts to the contrary, it cannot be assumed that 
prosecutors will be motivated in their charging decision 
by factors other than the strength of their case and the 
likelihood that a jury would impose the death penalty 
if it convicts. Unless prosecutors are incompetent in 
their judgments, the standards by which they decide 
whether to charge a capital felony will be the same as 
those by which the jury will decide the questions of 
guilt and sentence.

45/
428 U.S. at 225. Though the opinion assumed this would

mean that capital charging decisions would be based on the ser­
iousness of the offense and the strength of the evidence, ibid., 
that logic equally leads to the hypothesis the evidence sup­
ports here: if jury decisions are influenced by racial factors,
prosecutorial decisions will be as well. It would ignore that 

commonsense assumption to view these decision points in isola­
tion.

It would also mask discrimination, for a reason strongly 

suggested by the evidence here: by anticipating the unequal
treatment cases will receive from juries, based on the racial 

makeup of the defendant and victim, prosecutorial charging 
decisions may well reduce the apparent impact of jury discrim­
ination, though in that process the impact is no less real. For 
whether prosecutors treat cases differently based on race because 

of their own unconscious or conscious racial prejudices, or

cases to try as capital felonies [as] unsupported by any facts,"

45/ Lewis Slayton said much the same thing in his deposition 
testimony. (See Dep. 31, 38-39.)



whether they simply respond to the racial propensity of juries, 
the Equal Protection clause is no less offended. See United 

States v. Texas Education Agency, supra, 579 F.2d at 914 n.7.

To subdivide cases according to decision level would sim­
ilarly contravene the methods used in analysing discrimination 
in analogous contexts elsewhere. Like Furman, the major cases 
evaluating claims of discriminatory intent have not distinguished 
between the actions of particular decisionmakers at the various 

discretionary levels of the enforcement process. See Yick Wo 
v. Hopkins, supra; Jean v. Nelson, supra. Again, if any such 
subdivision is relevant, the message of these other equal 
protection cases is that the state bears the burden of showing 

some difference at one level that might dispel the presumption 
of discrimination created by the overall results. See Jean v . 

Nelson, supra, 711 F.2d at 1494.
Were petitioner's claim based upon the statements or actions 

of a single decisionmaker, of course that alone would not be 

sufficient and petitioner would bear the burden of showing the 
controlling influence of that factor on the process and the 
outcome of the system generally. United States v. Texas Edu­
cation Agency, supra, 579 F.2d at 913. But it clearly is not: 
it is based on an overall, pervasive showing of stark racial 

discrepancies in the Georgia capital sentencing system. Against 
such a showing, it is the State's burden to establish that

112



petitioner was somehow insulated from the system at some level. 
That showing has not been, and cannot be, made in this case.

3. Consideration of the Aggravation Level
Petitioner has never contended that racial factors are 

the exclusive determinant of the outcome of capital cases in 
Georgia. The evidence clearly shows that other factors have 
a clear influence as well. Principal among these is the level 
of aggravation present in the particular cases, measured by 
statutory criteria or nonstatutory factors.

Petitioner's evidence also shows that there are interac­
tions between racial factors and these legitimate influences.

One important aspect of this interaction is that discrimina­

tion is most evident in cases which fall in the statistical 
mid-range of aggravation. In those cases so aggravated that 

a death sentence is extremely likely, or so mitigated it is 
statistically or legally impossible, few racial disparities 

are seen. It is in the mid-range of cases where disparities 
are most strikingly evident. (See GW 8.) As Professor Baldus 
has testified, this finding supports the "liberation hypoth­
esis": "doubts about the evidence free the jury to follow

sentiment," H. KALVAN AND H. ZEISEL, THE AMERICAN JURY 166 
(1965), and "sentiment" is influenced by racial factors in 

those cases in which it holds sway.
Dr. Woodworth's testimony establishes that Warren McCleskey's 

case falls at the aggravation level where racial factors have

113



perhaps their greatest statistical influence, over 20 points. 

Because of that, the Court need not decide here the legal out­
come in a case which fell at the extreme level of aggravation.
Were this such a case, petitioner would submit that the burden 
would rest on the State to rebut a prima facie showing of sys­
temwide discrimination. The law seems particularly clear that 
it is the state which shoulders

the burden of establishing that the same decision 
would have resulted even had the impermissible 
purpose not been considered. If this were estab­
lished, the complaining party in a case of this 
kind no longer fairly could attribute the injury 
complained of to improper consideration of a dis­
criminatory purpose. In such circumstances, there 
would be no justification for judicial interference 
with the charged decision.

Arlington Heights v. Metropolitan Housing Corp., supra, 429 
U.S. at 271 n.21; see Teamsters v. United States, supra, 431 
U.S. at 359; Mt. Healthy Board of Education v. Doyle, 429 U.S. 
274, 287 (1977).

