Trial Proceedings Transcript Vol. 4
Public Court Documents
August 11, 1983
207 pages
Cite this item
-
Case Files, McCleskey Legal Records. Trial Proceedings Transcript Vol. 4, 1983. 914714fe-5aa7-ef11-8a69-7c1e5266b018. LDF Archives, Thurgood Marshall Institute. https://ldfrecollection.org/archives/archives-search/archives-item/e6cc6055-b6cb-40d7-9606-4f48a8795933/trial-proceedings-transcript-vol-4. Accessed July 02, 2025.
Copied!
640
kr
IN THE UNITED STATES DISTRICT COURT
3 FOR THE NORTHERN DISTRICT OF GEORGIA
4 ATLANTA DIVISION
= i?
w & WARREN MCCLESKEY ) DOCKET NO. C31-2434A
)
7 }
PETITIONER, } ATLANTA. GEORGIA
a )
lf ) AUGUST 11, 1983
? )
WALTER D. ZANT, WARDEN. )
10 )
)
11 RESPONDENT. )
12 lw.
13 VOLUME IV
14 TRANSCRIPT OF PROCEEDINGS
13 BEFORE THE HONORABLE J. OWEN FORRESTER, UNITED STATES DISTRICT
ié6 JUDGE. 18 APPEARANCES OF COUNSEL:
|
: Ww 19 FOR THE PLAINTIFF: JOHN CHARLES BOGCER3; TIMOTHY K. FORD
AND ROBERT H. STROUF.
21 FOR THE DEFENDANT: MARY BETH WESTMORELAND AND SUSAN V,
BOLEYN.
22
23
|
JIM PUGH
24 OFFICIAL COURT REPORTER
ROOM 2347. 75 SPRING STREET. S5.W.
25 ATLANTA, GEORGIA 20303
13
a
Wl TNESSES
WITNESSES FOR THE PLAINTIFF:
BALDUS, DAVID C.
WITNESSES FOR THE DEFENSE:
DIRECT CROSS
643
REDIRECT
641
RECROSS
Ta
y
ry
O
B
d
e
A
d
g
p
b
—— p—— f—— ——————— —
DB-3%
DB~&60
DB-461
DB-é4
DB-&3
np-67
DB-48
DB-12
DB-62
DB-63
DB-49
DB-70
DB-71
DB-72
DB-73
DB-74
DB-73
DB=76
DB-77
DB-78
DB-7%
DB-80
DB~-21
DB-32
DB-83
DB-84
DB-83
DB-36
DOCUMENTARY EVIDENCE
MARKED
633
639
L622
681
&97
701
709
725
73
732
733
733
741
735
7460
767
768
774
780
782
788
800
811
316
819
832
839
843
&542
RECEIVED
658
&60
b64
627
701
716
725
729
732
733
737
739
735
760
7463
768
773
782
7387
7388
310
819
819
231
838
842
543
(ATLANTA, FULTON COUNTY, GEORGIA: AUGUST 11, 1983,
IN OPEN COURT.)
THE COURT: BEFORE YOU START. MR. BOGER. LET ME A3K
PROFESSOR BALDUS A FEW QUESTIONS.
ON YOUR WEIGHTING, --—
MR. FORD: YOUR HONOR, COULD I POSSIBLY, I DON’T MEAN
TO INTERRUPT THE COURT, BUT I DO HAVE JUST ONE MATTER I WOULD
LIKE TO PERHAPS, IF I COULD. INTERJECT FOR ONE MOMENT BEFORE
PROFESSOR BALDUS BEGINS.
IT DEALS WITH A MATTER THAT CAME UP YESTERDAY. THERE
MUST BE SOME CONFUSION.
COULD I HAVE JUST A MOMENT?
THE COURT: I DON’T KNOW WHAT IT IS.
G0 AHEAD.
MR. FORD: IT HAS TO DO, YOUR HONOR, WITH THE QUESTION
THAT AROSE YESTERDAY WITH REGARD TO ONE OF THE QUESTIONNAIRES.
1 BELIEVE IT WAS NUMBER EG-SA, WHICH APPEARED TO HAVE BEEN CODED
AT THE BOARD OF PARDONS, BUT A DEATH SENTENCE WAS CODED AS
HAVING BEEN IMPOSED. AND THERE WAS SOME QUESTION, AND OF COURSE,
MR. GATES DIDN“T HAVE THE KNOWLEDGE OF WHAT OCCURRED. AND I
THINK YOUR HONOR RECALLS ASKING MRS. WESTMORELAND IF THERE WAS A
POSSIBLE STIPULATION, AND I INFORMED THE COURT OF MY
RECOLLECTION. AND I DO HAVE AND AM HANDING TO THE COURT A COPY
FOR THE COURT AND A COPY FOR THE CLERK OF, A XEROX COPY OF A
H44
1 CERTIFIED COPY OF A TRANSCRIPT, AND AN ORDER IN THE. IN A CASE
2 IN THE SUPERIOR COURT OF CHATHAM COUNTY, STATE OF GEORGIA,
3 VERSUS JOHN ARTHUR BROWN. DECEMBER 4, 1778.
THE DOCUMENT IS A S~PAGE, I BELIEVE, HEARING IN WHICH
THE, AT PAGE 3, THE JUDGE INDICATES THAT MR. BROWN HAD BEEN
SENTENCED TO DEATH, AND AFTER SOME DISCUSSION WITH REGARD TO THE
4
S
é&
¥ § WISHES OF THE PROSECUTION AND THE, 1 BELIEVE THE FAMILY OF THE
38 VICTIM, THE DEFENDANT IS RESENTENCED TO LIFE IMPRISONMENT.
9 ATTACHED ALSO TO THE DOCUMENT IS A COPY OF THE ORDER
) REDUCING THE SENTENCE OR IMPOSING A LIFE SENTENCE AT THE CLOSE
11 OF THAT HEARING.
12 AND MS. WESTMORELAND WEST INDICATED TO ME SHE HAD
13 CHECKED ON THIS CASE AS WELL, APPARENTLY YESTERDAY, AND I ASKED
14 IF SHE WOULD STIPULATE TO THE ADMISSION OF THIS. JUST TO CLARIFY
1% THE RECORD AS TO DOCUMENT EG-7.
16 MS. WESTMORELAND: YOUR HONOR, MY ONLY COMMENT AT THIS
17 POINT, I DON’T SEE THE NEED FOR INTRODUCING THIS DOCUMENT. MY
13 UNDERSTANDING AS WE WERE GOING THROUGH THE QUESTIONNAIRES
» 19 YESTERDAY, THEY WERE BEING UTILIZED AS EXAMPLES. AND THAT WAS
20 THE WHOLE PURPOSE OF THE QUESTIONING YESTERDAY, WAS TO EXAMINE
21 THE QUESTIONNAIRES THEMSELVES. SO I DON’T SEE ANY NEED IN
22 INTRODUCING THIS PARTICULAR DOCUMENT. AT THIS TIME, I DON'T SEE
23 ITS RELEVANCE.
24 MR. FORD: PERHAPS, YOUR HONOR, IT’S JUST AN EXAMPLE OF
25 THE KIND OF THINGS THAT COULD HAVE OCCURRED THAT COULD RESULT IN
As Seri rn
645
BALDUS - DIRECT
THE CASES BEING CODED IN THE MANNER THAT THAT CASE APPEARED TO
HAVE BEEN.
IM NOT JUST TRYING TO DO ANYTHING BUT CLEAR UP THE
CONFUSION, AND MAKE SURE THE COURT IS FULLY INFORMED TO THE
EXTENT WE CAN.
THE COURT: I DON’T KNOW WHETHER THE CONFUSION IS
CLEARED UP OR NOT. BUT AT LEAST WE NOW KNOW MORE ABOUT THE CASE
THAN WE DID.
ILL ADMIT IT.
MR. FORD: THANK YOU, YOUR HONOR.
DAVID C. BALDUS,
BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
TESTIFIED FURTHER AS FOLLOWS:
THE COURT: IN YOUR WEIGHTING. DOCTOR BOWERS, --
THE WITNESS: IF YOUR HONOR MAY INDULGE ME. BALDUS IS
THE LAST NAME. I KNOW I HAVE A DISTINGUISHED FORERUNNER. --
THE COURT: YOU WOULDN‘T WANT TO BE DISTINGUISHED IN
THE WAY HE WAS, SIR. I“M SORRY.
| LET-S SEE, ITS DB-5é.
GOING BACK TO THE ATLANTA JUDICIAL CIRCUIT. IF I
UNDERSTAND THIS CORRECTLY. YOUR SAMPLE HAD ALL OF THE DEATH
CASES FROM THE ATLANTA JUDICIAL CIRCUIT. WHICH WERE TEN.
THE WITNESS: YES.
THE COURT: SO THAT REQUIRED NO RE-WEIGHTING -—-
nN
4]
a
0
.
N
O
646
BALDUS - DIRECT
THE WITNESS: THAT”S CORRECT.
THE COURT: ~-— OF THAT.
THEN BECAUSE YOU HAD ONLY 142 OF 1462, I UNDERSTAND THAT
AS TO THE RESULTS IN THOSE FORTY-TWO CASES. YOU WOULD HAVE
APPLIED, IF YOU WILL, A MULTIPLE OF 3.85 OR 3.86 TO GET THOSE,
THAT NUMBER OF CASES INTO A PROPER PROPORTION WITH THE ACTUAL?
THE WITNESS: THAT’S CORRECT.
THE COURT: AND THE SAME WOULD BE TRUE FOR WHATEVER IT
IS. THE HOMICIDE.
THE WITNESS: VOLUNTARY MANSLAUGHTER. YOUR HONOR.
THE COURT: ALL RIGHT. THEN TURNING TQ DB-37.
THE WITNESS: YES.
"THE COURT: WHERE YOU HAVE GIVEN THE RAW NUMBERS.
I UNDERSTAND YOUR TESTIMONY TO BE THAT THOSE RAW
NUMBERS IN EVERY INSTANCE ARE ESTIMATES OF THE RAW NUMBERS
INSTEAD OF ACTUAL, IS THAT CORRECT?
THE WITNESS: YES, WITH THE EXCEPTION OF THE DEATH
SENTENCES, YOUR HONOR. THE DEATH SENTENCES INCLUDE THE WHOLE
UNIVERSE.
THE PURPOSE OF THIS TABLE IS TO MAKE ESTIMATES OF WHAT
IS IN THE UNIVERSE OF THE STUDY, WHICH IS THE, ALL THE CASES
THAT WERE LISTED IN THE BOARD OF PARDONS AND PAROLES.
THE COURT: AS YOU KNEW WHAT WAS IN THE UNIVERSE. WHY
DID YOU FEEL IT WAS NECESSARY TO MAKE ESTIMATES?
THE WITNESS: WELL, THE IDEA IS TO MAKE ESTIMATES AS TO
Pa
y
v
o
.
~
0
4
Bb
»
O
N
10
il
647
BALDUS - DIRECT
HOW MANY WERE CONVICTED AND DISPOSED OF IN VARIOUS WAYS DURING
THE RESPECTIVE YEARS. WE WANT TO TRY AND MAKE AN ESTIMATE OF
HOW ALL THESE CASES WERE TREATED OVER A PERIOD OF TIME.
THE SAMPLE WOULD NOT GIVE YOU A TRUE PICTURE IN AND OF
ITSELF OF THE NUMBERS OF CASES.
THE COURT: 1 UNDERSTAND THAT. BUT MY QUESTION IS, IF
YOU’VE GOT THE RAW DATA, WHY DID YOU ESTIMATE IT?
THE WITNESS: WE HAD TO ESTIMATE THE NUMBER OF CASES
THAT RESULTED IN A, OH. I SEE, WITH RESPECT TO,» WITH RESPECT TO
THE OUTCOME, YOU MEAN? YES.
WELL. I CAN’T ANSWER THAT TO A CERTAINTY, BUT I BELIEVE
THAT THE RESULTS WOULD BE THE SAME, IF WE HAD GONE BACK AND
ACTUALLY COUNTED THAT.
WHAT WE WERE INTERESTED IN GENERALLY WERE THE
CHARACTERISTICS OF THE CASES THAT COULD NOT BE ESTIMATED FROM
THE ACTUAL UNIVERSE OF THE CASES. THAT WAS THE JUSTIFICATION
GENERALLY FOR USING THAT PROCEDURE.
BUT I SEE YOUR POINT. THOSE CALCULATIONS COULD HAVE
BEEN MADE USING THE SAME METHOD, BUT 1 BELIEVE THE RESULTS WOULD
BE THE SAME. YOUR HONOR. GIVEN THE NATURE OF THE WEIGHTING
PROCESS THAT WAS USED.
THE COURT: ALL RIGHT. WE WERE ON DB-58, I BELIEVE,
WHEN WE STOPPED YESTERDAY. AND MS. WESTMORELAND, IF I“M RIGHT
ABOUT THE EXHIBIT, MS. WESTMORELAND STATED SHE HAD SOME PROBLEMS
WITH IT. IS THAT THE RIGHT EXHIBIT? IS THAT WHERE WE WERE?
r——— — ————— ——. —— —— neg A——, S———.. p—— (—— ——
648
BALDUS - DIRECT
fl
e MR. BOGER: THAT’S CORRECT, YOUR HONOR.
MS. WESTMORELAND: YES, YOUR HONOR.
THE COURT: THE PROBLEM WITH THAT EXHIBIT AND HAVING
GIVEN SOME MORE THOUGHT TO IT OVERNIGHT, WOULD YOU LIKE TO STATE
YOUR POSITION? ‘
MS. WESTMORELAND: YES. YOUR HONOR.
IN THINKING OVERNIGHT AND ALSO IN READING THE CASE
REFERRED TO. OR AT LEAST THE PERTINENT PORTION OF THE CASE
O
W
T
N
E
Ne
Y
N
REFERRED TO BY YOUR HONOR YESTERDAY AFTERNOON, I BELIEVE THE
10 CONSTRUCTION AGGREGATE TRANSPORT, INCORPORATED. IS THE CASE,
11 WAS THE ONE --
12 THE COURT: I MAY HAVE OVERSTATED IT A BIT. BUT AS I
13 TOLD MR. STROUP TO TELL YOU, I WAS TRYING TO SMOKE YOU OUT ON
14 WHETHER OR NOT YOU WERE GOING TO CONTEST THE UNDERLYING
13 VALIDITY. THE VALIDITY OF THE UNDERLYING DATA OR NOT.
16 MS. WESTMORELAND: THAT'S WHERE I WAS GETTING AT THIS
17 POINT. YOUR HONOR.
18 THE COURT: OKAY.
% 19 : MS. WESTMORELAND: IF I MIGHT JUST MAKE A BRIEF
20 STATEMENT TO THE COURT AND I THINK MAYBE THIS WILL CLARIFY
21 THINGS SOMEWHAT.
22 I UNDERSTAND THE COURT’S NOT BEEN PRIVY TO ALL OF THE
23 DEPOSITIONS AND CRITICISMS THAT HAVE BEEN MADE.
24 WERE NOT IN POSITION AT THIS POINT TO CHALLENGE THE
25 UNDERLYING DATA SOURCES. THAT IS, THE INFORMATION FROM THE
L429
BALDUS - DIRECT
PARDONS AND PAROLES BOARD AND GEORGIA SUPREME COURT. NOT
Po
e
NECESSARILY CONCEDING IT’S ACCURATE BUT IT’S NOT A CHALLENGE
THAT WERE MAKING AT THIS TIME. WE ARE MAKING CHALLENGES. WILL
BE MAKING CHALLENGES THROUGH COUR EXPERTS AS TO CERTAIN THINGS
THAT WE THINK WOULD MAKE THE DATA. REFLECT ON THE CREDIBILITY
AND RELIABILITY OF THE DATA. 1 THINK THAT IS SOME OF THE THINGS
THAT WERE MENTIONED DURING THE QUESTIONING YESTERDAY IN REGARD
TO. FOR INSTANCE, THE QUESTIONNAIRE DESIGN.
OTHER THINGS WILL COME UP LATER WITH RELATION TO THE
—-
:
oo
RU
SE
WE
RE
g
o
.
Fe
e
OE
E
CR
S
E
Fa
d
NUMBER OF UNKNOWNS AND THAT TYPE OF THING.
iX YES. IN THAT RECORD, WE ARE INTENDING TO CHALLENGE THE
1z RELIABILITY OF THE DATA, BUT NOT THE UNDERLYING DATA SOURCES.
13 THE COURT: WELL. I UNDERSTAND WHAT YOU SAID ABOUT THE
14 UNDERLYING DATA SOURCES.
15 WHAT I AM TRYING TO FIND OUT FROM YOU IS WHETHER OR NOT
16 YOU CHALLENGE WHETHER THE INFORMATION ON THIS TAPE AS CLEANED
17 FAIRLY REPRESENTS WHAT IS AVAILABLE IN THE DATA SOURCES RELIED
18 UFQON.
w 19 MS. WESTMORELAND: PHRASED THAT WAY, IT’S SOMEWHAT OF A
20 DIFFICULT QUESTION TO ANSWER. YOUR HONOR.
21 THE COURT: I THINK FROM WHAT HAS COME OUT IN MR.
22 GATES” TESTIMONY. AND FROM WHAT IS OBVIOUS FROM A FEW THINGS
23 THAT PROFESSOR BALDUS HAS TESTIFIED ABOUT, IT IS OBVIOUSLY NOT
24 PERFECT, BUT I WANT TQ FIND OUT IF YOU THINK IT IS SUBSTANTIALLY
25 INVALID 30 AS TO BE UNTRUSTWORTHY, 350 AS TO MAKE ANY CONCLUSIONS
630
BALDUS - DIRECT
Py
DRAWN FROM THE DATA UNTRUSTWORTHY. ASSUMING THAT THE RIGHT SORT
OF ANALYSIS WAS USED TO DRAW THE CONCLUSIONS.
MS. WESTMORELAND: YOUR HONOR. AND I GUESS, ONCE AGAIN,
ITS HARD TO ANSWER YES OR NO TO THAT QUESTION. I“M NOT TRYING
TO AVOID THE QUESTION OF THE COURT, IM TRYING TO PHRASE IN MY
MIND EXACTLY WHAT. HOW OUR CRITICISMS RELATE TO THE DATA
THEMSELVES.
AS TO THE FACT THAT THAT TAPE REFLECTS WHAT WAS CODED
vv
©
N
N
oc
UO
»
O
N
ON THE QUESTIONNAIRES, THAT’S NOT THE CHALLENGE THAT WE’RE
rr
oO
MAKING.
31 THAT’S NOT THE QUESTION. THE QUESTION I THINK THAT
12 WERE RAISING IN THIS PARTICULAR REGARD IS THAT, OR A PORTION OF
13 IT AT ANY RATE, IS THAT A COMBINATION OF FACTORS. INCLUDING THE
14 DATA GATHERING METHODS AND THINGS THAT WERE SUBSEQUENTLY DONE
15 WITH THE DATA, INCLUDING SOME, FOR INSTANCE, THE RECODES THAT
146 WERE UTILIZED LATER ON, CERTAIN THINGS THAT WERE THEN DONE WITH
17 THE DATA THEN DO MAKE IT UNRELIABLE FOR PURPOSES OF FURTHER
1& STATISTICAL ANALYSES, AT LEAST FOR THE TYPE OF ANALYSES THAT
% 19 WERE DONE.
20 : AND THE I THINK IT IS PERHAPS A CUMULATIVE THING, AND
21 WHAT OUR POSITION WOULD BE IS THAT THESE VARIOUS THINGS ADD UP,
22. RATHER THAN SPECIFICALLY POINTING AND SAYING THIS TAPE AS IT
23 EXISTS IS UNRELIABLE. I THINK IT IS NECESSARY TO LOOK AT NOT
24 ONLY THAT, BUT THINGS THAT HAVE BEEN DONE WITH IT AND HOW IT WAS
25 UTILIZED.
631
BALDUS - DIRECT
fi
b THE COURT: WELL. LET ME TALK OUT LOUD A LITTLE BIT,
2 NOT TRYING TO INDICATE HOW I VIEW IT, BUT REALLY TRYING TO TAKE
3 YOUR POSITION AS AN ADVOCATE.
4 IT IS CLEAR THAT THE CODERS HAD TO USE A LOT OF
S JUDGMENTS. .
® & IT IS ARGUABLE THAT THEY WERE NOT PREPARED BY VIRTUE OF
7 EXPERIENCE TO MAKE THE KINDS OF SUBTLE JUDGMENTS THAT ARE
3 NECESSARY.
? THERE ARE A LOT OF INFERENCES BUILT INTO THE STUDY THAT
10 ARE GUESTIONABLE. AND INTO THE PROCESS BY WHICH CODING DECISIONS
11 WERE MADE ARE QUESTIONABLE.
12 AND OF COURSE. THERE IS ALWAYS THE POSSIBILITY THAT THE
13 CODER MADE MISTAKES. AND, IN FACT, OUT OF THE FEW THAT WE HAVE
14 SEEN, WE HAVE SEEN MISTAKES BEING MADE.
13 NOW THAT. BASICALLY, IS A CHALLENGE TO THE DATA
16 GATHERED ON THE QUESTIONNAIRE.
17 I DON‘T KNOW WHAT PROBLEMS YOU HAVE WITH WHAT HAPPENED
18 TO IT THEREAFTER. I DON’T IMMEDIATELY SEE ANY PARTICULAR
% 19 PROBLEMS WITH THE INFORMATION GATHERING THAT HE TESTIFIED ABOUT
20 YESTERDAY WITH THE D.A.“S, THE DEFENSE ATTORNEYS, THE VITAL
21 STATISTICS BUREAU, THAT SORT OF THING. THEN THIS MASSAGING
22 PROCESS BEGINS, AND THEN ONE THING THAT DOES OCCUR THAT I CAN
23 UNDERSTAND YOU MIGHT WISH TO CHALLENGE WOULD BE THAT THAT WHICH
24 WAS TENUOUS AT BEST UNDER A CODING OF "2", RATHER THAN "1" ALL
235 OF A SUDDEN BECOMES EITHER AN ABSOLUTE YES OR NO. AND I
6352
BALDUS - DIRECT
1 UNDERSTAND YOU CRITICIZE THAT.
[3
MS. WESTMORELAND: YES, YOUR HONOR. THAT’S CORRECT.
3 THE COURT: WHAT I“M TRYING TO SAY. HAVE I HIT ON ANY
4 OF THE THINGS THAT ARE BOTHERING YOU OR IS SOMETHING ELSE
4]
BOTHERING YOU?
* MS. WESTMORELAND: THOSE ARE SOME OF THE PRIMARY
CONCERNS THAT WE HAVE. YOUR HONOR.
ANOTHER, PERHAPS ANOTHER EXAMPLE WOULD BE THE
0
N
D
SUBSEQUENT RECODING. ALONG THE SAME LINES OF WHAT YOU WERE
10 INDICATING WITH A 1 OR 2 THAT WAS THEN CODED TO BE AN ABSOLUTE
11 VALUE. I THINK CERTAIN CIRCUMSTANCES, PERHAPS UNKNOWN. WERE
12 RECODED, AND RATHER THAN BEING REFLECTED AS UNKNOWN,
13 REFLECTING IT AS NOT EXISTING AT ALL. IT’S THE SAME TYPE OF
14 THING, AND THOSE ARE NOT ALL OF OUR CRITICISMS. WE HAVE
15 CRITICISMS AS TO THE METHODOLOGY AS WELL.
16 BUT I THINK THOSE WOULD BE THE SUBSTANTIAL CRITICISMS
17 AS TO THE DATA ITSELF. YES.
18 THE COURT: WELL. CAN YOU QUANTIFY -- ARE YOU IN A
% 19 POSITION, EITHER BY EXAMPLE OR BY ANY OTHER MEANS TO
20 DEMONSTRATE THAT THE DATA ON THAT TAFE DOES NOT SUBSTANTIALLY
21 REFLECT REALITY BEYOND SUGGESTING THE POSSIBILITY IT MAY NOT?
22 MS. WESTMORELAND: INSOFAR AS BEING ABLE TO SHOW ACTUAL
23 CASES, AND WHAT THE CASES, HOW THEY SHOULD HAVE BEEN
24 DONE, WERE NOT IN POSITION TO DO THAT, IF THAT’S THE QUESTION.
25 THE COURT: WELL, THAT WOULD BE ONE WAY TO DO IT.
SA ——_———_ A ——— ap en mpmpr— mn ——
BALDUS - DIRECT
1 ANOTHER WAY TO DO IT WOULD BE SOME SORT OF A DETAILED
ANALYSIS OF THE KINDS OF DECISIONS THAT WERE BEING MADE IN THE N
DATA MASSAGING PROCESS AND AN EFFORT TO QUANTIFY THE RANGE OF
ERROR THAT COULD BE INTRODUCED USING THOSE SORTS OF TECHNIGUES.
ARE YOU GOING TO DO THAT?
MS. WESTMORELAND: TO THE EXTENT THAT WE HAVE ALREADY
INDICATED, THE SAME TYPES OF PROBLEMS THAT HAVE ALREADY BEEN
INDICATED, YOUR HONOR, 1 THINK THOSE ARE GOING TO BE OUR
FOCUSES, AND, YES, WE WILL FOCUS ON SHOWING THAT THOSE DIFFERENT
w
a
l
e
R
E
H
E
C
a
l
G
R
S
t
e
S
E
TW
PROBLEMS DID HAVE A SUBSTANTIAL EFFECT. THAT IS ONE OF THE
fo
e
an
d FOCUSES WE INTEND TO MAKE.
12 STILL, I’M NOT SURE IF I“M ANSWERING THE COURT-S
13 RUEST ION, I‘M TRYING.
14 THE COURT: I FEEL A LITTLE BIT WITH YOU LIKE I DID
15 WITH MR. BOGER. I TOLD YOU ALL YESTERDAY I DIDN‘T KNOW WHAT THE
16 APT ANALOGY WAS. BUT IT MAY BE LIKE AN ISRAELITE HEARING JOHN
17 THE BAPTIST DESCRIBE THE COMING OF THE MESSIAH. I“M JUST REAL
13 ANXIOUS TO SEE THE MESSIAH THAT BOTH OF YOU ARE SUPPOSED TO
® 19 PRESENT, BUT YOU WON‘T EVEN LET ME SEE EVEN A LITTLE PIECE OF |
|
20 HIM.
21 WELL, UNDER THOSE VERY GENERAL STATEMENTS I“LL HAVE TO
22 CONDITIONALLY ADMIT THE DOCUMENTS ON THE BASIS THAT THE ONLY
23 CHALLENGE YOU HAVE IS THAT THE UNDERLYING DATA IS UNRELIABLE. I AM
24 NOT IN POSITION TO MAKE THAT DETERMINATION AT THE MINUTE, AND
23 HAVE TO CONDITIONALLY ADMIT THEM SUBJECT TO A MOTION TO STRIKE
— ——— — i p— —— a A r— ———— op —— p—
BALDUS - DIRECT
pod
e IF YOU DEMONSTRATE THEY RE UNRELIABLE.
MS. WESTMORELAND: THAT WOULD BE ALL RIGHT, YOUR HONOR.
YOUR HONOR. MAY I ASK AT THIS TIME. RATHER THAN ——
COULD I HAVE A CONTINUING OBJECTION ALONG THOSE LINES, RATHER
THAN MAKING AN OBJECTION EACH DOCUMENT AT THE TIME IT MIGHT BE
PROFFERED, OR DO YOu PREFER THAT I MAKE AN OBJECTION EACH TIME?
THE COURT: TO THE EXTENT THAT YOUR OBJECTION IS AS
GENERALIZED AS THE ONE YOU HAVE MADE THIS MORNING, THAT WOULD BE
$
0
N
o
W
N
RUITE AGREEABLE.
Py
<Q
TO THE EXTENT THAT YOU HAVE ANY OTHER REASON FOR
Pr
y
-
DOUBTING THE RELIABILITY OF AN EXHIBIT OR TESTIMONY, I WOULD
12 WANT YOU TO STATE THAT THEN AND THERE.
13 MS. WESTMORELAND: THAT WOULD BE FINE. YOUR HONOR. AS
14 LONG AS FOR THE GENERAL PURPOSE. I WOULD LIKE THAT TO BE A
13 CONTINUING OBJECTION.
16 THE COURT: ALL RIGHT.
17 MS. WESTMORELAND: THANK YOU.
13 THE COURT: GO AHEAD.
* 19 MR. BOGER: FINE, YOUR HONOR. AND I HOPE TODAY IS THE
20 DAY, AND THE HOUR IS THE HOUR.
21 DIRECT EXAMINATION (CONTINUED)
22 BY MR. BOGER:
23 |@. PROFESSOR BALDUS., YESTERDAY YOU WERE TALKING ABOUT HOW YOU
24 HAD BEGUN TO ANALYZE THE DATA. YOU SAID, I BELIEVE. THAT YOU
J
iA
WANTED FIRST TO SEE A PICTURE OF HOW THE CASES HAD FLOWED
6355
BALDUS - DIRECT
THROUGH THE SYSTEM, AND WE LOOKED AT THAT IN DB-37.
AND IN DB-58 WE LOOK AT THE INCREASING RISK FOR A DEATH
SENTENCE AT DIFFERENT STAGES IN THE PROCESS.
WAS THERE ANY OTHER GENERAL MEANS BY WHICH YOU GOT A
PICTURE AT THE OUTSET THAT HELPED YOU FORM YOUR DIFFERENT
ANALYSES OF WHAT THE RISK OF RECEIVING A DEATH SENTENCE WAS?
A. YES. WE USED ONE OTHER METHOD. AND THAT WAS TO QUANTIFY THE
RISK THAT OFFENDERS RAN IF THEY MOVED THROUGH THE VARIOUS STAGES
OF THE PROCESS.
@. LET ME DIRECT YOUR ATTENTION. THEN. TO DB-59 MARKED FOR
IDENTIFICATION. CAN YOU IDENTIFY THAT DOCUMENT?
A. YES. THIS IS AN EXTENSION OF DB-58. AND WHAT IT DOES VERY
SIMPLY IS LOOK AT THE CASES THAT REMAIN IN THE SYSTEM AT EACH OF
THE SUCCESSIVE SIX STAGES THAT WE HAVE IDENTIFIED AND CALCULATE
THE RISK THAT OFFENDERS RUN AT EACH ONE OF THOSE STAGES. THAT
IS, THE PROBABILITY OF A DEATH SENTENCE HAS BEEN CALCULATED IN
THE RIGHT-HAND COLUMN OF DB-38, AT EACH STAGE.
Q. EXCUSE ME. ARE YOU REFERRING TO DB-53 OR fh 4
A. 1 BEG YOUR PARDON. DB-59.
IN THE RIGHT-HAND COLUMN OF DB-5%, YOUR HONOR, YOU CAN
SEE THE DEATH SENTENCING RISK FOR OFFENDERS AT EACH STAGE
INCREASE AS ONE MOVES THROUGH THE SYSTEM. COMMENCING FROM .0S5
‘FOR A PERSON INDICTED AT THE OUTSET FOR EITHER MURDER OR
VOLUNTARY MANSLAUGHTER, AND STEADILY INCREASING UNTIL THE RISK
IS IN EXCESS OF FIFTY PERCENT AT THE PENALTY TRIAL.
636
BALDUS - DIRECT
p>
IT’S CLEAR THAT AT THAT POINT AT WHICH ONE RUNS THE
2 | GREATEST RISK OF INCURRING A DEATH SENTENCE.
3 THE NEXT POINT AT WHICH THERE IS A SIGNIFICANT EFFECT
4 |ON THAT RISK IS A DECISION OF THE PROSECUTOR TO ADVANCE THE CASE
5 |TO A PENALTY TRIAL WHEN A MURDER CONVICTION HAS BEEN OBTAINED AT
* 5 lYRIAL,
7 AND AS YOU CAN SEE. THAT AFTER A MURDER CONVICTION AT
Ss |TRIAL, THE RISK OF THE DEATH SENTENCE FOR ALL THOSE CONVICTED AT
y |TRIAL 1S .17.
10 FOR THOSE WHO MOVED FORWARD. THE. TO A PENALTY TRIAL.
11 |THE RISK INCREASES SUBSTANTIALLY FROM .17 TO .S3.
12 MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE THE
13 |ADMISSION OF DB-59 INTO EVIDENCE.
14 THE COURT: YOUR FIRST LINE IS INDICTED FOR MURDER OR
15 MANSLAUGHTER: AND THE OTHER LINE IS INDICTED FOR MURDER; AND THE
1&6 FIGURES BOTH IN TERMS OF PROPORTION AND IN TERMS OF RAW NUMBERS
17 ARE THE SAME.
18 I DOUBT THAT THAT WOULD BE TRUE IF I UNDERSTAND THE
* 19 NOMENCLATURE.
20 : IN OTHER WORDS, THERE OUGHT TO BE A LARGER NUMBER
21 INDICTED FOR MURDER OR VOLUNTARY MANSLAUGHTER THAN ARE INDICTED
22 FOR MURDER.
23. CAN YOU EXPLAIN THAT?
24 THE WITNESS: THAT SHOULD BE WHAT’S ON THE EXHIBIT.
23 YOUR HONOR.
&37
BALDUS - DIRECT
fo
e THE COURT: WELL, THE DISTINCTION, THE NUMERICAL
DIFFERENCE IS A RAW DIFFERENCE OF FORTY-TWO.
THE WITNESS: NO, I THINK, I THINK IT“S A HUNDRED --
DOESN’T THE —-
THE COURT: EXCUSE ME. 142.
THE WITNESS: YES.
THE COURT: THAT STRIKES ME AS BEING KIND OF SMALL. IS
THAT —-
THE WITNESS: THAT WAS ONE OF THE BIG SURPRISES I HAD
pa
s O
M
0
S
W
N
WHEN I LOOKED AT THESE DATA, YOUR HONOR, WAS EXACTLY THAT FACT,
THAT APPLYING IT IN THIS STAGE, THAT MOST HOMICIDES RESULT IN A
po
e
ph
12 |MURDER INDICTMENT. AND EVEN CASES THAT ARE ULTIMATELY DISPOSED
13 OF ON EITHER A VERDICT OR A PLEA OF VOLUNTARY MANSLAUGHTER. HAVE
14 INITIALLY BEEN COMMENCED WITH A MURDER INDICTMENT.
13 BY MR. BOGER:
16 R. 1S THAT REFLECTED, BY THE WAY. IN DB-3S8 AS WELL?
17 A. YES. DB-58 SHOWS IT VERY GRAPHICALLY. YOU SEE IN DB-53 ON
18 THE SECOND BAR. WE HAVE A VERY SUBSTANTIAL PROPORTION OF THE
% 19 CASES, 94 PERCENT OF THE CASES COMMENCE WITH A MURDER
20 INDICTMENT, AND AFTER PLEAS TO VOLUNTARY MANSLAUGHTER ARE
21 RECEIVED AND ACCEPTED, THE POOL OF PEOPLE HAS BEEN REDUCED TO DO
22 =5 PERCENT OF THOSE THAT ORIGINALLY STARTED IN THE PROCESS.
23 THE COURT: ALL RIGHT. I SEE.
24 BY MR. BOGER:
25 |G. LET ME RENEW --
om — | a pense ep——— eee et ret. ee. erent ete 3 ——
rn S——— —— — — r—— t— —— —
BALDUS - DIRECT
fo
re
THE COURT: THEY WILL BE ADMITTED.
MR. BOGER: THANK YOU, YOUR HONOR.
BY MR. BOGER:
@. WAS THERE ANY OTHER WAY IN WHICH YOU ATTEMPTED, PROFESSOR
BALDUS., AGAIN, AT THE OUTSET OF YOUR ANALYSIS, TO GET A SENSE
FOR WHAT YOUR CASES LOOKED LIKE, OR WHAT THE UNIVERSE WAS IN THE
CHARGE AND SENTENCING STUDY?
A. YES. THE INITIAL STEP WAS TO PRODUCE WHAT IS KNOWN AS A
FREQUENCY DISTRIBUTION OF THE NON-RACIAL FACTORS THAT COULD
pe
(®
)
vg
@
wj
0
H
B
MN
O
N
POSSIBLY EXPLAIN ANY RACIAL DISPARITIES THAT WE OBSERVED IN THE
p
o
po
t PROCESS.
oy
Ka
IN STUDYING ANY SYSTEM. IT”3 IMPORTANT TO GET AN
13 UNDERSTANDING OF THE CHARACTERISTICS OF THE UNDERLYING PROCESS.
14 AND HERE WE WERE INTERESTED IN HOW HOMICIDES ARE
15 CREATED, WHO THE ACTORS ARE, WHAT KIND OF BACKGROUND THEY BRING
16 TO THE CIRCUMSTANCES OF THE CRIME, WHAT THE CHARACTERISTICS OF
17 THE CRIME ARE. SO IN ONE SENSE, WE EXAMINED THESE DISTRIBUTIONS
1a ON THE VARIABLES THAT WE HAD CREATED TO CONDUCT THE STATISTICAL
® 19 ANALYSIS, SO WE COULD GET THAT SORT OF INSIGHT INTO HOW THIS
20 SYSTEM WAS GOING FORWARD, AND I MEAN THE SYSTEM. THE PROCESS BY
21 WHICH PEOPLE ARE KILLING ONE ANOTHER IN THIS STATE.
al IN ADDITION, WE WANTED TQ GET A SENSE OF HOW THE
23 CHARACTERISTICS OF THE POPULATION VARIED WITH RESPECT TO EACH OF
24 THESE CHARACTERISTICS.
23 @. NOW WE'RE TALKING ABOUT THE POPULATION HERE OF THE HOMICIDE
639
BALDUS - DIRECT
DEFENDANTS. IS THAT CORRECT?
A. THAT“S RIGHT, POPULATION OF DEFENDANTS WHO WERE THE SUBJECT
OF OUR STUDY.
IF YOU COULD TURN, YOUR HONOR, TO PAGE 4 OF DB-40, I
CAN GIVE YOU AN ILLUSTRATION.
THE COURT: SINCE WE’RE GOING TO TESTIFY FROM THE
DOCUMENT. IT OUGHT TO BE IN EVIDENCE.
OTHER THAN YOUR GENERAL OBJECTION, DO YOU HAVE ANY
OBJECTION TO THE RECEIPT OF -607?
MS. WESTMORELAND: YOUR HONOR, THE OTHER OBJECTION I
WOULD HAVE AT THIS POINT IS THAT WE WOULD OBJECT, I SUPPOSE THIS
WOULD OBJECT A CONTINUING OBJECTION AS WELL, TO ANYTHING THAT
DOES NOT SPECIFICALLY ADDRESS FULTON COUNTY, AS WE CONSIDER THAT
TO BE THE AREA OF IMPORTANCE IN THIS CASE.
THE COURT: THAT OBJECTION IS NOT WITHOUT SOME MERIT.
BUT I NECESSARILY AM GOING TO HAVE TO RESERVE RULING ON THAT. SO
WE WILL GO AHEAD.
BY MR. BOGER:
QR. PROFESSOR BALDUS, CAN YOU IDENTIFY, THEN. DB-607?
A. YES. DB-40 IS A FREQUENCY DISTRIBUTION ON THE MORE THAN 200
VARIABLES THAT WE SELECTED FOR INCLUSION IN THE FINAL
STATISTICAL ANALYSIS OF THESE DATA.
@. AND THAT’ S THE FREQUENCY DISTRIBUTION TABLE YOU SPOKE OF
JUST A MOMENT AGO?
A. THATS RIGHT.
rn —— rt; poe. arr —— ———, seine, tree Ser e— J ——
660
BALDUS - DIRECT
1 MR. BOGER: YOUR HONOR, I DO MOVE ITS ADMISSION INTO
2 EVIDENCE NOW.
3 THE COURT: IT-LL BE ADMITTED.
4 MR. BOGER: THANK YOU.
> THE WITNESS: NOW, TURNING TO PAGE 4, I CAN ILLUSTRATE
& THE NATURE OF THE FREQUENCY DISTRIBUTION, AND ALSO THE IMPORTANT
7 PROPERTY OF IT WITH RESPECT TO THE CONDUCT DF THE STATISTICAL
8 ANALYSIS.
9 UNDER ROW K, IT INDICATES NUMBER OF CONVICTIONS:
10 VIOLENT PERSONAL CRIME, BURGLARY. ARSON FIRST DEGREE. AND GIVES
11 THE NAME OF THAT VARIABLE.
12 AND THEN BENEATH THAT, IT LISTS THE PART WE
13 CHARACTERIZE AS THE FREQUENCY DISTRIBUTION ITSELF, THAT IS.
14 |NUMBER OF CONVICTIONS, HOW MANY OFFENDERS IN OUR UNIVERSE DO WE
13 ESTIMATE HAD NO CONVICTIONS. AND THESE ESTIMATES ARE BASED ON
16 THE SAMPLE OF CASES WE COLLECTED DATA ON. AND YOU CAN SEE THERE
17 THAT 1,680 PEOPLE HAD A RECORD IN THE PAROLE BOARD THAT
13 INDICATED THAT THEY HAD NO CONVICTIONS FOR THIS SORT OF CRIME.
4 19 THE COURT: THIS KIND -- OH. OKAY, FOR PERSONAL CRIME,
20 BURGLARY. ARSON, FIRST DEGREE.
21 THE WITNESS: YES, YOUR HONOR. VIOLENT PERSONAL CRIME.
22 THESE ARE REALLY THE VERY SERIQUS CRIMES. THIS IS AN IMPORTANT
23 VARIABLE WE FOUND IN QUR LATER ANALYSIS. AND THAT THERE WERE AN
24 |ESTIMATED S73 WHO HAD FROM ONE TO TWO.
25 23 PERCENT OF THE POPULATION.
&61
BALDUS - DIRECT
AND 184 IS THE ESTIMATED GROUP OF THOSE WHO HAD THREE
Pt
e
TO FIVE. AND SO ON.
IT GIVES YOU A RANGE OF THE DISTRIBUTION OF THOSE
CHARACTERISTICS. AND WHATS IMPORTANT IS THAT THIS RANGE IS
WHAT WE CALL VARIATION. IN EXPLANATORY VARIABLE.
THIS IS THE TYPE OF VARIABLE. AND ALL OF THESE
VARIABLES HERE ARE THE VARIABLES THAT WE WOULD EXPECT TO EXPLAIN
oO
~N
¢
A
PP
»
B
N
WHO RECEIVED DEATH SENTENCES IN THE JURISDICTION.
? THAT WAS THE PRINCIPAL CRITERION UPON WHICH THESE
10 VARIABLES WERE SELECTED. AND THIS ILLUSTRATES IT. SO WHEN WE DO
5) OUR ANALYSIS WE TALK ABOUT THE EXTENT TO WHICH VARIATIONS IN
12 THIS UNDERLYING VARIABLE WILL EXPLAIN WHO RECEIVES A DEATH
13 SENTENCE, AND WHO DOES NOT.
14 | THAT WAS THE OTHER PRINCIPAL PURPOSE OF THIS ANALYSIS,
13 TO GIVE ONE A SENSE OF HOW THESE CHARACTERISTICS IN THE
16 POPULATION VARY BETWEEN OFFENDERS.
17 BY MR. BOGER:
13 @. AND WHAT GENERAL AREAS AS REFLECTED BY YOUR TABLE oF
19 CONTENTS DID YOU OBTAIN THESE FREQUENCY DISTRIBUTIONS ON?
20 A. THE, THE LIST OF VARIABLES INCLUDE THE MAJOR CHARACTERISTICS
21 OF, THAT ARE IMPORTANT IN THE DISPOSITION OF THESE CASES, THAT
_— IS. THE CHARACTERISTICS OF THE DEFENDANT. THE NUMBER AND ROLE OF
23 CO-PERPETRATORS, CHARACTERISTICS OF THE VICTIM, THE
24 CHARACTERISTICS OF THE CRIME ITSELF. OTHER SPECIAL AGGRAVATING
25 AND MITIGATING FACTORS. AND EVIDENTIARY FACTORS.
— — ———— ——— ——. —— A ——— ——— A— ——— _————", W—.. —— —— p—— N— —
po
p
t
oO
-
Py
po
h nN
W
M
a
o
e
W
B
N
BALDUS — DIRECT
FINALLY» WE HAVE A DISTRIBUTION ON THE RACIAL
CHARACTERISTICS AS WELL.
0. ALL RIGHT. THEN. YOUVE GOT THESE BACKGROUND INDICES OR, I
GUESS INDICES IS AN INAPPROPRIATE WORD, LOOKS OR PICTURES, IF
YOU WOULD, OF THE HOMICIDE FREQUENCY OVER THE LAST EIGHT OR NINE
YEARS IN GEORGIA. DEATH SENTENCING OUTCOMES, FREQUENCY
DISTRIBUTION OF THE CHARACTERISTICS OF THE VARIABLES IN THE
CASES YOU ARE STUDYING.
BUT THE PARTICULAR FOCUS OF INTEREST, AS I LNDERSTAND
YOUR PREVIOUS TESTIMONY. OF THE PROCEDURAL REFORM, AND CHARGING
AND SENTENCING STUDY, AT LEAST THOSE PORTIONS BEFORE THE COURT
IN THIS HEARING. WAS THE IMPACT OF THE RACIAL VARIABLES ON
SENTENCING OUTCOMES. AND WHETHER OTHER VARIABLES COULD EXPLAIN
THOSE OUTCOMES.
WHAT DID YOU FIND TO BE NECESSARY IN ORDER TO DETERMINE
WHETHER RACIAL FACTORS HAD AN OUTCOME EFFECT?
A. THE FIRST DETERMINATION THAT HAD TO BE MADE WAS TO SPECIFY A
MEASURE OF IMPACT. MEASURE OF STATISTICAL EFFECT.
@. WHAT DO YOU MEAN BY A MEASURE OF IMPACT?
A. 1 CAN ILLUSTRATE THIS BY EXAMINING DB-&1.
3. LET ME ASK YOU THEN TO TURN TO DB-61 AND IDENTIFY IT.
A. DB-61 IS A TABLE LABELED "UNADJUSTED MEASURES OF THE IMPACT
OF GEORGIA‘S STATUTORY AGGRAVATING FACTORS ON DEATH SENTENCING
RATE".
@. WHERE DOES THIS TABLE COME FROM?
&63
BALDUS - DIRECT
1 A. THIS IS TABLE EE FROM OUR FINAL REPORT.
2 @. FINAL REPORT, CHARGING AND SENTENCING STUDY?
3 A. YES.
@. AND DID YOU RELY UPON THAT REPORT IN GIVING THE EXPERT
OPINIONS WHICH YOU‘RE GOING TO GIVE IN THIS HEARING?
A. YES.
4
bo
b
7 MR. BOGER: YOUR HONOR, I MOVE THE ADMISSION OF
e DB-61 INTO EVIDENCE. PROFESSOR BALDUS INDICATED HES GOING TO
? USE IT TO ILLUSTRATE HIS TESTIMONY. AND THATS THE PURFOSE FOR
0 WHICH I WILL OFFER IT AT THIS POINT.
11 MS. WESTMORELAND: YOUR HONOR, IF IT’S JUST BEING
12 OFFERED FOR THE PURPOSE OF ILLUSTRATING PROFESSOR BALDUS”
13 TESTIMONY, THEN IT MIGHT BE ADMISSIBLE. OTHERWISE. I DONT
14 THINK A PROPER FOUNDATION HAS BEEN SHOWN OF HOW THE NUMBERS ON
13 THIS TABLE WERE OBTAINED AT THIS STAGE OF THE PROCEEDINGS.
16 THE COURT: I WOULD NOT ADMIT IT FOR THE PURPOSE OF
17 ILLUSTRATING HIS TESTIMONY.
18 I THINK IT IS BEING ADMITTED BOTH TO SHOW PROCESS AND
a 39 TO SHOW PREDICATE FACTS.
