McCleskey v. Kemp Motion for Leave to File Brief Amici Curiae and Brief Amici Curiae in Support of Petitioner

    No. 84-6811

I n  th e

Bnpvmv tour! of tfje United States
O ctober T eem , 1985

W arren  M cClesk ey ,

— v._
Petitioner,

R a l p h  M. K e m p , Superintendent, 
Georgia Diagnostic & Classification Center.

ON WRIT OP CERTIORARI TO THE UNITED STATES 
COURT OF APPEALS FOR THE ELEVENTH CIRCUIT

MOTION FOR LEAVE TO FILE BRIEF AMICI CURIAE 
AND BRIEF AMICI CUMAE FOR DR. FRANKLIN M. FISHER, 
DR. RICHARD O. LEMPERT, DR. PETER W. SPERLICH, 
DR. MARVIN E. WOLFGANG, PROFESSOR HANS ZEISEL 
& PROFESSOR FRANKLIN E. ZIMRING IN SUPPORT 

OF PETITIONER WARREN McCLESKEY

M ich ael  O. F in k elstein  
M artin  F . R ic h m a n *
B arrett S m it h  S chapiro 

S im on  & A rmstrong  
26 Broadway
New York, New York 10004 
(212) 422-8180

Attorneys for  Amici Curiae 

^Counsel of Record



TABLE OF CONTENTS
TABLE OF AUTHORITIES.................  a
MOTION FOR LEAVE TO FILE BRIEF AMICI

CURIAE.....................   i
BRIEF AMICI CURIAE ...................  1

SUMMARY OF ARGUMENT .............  1
ARGUMENT ........................ 7
I. THE BALDUS STUDIES

DEMONSTRATE THAT IN THE 
STATE OF GEORGIA, THE RACE 
OF THE HOMICIDE VICTIM HAS
BEEN AN IMPLICIT AGGRAVATING 
CIRCUMSTANCE WITH A 
SUBSTANTIAL IMPACT ON THE 
RATE OF CAPITAL SENTENCING 7

II. THE BALDUS STUDIES EMPLOYED 
EXCELLENT, PROFESSIONAL 
METHODS OF EMPIRICAL 
RESEARCH AND PRODUCED 
STRONG, RELIABLE FINDINGS ON 
THE ROLE OF RACE IN 
GEORGIA'S CAPITAL SENTENCING
S YSTEM.....................  18

C O N C L U S I O N .....................  2 6



TABLE OF AUTHORITIES
Cases Pages
Ballew v. Georgia, 435 U.S.

223 (1972)..................... iv, vi
Bazemore v. Friday, ___U.S.___,

___L.Ed. 2d___(1986).............  26
Hazelwood School District v. United

States, 433 U.S. 299 (1977). . . .  6
McCleskey v. Kemp, 753 F. 2d 877

(11th Cir. 1985) (en banc) . . .v,3,4,27
McCleskey v. Zant, 580 F. Supp.

388 (N.D. Ga. 1984)............. 9
Segar v. Smith,

738 F. 2d 1249 (D.C. Cir. 1984). 6
Teamsters v. United States,

431 U.S. 324. .................... 6
Vuyanich v. Republic National

Bank, 505 F.Supp.244 (N.D. Tex.
1980), vacated on other grounds,
723 F.2d 1195 (5th Cir. 1984). 26,27

Statutes
Former Ga. Code Ann.

§27-2534.1(6) ) (2)........... 9n.l

Other Authorities
Fisher, Multiple Regression

in Legal Proceedings, 80 Colum.
L Rev. 702 (1980)............... iii

a.



vi
H. Kalven & H. Zeisel, The American 

Jury (1966)...................
R. Lempert, An Invitation to Law 

and Social Science: Desert,
Disputes and Distribution (1986). iv

b.