But clearly, in this case that burden cannot be met: peti­
tioner's evidence makes it clear that the sentencing outcome 
of his case, absent the influence of racial factors, was far 

from preordained —  and the State offered no evidence to the 

contrary. In this case, at least, separation of the data by 

aggravation level actually enhances the strength and signif­
icance of petitioner's discrimination claim.

114



D. The State's Burden of Proof

The evidence petitioner has offered is clearly sufficient
to support a finding of discrimination. Like any other evidence
making out a prima facie case of discrimination, however, it is
subject to rebuttal. But the State has offered virtually no
contrary evidence to rebut petitioner's case.

What evidence the State did offer should be evaluated
under settled standards, most recently applied in Jean v. Nelson,

supra. In Jean the government's rebuttal position was strikingly
similar to that of the State here:

The government's case rested on two independent gen­
eral arguments. The first attacked the validity of 
plaintiff's statistical evidence, and the second 
proffered explanations to dispel the appearance of 
disparate impact. Both government arguments are 
considered properly as rebuttal. See Johnson v .
Uncle Ben's, Inc., 628 F.2d 419, 424 (5th Cir. 1980) 
(rebuttal includes existence of "legitimate nondis- 
criminatory reason for defendant's action, or evi­
dence of "total unacceptability of plaintiff's 
statistical evidence"), vacated and remanded, 451 
U.S. 902, 101 S.Ct. 1967, 68 L.Ed.2d 290 (1981), 
modified in part, rev'd in part, 657 F.2d 750 (5th 
Cir. 1981), cert, denied, U.S. , 103 S.Ct.
293, 74 L.Ed.2d 277 (1982).

711 F .2d at 1487.

In this case, it remains unclear to petitioner on what 

basis the State can challenge his statistical evidence as unac­
ceptable. All three of petitioner's experts testified that the 

data here was gathered and analyzed according to accepted, 
"state-of-the-art" social science techniques. The State's only 
witness qualified to speak to the social science methodology

115



used in gathering the evidence, Dr. Burford, said nothing to 
the contrary.

The only pervasive criticism of the database the State 

appeared to make involved the relevance and accuracy of the 

Parole Board files from which most of the data was gathered.
These files were compiled under a state law and a set of guide­
lines that required careful and accurate data collection, and 

the records themselves were in the possession of the State.
No expert opinion was presented to indicate the records were 
materially inaccurate in any way —  certainly not in any way 
which would systematically influence the outcome of the research, 
or change the finding of discrimination. To the contrary, 

petitioner called a witness and presented evidence showing 
that the Parole Board files of this state are compiled with 
extraordinary care. Similarly, though the state appeared to 

challenge Dr. Baldus' well-founded assumption on the proper 
coding of "unknowns" —  that, in most circumstances, it could 

be assured that the information would be in the file if it 
was before the decisionmaker —  no qualified expert testified 

that this assumption was untenable. No attempt was made to 

show that the recoding of these entries changed the results 
Professor Baldus obtained. Indeed, Professor Baldus' worst 

case and listwise deletion analyses suggested no impact on the 
racial disparities. And the state made no attempt to call any

116



witnesses, or to present prosecutors' files or police records,

showing in any way that the data were inaccurate.
In this respect, again, this case is much like Jean. There,

as here, the statistical evidence utilized was derived from
various government records. 711 F.2d at 1488-89. There, as

here, the government apparently argued that these records did
not reflect all possible factors influencing parole decisions.
711 F.2d at 1494. But there, as here, the government called
no witnesses to show that critical information was actually
left out of the files, or what such information might show. The
Court held the government's rebuttal to be utterly insufficient:

[T]he government failed to demonstrate that parole 
decisions were made in a non-discriminatory manner.
Although the evidence at trial was clear that parole 
decisions were made, if at all, at a rather low 
level of authority in the immigration corps, the 
government declined to offer as a witness any individ­
ual who could explain or demonstrate exactly which 
factors were considered in granting parole. See 
Castaneda v. Partida, 430 U.S. at 497-99, 97 S.Ct. 
at 1281-82, 51 L.Ed.2d at 512-13 (discussing value of 
jury commissioner testimony to rebut statistical 
evidence of discrimination). In this regard the 
opinion of the Second Circuit in Betrand v. Sava,
684 F.2d 204 (2d Cir. 1982) is instructive. In 
Sava the Court of Appeals admonished the district 
judge for substituting his judgment for the dis­
cretion of the District Director, who took the 
stand to explain in specific cases why members of 
the plaintiff class were not paroled. The record 
before us is devoid of any such evidence.

711 F.2d at 1497 (footnotes omitted).
Aside from this question, all the State has asserted against

the database itself is a series of minor quibbles —  unsupported
by any qualified witness —  with some of the methodological and

117



coding decisions made in the course of the research. Peti­

tioner's expert testimony showed the exercise of professional 
judgment is inevitable in research of this type and magnitude. 
But no expert testified that what was done in petitioner's 

studies was unreasonable, or outside of the range of accepted 
social science methodology. Nothing was shown, or indeed the 
State made no effort at all to demonstrate, that different 
decisions or methods would have produced a different result.