20 : THE WITNESS: YES, YOUR HONOR.
2} THE COURT: SO IT IS BEING OFFERED FOR THE TRUTH OF THE
22 MATTER ASSERTED THEREIN.
23 ASSUMING THAT, WHAT IS YOUR OBJECTION, OTHER THAN
24 GENERAL ONES PREVIOUSLY STATED?
23 MS. WESTMORELAND: OTHER THAN THAT, YOUR HONOR. I
et. ren, Se —— Wo———at” " — —— © — So— — le
664
BALDUS - DIRECT
fy
ASSERT THERE IS NO FOUNDATION THAT HAS BEEN LAID TO SHOW HOW THE
2 NUMBERS REFLECT IN THE TABLE, FOR INSTANCE. DATA NUMBER 3, HOW
3 THE REGRESSION RESULTS WERE OBTAINED, WHAT FIGURES WERE USED TO
4 OBTAIN THE VARIOUS FACTORS IN THIS TABLE. I DON‘T KNOW IF THAT’S
3 GOING TO BE THE SUBSTANCE OF THE TESTIMONY FORTHCOMING OR NOT.
. é THE COURT: I THINK THATS WHERE WE‘RE GOING TO GET.
7 I WOULD LIKE, I THINK I WOULD LIKE TO —= I THINK I
8 WOULD LIKE TO GET A BIG PICTURE BEFORE I START LOOKING AT THE
4 PIECES.
10 S0 IM GOING TO RESERVE RULING ON YOUR OBJECTION. ADMIT
11 IT CONDITIONALLY, SO THAT YOU CAN GO ON THROUGH AND GIVE ME THE
12 BIG PICTURE, AND THEN WELL GO BACK AND LAY THAT SORT OF
13 FOUNDATION.
14 MR. BOGER: FINE, YOUR HONOR, ALTHOUGH SOME OF THAT
13 FOUNDATION, I“M NOT SURE IT’S NECESSARY FOUNDATION SUGGESTED BY
16 THE STATE’S RESPONSE. BUT WILL PERHAPS BE LAID BY PROFESSOR
17 WOODWORTH AND SOME BY PROFESSOR BALDUS.
18 BY MR. BOGER:
® 1? QR. PROFESSOR BALDUS, YOU’VE INDICATED YOU CAN ILLUSTRATE THIS
20 NOTION OF MEASURE OF IMPACT THROUGH DB-&1.
21 CAN YOU EXPLAIN FOR US?
22 A. YES. DB-41 PRESENTS INFORMATION THAT WE COLLECTED ON EACH
23 OF GEORGIA STATUTORY AGGRAVATING FACTORS.
24 WHAT I MEAN BY THAT IS THAT WE EXAMINED EACH CASE IN
25 THE UNIVERSE OF CASES, AND DETERMINED WHETHER OR NOT THE FACTORS
&65
BALDUS - DIRECT
THAT THE GEORGIA CODE SPECIFIES AS CONSTITUTING A STATUTORY
AGGRAVATING CIRCUMSTANCE WERE PRESENT IN THOSE CASES.
FOR EXAMPLE, UNDER THE STATUTORY AGGRAVATING FACTOR BZ.
ENUMERATED CONTEMPORANEOUS OFFENSE, IN OUR RECODING OF THE DATA.
WE ASKED THE MACHINE TO IDENTIFY CASES WHICH HAD ONE OF THOSE
CONTEMPORANEOUS OFFENSES WHICH MAKES A CASE DEATH ELIGIBLE UNDER
THE GEORGIA DEATH SENTENCING STATUTE.
IN COLUMN B., WE INDICATE THE NUMBER OF SUCH CASES.
DEATH SENTENCING RATES IS A FACTOR PRESENT IN THE CASE.
COLUMN B., ESTIMATES THE 497 CASES WHERE THE B2
STATUTORY FACTOR IS PRESENT IN THE CASE.
WE THEN ASKED HOW MANY OF THOSE CASES RESULTED IN DEATH
SENTENCE. AS You CAN SEE THERE. THERE WERE 105 OF THOSE CASES
THAT RESULTED IN A DEATH SENTENCE, PRODUCING AN OVERALL DEATH
SENTENCING RATE FOR THAT CATEGORY OF CASES OF .Z21.
WE THEN —-
THE COURT: WAIT A MINUTE.
IS THE FACTOR PRESENT IN THE CASE? THERE WERE 4%7
CASES IN WHICH, WHAT?
THE WITNESS: IN WHICH THERE WAS A CONTEMPORANEOUS
| OFFENSE.
THE COURT: ALL RIGHT. AND 105 OF THOSE. WHAT
HAPFENED?
THE WITNESS: THERE WAS A DEATH SENTENCE IMPOSED.
THE COURT: WHATS IN COLUMN 27?
wr. ——— p—— o——— a—— —— ——— paren ieee Fot—— ——— eer". SS pnt Sr——— se — a ——— hs, FOS — ———— —n | — ———
BALDUS - DIRECT
1 THE WITNESS: COLUMN 2 ARE THE CASES THAT DID NOT HAVE
A B2 CONTEMPORANEOUS OFFENSE PRESENT. THEY MAY HAVE HAD SOME
OTHER STATUTORY FACTOR PRESENT, YOUR HONOR. BUT THEY DIDN’T HAVE
A BZ.
SO YOU CAN SEE IN THAT POPULATION OF CASES WHERE THE BZ
FACTOR WAS NOT PRESENT. THAT CONSTITUTES THE GREAT BULK OF THE
CASES, AN ESTIMATED ROUGHLY TWO THOUSAND CASES, AND 23 DEATH
0
~N
6
A
H+
O
N
SENTENCES WERE FOUND AMONG THOSE CASES.
? SO, THIS SORT OF INITIAL CALCULATION CAN BE MADE WITH
10 RESPECT TO EACH OF THESE STATUTORY AGGRAVATING FACTORS.
11 THE COURT: LET’S MOVE TO COLUMN D. WHAT DOES THAT
12 TELL ME?
13 THE WITNESS: D ILLUSTRATES THE FIRST MEASURE THAT WE
14 USED TO SHOW THE IMPACT OF, THE STATISTICAL IMPACT OF BZ
13 STATUTES, OTHER AGGRAVATING FACTORS. AND THAT MEASURE IS SIMPLY
1&6 THE ARITHMATIC DIFFERENCE BETWEEN THE TWO RATES PRESENTED IB
37 COLUMNS B AND C, DIFFERENCE BETWEEN . 21 AND .01. VERY SIMPLY,
18 THIS IS THE BASIC MEASURE OF STATISTICAL IMPACT THAT IS USED IN
® 19 A WIDE VARIETY OF CONTEXTS. JUDICIAL CONTEXTS CONCERNING
20 DISCRIMINATION. THE MOST VISIBLE USE OF THIS MEASURE OF
21 DISPARITY AND TREATMENT IS SWAYNE V. ALABAMA. IN THAT CASE, THE
22 UNITED STATES SUPREME COURT —-
23 MS. WESTMORELAND: YOUR HONOR, ILL OBJECT TO THE
24 WITNESS ARGUING LAW TO THE COURT AT THIS POINT.
23 THE COURT: SUSTAINED.
667
BALDUS - DIRECT
MR. BOGER: YOUR HONOR, I DON’T THINK THE WITNESS IS
ARGUING LAW TO THE COURT. THE WITNESS IS A LAW PROFESSOR WHO IS
SIMPLY, WHO HAS WRITTEN A BOOK ON DISCRIMINATION AND HOW COURTS
USE DISCRIMINATION —- I THINK HES SIMPLY TRYING TQ EXPLAIN THE
CONTEXT IN WHICH THIS MEASUREMENT IS USED. I DON’T THINK HES
URGING THAT MEASUREMENT EVEN ON THE COURT.
HES SIMPLY =
THE COURT: I SUSTAIN THE OBJECTION.
BY MR. BOGER:
R. ALL RIGHT, PROFESSOR BALDUS, YOUVE INDICATED THAT MEASURE
ONE IS A MEASUREMENT OF THE IMPACT OR DISPARITY OF IMPACT
BETWEEN CASES WITH AND WITHOUT THE B2 FACTOR PRESENT.
ARE THERE OTHER WAYS IN WHICH ONE MIGHT MEASURE THAT
SAME IMPACT?
A. YES. WHEN EVERYONE IS EXAMINING THE EFFECT OF A FACTOR, THE
EFFECT CAN BE VIEWED FROM DIFFERENT PERSPECTIVES. THE FIRST
MEASURE THAT WE HAVE HERE INDICATES THE RISE IN PROBABILITY THAT
THE AVERAGE DEFENDANT IS EXPOSED TO IF HIS CASE INVOLVES A
CONTEMPORANEOUS OFFENSE.
NOW, ANOTHER WAY TO LOOK AT IT IS FRESENTED IN MEASURE
2, WHICH IS FOUND IN COLUMN E. AND THIS IS THE RATIO. IT
INDICATES A MULTIPLIER, IT PRESENTS A MULTIPLIER EFFECT. SHOWS
HOW MANY MORE TIMES ONE GROUP OF OFFENDERS IS LIKELY TO RECEIVE
A DEATH SENTENCE THAN ANOTHER GROUR.
AND THIS ALSO IS A MEASURE THAT IS COMMONLY UZED IN
668
BALDUS - DIRECT
to
ed
SOCIAL SCIENCE RESEARCH AND IT“S USED IN JUDICIAL CONTEXTS AS
2 WEL.L..
3 AND IT SIMPLY ASKS A DIFFERENT QUESTION. THESE ARE NOT
4 STATISTICAL QUESTIONS, THESE ARE LEGAL QUESTIONS, THAT IS, WHAT
3 IS THE APPROPRIATE MEASURE.
. & THE LAW DEFINES THE ISSUE, AND THE METHODOLOGY IS
7 SELECTED THAT GIVES THE MOST RELEVANT ANSWER TO IT.
8 AND IN THE CONTEXT OF DISCRIMINATION LITIGATION.
7? LITIGANTS AND COURTS HAVE FOUND THAT THESE TWO MEASURES
10 FREQUENTLY PROVIDE RELEVANT INFORMATION TO ANSWER THE FACTUAL
11 QUESTIONS THAT ARE POSED BY THE CLAIMS OF LITIGANTS.
12 MS. WESTMORELAND: YOUR HONOR. I OBJECT TO THE --
13 THE COURT: I SUSTAIN THE DBJECTION.
14 MR. BOGER, WE WENT THROUGH THIS THE LAST TIME YOU AND I
13 WERE TOGETHER. AS FAR AS I“M CONCERNED, THERE IS NO SUCH THING
16 AS AN EXPERT WITNESS IN LAW. OKAY?
17 MR. BOGER: FINE, YOUR HONOR.
18 THE COURT: UNLESS IT HAPPENS TO BE FOREIGN LAW.
® 19 MR. BOGER: ALL RIGHT. YOUR HONOR.
20 BY MR. BOGER:
23 RA. PROFESSOR BALDUS, WE‘VE TALKED NOW ABOUT TWO KINDS OF
22 MEASUREMENTS THAT ARE USED IN SOCIAL SCIENCE TO MEASURE
23 DISCRIMINATION, AND WHICH YOU‘VE TESTIFIED YOU HAVE KNOWLEDGE.
24 WHY WOULD ONE USE MORE THAN ONE METHOD IF THERE IS ONE
23 METHOD AVAILABLE?
669
BALDUS - DIRECT
A. WELL. FOR THE REASON I JUST INDICATED. THEY ADDRESS
DIFFERENT QUESTIONS, THAT IS, THAT MEASURES PROVIDE DIFFERENT
INFORMATION ABOUT THE EFFECT OF A PARTICULAR FACTOR IN A
PROCESS, AND HERE THE FIRST MEASURE IN "D" INDICATES THE EFFECT
THAT THE PRESENCE OF THE FACTOR HAS ON THE AVERAGE
DEFENDANTS PROBABILITY OF RECEIVING A DEATH SENTENCE.
AND THE SECOND ONE, IN COLUMN E WHICH WE REFER TO AS
MEASURE 2, PRESENTS A MULTIPLICATIVE MEASURE OF THE
RELATIONSHIPS BETWEEN THE RISKS OF A DEATH SENTENCE EXPERIENCED
BY THOSE TWO GROUPS. SO THAT ON THE ONE HAND, WE SAY THERE’S A
TWENTY PERCENTAGE POINT DIFFERENCE IN THE RATES AT WHICH B2
CASES RECEIVE DEATH SENTENCES AND OTHER CASES RECEIVE DEATH
SENTENCES.
AND THE SECOND MEASURE. WE CAN SAY THAT THE RATE OF
DEATH SENTENCING AMONG B2 CASES IS 21 TIMES HIGHER THAN IT IS IN
THE POPULATION OF PEOPLE WHO DO NOT HAVE A B2 FACTOR IN THEIR
CASE.
@. NOW, ARE THERE ANY OTHER METHODS OF MEASUREMENT THAT CAN BE
USED?
A. YES. THERE ARE TWO OTHER COMMONLY USED METHODS OF
MEASUREMENT, AND THEY ARE DERIVED FROM REGRESSION ANALYSIS. AND
THEY ARE PRESENTED. ILLUSTRATIONS OF THEM ARE PRESENTED IN “F"
AND "G", AND THEY ARE CLOSELY ANALOGOUS TO THE MEASURES PRESENTED
IN COLUMN D AND COLUMN E, THAT IS, MEASURES 1 AND Z.
MEASURE 2 IS A LEAST SQUARES REGRESSION CO-EFFICIENT.
| e— ———. . mre ere S———
670
BALDUS ~ DIRECT
I
a
AND IN THE CONTEXT OF THIS EXAMPLE THAT-S GIVEN HER, IT MEASURES
THE SAME THING THAT MEASURE 1 IN COLUMN D DOES.
I WILL EXPLAIN IN MORE DETAIL SHORTLY ABOUT THE
REGRESSION PROCEDURES AND THE OTHER PROPERTIES THEY HAVE. BUT IN
TERMS OF THE BASIC LEAST SQUARES MEASURE, THE INFORMATION
CARRIED BY THE UNADJUSTED REGRESSION CO-EFFICIENT IS THE SAME
INFORMATION THAT-S CARRIED BY MEASURE 1, THE ARITHMETIC
DIFFERENCE IN DEATH SENTENCING RATES.
80 THAT'S A VERY COMMONLY USED MEASURE IN SOCIAL
O
O
m
o
r
e
n
SCIENCE. AND IN A WIDE VARIETY OF SCIENCES.
11 RA. IS THERE AN INITIAL METHOD YET AGAIN?
12 A. YES, THERES A FOURTH MEASURE THAT WE USED, WHICH IS DERIVED
13 FROM AN ALTERNATIVE REGRESSION PROCEDURE KNOWN AS LOGISTIC
14 REGRESSION PROCEDURE. AND IT PRESENTS A MULTIPLICATIVE MEASURE.
15 AS YOU CAN SEE. YOUR HONOR, UNDER COLUMNS G AND H, IVE
16 INDICATED THE REGRESSION CO-EFFICIENT FROM THE LOGISTIC
17 REGRESSION ANALYSIS. THAT CO-EFFICIENT ITSELF IS NOT
18 INTERPRETABLE WITHOUT FURTHER ANALYSIS. BUT UPON FURTHER
® 19 ANALYSIS, IT IS POSSIBLE TO CALCULATE A MEASURE WHICH WE IN THIS
20 CONTEXT REFER TO AS THE DEATH ODDS MULTIPLIER. AND THAT MEASURE
21 TELLS ONE THE NUMBER OF TIMES THAT ONE-‘S ODDS OF RECEIVING A
22 DEATH SENTENCE ARE INCREASED IF THE CHARACTERISTIC OF INTEREST
23 IS PRESENT.
24 TRANSLATED TO THE EXAMPLE THAT WE HAVE HERE CONCERNING
25 THE B2 CONTEMPORANEOUS OFFENSE, IT TELLS US IN THIS CONTEXT THAT
OO
WD
:
0
N
O
R
R
N
»
po
s
po
t
-
—-
3
MB
TY
E+
571
BALDUS —- DIRECT
BEFORE WE MAKE ADJUSTMENTS FOR ANY OTHER BACKGROUND FACTOR AND
WE JUST LOOK AT THE STATISTICAL EFFECT OF THE B2 FACTOR, THAT
THE AVERAGE OFFENDER‘S ODDS WHO HAS A CONTEMPORANEOUS B2 OFFENSE
IN HIS CASE IS 17 TIMES LARGER THAN OFFENDERS WHO DO NOT HAVE
THAT CHARACTERISTIC PRESENT IN THEIR CASE.
3. NOW. DURING THE COURSE OF YOUR CHARGING AND SENTENCING STUDY
HAVE YOU EMPLOYED ANY ONE OF THESE MEASURES?
A. THROUGHOUT THE STUDY, WE EMPLOYED ALL OF THESE MEASURES.
THE PRINCIPAL METHODS THAT WE USED TO CONTROL FOR
BACKGROUND FACTORS ARE CROSS TABULATION METHODS AND REGRESSION
METHODS.
THE CROSS TABULATION METHODS RELY ON MEASURES ONE AND
TWO.
AND THE REGRESSION MEASURES. METHODS. EMPLOY LEAST
SQUARES ANALYSIS AND LOGISTIC ANALYSIS, AND THEY RELY ON
MEASURES 3 AND 4.
AND THEY WERE USED IN EACH CONTACT. IN EACH MAJOR
SECTION OF THE PAPER. SIMULTANEOUSLY.
3. WE’RE GOING TO COME BACK AND TALK ABOUT MEASUREMENT A LITTLE
BIT MORE IN A FEW MINUTES.
I WANT NOW TO DIRECT YOUR ATTENTION TO ROW 8 OF DE-61,
WHICH IS THE STATUTORY AGGRAVATING FACTOR INVOLVING THE SHOOTING
OF A POLICE OR FIRE PERSON.
CAN YOU TELL ME WHAT THE UNADJUSTED ARITHMETIC
DIFFERENCE IS IN SENTENCING RATE FOR A CAPITAL DEFENDANT BASED
rent Seer Se, ep— —————— —
BALDUS - DIRECT
1 UPON WHETHER THAT AGGRAVATING CIRCUMSTANCE IS PRESENT OR ABSENT?
2 A. YES, THE STATISTICAL IMPACT OF THE B& FACTOR. WHICH IS
3 WHETHER OR NOT THERE WAS A POLICE OR FIRE PERSON AS VICTIM. IS
4 AS CALCULATED BY MEASURE ONE. A 23 PERCENTAGE POINT DISPARITY IN
5 DEATH SENTENCING RATE.
b MN. WHAT WOULD THAT 23 POINT DIFFERENCE SUGGEST?
7 A. WELL, THAT WOULD SUGGEST THE HYPOTHESIS THAT THE SYSTEM IN
f GEORGIA, THE CHARGING AND SENTENCING SYSTEM, VIEWS THE KILLING
9 OF A POLICE OFFICER AS A SERIOUS MATTER. AND THAT AS A
10 CONSEQUENCE, DEATH SENTENCES WOULD BE MORE LIKELY SOUGHT IN
11 THOSE CASES, AND OBTAINED IN THOSE CASES BECAUSE OF THE PRESENCE
12 OF THE BS FACTOR. .
13 AN UNADJUSTED MEASURE OF THIS TYPE GENERATES A
14 HYPOTHESIS THAT WITHIN THE SYSTEM THAT WE“RE ANALYZING, THAT
15 THIS FACTOR WHICH PRODUCES A STATISTICAL EFFECT. HAS A REAL
14 EFFECT ON THE DECISIONS IN THE SYSTEM. SPECIFICALLY THE
17 |DECISIONS TO SEEK AND TO OBTAIN A DEATH SENTENCE.
18 @. IS THERE ANY WAY TO TEST THAT HYPOTHESIS?
% 19 THE COURT: LET ME ASK A QUESTION FIRST.
20 PROFESSOR BALDUS TESTIFIED JUST NOW THAT THIS INDICATED
21 TO HIM A GREATER LIKELIHOOD OR AN ENHANCED LIKELIHOOD THAT THE
22 DEATH PENALTY WOULD BE SOUGHT AND IMPOSED.
23 I DON’T UNDERSTAND HOW THIS SHOWS THAT THERE IS AN
24 INCREASED LIKELIHOOD THAT IT WOULD BE SOUGHT.
23 BY MR. BOGER:
&73
BALDUS - DIRECT
@. CAN YOU ADDRESS THAT QUESTION. PROFESSOR BALDUS?
A. CERTAINLY.
YOUR HONOR, THESE NUMBERS HERE REFLECT THE IMPACT OF
ALL SIX DECISIONS IN THE -- SIX DECISION POINTS IN THE CHARGING
AND SENTENCING PROCESS. AND TO RECEIVE A DEATH SENTENCE. IT IS
NECESSARY THAT THE. THE CASE REMAIN IN THE SYSTEM UP TO THE
POINT OF THE PENALTY TRIAL.
WHEN I WAS USING THE TERM "SOUGHT" I WAS USING IT IN A
GENERAL SENSE, THAT THE PROSECUTORS PROCESSING SUCH A CASE WOULD
MAINTAIN THE CASE IN THE SYSTEM UP TO AND INCLUDING THE POINT OF
A PENALTY TRIAL.
BUT FOR THAT, THERE WOULD NOT BE POSSIBLE TO HAVE A
DEATH SENTENCE, BUT IT IS CORRECT THAT THIS, THESE NUMBERS HERE
DO NOT REFLECT THE. THE DECISION TO SEEK A DEATH SENTENCE. NOR
SIMPLY THE COMBINED EFFECTS. THESE DECISIONS REFLECT THE IMPACT
OF ALL THE DECISIONS MADE FROM THE POINT OF INDICTMENT DOWN TO
THE DEATH SENTENCING OUTCOME ITSELF.
THE COURT: IT’S ONLY INDIRECTLY. ISN‘T THAT S07?
THE WITNESS: 1 DIDNT UNDERSTAND?
THE COURT: ONLY INDIRECTLY.
ISN“T THAT SO?
THE WITNESS: I DON‘T THINK SO.
THE COURT: IF I UNDERSTAND IT, THESE ARE THE CASES IN
WHICH THE DEATH PENALTY WAS IMPOSED BY THE SENTENCING OFFICER,
WHETHER THAT BE THE JURY OR JUDGE.
&74
BALDUS - DIRECT
1 THE WITNESS: YES, YOUR HONOR, BUT THE NUMERATOR
2 INDICATES ALL THE PEOPLE WHO WERE INDICTED FOR VOLUNTARY
MANSLAUGHTER OR MURDER.
MR. BOGER: EXCUSE ME, PROFESSOR BALDUS. YOU SAID THE
NUMERATOR. DO YOU MEAN THE DENOMINATOR?
3
4
5
* & THE WITNESS: I BEG YOUR PARDON. THE DENOMINATOR
7 INDICATES ALL THE PEOPLE WHO WERE INDICTED TQ, PERHAPS I COULD
8 ILLUSTRATE IT.
$ |BY MR. BOGER:
10 @. LET’S GO THROUGH BS. PROFESSOR BALDUS. IF WE COULD, JUST TO
11 ILLUSTRATE THAT.
12 THE FIGURES IN COLUMN B, REPRESENT WHAT? LET’S TALK
13 ABOUT THE NUMBER OUT OF THE PARENTHESIS, AND THE NUMBER IN THE
14 PARENTHESIS.
5 A. COLUMN B INDICATES THE DEATH SENTENCING RATE AMONG THE CASES
16 IN OUR POPULATION ESTIMATED TO HAVE A POLICE OFFICER VICTIM IN
17 THE ENTIRE UNIVERSE OF CASES THAT WE“RE CONCERNED WITH. THAT
18 IS, AMONG OUR 2484 CASES CONSTITUTE OUR -—-
. 19 THE COURT: THERE ARE 29 CASES YOU PROJECT IN WHICH A
20 POLICE OFFICER OR A FIREMAN WOULD HAVE BEEN A VICTIM.
21 THE WITNESS: THAT”S RIGHT. |
22 THE COURT: AND YOU HAVE ACTUAL. EIGHT WHERE A DEATH
23 PENALTY WAS IMPOSED.
24 THE WITNESS: RIGHT, THOSE EIGHT DECISIONS, YOUR HONOR,
25 REFLECT THE IMPACT OF ALL THE DECISIONS TAKEN FROM THE POINT OF
&75
BALDUS - DIRECT
ORIGINAL INDICTMENT DOWN TO AND INCLUDING THE EFFECT OF THE JURY
-
2 SENTENCING DECISION.
3 THEY REFLECT THE COMBINED CONSEQUENCES OF THAT WHOLE
4 STREAM OF CASES. IF WE WANTED TO LOOK AT, THE EFFECT OF A
3 SMALLER SERIES OF CASES, FOR EXAMPLE. IF WE WANTED TO LOOK AT
® é SIMPLY THE IMPACT OF THE PROSECUTORIAL DECISION TO SEEK A
7 DEATH SENTENCE AFTER OBTAINING A CONVICTION IN A MURDER TRIAL.
a OUR NUMERATOR IN THESE ANALYSES WOULD BE SHARPLY REDUCED. IN
y 4 OUR LATER ANALYSIS. WE DO LOOK AT SUCH ANALYSES.
10 BUT HERE FOR THE PURPOSES OF THIS, WE‘RE TALKING ABOUT
31 THE PROBABILITY. THATS WHAT THAT FIGURE .28 INDICATES. WHATS
12 THE PROBABILITY IF A PERSON WHO IS INDICTED FOR VOLUNTARY
12 MANSLAUGHTER OR MURDER AND HAS A POLICE OFFICER VICTIM. THAT
14 THEY WILL RECEIVE A DEATH SENTENCE. AND THE IMPOSITION OF THAT
135 DEATH SENTENCE IS GOING TO REFLECT A SERIES OF DECISIONS.
14 TO BE SURE THE ACTUAL DEATH SENTENCE ITSELF IS IMPOSED
17 BY A JURY. BUT YOU CAN‘T GET TO THAT POINT EXCEPT FOR THE
13 EARLIER DECISIONS.
. 19 THE COURT: I GUESS WHAT IS WORRYING ME IS THAT BEFORE
20 I WOULD BE ABLE TO DRAW MUCH INFERENCE ABOUT THE CHARGING
21 DECISIONS, I WOULD HAVE TO SEE THE EFFECT OF THE JURY DECISIONS.
22 BY MR. BOGER:
23 ?. PROFESSOR BALDUS, IS THERE A PLACE LATER IN OUR TESTIMONY
24 AND IN YOUR REPORT WHERE WE WILL BREAK DOWN DECISIONS BY
23 STATUTORY AGGRAVATING CIRCUMSTANCE IN THE SIX DECISION POINTS
e
676
BALDUS - DIRECT
1 FROM INDICTMENT THROUGH JURY DECISIONS?
2 A. YES.
3 MR. BOGER: YOUR HONOR. WE WILL BE COMING TO THAT.
4 THE COURT: ALL RIGHT.
5S BY MR. BOGER:
é& @. PROFESSOR BALDUS, I WAS ASKING YOU THEN WHAT. WHAT
4 HYPOTHESIS YOU MIGHT DRAW FROM THIS 23 POINT ARITHMETIC
8 DIFFERENCE IN DEATH SENTENCING RATES. AND I BELIEVE YOUR
9 TESTIMONY WAS, WELL, WHY DON’T YOU GIVE ANOTHER ANSWER TO THAT
10 QUESTION?
11 A. VERY WELL.
12 AN UNADJUSTED RACIAL DISPARITY, CORRECTION. AN
13 UNADJUSTED DISPARITY OF THIS TYPE INDICATES THAT THERE IS AN
14 ASSOCIATION, STATISTICALLY, BETWEEN THE PRESENCE OF THIS
13 FACTOR, THE BS FACTOR IN A CASE, AND THE LIKELIHOOD OF A DEATH
16 SENTENCE.
17 ANOTHER WAY THAT SORT OF ASSOCIATION IS DESCRIBED IN
18 STATISTICAL TERMINOLOGY. IS THAT THERE IS A CORRELATION BETWEEN
THE PRESENCE OF THE POLICE OFFICER VICTIM IN A CASE, AND THE
20 DEATH SENTENCING RESULT.
4 : WHERE THERE-S A POLICE OFFICER VICTIM, IT’S MORE LIKELY
22 THAT THERE IS A DEATH SENTENCING RESULT. WITHOUT LOOKING AT ANY
23 OTHER FACTORS. THAT SUGGESTS THAT THESE TWO FACTORS ARE
24 CORRELATED. WHEN TWO FACTORS ARE CORRELATED. IT GENERATES A
23 HYPOTHESIS THAT IN FACT THE PRESENCE OF A POLICE OFFICER
;
&77
BALDUS - DIRECT
bo
t VICTIM IS INFLUENCING THE DECISIONS IN THE PROCESS.
2 @. ARE YOU ABLE TO MAKE ANY FINAL CONCLUSIONS BASED ON THE
3 HYPOTHESIS OF THAT SORT?
4 A. NO.
5 THE COURT: ISNT THAT A CORRELATION IMPOSED BY LAW?
. A ISN’T THE CORRELATION INEVITABLE?
7 THE WITNESS: NO» YOUR HONOR.
8 THE COURT: YOU COULDNT HAVE ANY DEATH SENTENCES IF IT
9 DON’T VIOLATE THE BS. IN OTHER WORDS, IT WOULDN’T BE
10 CHARGEABLE, IF THERE WERE NOT A BS POSSIBILITY.
11 BY MR. BOGER:
12 @. I THINK WE NEED A FEW CLARIFYING QUESTIONS.
13 PROFESSOR BALDUS, IN THE COLUMN C, ROW ©, WHICH IS, IS
14 THE FACTOR PRESENT IN THE CASE, ANSWER. NO, FOR BS.
13 DOES THE DENOMINATOR THERE INCLUDE CASES THAT HAVE NO
16 STATUTORY AGGRAVATING CIRCUMSTANCES OR DOES IT INCLUDE ALL THE
17 CASES, WHETHER THERE ARE AGGRAVATING CIRCUMSTANCES OR NOT
13 INSOFAR AS THEY DON’T HAVE A BS?
® 19 A. YES, IT INCLUDES ALL THE CASES. SOME HAVE STATUTORY FACTORS
20 UNDER OTHER SECTIONS AND SOME HAVE NONE.
21 I THINK, YOUR HONOR, I COULD MAKE A POINT OR RESPOND TO
22 YOUR QUESTION BY REFERENCE TO THE B3 FACTOR, WHICH IS IN ROW 3
23 OF THIS EXHIBIT.
24 YOU“LL NOTE THERE THAT AMONG THE 348 ESTIMATED CASES
25 WHERE A B3 FACTOR EXISTS, THE DEATH SENTENCING RATE IS .04.
678
BALDUS - DIRECT
1 THAT IS ONLY 14 OF THOSE CASES RESULTED IN A DEATH SENTENCE.
2 WHEREAS AMONG THE ESTIMATED 2114 CASES WHERE THAT FACTOR WAS NOT
3 PRESENT. WE SEE A DEATH SENTENCING RATE OF .035, WHICH IS ONLY
ONE PERCENTAGE POINT DIFFERENCE. SO, -——
THE COURT: WHAT I“M TRYING TO SUGGEST TO YOU IS THIS:
IF YOU DIDN’T HAVE THE STATUTORY AGGRAVATING
4
Z
&
7 CIRCUMSTANCE, YOU COULDN’T HAVE A DEATH PENALTY. THEREFORE,
8 THERE IS GOING TO BE A CORRELATION BETWEEN THE FACT THERE IS
? SOME STATUTORY AGGRAVATING CIRCUMSTANCE AND THE DEATH PENALTY.
0 WHAT I UNDERSTAND YOU TO BE DEMONSTRATING HERE IS THAT
11 SOME AGGRAVATING CIRCUMSTANCES ARE APPARENTLY VIEWED BY THE
12 SYSTEM AS MORE IMPORTANT OR MORE LIKELY TO PRODUCE IT THAN
13 OTHERS, SO IT’S AN INTER-FACTOR COMPARISON, ISN'T IT?
14 THE WITNESS: WELL, THE NOTION HERE IS WHEN WE TALK
15 ABOUT A CORRELATION OR TALK ABOUT A MEASURE OF IMPACT IS THAT WE
16 ARE TALKING ABOUT HOW TWO THINGS MOVE TOGETHER. AND IF WE SEE,
17 IF WE MOVE FROM NO BS TO BS, AND WE DON’T SEE ANY CHANGE IN THE
18 DEATH SENTENCING RATE, THEN WE“RE GOING TQ SAY THERE’S NO
. 19 CORRELATION BETWEEN THOSE, AS IS THE SITUATION UP HERE WITH B3.
20 WHEN WE LOOK, IF WE MOVE FROM NO B3 TO BS, WE SEE JUST A TINY
21 CHANGE IN THE DEATH SENTENCING RATE. THEREFORE WE WOULD SAY
a2 THERES NO CORRELATION OF ANY MOMENT BETWEEN THOSE TWO
23 VARIABLES. AND THERE IS NO STATISTICAL EFFECT. THERE’S NO
24 DISPARITY IN DEATH SENTENCING RATE THAT’S ASSOCIATED WITH THE
23 PRESENCE OF THAT FACTOR.
(£1
BALDUS - DIRECT
THE CONCEPT OF AN ASSOCIATION OR A CORRELATION IS A
TECHNICAL TERM THAT HAS THAT DISTINCT MEANING, THAT WHEN ONE
VARIABLE CHANGES. THE OTHER VARIABLE CHANGES. AND IT’S THE
DEGREE OF CHANGE THATS ASSOCIATED WITH THE CHANGE IN ONE
VARIABLE AND REFLECTED IN THE OTHER, IT IS THE KEY FROPERTY OF
THE SYSTEM THAT WE'RE TRYING TO MEASURE. AND THAT“S WHAT THESE
DIFFERENT MEASURES OF IMPACT ARE FOCUSED ON.
BY MR. BOGER:
?. PROFESSOR BALDUS -- EXCUSE ME?
A. YEAH.
@. BUT DOES, HOWEVER, COLUMN D. THE MEASUREMENT ONE, OR COLUMN
E, THE MEASUREMENT TWO. SUGGEST THE HYPOTHESIS THAT THE
PRESENCE OR ABSENCE OF DIFFERENT AGGRAVATING CIRCUMSTANCES. BZ,
B4, B&6 MIGHT HAVE DIFFERENT IMPACTS —-
A. YES.
R. == DON THE LIKELIHOOD OF RECEIVING A DEATH SENTENCE?
A. EXACTLY. IF YOU LOOK DOWN COLUMN D, YOU CAN SEE THAT IN ROW
3 WITH RESPECT TO THE B3 FACTOR, THAT THERE IS NO REAL
STATISTICAL EFFECT MEASURED THERE.
THE COURT: IT“S ALMOST AN INVERSE CORRELATION.
THE WITNESS: YES. AND WHEN IT“S THAT SMALL. WHEN THE
SAMPLES OF THAT MAGNITUDE WE TOOK THAT AS REALLY SHOWING NO
ASSOCIATION.
IF YOU COMPARE THAT WITH THE MEASURES ASSOCIATED WITH A
NUMBER OF THESE OTHERS, VIRTUALLY ALL OF THE OTHERS, YOU SEE
Py
H
d
e
e
A
p
D
N
10
11
— p—————— —— y— . —
680
BALDUS - DIRECT
VERY STRONG STATISTICAL EFFECTS.
SO THAT THATS THE NOTION THAT UNDERLIES THE IDEA OF AN
ASSOCIATION OR A STATISTICAL EFFECT IS THAT WHEN THE FACTOR IS
PRESENT IN THE CASE, THE RATE GOES UP, IT’S HIGHER.
SIGNIFICANTLY, THAN IS THE CASE WHEN THE FACTOR IS NOT PRESENT
IN THE CASE.
BY MR. BOGER:
@. WELL, NOW. YOU“VE INDICATED THAT THAT'S A HYPOTHESIS AND I
BELIEVE I ASKED YOU A MOMENT EARLIER WHETHER THAT HYPOTHESIS WAS
CONCLUSIVE AT THIS POINT, AND I BELIEVE YOUR ANSWER WAS "NO."
CAN YOU EXPLAIN FURTHER?
A. YES, I CAN. THE REASON ONE WOULD HAVE NO CONFIDENCE IN
CONCLUDING THAT THAT UNADJUSTED ASSOCIATION THAT WE SEE THERE
BETWEEN TWO VARIABLES PROVES OR STRONGLY SUGGESTS THAT THAT BS
FACTOR IS AN IMPORTANT INFLUENCE IN THE SYSTEM IS THAT THERE MAY
EE OTHER FACTORS THAT ARE REALLY PRODUCING THAT RESULT. THAT
THERE ARE WHAT WE CALL RIVAL HYPOTHESES PLAUSIBLE RIVAL
| HYPOTHESES THAT COULD EXPLAIN THAT. AND THERE ARE TWO MAIN
CATEGORIES OF PLAUSIBLE RIVAL HYPOTHESES.
IT IS POSSIBLE TO GENERATE A WHOLE SERIES OF RIVAL
HYPOTHESES. PHASES OF THE MOON COULD BE OFFERED AS AN
EXPLANATION FOR IT. WE DON’T PUT THAT IN THE CATEGORIES OF
FLAUSIBLE RIVAL HYPOTHESES.
WE HAVE A SERIES OF METHODS FOR ANALYZING PLAUSIBLE
RIVAL HYPOTHESES. THE MAJOR CATEGORY THAT WE ARE CONCERNED
BALDUS - DIRECT
WITH. OR A MAJOR CATEGORY IN THIS CASE THAT WE ARE CONCERNED
WITH IS THAT OTHER FACTORS THAT ARE FREQUENTLY PRESENT IN CASES
MAY BE THE CAUSAL FACTOR, AND THAT IT’S THEIR INFLUENCE IN
CONJUNCTION WITH THE FACTOR, FOR EXAMPLE, THE B8 FACTOR. THAT IS
REALLY PRODUCING THE ELEVATED DEATH SENTENCING RATE THAT WE SEE
IN THE B8 CASES.
THE NOTION IS THAT IT’S SOME OTHER FACTOR THAT'S
PRODUCING THE EFFECT, AND IT JUST HAPPENS TO BE AROUND WHEN THE
BS FACTOR IS PRESENT. AND THAT WHEN YOU LOOK AT AN UNADJUSTED
ASSOCIATION, THE STATISTICS GIVE THE CREDIT TO THE B8 FACTOR AS
AN EXAMPLE, WHEN IT’S REALLY SOME OTHER FACTOR THAT’S PRODUCING
THE RESULT.
THE, A COMMONLY REFERRED WAY OF CHARACTERIZING THIS,
DID THIS RELATIONSHIP BETWEEN THE B8 FACTOR AND THE DEATH
SENTENCING RATE IS A SPURIOUS RELATIONSHIP, IT DOES NOT REFLECT
A REAL EFFECT GOING ON IN THE SYSTEM.
BY MR. BOGER:
2. WOULD YOU TURN, PROFESSOR BALDUS, TO DB-64 FOR
IDENTIFICATION, -&4, AND LET ME ASK YOU IF YOU CAN IDENTIFY THAT
DOCUMENT?
A. YES. DB-44 IS A
THE COURT: 447?
MR. BOGER: -64. I“M GOING TO SKIP -42 AND -43 FOR A
MOMENT. YOUR HONCR.
THE COURT: ALL RIGHT.
emer porate eee. epee, Aree, ei, . eet, SA Fearon. te. sete
BALDUS - DIRECT
PY
THE WITNESS: DB-64 IS A TABLE WHICH ILLUSTRATES THE
TWO PRINCIPAL METHODS THAT ARE USED IN SCCIAL SCIENCES TO
CONTROL FOR RIVAL HYPOTHESES OF THE TYPES THAT I JUST DESCRIBED
TO YOU, THAT THERE IS ANOTHER FACTOR, ANOTHER BACKGROUND FACTOR
THAT HAS NOT BEEN TAKEN INTO ACCOUNT. AND THAT UNTIL IT IS, IT’S
IMPOSSIBLE TO DISENTANGLE POSSIBLE IMPACT THAT THESE FACTORS MAY
BE HAVING IN THE SYSTEM.
THIS TABLE IS TABLE H FROM OUR FINAL REPORT. AND IT
PRESENTS MEASURES OF THE IMPACT OF THE STATUTORY AGGRAVATING
Q
D
0
B
B
N
FACTOR BS ON DEATH SENTENCING RATES AFFAIR CONTROLLING FOR THE
11 |PRESENCE OF STATUTORY AGGRAVATING FACTOR B10, WHICH IS A MOTIVE
12 |TO AVOID ARREST.
12 |BY MR. BOGER:
14 |@. WHAT DOES THE TABLE IN ROMAN NUMERAL ONE ASK? WHAT
1S |QUESTION DID IT ADDRESS, RATHER?
16 |A. IT ADDRESSES THE QUESTION OF WHAT STATISTICAL EFFECT IS
17 |OBSERVED FOR THE BS FACTOR AFTER ONE CONTROLS FOR THE BACKGROUND
18 |FACTOR B10. THE BACKGROUND FACTOR B10, PRESENTS A PLAUSIBLE
4 19 |RIVAL HYPOTHESIS FOR THE REASON THAT MANY POLICE OFFICERS ARE
20 |KILLED IN THE LINE OF DUTY BY SOMEONE WHO IS SEEKING TO AVOID
21 |ARREST. THAT IS ALSO ONE OF THE STATUTORY AGGRAVATING FACTORS
22 |NUMBER B10.
23 $0 WE ASK THE QUESTION IS IT THE PRESENCE OF THE POLICE
24 |OFFICER IN THE CASE THAT’S THE IMPORTANT FACTOR, OR IS IT THIS
25 MOTIVE TO AVOID ARREST THATS THE IMPORTANT FACTOR.
&83
BALDUS - DIRECT
1 AND THE CROSS TABULATION METHODOLOGY THAT WE HAVE HERE
2 PROVIDES A WAY TO DISENTANGLE THE IMPACT OF THESE TWO VARIABLES
ON THE DEATH SENTENCING RATES.
NOW, THE WAY IT WORKS IS QUITE STRAIGHTFORWARD. WE
3
4
5 TAKE THE POPULATION OF CASES AND WE CONTROL FOR THE B10 FACTOR.
& AND WE DO THAT BY BREAKING THE CASES DOWN INTO TWO CATEGORIES,
7 WHICH ARE. ILLUSTRATED IN COLUMNS B AND C OF DB-64.
8 IN COLUMN B, WE HAVE THE CASES WITH A DB FACTOR
4 PRESENT.
10 IN COLUMN C, WE HAVE THE CASES WHERE THE DB10 FACTOR IS
11 NOT PRESENT.
12 THEN WITHIN EACH OF THOSE SUB CROUFS OF CASES. WE
13 REFEAT THE ANALYSIS THAT WE SAW IN THE PRECEDING DISCUSSION,
14 THAT 1S. WE CALCULATE AN UNADJUSTED MEASURE OF THE IMPACT OF THE
1S BS FACTOR. AND WHEN WE DO THAT WITHIN THE CASES THAT ARE
16 ORGANIZED UNDER COLUMN B, WE SEE THAT AMONG THE CASES INVOLVING
17 A B10 FACTOR, THE DEATH SENTENCING RATE AMONG THE CASES WITH A
18 BS FACTOR ALSO PRESENT IS .3%.
Re 1? @. IN OTHER WORDS, THAT MEANS THESE ARE CASES, THESE 21 CASES
20 IN THE PARENTHESIS?
21 A. YES,
22 R. THE CASES IN WHICH THIS IS BOTH A B8 AND A B10 FACTOR?
23 A. THAT’S RIGHT.
24 @. AND WHAT IS THE 8, THE NUMERATOR IN THE PARENTHESIS. STAND
23 FOR?
£84
BALDUS - DIRECT
1 A. THAT REPRESENTS THE NUMBER OF THOSE CASES THAT RECEIVED THE
DEATH SENTENCE.
@. AND WHAT'S THE .397
2
3
4 |A. THAT’S THE RATE THAT WE CALCULATED. THAT’S THE MEASURE OF
5 | DEATH SENTENCING AMONG THAT POPULATION OF CASES.
6 THE COURT: THE RATIO.
7 THE WITNESS: THAT”S RIGHT. YOUR HONOR RATIO OR RATE.
8 |BOTH EQUIVALENT.
9 AND THEN WE LOOK DOWN HERE AT THE CASES WHERE THE B10
10 |FACTOR WAS PRESENT, BUT THE BS FACTOR WAS NOT PRESENT. AND WE
11 SEE THAT THE DEATH SENTENCING RATE IN THAT SUB-GROUP OF CASES
12 |wAsS .35.
13 THE COURT: NOW WAIT A MINUTE. THE NUMBER OF CASES IN
14 WHICH THE B8 FACTOR WAS NOT PRESENT?
15 THE WITNESS: THAT‘S RIGHT.
146 THE COURT: BUT B10 FACTOR WAS?
17 THE WITNESS: YES. |
18 THE COURT: ALL RIGHT.
® 19 THE WITNESS: THAT'S A POPULATION OF CASES OF ESTIMATED
20 CASES OF 144. AND OF THOSE CASES, FIFTY RECEIVED THE DEATH
21 SENTENCE, PRODUCING AN OVERALL DEATH SENTENCING RATE FOR THOSE
22 CASES OF .335.
23 NOW WE, WE LOOK WITHIN THIS B10 FOPULATION OF CASES,
24 AND WE SAY HOW IMPORTANT IS THE BS FACTOR WITHIN THAT SUB-GROUP
23 OF CASES.
685
BALDUS ~ DIRECT
BY MR. BOGER:
Q. YOU SAY THIS B10 POPULATION OF CASES. YOU MEAN THE
POPULATION IN COLUMN 2, WHERE THERE IS A B10 FACTOR PRESENT?
A. YES. IT’S LABELED COLUMN B ON THE TABLE, BUT IT’S
SECOND COLUMN OVER. THAT“S RIGHT.
AND WE LOOK WITHIN THOSE CASES AND WE COMPARE THE RATE
WHEN THE BS FACTOR IS PRESENT .AND WHEN IT“S ABSENT, AND WE SEE
THERES A FOUR PERCENTAGE POINTS DIFFERENCE. HERE WE‘RE
APPLYING, YOUR HONOR, WHAT WAS LABELED MEASURE ONE ON THE
PRECEDING TABLE WHERE WE WERE ILLUSTRATING THE MEASURES. SHOWS
THE ARITHMETIC DIFFERENCE IN THE DEATH SENTENCING RATE.
AMONG THESE CASES. THOSE WHO HAVE POLICE OFFICER
VICTIMS, HAVE A FOUR PERCENTAGE POINT HIGHER CHANCE OF DEATH
SENTENCE THAN THOSE WHO DON’T HAVE A POLICE OFFICER VICTIM.
@. SO IF YOU KILLED WHILE AVOIDING ARREST, IT MAKES ONLY A
FOUR-POINT DIFFERENCE WHETHER OR NOT THE PERSON YOU KILLED WAS A
POLICE OFFICER. IS THAT ANOTHER WAY TO RESTATE IT?