No. 84-6811
IN THE

SUPREME COURT OF THE UNITED STATES 
October Term, 1985

WARREN McCLESKEY,
Petitioner,
- v.-

RALPH M. KEMP, Superintendent, 
Georgia Diagnostic & Classification 
Center.

On Writ of Certiorari To The United States 
Court of Appeals for the Eleventh Circuit

MOTION FOR LEAVE TO 
FILE BRIEF AMICI CURIAE

Dr. Franklin M. Fisher, Dr. Richard O. 
Lempert, Dr. Peter W. Sperlich, Dr. Marvin 
E. Wolfgang, Professor Hans Zeisel and 
Professor Franklin E. Zimring respectfully 
move, pursuant to Rule 36.3 of the Rules of 
the Court, for leave to file the attached



bri®f amici curiae in support of the 
petitioner in this case. The consent of 
counsel for the petitioner has been 
obtained. The consent of counsel for 
respondent was requested but refused, 
necessitating this motion.

This case involves one of the most 
carefully studied criminal justice 
questions ever to come before the Court. At 
issue is research by Professor David Baldus 
and his colleagues on the influence of 
racial factors in the capital sentencing 
system of the State of Georgia. The 
underlying constitutional and policy 
questions are of great national 
significance, and the value and 
significance of social science evidence has 
been a central issue in the case.

Amici believe they could be of aid to 
the Court in the evaluation of; (i) the 
significance of the racial disparities 
reported in the Baldus studies and (ii) the

ii



essential validity and soundness of these 
studies. The competence of amici to address 
these issues stems from their distinguished 
professional work in the areas of 
econometrics, statistics, research 
methodology and criminal justice issues.

Dr. Fisher is Professor of Economics at 
the Massachusetts Institute of Technology. 
He is one of the nation's most 
distinguished econometricians, having 
taught, written and consulted on a wide 
range of econometric and legal issues for 
over three decades. His article Multiple 
Regression in Legal Proceedings. 80 Colum. 
L. Rev. 702 (1980) has had a major 
influence on the judicial use of 
statistical methods. His research on 
sentencing guidelines and on the legal 
context of various economic issues has 
provided major empirical contributions to 
the fields of law and economics. He has 
served as a member of the National Academy

iii



of Sciences Panels on Deterrence and 
Incapacitation and on Sentencing Research.

Dr. Lempert is Professor of Law and 
Sociology at the University of Michigan. 
He has studied and written widely on a 
variety of legal and criminal justice 
issues, including capital punishment. He 
has served on the editorial boards of 
several distinguished professional journals 
including the Journal of Law and Human 
Behavior and Evaluation Review. Dr. Lempert 
has recently completed a term as the editor 
of Law & Society Review. His most recent 
book is An Invitation to Law and Social 
Science; Desert. Disputes and Distribution 
(1986). His work on jury size was cited by 
the Court in Ballew v. Georgia. 435 U.S. 
223 (1978).

Dr. Sperlich is Professor of Political 
Science at the University of California at 
Berkeley. Dr. Sperlich has taught, 
consulted and published widely on many

iv



criminal justice issues, including the role 
of juries and the use of scientific 
evidence in legal settings. His writings 
were cited prominently by the Court of 
Appeals in McCleskey v. Kemp.

Dr. Wolfgang is Professor of 
Criminology and Criminal Law and Director 
of the Sellin Center for Studies in 
Criminology and Criminal Law at the 
University of Pennsylvania. During his 
distinguished career, Dr. Wolfgang has made 
numerous contributions to the development 
of empirical research on legal issues. His 
pioneering study on the influence of racial
factors in the imposition of death
sentences for rape was the obj ect of
intensive legal examination during the
Maxwell v. Bishop litigation of the 1960s. 
He is a former president of the American 
Society of Criminology.