The State has therefore not even approached the showing 
required to rebut a prima facie statistical case on method­
ological grounds: a showing that the methods were "totally

unacceptable." Whatever minor imperfections these data, might 
share with any complex study, the record shows they were gath­

ered and analyzed in the most sophisticated and careful scien­
tific manner. On this point, as on most, of its rebuttal the 
State's constant position

. . . consists of nothing but "mere protestation 
and arguments of counsel." These do not consti­
tute rebuttal of a prima facie case of discrim­
ination, especially one as strong as that proven 
by plaintiffs. See Burdine, supra, 450 U.S. at 
255 n .8, 101 S.Ct. at 1094 n.8, 67 L.Ed.2d at 
216 n.8; Castaneda, supra, 430 U.S. at 499 n.19,
97 S.Ct. at 1282 n.19, 51 L.Ed.2d at 513 n.19;
Alexander v. Louisiana, 405 U.S. 625, 632-33, 92 
S.Ct. 1221, 1226-27, 31 L.Ed.2d 536, 543 (1972).

Jean v. Nelson, supra, 711 F.2d at 1501-02.

The State's second set of arguments, "to dispel the appear­
ance of disparate impact," even more strikingly resembles the 

government's position in Jean. There, as here, the main thrust

118



of the government's argument was that "what is actually occur­

ring could be explained in any number of ways other than inten­
tional discrimination," by accounting for "a number of finely 
tuned factors or 'qualifications' which were not included in 
the plaintiffs' data base." 711 F.2d at 1494. The Court in 

Jean rejected this argument in no uncertain terms, reversing 
the District Court's reliance on it as clearly erroneous.

[I]t is nonsensical to require plaintiffs to establish 
qualifications of the plaintiff class when the relevant 
qualifications were vague, uncertain or unknown. The 
district court could not state the detention/parole 
policy with any certainty. The district court found 
that no guidelines for parole were ever promulgated.
The district court stated that because of the lack of 
guidelines government witnesses described one policy 
while their subordinates enforced another. See Louis 
III at 981 n.24.

Yet, despite its inability to locate a written 
parole policy, and its inability to specify exactly 
what the policy was, either of which would indicate 
those qualifications important to a parole determin­
ation, the court below swept away plaintiff's sta­
tistics on the basis that they did not reflect the 
relevant qualifications. The district court stated:

Factors that may be considered include the 
age and health of the alien as well as the 
reason he does not appear entitled to enter 
this country. Other factors include being 
accompanied by a minor and pendency of an 
1-130 application. These factors, with the 
exception of documentation, were not sep­
arately analyzed by Plaintiffs' expert.
With regard to his analysis of the signifi­
cance of possessing documents it was far 
too simplistic because it did not distinguish 
between the types of documents possessed 
and the facial validity thereof.

Louis III at 982. This determination is clearly 
erroneous and wholly unsupported by the evidence.

119



Pullman Standard v. Swint, ___ U.S. ___, 102 S.Ct.
1781, 72 L.Ed.2d 66 (1982). . . . Although factors 
similar to those mentioned by the district court 
were suggested piecemeal during the course of the 
trial, there is no evidence to suggest they formed 
a coherent policy. Rejecting a well-supported claim 
of discrimination on the basis of vaguely-asserted 
and ill-defined qualifications represents a clear 
error of law and fact.

711 F.2d at 1496 (footnotes omitted).

The decision in Jean thus again makes clear that a prima 
facie case of discrimination in this context can only be rebut­

ted by evidence, not just argument. That is consistent with 
analogous discrimination law elsewhere. See Village of 
Arlington Heights v. Metropolitan Housing Board, supra, Mt.
Healthy School District v. Doyle, supra. The only thing resem­

bling statistical evidence the State has offered is Dr. Katz' 
showing that, in general, cases involving white victims are 
more aggravated than black victim cases. Petitioner's experts all 

acknowledge this; but as their testimony shows, that does not 
dispel the showing of discrimination here. For it remains 
unrebutted —  and undeniable in light of Professor Baldus' 
findings and testimony —  that significant racial disparities 
remain, at levels sufficient to support an inference of dis­

crimination, when aggravation level is controlled for in every 

reasonably conceivable manner. The State's resort to such 
alternative explanations, without any evidence of the results 

when they are controlled for, cannot satisfy their burden 
of proof. Certainly, it does not dispel the inference of

120



discrimination "by preponderance of the evidence," showing that 

the same result would have obtained in the absence of racial 
discrimination. Cf. Mt. Healthy School Board of Education v . 

Doyle, supra, 429 U.S. at 286. Reference to factors which 

unrebutted evidence shows cannot fully account for discrimin­
ation patterns amounts to nothing more than a "pretext" expla­

nation for those results. Cf. McDonnell Douglas Corp. v. Green, 
411 U.S. 792, 804-06 (1973), supra. The State's meager unad­
justed showing —  which, in fact, reveals nothing more than 

petitioner's witnesses have readily acknowledged and fully 
accounted for in their own alternative analyses —  must fail 
as rebuttal as a matter of law.