A. YES. FOUR PERCENTAGE POINTS.
@. FOUR PERCENTAGE POINT DIFFERENCE.
THE COURT: NOW, WAIT A MINUTE. IS THAT RIGHT. MR.
BOGER?
MR. BOGER: WELL, THAT’S. I HATE TO BE TESTIFYING.
ILL ASK PROFESSOR BALDUS THAT.
THE COURT: IT WOULD SEEM TQ ME THAT THAT MIGHT BE
SHOWN BETTER BY COLUMN C, WHERE THE B10 FACTOR IS NOT PRESENT.
686
BALDUS - DIRECT
Pe
WOULDN'T IT?
MN
BY MR. BOGER:
@. PROFESSOR BALDUS, WHY DON’T YOU ADDRESS THAT QUESTION? I“VE
STRUGGLED WITH IT A LONG TIME MYSELF.
THE COURT: LET ME TELL YOU WHAT I THINK I UNDERSTAND
AND YOU CAN COMMENT ON IT.
IT WOULD SEEM TO ME IF YOU WERE CONTROLLING FOR THE B10
FACTOR IN THE POLICE ESCAPE SITUATION, THEN IF YOU TOOK OUT
THOSE CASES WHERE YOU HAD B10 WITH BS, AND LOOKED AT THE RATE
Pr
ot
+
SO
E
R
E
SO
RE
LY
Ge
S
C
R
a
Ea
WHERE YOU ONLY HAD BS, THAT THAT WOULD BE THE TRUEST MEASURE OF
11 THE EFFECT THAT B10 WAS HAVING.
12 THE WITNESS: NO, YOUR HONOR. WE“RE NOT TRYING TO
13 ESTIMATE THE IMPACT OF B10 HERE. YOU SEE, WE“RE TRYING TO
14 ESTIMATE THE IMPACT OF BS. YOURE QUITE RIGHT IF OUR OBJECT WAS
13 TO MEASURE THE IMPACT OF B10, AND THAT. WE COULD REORGANIZE THE
146 TABLE VERY SIMPLY TO DO THAT.
17 BUT HERE THE FOCUS IS ON THE IMPACT OF BS, CONTROLLING
18 FOR B10. 50 THE WAY WE DO THAT. WE LOOK AT THE IMPACT OF BS
® 12 AMONG THE CASES WHERE B10 IS PRESENT, AND THE IMPACT AMONG THE
20 CASES WHERE B10 IS NOT PRESENT. AND THEN WE SEE WHETHER OR NOT
21 B28 IMPACT HOLDS UP.
22 YOU RECALL WE SAW A 23 PERCENTAGE POINTS DISPARITY IN
23 THE DEATH SENTENCING RATES ASSOCIATED WITH A B& FACTOR WHEN WE
24 DIDN'T CONTROL FOR ANYTHING.
25 THE COURT: ALL RIGHT, I SEE WHAT YOU'RE SAYING.
rr ap——e———
rd
a
Lomas se ———
687
BALDUS — DIRECT
THE WITNESS: BUT THE THEORY WAS THAT THE ELEVATED
DEATH SENTENCING RATE IS REALLY A REFLECTION OF THE FACT THAT
THERE ARE A LOT OF B10 PROPERTIES IN THOSE CASES AS WELL. WHEN
WE LOOK AT THIS, THAT/S EXACTLY WHAT THAT INDICATES. IS THAT
ITS ONLY IN THE BS CASES. WITH B10 IS ALSO PRESENT, THAT THE
DEATH SENTENCING RATE FOR THE BS CASES IS VERY HIGH.
BUT WE ALSO SEE THAT THE B10 FACTOR PRODUCES A HIGH
DEATH SENTENCING RATE. .35, EVEN WHEN THE BS FACTOR ISN‘T THERE.
THE COURT: YOU’RE TALKING ABOUT COLUMN C NOW.
THE WITNESS: NO, B.
MR. BOGER: B. ON THE ANSWER TO THE BS FACTOR, NO.
THE WITNESS: LET ME TRY AGAIN, YOUR HONOR.
THE COURT: THERES AN INCREMENTAL EFFECT BY HAVING THE
POLICE OFFICER IN THE CASE, YES OR NO?
THE WITNESS: YES. WHEN THERE’S A B10 FACTOR PRESENT.
BUT IT“S A RELATIVELY SMALL INCREMENT, FOUR PERCENTAGE POINTS.
THE COURT: OKAY.
THE WITNESS: AND THEN WE GO OVER AND LOOK AT THE CASES
WHERE THE B10 FACTOR IS NOT PRESENT, AND WE SEE THAT ONCE YOU
TAKE THE B10 FACTOR OUT OF THE CASE ALTOGETHER. THAT IN FACT THE
DEATH SENTENCING RATE IS LOWER AMONG THE POLICE VICTIM CASES
THAN IT IS AMONG THE OTHER CASES, SUGGESTING THAT AMONG THIS
POPULATION OF CASES AS WELL, THE BS FACTOR DOES NOT HAVE A BIG
EFFECT ON DEATH SENTENCING RATES.
BY MR. BOGER:
488
BALDUS - DIRECT
TY
@. S00, PROFESSOR BALDUS. WHAT HAD BEEN THE UNASSOCIATED. OR.
2 EXCUSE ME.» UNADJUSTED MEASURE OF IMPACT OF B8 AS REFLECTED IN
3 DB-617
A A. THE UNADJUSTED MEASURE, USING THE SAME MEASURES, WERE A 23
3 PERCENTAGE POINT DIFFERENCE IN DEATH SENTENCING RATES.
@ 6 RA. WHAT IS THE ADJUSTED F IGURE, NOW. CONTROLLING FOR THE B10
7 STATUTORY AGGRAVATING CIRCUMSTANCE?
8 A. LOOKING AT THE CROSS TABULATION ALONE. YOU CANNOT GET ONE
9 NUMBER THAT GIVES YOU THAT MEASURE. YOU HAVE TO USE AN
10 ALTERNATIVE METHOD TO DO THAT. BUT YOU CAN SEE BY LOOKING AT
11 THE TWO DIFFERENT MEASURES THAT ARE PRODUCED WITHIN THE TWO
12 SUBJECT GROUPS OF CASES, THAT IS THE DIFFERENCE MEASURE THAT'S
13 CALCULATED AMONG THE CASES WHERE B10 IS PRESENT. AND THE
14 DIFFERENCE MEASURE CALCULATED AMONG THE CASES WHERE IT“S ABSENT
13 THAT YOU CAN SEE THAT THE IMPACT IS SUBSTANTIALLY REDUCED AMONG
1&6 THESE TWO POPULATIONS OF CASES. SO THAT WE COULD MAKE AN
17 INFERENCE FROM THIS THAT THE OVERALL EFFECT IS PROBABLY GOING TO
13 BE SOMEWHERE BETWEEN FOUR POINTS AND MINUS THREE POINTS,
% 19 PROBABLY CLOSER TO MINUS 3 POINTS BECAUSE THERE ARE MORE CASES
20 | IN THE CATEGORY WHERE THE B10 FACTOR IS NOT PRESENT. SO THAT
21 OVERALL, WE WOULD EXPECT TO SEE THAT THE RATE WOULD FROBABLY BE
22 SOMEWHERE BETWEEN THESE TWO ESTIMATES.
23 @. YOU MENTIONED THERE WAS ANOTHER WAY TO MEASURE THIS EFFECT
24 OF ANOTHER CONTROL OR OF A BACKGROUND FACTOR. WHAT IS THAT
25 OTHER METHOD OF MEASUREMENT?
689
BALDUS - DIRECT
THE COURT: ARE YOU ABOUT TO GC INTO THE MULTIPLE
PW
N
2 REGRESSION?
3 MR. BOGER: THAT-S RIGHT. YOUR HONOR.
4 THE COURT: LET“S TAKE A BREAK TILL ABOUT. OH. FIVE
3 MINUTES AFTER THE HOUR.
® A oh
7 (RECESS TAKEN.)
8 THE COURT: ALL RIGHT.
? - i -
10 DAVID C. BALDUS,
11 BEING PREVIOUSLY DULY SWORN. RESUMED THE WITNESS STAND AND
12 TESTIFIED FURTHER AS FOLLOWS:
13 DIRECT EXAMINATION (CONT D)
14 BY MR. BOGER:
13 @. PROFESSOR BALDUS, YOU INDICATED THERE IS ANOTHER METHOD TO
146 MEASURE THE IMPACT OF THE DEATH SENTENCING RATE, CONTROLLING FOR
17 ANOTHER FACTOR?
18 A. YES. THAT PROCEDURE IS KNOWN AS REGRESSION ANALYSIS. AND
® 19 BEFORE I TRY AND DESCRIBE MULTIPLE REGRESSION, I WOULD LIKE TO
20 MAKE AN ATTEMPT AT WHAT WE CALL BIVARIATE REGRESSION, THAT IS.
21 WHAT WERE TRYING TO MEASURE THE ASSOCIATION WITH A REGRESSION
22 PROCEDURE BETWEEN A SINGLE VARIABLE AND, THAT MIGHT EXPLAIN
23 THINGS, AND SINGLE OUTCOME VARIABLE.
24 REGRESSION ANALYSIS IS A COMPUTATIONAL PROCEDURE THAT
23 DESCRIBES HOW THE AVERAGE OUTCOME IN A PROCESS. IN THIS CASE,
es —. gp——— Fett. BO——— teeter. ety Vee sre ete —
620
BALDUS ~ DIRECT
WL
THE DEATH SENTENCING RATE. IS RELATED TO A PARTICULAR
CHARACTERISTIC OF THE CASES IN THE SYSTEM. 3)
3 THE MEASURE OF THAT RELATIONSHIP IS WHAT WE CALL THE
4 |REGRESSION CO-EFFICIENT. AND I ILLUSTRATED IT EARLIER WHEN WE
S |EXAMINED THE B2 FACTOR. WHERE WE WERE LOOKING AT THE UNADJUSTED
“ & |MEASURE OF IMPACT. AND WE SAW THAT THERE WAS A NINETEEN
2 | PERCENTAGE POINTS DIFFERENCE IN THE RATES AT WHICH B2 CASES WERE
8 |SENTENCED TO DEATH AS OPPOSED TO OTHER CASES.
9 THE REGRESSION CO-EFFICIENT FROM A LEAST SQUARES
10 |ANALYSIS. THATS JUST AN ARTFUL TERM THAT DESCRIBES THE TYPE OF
11 |REGRESSION ANALYSIS, THE REGRESSION CO-EFFICIENT IN THAT CONTEXT
12 |1S PERFECTLY ANALOGOUS TO THE ARITHMETIC DIFFERENCE BETWEEN
13 | THOSE TWO RATES.
14 THE LEAST SQUARES REGRESSION CO-EFFICIENT FOR A SINGLE,
1S |RATHER THAN FOR A BIVARIATE REGRESSION ANALYSIS WHERE THERE’S
16 |ONLY ONE EXPLANATORY VARIABLE IN THE ANALYSIS IS .19. AND THAT
17 |SHOWS THAT. TELLS US THAT FOR THE B2 CASES. THE DEATH SENTENCING
13 RATE IS NINETEEN PERCENTAGE POINTS HIGHER THAN IT IS FOR OTHER
% 19? CASES. TELLS YOU THE SAME THING AS 1 INDICATED AS THE
20 ARITHMETIC DIFFERENCE.
21 NOW THE GREAT STRENGTH OF THE REGRESSION PROCEDURE IS
22 THAT YOU CAN GET A SINGLE MEASURE OF IMPACT OF THE VARIABLE
23 WHILE ADJUSTING OR CONTROLLING FOR BACKGROUND VARIABLES. AND
24 THIS IS WHERE THE CONCEPT OF MULTIPLE REGRESSION ENTERS THE
23 PICTURE.
[O
Y
S
E
Se
d,
S
i
S
W
B
E
R
B
E
eR
671
BALDUS - DIRECT
@. CAN YOU USE THE EXAMPLE THAT WE WERE LOOKING AT IN DB-64 TO
EXPLAIN WHAT YOU MEAN BY BACKGROUND VARIABLE?
A. YES. IN DB-44, THE BACKGROUND VARIABLE THAT WE WERE
ATTEMPTING TO CONTROL FOR OR HOLD CONSTANT AS IT’S SOMETIMES
DESCRIBED. WAS THE B10 VARIABLE. AND WE SAW WHEN WE DID THAT
USING THE REGRESSION METHOD THAT THE EFFECT OF THE B8 VARIABLE
DELINED, BUT WE DIDN‘T GET ONE MEASURE OR ONE NUMBER THAT WOULD
DESCRIBE WHAT ITS IMPACT WAS AFTER CONTROLLING FOR THAT B10
VARIABLE.
@. EXCUSE ME, PROFESSOR BALDUS, I THINK YOU JUST TESTIFIED THAT
YOU SAW, USING THE REGRESSION METHOD. DO YOU MEAN PERHAPS
THE CROSS TABULATION METHOD?
A. I’M SORRY. IF THAT’S WHAT I SAID. I STAND CORRECTED.
WHEN WE DID IT WITH THE CROSS TABULATION METHOD WE DID
NOT PRODUCE A SINGLE MEASURE THAT SHOWED THE EFFECTS OF THE BS
FACTOR AFTER HOLDING CONSTANT THE CASES ON THE VARIABLE B10.
AND THE STRENGTH OF THE REGRESSION PROCEDURE IS THAT IT
ALLOWS YOU TO DEVELOP A SINGLE MEASURE THAT INDICATES THE IMPACT
IN THIS SITUATION OF THE BS FACTOR AFTER CONTROLLING FOR THE
BACKGROUND FACTOR, B10.
AND THAT 1S ILLUSTRATED. YOUR HONOR, IN PANEL 2 OF
DB-44. IN COLUMN B, IN THAT PART OF THE EXHIBIT. YOU CAN SEE
THAT WE HAVE INDICATED THE WEIGHTED LEAST SQUARES REGRESSION
CO-EFFICIENT.
THE TERM "WEIGHTED" IN THIS CONTEXT REFERS TO THE FACT
BALDUS - DIRECT
fy
THAT WE-RE USING WEIGHTED DATA OF THE TYPE WE DESCRIBED EARLIER.
THAT IS. THIS REFLECTS THE ADJUSTMENTS FOR THE DIFFERENT RATES
AT WHICH WE SAMPLED IN DIFFERENT PARTS OF THE STATE.
THE CO-EFFICIENT THAT’S ASSOCIATED WITH THE BS FACTOR
WHICH IS IDENTIFIED IN COLUMN A, AS YOU CAN SEE THERE, IS .02.
THAT TELLS YOU THAT AFTER YOU CONTROL FOR THE BACKGROUND FACTOR.
B10, THAT THE AVERAGE DIFFERENCE IN DEATH SENTENCING RATE
| BETWEEN CASES INVOLVING A BS FACTOR AND CASES NOT INVOLVING A BS
g
n
o
e
T
s
w
a
n
FACTOR IS TWO PERCENTAGE POINTS. AND YOU CAN SEE IF YOU COMPARE
a
Oo
THAT FIGURE WITH THE TWO DIFFERENCE MEASURES WE OBZERVED UP IN
TY
pb
THE CROSS TABULATION ANALYSIS IN THE FIRST PANEL OF DB-64, YOU
12 CAN SEE THAT THAT REGRESSION CO-EFFICIENT OF .02 FALLS JUST
13 ABOUT MIDWAY BETWEEN THE TWO MEASURES THAT WE SAW IN COLUMN B
14 AND C. THAT IS. BETWEEN FOUR POINTS AND MINUS 3 POINTS.
15 THE REGRESSION CO-EFFICIENT GIVES YOU THAT WEIGHTED
16 AVERAGE THAT THE CROSS TABULATION METHOD DOES NOT. AND THATS
17 ONE OF ITS GREAT PROPERTIES. THAT IT ENABLES YOU TO GET A SINGLE
18 NUMBER WHICH REFLECTS ITS IMPACT AFTER HOLDING CONSTANT OTHER
4 19 BACKGROUND FACTORS THAT YOU WANT TO CONTROL FOR.
20 THE COURT: NOW, WHAT DOES THAT MEAN, THAT YOU’VE GOT A
21 TWO PERCENT HIGHER CHANCE OF GETTING A DEATH PENALTY IF B3 IS
22 PRESENT?
23 THE WITNESS: TWO PERCENTAGE POINTS HIGHER THAN -——
24 THE COURT: THAN NORM?
8. THE WITNESS: THAT’S RIGHT. AFTER YOU ADJUST FOR THAT
693
BALDUS - DIRECT
BACKGROUND FACTOR BE CASES HAVE A TWO PERCENTAGE POINT HIGHER
CHANCE OF GETTING A DEATH SENTENCE THAN CASES THAT DON‘T HAVE BS
IN THEM AFTER YOU CONTROL FOR B10.
AND AS YOU CAN SEE WHEN WE STARTED OUT WITH AN ORIGINAL
UNADJUSTED IMPACT OF, FOR BS. OF 23 PERCENTAGE POINTS, THAT WAS
THE ORIGINAL MEASURE THAT WE OBTAINED WITH AN UNADJUSTED
REGRESSION CO-EFFICIENT, YOU CAN SEE THAT WHEN YOU INTRODUCE THE
BACKGROUND CONTROL FOR THE B10 FACTOR THAT THE IMPACT OF BS
FACTOR DROPS DRAMATICALLY.
WHATS HAPPENING HERE IS THAT WE SAY THE B10 FACTOR
EXPLAINS THE EFFECT THAT WE SAW EARLIER IN THE BS FACTOR. THAT
BS FACTOR IS SHARPLY DIMINISHED WHEN WE CONTROL FOR THE
BACKGROUND FACTOR. B10.
THE COURT: OR TO SAY IT IN ANOTHER WAY, IN THE PUREST
CASE. MEASURING AT LEAST THE JURY‘S REACTION AND YOU CONTEND THE
WHOLE SYSTEM‘S REACTION. IN A PURE CASE. YOU WOULD EXPECT A
PERSON CHARGED WITH KILLING A POLICE OFFICER TO HAVE A TWO
PERCENT GREATER CHANCE OF GETTING THE DEATH PENALTY THAN THE
AVERAGE PERSON WHG IS SUBJECTED TO THE DEATH PENALTY.
THE WITNESS: YES, TWO PERCENTAGE POINTS HIGHER.
THE COURT: THAN THE AVERAGE.
THE WITNESS: THAN ——
THE COURT: THAN THE AVERAGE PROBABILITY.
THE WITNESS: WELL. NOT NECESSARILY THE AVERAGE, BUT THE
AVERAGE, THE PROBABILITY AMONG ALL OTHER CASES THAT DON'T HAVE
Fa
y
kr
pt
O
M
0
YN
Oo
0
0
694
BALDUS - DIRECT
Ba.
THE COURT: OKAY.
THE WITNESS: THAT’S THE MESSAGE THAT CO-EFFICIENT
CARRIES FOR YOU. AND IT GIVES AN AVERAGE ACROSS ALL THE CASES.
IT GIVES YOU A GLOBAL MEASURE OF THE IMPACT OF THAT ONE VARIABLE
ON ALL THE DATA. |
THE COURT: LET ME ASK YOU A QUESTION. WE HAVE
OBSERVED. I THINK: IN THE RAW DATA. THAT B2, WHICH IS THE
CONCURRENT FELONY. IN RAW NUMBERS IS A LARGE PERCENTAGE OF THE
TOTAL DEATH PENALTY CASES. OR IS PRESENT IN A LARGE PERCENTAGE.
THE WITNESS: YES.
THE COURT: AND FROM THE RAW NUMBERS WE WOULD GATHER
THAT THE SYSTEM REACTS PARTICULARLY STRONGLY TO THAT.
THE WITNESS: YES, IT DOES.
THE COURT: SINCE SUCH A GREAT PROPORTION OF THE DEATH
PENALTY CASES CARRY A FACTOR WHICH THE SYSTEM SEEMS TO REACT
PARTICULARLY ADVERSELY TO, WOULD THAT. IN A SENSE, HAVE THE
EFFECT OF ARTIFICIALLY RAISING THE AVERAGE NON-BS RESPONSE. DO
YOU UNDERSTAND WHAT I°M SAYING?
THE WITNESS: NON-BS RESPONSE, OH. 1 SEE WHAT YOU MEAN.
YES. IT. THE NON-BS RESPONSE IS GOING TO REFLECT THE IMPACT OF
THE B2 CASES.
THE COURT: AND SINCE THERE ARE SO MANY OF THEM AND
SINCE SOCIETY REACTS SO STRONGLY THAT AT LEAST ON THIS TABLE,
WERE NOT REALLY COMPARING IT AGAINST AN AVERAGE RESPONSE
&95
BALDUS - DIRECT
1 BECAUSE THAT RESPONSE HAS BEEN HEIGHTENED BY HAVING A FACTOR IN
2 IT THAT SOCIETY REACTS STRONGLY AGAINST OR THE SYSTEM OR
3 WHATEVER. IS THAT FAIR?
THE WITNESS: THATS RIGHT. YOUR HONOR.
THE COURT: SO YOU WOULD THEN HAVE TO GET THE CONTROL
OUT FOR B2, TO GET A LITTLE BIT MORE AVERAGE SORT OF A FEELING
4
3
b
7 OF HOW JURIES AND PROSECUTORS REACT TO THE BS FACTOR.
8 THE WITNESS: EXACTLY. YOUR HONOR. AND WHAT YOUVE
2? IDENTIFIED IS THAT JUST CONTROLLING FOR ONE BACKGROUND FACTOR
0 DOESN’T TELL YOU THE WHOLE PICTURE. IT’S THE IMPACT OF A
11 MULTITUDE OF FACTORS IN A SYSTEM THAT DETERMINE WHAT HAPPENED.
12 AND TO GET, TO ESTIMATE PROPERLY THE EFFECT OF ANY ONE VARIABLE.
13 YOU’VE GOT TO CONTROL FOR THE BACKGROUND EFFECTS OF A VARIETY OF
14 OTHER FACTORS AT ONCE. YOU WANT TO IDENTIFY WHAT ARE THE
1S IMPORTANT FACTORS IN A SYSTEM AND CONTROL FOR THEM.
1& | THE COURT: YOUR ANSWER TO ME POINTS UP THE FROBLEM I'M
7 HAVING.
18 WHEN YOURE SAYING YOU ARE CONTROLLING FOR, FOR
“» 19 BACKGROUND FACTORS, I PRESUMED THAT YOU MEANT YOU WERE
20 CONTROLLING FOR BACKGROUND FACTORS IN THE CASE UNDER STUDY. THE
21 BS CASE, NOT CONTROLLING FOR BACKGROUND FACTORS IN THE NON-BS
22 POPULATION THAT YOU WERE MAKING A COMPARISON WITH.
23 THE WITNESS: WERE CONTROLLING HERE FOR ONE
24 BACKGROUND FACTOR: THE PRESENCE OF A B10 FACTOR.
25 WE“RE HOLDING THE CASES CONSTANT WITH RESPECT TO A B10 FACTOR
BALDUS - DIRECT
1 AND THAT'S ALL.
2 AND WE“RE LOOKING, THE CO-EFFICIENT MEASURES. THE
3 DIFFERENCE BETWEEN PRESENCE OF BS AND ITS ABSENCE, HOLDING
4 CONSTANT ON THE B10 FACTOR ONLY.
S THE COURT: I THINK I UNDERSTAND.
. & GO AHEAD.
7 BY MR. BOGER:
8 @. DOES MULTIPLE REGRESSION PERMIT YOU TO CONTROL FOR MORE THAN
? ONE BACKGROUND FACTOR AT THE TIME, PROFESSOR BALDUS?
10 A. YES, IT DOES.
11 THE COURT: I PRESUME THAT'S WHERE WE“RE GOING.
12 MR. BOGER: THAT’S EXACTLY WHERE WERE GOING, YOUR
13 HONOR. IF WE COULDN'T GO FURTHER THAN THIS, WE PROBABLY
14 WOULDN’T HAVE STARTED.
15 LET ME MOVE FOR THE ADMISSION OF DB-&4 INTO EVIDENCE AS
14 REFLECTING PROFESSOR BALDUS- TESTIMONY ABOUT THE IMPACT OF B10
37 ON BS.
13 MS. WESTMORELAND: YOUR HONOR, SUBJECT TO THE SAME
w 19 OBJECTIONS THAT I PREVIOUSLY NOTED, I HAVE NO FURTHER CBJECTION.
20 THE COURT: I TAKE IT THAT DB-64 IS MORE OF A
21 DEMONSTRATIVE EXHIBIT THAN IT IS AN EVIDENTIARY EXHIBIT IN THE
22 SENSE THAT TO GET THE FINDER OF FACT TO UNDERSTAND WHATS GOING
23 ON, THIS IS THE FIRST BUILDING BLOCK OF KNOWLEDGE.
24 IS THAT REALLY WHAT YOURE GETTING AT HERE?
23 MR. BOGER: YES, YOUR HONOR.
697
BALDUS - DIRECT
1 THE COURT: WE“RE NOT GOING TO DRAW ANY FINAL
2 CONCLUSIONS FROM DB-44, ARE WE?
3 MR. BOGER: NO, YOUR HONOR. I THINK THERE WILL BE
OTHER TABLES THAT REFLECT THE SAME KIND OF INFORMATION MORE
THOROUGHLY .
THE COURT: ALL RIGHT. WE“LL INTRODUCE DB-64 AS AN
EXHIBIT OF PROCESS.
MR. BOGER: THANK YOU, YOUR HONOR.
YM
0
.
N
A
BY MR. BOGER:
10 Q. DOES THE REGRESSION PROCEDURE GENERATE ANYTHING MORE THAN
11 THE NUMBERS THAT YOU“VE MENTIONED, PROFESSOR BALDUS?
12 A. YES, IT PRODUCES AN EQUATION.
13 @. LET ME ASK YOU TO TURN TO DB-45 FOR IDENTIFICATION, AND
14 IDENTIFY THAT.
13 A. 6&5, DB-&5, PRESENTS TWO EXAMPLES OF REGRESSION EQUATIONS.
1&6 THE FIRST ONE 1S A BIVARIATE REGRESSION EQUATION AND BY THAT 1
37 MEAN IT HAS, REPRESENTS THE VALUES ESTIMATED IN AN ANALYSIS
1a INVOLVING ONE OUTCOME VARIABLE AND ONE EXPLANATORY VARIABLE.
a 32 IN THE TERMINOLOGY OF THE FIELD, THE OUTCOME VARIABLE,
20 WHICH IN THIS CONTEXT WOULD BE THE DEATH SENTENCING RESULT, IS
21 KNOWN AS THE DEPENDENT VARIABLE.
22 THE OUTCOME MEASURE. AND IT“S DESIGNATED IN THIS
23 FORMULA BY THE SYMBOL "Y". AND IT IS THE ULTIMATE. THE ULTIMATE
24 POINT IF ONE IS TRYING TO MAKE AN ESTIMATE OF HOW LIKELY IT IS |
25 IF SOMEONE WITH CERTAIN CHARACTERISTICS WOULD RECEIVE A DEATH
— — ——— —— ——— S— F——n So — —— A —— ——— ere. Veen et . S—— — er . - ———r ——
498
BALDUS - DIRECT
SENTENCE. GIVEN THE CHARACTERISTICS OF THEIR CASE.
THE "A" IN THIS EQUATION IS A CONSTANT TERM THAT DOES
NOT CHANGE IN ANY OF THE ANALYSIS, IS NOT REALLY IMPORTANT FOR
THE PURPOSES OF QUR DISCUSSION HERE.
THE "X-1" IS THE SYMBOL FOR THE VARIABLE,
CONTEMPORANEOUS OFFENSE. AND "B-1" IS THE REGRESSION
CO~EFFICIENT FOR THAT VARIABLE.
AND THE REGRESSION PROCEDURE THAT WAS USED TO ESTIMATE
A CO-EFFICIENT. AND CONSTANT TERM FOR THESE VARIABLES IS SET OUT
IMMEDIATELY BELOW THE FORMULA. AND THE DEFINITION OF ITS TERMS.
AND IT READS. Y=.01 + .18 TIMES X1. SO THAT MEANS TO
ESTIMATE THE PROBABILITY OF A DEATH SENTENCE FOR SOMEONE WHO HAS
A CONTEMPORANEOUS OFFENSE, THAT IS "Y", WHAT IS THE AVERAGE
LIKELIHOOD THAT A PERSON WITH A CONTEMPORANEOUS OFFENSE. LOOKING
AT NO OTHER FACTORS. WILL RECEIVE A DEATH SENTENCE, WE SIMPLY
ADD UP THE .01 AND .18 TIMES THE CODING THAT IS ENTERED FOR THE
CONTEMPORANEOUS OFFENSE, WHEN THE CONTEMPORANEDUS OFFENSE IS
PRESENT, WE CODE THAT AS A "1". WHEN THERE“S NO EVIDENCE OF A
CONTEMPORANEOUS OFFENSE, ITS CODED AS ZERD.
SO IN THIS CASE, WE WOULD MULTIPLY THE .18 TIMES 1,
WHICH WOULD GIVE US .18. WE ADD THAT TO THE .01,. AND THAT GIVES
US THE ESTIMATED PROBABILITY OF THE .1%9, WHICH IS THE
CO-EFFICIENT FOR THE CONTEMPORANEOUS OFFENSE VARIABLE IN THE
LEAST SGIJARES PROCEDURE.
THE COURT: WHERE DID YOU GET "B-1" FROM?
6579
BALDUS - DIRECT
f
u
THE WITNESS: OH. THAT "B-1" HAS NOTHING TO DO WITH THE
STATUTE. IS THAT WHAT YOURE SUGGESTING, YOUR HONOR?
THE COURT: NO NO NO. I WASN‘T. I JUST WONDERED WHERE
YOU GOT THE NUMBER .187
THE WITNESS: OH, THE "B~1" IS EQUIVALENT TO, IN THIS
EXAMPLE, .18, THAT'S WHATS ESTIMATED BY THE REGRESSION
PROCEDURE, YOUR HONOR. THE REGRESSION PROCEDURE PRODUCES A
VALUE FOR THE "B-1", WHICH IS THE REGRESSION CO-EFFICIENT. THAT
"B-1" IS THE VALUE THAT WAS PRESENTED OVER ON DB-&4 AS THE
no
CO
RE
E
GR
E
E
,
GR
R
Y
H
e
B
E
R
(R
OR
WEIGHTED LEAST SQUARES REGRESSION CO-EFFICIENT. THAT’S THE
11 EQUIVALENT THERETO. IN THIS EXAMPLE.
12 BY MR. BOGER:
13 @. DOES THIS EQUATION. PROFESSOR BALDUS. REFLECT HOW THE EQUATION
14 WAS SOLVED, AND HOW THAT CO-EFFICIENT WAS OBTAINED OR DOES IT
15 SIMPLY REFLECT THE OUTCOME?
16 A. NO, IT’S JUST THE RESULTS. THIS IS WHAT YOU GET. YOUR
17 HONOR, WHEN YOU LOOK AT A PIECE OF COMPUTER PRINTOUT, IT
18 REFLECTS THE ANALYSIS THAT-S BEEN DONE BY THE MACHINE.
%® 19 THE COURT: ALL RIGHT.
20 . THE WITNESS: AND PART TWO. I JUST MENTION IN PASSING,
21 IS AN EXTENSION OF THAT. IT INDICATES HOW THIS PROCEDURE CAN BE
22 REPRESENTED. THE RESULTS OF THE PROCEDURE CAN BE REPRESENTED
23 WHEN ONE INCLUDES MORE THAN ONE INDEPENDENT MEASURE.
24 HERE WOULD BE A SITUATION WHERE WE WOULD HAVE AN
23 ADDITIONAL BACKGROUND FACTOR CONTROLLED FOR BEYOND THE VARIABLE
C
B
B
a
a
B
E
S
t
1
E
e
C
R
pt
oO
700
BALDUS - DIRECT
X~1 AND CO-EFFICIENT FOR IT. WHICH I3 "B-1". HERE WE HAVE
MULTIPLE REGRESSION ANALYSIS, WHEREAS THE FIRST EXAMFLE WAS
KNOWN AS BIVARIATE REGRESSION ANALYSIS BECAUSE IT INVOLVES
ONLY TWO VARIABLES.
@. PROFESSOR BALDUS, WHEN YOU CONTROL FOR EIGHT, TEN OR TWELVE
FACTORS, DO YOU GENERATE AN EQUATION THEN WITH EIGHT OR TEN OR
TWELVE CO-EFFICIENTS NEXT TO THE X-1“S AND X-27S, ET CETERA. IS
THAT THE WAY THAT WORKS?
A. EXACTLY. THE RESULTS OF THE ANALYSIS WILL GIVE YOU A
CO-EFFICIENT FOR EACH VARIABLE. AND THAT CO-EFFICIENT WILL TELL
YOU WHAT IS THE IMPACT STATISTICALLY OF THAT VARIABLE AFTER ONE
HOLDS CONSTANT ALL THE OTHER FACTORS THAT ARE INCLUDED IN THE
ANALYSIS.
MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE THE
ADMISSION OF DB-45 INTO EVIDENCE AS DEMONSTRATIVE EVIDENCE
ILLUSTRATING PROFESSOR BALDUS‘ TESTIMONY.
MS. WESTMORELAND: YOUR HONOR. I WOULD LIKE SOME
INDICATION AS TO WHETHER PROFESSOR BALDUS PREPARED THIS EXHIBIT
OR WHETHER THERE WAS SOMETHING DONE BY PROFESSOR WOODWORTH TO
CLARIFY WHETHER THIS WAS IN HIS AREA OF EXPERTISE.
MR. BOGER: I-LL BE HAPPY TO ASK HIM A FEW QUESTIONS ON
THAT SCORE.
BY MR. BOGER:
@. PROFESSOR BALDUS, WHO PREPARED THIS DOCUMENT?
A. 1 DID.
BALDUS - DIRECT
@. DID YOU RELY ON YOUR OWN KNOWLEDGE?
A. YES.
MR. BOGER: NOD FURTHER QUESTIONS ON THAT.
MS. WESTMORELAND: THEN AS A DEMONSTRATIVE EXHIBIT.
YOUR HONOR, I WOULD HAVE NO OBJECTION.
THE COURT: IT WILL BE ADMITTED.
BY MR. BOGER:
2. PROFESSOR BALDUS, YOU INDICATED THAT IT MAY BE POSSIBLE TO
CONTROL FOR MORE THAN ONE BACKGROUND VARIABLE AT A TIME.
CAN YOU TELL ME WHAT METHODS ONE MIGHT USE TO HAVE SUCH
CONTROL?
A. YES. THERE ARE TWO PRINCIPAL METHODS ONE CAN USE. THE
CROSS TABULATION METHOD, AND MULTIREGRESSION METHOD.
@. THE SAME TWO METHODS WE TALKED ABOUT WHEN THERE WAS ONLY ONE
BACKGROUND CONTROLLED FOR?
A. YES, THESE ARE THE TWO PRINCIPAL METHODS THAT ARE USED IN
SOCIAL SCIENCE TO CONTROL FOR THE BACKGROUND FACTORS WHILE
TRYING TO ESTIMATE THE IMPACT STATISTICALLY OF ONE VARIABLE ON
ANOTHER.
@. LET ME ASK YOU TQ TURN TO DB-47, MARKED FOR IDENTIFICATION.
CAN YOU IDENTIFY THAT DOCUMENT?
A. YES. THIS IS AN EXAMPLE DRAWN FROM OUR REPORT OF A CROSS
TABULATION ANALYSIS WHICH CONTROLS SIMULTANEOUSLY FOR TWO
BACKGROUND FACTORS.
THE OBJECT OF THIS ANALYSIS WAS TO ESTIMATE THE IMPACT
702
BALDUS - DIRECT
Ty
OF A VARIABLE WHETHER A DEFENDANT SURRENDERED WITHIN 24 HOURS ON
THE DEATH SENTENCING RATE WHILE CONTROLLING FOR THE BACKGROLIND
VARIABLES OF WAS THERE A SERIOUS PRIOR RECORD ASSOCIATED WITH
THE DEFENDANT IN THE CASE, AND DID THE CASE INVOLVE A
FAMILY-LOVER-LIQUOR OR BAR ROCM QUARREL.
AND THE WAY THE PROCEDURE WORKS IS STRAIGHTFORWARD. AT
THE TOP, YOU CAN SEE WE HAVE THE ENTIRE POPULATION OF CASES.
AND THE DEATH SENTENCE RATE CALCULATED WITHIN THAT GROUP OF
0
o
N
e
o
s
W
N
CASES.
pr
o @. I TAKE 1T THE ENTIRE POPULATION OF CASES IS REFLECTED WITHIN
Po
d
pt
THE PARENTHESIS?
12 A. YES. WITH THE EXCEPTION OF ONE DEATH SENTENCE CASE. THERE
13 WAS ONE ESTIMATED DEATH CASE THAT WAS LOST BY ROUNDING HERE.
14 I INDICATED EARLIER ON THAT THERE WERE, WE HAD ALL THE
15 DEATH SENTENCE CASES IN THE UNIVERSE OF CASES WITH ONE EXCEPTION
1&6 TO THAT. THERE WAS ONE CASE WE WERE UNABLE TO LOCATE IN THE
17 FILES OF THE BOARD OF PARDONS AND PAROLES. SO ACTUALLY WE HAVE
13 127 OUT OF 128, AND SOME OF THE ANALYSES, BY VIRTUE OF THE
( 19 ROUNDING INVOLVED WITH RESPECT TO THAT ONE CASE, IT DROPS QUT OF
20 THE ANALYSIS ON OCCASION. AND THATS WHY THERE ARE ONLY 127
21 HERE, WHEREAS OUR UNIVERSE IS ACTUALLY 128.
22 THEN WE BREAK DOWN THE CASES ACCORDING TO WHETHER OR
23 NOT A QUARREL OF THIS TYPE EXISTED IN THE CASE. AND YOU CAN SEE
24 THAT THE DEATH SENTENCING RATE AMONG CASES INVOLVING SUCH
23 QUARRELS IS QUITE LOW. ONLY 11 DEATH SENTENCES WERE IMPOSED IN
Ty
Ww
nN
703
BALDUS - DIRECT
SUCH CASES. SUBSTANTIALLY HIGHER IN OTHER CATEGORIES OF CASES.
THAT IS TO CONTROL THAT VARIABLE. EACH TIME YOU
CONTROL ON A VARIABLE, IT RESULTS IN BREAKING UP, BREAKING DOWN
THE DATA SET INTO A NUMBER OF SUB GROUPS. IT“S A VARIABLE. HAS
ONLY TWO CATEGORIES. THIS ONE DOES. YOU BREAK DOWN THE FILE INTO
TWO SUB GROUPS.
THEN AT THE NEXT LEVEL, YOU CAN SEE THE EFFECTS OF
CONTROLLING ON THE SECOND BACKGROUND VARIABLE, WHICH IS SERIOUS
PRIOR RECORD. AND YOU CAN SEE THAT AMONG THE CASES INVOLVING A
FAMILY QUARREL THAT THE PRESENCE OF A SERIOUS RECORD DOES NOT
SUBSTANTIALLY AFFECT THE LIKELIHOOD OF A DEATH SENTENCE.
WHEN YOU GET OVER AMONG THE CASES THAT DID NOT INVOLVE
SUCH A QUARREL. YOU SEE THAT THE OVERALL RATE IS .07 WHEN THE
PRIOR RECORD WAS NOT PRESENT, AND .13. SO THAT VARIABLE SEEMS
TO HAVE AN EFFECT ON THE SYSTEM.
AND NOW THE PRINCIPAL POINT ON THIS IS TO FOCUS ON THE
IMPACT OF WHETHER OR NOT THE DEFENDANT SURRENDERED WITHIN Z4
HOURS. AND THAT’S WHAT IS PRESENTED IN THE BOTTOM LINE OF THIS
TABULATION.
IN THE FIRST COLUMN, YOU CAN SEE THAT THERE IS
STATISTICALLY LITTLE EFFECT, EVEN AMONG THIS GROUP OF CASES.
FIVE DEATH SENTENCES OUT OF SIX THAT WERE IMPOSED IN THIS
CATEGORY, WERE IMPOSED AGAINST OFFENDERS WHO HAD NOT SURRENDERED
WITHIN 24 HOURS. BUT BECAUSE THE OVERALL DEATH SENTENCING RATE
IS SO LOW IN THIS CATEGORY. EVEN THOUGH ALL THOSE CASES, THOSE
704
BALDUS - DIRECT
CASES ARE DISPROPORTIONATELY REPRESENTED. THAT IS. THE DEATH
CASES ARE DISPROPORTIONATELY REPRESENTED IN THAT CATEGORY.
NUMBERS OF CASES. BOTH OF THESE CATEGORIES ARE SO LARGE. THAT
THOSE DO NOT PRODUCE ANY STATISTICAL EFFECT.
SIMILARLY. IN THE NEXT CATEGORY. WE SEE A VERY. VERY
SMALL DEATH SENTENCING RATE, AND NO STATISTICAL EFFECT
ASSOCIATED WITH THIS VARIABLE.
HOWEVER WHEN WE MOVE OVER INTO THE CATEGORY WHERE THE
DEATH SENTENCING RATES ARE HIGHER, WE SEE THAT THE FACT OF THE
SURRENDERING WITHIN 24 HOURS DOES PRODUCE A STATISTICAL EFFECT.
AND THAT IT IS SEVEN PERCENTAGE POINTS LOWER, THE DEATH
SENTENCING RATE IS SEVEN PERCENTAGE POINTS LOWER IN CASES WHERE
THE DEFENDANT DID SURRENDER THAN IN OTHER CASES, CONTROLLING FOR
THESE. I“M SORRY. WITHIN THIS SUB-GROUP OF CASES, WHICH IS
COMPARABLE WITH RESPECT TO THE KIND OF RUARREL., WHETHER THERE
WAS A QUARREL, AND WHETHER THE TYPE OF RECORD THAT THE OFFENDER
HAD. AND YOU CAN SEE IN THE MOST AGGRAVATED CLASS OF CASES OVER
HERE, CASES WHERE THERE WAS NO FAMILY QUARREL. AND WHERE THE
OFFENDER HAD A SERIOUS PRIOR RECORD. THAT IN THE CONTEXT OF THIS
GROUP OF CASES, THAT THE PRESENCE OR ABSENCE OF A SURRENDER
WITHIN 24 HOURS SEEMS TO HAVE A MUCH LARGER EFFECT.
@. WHAT’S THE ADVANTAGE OF THE USE OF CROSS TABULATION METHOD
HERE?
A. THE ADVANTAGE OF THIS METHOD I3 IT GIVES YOU A VERY
STRAIGHTFORWARD PICTURE OF HOW THE DEATH SENTENCING RATES VARY
pb
2
3
4
=
6
7
a
9
10
705
BALDUS - DIRECT
WITH RESPECT TO THE DIFFERENT COMBINATIONS OF VARIABLES.
YOU SEE THE PATTERN OF DEATH SENTENCING RATES AS YOU
MOVE FROM ONE GROUP TO ANOTHER, AND YOU CAN SEE AS YOU MOVE
ACROSS THE, THIS ROW. FOR EXAMPLE. LABELED "SERIOUS PRIOR
RECORD," YOU CAN SEE HOW THE DEATH SENTENCING RATE INCREASED AS
THE CASES BECOME MORE AGGRAVATED. IT GIVES INSIGHT INTO THE
OPERATION OF THE SYSTEM AND ALSO IT PROVIDES YOU A BASIS FOR
MAKING MEASUREMENTS OF THE DISPARITIES OF THE INDIVIDUAL
VARIABLES AT THE BOTTOM LINE AS IS INDICATED HERE.
@. IS THERE A LIMITATION ON THE CROSS TABULATION METHOD THAT
PERHAPS DOES NOT EXIST WITH A REGRESSION METHOD?
A. YES. THE BIG LIMITATION IS THAT YOU ARE REQUIRED IN THE USE
| OF THIS METHOD TO BREAK DOWN EACH CELL OF CASES EACH TIME YOU
INTRODUCE A NEW CONTROL VARIABLE, SO THAT EVEN THOUGH WE HAVE
HERE QUITE LARGE SAMPLE SIZES IN THE ROW LABELED “SERIOUS PRIOR
FELONY," YOU CAN SEE DOWN IN THE NEXT ROW WHERE WE HAVE BROKEN
THOSE GROUPS OF CASES IN TWO, CONTROLLING FOR THE DEFENDANTS
SURRENDERED WITHIN 24 HOUR VARIABLE; IF WE FURTHER BREAK THOSE
CATEGORIES DOWN. VERY S00N WE WOULD BE EXHAUSTING OUR SAMPLE.
AND WHAT I MEAN BY THAT IS THAT YOU WOULD HAVE INSUFFICIENT
GROUPS OF CASES.
THE COURT: YOU WOULDN’T GET STATISTICALLY SIGNIFICANT
RESULTS?
THE WITNESS: THAT’S RIGHT. YOUR HONOR. THERE WOULD BE
NO STABILITY IN THE ESTIMATES.
706
BALDUS - DIRECT
1 MR. BOGER: YOUR HONOR PLEASES. AT THIS TIME I MOVE THE
ADMISSION OF DB-47 INTO EVIDENCE AS DEMONSTRATIVE EVIDENCE.
ILLUSTRATIVE OF PROFESSOR BALDUS” TESTIMONY.
MS. WESTMORELAND: FOR THOSE PURPOSES. OTHER THAN THE
CONTINUING OBJECTION, I HAVE NOTHING FURTHER. YOUR HONOR.
THE COURT: ALL RIGHT. IT WILL BE ADMITTED AS
DEMONSTRATIVE OF THE PROCESS ABOUT WHICH HES TESTIFYING.
E
T
S
a
Le
EE
LET ME ASK YOU A QUESTION AT THIS POINT.
El MY QUESTION IS PREDICATED BASED SOLELY ON MY SENSE OF
10 WHAT MIGHT GO ON IN A TRIAL SUCH AS THE. OF THE TYPES THAT WE“RE
11 DISCUSSING.
12 I COULD UNDERSTAND THAT THE JURY. PERHAPS THE
13 PROSECUTOR, MIGHT VERY WELL KNOW SOMETHING OF THE CONTEXT OF THE
14 CRIME, SUCH AS YOUR FIRST SORT THERE ON FAMILY-LOVER-LIRQUOR OR
13 BAR ROOM.
16 I COULD UNDERSTAND THAT THERE MIGHT BE SOME
17 POSSIBILITY, AT LEAST WHERE A DEATH PENALTY IS IMPOSED. THAT THE
18 DECISION MAKERS MIGHT KNOW SOMETHING ABOUT THE PRIOR RECORD.
% 19 MY UNEDUCATED GUESS WOULD BE THERE’S PROBABLY LITTLE
20 LIKELIHOOD THAT AT LEAST THE JURY WOULD KNOW MUCH ABOUT
ik SURRENDER TIME. THEY MAY, MAY NOT. BUT THERE’S SOME DOUBT
22 ABOUT WHETHER OR NOT THEY WOULD KNOW IT IN EVERY CASE BY ANY
23 MEANS. THERE“S JUST NOTHING INHERENT IN THE PROCESS THAT WOULD
24 MAKE THAT COME OUT OR EVEN SUGGEST THAT IT WOULD COME OUT.
SO WHEN YOU GET SOMETHING THAT WOULD APFEAR TO BE AS XR)
(8
707
BALDUS - DIRECT
1 STATISTICALLY IMPORTANT AS WHETHER OR NOT THE DEFENDANT
2 SIJRRENDERED WITHIN 24 HOURS AND YOUR INSTINCTS AT LEAST TELL YOU
THERE IS SOME LIKELIHOOD IN THE FIRST INSTANCE IT MAY NOT MAKE
ANY DIFFERENCE TO THE PROSECUTOR AND IN THE SECOND INSTANCE. THE
JURY MAY NOT HAVE KNOWN ABOUT IT, WHAT DO YOU DO WITH THAT SORT
3
4
3
A 6 OF DOUBT?