Professor Hans Zeisel is Emeritus 
Professor of Law and Sociology and

v



Associate of the Center for Criminal
Justice Studies at the University of 
Chicago. He is co-author of The American 
Jury, widely recognized as one of the most 
influential empirical studies of the legal 
system ever published. Professor Zeisel is 
a fellow of the American Statistical 
Association and the American Academy of 
Arts and Sciences. His empirical research 
on the functioning of juries was relied 
upon by this Court in Ballew v. Georgia. 
435 U.S. 223 (1978).

Professor Zimring is Professor of Law 
and Director of the Earl Warren Institute 
at Boalt Hall, University of California at 
Berkeley. He has written extensively on 
criminal justice issues, including juvenile 
crime and sentencing, the deterrent value 
of punishment, and the control of firearms. 
Professor Zimring served as Director of 
Research for the Task Force on Firearms of 
the National Commission on the Causes and

vi



Prevention of Violence, and has also served 
as consultant to many private and public 
organizations.

In view of their long-standing
professional interest in the legal use of
social scientific evidence and their
extraordinary professional competence to
address those issues, amici curiae believe
that their views might be of assistance to
the Court. They therefore urge the Court
to grant their motion and permit the
submission of this brief amici_curiae.
Dated: New York, New York

August 21, 1986
Respectfully submitted,
MICHAEL 0. FINKELSTEIN
MARTIN F. RICHMAN *Barrett Smith Schapiro 

Simon & Armstrong 
26 Broadway 
New York, N.Y. 10004 
(212) 422-8180

ATTORNEYS FOR AMICI CURIAE
gy • _____ ________ _______________ _

MARTIN F. RICHMAN 
♦Counsel of Record

vii



No. 84-6811
IN THE

SUPREME COURT OF THE UNITED STATES 
October Term, 1985

WARREN McCLESKEY,
Petitioner,

- v.-
RALPH M. KEMP, Superintendent, 
Georgia Diagnostic & Classification 
Center.

On Writ of Certiorari To The United States 
Court of Appeals for the Eleventh Circuit

BRIEF AMICI CURIAE OF DR. FRANKLIN 
M. FISHER, DR. RICHARD 0. LEMPERT, 
DR. PETER W. SPERLICH, DR. MARVIN 
E. WOLFGANG, PROFESSOR HANS ZEISEL 
& PROFESSOR FRANKLIN E. ZIMRING

SUMMARY OF ARGUMENT
The factual questions presented by this 

case are among those that can be tested by 
established social science methods. At 
issue was a series of decisions and actions



carried out in a single state over a 
limited period of time. The sources of 
information about those decisions were, in 
this case, official state files containing 
unusually rich and detailed data. The 
scientific techniques for the collection 
and analysis of such data are well- 
developed and highly reliable. 
Furthermore, the social science professions 
have developed criteria for the evaluation 
of such research. This is, in sum, the 
kind of research that social scientists 
know how to do, and it can be critically 
evaluated according to well-established 
standards with considerable confidence.

The studies at issue in this case were 
conducted in the State of Georgia by 
Professor David Baldus, George Woodworth 
and Charles Pulaski. The results of the 
Baldus studies are consistent with a solid 
body of previous research in this area? 
Georgia defendants whose victims are white,

2



especially black defendants, face death- 
sentencing rates many times higher than 
those whose victims are black. A natural 
question provoked by these findings is 
whether other legitimate sentencing factors 
might explain what initially appear to be 
racial differences. Yet these striking 
results did not disappear after searching
statistical analysis by Baldus and his
colleagues. Neither chance nor any
legitimate sentencing considerations can
explain the powerful influence of these
racial factors.

The Baldus studies were conducted in 
careful compliance with accepted research 
techniques. Their design and execution 
were meticulous and their analytical
methods are appropriate. They are among the 
best empirical studies on criminal
sentencing ever conducted, and their
results are entitled to a high degree of 
confidence.