So too, must the only other item of possible rebuttal 

evidence the State could rely on here: District Attorney Lewis
Slayton's deposition testimony that he did not "recall any case 

where race has ever been a factor." (Dep. at 78.) Mr. Slayton 

testified to no contrary statistics, and offered no set policies 
or alternative explanations for the discrimination patterns 
shown here. However honestly felt, his testimony is nothing 
more than a "mere protestation" of his lack of intent to dis­

criminate :
The law is clear that "mere protestation" is insuffficient 
to rebut a plaintiff's prima facie case. E.g., Castaneda 
v. Partida, 430 U.S. at 499 n.19; Alexander v, Louisiana, 
405 U.S. 625, 633, 92 S.Ct. 1221, 1226-27, 31 L.Ed.2d 536, 
543 (1972). Yet, the record contains nothing more than 
the self-serving testimony of high government officials 
that the policy was not intended to be discriminatory . . .

121



as well as a variety of vague, contradictory state­
ments concerning parole factors. . . . Beyond this, 
the record is clear that whatever the officials 
responsible for policy formulation intended, those 
intentions never were conveyed to immigration offi­
cials responsible for implementation of the policy.

Jean v. Nelson, supra, 711 F.2d at 1496-97.

Mr. Slayton made it clear his office did not have any 
set policy for decisions in capital cases (Dep. at 31), let 
alone one uniformly communicated to the other decisionmakers 
in his office or consistent with the policies of prosecutors 
elsewhere in the United States. Under Jean, in those circum­
stances, even if his own actions are entirely in good faith, 
his testimony adds no rebuttal weight.

As in Jean, therefore, this case must ultimately be 
decided on the strength of petitioner's statistical evi­
dence —  much of which anticipates potential counterarguments 

that can be made against his claim —  and the background 

of race discrimination in Georgia he has offered to prove 
and maintains the Court can judicially notice. The ques­

tion, as in any other discriminatory case, is whether a 

preponderance of all the evidence before the court shows 
racial considerations have influenced capital sentencing 
in this State.

It is inevitably the task of the trier of fact 
to decide whether plaintiffs have succeeded, in 
the face of the defendant's rebuttal, in proving, 
by a preponderance of the evidence, that defendant's 
actions were taken in part because of a discrimi­
natory intent or purpose. Cf. Burdine, 450 U.S. at 
252-53, 101 S.Ct. at 1093.

122



Jean v. Nelson, supra, 711 F.2d at 1486. Petitioner submits 

he has succeeded in that showing here.

123



III. The Appropriate Relief
Though the evidence presented here demonstrates racial 

disparities which plague the entire Georgia death sentencing 
system, ultimately the only issue this Court need decide 
in this case is the constitutionality of the death sentence 
imposed on petitioner Warren McCleskey. Indeed, it is the only 

question this Court can decide in this case, for its jurisdic­
tion in this matter lies under 28 U.S.C. § 2241, which grants it 
the power to grant a writ of habeas corpus to release a person 
confined in violation of the law, see Fay v. Noia, 372 U.S. 391 
(1963), upon a finding that "[h]e is in custody in violation of 
the Constitution or laws or treaties of the United States," 28 

U.S.C. § 2241(c)(3). That is the limit of the Court's remedial 

power under the habeas statute. See Anderson v. Beto, 469 F.2d 
1076 (5th Cir. 1972); Dixon v. Beto, 472 F.2d 598 (5th Cir. 1973).

In a death penalty case, upon a finding that an individual 

habeas petitioner's sentence has been imposed under an uncon­
stitutionally administered statute, the appropriate remedy is 
thus "to grant the writ if the state .... should fail or refuse 
within a reasonable period of time to reduce [the] ... death 
sentence to a period not exceeding life imprisonment." Newman 

v. Wainwright, 464 F.2d 615, 616 (5th Cir. 1972). The Court 
in Newman so held in a case directly analogous to this: a death

sentence held invalid in habeas corpus proceedings after Furman 

v. Georgia, supra, because it was pronounced under a state law 

"which ... has been inequitably, arbitrarily, and infrequently

124



Newman remains the law of this Circuit, and the lawimposed."
generally, as to the appropriate habeas relif in such a case.
See also Smith v. Estelle, 445 F.Supp. 647, 654 (N.D. Texas), 
aff'd, 451 U.S. 454 (1981); Goode v. Wainwright, 704 F.2d 593, 

612-13 (11th Cir. 1983) .
Petitioner recognizes that any favorable decision the 

Court might render in this case might have ramifications for 

other death penalty cases in the State of Georgia. As we have 

noted above, however, it is not wholly clear what that impact 
might be in different factual contexts —  where the race of the 
defendant or the victim might be different than it is here, for 
example, or where the aggravation level was markedly higher.