YJ THE WITNESS: WE TREAT CASES HAVING THIS PROPERTY,
8 SURRENDERED WITHIN 24 HOURS, ONLY IN CIRCUMSTANCES WHERE IT’S
9 PLAIN FROM THE RECORDS WE CONSULTED THAT THAT OCCURRED.
10 WHAT WE’RE INTERESTED IN DOING IS IDENTIFYING
3131 AGGRAVATING AND MITIGATING FACTORS THAT THE RECORD INDICATES
12 WERE KNOWN TO THE PEOPLE PROCESSING THE CASES, LIKELY KNOWN TO
13 THE PEOPLE PROCESSING THE CASES.
14 IF THERE“S NO EVIDENCE ONE WAY OR ANOTHER ABOUT THE
15 MATTER, WE TREAT THAT AS NOT A FACTOR IN THE CASE, BECAUSE IF
16 THERE’S NO EVIDENCE TO SUGGEST THAT WAS A FACTOR IN THE CASE. IT
17 IS UNLIKELY. IT”S IMPOSSIBLE FOR IT TO HAVE HAD ANY INFLUENCE
18 OVER THE DECISION. IF NO ONE KNEW ONE WAY CR ANOTHER WHEN THE
4 19 PERSON SURRENDERED, WHICH IS THE CIRCUMSTANCE IN MANY CASES. NOT
20 BROUGHT QUT, THAT WOULD NOT SERVE AS A MITIGATING FACTOR IN THAT
22 THE COURT: WELL, THERE‘S ABSOLUTELY NO TESTIMONY TO
23 THE EFFECT THAT THE JURY HAD ANY INFORMATION ABOUT WHEN THE
24 DEFENDANT SURRENDERED. AND THE PROCESS THAT YOU EMPLOYED WOULD
23 SUGGEST THAT IT WOULD NOT LIKELY BE ADEQUATE TO GET THAT SORT OF R
e
t
i
n
a
it
e
i
n
e
s
m—
—
BALDUS - DIRECT
1 INFORMATION, BECAUSE YOU REALLY DIDM’T LOOK AT THE TRANSCRIPTS
IN EVERY CASE, SO THAT’S WHY I“M RAISING THE QUESTION AS TO THIS
SORT OF INSIGNIFICANT FACTOR.
I DON‘T MEAN TO SAY IT“S INSIGNIFICANT. BUT MAYBE IT IS
NOT A FACTOR THAT IT IS AS LIKELY TO BE KNOWN TO DECISION MAKERS
AS A LOT OF OTHER FACTORS.
THE WITNESS: THAT’S RIGHT. IT“S NOT ONE OF THE MAJOR
0
-
8
0
3
p
W
WN
FACTORS IN THE SYSTEM. YOUR HONOR. IT“S TRUE. BUT IT WOULD BE
? A MITIGATING FACTOR, IS MY UNDERSTANDING OF THE LAW.» AS I“VE
10 BEEN ADVISED AND STUDIED MYSELF, THAT IS, THIS IS THE KIND OF
11 MITIGATING FACTOR THAT A DEFENDER COULD PRESENT TO A SENTENCING
12 JURY IN A PENALTY TRIAL AND IT WOULD BE THE KIND OF FACTOR THAT
13 THE DEFENDANT WOULD HAVE AN INCENTIVE TO PRESENT, AND I THINK
14 THATS THE EXPLANATION OF WHY WE FIND IT IN CERTAIN CASES, IS
13 THAT THE DEFENDANTS HAVE INCENTIVE TO PRESENT THE MATERIAL INTO
16 THE PROCESS OR INJECT THIS MATERIAL INTO THE PROCESS, WHEN IT
17 DOES MITIGATE THE CASE IN THEIR DIRECTION.
13 THERE ARE OTHER FACTORS THAT THE DEFENDANTS HAVE NO
, 1° INCENTIVE TO INJECT INTO THE PROCESS, BUT THIS IS ONE THEY DO.
20 THE COURT: IT MAY BE THAT MY VIEW OF REALITY IS SKEWED
21 BY THE DEATH CASES THAT IVE REVIEWED. BUT INCLUDING THIS ONE
22 AND AT LEAST ONE OTHER THAT MR. BOGER HAS PRESENTED BEFORE ME,
23 THERE WAS NO PRESENTATION OF EITHER SUBSTANTIAL MITIGATING
24 FACTORS ACCORDING TO MR. BOGER OR IN ONE CASE. BEYOND
23 PREADVENTURE THERE WAS NO MENTION OF ANY MITIGATING FACTOR.
709
BALDUS - DIRECT
1 MR. BOGER: NOT EVEN AN AWARENESS THAT THERE WAS A
2 SENTENCE IMPOSED IN THE CASE.
3 THE COURT: MAYBE I HAVE A DIFFERENT PERCEPTION OF
4 HOW THE SYSTEM WORKS THAN YOU.
3 GO AHEAD.
“ b BY MR. BOGER:
7 Q. PROFESSOR BALDUS, I WANT TO ASK YOU NOW, YOU‘VE TALKED ABOUT
a THE DISADVANTAGES OR LIABILITIES, LIMITATIONS OF THE CROSS
9 TABULATION METHOD, AND YOU“VE SPOKEN ABOUT MULTIREGRESSION’S
10 ABILITY TO LOOK AT MORE THAN ONE BACKGROUND FACTOR AT A TIME.
13 LET ME ASK YOU TO TURN TO DB-48 MARKED FOR
12 IDENTIFICATION, AND ASK YOU TO IDENTIFY THAT DOCUMENT.
13 A. YES. DB-68 IS A TABLE FROM OUR REPORT WHICH MEASURES IN TWD
14 DIFFERENT REGRESSION ANALYSES THE IMPACT OF EACH STATUTORY
13 AGGRAVATING FACTOR AFTER HOLDING CONSTANT THE STATISTICAL EFFECT
16 OF ALL THE OTHER STATUTORY AGGRAVATING FACTORS.
17 THE CO-EFFICIENTS FROM THIS ANALYSIS, THE PORTION THAT
18 WAS CONDUCTED WITH THE LEAST SQUARES STATISTICAL PROCEDURE. ARE
“ 19 REPORTED IN COLUMN C.
20 @. LET ME ASK YOU FIRST ABOUT COLUMN B. WHERE. ARE THOSE
2} FIGURES REFLECTED ANYWHERE ELSE IN THE MATERIAL FROM WHICH
22 YOURE TESTIFYING?
23 A. YES. IN COLUMN B, WE HAVE THE UNADJUSTED REGRESSION
24 CO-EFFICIENT, WHICH IS, AS I INDICATED EARLIER, THE EQUIVALENT
25 OF THE ARITHMETIC DIFFERENCE IN THE DEATH SENTENCING RATE
710
BALDUS - DIRECT
Ty
BETWEEN THE CASES WHERE THE AGGRAVATING CIRCUMSTANCE I'S PRESENT
rN
IN THE CASE AND THOSE CASES WHERE IT’S NOT.
THOSE WERE THE FIGURES THAT WERE REPORTED IN THE -—
®@. WOULD THAT BE DB-417
A. DB-61, YES. THESE ARE THE THE REGRESSION CO-EFFICIENTS.
UNADJUSTED, THAT WERE REPORTED IN DB-é1.
AND THE CO-EFFICIENTS REPORTED IN COLUMN C INDICATE A
VERY COMMON PROPERTY OF REGRESSION RESULTS WHEN ONE INTRODUCES
CONTROLS FOR BACKGROUND FACTORS.
O
V
oH
N
0
0
h
d
W
[o
y R. WHAT DOES COLUMN C REFLECT?
11 A. IT REFLECTS WHEN ONE COMPARES THE CO-EFFICIENT THERE WITH
12 |THE CO~-EFFICIENTS IN COLUMN B, WHICH ARE UNWEIGHTED. --
13 THE COURT: ALL RIGHT.
14 THE WITNESS: -- A COMPARISON OF THOSE TWO COLUMNS
15 INDICATES THAT WITHOUT EXCEPTION THE CO-EFFICIENT DROPS AT A,
16 SUBSTANTIALLY IN TERMS OF ITS STATISTICAL EFFECT.
17 ONE EXCEPTION TO THAT IS. ILL CORRECT MY REMARK, ONE
18 EXCEPTION IS THE B3 FACTOR WHICH SHOWED NO STATISTICAL EFFECT IN
% 19 THE FIRST INSTANCE. THAT REMAINS BASICALLY UNCHANGED.
20 BUT ALL THE OTHER ONES REFLECT THE FACT THAT THESE
21 VARIABLES ARE HIGHLY RELATED TO ONE ANOTHER. CORRELATED WITH ONE
22 ANOTHER AND AS A CONSEQUENCE, WHEN BACKGROUND CONTROLS ARE
23 INTRODUCED THROUGH THE REGRESSION PROCEDURE. THAT THE
24 STATISTICAL IMPACT OF EACH TENDS TO DECLINE.
23 BY MR. BOGER:
Uo
y
f
711
BALDUS - DIRECT
QR. DO THEY DECLINE AT A SIMILAR RATE OR TO A SIMILAR EXTENT?
A. NO, THEY DON“T. THERE’S A SUBSTANTIAL VARIATION IN THE
DEGREE TO WHICH THEY DECLINE. FOR EXAMPLE, THE B4 FACTOR. WHICH
IS VERY HIGHLY CORRELATED WITH THE B2 FACTOR, DROPS FROM 17
PERCENTAGE POINTS TO VIRTUALLY ZERO.
AND WE SEE ALSO A, FURTHER EVIDENCE OF WHAT WAS
OBSERVED BEFORE WHEN WE INTRODUCED BACKGROUND FACTORS FOR THE BS
FACTOR. AND HERE THE ORIGINAL QUITE STRONG REGRESSION
CO-EFFICIENT OF 23 POINTS DROPS TOD ESSENTIALLY ZERO.
THE, SOME OF THE OTHER FACTORS, FOR EXAMPLE. THE B2
FACTOR, DROPS SIGNIFICANTLY IN ITS MAGNITUDE. BUT IT STILL IS
STRONG. AS ARE A NUMBER OF THE OTHERS.
Q. WHAT CONCLUSIONS OR HYPOTHESES CAN ONE DRAW FROM LOOKING AT
THE CHANGE IN THE IMPACT OF THESE STATUTORY AGGRAVATING
CIRCUMSTANCES, WHEN CONTROLLING FOR EACH OTHER?
A. WHAT THIS WOULD TELL YOU IS THAT IF THE EFFECTS OF THE
VARIABLE ASSOCIATED WITH A GIVEN FACTOR REMAIN STRONG AFTER
CONTROLS HAVE BEEN INTRODUCED FOR THESE BACKGROUND FACTORS, THAT
IT INCREASES THE LIKELIHOOD THAT THOSE FACTORS ARE HAVING AN
IMPACT IN THE SYSTEM. THAT IS, IT INCREASES YOUR CONFIDENCE
THAT THOSE FACTORS ARE HAVING IMPACT IN THE SYSTEM. BECAUSE WHAT
YOU“VE DONE, YOU HAVE CONTROLLED FOR A SERIES HERE OF NINE
PLAUSIBLE RIVAL HYPOTHESES, AND AFTER CONTROLLING FOR, ACTUALLY
EIGHT, BECAUSE THERE ARE NO B4 FACTORS, AFTER CONTROLLING FOR IT
PLAUSIBLE RIVAL HYPOTHESES. WE SEE FOR EXAMPLE, B2 FACTOR STILL
BALDUS - DIRECT
fo
n HAS A STRONG EFFECT IN THE SYSTEM. THAT GIVE US MORE CONFIDENCE
THAT THATS HAVING A REAL IMPACT ON THE BEHAVIOR OF PEOPLE IN
THE SYSTEM.
MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE THE
ADMISSION OF DB-68 INTO EVIDENCE, DEMONSTRATATIVE EVIDENCE.
ILLUSTRATATIVE OF PROFESSOR BALDUS’ TESTIMONY.
THE COURT: LET ME ASK YOU A QUESTION, PROFESSOR.
THE STATISTICAL SIGNIFICANCE IS IN PARENTHESES. AM I
go
o0
ow
0
W
CORRECT IN ASSUMING THAT AS THAT NUMBER APPROACHES ONE, THE
ee
oO
SIGNIFICANCE INCREASES, 1.0, NOT —-—
Po
y
po
t THE WITNESS: NO. IT“S THE OTHER WAY AROUND. YOUR
12 HONOR. AS THAT NUMBER GETS SMALLER, THE STATISTICAL
13 SIGNIFICANCE INCREASES.
14 THE COURT: AS IT APPROACHES ONE. IT BECOMES LESS
15 SIGNIFICANT?
16 THE WITNESS: THATS RIGHT. YOUR HONOR.
17 THE COURT: 1I“M SORRY. I KNEW BETTER.
13 ALL RIGHT.
% 19 BY MR. BOGER:
20 @. LET ME JUST ASK A QUESTION OR TWO ON THAT SCORE.
21 WHAT DO THEY, WHAT DO THOSE NUMBERS IN PARENTHESIS
22 BELOW THE CO-EFFICIENTS WE“VE BEEN LOOKING AT REPRESENT?
23 A. THOSE ARE MEASURES OF STATISTICAL SIGNIFICANCE.
24 R. WHAT’ S THE IMPORTANCE OF MEASURING FOR STATISTICAL
23 SIGNIFICANCE?
713
BALDUS —- DIRECT
A. AS I INDICATED EARLIER. THAT WHEN ONE LOOKS AT A CORRELATION
-
2 BETWEEN TWO VARIABLES, THAT IS, AN ASSOCIATION MEASURED BY A
3 VARIETY OF DIFFERENT METHODS. THE ONE THAT WE“RE LOOKING AT HERE
4 ARE RATIOS OF DIFFERENCES IN CORRELATION CO-EFFICIENTS, THAT IF
3 YOU SEE AN: ASSOCIATION BETWEEN TWO VARIABLES. THERE ARE, THERE
0 5 WERE TWO SETS OF RIVAL HYPOTHESES. IT MAY LEAD YOU TQ THINK THAT
7 THAT ASSOCIATION DOES NOT TRULY REFLECT A REAL EFFECT OR REAL
8 RELATIONSHIP IN THE SYSTEM BETWEEN THESE TWO VARIABLES, AND ONE
? OF THE WAYS THAT WE WERE ABLE TO TEST FOR THE PLAUSIBILITY OF
10 RIVAL HYPOTHESES WAS TO CONTROL FOR BACKGROUND FACTORS.
11 ANOTHER MAJOR RIVAL HYPOTHESIS THAT AFFECTS ALL
12 RESEARCH OF THIS TYPE IS THAT THE DISPARITY THAT YOU ARE SEEING
13 IS SIMPLY THE RESULT OF CHANCE, THAT THE DIFFERENCE AND EFFECT
14 IS A CHANCE RESULT.
135 NOW THOSE CHANCE RESULTS CAN BE PRODUCED IN TWO WAYS.
16 AND IF WE COULD, COUNSEL, COULD WE TURN TO DB-é1.
17 BY MR. BOGER:
18 @. FINE, THAT’S BACK TO THE EXHIBIT WE LOOKED AT EARLIER TO
— 19 TALK ABOUT MEASURE?
| 20 A. RIGHT. I CAN ILLUSTRATE WITH THE STATUTE AGGRAVATING FACTOR
<1 B2. NOW THIS PRESENTATION HERE IN COLUMNS B AND C SUGGEST THAT
22 IN THE UNIVERSE OF CASES FROM WHICH WE DREW OUR SAMPLE, THAT
23 AMONG CASES WHERE THE B2 FACTOR WAS PRESENT, THAT WE HAVE A
24 DEATH SENTENCING RATE OF .21, AND THAT AMONG THE CASES WHERE THE
235 B2 FACTOR IS NOT PRESENT, WE HAVE A DEATH SENTENCING RATE OF
714
BALDUS -~ DIRECT
1 «01.
2 THAT SUGGESTS THERE“S A BIG EFFECT ON DEATH SENTENCING
RATE THATS ASSOCIATED WITH THIS, THIS FACTOR.
NOW. WE HAVE TO RECALL. HOWEVER, THAT THESE ESTIMATES
ARE BASED UPON A SAMPLE. AND THERE IS THE POSSIBILITY THAT IN
3
a
=
= & FACT THE DEATH SENTENCING RATES IN THE UNIVERSE OF CASES IS THE
7 SAME IN THE POPULATION OF CASES THAT HAVE THE B2 FACTOR PRESENT
8 AND THOSE WHERE 1T”S ABSENT. THAT’S A POSSIBILITY.
@ LET”S TAKE, FOR PURPOSES OF ILLUSTRATION, A SITUATION
10 WHERE THE DEATH SENTENCING RATE IN THOSE TWO SUB-GROUPS OF CASES
11 IS .01, LOW IN BOTH CATEGORIES.
12 NOW. IT IS A POSSIBILITY THAT AS A RESULT OF THE LUCK
13 OF THE DRAW ON THE SAMPLING PROCEDURAL. WHICH IS DONE RANDOMLY »
14 WE COULD TEND TO OVERSELECT THE CASES WITH THE BZ FACTOR
15 PRESENT, WHERE THERE WERE DEATH SENTENCES PRESENT. WE GOT ALL
16 THOSE THAT INVOLVED THE DEATH SENTENCE. BUT IN THE PROCESS WE
17 GOT VERY FEW OF THE CASES THAT INVOLVED LIFE SENTENCES, THAT WE
13 TENDED TO OVERREPRESENT SUBSTANTIALLY THE DEATH CASES, BUT
® 19 UNDERREPRESENT SUBSTANTIALLY THE LIFE CASES.
20 THE EFFECT WOULD BE THAT WE WOULD PRESENT A MISLEADING
21 PICTURE WITH THIS, AND .21 WOULD BE WILDLY OFF MARK IN TERMS OF
22 WHAT WAS ACTUALLY DUT THERE.
23 NOW THE FURPOSE OF A TEST OF STATISTICAL SIGNIFICANCE
24 IS TO PROVIDE A MEASURE OF THE PROBABILITY THAT YOU WOULD GET A
23 DISPARITY OF THE TYPE WE SEE HERE IF IN FACT THE RATES OF DEATH
715
BALDUS - DIRECT
1 SENTENCING IN THE UNIVERSE OF CASES WHICH ARE OF PRINCIPAL
2 INTEREST WERE IN FACT THE SAME. THAT'S WHAT THE TEST TELLS US.
HOW LIKELY WOULD WE SEE A DISPARITY IN RACE OF THIS MAGNITUDE IF
IN FACT THE RATES WERE IDENTICALS THERE WAS NO DIFFERENCE IN THE
RATES IN THE POPULATION OF CASES FROM WHICH WE DREW OUR SAMPLE.
3
4
5
i b HOW LIKELY WOULD THIS OCCUR IS A RESULT OF JUST THE LUCK OF
7 DRAW. BY THE COINCIDENCE OF THE DRAW. THAT’S THE ONE THING THAT
8 IT TELLS US.
? BY MR. BOGER:
10 @. LET ME ASK YOU HOW THAT IS MEASURED?
11 A. YES. IT IS MEASURED THROUGH A PROCEDURE WHICH GENERATES A
12 STATISTICAL TEST. AND THAT THEN IS TRANSLATED INTO THE STATISTIC
13 | THAT WE REPORT THROUGHOUT THIS REPORT AND IS COMMONLY REPORTED |
14 IN SOCIAL SCIENCE RESEARCH. AND THAT'S --
15 THE COURT: I'M FAMILIAR WITH IT.
16 THE WITNESS: ALL RIGHT. THE "P" VALUE, AND THAT'S
17 WHAT WE HAVE IN THESE PARENTHESES HERE, YOUR HONCR.
18 THE COURT: YDU ASSUMED I DIDN‘T UNDERSTAND IT BECAUSE
e 19 I GOT IT BACKWARDS WHILE AGO, BUT I KNOW WHAT YOU“RE TALKING
20 ABOUT.
3 GO AHEAD.
22 THE WITNESS: I WOULD LIKE TO MENTION ONE OTHER THING.
23 IF I COULD, YOUR HONOR, ABOUT THE FUNCTION OF THESE STATISTICAL
24 TESTS IN THE CONTEXT OF THIS KIND OF RESEARCH THAT WE’RE GOING
23 TO BE PRESENTING HERE.
716
BALDUS - DIRECT
1 I INDICATED THAT ONE SOURCE OF VARIABILITY, ONE SDURCE
OF THREAT TO VALIDITY THAT MIGHT PRODUCE A MISLEADING PICTURE IS 2
3 THE IMPACT OF THE SAMPLING PROCEDURE ITSELF.
FF
5 THERE IS ALSO A RISK THAT CHANCE CAN PRODUCE A
ba MISLEADING RESULT EVEN IF YOU HAVE ALL THE CASES IN THE
& UNIVERSE. FOR EXAMPLE. IN THE CHARGING. PARDON ME, IN THE
7 PROCEDURAL REFORM STUDY. WE HAVE ALL THE CASES. WE DIDN’T DRAW
3 A SAMPLE, SO WE DON’T HAVE ANY RISKS THAT WE“RE GOING TO BE
? GETTING MISLEADING RATES AS A RESULT OF LUCK OF THE DRAW FOR THE
10 SAMPLE.
13 HOWEVER. THERE ARE STILL A NUMBER OF CHANCE FACTORS
12 THAT ARE OPERATING IN A SYSTEM AS LARGE AND COMPLEX AS THIS
13 CHARGING AND SENTENCING SYSTEM THAT WE HAVE HERE, AND THE. THE
14 CHARGING AND SENTENCING SYSTEM IN THIS JURISDICTION COULD BE
13 OPERATING IN A COMPLETELY EVEN HANDED WAY. AND YET WE COULD SEE,
146 FOR EXAMPLE, ON THE RACIAL, WE CAN SEE RACIAL DISPARITIES THAT
17 EXIST. AND WHY DO THEY EXIST, WHY WOULD WE SEE THEM IN AN EVEN
18 HANDED SYSTEM? WE MIGHT SEE THEM BECAUSE THE RANDOM FACTORS
Ke 19 THAT WE DON’T HAVE ANY DATA ON IN THE STUDY THAT SHOCK THE
20 SYSTEM, THE UNKNOWN, THE UNKNOWABLES THAT ARE WORKING IN A
21 SYSTEM AS COMPLEX AS THIS, THAT THOSE BY CHANCE. BY PURE
22 COINCIDENCE COULD COME TO BEAR ON ONE CLASS OF CASES AND MAKE IT
23 APPEAR THAT THOSE CASES WERE BEING TREATED MORE PUNITIVELY IN
24 THE SYSTEM THAN ANOTHER CLASS OF CASES. THAT'S ALWAYS A
25 POSSIBILITY. AND THE TEST OF STATISTICAL SIGNIFICANCE AND THE
O
M
M
o
O
B
h
N
e
72
po
te
[w
y
717
BALDUS - DIRECT
“P" VALUE GIVES YOU SOME WAY OF ESTIMATING THE LIKELIHOOD THAT
THESE UNKNOWABLES THAT ARE IMPACTING ON THE SYSTEM, MIGHT
PRODUCE THE DISPARITY YOU OBSERVE. EVEN THOUGH THE SYSTEM IS
OPERATING IN A PERFECTLY EVEN HANDED WAY.
. OR HERE, FOR EXAMPLE, IT MIGHT BE. IN THE CONTEXT OF
THE B2 FACTOR, IT MAY BE THAT THE B2 FACTOR HAS NO EFFECT
WHATEVER ON THE SYSTEM. AND ASSUMING WE HAD INFORMATION AN ALL
THE CASES. YOU STILL MIGHT SEE A DIFFERENCE IN THE DEATH
SENTENCING RATES ON THE B2 FACTORS, EVEN THOUGH IT HAD NO
EFFECT, SIMPLY BECAUSE OF THE COINCIDENCE OF THE CO-EXISTENCE OF
THESE FACTORS ABOUT WHICH WE KNOW NOTHING, AND THE B2 FACTOR.
THAT IT WAS. THE UNKNOWABLE, THE UNKNOWNS THAT ARE AFFECTING
THINGS, WOULD TEND TO BE PRESENT MORE FREQUENTLY IN B2 CASES
THAN IN OTHER CASES. AND THAT WOULD TEND TO ELEVATE THE BZ
FACTOR DEATH SENTENCING RATE, EVEN THOUGH THAT WAS NOT A FACTOR
IN THE SYSTEM.
SO THOSE ARE THE TWO. THERE ARE OTHER TECHNICAL SOURCES
OF CHANCE VARIATION, BUT FOR THE PURPOSES OF THIS KIND OF
RESEARCH AND IN THIS PROCEEDING, THE TWO THINGS THAT THESE TESTS
OF SIGNIFICANCE GUO TO, ARE THE CHANCE VARIABILITY IN THE SYSTEM
PRODUCED BY THE SAMPLING PROCEDURE THAT WE USED, AND AL30 BY THE
PRESENCE OF THESE UNKNOWNS AND UNKNOWABLES SHOCK EFFECTS IN
THE SYSTEM.
‘MR. BOGCER: YOUR HONOR, AT THIS TIME, I —-
THE COURT: HE HAS OFFERED -48. WHAT'S YOUR POSITION
718
BALDUS - DIRECT
Ty
ON THAT, MS. WESTMORELAND?
MS. WESTMORELAND? YOUR HONOR, I UNDERSTAND IT WAS
3)
5s |OFFERED AS A DEMONSTRATIVE EXHIBIT. FROM MR. BOGER‘S PREVIOUS
4 |STATEMENT.
5 THE COURT: DID HE SAY THAT?
“ 5 MR. BOGER: THAT‘S CORRECT. YOUR HONOR.
7 THE COURT: YOU’RE NOT OFFERING IT FOR THE TRUTH OF THE
3 |MATTER ASSERTED THEREIN?
9 MR. BOGER: NOT AT THIS TIME. YOUR HONOR.
10 THE COURT: WELL, DO YOU INTEND TO?
11 MR. BOGER: YES. YOUR HONOR. WELL. I WILL OFFER IT FOR
12 |BOTH PURPOSES: HEAR THE STATES POSITION ON THAT.
13 MS. WESTMORELAND: YOUR HONOR, AT THIS TIME, OBVIOUSLY,
14 THE SAME OBJECTIONS THAT HAVE PREVIOUSLY BEEN NOTED. BUT I
Pa
y on
ALSO HAVE CONCERNS AT THIS TIME CONCERNING PROFESSOR BALDUS,
146 WHETHER HE WAS THE ONE CONDUCTING THESE TESTS DEVELOPING THESE
37 STATISTICS. AND ALSO WHETHER HIS EXPERTISE GOES 50 FAR AS TO
18 DETERMINE WHAT APPROPRIATE TESTS WERE TO BE UTILIZED TO DEVELOP
» 19 THIS, AND IF HE WAS NOT THE ONE WHO MADE THESE DETERMINATIONS, I
20 DON‘T THINK IT“S APPROPRIATE TO ADMIT THE TABLE AT THIS TIME FOR
21 THE TRUTH OF THE MATTER THEREIN.
22 MR. BOGER: YOUR HONOR, I DO THINK THIS IS WITHIN -—
23 THE COURT: ARE YOU ASKING FOR AUTHENTICITY?
24 MS. WESTMORELAND: YES, YOUR HONOR.
23 THE COURT: DID YOU CREATE THIS TABLE?
719
BALDUS - DIRECT
1 MR. BOGER: PROFESSOR BALDUS?
MN
THE WITNESS: YES, YOUR HONOR, I CREATED THE TABLE.
3 THE COURT: IT’S HIS TABLE, S0 ITS AUTHENTICATED.
4 MS. WESTMORELAND: BUT MY OTHER QUESTION THEN WOULD BE,
(&)
YOUR HONOR. IS TO JUST HIS ACTUALLY CREATING THE TABLE, 1
SUPPOSE, BUT THE ACTUAL FIGURES THAT WERE IN THE TABLE. AND
WHETHER THOSE WERE GENERATED BY PROFESSOR BALDUS OR BY PROFESSOR
BALDUS PURSUANT TO DIRECTIONS FROM OTHER PARTIES, WHAT THE
S
N
1
SH
E
SE
E
ACTUAL SOURCE OF THOSE FIGURES IS.
10 BY MR. BOGER:
11 @. PROFESSOR BALDUS. ARE YOU RELYING ON THESE TABLES IN YOUR,
12 IN YOUR EXPERT OPINIONS, IN YOUR FINAL REPORT?
13 A. YES.
14 @. AND HAVE YOU DRAWN ON YOUR OWN SOURCES AND SOURCES OF SDCIAL
13 SCIENCE LITERATURE TO CREATE THESE TABLES?
1&6 A. YES. 1 CREATED THESE TABLES IN AN ANALYSIS CONDUCTED ON
17 COMPUTER.
18 MR. BOGER: YOUR HONOR, I SUSPECT THAT THE STATE MIGHT
® 19 BE ABLE TO CROSS-EXAMINE PROFESSOR BALDUS ON HIS OWN CREATION OF
20 THESE TABLES. BUT I DON’T SEE ANY PROBLEM IN TERMS OF THEIR
21 ADMISSIBILITY.
22 THE COURT: IT SEEMS TO ME THAT EITHER YOU OR I HAVE
23 GOT THAT RULE WRONG. HE MAY RELY ON INADMISSIBLE EVIDENCE FOR
24 THE PURPOSES OF EXPRESSING AN OPINION. BUT THE FACT THAT HE
23 RELIED ON IT DOES NOT MAKE IT ADMISSIBLE.
720
BALDUS - DIRECT
MR. BOGER: NO, YOUR HONOR. I THINK WHAT I TRIED TO
[w
y
ELICIT FROM PROFESSOR BALDUS WAS NOT SIMPLY WHETHER HE RELIES ON
THE TABLE BUT WHETHER HE RELIED ON HIS BACKGROUND IN ORDER TO
GENERATE THE TABLE. AND HE INDICATED THAT HE HAS NOT SIMPLY
KNOWN OF STATISTICAL PROCEDURES, BUT THAT HE TEACHES AND HAS
WRITTEN AND REVIEWED OTHER PEOPLE. HE“S NOT QUALIFIED AS A
DOCTOR OF STATISTICS. BUT HE’S QUALIFIED AS SOMEONE WHO USES
STATISTICAL METHODS IN THIS AREA. AND I REALLY DON‘T SEE ANY
DIFFICULTY AT ALL, GIVEN THAT KIND OF QUALIFICATION TESTIMONY. AND
HIS TESTIMONY THAT HE IN FACT GENERATED THESE TABLES. WITH THEIR
T
s
OO
WO
B
N
h
n
ADMISSIBILITY. pa
-
12 THE COURT: I“M NOT SURE I REALLY UNDERSTAND WHAT MS.
13 WESTMORELAND IS CONCERNED ABOUT.
14 DO YOU WANT TO KNOW IF HE KNOWS IF THE COMPUTER DID IT
13 RIGHT OR DO YOU WANT TO KNOW IF HE KNOWS WHETHER THE COMPUTER
16 SHOULD HAVE DONE IT AT ALL?
17 MS. WESTMORELAND: YOUR HONOR, WHAT I THINK I REALLY
13 WANT TO KNOW IS IF HE KNEW PERHAPS EITHER HOW THE TESTS WERE
% 19 DONE OR WHY THESE PARTICULAR TESTS WERE UTILIZED.
20 THE COURT: WHY DID YOU UTILIZE THESE TESTS?
23 THE WITNESS: BECAUSE THESE ARE THE APPROPRIATE
22 PROCEDURES FOR DETERMINING THE EFFECTS OF VARIABLES.
23 THE COURT: WE’LL GET TO THE APPROPRIATENESS OF THEM
24 LATER.
23 DID YOU SELECT THESE TESTS OR SOMEBODY ELSE SELECT THE
BALDUS - DIRECT
TESTS? —
THE WITNESS: NO. GEORGE WOODWORTH ADVISED ME ABOUT
THESE TESTS. AND UPON HIS GENERAL ADVICE, I IDENTIFIED THE
SPECIFIC PROGRAM ON WHICH IT WAS USED IN THE COMPUTER.
BUT THE DECISION TO USE LEAST SQUARES ANALYSIS OF THE
TYPE RFEPORTED HERE AND THE LOGISTIC ANALYSIS, WAS DONE UPON THE
ADVICE OF PROFESSOR WOODWORTH.
MR. BOGER: YOUR HONOR, IT SEEMS TO ME IT REMAINS FREE
M
0
W
N
TO THE STATE AFTER THESE DOCUMENTS ARE ADMITTED, IF THEY ARE, TO
SHOW THAT THE PROCEDURES THAT WERE EMPLOYED WERE WRONG. pot
e oO
a
pp
BUT I“M NOT SURE I UNDERSTAND THE BASIS FOR NOT
Ta
y kK
ADMITTING THEM. IF THERE“S NO CHALLENGE TQ PROFESSOR BALDUS”
TN
Ww
HAVING DONE THEM HIMSELF, OR ACTUALLY ASKED A COMPUTER ANALYST
14 TO RUN THEM FOR HIM.
15 THE COURT: BASED ON YOUR KNOWLEDGE OF STATISTICS. DO
16 YOU BELIEVE THAT YOU WOULD KNOW WHETHER OR NOT THE TESTS
17 EMPLOYED ON -438 WOULD ANSWER THE SOCIAL SCIENCE QUESTION THAT
18 YOU POSED?
% 1% THE WITNESS: YES, I BELIEVE I KNOW.
20 THE COURT: WHAT’S YOUR OPINION?
21 THE WITNESS: I BELIEVE THAT THEY DO.
22 THE COURT: ALL RIGHT. THE NEXT QUESTION IS THIS: UWE
23 ARE CONCERNED WITH A BS CASE HERE, 1 GUESS. THE STATISTICAL
25 SIGNIFICANCE IS .94.
23 DOES THAT MAKE THAT STATISTICAL CONCLUSION HAVE EVEN
BALDUS - DIRECT
1 MINIMAL WEIGHT?
2 THE WITNESS: IT DOES NOT HAVE MUCH WEIGHT. YOUR HONOR,
3 BUT I WOULD ADD, HOWEVER. THAT WERE DEALING WITH A RELATIVELY
4 FEW NUMBER OF B8 CASES. I WOULD —-
=] THE COURT: WELL. THAT’S, WHAT“S THE NAME OF THAT CASE.
a & |I CITED YOU TO. MR. BOGER., THAT YOU AND MS. WESTMORELAND HADN'T
7 SEEN AND I HADN’T READ?
8 MR. BOGER: YOURE NOT TALKING ABOUT THE ADAMS CASE.
9 YOURE TALKING ABOUT THE CASE YOU MENTIONED YESTERDAY?
10 THE COURT: ADAMS VERSUS WAINWRIGHT.
11 MR. BOGER: THAT-S CORRECT. YOUR HONOR.
12 THE COURT: THE MERE FACT THAT THE SAMPLE IS S0 SMALL
13 AS TO BE STATISTICALLY INSIGNIFICANT IS A REASON THAT THE COURT
14 WOULD NOT RELY ON THE STATISTICS AT ALL. THAT'S WHAT IM TRYING
15 TO GET YOU TO ADDRESS. THAT’S WHAT ADAMS BASICALLY SAYS. AND
146 THERE ARE SOME OTHER CASES ON THE GRAND JURY FOREPERSON THAT SAY
17 EXACTLY THAT QUT OF THIS CIRCUIT.
13 MR. BOGER: WELL, YOUR HONDR, WE DO NOT PURPORT TO REST
& 19 OUR PETITIONER’S CASE, AND I‘LL ASK PROFESSOR BALDUS, BUT I
20 DON’T THINK HE PURPORTS TO REST HIS OPINIONS OR HIS ANALYSIS ON
23 ANY ONE SET OF TABLES OR FIGURES ALONE.
22 INDEED DUR WHOLE APPROACH AND WE ARE LITERALLY AT THE
23 POINT OF GETTING INTO IT, MY NEXT EXHIBIT WILL BE A REQUEST FOR
24 HIS EXPERT OPINIONS ON THE MATTERS AT ISSUE BEFORE THE COURT.
23 BUT OUR CONTENTION WILL BE THAT IT’S THE PATTERN OF EVIDENCE,
~4
N £0)
BALDUS - DIRECT
SOME OF WHICH HAS MARGINAL SIGNIFICANCE, SOME OF WHICH MAY HAVE
NO SIGNIFICANCE, BUT ALL OF WHICH IS REFLECTED IN A SERIES OF
GRAPHS AND TABLES THAT BEARS THE BURDEN OF OUR PROOF.
AND I DON‘T BELIEVE THAT WE’RE REQUIRED IN EVERY SINGLE
TABLE WE PRESENT TO HAVE A STATISTICALLY SIGNIFICANT '
RELATIONSHIP OR TO PROVE OUR CASE. INDEED, I THINK IT”S FOR
THE COURT TO SEE THE PLACES IN WHICH OUR CONTENTIONS MAY BE
STRONGER OR MAY BE WEAKER, AND THAT IT IS THE PATTERN AND
PRACTICE REFLECTED IN THE STATISTICS OVERALL THAT WILL EITHER
PERMIT US TO PREVAIL OR NOT.
SO MY CONTENTION SUMMED UP TO YOUR HONOR IS THAT THIS
TABLE IS NOT NECESSARILY ADMISSIBLE BECAUSE THE BS FACTOR IN IT
1S ITSELF NOT SIGNIFICANT. INDEED IF ONE LOOKS AT THIS CASE
ALSO AS A B2 CASE. AS AN ARMED ROBBERY CASE, WHICH IT WAS. THE
B2 CASE IS SIGNIFICANT AT THE ONE IN TEN THOUSANDTH LEVEL. 30
THAT REFLECTS THE INHERENT DIFFICULTY IT SEEMS TO ME OF AN
APPROACH THAT WOULD SAY YOU CAN‘T PUT A TABLE IN IF IT DOESN'T
POSITIVELY PROVE YOUR POINT. THESE. I THINK THE BURDEN OF
PROFESSOR BALDUS” TESTIMONY WILL BE THERE ARE LOTS OF FACTORS
THAT ONE HAS TO LOOK AT SIMULTANEOUSLY.
THE WITNESS: YOUR HONOR, -—
THE COURT: MR. BOGER, FRANKLY I“M TOYING WITH THOSE
CRYPTIC PARAGRAPHS FROM THE ANTI-TRUST CASE THAT I SHOWED MR.
STROUP THE QTHER DAY.
IT SEEMS TO ME THAT THE BASIC RULE OF RELEVANCY IS THAT
BALDUS - DIRECT
1 IF IT TENDS TO MAKE A FACTOR OF CONSEQUENCE MORE LIKELY THAN
NOT, THAT IT COMES IN, AND THAT’S A WHOLE LOT DIFFERENT FROM THE
SUBSTANTIALLY RELIABLE OR WHATEVER LANGUAGE THE ELEVENTH CIRCUIT
EMPLOYED.
THE CHALLENGE, MS. WESTMORELAND, I GUESS I“M HAVING
REAL PROBLEMS UNDERSTANDING WHAT IT IS YOU“RE OBJECTING TO OTHER
THAN AUTHENTICITY, WHICH I THINK WE-RE CLEANED UP.
0
N
o
d
b
s
N
MS. WESTMORELAND: YOUR HONOR, ASIDE FROM THE POINTS
9? WE“VE ALREADY MADE. I THINK PERHAPS WITH THIS TABLE BECOMING
10 MORE READILY APPARENT. ONCE AGAIN WE“RE NOT ADDRESSING FULTON
11 COUNTY. WHERE THE NUMBERS AS YOUR HONOR HAS ALREADY POINTED OUT,
12 ADDRESS THE PARTICULAR AGGRAVATING CIRCUMSTANCES OF THIS CASE,
13 THEY DID NOT SHOW ANY STATISTICAL SIGNIFICANCE. THAT IN
14 CONJUNCTION WITH THE FACT THAT PROFESSOR BALDUS DID NOT
13 PERSONALLY SELECT THE TEST TO BE EMPLOYED IN COMING UP WITH THIS
146 TABLE, ALL OF THESE FACTORS, I THINK SHOULD BE CONSIDERED BY THE
17 COURT AND SHOULD BE DETERMINATIVE OF WHETHER OR NOT THIS
18 PARTICULAR TABLE SHOULD BE ADMITTED.
pe 19 AND PERHAPS WHEN WE DETERMINE IT, THE REMAINDER OF
20 TABLE TO BE SUBMITTED, I PERCEIVE SIMILAR OBJECTIONS BEING
21 RAISED TO SUCCEEDING TABLES.
22 THE COURT: WELL, AS TO YOUR GENERAL OBJECTIONS. WE-LL
23 HAVE TO DEAL WITH THOSE. I THINK, AT THE TAIL END OF THE WHOLE
24 CASE.
25 BUT AS TO THE OTHER, I THINK THE BETTER PART OF VALOR
BALDUS - DIRECT
re
WOULD BE TO LET IT IN SUBJECT TO YOUR REPRESENTATION THAT YOUR
STATISTICIAN WILL SUPFLY ANY MISSING CONNECTION.
MR. BOGER: I THINK WE CAN CONNECT THAT UP LATER. YOUR
HONCR.
THE COURT: IF HE DOESN‘T, YOU MAY MOVE TO STRIKE ON
WHATEVER BASIS YOU CAN IMAGINE.
MS. WESTMORELAND: OKAY.
THE COURT: ALL RIGHT. ILL ADMIT IT.
MR. BOGER: THANK YOU. YOUR HONOR.
pt
S
W
o
N
0
8
MH
»
W
N
BY MR. BOGER:
11 @. NOW, PROFESSOR BALDUS, YOU HAVE TESTIFIED AT LENGTH, GREAT
12 LENGTH ABOUT THE DESIGN OF YOUR STUDIES, ABOUT THE COLLECTION OF
13 YOUR DATA, ABOUT THE CLEANING OF YOUR DATA, HOW TO GET ENTRY ON
14 THE COMPUTER. ABOUT THE METHODS YOU USED IN APPROACHING THE
15 QUESTIONS OF ANALYSIS, ABOUT THE MEASURES YOU USED TO CONDUCT
16 ANALYSIS.
17 NOW I WOULD LIKE TO BRING YOU TO THE HEART OF THIS CASE
18 BY ASKING YOU THE QUESTIONS ON WHICH WE HAVE SOLICITED YOUR
pM 1? EXPERT OFINION ON BEHALF OF PETITIONER.
20 LET ME DIRECT YOUR ATTENTION TO THE FIRST VOLUME OF
21 YOUR EXHIBIT BOOK TO AN ITEM MARKED AS DB-12 FOR IDENTIFICATION.
22 CAN YOU IDENTIFY THAT DOCUMENT?
23 A. YES. THIS IS A LETTER FROM YOU AS COUNSEL FOR PETITIONER IN
24 THIS PROCEEDING TO ME, DATED AUGUST 1, 1983.
23 @. AND HAVE YOU RECEIVED THIS DOCUMENT IN THE PAST?
ft
WO
0H
N
B
p
d
O
N
10
11
726
BALDUS - DIRECT
A. YES. IT WAS DELIVERED TO ME WITHIN THE LAST WEEK OR TEN
DAYS.
@. HAVE YOU READ IT?
A. YES. I HAVE.
@. LET ME DIRECT YOUR ATTENTION TO THE FIRST NUMBERED PARAGRAPH
IN THAT LETTER, AND ILL READ THAT PARAGRAPH TO YOU.
"PROFESSOR BALDUS, DO DISPARITIES EXIST IN THE
PENALTIES IMPOSED UPON HOMICIDE DEFENDANTS IN THE STATE
OF GEORGIA. BASED UPON THE RACT OF THE HOMICIDE VICTIM?"
DO YOU HAVE A PROFESSIONAL OPINION ON THAT QUESTION.
BASED ON YOUR OWN STUDY AND RESEARCH AND THE DATA WHICH YOU-VE
GATHERED AND ANALYZED?
A. YES. I DO.
@. WHAT IS YOUR EXPERT OFINION?
MS. WESTMORELAND: YOUR HONOR, I WILL OBJECT UNLESS
PROFESSOR BALDUS IS READY TO STATE THE BASIS FOR THAT OPINICN AT
THIS TIME.
THE COURT: YOU CAN‘T BUILD A BRICK WALL BUT ONE BRICK
AT A TIME.
GO AHEAD.
THE WITNESS: SIR?
THE COURT: GO AHEAD.
THE WITNESS: IT IS MY OPINION THAT SYSTEMATIC AND
SUBSTANTIAL DISPARITIES EXIST IN THE PENALTIES IMPOSED UPON
HOMICIDE DEFENDANTS IN THE STATE OF GEORGIA, BASED UPON THE
EE C————.
TW
A
To
ry
©
W
M
N
A
S
O
N
727
BALDUS -~ DIRECT
RACE OF THE HOMICIDE VICTIM.
BY MR. BOGER:
@. LET ME ASK YDU THE SECOND QUESTION ENUMERATED HERE IN DB-12.
"nO DISPARITIES EXIST IN THE PENALTIES IMPOSED UPON
HOMICIDE DEFENDANTS IN THE STATE OF GEORGIA. BASED UPON
THE RACE OF THE DEFENDANT?"
A. IT IS MY OPINION THAT DISPARITIES IN DEATH SENTENCING RATES
DO EXIST BASED UPON THE RACE OF THE DEFENDANT. BUT THEY ARE NOT
AS SUBSTANTIAL AND NOT AS SYSTEMATIC AS THE RACE OF VICTIM
DISPARITIES.
@. WELL, DO THESE -— ESCUSE ME,
DO DISPARITIES IN BOTH RACE OF VICTIM, RACE OF
DEFENDANT PERSIST. EVEN WHEN THE AGGRAVATING CIRCUMSTANCES
DEFINED BY GEORGIA‘S CAPITAL PUNISHMENT STATUTE ARE TAKEN INTQ
ACCOUNT?
A. YES, THEY DO.
R. DO THESE DISPARITIES PERSIST EVEN WHEN ONE CONSIDERS ONLY
THOSE HOMICIDE CASES WHICH RESULT IN A MURDER INDICTMENT?
A. YES.
@. DO THESE DISPARITIES PERSIST EVEN WHEN ONE CONTROLS
SIMULTANEOUSLY FOR STATUTORY AND UNSTATUTORY AGGRAVATING AND
MITIGATING CIRCUMSTANCES AND MEASURES OF THE STRENGTH OF
EVIDENCE?