3



The lower courts nevertheless displayed 
a profound and unwarranted mistrust of the 
Baldus studies and a misunderstanding of 
their results. The District Court judged 
the Baldus data sources by unrealistic and 
unjustified standards. It quarreled with 
data collection and coding methods that are 
well-established and widely used. It 
evinced a hostility towards methods of 
statistical analysis —  especially multiple 
regression analysis —  that is utterly 
unwarranted, expressing a skepticism toward 
techniques of statistical modeling, 
especially analyses conducted with 
parsimonious models, that is uninformed and 
indefensible. Finally, it faulted Baldus1s 
results on a variety of minor statistical 
grounds that reflect, at best, a partial 
understanding of the deficiencies that can 
afflict such research and a failure to 
appreciate the negligible extent to which 
those problems were likely to affect the

4



essential findings reported by Baldus. As 
a result of this series of errors, the 
District Court inappropriately devalued a 
first-rate body of research that sheds 
significant light on the issues before it.

The Court of Appeals, by contrast, 
purported to accept the validity of the 
Baldus studies and to address the legal 
implications of their results. Yet that 
court seriously underestimated the 
magnitude of the racial effects Baldus 
reported —  misconceiving both the actual 
size of the racial disparities and their 
relative significance as a force in Georgia 
sentencing decisions. Further, even while 
purportedly accepting the Baldus research, 
the Court of Appeals demanded a level of 
certainty that exceeds the powers of any 
statistical research to achieve —  a level 
of certainty not approached in most 
employment discrimination cases or in 
business litigation where such statistical

5



evidence is routinely received and often 
dispositive.

The Baldus results demonstrate that 
racial factors —  race of the defendant in 
white-victim cases and race of the victim 
throughout —  played a real, substantial 
and persistent role in death-sentencing 
decisions in the State of Georgia during 
the period studied. The State's evidence 
did not contradict these strong findings, 
which replicate less detailed, though 
similar conclusions reached in other 
studies. Whatever the legal implications 
of these facts, they should be accepted as 
proven to scientific satisfaction.

6



ARGUMENT
I

THE BALDUS STUDIES DEMONSTRATE THAT 
IN THE STATE OF GEORGIA, THE RACE 
OF THE HOMICIDE VICTIM HAS BEEN AN 
IMPLICIT AGGRAVATING CIRCUMSTANCE 
WITH A SUBSTANTIAL IMPACT ON THE 
RATE OF CAPITAL SENTENCING

The unadjusted results reported by 
Professor Baldus for the various 
combinations of race-of-defendant and race- 
of-victim in the State of Georgia are as 
follows:

Defendant / Defendants Percent
Victim Who Received

the Death 
Penalty_____

black/white 50 of 223 22white/white 58 of 748 8black/black 18 of 1443 1white/black 2 of 60 3
* * *
Totals bv 
Victim

* * * * * *

white victim 108 of 981 11black victim 20 of 1503 1

7



In particular, as the table shows, 
blacks who killed whites were sentenced to 
death at nearly 22 times the rate of blacks 
who killed blacks, and more than 7 times 
the rate of whites who killed blacks. The 
capital sentencing rate for all white- 
victim cases was almost 11 times the rate 
for all black-victim cases. Unless there 
is an extraordinarily perfect confounding 
with other factors correlated with race 
these very large racial disparities 
indicate that race is an implicit 
aggravating factor in the capital 
sentencing decision.

The first step in testing whether the 
disparities in capital sentencing rates 
were due to factors confounded with race, 
was to make cross-tabulations, based on the 
most important sentencing factors that 
might have been confounders. The racial 
disparities did not disappear. For example, 
by analyzing all cases that were death-

8



eligible under statutory aggravating factor 
(b)(2)1 —  murder by a defendant in the 
course of a contemporaneous felony, a 
category which included petitioner 
McCleskey's case —  Baldus found that 38 
percent (60 out of 160) of the blacks who 
murdered whites received the death penalty, 
while only 14 percent (15 out of 104) of 
the blacks who murdered blacks received 
this penalty. (See DB 87) Thus, blacks 
who murdered whites were sentenced to death 
at more than 2.5 times the rate of black- 
on-black cases in this category.