As in any other area of the law, the application of a legal rule 
to subsequent cases is not the responsibility of the Court which 

applies the rule in an initial case.
Petitioner Warren McCleskey has shown disparities in 

Georgia's administration of its capital statute strong enough to 
support a finding of intentional discrimination. He has also 

shown that the characteristics of this case are such that there 
is a real likelihood his sentence may have been the product of 

such discrimination. The State had presented nothing which comes 

close to justifying or explaining away those facts. Under the 

Equal Protection Clause of the Fourteenth Amendment, petitioner 

is entitled to habeas relief.

125



CONCLUSION
The writ should therefore issue, ordering petitioner 

to be released unless, within a reasonable time, he is re­
sentenced to life imprisonment.

Dated: September 26, 1983

Respectfully submitted,

ROBERT H. STROUP
1515 Healy Building 
Atlanta, Georgia 30303

JOHN CHARLES BOGER 
10 Columbus Circle 
New York, New York 10019

TIMOTHY K. FORD
600 Pioneer Building 
Seattle, Washington 94305

ANTHONY G. AMSTERDAM
New York University Law School 
40 Washington Square South 
New York, New York 10012

ATTORNEYS FOR PETITIONER
BY:

126



Appendix A: Summary of the Relevant
Legislative History of the Equal 

Protection Clause of the Fourteenth Amendment

The legislative history of the Equal Protection 
Clause reveals that the Framers of the Fourteenth Amendment 
intended it to prohibit unequal administrations of state 

criminal justice that adjusted the severity of punishment to 
the race of the victim. It is clear that the Congressional 
authors of the Fourteenth Amendment were aware of the inequal­
ity in the protection which the Southern criminal justice

1/systems accorded to whites and to blacks: the members
of the Congressional Joint Committee on Reconstruction —  the
committee that drafted the bill which was eventually adopted

2/
as the Fourteenth Amendment —  heard extensive testimony

V  As the Court in Strauder v. West Virginia, 100 U.S. 303 
(1886) asserted, "[The Black community] especially needed 
protection against unfriendly action in the State where they 
were resident. It was in view of these considerations that 
the 14th Amendment was framed and adopted." 100 U.S. at 306.
"It is well known that prejudices often exist against parti­
cular classes in the community, which sway the judgement of 
the jurors, and which, therefore, operate in some cases to 
deny to persons of those classes the full enjoyment of that 
protection which others enjoy ... and the apprehension that, 
through prejudice, they might be denied equal protection, 
that is, that there might be discrimination against them, was 
the inducement to bestow upon the National Government the 
power to enforce the provision that no State shall deny to 
them the Equal Protection of the Laws." Id. at 309.
2/ For discussion of the role of the Joint Committee on Re­
construction (also known as the Committee of Fifteen), see H. FLACK, 
THE ADOPTION OF THE FOURTEENTH AMENDMENT 60-74 (1908); B. KENDRICK, 
THE JOURNAL OF THE JOINT COMMITTEE OF FIFTEEN ON RECONSTRUCTION 
(1914); Fairman, Does the Fourteenth Amendment Incorporate the Bill 
of Rights?, 2 Stan. L. Rev. 5, 19-21, 124-25, 41-42 (1949).

l



on the subject; and the members of Congress received
4/

copies of the transcripts of this testimony.

The Congressional debates on the Fourteenth Amendment
make clear that the members of the Joint Committee on

Reconstruction and the other members of the House and Senate
intended to establish the Amendment as a remedy for these
abuses, and thereby to end the Southern States' failure to
provide equal protection for blacks through their criminal
justice systems. Representative John Bingham, member of the

5/
Committee and the drafter of Section 1 of the Amendment,

3/

3/ See Report of the Joint Committee on Reconstruction at 
the First Session, Thirty-Ninth Congress, Part II, at 25, 62, 
143, 209, 210, 213; Part II, at 141; Part IV, at 46-48, 50, 
75-76. For examples of the testimony heard by the Committee, 
see notes lg-2g supra.
4/ The Joint Committee on Reconstruction "ordered all 
evidence taken to be brought forward as rapidly as might be, 
to be printed" and distributed to the members of Congress. 
Cong. Globe, 39th Cong., 1st Sess. 1368 (1865-1866). Members 
of the House delegation of the Joint Committee submitted 
substantial portions of the testimony to the House as early 
as March 7, 1866. Id. at 1240 (remarks of Rep. Washburne).
On March 13, 1866, the House passed a resolution by the Joint 
Committee on Printing that; "twenty-five thousand extra 
copies of each of the reports of testimony taken by the joint 
select committee on reconstruction, together with the 
accompanying documents, be printed without covers for the use 
of the House." Id. at 1368. Although it is not clear when 
the testimony was submitted to the Senate, it appears that 
the presentation of the testimony occurred before the major 
Senate debates over the proposed Fourteenth Amendment in 
June, 1866. See, id. at 1368 (remarks of Rep. Conkling); id. 
at 2765 (remarks of Sen. Howard); B. KENDRICK, supra, 
note 3g, at 264-265.
5/ Fairman, supra, note 3g, at 19-20, 41-43.

li



argued to the House that a constitutional amendment was
required to eliminate the denial to certain citizens of the
protection of the courts, and the deprivation of these
citizens' right to procure redress of injuries through the

6/
courts. Similarly, Representative Thaddeus Stevens,

Chairman of the House delegation of the Joint Committee
7/

and drafter of an earlier version of the Amendment,
explained on the floor of the House that the Amendment

was necessary to afford equal means of redress to blacks 
8/

and whites. Statements by other Representatives and 

by Senators reflect the same intention to end unequal admini­
stration of southern criminal justice systems that resultedV
in inadequate protection of blacks.