A. YES.
@. DO THESE DISPARITIES PERSIST WITHIN THE JURISDICTION OF
728
BALDUS - DIRECT
J
y
FULTON COUNTY GEORGIA. WHERE PETITIONER WARREN MCCLESKEY WAS
2 TRIED?
3 A. THE RACE OF VICTIM DISPARITIES ONLY PERSIST WITHIN THE
4 JURISDICTION OF FULTON COUNTY.
9 @. PROFESSOR BALDUS, IN YOUR EXPERT OPINION. READING NOW
4 é ENUMERATED QUESTION NUMBER 7:
7 "ARE THERE OTHER LEGITIMATE FACTORS NOT CONTROLLED FOR
a IN YOUR ANALYSES WHICH COULD PLAUSIBLY EXPLAIN THE
? PERSISTENCE OF THESE RACIAL DISPARITIES IN THE STATE
10 OF GEORGIA AND IN FULTON COUNTY?"
11 A. NO.
12 QR. AND FINALLY, QUESTION 8, PROFESSOR BALDUS:
13 "DO YOU THINK THAT RACIAL FACTORS HAVE A
14 REAL EFFECT IN CAPITAL CHARGING AND SENTENCING
13 SYSTEM OF THE STATE CF GEORGIA?"
16 A. YES, 1 DO.
17 THE COURT: HOW ABOUT IN FULTON COUNTY?
18 THE WITNESS: I BELIEVE THAT THE RACE, PARDON ME, I
% A BELIEVE THAT THE EVIDENCE SUGGESTS THAT RACE OF VICTIM HAS AN
20 EFFECT, REAL EFFECT IN THE SYSTEM HERE.
21 MR. BOGER: YOUR HONOR, I MOVE THE ADMISSION OF DB-12
22 INTO EVIDENCE.
23 MS. WESTMORELAND: YOUR HONOR, I HAVE NO ACTUAL
24 OBJECTION, ALTHOUGH I DONT FELL IT’S NECESSARY, SINCE MR. BOGER
2 HAS READ THE ENTIRE DOCUMENT INTO THE RECORD.
Oo
0
0
~
or
OR
+
Ww
nN
__
[
=
[2
8
T
R
I
Y
GW
14
13
24
25
729
BALDUS - DIRECT
THE COURT: IN THE SENSE I HAVE ADMITTED A BUNCH OF
OTHER THINGS, NOT FOR THE TRUTH OF THE MATTER STATED THEREIN.
BUT AS A DEMONSTRATION OR A SUMMARY OF WHATEVER. I WILL LET IT
IN, BECAUSE THERE ARE MANY THINGS WHICH ARE MORE CONVENIENTLY
REVIEWED LATER IN THE BRIEFING STAGE AND APPELLATE STAGE AND
EVEN THE STAGE I“VE GOT TO DO IT, IF I CAN LOOK AT THE EXHIBIT.
RATHER THAN HAVE TO FIND IT IN THE TRANSCRIPT.
SO TO THE EXTENT THAT IT IS A CHART, GRAPH OR SUMMARY
OR WHATEVER OF THE QUESTIONS ON WHICH YOU ASKED HIS OPINION. I
WILL ALLOW IT IN.
MR. BOGER: THANK YOU, YOUR HONOR.
BY MR. BOGER:
@. NOW, PROFESSOR BALDUS, LET ME DIRECT YOUR ATTENTION BACK TO
DB-62 FOR IDENTIFICATION.
CAN YOU IDENTIFY THAT DOCUMENT?
@. WHAT IS IT?
THE COURT: NOW WAIT A MINUTE, MR. BOGER. I HAVEN'T
FOUND IT YET.
MR. BOGER: I“M SORRY. YOUR HONOR.
THE COURT: YOUR SYSTEM IS REAL GOOD, AS LONG AS YOU GO
IN SEALENT 10 ORDER, BUT WITHOUT TABS. IT’S HARD TO BACK UP.
MR. BOGER: THAT’S CORRECT, AND I THINK I WOULD HAVE
BEEN SLAUGHTERED BY TWO LAW STUDENTS IF I ASKED FOR TABS ON THAT
INSTEAD OF SIMPLY PUTTING IT TOGETHER.
[PRES SSS — e————— ——— Se———- ————— — ———— ———————— — — ——— S—————. ——— ——
730
BALDUS -~ DIRECT
po
ke
THE COURT: LEST YOU THINK I“M BEING UNDULY CRITICAL. I
2 TOLD MY COURTROOM DEPUTY THAT I MAY ADOPT THIS NATEBUOK
3 PROCEDURE FOR HEAVY DOCUMENT CASES IN THE FUTURE. WELL TRY TO
4 GIVE YOU APPROPRIATE CREDIT.
3S MR. BOGER: THANK YOU KINDLY.
® 1) THE COURT: AND ALL THE LAWYERS OF GEORGIA WILL THANK
7 YOU.
8 BY MR. BOGER:
? R. PROFESSOR BALDUS. CAN YOU IDENTIFY DB-627
10 A. YES. DB-462 PRESENTS UNADJUSTED RACE OF VICTIM AND RACE OF
1 DEFENDANT DISPARITIES IN DEATH SENTENCING RATE ESTIMATED IN THE
12 CHARGING AND SENTENCING STUDY.
13 @. ARE THESE RESULTS FROM YOUR DATA COLLECTION EFFORTS AND YOUR
14 REPORT?
13 A. YES, THEY ARE.
14 @. AND WHAT DO THEY REFLECT?
17 A. THEY REFLECT THAT WITHOUT ANY OTHER ADJUSTMENT. THAT THE.
18 THERE IS A 10 PERCENTAGE POINT HIGHER CHANCE THAT A DEFENDANT
PW
S 0 WHOSE VICTIM IS WHITE WILL RECEIVE THE DEATH SENTENCE THAN A
~)
Oo
DEFENDANT WHOSE VICTIM IS BLACK.
21 AND PANEL 2 OF THIS TABLE INDICATES THAT WITHOUT
22 ADJUSTMENT FOR ANY BACKGROUND FACTORS. A BLACK DEFENDANT IN THE
23 STATE OF GEORGIA HAS A 3 PERCENTAGE POINT LOWER PROBABILITY OF
24 RECEIVING A DEATH SENTENCE THAN DOES A WHITE DEFENDANT.
23 THE COURT: MR. BOGER. JUST FOR MY CLARIFICATION, THIS
731
BALDUS - DIRECT
1 IS ABOUT WHERE WE LEFT OFF WITH DOCTOR BOWERS, ISN'T IT?
nN
MR. BOGER: THAT“S CORRECT, YOUR HONOR. ACTUALLY WE“VE
3 GOT ONE MORE STEP DOCTOR BOWERS DID CONTROL FOR.
4 THE COURT: ONE AGGRAVATING CIRCUMSTANCE?
MR. BOGER: THAT’S CORRECT. tl)
THE COURT: OKAY.
BY MR. BOGER:
@. DO YOU BASE YOUR EXPERT OPINIONS ON THIS UNADJUSTED RACE OF
9
B
N
VICTIM AND RACE OF DEFENDANT DISPARITY TABLE THAT WE SEE HERE?
10 A. NO.
11 G. WHY NOT?
12 A. THIS TABLE MERELY GENERATES A HYPOTHESIS. IT“S, IT HAS NO
13 CONTROLS. THERE ARE MANY PLAUSIBLE RIVAL HYPOTHESES THAT COULD
14 EXPLAIN THESE RELATIONSHIPS.
15 ONE CAN‘T MAKE ANY SORT OF INFERENCE OF THE TYPE I
16 EXPRESSED IN MY OPINION WITHOUT SYSTEMATICALLY TESTING FOR THE
17 PLAUSIBILITY OF THOSE RIVAL HYPOTHESES.
13 MR. BOGER: YOUR HONOR, I MOVE THE ADMISSION OF DB-62
4 19 INTO EVIDENCE. z
20 MS. WESTMORELAND: YOUR HONOR, I GUESS I HAVE MY
21 CONTINUING OBJECTION THAT 1°VE BEEN MAKING ALL ALONG,
22 PARTICULARLY IN LIGHT OF THE PREVIOUS ANSWERS CONCERNING FULTON
23 COUNTY AND THE RACE OF DEFENDANT DISPARITIES FOUND IN FULTON
24 COUNTY.
23 OTHER THAN THOSE SAME OBJECTIONS THAT IVE BEEN MAKING ALL
~. —. ret” at —— — p———, S—————— ————— y——— ——— —— ———— ————" T——— W—————————— W———"—)—— | I———— Te ————, ——— el, Yt
BALDUS - DIRECT
1 ALONG. I HAVE NO FURTHER OBJECTION.
2 THE COURT: ALL RIGHT. I THINK DOCTOR BALDUS,
3 PROFESSOR BALDUS WOULD AGREE THAT THIS DOES NOT IN A SCHOLARLY
4 WAY MAKE A FACT OF CONSEQUENCE IN THIS CASE MORE OR LESS LIKELY
S AND IT WOULD NOT BE RECEIVED FOR THE TRUTH OF THE MATTERS
b ASSERTED THEREIN.
7 IT WILL, HOWEVER, AS A DEMONSTRATION OF HIS THOUGHT
8 PROCESSES BE ADMITTED.
? MR. BOGER: THANK YOU, YOUR HONOR.
10 BY MR. BOGER:
11 @. PROFESSOR BALDUS. LET ME ASK YOU TO TURN TO DB-63 FOR
12 IDENTIFICATION AND ASK YOU TO IDENTIFY AND DESCRIBE THAT
13 DOCUMENT?
14 A. YES. DB-63 IS A DISTRIBUTION OF DEATH SENTENCING RATES
13 ACCORDING TO THE DEFENDANT-VICTIM RACIAL COMBINATION OF
16 OFFENDERS IN OUR CHARGING AND SENTENCING STUDY.
17 IT SHOWS THAT IN CASES INVOLVING BLACK DEFENDANTS AND
18 WHITE VICTIMS THAT THE DEATH SENTENCING RATE IS .Z22.
% i9 CASES INVOLVING WHITE DEFENDANTS AND WHITE VICTIMS IT’S
20 . 08.
21 IN CASES INVOLVING BLACK DEFENDANTS AND BLACK VICTIMS
22 IT IS .01.
23 AND IN THE HANDFUL OF CASES. 64 IN NUMBER ESTIMATED.
24 INVOLVING WHITE DEFENDANTS AND BLACK VICTIMS, THE DEATH
23 SENTENCING RATE IS .Q3.
733
BALDUS - DIRECT
Q. DOES THIS TABLE REFLECT ANY CONTROLS FOR OTHER BACKGROUND
FACTORS?
A. NO, IT DOES NOT.
@. HOW WOULD YOU SUMMARIZE THE SIGNIFICANCE OF THIS TABLE?
A. THIS TABLE PROVIDES SUPPORT FOR A HYPOTHESIS THAT AMONG
WHITE VICTIM CASES. THAT BLACK DEFENDANTS HAVE A HIGHER RISK OF
BEING SENTENCED TO DEATH THAN DO WHITE DEFENDANTS.
AND WHEN ONE COMPARES COLUMNS A AND B WITH COLUMNS C
AND D, IT SUGGESTS THE HYPOTHESIS THAT THE RACE OF VICTIM IS A
FACTOR IN THE CHARGING AND SENTENCING PROCESS LEADING TO CAPITAL
PUNISHMENT.
MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE DB-63 INTO
EVIDENCE.
THE COURT: FOR THE PURPOSES OF DEMONSTRATING HOW THE
HYPOTHESIS WAS ARRIVED AT. BUT NOT FOR THE PROOF OF THE MATTERS
ASSERTED THEREIN. I WILL ADMIT IT.
MR. BOGER: YOUR HONOR, LET ME ASK FOR CLARIFICATION.
AND THAT IS. IS THIS DOCUMENT IN EVIDENCE FOR FROOF
THAT PROFESSOR BALDUS” STUDY, CHARGING AND SENTENCING STUDY
GENERATED THESE PERCENTAGES FOR BLACK DEFENDANT, WHITE VICTIM
COMBINATIONS AND SQ FORTH?
THE COURT: YES.
MR. BOGER: ALL RIGHT.
THE COURT: 1 THINK THAT-S IMPLICIT IN WHAT I HAVE
SAID. IN OTHER WORDS, HE STARTED OUT WITH A ROUGH TEST OF HIS
O
N
t
s
0
4h
3
N
w
O
E
a
NE
Y
A
R
E
E
E
o
b
B
BALDUS - DIRECT
HYPOTHESIS TO SEE WHETHER EVEN THE RAW DATA WOULD HELP. AND HE
GOT THIS.
NOW, THIS DOES NOT IN THE COURT’S VIEW MAKE MORE OR
LESS LIKELY ANY FACT OF EVIDENCE, ANY FACT OF CONTROVERSY IN
THIS CASE BECAUSE OF THE REASONS THAT HAVE REALLY BECOME MATTER
OF LAW REASONS AFTER I HEARD THE EVIDENCE IN MCCORQUODALE. AND
THAT IS. THAT THIS IS SO CRUDE. SO UNADJUSTED AND ACADEMICALLY
S0 UNRELIABLE THAT IT IS NOT WORTHY OF ANY WEIGHT AS EVIDENCE IN
A COURT OF LAW, ALTHOUGH FOR PURPOSES OF DESIGNING THE
SUBSEQUENT STEPS, I THINK IT“S USEFUL TO SEE HOW HE GOT FROM
POINT A TO FOINT B.
MR. BOGER: ALL RIGHT. THANK YOU. THATS WHAT I
THOUGHT YOUR HONOR MEANT. I JUST WANTED TO BE CLEAR.
BY MR. BOGER:
@. PROFESSOR BALDUS, LET ME ASK YOU IF YOU HAD EVER AT ANY TIME
CONTROLLED FOR THE THE PRESENCE OF ANY BACKGROUND VARIABLE WHEN
LOOKING AT THESE RACIAL FACTORS?
A. YES, 1 DID. THE ENTIRE STRATEGY OF THE RESEARCH WITH
RESPECT TO THE RACIAL EFFECT WAS TO TEST AS MANY DIFFERENT
POSSIBLE COMBINATIONS OF BACKGROUND FACTORS THAT WE COULD
CONCEIVE OF TO SEE IF THEY WOULD AFFECT THE UNADJUSTED DISPARITY
THAT WE’VE JUST SEEN IN THESE PRICR TABLES. THAT WAS THE WHOLE
BURDEN OF THE ENTERPRISE.
THE COURT: THATS WHAT I THOUGHT YOU WERE WORKING ON
ALL THESE YEARS.
HE
E
BE
a
E
E
SE
735
BALDUS - DIRECT
BY MR. BOGER:
@. LET’S LOOK NOW AT DB-69 IDENTIFICATION AND ILL ASK YOU TO
IDENTIFY IT. PROFESSOR BALDUS?
A. THE EXHIBIT DB-6% PRESENTS THE RACE OF VICTIM AND RACE OF
DEFENDANT DISPARITIES, ESTIMATED, WITH THE DATA IN THE CHARGING
AND SENTENCING STUDY. AFTER CONTROLLING FOR EACH RACIAL VARIABLE
RESPECTIVELY.
PANEL 1 SHOWS THE RACE OF VICTIM DISPARITIES. AFFAIR
CONTROLLING FOR THE RACE OF THE DEFENDANT. AND IT INDICATES
THAT WHEN A BACKGROUND CONTROL IS INTRODUCED FOR THE RACE OF
DEFENDANT THAT THE RACE OF VICTIM EFFECT INCREASES.
SPECIFICALLY WHAT WE SEE IS THAT AMONG BLACK DEFENDANT
CASES, THERE IS A 21 PERCENTAGE POINT DIFFERENCE IN DEATH
SENTENCING RATE FOR WHITE VICTIM CASES AND BLACK VICTIM CASES,
WITH THE RATE HIGHER IN THE WHITE VICTIM CASES.
AMONG WHITE DEFENDANT CASES. YOU SEE A FIVE PERCENTAGE
POINT DISPARITY IN THE DEATH SENTENCING RATE BETWEEN WHITE
VICTIM CASES AND BLACK VICTIM CASES. AGAIN THE RATE IS HIGHER
FOR THE WHITE VICTIM CASES.
PANEL 2 INDICATES THAT WHEN A BACKGROUND CONTROL FOR
RACE OF VICTIM IS INTRODUCED THAT THERE IS A 14 PERCENTAGE POINT
DISPARITY ASSOCIATED WITH THE RACE OF THE DEFENDANT AMONG WHITE
VICTIM CASES.
AND THAT THERE 1S. HOWEVER, MINUS. OR A NEGATIVE
DIFFERENCE AMONG BLACK VICTIM CASES.
——— A——— ———— ———— — —— — ——— ——— | —— —, ——— T— —— p—— —
736
BALDUS - DIRECT
1 THESE RESULTS ARE A FURTHER REFINED HYPOTHESIS AND
2 SUGGEST THAT THE PRINCIPAL SOURCE OF RACE OF VICTIM DISPARITIES
3 1S PROBABLY FOUND AMONG THE BLACK DEFENDANT CASES AND THAT THE
4 PRINCIPAL SOURCE OF THE RACE OF DEFENDANT DISPARITIES IS A
3 PRODUCT OF THE TREATMENT OF WHITE VICTIM CASES. THAT'S THE
& MESSAGE THAT THESE TABLES SUGGEST. BY WAY OF FURTHER DEVELOPMENT
7 OF THE HYPOTHESIS.
38 MR. BOGER: YOUR HONOR. AT THIS TIME, --
? THE COURT: LET ME THINK ABOUT WHAT YOU JUST SAID.
10 THE WITNESS: YES.
11 THE COURT: YOU WOULD LOOK AT THE SYSTEM'S TREATMENT OF
12 THE KILLING OF A WHITE VICTIM TO DETERMINE THE SYSTEM’S
13 TREATMENT OF BLACK AND WHITE DEFENDANTS. IN OTHER WORDS. YOU
14 WOULD LOOK AT THE WHITE VICTIM POPULATION TO SEE HOW THE RACE OF
13 THE DEFENDANT BECOMES A FACTOR.
1% THE WITNESS: THAT’S WHATS SUGGESTED BY THIS. YOUR
37 HONOR, THAT THE --—
18 THE COURT: THAT’S WHAT YOU SAID? I DIDN‘T WANT TO GET
“ 19 IN TO THE ~~
2Q THE WITNESS: QUITE RIGHT, YOUR HONOR. THAT THE RACE
21 OF DEFENDANT EFFECTS HERE APPEAR TO BE CONCENTRATED IM A
22 SUUB-GROUP OF THE CASES.
23 BY MR. BOGER:
24 @. DO THESE FIGURES IN TABLE &, MARKED AS DB-69, SUGGEST THAT
23 RACE OF VICTIM EFFECTS GO AWAY WHEN RACE OF DEFENDANT IS TAKEN
v
O
N
N
h
n
——— — —————————. ————— ————— ———————— —r—
737
BALDUS - DIRECT
INTO ACCOUNT OR CONTROLLED FOR?
A. NO, THEY DO NOT. IN FACT, THEY INCREASE. THAT IS. THE
UNADJUSTED MEASURE OF DISPARITY FOR RACE OF VICTIM WAS, I
BELIEVE, TEN PERCENTAGE POINTS, AND THAT DISPARITY INCREASES.
SOMEWHAT, AS YOU CAN SEE HERE, WHEN YOU INTRODUCE A CONTROL FOR
RACE OF DEFENDANT.
AND SIMILARLY. YOU CAN SEE THAT THE NEGATIVE DIFFERENCE
ASSOCIATED WITH BEING A BLACK DEFENDANT THAT 1S ESTIMATED WHEN
YOU CALCULATE THE DIFFERENCE WITHOUT ANY BACKGROUND CONTROL AT
ALL, CHANGES WHEN YOU CONTROL FOR THE RACE OF THE VICTIM, SO
THAT IN TABLE & YOU SEE WHEN THE RACE OF VICTIM CONTROL IS
INTRODUCED THAT THERE IS A RACE OF VICTIM. CORRECTION. RACE OF
DEFENDANT DISPARITY THAT IS CONCENTRATED AS I INDICATED IN THE
WHITE VICTIM CASES.
MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE THE
ADMISSION OF DB-469 INTO EVIDENCE.
THE COURT: I WOULD NOT ADMIT IT FOR ANY REASON OTHER
THAN THAT WHICH I HAVE JUST ADMITTED, WHAT. -437
MR. BOGER: -43, I BELIEVE THAT-S CORRECT, YOUR HONOR.
THE COURT: DO YOU HAVE ANY OTHER SPECIAL OBJECTIONS,
MS. WESTMORELAND. UNDERSTANDING THAT THAT IS THE ONLY PURPOSE
FOR WHICH IT IS BEING ADMITTED?
MS. WESTMORELAND: NO, YOUR HONOR.
THE COURT: DEMONSTRATE PROCESS.
BY MR. BOGER:
738
BALDUS - DIRECT
1 @. NOW. PROFESSOR BALDUS, DB-6%, EMPLOYS WHAT MEASUREMENT OR
2 |METHOD OF MEASURING THE DISPARITIES IN THE CONTROL?
3 |A. THIS IS A CROSS TABULATION METHOD FOR ADJUSTING FOR THE
4 BACKGROUND EFFECTS OF THE OTHER RACIAL VARIABLE.
s |@. DO YOU EMPLOY OR HAVE YOU EMPLOYED ANY OTHER METHOD FOR
b MEASURING THE, OR ADJUSTING FOR THE EFFECTS OF THE BACKGROUND
7 VARIABLE OF RACE?
8 A. YES. 1 HAVE.
9 |@. CAN YOU IDENTIFY DB-70 MARKED FOR IDENTIFICATION?
10 |A. YES. DB-70 IS A TABLE FROM OUR FINAL REPORT WHICH ESTIMATES
11 USING MULTIPLE REGRESSION PROCEDURES RACE OF VICTIM, AND RACE OF
12 DEFENDANT CO-EFFICIENTS, EACH ONE HOLDING CONSTANT THE OTHER
13 |RACIAL VARIABLE.
14 AND AS ONE CAN SEE IN THE SECOND COLUMN OF THE TABLE
15 THAT WHEN THE RACE OF DEFENDANT IS CONTROLLED FOR, THAT THE RACE
16 |OF VICTIM CO-EFFICIENT ESTIMATED WITH A WEIGHTED LEAST SQUARES
17 PROCEDURE, SUGGESTS THAT THE DEFENDANTS WITH WHITE VICTIMS, HAVE
12 |A 17 PERCENTAGE POINTS HIGHER CHANCE OF RECEIVING A DEATH
» 19 |SENTENCE THAN DEFENDANTS WHOSE VICTIMS BLACK.
20 AND THE LOGISTIC REGRESSION RESULTS SHOW COMPARABLE
21 EFFECTS, INDICATING THAT THE DEFENDANT WHOSE VICTIM IS WHITE HAS
22 |A 13.5 TIMES GREATER ODDS OF RECEIVING THE DEATH SENTENCE THAN
23 THE DEFENDANT WHOSE VICTIM IS BLACK.
24 THE SECOND ROW OF THE ANALYSIS REPORTED HERE INDICATES
<3 THAT THE ESTIMATED EFFECT OF BEING BLACK ON A DEFENDANTS
739
BALDUS - DIRECT
1 CHANCES OF RECEIVING A DEATH SENTENCE ARE TEN PERCENTAGE FOINTS
2 HIGHER THAN WHEN THE RACE OF VICTIM IS ADJUSTED.
3 CORRECTION, WHEN ADJUSTMENT IS MADE FOR THE BACKGROUND VARIABLE.
4 RACE OF VICTIM, AND THAT THE ODDS OF RECEIVING DEATH SENTENCE
bw’ ARE ALSO ELEVATED BY A FACTOR OF 3.
% & MR. BOGER: YOUR HONOR, AT THIS TIME, I WOULD MOVE
7 DB-70 INTO EVIDENCE.
8 THE COURT: FOR THE SAME PURPOSES AS THE LAST TWO
© EXHIBITS, I WILL ADMIT IT.
10 MR. BDGER: THANK YOU. YOUR HONOR.
11 BY MR. BOGER:
12 @. PROFESSOR BALDUS, HAVE YOU AT ANY POINT ATTEMFTED TO CONTROL
13 FOR THE RACE, NOT ONLY THE OTHER RACIAL FACTOR, BUT TO CONTROL
14 FOR OTHER BACKGROUND FACTORS AS WELL, WHICH MIGHT OTHERWISE
15 PRESENT A RIVAL HYPOTHESIS?
16 A. YES. OUR FIRST STEF WAS TO IDENTIFY WHAT WE BELIEVED ON THE
17 BASIS OF OUR EXPERIENCE AND WORK IN THIS AREA, THE FIVE FACTORS
13 THAT PRESENTED THE MOST PLAUSIBLE RIVAL HYPOTHESES. AND THIS
“ 19 JUDGMENT WAS BASED ON MY KNOWLEDGE OF THE CRIMINAL JUSTICE
20 SYSTEM.
21 HERE THE STUDY OF THE GEORGIA SUPREME COURT REPORTS,
22 STUDY OF THE EMPIRICAL LITERATURE, ON THE FACTS THAT TEND TO
23 EXPLAIN DEATH SENTENCING RESULTS. BUT MORE ESPECIALLY UPON MY
24 PRIOR ANALYSIS OF CAPITAL SENTENCING. DATA SETS IN CALIFORNIA,
23 IN ARKANSAS, AND FARTICULARLY HERE.
Rd)
(43
)
~
oo
B
Ww
nN
>
oO
a
0)
BALDUS - DIRECT
@. WHAT WERE THE FIVE FACTORS THAT YOU HAD IDENTIFIED AS
POSSIBLE LIKELY RIVAL HYPOTHESES?
A. THE FIVE FACTORS WERE THE PRESENCE OF FELONY
CIRCUMSTANCES IN THE CASE. THE PRIOR RECORD, --
@. ARE THOSE FIVE REFLECTED IN YOUR REPORT, PROFESSOR BALDUS?
A. YES, THEY ARE. I WANT TO MAKE SURE I“M READING OFF THE
RIGHT ONES.
@. WELL LET ME DEFER THAT QUESTION. WE WILL GET TO THAT LATER.
IF YOU WILL.
A. YES.
@. DO YOU HAVE ANY RESULTS FROM THE ATTEMPT TO CONTROL FOR
FELONY CIRCUMSTANCE. AND RACIAL VARIABLES?
A. YES, I DQ.
THE COURT: WHAT IS FELONY CIRCUMSTANCE? IS THAT THE
SAME THING AS CONTEMPORANEOUS OFFENSE?
THE WITNESS: YES, YOUR HONOR, IT’S THE FRESENCE OF
A CONTEMPORANEOUS OFFENSE.
THE COURT: BZ.
THE WITNESS: DEFINED AS A FELONY. NOT NECESSARILY B2,
YOUR HONOR, BECAUSE THERE ARE OTHER FELONIES INVOLVED IN CASES
THAT DO NOT MAKE ONE DEATH ELIGIBLE UNDER B2. IT”3 SLIGHTLY
BROADER THAN A B2 FACTOR.
BY MR. BOGER:
Q. SO FOR THIS DEFINITION AS OPPOSED TO OTHERS THAT MIGHT COME,
THE FELONY CIRCUMSTANCES VARIABLE IS BROADER THAN THE STATUTORY
BALDUS - DIRECT
1 B2 FACTOR?
2 A. YES. JUST SLIGHTLY BROADER.
3 @. ALL RIGHT. AND CAN YOU IDENTIFY DB-717
4 A. YES. DB-71 IS A TABLE. 8 FROM OUR REPORT. WHICH INDICATES THE
5 RACE OF VICTIM DISPARITIES AFTER ADJUSTING FOR THE PRESENCE OF
% & FELONY CIRCUMSTANCES IN THE CASE. AND RACE OF THE DEFENDANT.
7 R. AND WHAT DO THE FIGURES. REFLECT?
3 A. THE FIGURES INDICATE THAT AMONG CASES WHERE A FELONY
9 CIRCUMSTANCE IS PRESENT, THAT THE RACE OF VICTIM EFFECT PERSISTS
10 AND IS STRONGER THAN IT IS AMONG ALL CASES, BEFORE ONE ADJUSTS
11 FOR THIS BACKGROUND FACTOR.
2 HOWEVER, THE TABLE ALSO INDICATES THAT AMONG THE CASES.
13 AND THESE ARE INDICATED IN PART B OF THE TABLE, IN WHICH THERE
14 IS NO FELONY CIRCUMSTANCE PRESENT IN THE CASE, THE DEATH
13 SENTENCING RATES ARE SUBSTANTIALLY REDUCED. RANGING IN THESE
16 FOUR SUB-GROUPS OF CASES FROM ZERO TO .0S.
17 AND IN ADDITION, THE RACE OF VICTIM EFFECTS ARE VERY
18 SMALL AS WELL.
w 19 THE COURT: IM SORRY, PROFESSOR BALDUS, I LET MY MIND
20 WANDER, WHICH IS A GOOD INDICATION ITS TIME TO GO TO LUNCH.
21 LETS GO TO LUNCH RIGHT NOW. WHEN YOU COME BACK. IM
22 GOING TO GET YOU TO EXPLAIN WHAT YOU JUST SAID.
23 THE WITNESS: VERY WELL, YOUR HONOR.
24 THE COURT: WE-LL RECESS UNTIL QUARTER OF.
25 GEIL
BALDUS - DIRECT
Fy
(COURT RECESSED FOR LUNCH.)
THE COURT: MR. BOGER. YOU-VE GOT THE FRONT LINE 2
3 COURTROOM DEPUTY CORPS IN HERE TODAY, AND IF YOU MESS AROUND
4 WITH THE EXHIBITS, YOU‘RE GOING TO BE IN BIG TROUBLE.
5 : MR. BOGER: I BELIEVE IVE BEEN BEFORE HIM ONCE BEFORE.
EA & THE COURT: OKAY. GO AHEAD.
7 MR. BOGER: I WOULD LIKE TO CLARIFY AT THE OUTSET, YOUR
9 HONOR, SOME MATTERS IN REGARD TO THE LAST FEW EXHIBITS, THE ONES
9 HAVING TO DO WITH RACIAL DISPARITY AND WOULD LIKE TO ASK
10 PROFESSOR BALDUS ANOTHER QUESTION OR TWO ABOUT THOSE EXHIBITS IF
11 I MIGHT.
12 THE COURT: YES.
13 MR. BOGER: I THINK I PERHAPS FAILED TO A LAY
14 FOUNDATION FOR THEIR COMPLETE ADMISSIBILITY.
15 -ll. -
16 DAVID C. BALDUS,
17 BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
13 TESTIFIED FURTHER AS FOLLOWS:
19 DIRECT EXAMINATION (CONT D)
20 BY MR. BOGCER:
21 @. PROFESSOR BALDUS, WITH REGARD TO THE UNADJUSTED RACE OF
22 VICTIM AND RACE OF DEFENDANT DISPARITIES IN DEATH SENTENCING
23 RATES, AND THE CROSS TABULATIONS AND REGRESSIONS OF THOSE RACIAL
24 FACTORS WHERE YOU CONTROL FOR ONE OR ANOTHER VARIABLE. THE
23 VARIABLE OF THE OTHER RACIAL INTERESTS THAT ARE REFLECTED IN
BALDUS - DIRECT
DB-62 AND --3, AND DB. I BELIEVE. =-6% AND -70, DID YOU GENERATE
THOSE TABLES?
A. YES.
@. AND DO THEY PLAY ANY PART AT ALL IN YOUR EXPERT OPINION
EXPRESSED EARLIER THIS MORNING ABOUT THE ULTIMATE CONCLUSIONS
THAT RACIAL FACTORS PLAY A PART IN THE GEORGIA CHARGING AND
SENTENCING SYSTEM?
A. YES.
@. ARE THOSE DATA ALONE SUFFICIENT TO SUPPORT YOUR CONCLUSIONS?
A. NO.
@. WHAT, DO THEY NEVERTHELESS FORM SOME PART OR SOME BASIS FOR
YOUR CONCLUSION?
A. YES, THE ENTIRE ANALYSIS WAS FOCUSED AROUND THOSE INITIAL
UNADJUSTED DISPARITIES. THEY WERE THE POINT TO WHICH EVERY
COMPARISON WAS MADE WITH, OF THE REGRESSION CO-EFFICIENTS THAT
WERE ESTIMATED AFTER CONTROLLING FOR A VARIETY OF FACTORS. THEY
WERE A CENTRAL PART OF THE WHOLE ANALYSIS.
MR. BOGERS YOUR HONOR, LET ME RENEW MY MOTION TO ADMIT
THEM GENERALLY, AND SAY AT THE OUTSET THAT I FULLY UNDERSTAND.
PERHAPS I DON‘T FULLY UNDERSTAND, BUT I HOPE TO UNDERSTAND YOUR
HONOR‘S POSITION ABOUT THE MCCORQUODALE CASE AND THE HOLDING IN
THAT CASE. AND ADAMS AND SMITH V. BALKCOM, THAT THESE DATA ALONE
ARE NOT SUFFICIENT TO PROVE THE CONTENTION THE PETITIONER HAS
SET FORWARD, BUT WOULD URGE UPON THE COURT THAT PROFESSOR BALDUS
HAS TESTIFIED THAT THEY ARE PART, FORM PART OF THE BASIS OF HIS
744
BALDUS - DIRECT
CONCLUSION, AND THE ANALOGY THAT I WOULD USE TO THE COURT IS THE
py
nN
FOLLOWING ONE.
IN A MURDER CASE, THERE WOULD BE NO CONVICTION BASED
UPON A MURDER WEAPON ALONE WITHOUT OTHER EVIDENCE THAT WOULD
SHOW WHO PULLED THE TRIGGER OF THE WEAPON AND WHETHER THERE WAS
INTENT AND SO FORTH.
IN THE SAME SENSE, ALTHOUGH PERHAPS A SLIGHTLY
DIFFERENT SENSE. THESE DATA. ALTHOUGH NOT SUFFICIENT TO SUPPORT
ANY PRINCIPLE THAT COULD BE RECOGNIZED UNDER LAW, NEVERTHELESS
o
O
M
o
o
8
M
W
Fy
ARE PART OF OR TO USE YOUR HONOR‘S TERMS THIS MORNING, ARE
fo
ry
BRICKS IN THE BUILDING, THE OPINION OF PROFESSOR BALDUS. AND
12 THEREFORE WE WOULD RESPECTFULLY RENEW THE MOTION TO ADMIT THEM
13 GENERALLY AT THIS TIME. SUBJECT, OF COURSE, TO THE STATE'S
14 OBJECTIONS IN BRIEFS OR AT ANY LATER POINT AS TO THEIR WEIGHT.
15 THE COURT: I GATHER THAT YOU DEEM IT VERY IMPORTANT
14 FOR EVERYONE TO BE IMPRESSED WITH THE THOUGHTFUL AND SCHOLARLY
17 Way IN WHICH THIS WAS UNDERTAKEN, THIS STUDY.
18 MR. BOGER: I HAVE A SENSE IT’S IMPORTANT. YES, YOUR
e 19 HONDR.
20 THE COURT: TO THE EXTENT THAT THAT IS IMPORTANT, I
21 THINK IT IS INTERESTING TO LEARN HOW THE PROCESS EVOLVED. 1
22 UNDERSTAND THAT PROFESSOR BALDUS HAS TESTIFIED THAT THESE WERE
23 IMPORTANT TO HIM AT LEAST IN TERMS OF PRELIMINARY TESTING OF
24 HYPOTHESES TRYING TO FORMULATE HYPOTHESES TO TEST LATER.
28 TO THE EXTENT THEY SHOW PROCESS. THEY RE ADMITTED.
745
BALDUS - DIRECT
BECAUSE OF THEIR SCIENTIFIC UNSOPHISTICATION AS
pt
e
[3
ATTEMPTS TO MEASURE THE FACTS OF CONSEQUENCE IN THIS CASE, THEY
ARE AS A MATTER OF LAW IN THIS CIRCUIT. BASED ON THE MANY TIMES
THAT YOU HAVE BROUGHT BOWERS AND -- WHO IS THE OTHER GUY?
MR. BOGER: PIERCE. .
THE COURT: PIERCE‘S STUDY, WHICH IS, MY PERSONAL VIEW
1S THAT DOCTOR BALDUS HAS DEMONSTRATED A GREAT DEAL MORE CARE
THAN THEY DID. BUT ASSUMING. PUTTING THAT ASIDE. BASICALLY.
WHERE WE WERE IN THESE EARLY THINGS IS ABOUT WHERE BALDUS AND
©
Wu
£0
N
o
e
o
A
d
W
P
y
PIERCE LEFT OFF. WE HAVE SEEN THE COURT RULE TIME AND TIME
ey
Ty
AGAIN THAT IF THATS ALL YOU GOT, YOU HAVEN‘T FROVED ANYTHING.
12 MY VIEW, EVEN BEYOND THAT INTERPRETATION OF DECISIONS
13 OF OUR CIRCUIT IS THAT BECAUSE OF THEIR RELATIVE
14 UNSOPHISTICATION IN ATTEMPTING TO MEASURE THE MANY MANY OTHER
15 REAL VARIABLES THAT ARE AT WORK IN THE SYSTEM. THAT THEY DO NOT
16 EVEN MAKE THE FACT IN CONSEQUENCE MORE OR LESS LIKELY.
17 IN OTHER WORDS, I FIND THAT THEY ARE NOT RELEVANT,
18 BECAUSE THEY, TO PROVE THE CONTENTION, BECAUSE IN MY VIEW. THEY
% 19 ARE SO WITHOUT PROBATIVE FORCE AS TQ LEAVE US WITH NOTHING AS A
20 MATTER OF LAW DEMONSTRATED. ONCE THEY ARE PUT UPON THE RECORD OF
21 THE COURT.
22 THE BASIS OF MY VIEW, YOU OBVIOUSLY HAVE THEM IN THE
23 RECORD, AND IF THE COURT OF APPEALS OR EVEN THE SUPREME COURT
24 DOES NOT AGREE WITH MY VIEW, THEN THEY CAN REVERSE ME. ADMIT
23 THEM AND USE THEM FOR ANY PURPOSE THEY WANT.
7446
BALDUS - DIRECT
MR. BOGER: ALL RIGHT, YOUR HONOR. I CERTAINLY WILL
ACCEPT YOUR RULING, WITH THE FOUNDATION WE“VE LAID AND WITH THE
STATEMENT WE“VE MADE ABOUT IT, AND PROCEED FROM THERE.
THE COURT: LET ME SAY THIS. I UNDERSTAND YOU TO HAVE
ASSERTED THROUGHOUT THAT ANYTHING THAT AN EXPERT RELIES UPON IS
ADMISSIBLE IN EVIDENCE. IF THAT IS CORRECT, THEN I HAVE
MISUNDERSTOOD THE RULES OF EVIDENCE.
I DON’T UNDERSTAND THAT TO BE THE LAW.
BEYOND THAT, -——
MR. BOGER: WELL, THAT”S NOT THE BASIS FOR OUR --
THE COURT: LET ME SAY, IF YOU HAVEN‘T GOT ANYTHING
MORE TO BASE DOCTOR BALDUS” OPINION ON THAN THESE STUDIES. YOU
SURE HAVE WASTED A LOT OF MY TIME.
MR. BOGER: 1 DO HOPE THE REST OF OUR BOOK WILL GO
FURTHER THAN THAT. YOUR HONOR.
"THE COURT: OKAY.
BY MR. BOGER:
Q@. PROFESSOR BALDUS. YOU TESTIFIED JUST BEFORE LUNCH THAT YOU
HAD IN FACT TAKEN SOME STEPS TO CONTROL NOT ONLY FOR THE RACE
FACTORS BUT FOR OTHER IMPORTANT BACKGROUND OR CONTROL FACTORS AS
WELL. AND I BELIEVE WE HAD REFERRED YOUR ATTENTION TO DB-71 FOR
IDENTIFICATION.
CAN YOU IDENTIFY THAT DOCUMENT?
A. YES. DB-71 IS TABLE 8 FROM OUR REPORT, OUR REPORT FROM THE
CHARGING AND SENTENCING STUDY. RACE OF VICTIM. DISPARITIES IN
747
BALDUS - DIRECT
DEATH SENTENCING RATES AFTER ADJUSTMENT FOR THE PRESENCE OF
pa
2 FELONY CIRCUMSTANCES IN THE CASE.
3 I ESTIMATED WITH A CROSS TABULATION METHODOLOGY. THESE
4 RESULTS SHOW THE RACE OF VICTIM EFFECTS AFTER ADJUSTMENT WAS
3 MADE FOR ONE BACKGROUND FACTOR. THAT IS, WHETHER OR NOT FELONY
» & CIRCUMSTANCES WERE INVOLVED IN THE CASE.
7 IF YOU CAN LOOK AT PART A OF THE TABLE. YOU SEE
a ORGANIZED THERE THE CASES IN WHICH THERE WAS A FELONY
g CONTEMPORANEOUS CIRCUMSTANCE INVOLVED IN THE CASE.
10 THE CASES IN THAT PANEL OF THE TABLE ARE BROKEN DOWN
-31 ACCORDING TO WHETHER THE DEFENDANT WAS BLACK OR WHITE.
12 WITHIN THE CATEGORY IN THE SECOND COLUMN, WHERE THE
13 BLACK DEFENDANT CASES ARE LOCATED. THERES A FURTHER BREAKDOWN
14 BETWEEN THOSE CASES INVOLVING WHITE AND BLACK VICTIMS.
135 THE SIMPLE MEASURES OF THE DIFFERENCE IN DEATH
16 SENTENCING RATES, AND THE RATIO BETWEEN THOSE RATES. SHOW THAT
17 WITHIN THESE CASES INVOLVING BLACK DEFENDANTS WHERE FELONY
18 CIRCUMSTANCES WERE PRESENT, THAT THERE IS A 21 PERCENTAGE POINT
. 19 RACE OF VICTIM DISPARITY. AND A RATIO OF FOUR TO ONE, WHICH
20 PROVIDE MEASURES OF THE HEIGHTENED RISK FOR OFFENDERS WHEN THEIR
21 VICTIM IS WHITE.
22 THE THIRD COLUMN IN THE FIRST PART OF THE TABLE SHOWS
23 THE EFFECT WHEN THE DEFENDANT IS WHITE. THERE YOU SEE THAT THE,
24 THERE IS A SLIGHT RACE OF VICTIM EFFECT. I USE THE WORD SLIGHT,
25 BECAUSE THE NUMBER OF BLACK VICTIM CASES IS QUITE SMALL.
748
BALDUS - DIRECT
ESTIMATED TO BE ONLY TWELVE IN NUMBER, EVEN THOUGH THE DISPARITY
IS IN THE ORDER OF SIX PERCENTAGE POINTS.
IN CONTRAST TO THOSE DISPARITIES THAT ONE OBSERVES WHEN
A FELONY CIRCUMSTANCE IS PRESENT. WE SEE AMONG THE CASES WHERE A
FELONY CIRCUMSTANCE IS NOT PRESENT SIGNIFICANTLY SMALLER RACE OF
VICTIM EFFECTS. AMONG BLACK DEFENDANTS THERE. THE RACE OF
VICTIM DISPARITY IS FOUR PERCENTAGE POINTS.
AND AMONG WHITE DEFENDANT CASES, IT’S 3 POINTS.
WHAT WE SEE HERE IS A PATTERN THAT RUNS THROUGHOUT THE
STUDY. THAT IS, THAT AS THE CASES BECOME MORE AGGRAVATED WHEN.
BECAUSE OF THE PRESENCE OF AGGRAVATING CIRCUMSTANCES. THE
DEATH SENTENCING RATE GOES UP.
ALONG WITH THAT RISE, WE SEE A RISE IN THE RACE OF
VICTIM DISPARITIES.
THAT”S WHAT’S APPARENT HERE. THAT’S WHY THE RACE OF
VICTIM DISPARITIES ARE SO MUCH SHARPER IN THE FIRST FART OF THE
TABLE THAN IN THE SECOND PART.
@. SO NOT ONLY ARE THE SENTENCING RATES HIGHER WHERE FELONY
CIRCUMSTANCES ARE, OR ACCOMPANY THE HOMICIDE, BUT YOUR TESTIMONY
IS THAT THE DISPARITY BETWEEN THE TREATMENT OF RACES, BASED ON
THE RACE OF VICTIM. RISES AS WELL?
A. YES, IN A CASE LIKE THIS, WHERE THIS FACTOR IS OF TREMENDOUS
IMPORTANCE TO THE SYSTEM, THAT“S REFLECTED BY THE QUITE LOW
DEATH SENTENCING RATE AMONG CASES THAT DO NOT INVOLVE FELUNY
CIRCUMSTANCES, 1T-°S VERY DIFFICULT FOR ANY VARIABLE TD HAVE A
Ww
6
N
r
749
BALDUS —- DIRECT
SIGNIFICANT STATISTICAL EFFECT AMONG THAT GROUP OF CASES WHERE
THE DEATH SENTENCING RATE RANGES ONLY FROM ZERO TO FIVE.
ITS ONLY IN THE CATEGORY OF CASES WHERE YOU HAVE AN
ELEVATED DEATH SENTENCING RATE ARE YOU GOING TO OBSERVE A
STATISTICAL EFFECT OF ANY PARTICULAR FACTOR, AND YOU SEE IT HERE
WITH RESPECT TO THE RACE OF THE VICTIM.
MR. BOGER: YOUR HONOR, AT THIS TIME, I MOVE THE
ADMISSION OF DB-71 INTO EVIDENCE.
THE COURT: LET ME SEE IF I CAN UNDERSTAND THIS.
THERE WERE, LOOKING AT "A," 47 CASES WHERE THE DEATH
PENALTY WAS IMPOSED, WHERE THE DEFENDANT WAS BLACK, THE VICTIM
WAS WHITE, AND THERE WERE, QUOTE, FELONY CIRCUMSTANCES.
THE WITNESS: YES, YOUR HONOR.
THE COURT: THERE WERE 149 CASES WHERE ALL OF THAT WAS
PRESENT, BUT THE DEATH PENALTY WAS NOT IMPOSED.
THE WITNESS: NO, YOUR HONOR. I“M SORRY. 149 IS THE
DENOMINATOR: WHICH INCLUDES THE DEATH AND LIFE CASES.
THE COURT: YOU“RE RIGHT. I“M SORRY. THAT IS WHAT I
MEANT TO SAY. THAT’S THE UNIVERSE.
ALL RIGHT. TO SEE THE EFFECT THEN ON RACE OF THE
VICTIM, IS THAT WHAT YOU’RE MEASURING?
THE WITNESS: YES, YOUR HONOR. WERE LOOKING AT THE
RACE OF THE VICTIM EFFECT CONTROLLING FOR THE RACE OF DEFENDANT.
THATS WHY WEVE DIVIDED THE CASES ALONG THE DIMENSION OF THE
RACE OF THE DEFENDANT.
fo
b
Se
M
e
,
S
l
SE
S
E
E
T
E
OE
BALDUS - DIRECT
THE. THE UNADJUSTED MEASURES THAT WE LOOKED AT EARLIER
INDICATED THAT THE RACE OF VICTIM EFFECT IS SOMETHING THAT IS
MOST APPARENT IN CASES INVOLVING BLACK DEFENDANTS.
THE COURT: ALL RIGHT.
THE WITNESS: AND —-
BY MR. BOGER:
Q. SO PROFESSOR BALDUS, IF A FELONY CIRCUMSTANCE IS PRESENT,
AND THE DEFENDANT IS BLACK. YOUR DATA SHOW THAT A BLACK
DEFENDANT WHO HAS KILLED A WHITE VICTIM RUNS A .28, OR DEATH
SENTENCES ARE IMPOSED AT A .28 PERCENT. WHEREAS IF THE VICTIM IS
BLACK WITH A BLACK DEFENDANT WHO HAS COMMITTED A HOMICIDE WITH A
FELONY CIRCUMSTANCE IT’S .07. IS THAT CORRECT?