When Professor Baldus separated out 
only those, like McCleskey, whose 
contemporary felony was armed robbery, the 
disparities were even more pronounced: 34 
percent (42/123) of blacks who killed 
whites received a death sentence, while 
only 5 percent (3/57) of blacks who killed 
blacks did. (See DB 87) . These cross-

former Ga. Code Ann. §27-2534.1 (b) (2).
9



tabulations tell the basic story of the 
magnitudes of racial effects. Felony 
murders with white victims produce death 
sentences in Georgia more than twice as 
often as felony murders with black victims. 
This finding alone is strong evidence of 
racial impact.

Other data from these studies not only 
establish the fact of racial discrimination 
but tell us largely where it occurs. 
Cross-tabular results reveal noticeably 
different treatment of cases, by race, at 
various decision points from indictment 
forward. The following table, for example, 
addressing only Georgia cases in which a 
murder conviction had been obtained, 
reveals, by racial category, the rate at 
which Georgia prosecutors chose to advance 
cases to a capital sentencing hearing—  
where a death sentence was a possible 
outcome —  rather than permit an automatic 
life sentence.

10



Defendant / Victim
black / white .70 (87/124)
white / white .32 (99/312)
black / black .15 (38/250)
white / black .19 (4/21)

(DB 94) . Thus even among convicted
defendants, where strength of the evidence 
factors presumably no longer played a major 
role, Georgia prosecutors advanced black 
defendants to a penalty trial, if their 
victims were white, at nearly five times 
the rate they advanced black defendants 
whose victims were black (.70 vs. 15), and 
over three times the rate of whites who 
killed blacks (.70 vs. .19).

Because there were insufficient numbers 
of cases, Baldus could not use cross­
tabulations to control simultaneously for 
all combinations of possible confounding 
factors. This is a common problem in

11



social science research, and to deal with 
it, he resorted to multiple regression 
analysis, using both weighted least squares 
and logistic regressions. These are 
standard statistical methods for this type 
of analysis. Both forms of analysis showed 
substantial racial disparities in capital 
sentencing rates.

It is important to place the regression 
effort accurately in the context of the 
larger study. The regression exercise was 
intended principally to check the basic 
finding of the cross-tabular approach 
against a wide variety of possible 
alternative explanations. Multiple 
regression analysis permitted Baldus to 
take over 230 factors simultaneously into 
account to see whether any combination of 
them might be confounded with race. His 
basic finding was that white-victim cases 
remained more than twice as likely to 
produce death sentences when all these

12



other factors were accounted for. Among 
the regression results reported are many 
highly statistically significant regression 
coefficients for the race of the victim and 
the race of the defendant, employing 
statistical models of varying sizes. (See 
DB 83). These results indicate that racial 
factors have an independent influence on 
death-sentencing rates after the effects of 
all other legitimate sentencing variables 
included in the models have been taken into 
account.

In its discussion of the magnitude of 
the average race-of-victim effect in 
Georgia's capital sentencing system, the 
Court of Appeals focused almost exclusively 
on what it styled a "6%" disparity. This 
figure was presumably derived from the .06 
least squares regression coefficient 
estimated for the race-of-victim variable 
in the 230 - variable large scale multiple 
regression model in the Baldus studies.

13



(DB 83) . The court, confusing percent and 
percentage point, interpreted this "6%" 
average disparity to mean that "a white 
victim crime is 6% more likely to result in 
the [death] sentence than a comparable 
black victim crime.H McCleskev v. Kemp, 753 
F.2d 877, 896 (11th Cir. 1985) (en banc) . 
The assumption of the statement is that the 
death sentencing rate in white victim cases 
would on average be 6% higher than the rate 
for similarly situated black-victim cases. 
Thus, for example, if the death sentencing 
rate in a given class of black-victim cases 
were 10%, the white victim rate would be 6% 
higher or 10.6%.