6/ Cong. Globe, 39th Cong., 1st Sess. 1064, 1090 (1865-1866) 
(remarks of Rep. Bingham).

7/ Fairman, supra note 3g, at 19-21, 41-43.
8/ Cong. Globe, 39th Cong., 1st Sess. 2459 (1865-1866) 
(remarks of Rep. Stevens).

9/ Several of the Representatives and Senators declared that 
the constitutional amendment was necessary to provide equal 
"protection of the laws" to the blacks in the South. See, 
e.g., Cong. Globe, 39th Cong., 1st Sess. 2459 (1865-1866) 
(remarks of Rep. Stevens; ui. at 2539 (remarks of Rep. 
Farnsworth); iji. at 2765-2766 (remarks of Sen. Howard); id. 
at App. 227 (remarks of Rep. Defrees). The term "protection 
of the laws" plainly encompassed the concept of protecting 
citizens from criminal acts by other private citizens. See, 
e.g., id. at 2890 (remarks of Sen. Cowan) ("If a traveler 
comes here from Ethiopia, from Australia, or from Great

(Continued)

in



The legislative history reveals that the Framers 
were particularly intent upon outlawing the statutory embodi­
ment of that inequality: the Black Codes that were enacted
in Southern states after the Civil War as an effort to 
preserve white supremacy. The Codes often provided more 
drastic punishments for crimes when the victims were white,
as well as different punishments for black and white defen-

W
dants.

Congressional debates upon legislation provide further 
evidence of this basic purpose of the Fourteenth Amendment.

The debates over the proposed Civil Rights Act of 1866 —  which 

occurred in the same session as the debates over the Fourteenth

9/ (Continued)

Britain, he is entitled, to a certain extent, to the protec­
tion of the laws. You cannot murder him with impunity. It 
is murder to kill him, the same as it is to kill another man. 
You cannot commit an assault and battery on him, I apprehend. 
He has a right to the protection of the laws ....") It was 
deprivation of this type of "protection of the laws" that was 
described in the hearings before this Joint Committee 
on Reconstruction, see notes 1-2 supra; as Senator Howard, 
a member of the Committee, explained in introducing the bill 
of the Senate, the Committee had drafted the bill to end the 
types of evils described in the hearings. See id. at 2765.
For other discussion of the need to provide this sort of 
protection to blacks, see, e.g ., id. at 2964 (remarks of Sen. 
Stewart) (provision of the amendment necessary to ensure 
continued protection of blacks and white loyalists in the 
South from "sure violence at the hands of rebels").

10/ 1 Fleming, Documentary History of Reconstruction, 273-312
(1906) and McPherson, History of the Reconstruction, 29-44 
(1971). See also Bichel, The Original Understanding and The 
Segregation Decision, 69 Harv. L. Rev. 7, 56-58 (1956).

IV



Amendment, and which historians recognize as reflective
12/

of the intent of the Amendment — ■ bespeak a congressional

determination to correct the failings of Southern justice so as 
to provide to blacks and unionists adequate protection against

1 1 /

11/ The debates over both the proposed Fourteenth Amendment and 
the proposed Civil Rights Act occurred in the first session 
of the 39th Congress. The proposed Fourteenth Amendment was 
introduced into both houses of Congress on Feb. 13, 1866 (as 
S.R. No. 30 in the Senate, Cong. Globe, 39th Cong., 1st Sess.
806 (1865-1866); and as H.R. No. 63 in the House, id. at 
813). The bill was immediately tabled in the Senate, i_d. at 
806, and remained tabled; the bill was debated in the House 
but then postponed for further consideration on Feb. 28,
1866, icL at 1095. On April 30, 1866, a new version of 
the proposed Fourteenth Amendment was introduced into both 
houses (as S.R. No. 78 in the Senate, ijl. at 2265; and as 
H.R. No. 127 in the House, id., at 2286). The Amendment 
passed the House on May 10, 1866, ijl. at 2545; passed the 
Senate with revisions on June 8, 1866, iji. at 3042; and was 
approved in its revised form by the House on June 13,
1866, ijd. at 3149. The proposed Civil Rights Act of 1866 was 
introduced into the Senate (as S. No. 61) on Jan. 5, 1866, 
id. at 129; passed by the Senate on Feb. 2, 1866, _id. at 606; 
passed with amendments by the House on March 13, 1866, id. at 
1367; and approved in its amended form by the Senate on March 
14, 1866, id̂ . at 1416. President Johnson thereafter vetoed 
the bill, but the Senate carried it over his veto on April 6, 
1866, id. at 1809, and the House carried it over the veto on 
April 9, 1866, id. at 1861.