A. THAT’S CORRECT.
@. SO THE DISPARITY YOU CALCULATE BY SUBTRACTING THAT .28 FROM
THE .07. TO REACH YOUR RESULT OF .21, OR 21.
A. RIGHT.
THE COURT: IF THERE’S BLACK VICTIM AND A WHITE PERSON
KILLS HIM, HE HAS A GREATER LIKELIHOOD THAN IF A BLACK PERSON
KILLS A BLACK PERSON.
THE WITNESS: YES. YOUR HONOR. AGAIN THATS BASED ON
ESTIMATED POPULATION OF 12 CASES.
THE COURT: THERE'S A QUESTION AS TO WHETHER THAT'S A
STATISTICALLY SIGNIFICANT RESULT, BUT THAT'S WHAT IT SHOWS.
THE WITNESS: YES, THAT'S RIGHT. EXACTLY.
THE CONCLUSION ONE REACHES FROM EXAMINING THIS TABLE IZ
An. J —— SO ———— rp —— ——" ————— ese eererenet| ee. So———— ee ed PO, pre ee a—
-
0
N
O
s
W
M
10
11
731
BALDUS - DIRECT
THAT THE INTRODUCTION OF A BACKGROUND CONTROL FOR THIS VARIABLE
DOES NOT EXPLAIN THE RACE OF VICTIM EFFECT THAT WE SAW IN THE
UNADJUSTED MEASURES.
IT’S TRUE THAT WITHIN THE ONE SUB-POPULATION OF CASES,
IT’S CREATED BY THE INTRODUCTION OF THIS CONTROL. THAT THE
DISPARITIES DROPPED SUBSTANTIALLY, THAT IS, AMONG CASES
INVOLVING NO FELONY CIRCUMSTANCES. THERE IS VERY SHARP REDUCTION
IN TERMS OF THE AVERAGE EFFECT.
BUT AMONG THE CASES INVOLVING FELONY CIRCUMSTANCES. THE
EFFECT IS ENHANCED AMONG THE BLACK VICTIM CASES, WHICH IS THE
PRINCIPAL SOURCE OF THE RACE OF VICTIM DISPARITIES.
MR. BOGER: YOUR HONDR., AT THIS TIME. I WOULD RENEW MY
MOTION FOR THE ADMISSION OF DB-71 INTO EVIDENCE.
THE COURT: OTHER THAN YOUR CONTINUING OBJECTION. DO
YOU HAVE ANY PROBLEM?
MS. WESTMORELAND: NO, YOUR HONOR. THE CONTINUING
OBJECTIONS AND ALSO I WOULD JUST NOTE THAT I WOULD SUBMIT THE
CONTROLLING FOR FELONY CIRCUMSTANCES STILL IS INSUFFICIENT TO
SHOW THE FINAL ISSUE THAT IS BEFORE THE COURT.
THE COURT: ALL RIGHT. LET ME ASK YOU ONE OTHER
GUESTION. I THINK YOU HAD BEST DESCRIBE PRECISELY HOW
CONTROLLING FOR FELONY CIRCUMSTANCES EXCEEDS THE B2 NOTION.
WHAT DID YOU, WHAT FELONY CIRCUMSTANCES DID YOU CONTROL
FOR, ADD IN, WOULD TAKE OUT AS THE CASE MAY BE, THAT WOULD NOT
HAVE BEEN ADDED IN OR TAKEN OUT IF YOU HAD BEEN USING THE
A——. p— — —
BALDUS - DIRECT
1 CRITERIA OF B27
2 THE WITNESS: WELL, THERE ARE FELONIES, YOUR HONOR,
3 THAT ARE NOT EMBRACED WITHIN THE DEFINITION OF A DEATH ELIGIBLE
4 CONTEMPORANEOUS OFFENSE. I WOULD BE GLAD TO IDENTIFY WHAT THOSE
3S ARE, AND THE THEN, I THINK THERE ARE FIVE FELONIES THAT MAKE ONE
» 6 DEATH ELIGIBLE UNDER BZ.
7 THE COURT: WHEN YOU SAY YOU CONTROLLED FOR FELONY
8 CIRCUMSTANCES, WHAT YOU MEAN IS YOU ADDED IN OR EXCLUDED ALL
9 WHERE THERE WAS ANY CONTEMPORANEOUS FELONY.
10 THE WITNESS: YES. SIR.
11 THE COURT: UNDER GEORGIA LAW.
12 THE WITNESS: THAT’S RIGHT.
13 THE COURT: AS A FELONY CONVICTION CANNOT BE USED
14 AGAINST A DEFENDANT, I THINK I KNOW THE ANSWER TO THAT QUESTION.
15 MR. BOGER: YOUR HONOR. IF THE --
16 THE COURT: IF THE FELONY RECORD WAS THE RECORD THAT WE
17 HEARD MR. GATES TESTIFY TO RECEIVING WHICH WAS A COMBINATION OF
18 THE DEFENDANT’S STATEMENT, THE F.B.I. RAP SHEET AND THE LOCAL
Ww 19 POLICE DEPARTMENT, THOSE WERE THE SOURCES. IDEALLY. ALL THREE
20 FOR THE FELONY RECORD INFORMATION.
21 THE WITNESS: YES, SIR.
22 THE COURT: AND IF I REMEMBER CORRECTLY, YOU DROPPED
23 THE COLUMN IN YOUR FIRST STUDY WHICH INQUIRED AS TO WHO KNEW
24 WHAT, WHEN. SO YOU HAVE NO IDEA ABOUT WHAT DECISION MAKERS KNEW
23 ABOUT THE FELONY BACKGROUND OF THE FOLK IN THIS CASE.
nN
OO
WH
0
NN
.
>
OB
BH
W
-
BALDUS - DIRECT
THE WITNESS: NO, I DO NOT HAVE DOCUMENTED KNOWLEDGE
WITH RESPECT TO EACH CASE AS TO WHO KNEW WHAT ABOUT EACH CASE.
BUT WHAT I DO KNOW, THIS EVIDENCE WAS IN THE RECORDS OF
THESE DEFENDANTS, AND ON THE BASIS OF MY KNOWLEDGE OF HOW
CRIMINAL JUSTICE SYSTEMS OPERATE, I BELIEVE THERES A FAIR BASIS
FOR ASSUMING THAT THIS IS THE SORT OF INFORMATION THAT CERTAINLY
IS BROUGHT TO THE ATTENTION OF PROSECUTORS. AND ALBEIT, IT’S
CORRECT THAT SUCH EVIDENCE IS NOT ADMISSIBLE ON A GUILT TRIAL
AGAINST A DEFENDANT, EXCEPT FOR MATTERS OF IMPEACHMENT. THAT
THIS 1S THE SORT OF INFORMATION THAT IS BROUGHT TO PENALTY
TRIAL, IS MY UNDERSTANDING OF THE LAW.
THE COURT: CAN BE.
THE WITNESS: CAN BE. YES.
THE COURT: ASSUMING IT’S AVAILABLE.
I WOULD SAY TO YOU, MR. BOGER, AS A PROBLEM TO
CONTEMPLATE, MY PERCEPTION OF THE WAY THE GEORGIA SYSTEM
OPERATES. I WOULD NOT COME ANYWHERE CLOSE TO JUDICIALLY NOTICING
THAT THAT INFORMATION WAS AVAILABLE EITHER AT THE TIME OF
INDICTMENT DECISION, PLEA BARGAINING DECISION, TRIAL OR
SENTENCING.
MR. BOGER: AVAILABLE TO WHO, YOUR HONOR, THE DECISION
MAKERS?
THE COURT: THE DECISION MAKERS, WHETHER IT BE
PROSECUTOR, JURY OR JUDGE.
FOR EXAMPLE, I HAVE HAD ANY NUMBER OF STATE JUDGES TELL
BALDUS - DIRECT
Pa
v ME THAT THEY HAVE NEVER SEEN A FRESENTENCE REPORT AND NEVER SEEN
2 A RAP SHEET. IN HUGE NUMBERS OF THEIR CASES. I AIN‘T GOING TO
3 JUDICIALLY NOTICE WHAT SOME GUY TOLD ME. BUT TO LET YOU KNOW
4 THAT THAT’S KIND OF MY VIEW OF THE WORLD, I THOUGHT I WOULD
3 MENTION IT.
% B MR. BOGER: ALL RIGHT, YOUR HONOR. I THINK WE WILL
7 HAVE EVIDENCE THAT PROSECUTORS MAY HAVE BETTER ACCESS TQ THAT
8 KIND OF INFORMATION INSOFAR AS THEIR DECISION MAKING IN THE
9 SYSTEM.
10 THE COURT: I DON’T DENY THEY HAVE THEORETICAL
11 ACCESS: HAVE, AT THE TIME THE DECISION MADE. IS A PROBLEM THAT I
12 WOULD NOT NOTICE BASED ON MY KNOWLEDGE OF HOW THE STATE CRIMINAL
13 JUSTICE SYSTEM OPERATES. I WOULD NOT MAKE THAT ASSUMPTION.
14 MR. BOGER: ALL RIGHT, I APPRECIATE THAT INFORMATION,
15 YOUR HONOR.
16 I KNOW FROM BITTER PERSONAL EXPERIENCES IN THE SUPREME
17 COURT THAT JURIES AT LEAST HAVE THE ABILITY TO HEAR THAT KIND OF
13 EVIDENCE. CERTAINLY. AS PROFESSOR BALDUS HAS STATED. IT“S NOT
» 19 PRESENT IN EVERY CASE. BUT IN THE ZANT VERSUS STEPHENS CASE.
20 WHICH CAME DOWN HEAVILY, SEVEN TO TWO AGAINST ME. THE SUPREME
21 COURT MADE IT CLEAR THAT NON-STATUTORY AGGRAVATING EVIDENCE. SO
22 CALLED, IS AT LEAST ADMISSIBLE IN THE STATE OF GEORGIA IN A
23 CAPITAL CASE. BUT =--
24 THE COURT: WELL, I STARTED, THE PENALTY PHASE IS WHAT
23 THROWS YOU A LOT OF TIMES IN THINKING ABOUT THIS, BECAUSE I
735
BALDUS - DIRECT
KNOW THEY WOULDN‘T BE ADMISSIBLE IN GUILT PHASE IN GEORGIA.
BECAUSE YOU CAN“T EVEN IMPEACH THE DEFENDANT WITH PRIOR RECORD
UNLESS THEY‘VE CHANGED THAT RULE LATELY. BUT I REMEMBERED IT
WAS ADMISSIBLE IN THE SENTENCING PHASE.
MR. BOGER: WE MADE AN ARGUMENT TO THE CONTRARY. AND
LOST THAT ARGUMENT IN THE HIGHEST PLACES.
"BUT I DO NOW RENEW DUR MOTION TO ADMIT DB-7.
THE COURT: ALL RIGHT, FOR THE PURPOSES I HAVE
PREVIOUSLY STATED. IT IS ADMITTED. FOR ALL OF THE REASONS THAT
1 HAVE PREVIOUSLY STATED THAT IT IS NOT ADMITTED FOR THE TRUTH
OF THE MATTERS ASSERTED THEREIN WITH THE ADDITIONAL CAVEAT THAT
IT HAS NOT DEMONSTRATED WHAT THE DECISION MAKERS KNEW.
BY MR. BOGER:?
@. PROFESSOR BALDUS. DID YOU DO A SIMILAR ANALYSIS OF THE
EFFECT oF THE PRESENCE OF FELONY CIRCUMSTANCES ON RACE OF
DEFENDANT DISPARITIES. AS OPPOSED TO RACE OF VICTIM DISPARITIES?
A. YES.
@. LET ME ASK You TO LOOK AT DB-72 MARKED FOR IDENTIFICATION.
CAN YOU IDENTIFY THAT DOCUMENT?
A. YE3. THIS IS TABLE ¢ FROM THE FINAL REPORT SUBMITTED IN
THIS CASE, AND IT MEASURES USING CROSS TABULATION METHOD, RACE
OF DEFENDANT DISPARITIES IN DEATH SENTENCING CASES, CONTROLLING
FOR THE PRESENCE OF FELONY CIRCUMSTANCES, AND THE VICTIM’S RACE.
AND IT SHOWS AMONG CASES INVOLVING FELONY CIRCUMSTANCES
A RACE OF DEFENDANT DIFFERENTIAL IN SENTENCING RATES OF FIVE
736
BALDUS ~ DIRECT
1 PERCENTAGE POINTS.
2 AND THAT IS. BLACK DEFENDANTS IN CASES INVOLVING FELONY
3 CIRCUMSTANCES HAVE A FIVE PERCENTAGE POINT HIGHER PROBABILITY OF
4 DEATH SENTENCE THAN DO WHITE DEFENDANTS.
9 THE COURT: THESE ARE THOSE EARLIER TABLES. STOOD ON
% & END, IF YOU WILL.
7 THE WITNESS: THAT IS RIGHT, THE PRECEDING TABLE. THE
8 READJUSTMENT OF THE CELLS, YOUR HONOR. THATS CORRECT.
? AND THE CASES INVOLVING A BLACK VICTIM, THE RATE IS
10 HIGHER FOR A WHITE DEFENDANT THAN IT IS FOR BLACK DEFENDANTS.
11 CASES NOT INVOLVING FELONY CIRCUMSTANCES. THERE IS JUST
12 A SLIGHT RACE OF DEFENDANT EFFECT IN WHITE VICTIM CASES.
13 AND ONE POINT DISPARITY IN BLACK VICTIM CASES.
14 THE COURT: NOW. NOTICING YOUR FOOTNOTE THERE, IT SAYS
135 IT’S SIGNIFICANT AT THE .S LEVEL, WHICH IS FAIRLY HIGH.
16 WOULD YOU COMMENT ON THE. THAT DEGREE OF UNRELIABILITY?
17 THE WITNESS: YES. WHEN WE APPLIED HERE. YOUR HONOR,
18 |AN OVERALL MEASURE, WHICH IS ANALOGOUS TO A REGRESSION MEASURE.
» 19 THE STATISTICAL SIGNIFICANCE OF THIS RACE OF DEFENDANT DISPARITY
20 DID NOT ATTAIN WHAT GENERALLY ARE ACCEPTED AS AN APPROPRIATE
21 LEVEL OF STATISTICAL SIGNIFICANCE IN WHICH ONE COULD HAVE
22 SUBSTANTIAL CONFIDENCE.
23 THE COURT: WELL, I GUESS ONE OF THE PROBLEMS I MEANT
24 TO RAISE TO YOU IN THAT QUESTION IS THAT HERE YOU RE DEALING
25 WITH THE SAME BODY OF DATA, THE SAME SAMPLE. THE SAME UNIVERSE.
BALDUS - DIRECT
CR WHAT HAVE YOU.
WHY IS THE SIGNIFICANCE LEVEL DIFFERENT WHEN YOUVE GOT
THE SAME GUANTITY OF THE THINGS YOU“RE MEASURING?
THE WITNESS: YOU MEAN IN CONTRAST. YOUR HONOR?
THE COURT: FOR EXAMPLE, MY RECOLLECTION WAS THAT IN
FIGURING, I DON’T KNOW WHETHER IT’S IN THE "P" FACTOR. BUT IN
DISCUSSING SIGNIFICANCE, THE GREATER THE NUMBER WAS TOWARD
INFINITY, THE LESS THE PROBABILITY IS THAT THE RESULT WAS BY
CHANCE. IS THAT FAIR?
IN OTHER WORDS. THE GREATER THE NUMBER, THE NUMBERS
YOU“RE DEALING WITH, THE LESS THE CHANCE IS THAT THE RESULT
YOU“RE OBSERVING IS A FUNCTION OF CHANCE.
THE WITNESS: IT’S THE SMALLER OF THE NUMBER.
THE COURT: THE SMALLER THE NUMBER YOU“RE OBSERVING.
THE GREATER THE CHANCE THAT THE RESULT IS INFLUENCED BY CHANCE.
THE WITNESS: IT’S THE OTHER WAY AROUND, YOUR HONOR. I
THINK -—-—
BY MR. BOGER:
@. PROFESSOR BALDUS. FIRST OF ALL, LET ME ASK YOU A QUESTION
JUST SAY TO CLARIFY.
THE NUMBER OF OBSERVATIONS. AS THE NUMBER OF
OBSERVATIONS INCREASES OR SAMPLE SIZE INCREASES, DOES THAT MAKE
FOR GREATER RELIANCE OR LESSER RELIANCE UPON THE STATISTICAL
SIGNIFICANCE OF THE QUTCOME?
IN OTHER WORDS, IF THE SIZE OF THE SAMPLE IS FIVE OF A
G3
H
M
oN
7358
BALDUS - DIRECT
THOUSAND ARE YOU MORE OR LESS CERTAIN ABOUT THE RESULTS THAN IF
A SAMPLE IS FIVE HUNDRED OF A THOUSAND?
A. IM NOT SURE I COMPLETELY UNDERSTAND THE QUESTION, COUNSEL.
I CAN SAY THAT THE LIKELIHOOD OF GETTING A
STATISTICALLY SIGNIFICANT RESULT. THAT IS, A "P" VALUE THAT IS
VERY SMALL. IS ENHANCED AS THE SAMPLE SIZE INCREASES.
THE COURT: THE NUMBER GETS BIGGER.
THE WITNESS: ANY DISPARITY WILL BECOME STATISTICALLY
SIGNIFICANT IF THE SAMPLE SIZE IS LARGE ENOUGH.
THE COURT: THAT’S KIND OF WHAT I WAS SAYING.
BUT THE PROBLEM I“M HAVING IS THAT I UNDERSTOOD THE
SAMPLE SIZE WAS THE SAME IN DB-71 1 AND DB-72, AND YET YOU“VE
GOT A DIFFERENT SIGNIFICANCE LEVEL. AND I DON’T IMMEDIATELY
UNDERSTAND WHAT YOU ARE ——
THE WITNESS: OH. THE REASON. YOUR HONOR, WE“RE
MEASURING A DIFFERENT DISPARITY. ONE IS RACE OF VICTIM
DISPARITY AND OTHER DISPARITY IS RACE OF DEFENDANT DISPARITY.
AND WHAT WE SEE IN DB-71 IS THE RACE OF VICTIM DISPARITY IN THE
FOOTNOTE, TEST OF SIGNIFICANCE FOR IT IN THE FOOTNOTE.
AND IN THE DB-72, WE SEE A MEASURE OF STATISTICAL
SIGNIFICANCE FOR THE RACE OF DEFENDANT DISPARITY.
THE COURT: IS THAT BECAUSE THERE ARE RELATIVELY FEW
CAPITAL PUNISHMENTS IMPOSED FOR THE KILLING OF A BLACK VICTIM?
IS THAT THE SIMPLE ANSWER?
THE WITNESS: NO, I DON’T BELIEVE THATS THE CASE, YOUR
“
f
e
d
e
o
N
a
©
10
24
23
759
BALDUS - DIRECT
HONOR, BECAUSE HERE WERE FOCUSING IN ON THE WHITE VICTIM CASES
AND THE DIFFERENTIAL IN TREATMENT OF BLACK AND WHITE. WHITE
DEFENDANTS WITHIN THE CATEGORY OF WHITE VICTIM CASES. WHICH
COMPRISE THE BIGGEST POPULATION OF CASES IN WHICH DEATH SENTENCE
IS IMPOSED.
THE REASON FOR IT. YOUR HONOR. IS THAT THE DIFFERENCE
IN THE RACE, WITHIN THAT. WITHIN EACH OF THESE SUB-GROUPS, THE
DIFFERENCE IN THE DEATH SENTENCING RATES BETWEEN THE WHITE AND
BLACK DEFENDANTS IS SMALL. THAT’S WHY WE DON’T GET STATISTICAL
SIGNIFICANCE WHEN WE LOOK AT THE OVERALL EFFECTS.
THE COURT: BECAUSE THE SPREAD IS SMALL.
THE WITNESS: THAT’S RIGHT. THAT’S THE PRINCIPAL
THING. YOU SEE. THERE ARE TWO PRINCIPAL THINGS THAT INFLUENCE
THE LEVEL OF STATISTICAL SIGNIFICANCE. WOULD BE THE MAGNITUDE OF
THE DIFFERENCE IN THE RATES AND THE SAMPLE SIZE OVER THE TWO ——
THE COURT: IN OTHER WORDS, IT’S KIND OF THE OLD NOTION
OF THE STANDARD DEVIATION, CERTAIN AMOUNT OF DEVIATION IS
STATISTICALLY ACCEPTED, BUT IF YOU GET OUTSIDE OF THAT, IF 1
REMEMBER CORRECTLY. STANDARD DEVIATION, THEN IT BECOMES MORE
SIGNIFICANT BECAUSE IT IS LESS PREDICTABLE IF THAT DEVIATION
WOULD OCCUR --—
THE WITNESS: BY CHANCE.
THE COURT: -- BY CHANCE.
THE WITNESS: YES.
THE COURT: OKAY. ALL RIGHT.
BE
eee. . feet” —. ——_ f— P— ————
7&0
BALDUS - DIRECT
1 BY MR. BOGER:
2 @. S0 PROFESSOR BALDUS, IT“S YOUR TESTIMONY THAT THE RACE OF
3 DEFENDANT DISPARITIES. WHEN CONTROLLING FOR FELONY CIRCUMSTANCE
4 AND RACE OF THE VICTIM. ARE LESS STATISTICALLY SIGNIFICANT BUT
9 STILL ALSO OBSERVABLE AT SOME POINT IN THIS CROSS TABULATION, IS
% b THAT CORRECT?
7 A. YES.
8 MR. BOGER: YOUR HONOR, I MOVE THE ADMISSION OF DB-72
9? IN EVIDENCE.
10 THE COURT: FOR THE PURPOSES PREVIOUSLY STATED. IT WILL
11 BE ADMITTED.
12 MR. BOGER: THANK YOU, YOUR HONOR.
13 BY MR. BOGER:
14 @. PROFESSOR BALDUS., HAVE YOU CONDUCTED AN ANALYSIS THAT WOULD
15 MEASURE THE SAME FACTORS USING ANY OTHER FORM OF MEASUREMENT?
146 A. YES. WE CONDUCTED A SERIES OF MULTIPLE REGRESSION ANALYSES.
17 @. LET ME ASK YOU TO TURN TO WHAT’S BEEN MARKED AS DB-73.
13 CAN YOU IDENTIFY THAT DOCUMENT?
R 8 19 A. YES. DB-73 IS A MULTIPLE REGRESSION ANALYSIS WHICH
20 ESTIMATES THE RACE OF VICTIM EFFECTS CONTROLLING FOR RACE OF
21 DEFENDANT. AND THE PRESENCE OR ABSENCE OF FELONY
22 CIRCUMSTANCES IN A CASE, AS WELL AS A RACE OF DEFENDANT EFFECT.
23 AFTER CONTROLLING FOR THE RACE OF THE VICTIM AND ALSO FELONY
24 CIRCUMSTANCES.
23 THIS IS TABLE 10 FROM THE REPORT.
BALDUS - DIRECT
@. WHAT DDES THE TABLE REFLECT? P
y
A. IT REFLECTS STRONG, A STRONG RACE OF VICTIM EFFECT. THE,
THE LEAST SQUARES REGRESSION CO-EFFICIENT IS .09, AND IT”S
HIGHLY SIGNIFICANT FROM A STATISTICAL STANDPOINT. ITS
IMPORTANT HERE TO COMPARE IT WITH WHAT THE RACE OF VICTIM EFFECT
WAS BEFORE WE INTRODUCED THE CONTROL FOR FELONY CIRCUMSTANCES.
IF YOU RECALL, IT WAS 17 PERCENTAGE POINTS. IT WAS
.17. THE INTRODUCTION OF THIS SINGLE CONTROL FOR THE FELONY
CIRCUMSTANCES BACKGROUND FACTOR REDUCES THE LEAST SQUARES
po
t e
e
e
Sa
S
E
GR
E
c
BE
R
CO
A
E
CO-EFFICIENT FROM .17 TO .09.
33 AND SIMILARLY WITH RESPECT TO THE LOGISTIC REGRESSION
12 ANALYSIS, WHICH IS REPORTED IN COLUMNS C AND D, OF DB-73, WE SEE
13 THERE AS WELL A SUBSTANTIAL AND STATISTICALLY SIGNIFICANT
14 CO-EFFICIENT.
15 HERE AGAIN, THE UNADJUSTED RACE OF VICTIM DEATH ODDS
16 MULTIPLIER WAS IN THE ORDER OF 13 BEFORE WE INTRODUCED THIS
37 BACKGROUND CONTROL FOR FELONY CIRCUMSTANCES. HERE WE SEE THE
18 FELONY CIRCUMSTANCES VARIABLE REDUCES CONSIDERABLY THE MAGNITUDE
» 19 OF THE RACE OF VICTIM EFFECT, BUT IT IS STILL APPARENT THAT A
20 PERSON, EVEN CONTROLLING FOR THIS BACKGROUND FACTOR, IF HIS
21 VICTIM IS WHITE, HAS SIGNIFICANTLY HIGHER ODDS OF RECEIVING A
22 DEATH SENTENCE THAN A DEFENDANT WHOSE VICTIM IS WHITE.
23 THE REASON FOR THIS SHARP DROP IN THE MAGNITUDE OF
24 THESE TWO RACE OF VICTIM CO-EFFICIENTS IS THAT THE FELONY
CIRCUMSTANCES VARIABLE IS HIGHLY CORRELATED WITH BOTH THE DEATH M 41}
BALDUS - DIRECT
1 SENTENCING RESULTS AND WITH RACE OF THE VICTIM.
2 AND IT’S THAT PROPERTY OF A BACKGROUND CONTROL VARIABLE
THAT ENABLES IT TO REDUCE THE EFFECTS OF THE RACE OF THE VICTIM
VARIABLE. IF THE VARIABLE WERE CORRELATED WITH THE DEATH
SENTENCING RESULT. BUT IT WAS NOT ALSO CORRELATED WITH THE RACE
3
3
S
& OF VICTIM. THEN A BACKGROUND CONTROL FOR THAT WOULD TEND NOT TO
7 SUPPRESS THE MAGNITUDE OF THE RACE OF VICTIM CO-EFFICIENT.
8 AND SIMILARLY, IF A VARIABLE WERE CORRELATED WITH THE
9 RACE OF VICTIM, BUT NOT ALSO CORRELATED WITH THE DEATH
10 SENTENCING OUTCOME, IT ALSO WOULD TEND NOT TO SUPPRESS THE
11 MAGNITUDE OF THE RACE OF VICTIM CO-EFFICIENT.
12 THE FACT THAT THE RACE OF THE VICTIM, PARDON ME. THAT
13 THE FELONY CIRCUMSTANCES VARIABLE IS CORRELATED WITH BOTH THE
14 OUTCOME VARIABLE AND WITH THE RACE OF THE VICTIM WHICH TENDS TO
13 SUPPRESS THE MAGNITUDE OF THE RACE OF VICTIM EFFECT WHEN THAT IS
16 INTRODUCED AS A BACKGROUND CONTROL.
17 NOW YOU WILL NOTE HERE IN RESPECT TO THE RACE OF THE
138 DEFENDANT EFFECT, THAT THE ORIGINAL MAGNITUDE OF THAT
®» 1¥ CO-EFFICIENT WHEN THERE WAS ONLY A CONTROL FOR RACE OF VICTIM
20 WITH .10, IT“S REDUCED BY FIFTY PERCENT TO .03 WHEN THE SINGLE
21 BACKGROLIND FACTOR I3 INTRODUCED.
22 AND INDEED, IN THE LOGISTIC REGRESSION CO-EFFICIENT.,
23 YOU) NOTE THAT IT IS. IT HAS ALSO DROPPED AND IT. THERE IS
24 REPORTED THERE A LEVEL OF STATISTICAL SIGNIFICANCE FOR THAT OF
23 ONLY .29. THIS, THIS RESULT HERE, YOUR HONOR, REFLECTS A
763
BALDUS - DIRECT
1 PATTERN WE SEE THROUGHOUT ALL THESE RESULTS. THAT THE RACE OF
2 VICTIM CO-EFFICIENTS, THE RACE OF VICTIM EFFECTS ARE STABLE
3 ACROSS THE DIFFERENT METHODS OF ANALYSIS. AND WE‘LL SEE THEY’RE
STABLE WITH DIFFERENT VARIETIES OF BACKGROUND FACTORS CONTROLLED
FOR.
THE RACE OF DEFENDANT EFFECTS, HOWEVER. ARE DIFFERENT.
4
S
6
7 THEY“RE LESS STABLE AND THEY VARY AND BOUNCE AROUND FROM ONE
8 ANALYSIS TO ANOTHER.
9 THE COURT: I NEED FOR YOU TO TELL ME. WHILE YOURE AT
oO THIS POINT, HOW TO READ THIS. IT SAYS "FELONY CIRCUMSTANCES?"
11 NOW, TELL ME IN GOOD LAYMAN LANGUAGE WHAT TO DO WITH
12 THAT .13.
13 THE WITNESS: RIGHT. THAT .13 MEANS, YOUR HONOR, WHEN
14 YOU CONTROL FOR THE BACKGROUND FACTORS OF RACE OF VICTIM AND
15 RACE OF DEFENDANT, THAT A DEFENDANT WHOSE HOMICIDE INVOLVES A
156 FELONY CIRCUMSTANCE HAS A 15 PERCENTAGE POINTS HIGHER LIKELIHOOD
17 OF RECEIVING A DEATH SENTENCE THAN SOMEONE WHOSE CASE DID NOT
18 INVOLVE FELONY CIRCUMSTANCES.
® 12 AND THE LOGISTIC ANALYSIS ON THE RIGHT SIDE. YOUR
20 HONOR, THE KEY STATISTIC THERE IS WHAT WE CALL THE DEATH ODDS
21 MULTIPLIER. THAT INDICATES HOW MUCH THE PRESENCE OF A FELONY
<2 CIRCUMSTANCE IN THE CASE. ON THE AVERAGE, INCREASES THE
23 DEFENDANTS ODDS OF RECEIVING A DEATH SENTENCE. THOSE ARE TWO —-—
24 THE COURT: THAT DOESN‘T REALLY SHOW ME ANYTHING RACIAL - Y
23 THE WITNESS: THAT’S RIGHT.
rent, meee, pete Gg eet. bet ee pe ts eter. meester’. “oem etree. eet. “Je Se —— ————. —— p— ————— Ailey, . ft
764
BALDUS — DIRECT
1 THE COURT: -- EXCEPT THAT YOU HELD THEM CONSTANT.
2 ALL RIGHT. LET”S GO DOWN TO THE RACE OF VICTIM. .09
MEANS WHAT?
THE WITNESS: THAT MEANS THAT AFTER ONE CONTROLS FOR
THE FELONY CIRCUMSTANCES VARIABLE AND THE RACE OF DEFENDANT
3
4
S
ha. 6 VARIABLE, THAT IS. AFTER THOSE TWO FACTORS ARE HELD CONSTANT
7 THROUGH THE MULTIPLE REGRESSION PROCEDURE. THAT ON THE AVERAGE.
8 DEFENDANTS WHOSE VICTIMS ARE WHITE HAVE A 9 PERCENTAGE POINTS
@ HIGHER CHANCE OF RECEIVING A DEATH SENTENCE THAN DEFENDANTS
10 WHOSE VICTIMS BLACK.
11 THE COURT: OKAY. THAT'S WHAT I THOUGHT. BUT I WANTED
12 TO MAKE SURE I UNDERSTOOD YOU.
13 AND WHERE YOU CONTROL FOR RACE OF VICTIM AND FOR FELONY
14 CIRCUMSTANCES, A BLACK DEFENDANT HAS GOT A FIVE PERCENT GREATER
15 CHANCE THAN A WHITE DEFENDANT OF RECEIVING THE DEATH PENALTY.
146 THE WITNESS: WELL, IT“S TECHNICALLY NOT A FIVE
17 PERCENT, YOUR HONOR, BECAUSE THAT SUGGESTS THERE’S SOME OTHER
18 NUMBER THAT YOU’RE MAKING A CALCULATION AGAINST. IT’S THE, THE
4 19 AVERAGE RATE IS FIVE PERCENTAGE POINTS HIGHER FOR THE BLACK
20 DEFENDANT THAN A WHITE DEFENDANT.
21 THE COURT: BLACK DEFENDANT IS FIVE PERCENT GREATER
22 THAN ALL BLACK AND WHITE DEFENDANTS TOGETHER?
23 THE WITNESS: NO, THE -—
24 BY MR. BOGER:
25 @. LET ME ASK YOU A QUESTION, PROFESSOR BALDUS, THAT MIGHT
“
Ov
{8
8
OW
8
9
10
765
BALDUS - DIRECT
CLARIFY IT.
IF ALL DEFENDANTS HAD A CONTROLLING FOR FELONY
CIRCUMSTANCES. AND RACE OF VICTIM, HAD A .27 LIKELIHOOD OF A
DEATH SENTENCE, WHAT WOULD A BLACK DEFENDANT HAVE UNDER THOSE
CIRCUMSTANCES?
A. WELL, IT WOULD BE FIVE PERCENTAGE POINTS HIGHER.
@. WOULD IT BE .327
Pe YES, THAT WOULD BE MY UNDERSTANDING OF IT.
I THINK THE POINT, YOUR HONOR, WHEN YOU SPEAK ABOUT A
PERCENTAGE INCREASE, WE'RE USUALLY TALKING ABOUT A GIVEN
PROBABILITY TO BEGIN WITH, AND WHAT WE“RE REALLY TALKING ABOUT
IS A FIVE PERCENTAGE POINT INCREASE ON THE AVERAGE. AND A FIVE
PERCENTAGE POINT INCREASE ON THE AVERAGE MIGHT REPRESENT A FIFTY
PERCENT INCREASE IF THE UNDERLYING VALUE WERE .10 AND A MUCH
SMALLER INCREASE IF IT WERE MUCH HIGHER. THAT’S., IT’S A
TECHNICAL POINT, BUT I THINK IT’S —-—
THE COURT: IT“S NOT A FIVE PERCENT INCREASE, IT’S A
FIVE PERCENTAGE POINT INCREASE IN THE PROBABILITY.
THE WITNESS: YES, I THINK THAT’S THE POINT, EXACTLY.
THE COURT: OKAY.
MR. BOGER: YOUR HONOR, I MOVE THE ADMISSION OF DB-73 IN
EVIDENCE.
THE COURT: ALL RIGHT. I WILL ADMIT IT FOR THE PURPOSES
PREVIOUSLY STATED AND WITH THE RESTRICTIONS PREVIOUSLY STATED
UNLESS YOU HAVE ANY.
i ee. fen... Set it. reat en. ed, i btr— ——_ it rb a. teen’ Seer.
BALDUS -~ DIRECT
or
y MR. BOGER2 NO. YOUR HONOR.
nN
THE COURT: IM TALKING TO THE STATE.
MS. WESTMORELAND: NO, YOUR HONOR, SAME OBJECTION. 0)
THE COURT: ALL RIGHT. ITLL BE ADMITTED FOR THOSE
PURPOSES.
BY MR. BOGER:
@. PROFESSOR BALDUS., HAVE YOU CONTROLLED FOR OTHER IMPORTANT.
BACKGROUND VARIABLES BESIDES FELONY MURDER CIRCUMSTANCE IN RACE
3
0
0
th
A
OF VICTIM CASES?
10 A. YES. AS I SUGGESTED EARLIER, WE TOOK THE FIVE MOST. NOT THE
11 MOST IMPORTANT, FIVE BACKGROUND FACTORS THAT APPEARED TO BE
12 IMPORTANT FACTORS IN THE SYSTEM, AND THEY WERE THE PRESENCE OF
13 FELONY CIRCUMSTANCES. THE PRESENCE OF SERIOUS PRIOR RECORD,
14 PRESENCE OF A FAMILY-LOVER-L IQUOR-BAR ROOM QUARREL: MULTIPLE
13 VICTIMS IN A CASE AND PRESENCE OF A STRANGER VICTIM. AND
16 CONDUCTED AN ANALYSIS COMPARABLE TO THE ONE I JUST WENT THROUGH
17 ON FELONY CIRCUMSTANCES FOR EACH ONE OF THOSE FACTORS, TO SEE
13 THE EXTENT TO WHICH ONE OF THOSE FACTORS. WHEN ADJUSTED FOR,
» 19 WOULD TEND TO SUPPRESS THE RACE OF DEFENDANT AND RACE OF VICTIM
2Q EFFECTS.
21 2. DID ANY OF FIVE SUPPRESS THOSE EFFECTS ENTIRELY?
22 A. NO.
23 @. DOES YOUR REPORT CONTAIN THE RESULTS OF ALL OF THOSE
24 ANALYSES THAT YOU‘ VE CONDUCTED?
23 A. YES.
Og
0
Y
O
N
(W
y oO
11
12
767
BALDUS ~ DIRECT
@. LET ME DIRECT YOUR ATTENTION TO DB-74 MARKED FOR
IDENTIFICATION, AND ASK IF YOU CAN IDENTIFY THAT DOCUMENT?
A. YES. THIS IS THE RESULT OF THE ANALYSIS OF THE SERIOUS
PRIOR RECORD VARIABLE. AND IT IS TABLE 13 FROM THE REPORT. AND
IT SHOWS THAT BACKGROUND CONTROL FOR THAT FACTOR HAS NO EFFECT
WHATEVER ON THE MAGNITUDE OF THE RACE OF VICTIM, AND RACE OF
DEFENDANT EFFECT. THEY STILL ARE AT THE SAME MAGNITUDE HERE AS
THEY WERE WITHOUT THE INTRODUCTION OF ANY BACKGROUND CONTROL FOR
NON-RACIAL FACTORS.
THAT. THE INTRODUCTION OF THIS BACKGROUND FACTOR STANDS
IN SHARP CONTRAST TO THE INTRODUCTION OF THE BACKGROUND FACTOR
FOR FELONY CIRCUMSTANCES WHICH HAD A SIGNIFICANT DEPRESSIVE
EFFECT ON THE MAGNITUDE OF THE RACIAL CO-EFFICIENTS.
@. BUT WHEN ONE CONTROLS SIMULTANEOUSLY FOR THE RACE OF
DEFENDANT, AND RACE OF VICTIM. AND SERIOUS PRIOR RECORD, WHICH.
WHICH FACTOR IS MORE TELLING OR IMPORTANT IN PREDICTING A DEATH
SENTENCE OUTCOME. RACE OF VICTIM OR SERIQUS PRIOR RECORD AS
REFLECTED IN THIS CHART?
A. WELL, THESE CO-EFFICIENTS THAT ARE REPORTED HERE GIVE ONE A
BASIS FOR APPROXIMATING AND IT IS ONLY AN APPROXIMATION OF THE
RELATIVE IMPORTANCE OF THESE FACTORS IN DETERMINING THE
LIKELIHOOD OF A DEATH SENTENCE. BUT IT-S PLAIN AT LEAST WHEN
ONLY THESE FACTORS ARE TAKEN INTO THE ACCOUNTS THAT THE TWO
RACIAL FACTORS APPEAR TO HAVE A MORE SUBSTANTIAL EFFECT THAN THE
PRIOR RECORD VARIABLE.
768
BALDUS - DIRECT
fo
e MR. os]
0 & m z YOUR HONOR, I MOVE THE ADMISSION OF DB-74
EVIDENCE.
THE COURT: SAME SITUATION?
MS. WESTMORELAND: SAME SITUATION, YOUR HONOR.
THE COURT: IT WILL BE ADMITTED AS DEMONSTRATIVE OF
SUMMARY EVIDENCE.
BY MR. BOGER:
@. PROFESSOR BALDUS, SOME COURTS HAVE INDICATED INTEREST OF
N
0
0
NM
0
UB
H
N
SOME SORT IN WHETHER RACIAL FACTORS AND THEIR IMPACT ON
10 SENTENCING OUTCOME MIGHT BE EXPLAINED BY THE PRESENCE OR ABSENCE
11 OF GEORGIA STATUTORY AGGRAVATING CIRCUMSTANCES.
312 HAVE YOU AT ANY TIME DONE AN ANALYSIS CONTROLLING FOR
13 THE STATUTORY AGGRAVATING CIRCUMSTANCES UNDER THE GEORGIA
14 SYSTEM?
135 A. YES. WE HAVE CONDUCTED ANALYSES IN WHICH WE ESTIMATE RACIAL
14 EFFECTS FOR BOTH THE DEFENDANT AND THE VICTIM. CONTROLLING FOR
17 EACH STATUTORY AGGRAVATING FACTOR SEPARATELY. WE“VE USED BOTH
18 CROSS TABULATION METHODS TO CONDUCT THOSE ANALYSES AND MULTIPLE
& 19 REGRESSION METHODS.
20 @. LET ME DIRECT YOUR ATTENTION TO DB-75 MARKED FOR
21 IDENTIFICATION. ILL ASK YOU TO IDENTIFY THAT DOCUMENT. IF YOU
<2 CAN?
23 A. YES. THIS TABLE, WHICH IS NUMBER 18 IN OUR REPORT,
24 ESTIMATES RACE OF VICTIM AND RACE OF DEFENDANT DISPARITIES IN
23 THE CHARGING AND SENTENCING STUDY AMONG GROUPS OF CASES WHICH
T
E
I
E
R
T
E
ry
oO
Te
y
po
d
769
BALDUS - DIRECT
OUR ANALYSIS IDENTIFIED AS HAVING PRESENT IN THEM THE
CHARACTERISTICS THAT WOULD MAKE THEM DEATH ELIGIBLE UNDER
GEORGIA’S DEATH SENTENCING STATUTE.
TO ILLUSTRATE, IF ONE LOOKS IN THE FIRST ROW, THE
STATUTORY AGGRAVATING FACTOR Bl, IS IDENTIFIED.
IN THE SAMPLE, THERE WERE ONE HUNDRED -- PARDON ME --IN
THE UNIVERSE, THERE WERE 104 CASES THAT WERE DEATH ELIGIBLE
UNDER THIS STATUTORY AGGRAVATING FACTOR ACCORDING TO OUR
ESTIMATE. BASED ON THE SAMPLE.
AND WE TOOK THOSE CASES. ISOLATED THEM. AND ESTIMATED A
RACE OF VICTIM AND RACE OF DEFENDANT CO-EFFICIENT WHICH
EXPLAINED THE IMPACT OF THOSE TWO VARIABLES ON THE DEATH
SENTENCING OUTCOMES AMONG THAT SUB-GROUP OF CASES SO THAT YOU
CAN SEE THAT AMONG THAT SUB-GROUP OF CASES THAT DEFENDANTS WITH
A WHITE VICTIM HAD A. ON AVERAGE, HAD A FORTY-FIVE PERCENTAGE
POINT HIGHER LIKELIHOOD OF A DEATH SENTENCE THAN BLACK VICTIM
CASES. AND THAT THE BLACK DEFENDANTS AMONG THAT POPULATION HAD
A 14 PERCENTAGE POINT HIGHER LIKELIHOOD OF A DEATH SENTENCE.
THIS ANALYSIS DID NOT CONTROL FOR ANY OTHER FACTORS, OTHER THAN
THE PRESENCE OF THE SINGLE STATUTORY AGGRAVATING FACTOR.
@. JUST FOR THE RECORD. WHAT COLUMNS WERE YOU LOOKING IN TO
DERIVE THOSE FIGURES?
A. YES, I WAS LOOKING IN ROW Bi, D AND E, THE TWO RACIAL
CO-EFFICIENTS TO WHICH I JUST TESTIFIED.
AND ONE CAN SEE FROM EXAMINING COLUMNS D AND E AND ALSO i
|
|
770
BALDUS - DIRECT
1 COLUMNS H AND I THAT. PARDON ME, COLUMNS, COMPARING COLUMN D
2 WITH COLUMN G, YOU SEE WITH THE TWO MEASURES THE MAGNITUDE OF
3 THE RACE OF VICTIM EFFECTS.
4 @. DO THE RACE OF VICTIM EFFECTS VARY DEPENDING ON THE
3 STATUTORY AGGRAVATING CIRCUMSTANCE HELD A CONSTANT?
b A. YES, THEY DO VARY SUBSTANTIALLY. THEY“RE MUCH HIGHER IN
7 SOME CATEGORIES THAN IN OTHERS.
8 IN BS. STATEWIDE, WE SEE NO RACE OF VICTIM EFFECT NOR
4 DO WE SEE ANY IN B&, WHILE IN OTHERS WE SEE SIGNIFICANT RACE OF
10 VICTIM EFFECTS.
11 BZ, B7. IT’S APPROXIMATELY 20 POINTS. AND THOSE. THOSE
12 FIGURES ARE ALSO REFLECTED IN THE REGRESSION RESULTS ESTIMATED
13 WITH A LOGISTIC REGRESSION PROCEDURE. WHICH ARE REPORTED IN
14 COLUMNS F AND G OF DB-73.
13 ONE NOTES. HOWEVER, THAT WHEN YOU LOOK AT THE RACE OF
146 DEFENDANT EFFECTS ESTIMATED WITH A LOGISTIC REGRESSION ANALYSIS,
27 THAT THE RESULTS DO NOT SHOW THE SAME SYSTEMATIC PATTERN. THEY
18 WILL. THE ONLY SUB~-CATEGORY OF CASES WHERE YOU SEE STATISTICALLY
» 19 SIGNIFI CANT EFFECTS ARE UNDER THE B3 STATUTORY AGGRAVATING
20 FACTOR. AND ALSO UNDER THE B7 FACTOR. UNDER THE B7 FACTOR. WE
21 SEE EXTREMELY STRONG RACE OF DEFENDANT EFFECTS. AFTER
22 CONTROLLING FOR RACE OF THE VICTIM.
23 BUT AGAIN, THE SAME PATTERN EMERGES THAT THE RACE OF
24 DEFENDANT EFFECTS ARE NOT SYSTEMATIC TO THE SAME DEGREE THAT THE
25 RACE OF VICTIM EFFECTS ARE.
M
i
B
N
P
N
-
oO
Po
y
P
y
BALDUS - DIRECT
@. DOES THIS PROVIDE FURTHER SUPPORT FOR YOUR CONCLUSIONS ABOUT
THE OVERALL IMPACT OF RACIAL DISPARITIES THAT YOU TESTIFIED TO
THIS MORNING?
A. YES. THIS ANALYSIS ADDRESSES A SERIES OF RIVAL HYPOTHESES OR
TESTS A SERIES OF RIVAL HYPOTHESES.
AND THE FIRST ONE WOULD BE THAT IF YOU CONTROL FOR THE
STATUTORY AGGRAVATING FACTOR SEPARATELY THAT YOU WILL EXPLAIN
AWAY THE RACIAL EFFECTS.
YOU DO TO AN EXTENT AS WE“VE SEEN WITH RESPECT TO THE
SOME OF THE ANALYSES. BUT BY AND LARGE. PARTICULARLY WITH
RESPECT TO THE RACE OF VICTIM DISPARITIES. THEY PERSIST.
MR. BOGER: YOUR HONOR, WE MOVE THE ADMISSION OF DB-735
INTO EVIDENCE AND I MIGHT ADD THAT WHILE AS YOU KNOW THAT WE HAVE
UNSUCCESSFULLY CONTENDED THAT THE PRIOR FIGURES AND TABLES
SHOULD COME IN FOR ALL PURPOSES» AS TO THIS, WERE AWARE OF YOUR
HONCOR“S RULING. AND HAVE ABIDED BY IT.