Such an interpretation is incorrect and 
highly misleading. The .06 race of victim 
regression coefficient indicates that the 
average death-sentencing rate in the system 
is 6 percentage points higher in white- 
victim cases than it is in similarly 
situated black-victim cases. The

14



percentage increase in the rate is much 
greater than 6 percent at almost all levels 
of aggravation where the death penalty is 
given, because the base rates are low.

Having misunderstood the basic results 
of the Baldus studies, the lower courts,
not surprisingly, also misunderstood the
implications of those results for
McCleskey's case. To understand these
implications, one has to focus on the
disparity in sentencing rates disclosed by 
the studies for aggravation levels 
comparable to those in McCleskey's case. 
One can do this by looking at disparities 
in capital sentencing rates at the average 
aggravation levels for all white-victim 
cases (of which McCleskey' s is one) or, 
more precisely, at the cases in the mid­
range of aggravation (of which McCleskey's 
is also one). We examine both below.

The overall death-sentence rate in 
white-victim cases is 11%. Since the

15



weighted least squares regression model 
cited by the Court of Appeals tells us that 
the overall rate in comparably aggravated 
black-victim cases is six percentage points 
less, the rate in such cases is estimated 
at five percent. Thus, at the average level 
of aggravating circumstances represented by 
the white - victim cases, the rate of 
capital sentencing in a white - victim case 
is 120% greater than the rate in a black- 
victim case. Or to state the results 
differently: in six out of every 11 death 
penalty cases in which the victim was 
white, race of victim was a determining 
aggravating factor in the sense that the 
defendants would not have received the 
death penalty if the victims had been 
black.

The Court of Appeals properly points 
out that the race-of-victim effect is 
concentrated at the mid-range, where it is 
approximately 20 percentage points. In

16



that range, the average death sentencing 
rates (calculated from DB 90: col. D, 
levels 3-7) is 14.4% for black-victim cases 
and 34.4% for white-victim cases, an 
increase of 139%. This means that out of 
every 34 death-penalty cases in the mid­
range in which the victims were white, 20 
defendants would not have received a death 
penalty if their victims had been black.

McCleskey's case is, a white-victim 
death penalty case that is in the mid­
range in terms of aggravating 
circumstances. Since the statistical 
results show that in a majority of such 
cases the death penalty would not have been 
imposed if the victim were black, it is 
appropriate to conclude that in McCleskey's 
case (as in others of the same class) it is 
more likely than not that the victim's race 
was a determining aggravating factor in the 
imposition of the death penalty. This 
cannot be called a "marginal" difference.

17



II
THE BALDUS STUDIES EMPLOYED EXCELLENT, 
PROFESSIONAL METHODS OF EMPIRICAL 
RESEARCH AND PRODUCED STRONG, RELIABLE 
FINDINGS ON THE ROLE OF RACE IN 
GEORGIA'S CAPITAL SENTENCING SYSTEM

The District Court, as well as the 
Court of Appeals, appear to have rejected 
the Baldus studies in large measure because 
of their misapprehensions about the quality 
of the data gathered or the statistical 
methods employed to analyze that data. In 
our opinion, these reservations are 
unwarranted: the design of the research
followed accepted scientific practice, the 
research was carried out in a careful and 
thorough manner, the statistical methods 
employed were appropriate —  and the 
results, consequently, are reliable.

The District Court's opinion, in 
particular, raised a series of objections 
to empirical methods and procedures, almost 
none of which is well-founded. It asserts 
that Baldus's data base was "substantially

18



flaw[ed]," McCleskev v. Zant. 580 F. Supp. 
338, 360 (N.D. Ga. 1984) (emphasis 
omitted), because it "could not capture 
every nuance of every case." Id. at 356. 
None of Baldus's many models, even those 
with over 230 variables, was deemed 
sufficiently inclusive in the District 
Court's eyes, since they "have not 
accounted for . . . unaccounted-for 
factors." Id. at 362.