12/ As Fairman explains, the "same topics" were considered 
by Congress in the framing of the Civil Rights Act and the 
constitutional amendment, and "[t]hus much that was said on 
the Civil Rights Bill proves meaningful in a study of the 
understanding on which the Fourteenth Amendment was based." 
Fairman, supra note 3g, at 8. Flack observes; "The legis­
lation preceding the adoption of the Amendment will probably 
give an index to the objects Congress was striving to obtain, 
or the evils for which a remedy was being sought." H. FLACK, 
supra note 3g, at 11. See also J. TENBROEK, EQUAL UNDER 
LAW 201-203 (1965).

v



violence and murder. The debates concerning the pro­

posed Freedman's Bureau Act of 1866 —  which also occurred 

in the same session as the debates over the Fourteenth

12/

13/ See, e.g., Cong. Globe, 39th Cong., 1st Sess. 1159 (1865- 
1866) (remarks of Rep. Windom) (Civil Rights Bill "declares 
that henceforth the perjured white traitor shall have no 
civil rights or immunities which are denied to the black 
patriot; ... that in laws and proceedings for the security 
of person and property, and for the punishment of offenses, 
no discrimination shall be made in favor of traitors, because 
they are white and have always been petted and pampered by 
the Government, as against patriots who are black .... It 
merely provides safeguards to shield them from wrong and 
outrage, and to protect them in the enjoyment of that lowest 
right of human nature, the right to exist."); _id. at 1267 
(remarks of Rep. Raymond) (observing that previous speakers 
had traced the need for the Civil Rights Act to murder and 
violence perpetrated on blacks and unionists in the south, 
and arguing that accurate analysis of the conditions in 
the south must await the printing of the testimony heard by 
the Joint Committee on Reconstruction); ijd. at 1292 (remarks 
of Rep. Bingham) (arguing that the goals of the Civil Rights 
Bill could not be achieved except through the passage of his 
proposed constitutional amendment because, without it, the 
federal government does not have the constitutional authority 
to "punish the murderer for taking away in time of peace the 
life of the noblest, and the most unoffending as well, of 
your citizens, within the limits of any State of the Union"); 
id. at 1293 (remarks of Rep. Shellabarger) (explaining that 
the provisions of the bill guarantee that persons, "without 
regard to such race or condition, shall have the same right 
to contract, convey, sue, testify, inherit, and to claim 
benefit of the laws protecting person and property as white 
citizens ...."); see also, e.g., id. at 476 (remarks of Sen. 
Trumbull) (explaining that the civil rights protected by the 
bill include the right to "full and equal benefit to all laws 
and proceedings for the security of person and property"); 
id. at 1118 (remarks of Rep. Wilson) (explaining that the 
"right of personal security" as defined by Blackstone, 
"'[cjonsists in a person's legal and uninterrupted enjoyment 
of his life, his limbs, his body, his health, and his repu­
tation" ).

vi



Amendment, and which are also relevant in ascertaining
15/

the objectives of the constitutional amendment —  similarly
evidence the intention to establish some means of reforming

southern justice and of giving blacks equal protection
±6/

against violence and murder. The later debates over the

proposed "Ku Klux Klan Act" of 1871 provide clear evidence

14/

14/ For the chronology of the Fourteenth Amendment, see note 
llg supra. The Freedmen's Bureau Bill of 1866 was introduced 
in the Senate (as S. No. 60) on Jan. 5, 1866, _id. at 129, and 
passed by the Senate on Jan. 25, 1866, Id. at 421. The bill 
was introduced in the House of Representatives on Jan. 25, 
1866, Id. at 435, and passed by the House in an amended 
version on Feb. 6, 1866, id. at 688. The Senate concurred in 
the amendments and passed the bill as amended on Feb. 19,
1866. VI Message and Papers of the Presidents 398-405 
(1897). A new Freedmen's Bureau Bill was thereafter intro­
duced in the House on May 22, 1866, Cong. Globe, 39th Cong., 
1st Sess. 2743 (1865-1866) and in the Senate on June 11, 
1866, i^. at 3071. The House approved the second Act on 
May 29, 1866, _id. at 2878; and the Senate passed a modified 
version on June 26, 1866, ij3. at 3413; and the Conference was 
adopted on July 2 and 3, 1866, ^d. at 3524, 3562. President 
Johnson again vetoed the bill, VI Messages and Papers of the 
Presidents, supra at 422-426. Both houses, however, on July 
16, 1866, voted to override the veto. Cong. Globe, 39th 
Cong., 1st Sess. 3842, 3850 (1865-1866).
15/ See H. FLACK, supra note 3g, at 11; J. TENBROEK, supra 
note 12g, at 201-203 ("The one point upon which historians of 
the Fourteenth Amendment agree, and indeed, which the evidence 
places beyond cavil, is that the Fourteenth Amendment was 
designed to place the constitutionality of the Freedmen's 
Bureau and civil rights bills ... beyond doubt.")
16/ See, e.g., Cong. Globe, 39th Cong. 1st Sess. 339 (1865- 
1866) (remarks of Rep. Creswell) (arguing against a proposed 
amendment to the Freedmen's Bureau that would limit the bill's 
operation to the Rebel States, because: "There is assuredly
a necessity for the operation of this bill in the State which