BUT AS TO THIS TABLE, IT SEEMS TO ME WE NOW ENTER THE
AREA THAT’S SPECIFICALLY REFERRED TO BY THE FIFTH CIRCUIT IN THE
SMITH CASE. IN WHICH THERE IS CONTROL BEGINNING TO BE MADE FOR
STATUTORY AGGRAVATING CIRCUMSTANCES. THE CONTROLS HERE, ONE AT
A TIME, BUT THEY NEVERTHELESS ADDRESS THE STATUTORY AGGRAVATING
ISSUE, AND THEREFORE WOULD CONTEND THAT THIS IS AN ADDITIONAL
BASIS ON WHICH THIS DOCUMENT SHOULD BE ADMITTED FOR ALL
PURPOSES.
THE COURT: I DON’T SEE ANY REASON TO BELIEVE THAT
ae peastmasboet impo, go m— ASRS Sonn Yiohabiiue saseeibam oy S——— <— — —————artt ———— i —— Se. A ——— SO ————. Sr r=, SM VS H— SS» WT S—_——— ts sapere, eee. Soe fre —— ——
772
BALDUS - DIRECT
be
THESE UNADJUSTED -- IF I UNDERSTAND WHAT HES DONE. IS NOT
CONTROL FOR ALL OF THEM, BUT CONTROL ONE AT A TIME. IS THAT
RIGHT?
MR. BOGER: THAT’S CORRECT. YOUR HONOR.
1 MIGHT ADD THAT FOOTNOTE 33 OF SMITH VERSUS BALKCOM
IT WAS NOT CLEAR ON THE QUESTION OF WHETHER ONE HAS TO CONTROL
FOR ALL SIMULTANEOUSLY OR FOR ONE AT A TIME. WE CERTAINLY
SUBMIT THAT THIS IS PARTIAL BASIS FOR PROFESSOR BALDUS’ JUDGMENT
Ug
08
.
N
O
C
d
N
AND THAT IT IS EVIDENCE THAT’ S ADMISSIBLE AND RELEVANT, WHETHER
pb
o ITS SUFFICIENT UNTO ITSELF TO SUPPORT HIS CONCLUSION.
Pe
ry
Jo
t THE COURT: STATISTICAL EVIDENCE. CIRCUMSTANTIAL
12 EVIDENCE, SUPPORT HYPOTHESES. PHYSICAL EVIDENCE, I GUESS. IN
13 DISCRIMINATION CASES SAYS THERE IS NO OTHER REASONABLE
14 HYPOTHESIS THAT EXPLAINS THE CONDUCT BUT AN INTENT TO
15 DISCRIMINATE.
16 AM 1 ESSENTIALLY RIGHT SO FAR?
17 MR. BOGER: THAT’S CERTAINLY WHAT THE PREVAILING LAW IN
13 SOME CIRCUITS IS. INCLUDING THIS ONE.
pe 19 : THE COURT: WELL, I WASN’T THINKING ABOUT DEATH PENALTY
20 CASES ALONE, BUT THE WHOLE CIVIL RIGHTS AREA.
21 IM NOT LOOKING AT IT AS A QUESTION OF LAW. I'M
22 LOOKING AT IT AS A QUESTION OF LOGIC. I DON’T THINK WE ARE PAST
23 BOWERS~PIERCE AT THIS POINT, BECAUSE WE ALL KNOW, EVERYBODY IN
24 THIS COURTROOM. I DON’T KNOW THE GENTLEMAN SEATED BACK THERE,
23 BUT EVERYBODY ELSE IN THE COURTROOM KNOWS THAT THERE ARE JUST
773
BALDUS - DIRECT
ONE HELL OF A LOT OF PERMISSIBLE FACTORS THAT WOULD INFLUENCE
THE RESULTS BEYOND THAT WHICH IS CONTROLLED FOR HERE. IF I
UNDERSTAND THIS CHART. AND UNTIL YOU HAVE SUCCEEDED IN
CONTROLLING FOR THOSE WHICH MIGHT BE OTHER LEGITIMATE REASONS,
YOU HAVEN'T GOTTEN OVER THE THRESHOLD OF RELEVANCY WITH ME AND I
DON’T BELIEVE WITH THE CIRCUIT, AND I DOUBT WITH THE SUPREME
COURT.
MR. BOGER: WELL, WE WILL ABIDE BY YOUR HONOR’S RULING.
I THINK WE“VE ARGUED THAT THIS IS CLEARLY RELEVANT, WHETHER IT’S
SUFFICIENT OR NOT, BUT WE WILL ABIDE BY YOUR HONORS RULING. OF
COURSE.
I THINK WE CAN SAY THAT WE DO HAVE CHARTS DOWN THE
ROAD, NOT MUCH FURTHER DOWN THE ROAD, THAT WILL CONTROL FOR
QUITE A NUMBER OF FACTORS. SO THIS IS NOT THE END OF THE
HEARING WITH THIS RULING, EITHER.
BUT WILL YOUR HONOR ADMIT THIS DOCUMENT ON THE BASIS OF
THE PREVIOUSLY ——
THE COURT: SURELY.
MR. BOGER: THANK YOU, YOUR HONOR.
BY MR. BOGER:
@. PROFESSOR BALDUS, HERE WE HAVE BEEN CONTROLLING FOR
BACKGROUND FACTORS ONE BY ONE. YOUR FELONY CIRCUMSTANCE
BACKGROUND FACTOR, YOUR PRIOR RECORD BACKGROUND FACTOR. AND NOW
YOUR STATUTORY AGGRAVATING CIRCUMSTANCES.
IS IT POSSIBLE TO EMPLOY THE MULTIPLE REGRESSION METHOD
— ——— —— S——" —— —— fb ——————A iad Seta. mein . Weta S——— , ——. Fm——————— | RY. WATT—_—— | W—_————— T—r——. fr — S——— ey S———— —————"
774
BALDUS - DIRECT
Fw
y YOU TESTIFIED TO EARLIER TO CONTROL FOR MORE THAN ONE BACKGROUND
FACTOR AT THE TIME?
A. YES. IT’S POSSIBLE TO USE MULTIPLE REGRESSION AND CROSS
TABULATION METHODS TO CONTROL FOR ADDITIONAL BACKGROUND FACTORS.
@. AND HAVE YOU DONE SO IN YOUR ANALYSIS?
A. YES, WE HAVE USED BOTH APPROACHES.
@. LET ME DIRECT YOUR ATTENTION TO WHAT’S MARKED FOR
IDENTIFICATION AS DB-7&6 AND ASK YOU TO IDENTIFY THAT DOCUMENT?
W
N
0m
MN
6
R
P
h
B
N
A. YES. DB-76 IS FIGURE 2 FROM OUR REPORT. AND IT INDICATES
wy
oO
FOR THE CHARGING AND SENTENCING STUDY, RACE OF VICTIM
11 DISPARITIES CONTROLLING SIMULTANEOUSLY FOR FELONY CIRCUMSTANCES
12 AND PRIOR RECORD, AND IT FOLLOWS THE PATTERN OF THE, OF THE
13 CRNSS TABULATION ANALYSIS THAT WE EXAMINED EARLIER, BUT HERE. AT
14 THE BOTTOM LINE, WE ESTIMATE THE RACE OF VICTIM EFFECTS THAT ARE
15 PRODUCED BY AN ANALYSIS OF THE FOUR SUB-CATEGORIES OF CASES THAT
14 RESULT FROM THE INTRODUCTION OF THESE TWO BACKGROUND CONTROLS.
17 AS 1 INDICATED BEFORE, EACH BACKGROUND CONTROL
18 |SUBDIVIDES THE CASE, AND WE HAVE TWO OF THEM. WE END UP WITH
lt 19 |FOUR SUB-GROUPS.
20 AND AGAIN THE PATTERN IS CLEAR. AS YOU READ ACROSS THE
21 BOTTOM LINE HERE. THE CASES BECOME MORE AGGRAVATED.
22 THE TABLE WAS CONSTRUCTED IN A WAY SO THAT THE LESS
23 AGGRAVATED GROUPS OF CASES ARE ON THE LEFT SIDE, MOVING TO THE
24 MOST AGGRAVATED ON THE RIGHT SIDE.
23 AND AS ONE DOES THAT. IT IS APPARENT THAT THE DEATH
BALDUS - DIRECT
SENTENCING RATE SHARPLY INCREASES AMONG THE LAST TWO GROUPS.
ry
AND SIMILARLY. IF YOU LOOK DOWN AT THE NEXT-TO-THE-LAST
ROW. YOU SEE THE MEASURE, THE ARITHMETIC DIFFERENCE BETWEEN THE
DEATH SENTENCING RATES FOR THE WHITE AND BLACK VICTIM CASES
WITHIN THESE SUB-GROUPS WHICH HAVE COMMON FACTORS WITH RESPECT
TO THESE TWO CONTROL VARIABLES.
AS WE MOVE ACROSS, YOU SEE IN THE SUB-GROUP. WHERE THE
OVERALL DEATH SENTENCING RATE IS ONLY .01, 17 OUT OF 1281
ESTIMATED CASES RECEIVING DEATH SENTENCES. THE MEASURE OF RACE
pa
O
°
W
N
R
g
N
OF VICTIM DISPARITY IS VERY SMALL. ONE PERCENTAGE POINT.
AS WE MOVE OVER TO THE NEXT SUB-GROUP OF CASES. WHERE
Po
y
Pa
s
12 THE OVERALL DEATH SENTENCING RATE IS .02, WE SEE A SLIGHTLY
13 HIGHER DISPARITY OF FOUR PERCENTAGE POINTS.
14 HOWEVER, IN THE LAST TWO CATEGORIES WHERE WE SEE
13 SHARPLY ELEVATED DEATH SENTENCING RATES. ONE CATEGORY. .15 AND
1&6 IN THE MOST AGGRAVATED CATEGORY, .27. WE SEE SHARP RISES ALSO
17 IN THE RACE OF VICTIM DISPARITIES IN THOSE TWO SUB-GROUPS.
18 THIS IS A, ONE METHOD OF CONTROLLING FOR THE RACE OF
% 12? VICTIM EFFECTS, AND WHAT IT ILLUSTRATES HERE IS THAT RACE OF
20 VICTIM EFFECTS TEND NOT TO BE CONSTANT ACROSS ALL CATEGORIES OF
21 CASES. THAT THEY TEND TO BE CONCENTRATED IN THE MORE AGGRAVATED
22 CASES WHERE THE DEATH SENTENCING RATES ARE ELEVATED AS A RESULT
23 OF AGGRAVATING FACTORS.
24 BUT THIS ANALYSIS WOULD CLEARLY SUGGEST THAT BACKGROUND
25 CONTROL FOR THESE TWO FACTORS DOES NOT EXPLAIN THE RACE OF
on —_ Am———— renee p— " - in ete settee. eit. ite ira Ararat. “A—— S———— A——— —————— ———. — — ———| —— ———. S———— T———
776
BALDUS - DIRECT
1 VICTIM EFFECTS.
- THE COURT: LET ME ASK YOU A QUESTION.
3 THE FIRST TIER, THE ONE SOLITARY BLOCK AT THE TOP OF
4 THE PAGE IS THE RATIO OF CAPITAL CASES TO YOUR UNIVERSE?
LJ
o THE WITNESS: YES, YOUR HONOR.
pe ) THE COURT: AND YOU BREAK OUT IN THE "NO" BLOCK THOSE
7 THAT ARE NOT FELONY MURDER AND THOSE THAT ARE.
3 THE WITNESS: YES.
> THE COURT: AND THEN IN THE FIRST "NO" BLOCK YOU BREAK
10 THAT DOWN AS TO FOLKS WHO HAD BAD PRIOR RECORD AND FOLKS WHO HAD
11 G00D PRIOR RECORD?
12 THE WITNESS: THATS RIGHT. YOUR HONOR.
132 THE COURT: "GOOD" BEING “NO" AND "BAD" BEING "YES."
14 AND THE SAME BREAKDOWN AS TO THE FELONY MURDERS. BASED
13 ON THE RECORD.
16 THE WITNESS: THATS CORRECT. YOUR HONOR.
17 THE COURT: OKAY.
13 BY MR. BOGER:
» 19 |[Q. SO UNDER THAT —-
20 THE COURT: EXCUSE ME. JUST A MOMENT.
21 MR. BOGER: FORGIVE ME. YOUR HONOR.
22 THE COURT: I‘M TRYING TO ABSORB WHAT YOU ALREADY KNOW.
23 IF THERE WERE INSTITUTIONAL BIAS, WHY WOULD THAT NUMBER
24 INCREASE?
23 THE WITNESS: THE, THE IMPACT OF A FACTOR, MY REMARKS
777
BALDUS - DIRECT
APPLIED TO THE IMPACT OF ANY FACTOR WILL BECOME APPARENT ONLY
WHEN THERE IS REALLY ROOM FOR THE EXERCISE OF DISCRETION.
IF THE CASES ARE OF A TYPE THAT ARE BASICALLY
UNTHINKABLE WITH RESPECT TO WHETHER OR NOT THERE SHOULD BE A
DEATH PENALTY, THESE ARE NOT HIGHLY AGGRAVATED CASES. THERE’S NO
LIKELIHOOD THAT WE-RE GOING TO SENTENCE THESE CASES TO DEATH.
AS A CONSEQUENCE. NO ONE INDIVIDUAL FACTOR. EXCEPT THE BASIC
FACTOR THAT PUT THEM IN THAT CATEGORY WHERE THE LIKELIHOOD OF A
DEATH. PROSPECT OF A DEATH SENTENCING SEEMS UNTHINKABLE,
BASICALLY, NO ONE FACTOR IS GOING TO HAVE VERY MUCH EFFECT ON
THOSE CASES.
THE PLACE WHERE FACTORS HAVE AN INFLUENCE IN THE
DECISION MAKING PROCESS IS WHERE THERE IS ROOM FOR THE EXERCISE
OF DISCRETION, WHERE THE DECISION MAKER HAS A CHOICE. THE FACTS
ALLOW SOME SORT OF CHOICE. IF THE CASE IS SO MITIGATED THAT THE
DEATH SENTENCE IS UNTHINKABLE, THE RACE OF THE VICTIM IS NOT
GOING TO HAVE ANY EFFECT. IT’S ONLY AS THE CASES RISE IN THE
LEVEL OF AGGRAVATION THAT THE PROSPECT OF A DEATH SENTENCE
BECOMES THINKABLE., THAT THESE FACTORS BEGIN TO HAVE AN EFFECT.
AND I WOULD SAY. YOUR HONOR, IT“S NOT JUST THE RACE OF VICTIM
THAT SEEMS TO BE HAVING THAT EFFECT IN THIS SYSTEM, IT“S ALL THE
FACTORS, ALL THE OTHER MITIGATING FACTORS. WHEN YOU’RE LOOKING
AT THE VERY LOWLY AGGRAVATED CASES, THAT IF YOU LOOK FOR EXAMPLE
WHETHER THE DEFENDANT SURRENDERED WITHIN 24 HOURS. THE ONE WE
LOOKED AT HERE EARLIER, DOESN‘T SEEM TQ HAVE ANY EFFECT. IF YOU
|
{
a
nN
R
O
1
M
E
.
SA
773
BALDUS - DIRECT
LOOK AT ALMOST ALL THE OTHER, SORT OF MARGINAL AGORAVATING
FACTORS, STATUTORY AND NON-STATUTORY AGGRAVATING FACTORS, THEY
DON‘T SEEM TO HAVE MUCH EFFECT. YOU HAVE GOT TO GET THE CASE
AGGRAVATED UP TO A LEVEL WHERE THERES REALLY SERIOUS
CONSIDERATION GIVEN TO WHETHER OR NOT THIS IS A DEATH WORTHY
CASE.
THAT’S WHEN ALL SORTS OF FACTORS COME INTO PLAY. AND
SHOW THEIR EFFECTS STATISTICALLY, AND THAT’S WHY WE SEE A
STATISTICAL EFFECT IN ONLY THE RIGHT SIDE OF THIS TABLE. BECAUSE
ITS ONLY ON THE RIGHT SIDE OF THIS TABLE THAT SERIOUS
CONSIDERATION IS GIVEN TO THE IMPOSITION OF A DEATH SENTENCE.
IN THESE OTHER CASES OVER HERE. WE HAVE A HANDFUL OF
DEATH SENTENCES. BUT THEY AREN‘T ENQUGH STATISTICALLY TO HAVE
ANY EFFECT, GIVEN THE TREMENDOUS NUMBER OF CASES THAT ARE
INVOLVED IN THOSE CATEGORIES. AND THAT'S WHY WE SEE THESE
EFFECTS. 1 BELIEVE. AS THE CASES BECOME MORE AGGRAVATED.
AND AS I“LL POINT QUT IN A LATER. LATER TESTIMONY, YOUR
HONOR, WHAT FURTHER HAPPENS IS THAT WHEN THE CASES BECOME
TREMENDOUSLY AGGRAVATED 50 THAT EVERYBODY WOULD AGREE THAT IF
WE“RE GOING TO HAVE A DEATH SENTENCE, THESE ARE THE CASES THAT
SHOULD GET IT. THE RACE EFFECTS GO AWAY. IT“S ONLY IN THE
MID-RANGE OF CASES WHERE THE DECISION MAKERS HAVE A REAL CHOICE
AS TO WHAT TO DO. IF THERE‘S ROOM FOR THE EXERCISE OF
DISCRETION, THEN THE FACTORS BEGIN TO PLAY A ROLE.
THIS IS A PHENOMENON THATS WIDELY NOTED IN THE
779
BALDUS - DIRECT
[e
y LITERATURE IN VARINUS FIELDS THAT THE INFLUENCE OF FACTORS OVER
DECISIONS COMES TO BEAR ONLY WHEN THERES REAL ROOM FOR THE
EXERCISE OF DISCRETION.
THE COURT: DON’T FUDGE UNLESS YOU HAVE SOME LEEWAY TO
FUDGE, SOME WAY TO EXPLAIN WHAT YOURE DOING.
THE WITNESS: THAT’S RIGHT. MOST SYSTEMS ARE -—
THE COURT: THAT’S HUMAN NATURE. NOW THAT YOU“VE
EXPLAINED IT IN A SCHOLARLY WAY I RECOGNIZE IT AS BEING A
og
.
0
e
d
B
N
PRINCIPLE I HAVE OBSERVED IN LIFE.
10 ALL RIGHT. NOW I UNDERSTAND -74. AND UNDER THE SAME
11 GUIDELINES, IT“LL BE ADMITTED.
12 MR. BOGER: THANK YOU, YOUR HONOR.
13 BY MR. BOGER:
14 @. DID YOU DO THE SAME SORT OF ANALYSIS, USING A DIFFERENT
13 MEASURE. PROFESSOR BALDUS?
16 A. YES.
17 THE COURT: PROFESSOR BALDUS IS PROBABLY A CLEANER
13 LIVER THAN I AM, BUT WITH ALL OF THIS HEAVY WORK, I DO LIKE TO
® 19 TAKE BREAKS MORE OFTEN, EITHER FOR HIS MIND OR MINE OR YOURS, OR
20 SOMEBODY S, MR. FORD’S, I THINK.
21 LETS TAKE A TEN MINUTE BREAK.
22 MR. BOGER: THANK YOU, YOUR HONOR.
23 ---
24 {RECESS TAKEN.)
23 THE COURT: YOU MAY PROCEED.
be Sree bre ete meio ese Pent eee —
Eo ——————————————
730
BALDUS - DIRECT
3 -— wn —-—
2 DAVID C. BALDUS,
3 BEING PREVIOUSLY DULY SWORN. RESUMED THE WITNESS STAND AND
TESTIFIED FURTHER AS FOLLOWS:
DIRECT EXAMINATION (CONT'D)
BY MR. BOGER:
4
5
é
7 Q. PROFESSOR BALDUS, YOU INDICATED THAT YOU HAD EMPLOYED A
8 DIFFERENT MEASURE TO EXAMINE THE RACE OF DEFENDANT AND RACE OF
? VICTIM WHILE CONTROLLING SIMULTANEOUSLY FOR FELONY CIRCUMSTANCE
0 AND PRIOR RECORD.
i I BELIEVE I ASKED YOU TO IDENTIFY DB-77.
12 A. DB-77 1S TABLE 21 FROM OUR REPORT AND IT PRESENTS RACE OF
13 DEFENDANT AND RACE OF VICTIM DISPARITIES ESTIMATED WITH THE DATA
14 IN THE CHARGING AND SENTENCING STUDY.
15 AND IT INDICATES IN COLUMN B, ROW 1, THAT A REGRESSION
16 CO-EFFICIENT OF .09 WAS ESTIMATED IN THAT ANALYSIS,
17 SUGGESTING THAT WHEN THE RACE OF DEFENDANT, THE FELONY
18 CIRCUMSTANCES VARIABLE AND THE PRIOR RECORD VARIABLE ARE HELD
p 1? CONSTANT THAT THERE IS A NINE PERCENTAGE POINT DISPARITY IN THE
20 RATE AT WHICH WHITE VICTIM-BLACK VICTIM CASES RESULT IN DEATH
21 SENTENCES.
22 MR. BOGER: YOUR HONOR -- I’M SORRY.
23 THE WITNESS: I WAS GOING TO A ELABORATE ON A POINT.
24 BY MR. BOGER:
23 @. CONTINUE, PLEASE?
9g
.
0
4
BO
BA
751
BALDUS — DIRECT
A. I WAS GOING TO POINT OUT TO YOUR HONOR ONE INTERESTING
PROPERTY OF A REGRESSION CO-EFFICIENT THAT CAN BE OBSERVED BY
COMPARING THE RACE OF VICTIM CO-EFFICIENT IN TABLE 1 FOR, TABLE
| FOR THE LEAST SQUARES ANALYSIS, .09, IF YOU WOULD LOOK BACK ON
DB-76, THAT WAS THE TABLE, THE FIGURE THAT WE JUST EXAMINED,
YOU‘LL NOTICE ACROSS THE BOTTOM THAT THERE ARE DISPARITIES IN
THESE DIFFERENT CELLS THAT VARY SUBSTANTIALLY. ONE IS .01,
ANOTHER IS .04, .14, ANOTHER IS .25.
WHEN YOU AVERAGE THOSE. YOU GET THE .09. THAT’S WHAT
THE LEAST SGUARES REGRESSION CO-EFFICIENT DOES, IT GIVES YOU AN
AVERAGE OF THE RACE OF VICTIM EFFECTS. OR IF IT HOLDS WITH
RESPECT TO ANY VARIABLE THAT YOU‘RE CONCERNED WITH, VARIABLES
HAVE DIFFERENT EFFECTS AMONG DIFFERENT SUB-GROUPS OF CASES AND
THE REGRESSION CO-EFFICIENT. ESTIMATED IN THE LEAST SGUARES
ANALYSIS. GIVES ONE AN AVERAGE OF THE EFFECTS ACROSS ALL THESE
DIFFERENT GROUPS.
Q. WHAT ARE THE RACE OF DEFENDANT EFFECTS UNDER THE REGRESSION
ANALYSIS?
A. IN THE LEAST SQUARES ANALYSIS. THE RACE OF DEFENDANT EFFECT
1S FIVE PERCENTAGE POINTS. BUT IN THE LOGISTIC REGRESSION
ANALYSIS. IT IS SMALL AND IT‘S NOT STATISTICALLY SIGNIFICANT, IN
CONTRAST TO THE RACE OF VICTIM EFFECT. WHICH IN BOTH THE LEAST
SQUARES ANALYSIS AND IN THE LOGISTIC ANALYSIS IS A, SUBSTANTIAL
AND HIGHLY STATISTICALLY SIGNIFICANT.
MR. BOGER: YOUR HONOR, I MOVE DB-777S ADMISSION INTO
—— So——— ——— ———. —— ————— ———a— ——— ———. ——— ———
BALDUS ~ DIRECT
re
y EVIDENCE.
THE COURT: ALL RIGHT. LET ME ASK YOU AN ADDITIONAL
QUESTION, JUST TO MAKE SURE I‘M INTERPRETING IT RIGHT.
FELONY CIRCUMSTANCE, THAT LINE ON -77. INDICATES THE
DIFFERENCE ABOVE THE AVERAGE THAT. THAT THAT CIRCUMSTANCE MAKES
WHERE YOU HOLD THE OTHER 3 FACTORS CONSTANT?
THE WITNESS: YES, YOUR HONOR.
THE COURT: UNDER THE SAME CIRCUMSTANCES AND
gg
OO
NN
0
Aa
2
W
W
OBSERVATIONS, IT LL BE ADMITTED.
10 MR. BOGER: I“M SORRY. YOUR HONOR. I WAS --
11 THE COURT: I ADMITTED IT UNDER THE SAME OBSERVATIONS I
12 MADE PREVIOUSLY.
13 MR. BOGER: OH. THANK YOU, YOUR HONOR.
14 BY MR. BOGER:
15 @. PROFESSOR BALDUS, DID THERE COME A TIME WHEN YOU ATTEMPTED
146 TO CONTROL FOR THE THE PRESENCE OR ABSENCE OF ALL OF THESE
17 STATUTORY AGGRAVATING CIRCUMSTANCES, UNDER THE GEORGIA SYSTEM.
18 SIMULTANEOUSLY, IN ADDITION TO THE RACE OF THE DEFENDANT AND
19 RACE OF THE VICTIM?
20 A. YES.
~1 |@. LET ME ASK YOU TO TURN TO DB-72 MARKED FOR IDENTIFICATION.
22 I'LL ASK YOU TO IDENTIFY THAT DOCUMENT?
23 A. DB-78 IS TABLE 23 FROM OUR REPORT. AND IT PRESENTS
24 REGRESSION CO-EFFICIENTS ESTIMATED IN AN ANALYSIS, INCLUDING THE
23 VARIABLES OF RACE OF VICTIM, AND RACE OF DEFENDANT. PLUS THE
783
BALDUS - DIRECT
OTHER STATUTORY AGGRAVATING FACTORS. PROVIDED IN THE GEORGIA
pot
e
STATUTE AS WELL AS A VARIABLE FOR PRIOR RECORD. AND THE RESULTS
FOR THE RACIAL VARIABLES ARE SHOWN IN ROWS 1 AND 2, RATHER.
THEY/RE NOT NUMBERED. BUT THE FIRST TWO ROWS OF THE TABLE.
AND THE RACE OF VICTIM CO-EFFICIENTS IN THE LOGISTIC
AND REGRESSION, PARDON ME, IN THE LOGISTIC AND LEAST SQUARES
REGRESSION ANALYSES. SUGGEST STRONG STATISTICAL EFFECT, AND BOTH
OF WHICH ARE HIGHLY SIGNIFICANT STATISTICALLY.
IN ROW 2 WE SEE IN THE LEAST SQUARES ANALYSIS A FOUR POINT
Pe
ry
O
°
9
H
N
N
W
N
IMPACT ASSOCIATED WITH BEING A BLACK DEFENDANT. AND, HOWEVER. IT
IS ONLY SIGNIFICANT AT THE .09 LEVEL. tb
re
y
12 AND IN THE LOGISTIC RESULTS, THE RACE OF THE DEFENDANT
13 EFFECT IS QUITE SMALL AND NOT AT ALL SIGNIFICANT FROM A
14 STATISTICAL STANDPOINT.
13 THE COURT: YOU REALLY HAVE NOT TOLD ME MUCH ABOUT THE
14 LOGISTICAL REGRESSION RESULT. BUT HERE AS A JUDGE, I GUESS IVE
17 GOT TO MAKE A CHOICE AS TO WHICH IS MORE IMPORTANT AS BETWEEN
18 THE ANALYSIS IN B AND THE ANALYSIS IN C AND D. FOR EXAMPLE.
X 3 39 RACE OF THE DEFENDANT, THE ODDS MULTIPLIER IS THE SAME. BLACK
20 DEFENDANT AND WHITE DEFENDANT GOT THE SAME PROBABILITY.
21 THE WITNESS: THAT'S RIGHT.
22 THE COURT: DO YOU HAVE AN OPINION, BASED ON YOUR
23 ACRAUAINTANCE WITH STATISTICS AS TO WHICH, AS BETWEEN THE
24 ANALYSIS UNDER B AND THE ANALYSIS UNDER C, D, WHICH IS THE MORE |
23 RELIABLE?
734
BALDUS ~ DIRECT
1 THE WITNESS: I DO NOT CONSIDER MYSELF COMPETENT TO
2 EXPRESS AN OPINION AS TO THEIR RELIABILITY FROM A TECHNICAL
STANDPOINT.
THE COURT: YOU UNDERSTAND THE COMPARABLE RELIABILITY.
COMPARING THE ONE WITH THE OTHER, WHICH IS MORE LIKELY TO OBJECT
3
4
S
» 4 |PROBATIVE? YOU MAY NOT FEEL COMFORTABLE ABOUT IT. IF YOU
7 |DON’T, SAY SO.
8 THE WITNESS: 1 WILL, YOUR HONOR. WHAT I WAS GOING TO
9 | SUGGEST, YOUR HONOR, IS THAT THEY ADDRESS THE QUESTION FROM THE
10 |BASIS OF SOMEWHAT DIFFERENT ASSUMPTIONS, AND THEY BOTH SHED FROM
11 |MY PERSPECTIVE RELEVANT LIGHT ON THE QUESTION.
12 WHEN YOU SEE THAT THE RESULTS FROM BOTH PRODUCE CONSISTENT
13 | ANSWERS. IT GIVES YOU A GREAT DEAL OF CONFIDENCE THAT YOURE
14 |SEEING SOMETHING THATS REALLY THERE AND PERSISTENT.
15 . WHEN YOU SEE INSTABILITY BETWEEN THEM. IT GIVES YOU
16 |PAUSE ABOUT THE RESULTS. IT’S AT THAT LEVEL THAT I CAN EVALUATE
17 |IT. YOUR HONOR.
18 THE COURT: ALL RIGHT.
= 19 |BY MR. BOGER:
20 |@. WHAT SIGNIFICANCE. GENERALLY. DO YOU DRAW FROM THE
21 |REGRESSION RESULTS REPORTED IN DB-787
22 |A. THE RESULT INDICATES THAT INTRODUCTION OF SIMULTANEOUS
23 |CONTROLS FOR ALL THE STATUTORY AGGRAVATING FACTORS DOES NOT
24 |EXPLAIN THE RACE OF VICTIM EFFECT, WHICH REMAINS STRONG, BUT
29 THAT IT DOES HAVE THE EFFECT OF EXPLAINING PART. THE RACE OF
T
E
R
BE
E
E
R
a
v
i
h
o
a
0
10
11
7383
BALDUS - DIRECT
DEFENDANT EFFECT OVERALL IN THE SYSTEM.
@. YOU STRESS OVERALL IN THE SYSTEM. WHAT‘S THE REASON FOR
THAT?
A. WELL. WHAT WE OBSERVED IN OUR ANALYSIS OF THE RACE OF
DEFENDANT EFFECTS IS THAT THEY TEND TO BE CONCENTRATED IN WHITE
VICTIM CASES AMONG QUITE AGGRAVATED CASES. AND THAT WHEN YOU
AVERAGE IN ALL THE CASES, THE LACK OF RACE OF DEFENDANT EFFECT
THROUGHOUT MUCH OF THE SYSTEM TENDS TO SWAMP THE EFFECTS THAT
YOU SEE IN THIS NARROW CATEGORY OF CASES.
AS YOULL SEE. YOUR HONOR, AS WE PROCEED ALONG, THERE
ARE NOT THAT MANY CASES IN THIS ENTIRE SYSTEM THAT ARE HIGHLY
AGGRAVATED, AND THAT IS. AS A RELATIVE, TO THE PROPORTION OF THE
TOTAL NUMBER OF CASES, THERE JUST ARE NOT THAT MANY OF THESE
HOMICIDES THAT RAISE A HIGH PROBABILITY OF A DEATH SENTENCE. AND
AMONG THOSE CASES OUR EVIDENCE SUGGESTS. THAT THE RACE OF
DEFENDANT EFFECTS TEND TO BE CONCENTRATED AMONG THE WHITE VICTIM
CASES WITHIN THAT SUB-GROUP.
SO BECAUSE THESE CO-EFFICIENTS TEND TO REFLECT THE
AVERAGE EFFECT ACROSS THE WHOLE SYSTEM, BECAUSE IN SO MANY OF
|THE OTHER CASES. YOU DON’T SEE ANY EFFECTS. 1 BELIEVE THAT’S THE
REASON WE GET THIS INSTABILITY IN THE RACE OF DEFENDANT
CO-EFFICIENTS.
'@R. LET ME SIMPLY ADD FOR THE RECORD, WHAT IS THE LAST FACTOR
CONTROLLED FOR IN THIS TABLE?
A. OH, THE LAST FACTOR CONTROLLED FOR IS A VARIABLE WHICH
rn. —. | S—————t. Serer prerererret —
786
BALDUS - DIRECT
1 REFLECTS THE PRESENCE IN THE CASE OF EITHER A FELONY
CONVICTION OR A CONVICTION FOR TWO OR MORE VIOLENT MISDEMEANORS. nM
3 THAT VARIABLE IS INCLUDED IN THE ANALYSIS BECAUSE OF THE
4 PROVISION IN THE GEORGIA STATUTE WHICH SPECIFICALLY ALLOWS
EVIDENCE OF PRIOR RECORD TO BE ADMITTED AT THE PENALTY TRIAL (&
5 PHASE, AND, IN A DEATH SENTENCING CASE.
7 THE COURT: SO IN THIS, YOU HAVE NOT JUST CONTROLLED
= FOR THE STATUTORY AGGRAVATED CIRCUMSTANCES, YOU HAVE ALSO
7 CONTROLLED FOR THAT?
10 THE WITNESS: YES. YOUR HONOR. BUT AS YOU CAN SEE FROM
11 LOOKING AT THE CO-EFFICIENT OF THAT, IT REALLY DOESN‘T ADD VERY
12 MUCH. THE Bl FACTOR, PRIOR CAPITAL RECORD. THAT'S THE VARIABLE
13 THAT CARRIES THE INFORMATION THAT INFLUENCES THIS SYSTEM WITH
14 RESPECT TO PRIOR RECORD. BUT YOU CAN SEE THE CO-EFFICIENT. YES.
15 THE REGRESSION CO-EFFICIENT FOR THE, THIS LAST VARIABLE FOR
16 FELONY, VIOLENT MISDEMEANOR. DOESN‘T REALLY APPEAR TO HAVE MUCH
17 EFFECT. THE Bl FACTOR IS REALLY WHAT INFLUENCES THE PROCESS,
13 APPEARS TO.
% 19 BY MR. BOGER:
20 |@. LET ME ASK YOU AN ADDITIONAL QUESTION.
21 DO THESE FIGURES REFLECT THAT RACE OF THE VICTIM FACTOR
22 IS MORE SIGNIFICANT OR HAS MORE IMPACT ON THE SENTENCING OUTCOME
23 THAN SOME OF THE STATUTORY AGGRAVATING CIRCUMSTANCES THEMSELVES,
24 WHEN YOU CONTROL FOR ALL AGGRAVATING CIRCUMSTANCES IN THESE
23 PRIOR RECORDS?
O
B
W
N
A
l
R
P
N
e
fa
re
r
w
|
pa
787
BALDUS - DIRECT
A. YES. AGAIN I WANT TO CAUTION THE COMPARISON OF THESE
| REGRESSION CO-EFFICIENTS GIVES ONE JUST A ROUGH MEASURE OF THEIR
POSSIBLE IMPACT IN THE SYSTEM. AND IT IS PLAIN THAT THE RACE OF
THE VICTIM VARIABLE APPEARS TO HAVE MORE EFFECT THAN SEVERAL OF
THE STATUTORY AGGRAVATED FACTORS, BUT I DON‘T WANT TO MAKE TOO
MUCH OF THAT.
MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE DB-~78 INTO
EVIDENCE. AND WOULD SUGGEST THAT WE‘VE NOW TOPPED OUT, IF YOU
WOULD, ON WHAT MAY BE NECESSARY IN ORDER TO GET THIS ADMITTED
FOR ALL PURPOSES.
I THINK IT DOES ADDRESS ALL THE THE STATUTORY
AGGRAVATING FACTORS IN THE PRIOR RECORD WHICH IS ADMISSIBLE AT
THE PENALTY PHASE AND CONTROLLED FOR ALL OF THEM SIMUL TANEOUSLY.
MS. WESTMORELAND: YOUR HONOR. I WOULD OBVIQUSLY MAKE
THE SAME OBJECTION AS ALL ALONG, AND WE STILL ASSERT THAT WHILE
THIS TABLE DOES INCLUDE MORE FACTORS THAN HAS BEEN PREVIOUSLY
INCLUDED IN THE OTHER TABLES. IT FALLS FAR SHORT OF INCLUDING
ALL POSSIBLE FACTORS THAT COULD EXPLAIN ANY DISPARITIES THAT MAY
EXIST, AND IT STILL LACKS THE RELEVANCY NECESSARY AT THIS POINT.
THE COURT: THESE RESULTS, IF I UNDERSTAND IT
CORRECTLY. PROFESSOR BALDUS, INCLUDE DECISION MAKING PRIOR TO
JURY DECISION MAKING?
THE WITNESS: YES, YOUR HONOR. REFLECTS THE DECISION
THAT WE ARE EXPLAINING IN THIS ANALYSIS, IS THE DEATH SENTENCING
OUTCOME AMONG ALL CASES IN THE UNIVERSE THAT WE“RE STUDYING, IT
——— | —— — —— —————— | I—— —
788
BALDUS - DIRECT
1 EXPLAINS THE IMPACT OF THESE DECISIONS ON THE IDENTIFICATION OF
2 128 OUT OF 2484 TOTAL. WHAT ARE THE FACTORS THAT EXPLAIN HOW
3 THESE 128 GET PARTED FROM THE REST OF THAT COMPANY.
4 AND THAT REFLECTS THE DECISIONS AT SIX POINTS, AND THE
3 COMBINED EFFECTS OF DECISION AT SIX POINTS IN THIS PROCESS.
& THE COURT: NOW THE PROBLEM IS WEAKNESSES IN THE TABLE
7 WOULD INCLUDE THE FACT THAT IT IS. NOT ADJUSTED OR, FOR ANYTHING
a THAT WOULD NOT HAVE OR THAT MIGHT HAVE A SIGNIFICANT EFFECT ON
? THE PROSECUTOR IN HIS DECISION MAKING.
10 I WOULD BE THE FIRST TO AGREE THAT IT DOES. IF IT WERE
11 ONLY A VIEW OF WHAT THE JURY WAS DOING WOULD BE HIGHLY
12 SIGNIFICANT.
13 I WILL ADMIT IT. I THINK YOU MAY BE OVER THE
14 THRESHOLD.
15 MR. BOGER: THANK YOU, YOUR HONOR.
14 BY MR. BOGER:
17 @. PROFESSOR BALDUS, YOU MAY BE AWARE THAT MITIGATING FACTORS
18 ARE ALSO APPROPRIATE CONSIDERATION FOR JURIES IN CAFITAL CASES:
® 19 IT MAY EVEN COME INTO PROSECUTORIAL DECISION MAKING IN CAPITAL
20 CASES.
21 HAVE YOU DONE ANY ANALYSES THAT CONTROL FOR THE
22 PRESENCE OF MITIGATING FACTORS IN A CASE, AS WELL AS STATUTORY
23 AGGRAVATING FACTORS AND PRIOR RECORD?
24 A. YES. WE HAVE.
25 @. LET ME DIRECT YOUR ATTENTION TO DB-7% MARKED FOR
o
O
WM
N
T
0
N
e
Eo pb
ry
nN
13
14
73%
BALDUS - DIRECT
IDENTIFICATION, AND ASK YOU IF YOU CAN IDENTIFY THAT DOCUMENT?
A. DB-79 PRESENTS THE RESULTS OF TWO REGRESSION ANALYSES, ONE
EMPLOYING THE LEAST SQUARES METHOD, THE OTHER THE LOGISTIC
METHOD OF ANALYSIS.
IN THIS ANALYSIS WE INCLUDED ALL OF THE STATUTORY
AGGRAVATING FACTORS, PLUS THE FELONY RECORD, FINAL MISDEMEANOR
RECORD VARIABLE THAT WAS INCLUDED IN THE PRECEDING ANALYSIS.
AND IN ADDITION WE ADDED 75 MITIGATING FACTORS.
THE COURT: WHERE IS THAT?
THE WITNESS: THEY’RE NOT LISTED HERE IN THIS TABLE
ITSELF, YDUR HONOR. THEY‘RE LISTED IN A SEPARATE APPENDIX. IT
WAS JUST A MATTER OF ECONOMY IN PRESENTATION OF THE RESULTS.
THE COURT: ALL RIGHT.
THE WITNESS: BUT THOSE FACTORS WERE INCLUDED IN THE
ANALYSIS, AND THE RACIAL EFFECTS THAT ARE ESTIMATED HERE WERE
ESTIMATED AFTER SIMULTANEOUS ADJUSTMENT FOR ALL THOSE FACTORS.
ALL THOSE FACTORS WERE HELD CONSTANT BEFORE THESE RACIAL
CO-EFFICIENTS WERE ESTIMATED IN THIS ANALYSIS.
BY MR. BOGER:
A. ARE THE ENTIRE SET OF FACTORS THAT WERE CONTROLLED FOR
IN YOUR FINAL REPORT?
A. YES, THEY'RE LISTED IN THE TECHNICAL APPENDIX. THE NAMES OF
THE OTHER VARIABLES THAT WERE INCLUDED IN THIS ANALYSIS ARE
INCLUDED.
@. DO YOU RECALL SOME OF THE KINDS OF ADMITTING FACTORS THAT
790
BALDUS ~ DIRECT
WERE INCLUDED?
A. YES.
THE COURT: SOMEBODY SHOW ME THE LIST. IF YOU“RE NOT
PUTTING THAT IN EVIDENCE, YOU OUGHT TO.
MR. BOGER: WE SHALL, YOUR HONOR. IT“S GOING TO BE
PART OF THE FINAL REPORT, BUT I THOUGHT AT THIS POINT FOR
CLARIFICATION, --
THE COURT: I CAN‘T VERY WELL UNDERSTAND THE LIABILITY
OF IT UNTIL I KNOW WHAT HE’S ADJUSTED FOR.
MR. BOGER: I‘M AFRAID I HAVE NOTHING AT THIS MOMENT TO
HAND YOU, BUT WELL GET SOMETHING VERY SHORTLY.
PROFESSOR BALDUS HAS HIS COPY OF THE FULL REPORT.
FERHAPS HE CAN GIVE IT TO YQU AT THIS TIME.
THE COURT: IN THIS ONE, IS THIS THE ONE YOU SET UP A
DICHOTOMOUS CODING. THAT IS, IF ANY ONE OF THE 735 WAS PRESENT,
IT WAS COUNTED IN. AND YOU MADE NO ATTEMPT TO WEIGHT THEM,
AGGREGATE THEM. WHAT HAVE YOU?
THE WITNESS: MOST OF THE VARIABLES ARE DICHOTOMOUS
VARIABLES, THAT’S RIGHT, YOUR HONOR, BUT I BELIEVE --
THE COURT: FOR THE PURPOSE OF CONTROLLING ON 75, YOU
MADE THE DICHOTOMOUS DECISION IF ANY OF THE 735 ARE PRESENT, YOU
WILL COUNT A MITIGATING FACTOR?
THE WITNESS: 0H, NO, YOUR HONOR. EACH ONE, WITH
RESPECT TO EACH CASE, THE QUESTION WAS ASKED. WITH RESFECT TO
EACH VARIABLE, SO THAT EACH MITIGATING VARIABLE WAS ADJUSTED FOR
yy
.
0
t
w
T
E
R
d
N
(a
y o
11
12
791
BALIUS - DIRECT
SEPARATELY WITH RESPECT TO EACH CASE, IN THE CONDUCT OF THE
ANALYSIS.
SO IT’S NOT A QUESTION OF GOING THROUGH AND ASKING IF
ANY ONE OF THEM WAS THERE. THERE WERE 735 DIFFERENT QUESTIONS
ASKED WITH RESPECT TO EACH CASE IN THIS ANALYSIS.
THE COURT: WE‘RE PROBABLY SAYING THE SAME THING?
THE WITNESS: PERHAPS WE ARE.
THE COURT: IT GOT INTO THE BOX. IF YOU WILL. OF HAVING
A MITIGATING CIRCUMSTANCE IF IT ANSWERED ANY ONE OF THE 73
QUESTIONS "YES."
THE WITNESS: BUT LET ME TRY AND GIVE YDU AN
ILLUSTRATION, YOUR HONOR, OF WHAT CONCEPTUALLY THIS REGRESSION
DOES WHEN IT DEALS WITH A LARGE GROUP OF VARIABLES.
IT TENDS, WELL, NOT TENDS, WHAT IT DOES IS THROUGH AN
ALGEBRAIC PROCEDURE. THESE GROUPS OF VARIABLES THAT ARE THE SAME
WITH RESPECT TO ALL OF THE DIFFERENT FACTORS THAT WE‘RE
CONTROLLING FOR SIMULTANEOUSLY.
SO IN A LARGE, ON A LARGE SCALE, IT PRODUCES SOMETHING
THAT'S COMPARABLE TO THE CROSS TABULATION THAT WE LOOKED AT. IT
BROKE DOWN CASES INTO DIFFERENT SUB-CATEGORIES. AND EACH
CATEGORY OF CASES HAD CERTAIN CHARACTERISTICS IN COMMON.
WELL, THE REGRESSION PROCEDURE DOES THE SAME THING
THROUGH A METHOD OF ALGEBRAIC ADJUSTMENT. IT IDENTIFIES GROUPS
OF CASES, NOT PHYSICALLY SORTING THEM INTO CATEGORIES LIKE WE
DID HERE, BUT ALGEBRAICALLY PRODUCES THOSE SORTS OF SUB-GROUPS
9
0
N
N
N
D
pb
Tr
y
Po
y
792
BALDUS - DIRECT
OF CASES, AND WITHIN THOSE SUB-GROUPS OF CASES. THE RACIAL
EFFECTS ARE CALCULATED AND AVERAGED ACROSS THOSE DIFFERENT
SUB~-GROUPS OF CASES.
THATS IN AN INTUITIVE LEVEL THE WAY THE PROCEDURE
OPERATES.
THE COURT: WHAT YOU-RE SAYING IS YOU AVERAGE THE
RACIAL EFFECTS FOR THE 75 ADMITTING CIRCUMSTANCES, INDIVIDUALLY.
THE WITNESS: NO, YOUR HONOR. WHAT WE DO. IF YOU COULD -
THE COURT: AND ALL THE OTHER FACTORS. TOO.
THE WITNESS: WELL WHAT WE DO. WE MAKE A SIMULTANEOUS
ADJUSTMENT. THAT-S THE NOTION OF CONTROLLING SIMULTANEOUSLY FOR
ALL OF THESE FACTORS IS THAT WE -—-
THE COURT: LET ME TELL YOU WHAT I“M REALLY WORRIED
ABOUT.
THE WITNESS: ALL RIGHT.
THE COURT: IF YOU HAD MORE THAN, MORE THAN ONE
ADMITTING CIRCUMSTANCE OR MORE THAN TWO OR MORE THAN THREE. YOU
WEREN’T ABLE TO SHOW THE EFFECT OF THE PRESENCE OF MULTIPLE
MITIGATING CIRCUMSTANCES.