These objections are fundamentally 
misplaced. As a scientific matter, the 
likelihood that any omitted variable could 
significantly affect Baldus's robust racial 
findings —  especially when so many 
legitimate variables were taken into 
account —  is negligible. For any 
unaccounted for variables actually to make 
a difference in the Baldus findings, it 
would have to diminish a death sentencing 
rate in white-victim, felony murder cases 
more than double that of black-victim

19



cases. The 230 variables defined by 
Professor Baldus take into account every 
known, conceptually important legal and 
extra-legal factor that might affect 
capital sentencing. It is extremely 
unlikely that any factor that powerful has 
been overlooked in these studies. By 
insisting on a standard of "absolute 
knowledge" about every single case, the 
District Court implicitly rejected the 
value of all applied statistical analysis.

The District Court also expressed 
general skepticism toward a range of well- 
established social scientific methods 
employed by Baldus, including multiple 
regression analysis, which it found "ill 
suited to provide the court with 
circumstantial evidence of the presence of 
discrimination." Id. at 372 (emphasis 
omitted). Indeed the only statistical 
method that the District Court did seem to 
approve is the simple cross-tabular

20



approach, id. at 354, even though the Court 
acknowledged that the inherent nature of 
the problem under study here makes it 
•'impossible to get any statistically 
significant results in comparing exact 
cases using a cross tabulation method." Id. 
(emphasis omitted). This preference for 
cross-tabular methods lacks any scientific 
foundation. Baldus's use of multiple 
regression analysis is clearly valid and 
appropriate to his data. In any event, 
Baldus and his colleagues did use cross- 
tabular analysis extensively, and their 
findings using this method, as we have 
seen, are fully consistent with the 
regression results.

Finally, in evaluating Baldus's 
results, the District Court seized upon a 
somewhat confused welter of statistical 
issues, including Baldus's conventions for 
coding "unknown" data, id. at 357-59, the 
possible multicollinearity of his

21



variables, id. at 363-64, and the reported 
R2 of his model, id. at 351, 361, as 
reasons for its ultimate conclusion that 
Baldus's results cannot be relied upon. 
However, Baldus and his colleagues 
satisfactorily addressed each of these 
concerns and demonstrated that the racial 
results were not adversely affected by 
them. Baldus not only employed the correct 
method of treating "unknowns"; he also 
conducted alternative analyses to 
demonstrate that racial influences 
persisted irrespective of the method of 
treatment adopted. Multicoilinearity 
undoubtedly did affect some of the larger 
models employed by Baldus, but the District 
Court failed to realize that the presence 
of multicoilinearity would not change the 
estimate of the racial results reported. 
It would only affect the standard error of 
that estimate. Finally, the court's 
concern with the reported R2 of Baldus's

22



models is unfounded. Apart from the 
questionable relevance of the R2 measure 
for logistic models of the type he used, an 
R2 of .40 or higher is quite acceptable.

In sum, since the District Court's 
opinion was flawed by basic statistical 
errors and misunderstandings, its 
evaluation of the validity of the Baldus 
studies is simply off-target.

* *  *

The Court of Appeals took a different 
approach to Baldus's research: it 
announced that it would "assum[e] [the 
study's] validity and that it proves what 
it claims to prove," McCleskev v. Kemo. 
753 F.2d at 886, and would base its 
judgment solely on the legal consequences 
flowing from that research. Yet the 
skepticism which pervaded the District 
Court's analysis continued to dominate the 
treatment of Baldus's research by the Court 
of Appeals. After first knitting together

23



citations from several scholarly articles 
that caution courts against an unreflective 
use of social scientific evidence, id. at 
887-90, the Court announced "that 
generalized statistical studies are of 
little use in deciding whether a particular 
defendant has been unconstitutionally 
sentenced to death . . . [and] at most are 
probative of how much disparity is 
present." Id. at 893. That observation 
misses the point: Statistical evidence can 
determine with great reliability whether 
racial factors are playing a role in the 
sentencing system as a whole. Baldus's 
studies provide just such evidence.