(Continued)

v n



that the Fourteenth Amendment was designed to prohibit state 
discrimination on the basis of race of the victim of a 

criminal act: Senators and Congressmen repeatedly assert in
these debates that state authorities are violating the 
Fourteenth Amendment equal protection guarantees when they

16/ continued

I in part represent on this floor. I have received within 
the last two or three weeks letters from gentlemen of the 
highest respectability in my State asserting that combinations 
of returned rebel soldiers have been formed for the express 
purpose of persecuting, beating most cruelly, and in some 
cases actually murdering the returned colored soldiers of the 
Republic. In certain sections of my State the civil law 
affords no remedy at all. It is impossible there to enforce 
against these people so violating the law the penalties 
which the law has prescribed for these offenses."); ijl. at 
340 (remarks of Rep. Wilson)(arguing in support of the bill 
because: "wherever the Freedmen's Bureau does not reach, 
where its agents are not be found, there you will find 
injustice and cruelty and whippings and scourgings and 
murders that darken this continent..."); id. at 516-17 
(remarks of Rep. Eliot) ("[Tjhere is not one rebel state 
where these freedmen could live in safety if the arm of 
the Government is withheld .... In Mississippi houses have 
been burned and negroes have been murdered .... [I]f the
arm of the Government is withheld from protecting these men, 
and the powers of this bureau are not continued and enlarged, 
much injustice will be done to these freedmen, and there will 
be no one there to tell the story."); id. at 631, 633 
(remarks of Rep. Moulton) ("One object of the bill is to 
ameliorate the condition of the colored man and to protect 
him against the rapacity and violence of his southern prose­
cutors.... Suppose the Army was removed; suppose there was 
no Freedmen's Bureau for the purpose of protecting freedmen 
and white refugees there, what would be the consequence?
Why, sir, the entire body of freedmen would be annihilated, 
enslaved, or expatriated .... The testimony which will be 
published that has been exhibited before the committee of 
fiteen will astonish the world as to ... the condition of 
things in the South.").

viii



refuse to enforce their criminal laws to protect particular
12/classes of persons.

17/ See, e.g., Cong. Globe, 42nd Cong., 1st Sess. 697 
(1871) (remarks of Sen. Edmunds, Senate sponsor of the bill); 
id. App. 116 (remarks of Rep. Shellabarger, House sponsor of 
the bill). There were some members of Congress who read the 
Fourteenth Amendment restrictively to apply on to discrimina­
tion by state statutes. See, e.g., id., App. 118 (remarks of 
Sen. Blair); i^., App. 259 (remarks of Rep. Holman). The 
majority of Congress, however, rejected this restrictive 
interpretation and recognized that the Fourteenth Amendment 
applies to unequal protection of particular classes in the 
administration of the law. See, e.g., id. at 334 (remarks 
of Rep. Hoar); _icl. at 482 (remarks of Rep. Wilson); ijd. at 
505-06 (remarks of Sen. Pratt); id. at 606-08 (remarks of 
Sen. Pool); idL , App. 153 (remarks of Rep. Garfield); id., 
App. 300 (remarks of Rep. Stenvenson); _id. , App. 315 (remarks 
of Rep. Burchard).

IX



CERTIFICATE OF SERVICE

I hereby certify that I am one of the attorneys for 

petitioner and that I served the annexed Memorandum of Law 

on respondent by placing a copy in the United States mail, 
first-class mail, postage prepaid, addressed as follows:

Mary Beth Westmoreland, Esq.
Assistant Attorney General
132 State Judicial Building
Atlanta, Georgia 30334
Done this 26th day of September, 1983.

JOHN CHARLES BOGER

Copyright notice

© NAACP Legal Defense and Educational Fund, Inc.

This collection and the tools to navigate it (the “Collection”) are available to the public for general educational and research purposes, as well as to preserve and contextualize the history of the content and materials it contains (the “Materials”). Like other archival collections, such as those found in libraries, LDF owns the physical source Materials that have been digitized for the Collection; however, LDF does not own the underlying copyright or other rights in all items and there are limits on how you can use the Materials. By accessing and using the Material, you acknowledge your agreement to the Terms. If you do not agree, please do not use the Materials.


Additional info

To the extent that LDF includes information about the Materials’ origins or ownership or provides summaries or transcripts of original source Materials, LDF does not warrant or guarantee the accuracy of such information, transcripts or summaries, and shall not be responsible for any inaccuracies.

Return to top