When the Court turns to the Baldus 
studies, it relies primarily upon one 
summary figure drawn from the entire body 
of results —  a reported .06 disparity by 
race of victim in overall death- 
sentencing rates. As we showed above, this 
was but one of a number of important,

24



meaningful results indicating a consistent 
racial presence in the state of Georgia's 
capital sentencing system. More important, 
as also demonstrated earlier, the Court of 
Appeals seemed fundamentally to have 
misunderstood the magnitude and 
significance even of this single result 
upon which it focused: it took a pound for 
a penny.

Although Baldus and his colleagues have 
been consistently conservative in 
evaluating and reporting their findings, 
the adjusted influence of racial factors on 
Georgia's capital sentencing system remains 
both clear and significant. Race, 
especially the race of the victim, plays a 
large and recognizable part in determining 
who among Georgia defendants convicted of 
murder will be sentenced to life and who 
among them will be sentenced to death.

25



CONCLUSION
The contributions of social scientific 

evidence to the resolution of legal issues 
has increased significantly in recent 
decades, as statistical methods have 
improved and the confidence of the courts 
has grown. This Court has led the lower 
federal courts toward an appreciation of 
the nature of statistical evidence, and has 
developed legal principles —  including 
standards of proof for parties presenting 
such evidence —  that reflect a clear 
understanding of the powerful utility of 
valid social scientific evidence. See.
e.q. . Bazemore v. Friday. ___U.S.___ ,
___L.Ed.2d___,(1986); Hazelwood School
District v. United States. 433 U.S. 299 
(1977) ; Teamsters v. United States. 431 
U.S. 324 (1977) ; see also Seqar v. Smith.
738 F.2d 1249 (D.C. Cir. 1984); Vuvanich v. 
Republic National Bank. 505 F. Supp. 244 
(N.D. Tex. 1980), vacated on other grounds.

26



723 F.2d 1195 (5th Cir. 1984).
The Court of Appeals disregarded these

basic standards of proof fashioned by the
Court. Its opinion in McCleskev insists
upon a level of methodological purity in
data quality, model design, and analyse.
that can be achieved only in theory. T h e
cross-tabular and regression analyses of
Professor Baldus and his colleagues were
the correct analytical tools for the
research they undertook. Their studies
were undertaken with great care. Their
findings replicate the work of earlier,
less thorough scholars. The magnitude of
their findings is striking. This body of
research renders it far more likely than
not that racial factors have played a
significant role in Georgia's capital
sentencing system in the post-Furman era.
Dated: New York, New York

August 21, 1986
Respectfully submitted,

27



MICHAEL 0. FINKELSTEIN 
MARTIN F. RICHMAN *

Barrett Smith Schapiro 
Simon & Armstrong 

2 6 Broadway 
New York, N.Y. 10004 
(212) 422-8180

ATTORNEYS FOR AMICI CURIAE
BY:_________________________

MARTIN F. RICHMAN
*Counsel of Record

28



CERTIFICATE OF SERVICE

I hereby certify that I am a member of
the bar of this Court, and that I served
the annexed Motion for Leave to File Brief
Amici Curiae and Brief Amici Curiae on the
parties by placing copies in the United
States mail, first class mail, postage
prepaid, addressed as follows:

John Charles Boger, Inc.
NAACP Legal Defense Fund
99 Hudson Street
New York, New York 10013
Mary Beth Westmoreland, Esq.
132 State Judicial Building 
40 Capitol Square, S.W.
Atlanta, Georgia 30334

Done this ____  day of August, 1986.

MARTIN F. RICHMAN

29



Hamilton Graphics, Inc.— 200 Hudson Street, New York, N.Y.— (212) 966-4177