THE WITNESS: OH. YES, YOU WERE, YOUR HONCR.
THE COURT: ON THIS?
THE WITNESS: YES. EXACTLY. THAT’S THE PURPOSE OF IT,
SHOW THE PRESENCE OF MULTIPLE MITIGATING FACTORS 20 THAT A CASE
THAT HAD FOUR OR FIVE OR TEN OUT OF THE 73 POSSIBLE FACTORS THAT
WERE MITIGATING, THAT WOULD BE IN A CATEGORY ALL BY ITSELF. IT
R
ro
Ke
po
-
3
4
5
&
4
3
3
10
BALDUS - DIRECT
WOULD BE QUITE DIFFERENT FOR THE PURPOSES OF CALCULATING
DISPARITIES THAN WOULD BE CASES THAT HAD ONLY ONE OR TWO OR
NONE.
THE COURT: YOU MADE NO EFFORT TO WEIGHT MITIGATION,
RELATIVE MITIGATION IN ANY FACTOR?
THE WITNESS: NO.
THE COURT: FOR INSTANCE, I NOTICED YOU HAD ONE FACTOR
THAT MAY BE INCLUDED WHERE YOU CALLED THE UNDERLYING FELONY
MORALLY JUSTIFIED BECAUSE HE WAS STEALING BECAUSE HE HAD A POOR
CHILD. PROBABLY MANY JURORS IN GEORGIA THAT WOULD HAVE VIEWED
THAT AS BEING MORALLY MITIGATING. BUT THAT MIGHT BE WEIGHED
TOGETHER WITH SOMETHING THAT WOULD BE MORE UNIVERSALLY
COMPELLING.
THE WITNESS: THE IDEA HERE. YOUR HONOR, IN THIS SORT
OF REGRESSION ANALYSIS IS NOT TO ASSIGN ANY A PRIORI WEIGHTS TO
ANYTHING BUT RATHER TO SORT CASES INTO CATEGORIES WHERE CASES
ARE COMPARABLE WITH RESPECT TQ EACH OTHER. I CAN HARK BACK TO
THE DISCUSSION WE HAD THE OTHER DAY. THAT THE PURPOSE OF THE
ANALYSIS IN A RETROSPECTIVE STUDY OF THE TYPE WE RE CONDUCTING
IS TO TRY AND REPLICATE WHAT WENT ON IN A CONTROLLED EXPERIMENT.
THAT IS, TO TRY AND GET CASES SO THAT EVERYBODY IS THE SAME WITH
RESPECT TO ALL FACTORS, EXCEPT THE FACTOR WHOSE IMPACT WE'RE
TRYING TO ESTIMATE.
ONE OF THE PRINCIPAL METHODS FOR DOING THAT IS MULTIPLE
REGRESSION TECHNIQUE. AND WHAT IT DOES IS PRODUCES CASES THAT
PR A ————
nerve a ———
774
BALDUS - DIRECT
pt
e ARE COMPARABLE WITH RESPECT TO THE BACKGROUND FACTORS, THEN ASKS
WHETHER OR NOT THERE ARE DIFFERENCES WITH RESPECT TO THE OUTCOME
VARIABLE, THAT IS. THE DEATH SENTENCING RATES THAT ARE
ASSOCIATED WITH THE VARIABLE OF INTEREST, WHICH IN THIS CASE IS
THE RACIAL VARIABLE.
sn WE CAN SAY THAT THE PURPOSE OF THIS INQUIRY IS TO
| IDENTIFY GROUPS OF CASES THAT ARE COMPARABLE WITH RESPECT TO ALL
FACTORS, EXCEPT THE RACIAL FACTOR, AND THEN WE ASK TO WHAT
oy
6
N
N
2
0
a
U
N
EXTENT DOES THE RACE DIFFER WITHIN THESE SUB-GROUPS OF CASES ALONG
fy
oO
THE LINES OF THE RACE OF THE VICTIM, SO WHAT YOU HAVE
fa
y
PY
CONCEPTUALLY 1S A LARGE-SCALE BREAK DOWN OF THESE CASES AND THEN
12 THESE DISPARITIES ARE CALCULATED WITHIN THESE SUB-GROUPS. AND
13 THEN AN AVERAGE DISPARITY IS ESTIMATED.
14 THE COURT: WELL. THAT’S WHAT I WAS WONDERING, BECAUSE
13 IF THATS WHAT YOU‘RE DOING, AND I WAS JUST LOOKING AT THE TABLE
146 HERE, IF N.S. MEANS WHAT I THINK IT MEANS, WHICH IS WOT
17 SIGNIFICANT. IS THAT WHAT IT MEANS? 13 THE WITNESS: I‘M SORRY, —-
» 19 THE COURT: PAREN N.5.7 |
20 THE WITNESS: YES, THAT MEANS NOTHING SIGNIFICANT.
21 THE COURT: AREN‘T YOU BASICALLY DIVIDING UP INTO A
22 WHOLE BUNCH SUB-GROUPS, NO ONE OF WHICH OR THE VAST MAJORITY OF
23 WHICH WOULD NOT BE LARGE ENOUGH TO BE STATISTICALLY SIGNIFICANT
24 ON ITS OWN. AND FROM ALL OF THESE SUB-GROUPS, YOU MAKE AN
23 ANALYSIS OF WHAT HAPPENS TO ALL THESE LITTLE 3SUB-GROUPS THAT ARE
BALDUS ~ DIRECT
1 NOT STATISTICALLY SIGNIFICANT, AND PUT THEM ALL TOGETHER AND IT
2 BECOMES STATISTICALLY SIGNIFICANT.
THE WITNESS: WELL, YOU‘RE REACHING THE LIMIT OF MY
UNDERSTANDING OF THE UNDERLYING METHOD, YOUR HONCR.
BUT ONE OF THE PRINCIPAL PURPOSES OF REGRESSION
[A
d
3
3
2
wn & ANALYSIS IS TO BE ABLE TO ESTIMATE THE EFFECTS OF A VARIABLE WHILS
7 SIMULTANEOUSLY CONTROLLING FOR A SERIES OF BACKGROUND VARIABLES.
3 AND NOT RUN INTO THE PROBLEM YOU ENCOUNTER WHEN YOU DO ACTUALLY
9 PHYSICALLY BREAK THEM INTO SMALLER GROUPS. IT“S A METHOD THAT’S
10 COMPARABLE TO THAT BREAKDOWN PHYSICIALLY. IT“S DONE AN
11 ALGEBRAIC ADJUSTMENT WHICH IS GENERALLY ACCEPTED AS PERFECTLY
12 VALID.
13 THE COURT: THAT’S SOMETHING ID LIKE FOR THE OTHER
14 WITNESS TO BE ABLE TELL ME ABOUT.
13 DO YOU HAVE THAT LIST?
146 THE WITNESS: YES, YOUR HONOR. UNFORTUNATELY. THE LIST
17 ITSELF IS NOT WRITTEN OUT IN PLAIN ENGLISH, IT“S WRITTEN OUT IN
13 COMPUTERESE. I-M GOING TO HAVE TO GET YOU A TYPED UP A COPY OF
® 12 THIS, WHICH I WILL PROMPTLY.
20 THE COURT: ALL RIGHT.
21 MR. BOGER: YOUR HONOR. FOR YOUR CONVENIENCE, PROFESSOR
22 BALDUS, WHY DON’T YOU JUST READ A COUFLE OF THOSE INTO THE
23 RECORD IN ENGLISH FROM HIS REPORT? I HAVE A COPY OF THE REPORT
24 IN THE APPENDIX IF YOUR HONOR WOULD LIKE --
25 THE COURT: IS IT IN PLAIN ENGLISH?
—— —y ete. feel Se. free ett. ere, {ret ie: - —— — —— ————. ——— i ——————— ——— —— ——— — ———
79h
BALDUS - DIRECT
1 MR. BOGER: I“M AFRAID IT-S IN THE SAME KIND OF
< LANGUAGE.
3 THE COURT: I WILL RESERVE RULING ON DB-79 UNTIL YOU
SHOW ME IN PLAIN ENGLISH WHAT ITS BEEN ADJUSTED FOR.
MR. BOGER: ALL RIGHT, YQUR HONOR. I THINK WE CAN DO
THAT. BUT PERHAPS PROFESSOR BALDUS. IF PROFESSOR BALDUS
TESTIFIES AS TO WHAT IT’S BEEN ADJUSTED FOR AND THATS. YOU MADE
THOSE TABLES. DIDN’T YOU, PROFESSOR BALDUS?
V
e
0
SN
0
N
"
M
THE WITNESS: YES. WE CAN PROVIDE A COPY OF THE
10 COMPUTER PRINTOUT. THAT WOULD BE ~-
11 THE COURT: ALL I WANT IS A LIST OF THE 735 MITIGATING
} FACTORS THAT YOU ADJUSTED FOR.
13 MR. BOGER: YOUR HONOR. ONE OTHER THING THAT OCCURS.
14 AND WE CERTAINLY CAN DO THAT AS WELL, BUT IN EVIDENCE ALREADY IS
15 THE CODE BOOK WHICH LISTS ALL THE VARIABLES. FIRST IN
146 COMPUTERESE, AND THEN TRANSLATED INTO GOOD SOLID ENGLISH.
17 THE DOCUMENT THAT IS IN THE APPENDIX AT THIS POINT IS
18 IN THE CODE. THE CODE IS TRANSLATABLE. FOR THE CONVENIENCE OF
1? THE COURT WE CAN MAKE THE TRANSLATION, IF YOU WOULD. BUT --
20 THE COURT: I’VE GOT MORE THINGS TO DO THAN SIT HERE ON
21 THE BENCH AND TRANSLATE 75 THINGS BEFORE I MAKE A RULING ON THE
22 ADMISSIBILITY OF EVIDENCE. MR. BOGER.
23 MR. BOGER: ALL RIGHT, YOUR HONOR. WE CAN CERTAINLY
24 COMPLY WITH THAT REQUEST.
25 BY MR. BOGER:
797
BALDUS - DIRECT
1 @. PROFESSOR BALDUS, CAN YOU GIVE US A, I UNDERSTAND THE JUDGE
2 IS GOING TO RESERVE RULING ON DB-79 BUT WE MIGHT SIMPLY WANT AS
3 WE CONTINUE THIS AFTERNOON. WITH YOUR HONOR-‘S PERMISSION, TO GET
A SENSE FOR THE KINDS OF VARIABLES THAT WERE INCLUDED, OR WOULD
YOU HONOR PREFER SIMPLY TO DEFER THAT WHOLE —-
4
3
MS 1) THE COURT: I THINK I“VE PROBABLY GOT A ROUGH SENSE
7 FROM LOOKING AT THE QUESTIONNAIRE. IT“S THE MORE ESOTERIC ONES
8 THAT 1 WOULD BE MORE INTERESTED IN THAN THE OBVIOUS ONES.
9 WHY DON-T YOU JUST GO AHEAD AND LET HIM TESTIFY ABOUT
10 THE DOCUMENT. AND --
131 MR. BOGER: ALL RIGHT. YOUR HONOR.
12 THE COURT: -- I WILL ADMIT IT UNDER THE PREVIOUS
13 RESTRICTIONS AND NOT ADMIT IT FOR THE TRUTH OF MATTERS ASSERTED
14 THERE UNTIL I GET FURTHER COMPLETE EXPLANATION OF THE 75
15 VARIABLES.
16 MR. BOGER: ALL RIGHT. YOUR HONOR. WE-“LL ABIDE BY THAT
17 RULING.
13 BY MR. BOGER?
- 12 @. PROFESSOR BALDUS, CAN YOU TESTIFY NOW ABOUT DB~7%, WHAT IT
20 REFLECTS?
21 A. VES. WHAT IT REFLECTS ARE THE CO-EFFICIENTS THAT WERE
22 ESTIMATED IN THESE TWO REGRESSION ANALYSES.
23 THESE CO-EFFICIENTS REFLECT THE RACE OF VICTIM AND RACE
24 OF DEFENDANT EFFECTS. AFTER SIMULTANEOUS ADJUSTMENT FOR THE
23 STATUTORY AGGRAVATING AND MITIGATING FACTORS, AS WELL AS THIS
BALDUS - DIRECT
ADDITIONAL OROUP OF SOME 70 AGGRAVATING, FARDON ME. ADDITIONAL
MITIGATING FACTORS.
AND THE FIRST ROW OF THIS TABLE SHOWS A RACE
CO-EFFICIENT FOR RACE OF VICTIM OF .10 WHICH INDICATES THAT ON
THE AVERAGE, DEFENDANTS WITH WHITE VICTIMS HAVE A TEN PERCENTAGE
POINT HIGHER CHANCE OF RECEIVING THE DEATH SENTENCE AFTER
ADJUSTMENT FOR ALL OF THESE BACKGROUND FACTORS.
THE LOGISTIC REGRESSION RESULTS SIMILARLY SHOW A STRONG
RACE OF VICTIM EFFECT. SUGGESTING THAT AFTER ADJUSTMENT FOR
THESE FACTORS. THAT THE ODDS OF RECEIVING THE DEATH SENTENCE ARE
INCREASED BY A FACTOR OF EIGHT, IF THE VICTIM IS WHITE.
THE COURT: LET ME ASK YOU FOR YOUR VIEWS AS A SOCIAL
SCIENTIST. YOU HAVE HERETOFORE BEEN REDUCING THE DISPARITY AS
YOU ADDED FACTORS THAT YOU CONTROLLED FOR. AND NOW ALL OF A
SUDDEN WE HAVE AN INCREASE, AT LEAST FROM DB-78 WE HAVE AN
INCREASE.
HOW AS A SOCIAL SCIENTIST WOULD YOU INTERPRET WHAT'S
GOING ON THERE. WHY DOES THIS ADDITION OF THIS VARIABLE CAUSE
AN INCREASE? I“M NOT ASKING STATISTICALLY WHY, IM JUST ASKING
YOU TO INTERPRET THIS, AS A SOCIAL SCIENTIST.
THE WITNESS: WELL. I WOULD SAY WE WOULD BE. WE’RE
GETTING A BETTER PICTURE OF HOW THE SYSTEM OPERATES BY VIRTUE OF
YOUR BACKGROUND CONTROL FOR MORE OF THE FACTORS. THE MORE YOU
INCLUDE, THE BETTER PICTURE OF HOW IT OPERATES. THAT’S WHY WE
TRY TO BUILD INCREASINGLY LARGE ANALYSES. TO GET A BETTER
OO
W
O
N
0
S
N
Pu
ry
799
BALDUS - DIRECT
PICTURE. TO CAPTURE THE EFFECTS OF THE SYSTEM AS REFLECTED IN
ALL THE VARIABLES.
THE COURT: WELL, LET ME TELL YOU A HYPOTHESIS THAT
CAME TO MY MIND. AND PUTTING ASIDE THE FACT THAT YOU‘RE AN
EXPERT FOR AN ADVOCATE, RESPOND TO IT. : :
THIS WOULD BE AN OPPORTUNITY TO SEE WHETHER THE
DECISION MAKERS. AND WHOM WE‘RE ASSUMING BY AND LARGE ARE WHITE,
TO SHOW LEVELS OF FORGIVENESS IF THERE WERE REASONS TO FORGIVE,
AND IN FACT,» BY EXACERBATING IT. IT WOULD SUGGEST THAT THEY‘RE
LESS LIKELY TO FORGIVE THE KILLER, EVEN WHERE THERE 15 REASON.
NOW. IS THAT OVERDOING YOUR DATA?
THE WITNESS: WELL, I DON’T KNOW THAT THESE RESULTS
SPEAK DIRECTLY TO THAT QUESTION. YOUR HONOR. I WOULD SAY THAT
THE ANALYSES GENERALLY SHOW THE MITIGATING FACTORS DO NOT HAVE A
MAJOR IMPACT IN THE SYSTEM. IN THE SENSE THAT THEY ARE NOWHERE
COMPARABLE TO THE EFFECT THAT THE AGGRAVATING FACTORS DO.
BUT WHAT IT DOES SHOW IS THAT THE BACKGROUND CONTROL
FOR THESE MITIGATING FACTORS DOES NOT DIMINISH THE RACIAL
EFFECTS. AND AS YOU SUGGESTED. THEY TEND TO BE ENHANCED WHEN
ONE MAKES BACKGROUND CONTROLS FOR THOSE FACTORS.
I ALSO ADD THAT THIS IS A FAIRLY UNUSUAL RESULT IN THE
SENSE THAT IN MOST DATA ANALYSES THAT IVE DONE HERETOFORE, WHEN
ONE INCREASES THE NUMBER OF BACKGROUND VARIABLES IN AN ANALYSIS
SIGNIFICANTLY. THAT THE USUAL CONSEGUENCE IS TO REDUCE THE
EFFECT OF ALL THE VARIABLES IN THE ANALYSIS.
800
BALDUS - DIRECT
1 THE COURT: WELL, THAT’S KIND OF WHAT IM SAYING.
2 THE WITNESS: THAT PROVIDES A FURTHER BASIS FOR OUR
BELIEF THAT WHAT YOU’RE SEEING HERE IS A REAL EFFECT BECAUSE A
0
$n
NUMBER OF THESE MITIGATING FACTORS. AND VIRTUALLY ALL OF THE,
FIVE I BELIEVE, MORE THAN HALF OF THE AGGRAVATING FACTORS HERE. a
é STATUTORY AGGRAVATING FACTORS. REFLECT A FAIRLY SUBSTANTIAL
Y CORRELATION WITH THE RACE OF THE VICTIM, SO THAT TO SEE THE
8 PERSISTENCE OF THE RACE OF THE VICTIM EFFECT NOT ONLY IN TERMS
2 OF MAGNITUDE OF THE CO-EFFICIENT. BUT ALSO IN TERMS OF THE LEVEL
0 OF SIGNIFICANCE THAT IS APPARENT, THAT THIS IS SIGNIFICANT
11 EVIDENCE OF THE FACT THAT THESE OTHER FACTORS ARE NOT, DO NOT
12 PROVIDE AN ADEQUATE EXPLANATION FOR THESE RACE OF VICTIM
13 EFFECTS.
14 BY MR. BOGER:
15 @. PERHAPS WE CAN ILLUSTRATE THAT POINT. PROFESSOR BALDUS. BY
16 LOOKING AT DB-80 MARKED FOR IDENTIFICATION. I WAS GOING TO ASK
17 YOU HOW THE RACE OF DEFENDANT AND VICTIM MEASURES VARIED.
1a DEPENDING UPON THE DIFFERENT KINDS OF CONTROLS THAT YOU FLACED
8 1? ON THEM AS BACKGROUND?
20 A. YES. DB-80 IS TABLE 31 FROM OUR REPORT. AND IT PRESENTS
21 RACE OF VICTIM AND RACE OF DEFENDANT DISPARITIES, IN DEATH
22 SENTENCING RATES. AFTER ADJUSTMENTS FOR DIFFERENT SETS OF
23 BACKGROUND FACTORS.
24 AND IT SHOWS IN FIRST PANEL AT THE TOP OF THE PAGE,
23 WHICH UNFORTUNATELY. IT SHOULD BE LABELED 1 AND IT HAS A ROMAN
201
BALDUS - DIRECT
1 NUMERAL TWD, TYPE THERE, IT INDICATES THAT BEFORE ADJUSTMENT FOR
2 ANY BACKGROUND FACTORS, WE HAVE A TEN POINT RACE OF VICTIM
EFFECT ESTIMATED WITH LEAST SQUARES REGRESSION ANALYSIS AND A
MINUS .023 RACE OF DEFENDANT EFFECT. THESE ARE THE VERY SAME
FIGURES WE SAW EARLIER IN THE ANALYSIS. THIS IS JUST A
3
4
=
= & RECAPITULATION OF THAT ANALYSIS.
7 WE SEE IN THE SECOND COLUMN HOW THOSE RACE EFFECTS ARE
8 ENHANCED WHEN WE INTRODUCE CONTROLS FOR A SINGLE RACIAL FACTOR,
? THAT IS. WHEN WE ESTIMATE THE RACE OF VICTIM EFFECTS,
10 CONTROLLING FOR DEFENDANT, AND VICE VERSA. WE SEE THAT BOTH OF
11 THOSE EFFECTS BECOME ENHANCED.
12 THEN THE NEXT COLUMN SHOWS THE EFFECTS THAT ARE
13 APPARENT AFTER SIMULTANEOUS CONTROLS FOR NINE BACKGROUND
134 FACTORS, AND THOSE NINE BACKGROUND BACKGROUND FACTORS ARE FELONY
15 CIRCUMSTANCES VARIABLES. SERIOUS PRIOR RECORD, THE PRESENCE OF A
146 FAMILY-LOVER-LIQUOR-BAR ROOM QUARREL, MULTIFLE VICTIMS, WHETHER
17 THE VICTIM WAS A STRANGER. WHETHER THE DEFENDANT WAS A TRIGGER
18 MAN, PLUS THREE VARIABLES WHICH CAPTURE THE ELEMENTS OF
1? GRATUITOUS VIDLENCE IN THE CASES. AND AS YOU CAN SEE FROM THE
20 CO~-EFFICIENT ~-
21 R. WHAT ARE THOSE THREE?
28 A. OH, THOSE THREE ARE PHYSICAL TORTURE. MENTAL TORTURE. AND A
23 SERIOUS AGGRAVATING CIRCUMSTANCE ACCOMPANYING A CONTEMPORANEOUS
24 OFFENSE.
235 AS YOU CAN SEE FROM THIS THIRD COLUMN THAT THE
802
BALDUS - DIRECT
1 PRESENCE OF THOSE NINE VARIABLES AS BACKGROUND FACTORS IN THE
2 ANALYSIS TEND TO SUPPRESS THE MAGNITUDE OF BOTH THE RACE OF
VICTIM AND RACE OF DEFENDANT EFFECTS.
INDEED, THE RACE OF DEFENDANT EFFECTS IN THE FACE OF
THOSE NINE VARIABLES. LOSES STATISTICAL SIGNIFICANCE IN THE
3
4
=
- & LEAST SQUARES REGRESSION ANALYSIS.
7 THE RACE OF VICTIM EFFECT, HOWEVER, REMAINS STRONG AND
8 SUBSTANTIAL IN MAGNITUDE. AND HIGHLY SIGNIFICANT FROM A
? STATISTICAL STANDPOINT.
10 AFTER CONCLUDING THIS ANALYSIS WE UNDERTOOK AN ANALYSIS
11 WHICH INTRODUCED SIMULTANECUS BACKGROUND CONTROLS FOR OVER 230
12 VARIABLES. BASICALLY. WE TOOK THE NON-RACIAL FACTORS THAT WERE
13 LISTED IN THE TABLE THAT WE EXAMINED EARLIER THIS MORNING, WHICH
14 1S, THAT LISTED THE FREQUENCY DISTRIBUTION OF THE VARIABLES
18 USED FOR THE COMPUTER ANALYSIS. THAT IS TABLE. TABLE DD, AND -——
146 RQ. IS THAT DB-&07?
17 A. EXCUSE ME. EXHIBIT DB-60.
18 THE COURT: HAVE I SEEN THAT?
» 19 THE WITNESS: YES, YOUR HONOR, THAT WAS THE ONE THAT WE
20 LOOKED AT THIS MORNING THAT SHOWED THE DISTRIBUTION ON KEY
23 VARIABLES THAT WE WERE GOING TO EXAMINE TO SEE IF THEY REDUCED
22 THE RACE EFFECTS. WE LODKED AT PAGE 4, THE NUMBER OF
23 CONVICTIONS FOR VIOLENT PERSONAL CRIMES.
24 BY MR. BOGER:
23 0. AND YOUR TESTIMONY IS THAT THIS LAST ANALYSIS ON THE
a
vw
0
M
N
D
10
11
803
BALDUS - DIRECT
RIGHT-HAND COLUMN OF DB-80 CONTROLLED SIMULTANEOUSLY FOR ALL OF
THE FACTORS SET FORTH IN DB-4607
A. NOT ALL THE FACTORS. A FEW WERE OMITTED BECAUSE OF A HIGH
INTER-CORRELATION BETWEEN THEM. BUT IT EMBRACES THE SUBSTANTIAL
BODY OF THOSE VARIABLES, OVER 250. THAT LISTING HERE, I THINK,
INCLUDES 230.
WHEN YOU RUN AN ANALYSIS OF THIS MAGNITUDE. YOUR HONOR,
THE MACHINE PROCEDURES WILL NOT OPERATE CORRECTLY IF THERE IS
TOO HIGH CORRELATION BETWEEN SOME OF THE VARIABLES. THAT’S WHAT
PRODUCED THE REDUCTION IN THOSE VARIABLES.
AND WHAT YOU SEE THERE IS THAT AFTER SIMULTANEOUSLY
CONTROLLING FOR THE BACKGROUND EFFECTS OF ALL OF THOSE
VARIABLES. THAT WE SEE A FURTHER REDUCTION IN THE RACE OF VICTIM
MEASURE OF ONLY ONE PERCENTAGE POINT. STILL PERSISTS AT .06,
SUGGESTING EVEN IN THE FACE OF BACKGROUND CONTROL FOR ALL THESE
FACTORS THAT THE WHITE VICTIM CASES CONTINUE TO SHOW A HIGHER
DEATH SENTENCING RATE.
AND I WANT TO EMPHASIZE AGAIN THIS IS THE AVERAGE RATE
ACROSS ALL THE CASES. THAT THERE ARE NO EFFECTS OF, AMONG THE
LEAST AGGRAVATED CASES. WHAT THESE EFFECTS YOU SEE MANIFESTED
IN THIS CO-EFFICIENT REFLECT ARE THE IMPACTS WHICH ARE LARGER
THAN THIS AMONG THE AREAS WHERE THE RACE OF VICTIM IS HAVING ITS
EFFELT.
AND ALSO IT IS INDEED SURPRISING THAT THE LEAST SQUARES
REGRESSION CO-EFFICIENT FOR THE RACE OF THE DEFENDANT INCREASES
804
BALDUS - DIRECT
1 IN MAGNITUDE AND INCREASES IN STATISTICAL SIGNIFICANCE AS WE
2 CONTROL FOR ADDITIONAL BACKGROUND FACTORS.
AND I CAN SAY THAT ON THE BASIS OF MY EXPERIENCE. THAT
ITS QUITE UNUSUAL TO SEE AN EVENT LIKE THAT, THIS ESPECIALLY IN
3
4
5 THE FACE OF THE FACT THAT AMONG THESE 230 BACKGROUND VARIABLES
& THAT WE CONTROLLED FOR. THERES SOMETHING ON THE ORDER OF OVER A
7 HUNDRED OF THEM THAT SHOWED A STATISTICALLY SIGNIFICANT
8 CORRELATION WITH BOTH THE RACE OF THE VICTIM AND THE DEATH
? SENTENCING OUTCOME.
10 QR. WHAT'S THE SIGNIFICANCE OF THAT. PROFESSOR BALDUS?
11 A. WELL. THE SIGNIFICANCE OF THAT IS, AS I INDICATED TO YOU
12 EARLIER, YOUR HONOR, THE KIND OF VARIABLE THAT WOULD HAVE THE
13 CAPACITY TO EXPLAIN AWAY THE RACE OF VICTIM EFFECT IS A VARIABLE
14 THAT SHOWS A CORRELATION WITH BOTH THE DEATH SENTENCING OUTCOME
15 AND THE RACE OF VICTIM.
1& WHEN YOU INTRODUCE THAT AS A BACKGROUND CONTROL. THAT
17 WILL TEND TO SUPPRESS THE MAGNITUDE OF THE RACIAL VARIABLE. WE
13 SAW THAT WHEN WE DID A BACKGROUND CONTROL FOR THE FELONY
® 19 CIRCUMSTANCES VARIABLE. YOU RECALL THERE THAT VARIABLE ITSELF
20 HAD A SIGNIFICANT SUPPRESSANT EFFECT ON THE RACE OF VICTIM
21 CO-EFFICIENT.
22 THE REASON FOR THAT IS THAT THE FELONY CIRCUMSTANCE
23 VARIABLE IS STRONGLY CORRELATED WITH RACE OF VICTIM AND IT IS
24 ALSO STRONGLY CORRELATED WITH THE DEATH SENTENCING OUTCOME.
23 MOREOVER, AND THIS IS WHY I THINK THESE LARGE SCALE
805
BALDUS - DIRECT
oN
REGRESSIONS HAVE SOME SIGNIFICANCE. IS THAT THERE WERE A NUMBER
OF, A VERY SUBSTANTIAL NUMBER OF THE VARIABLES ON THIS LIST THAT
WERE HIGHLY CORRELATED WITH BOTH THE RACE OF THE VICTIM AND THE
RACE, PARDON ME, THE RACE OF THE VICTIM AND THE DEATH SENTENCING
OUTCOME. EVEN WHEN ALL THOSE ARE INTRODUCED AND THE CONTROLS
ARE MADE, BACKGROUND CONTROLS ARE MADE FOR ALL THOSE FACTORS,
YOU STILL SEE THIS RACE OF VICTIM EFFECT.
AND THAT'S WHAT WE CALL A PERSISTENT EFFECT. IT
W
O
0
A
Hs
W
N
CONTINUES IN SPITE OF THIS BACKGROUND CONTROL PROCESS FOR A VERY
po
i oO
SUBSTANTIAL NUMBER OF VARIABLES.
11 AND SIMILARLY THE NUMBER OF THE VARIABLES IN THIS
12 ANALYSIS. THE NUMBER IS NOT LARGE, I THINK IT”S FIFTY OR SIXTY
13 VARIABLES. PERHAPS, THAT ARE, HAVE A STATISTICALLY SIGNIFICANT
14 CORRELATION WITH BOTH THE RACE OF DEFENDANT AND THE DEATH
13 SENTENCE OUTCOME. THOSE WOULD BE THE GOOD CANDIDATES TO
16 SUPPRESS THE MAGNITUDE AND STATISTICAL SIGNIFICANCE OF THE RACE
17 OF DEFENDANT CO-EFFICIENT.
18 AGAIN. EVEN WITH ALL THOSE FACTORS IN THE MODELS. AND
A 19 THE ANALYSIS. WE FIND THAT THERE“S STILL A RACE OF DEFENDANT
20 CO-EFFICIENT THAT IS SUBSTANTIAL IN SIZE. AND HAS STATISTICAL
21 SIGNIFICANCE.
22 THE RESULTS OF THE LOGISTIC REGRESSION ANALYSIS ARE
23 CONSIST WITH THIS, THE ANALYSIS FROM THE LEAST SQUARES. THEY
24 SHOW -——
293 THE COURT: ARE YOU DDWN ON PANEL NUMBER 27
306
BALDUS - DIRECT
THE WITNESS: YES.
THE COURT: IS THIS WITH THE 230-PLUS VARIABLES?
THE WITNESS: NO, YOUR HONCR, ITS NOT GIVEN THE STATE
OF THE ART OF COMPUTER ANALYSIS AT THIS STAGE TO DO A LOGISTIC
ANALYSIS WITH THAT MANY VARIABLES CONTROLLED FOR SIMULTANEDUSLY.
IN THE PAGES TO COME, WELL DEMONSTRATE ONE THAT HAS
SOME 39 BACKGROUND VARIABLES CONTROLLED FOR, BUT LOGISTICS
SOFTWARE, THATS THE PROGRAMS THAT RUN THE ANALYSIS, CANNOT
HANDLE VARIABLES IN THAT QUANTITY IN AN ANALYSIS.
BUT WE SEE HERE WITH RESPECT TO THESE OTHER ANALYSES
THAT AGAIN THE RACE OF THE VICTIM DISPARITIES ARE STRONG WITH
THIS ALTERNATIVE METHOD OF ANALYSIS. BUT THIS PATTERN WE SAW
EARLIER AGAIN IS THAT THE RACE OF THE DEFENDANT DISPARITY
OVERALL, IN THOSE DISPARITIES ARE REFLECTING THE OVERALL IMPACT
OF THE RACE OF THE DEFENDANT IN THE SYSTEM, TEND TO BE SMALL AND
THEY LOST SIGNIFICANCE, AGAIN DEMONSTRATING THE LACK OF
STABILITY OF THAT RACE OF THE DEFENDANT EFFECT.
THE COURT: THESE CO-EFFICIENTS AND MULTIPLIERS APFLY
TO WHICH OF THE FOUR COLUMNS OF PANEL 17 |
THE WITNESS: OH, THE TWO MULTIPLIERS HERE RELATE TO |
THE SECOND AND THIRD COLUMNS. THAT IS, AFTER ADJUSTMENT FOR |
OTHER RACIAL VARIABLES, THOSE ARE THE RACE OF VICTIM MULTIPLIER, |
12.5 AND 3 FOR THE RACE OF THE DEFENDANT. AND THE THIRD COLUMN
INDICATES A MULTIPLIER OF 3.0, FOR THE RACE OF THE VICTIM. AFTER
A BACKGROUND CONTROL FOR NINE FACTORS.
a
Hr
WW
ON
9
.
0
NN
-
>
307
BALDUS - DIRECT
THE COURT: OKAY. IT REALLY RELATES TO COLUMN 3.
THE WITNESS: YES, YOUR HONOR, THAT'S RIGHT.
BY MR. BOGER:
@. PROFESSOR BALDUS, WHAT WAS YOUR RATIONALE FOR DEVELOPING THIS
ANALYSIS BASED ON NINE BACKGROUND VARIABLES?
A. OH, THE RATIONALE WAS TO IDENTIFY THE FACTORS THAT MY
EXPERIENCE IN WORKING - WITH THESE DATA, AND KNOWLEDGE OF THE
LITERATURE GENERALLY SUGGESTED WERE IMPORTANT FACTORS, TO SEE
WHAT THE IMPORTANT VARIABLES WERE IN THE SYSTEM AND TO SEE WHAT
ARE THE FACTORS THAT WOULD BE MOST LIKELY TO SUPPRESS THE
MAGNITUDE OF THE RACIAL CO-EFFICIENTS.
THE COURT: TELL ME WHAT THEY ARE, AGAIN.
THE WITNESS: OH, HERE, YOUR HONOR, I CAN HAND THIS UP
TO YOU, YOUR HONOR, IF YOU WOULD LIKE TO SEE THE LISTING RIGHT
THERE.
THE COURT: WHAT PROFESSOR BALDUS HAS HANDED ME. MR.
BOGER. IS TABLE 22A. IT IS INTERESTING, PROBABLY OUGHT TO BE
APPENDED TO DB-80.
MR. BOGER: YOUR HONOR, WE’LL BE HAPPY TO MAKE COPIES
OF THAT PARTICULAR DOCUMENT AND PROVIDE IT FOR THE COURT
TOMORROW OR WE CAN SUBMIT IT AT THIS POINT.
THE COURT: NO, I THINK JUST ANYTIME TOMORROW WILL BE
FINE.
MR. BOGER: FINE.
BY MR. BOGER:
808
BALDUS - DIRECT
.
@. PROFESSOR BALDUS, DID YOU IN THE CONDUCT OF YOUR ANALYSIS
3)
GROUP TOGETHER VARIOUS GROUPS OF VARIABLES TO SEE WHETHER WHEN
3 CONTROLLED FOR THEY COULD HAVE AN IMPACT IN SUPPRESSING THE
4 RACIAL FACTORS THAT YOU HAD IDENTIFIED?
3 A. YES, WE DID A WIDE VARIETY OF ANALYSES AND ONE OF THE GOALS
i & OF THIS APPROACH WAS TO DEVELOP AN ANALYSIS THAT WOULD EMBODY
7 THE FACTORS THAT APPEARED MOST IMPORTANT IN THE SYSTEM. AND WHAT
8 I MEAN BY MOST IMPORTANT IN THE SYSTEM, OR THE MEASURE OF
7? IMPORTANCE IN THE SYSTEM WAS STATISTICAL IMPORTANCE, BUT ALSO
10 IMPORTANCE ON A PRIORI STANDPOINT. FACTORS THAT COURTS. SCHOLARS
11 APPEAR TO BELIEVE WERE THE MOST IMPORTANT FACTORS IN THE SYSTEM.
12 AND ON THE BASIS OF THOSE CRITERIA, I SELECTED SOME 29
13 VARIABLES WHICH I THOUGHT CAPTURED THE ESSENCE OF THIS SYSTEM IN
14 TERMS OF THE MAIN DETERMINANTS OF THE PROCESS.
15 @. COULD YOU VIEW THOSE, PROFESSOR BALDUS. AS A GROUP OF
1&6 CONTROLS THAT TOGETHER MIGHT FORM A RIVAL HYPOTHESIS TO EXPLAIN
17 THE RACIAL VARIABLES? IS THAT -—-
1 A. YES. IN MY OPINION, THOSE VARIABLES CONSTITUTED THE MOST
19 PLAUSIBLE RIVAL HYPOTHESES. MOST PLAUSIBLE IN THE SENSE THAT
20 THEY WERE STATISTICALLY IMPORTANT AND ALSO THEY WERE
21 IMPORTANT IN TERMS OF OUR GENERAL UNDERSTANDING OF HOW SYSTEMS
22 OF THIS TYPE OPERATE. IN DETERMINING DEATH WORTHINESS OF
23 OFFENDERS.
24 @. BEFORE WE GET TO YOUR HYPOTHESIS, IF YOU WOULD, YOUR
23 ANALYSIS THAT INVOLVED 39 FACTORS. ARE THE NINE BACKGROUND
809
BALDUS ~ DIRECT
1 VARIABLE ANALYSIS THAT YOUVE RUN IN DB-80, AND THE Z30-PLUS
2 VARIABLE ANALYSIS THAT YOUVE RUN IN DB-80, ALSO RIVAL
3 HYPOTHESES. THAT THROUGH THIS METHOD YOU HAVE TESTED TO SEE
4 WHETHER THEY HAVE RACIAL EFFECTS?
3S A. YES.
pe & @. OR, EXCUSE ME, SUPPRESS THE RACIAL EFFECT?
7 A. YES, THAT’S THE PURPOSE OF EACH OF THESE ANALYSES. TO
3 DETERMINE WHETHER RACIAL EFFECTS PERSIST, AFTER BACKGROUND
ADJUSTMENT FOR THESE VARIABLES. AND THESE BACKGROUND ADJUSTMENTS
10 ARE DESIGNED TO TEST THE PLAUSIBILITY OF THESE RIVAL HYPOTHESES.
11 AND WHEN IN AN ANALYSIS OF THESE RIVAL HYPOTHESES. THAT
12 THESE ARE THE FACTORS WE TEST THE PLAUSIBILITY OF THESE RIVAL
13 HYPOTHESES. THAT THESE ARE THE FACTORS THAT WILL, ARE CAUSING
14 THE EFFECT. NOT THE RACIAL COMPONENTS OF THE CASES. AND WE TEST
15 THEM IN THIS WAY. AND WE SEE THE RACE EFFECTS PERSIST. THAT
16 PROVIDES EVIDENCE THAT THOSE ARE REAL EFFECTS IN THE SYSTEM.
17 PARTICULARLY AS I INDICATED BEFORE, BECAUSE THE OTHER BACKGROUND
18 FACTORS THAT ARE CONTROLLED FOR IN THESE ANALYSES ARE SO
» 19 STRONGLY CORRELATED WITH BOTH THE RACE OF THE VICTIM AND THE
20 DEATH SENTENCING OQUTCOME.
21 THE PERSISTENCE, PARTICULARLY THE RACE OF THE VICTIM
22 CO-EFFICIENT IN THE FACE OF THIS SORT OF ANALYSIS, THAT LEADS
232 ONE TO BELIEVE THAT THOSE OTHER FACTORS ARE NOT THE CAUSE, DO
24 NOT EXFLAIN THE RACE OF VICTIM EFFECT THAT WE SEE.
23 MR. BOGER: YOUR HONCR. AT THIS TIME I MOVE THE
810
BALDUS ~ DIRECT
Te
s ADMISSION OF DB-80 INTO EVIDENCE, WITH THE UNDERSTANDING WE-LL
APPEND TO IT BY TOMORROW MORNING TABLE 22A, FROM THE FINAL y)
REPORT.
THE COURT: MS. WESTMORELAND?
MS. WESTMORELAND: YOUR HONOR, I WOULD HAVE THE
SAME OBJECTION AS PREVIOUSLY NOTED. AND ALSO IT WOULD APPEAR
THAT THE ONLY COLUMN THAT PERHAPS WOULD HAVE ANY RELEVANCE AT
ALL, WOULD BE THE FOURTH COLUMN RELATING TO THE CONTROL FOR OVER
230 NON-RACIAL FACTORS. I DON’T SEE THE REST OF THE COLUMNS IN
Pa
y G
O
E
L
BE
E
S
R
S
L
Se
e
C
B
T
Se
THE FIRST PART OF THE TABLE HAVE ANY RELEVANCE. OTHER THAN THAT
11 AND THE GENERAL OBJECTION THAT HAS BEEN MADE.
12 MR. BOGER: YOUR HONOR, HERE’S WHERE I WANT TO RENEW MY
12 ARGUMENT .
14 : THE COURT: I“M GOING TO LET IT IN, BUT LET ME SAY
135 THIS. I1“M GOING TO LET THE THIRD AND FOURTH COLUMN IN FOR
16 EVIDENCE OF WHAT THEY CLAIM TO ESTABLISH.
17 I THINK WHAT MS. WESTMORELAND MAY BE SAYING IN A WAY.
13 AND IT’S SOMETHING SITTING HERE FEELING RIGHT NOW, AND THAT IS
19 PROFESSOR BALDUS HAS NOT YET ARRANGED ANY SET OF FACTORS THAT I
20 WOULD HAVE PERSONALLY PICKED, AND MAY BE THATS WHAT YOURE
21 SAYING, THAT’S KIND OF THE WAY I“VE BEEN SITTING HERE FEELING
22 FOR A FEW MINUTES. BUT THE POINT I THINK HE WAS JUST MAKING
23 ABOUT PERSISTENCE USING DIFFERENT COMBINATIONS. STARTS GIVING IT
24 SOMETHING APPROACHING AT LEAST ADMISSIBLE EVIDENCE.
235 THE REST OF IT GOES TO THE WEIGHT, I THINK. S0 I WILL
S811
BALDUS - DIRECT
ALLOW IT. THIS IS NOT TO SAY THAT IT“S THE 9 FACTORS THAT I
WOULD HAVE PICKED OR THAT I THINK THOSE NINE FACTORS ARE
CONCLUSIVE. I THINK WHAT HE WAS JUST SAYING ABOUT THE
PERSISTENCE, USING DIFFERENT COMBINATIONS OF FACTORS WHICH
OBVIOUSLY WE ALL AGREE DO ENTER INTO THE DECISION START MAKING
IT AT LEAST SOME EVIDENCE OF THE PERSISTENCE OF DECISION MAKING
IN PART INFLUENCED BY THE RACE OF THE VICTIM.
MR. BOGER: THANK YOU, YOUR HONOR.
BY MR. MR. BOGER?
@. PROFESSOR BALDUS., YOU HAD MENTIONED THIS 39 VARIABLE MODEL
THAT IN YOUR JUDGMENT REPRESENTED PERHAPS THE BEST ANALYSIS OF
WHATS GOING ON IN THE SYSTEM.
LET ME DIRECT YOUR ATTENTION TO DB-281 AND ASK YOU IF
YOU CAN IDENTIFY THAT DOCUMENT?
A. YES, DB~81 IS THE RESULT OF AN ANALYSIS WHICH INCLUDES THE
RACE OF VICTIM, RACE OF DEFENDANT, VARIABLES. ALONG WITH TWENTY
OF THAT ORIGINAL 23% THAT MAINTAINED STATISTICAL SIGNIFICANCE IN
A WEIGHTED LEAST SQUARES REGRESSION ANALYSIS.
@. EXPLAIN WHAT YOU MEAN BY THAT?
A. THAT IS, YOU STARTED OUT WITH 39 VARIABLES. AND A NUMBER OF
THOSE VARIABLES FAILED TO DEMONSTRATE A STATISTICALLY
SIGNIFICANT RELATIONSHIP WITH THE DEATH SENTENCING OUTCOME.
WHAT THAT MEANS IS THAT THEY PROVIDE RELATIVELY LITTLE
EXPLANATORY POWER IN THE SYSTEM. THEY DON‘T EXPLAIN VERY MUCH
AT ALL AS TO WHY CERTAIN PEOPLE GET A DEATH SENTENCE AND WHY
BALDUS ~- DIRECT
1 THEY DON‘T.
= IT INDICATES THAT THE VARIABLES THAT DISCRIMINATE ON A
QUESTION OF WHO IS SENTENCED TO DEATH AND WHO IS NOT SENTENCED
TO DEATH ARE EMBODIED IN THE VARIABLES THAT SHOW THE STRONG
STATISTICAL SIGNIFICANCE.
3
4
3
» 6 IN AN EFFORT TO MAKE A MORE PARSIMONIDUS,
7 UNDERSTANDABLE MODEL OF THE SYSTEM. OR AN ANALYSIS OF THE
8 SYSTEM. I RESTRICTED THEM TO THOSE VARIABLES THAT WERE
@ STATISTICALLY SIGNIFICANT.
10 IN ADDITION, IT WAS MY OBJECTIVE IN PREPARING THIS
11 ANALYSIS TO GIVE ONE A SENSE OF THE RELATIVE IMPORTANCE OF THESE
12 RACIAL EFFECTS. VIS-A-VIS OTHER FACTORS THAT ARE OPERATING IN
13 THE SYSTEM. AND TO DO THAT. I SORTED THE CASES. PARDON ME,
14 SORTED THESE VARIABLES THAT EMERGED FROM THIS ANALYSIS ACCORDING
15 TO THE MAGNITUDE OF THE REGRESSION CO-EFFICIENTS THAT WERE
16 STIMATED FOR EACH VARIABLE. THIS SORTING GIVES ONE A ROUGH,
17 AND I WANT TO EMPHASIZE THAT, A ROUGH SENSE OF THE RELATIVE
18 IMPORTANCE OF THE DIFFERENT VARIABLES IN THE SYSTEM. AND WHEN
RS 19 ONE SCANS THESE VARIABLES YOU CAN SEE THE KIND OF COMPANY THAT
20 THE RACE OF VICTIM VARIABLE KEEPS IN TERMS OF ITS STATISTICAL
21 EFFECT IN THE PROCESS. AND ONE CAN SEE IN LOOKING AT DB-81 THAT
22 IT WAS, WITH RESPECT TO THE RACE OF THE VICTIM. THAT THE
23 STATISTICAL EFFECT IS NOT TRIVIAL. IT’S IN THE SAME ORDER OF
24 MAGNITUDE OF VARIABLES SUCH AS WHETHER THERE WAS A Bl FACTOR,
235 STATUTORY AGGRAVATING FACTOR PRESENT IN THE CASE, WHETHER THE
813
BALDUS - DIRECT
DEFENDANT WAS A PRIME MOVER, WHETHER OR NOT THERE WAS AN L.D. ——
—
2 I MEAN, PARDON ME, A B10 STATUTORY AGGRAVATING FACTOR. AS YOU,
3 AS YD GO DOWN THE TABLE, HOWEVER. YOU BEGIN TO SEE THE TRULY.
4 HIGHLY AGGRAVATED CHARACTERISTICS OF THE CASES WHICH HAVE MUCH
3 HIGHER CO-EFFICIENTS ASSOCIATED WITH THEM. .
p 1 & SO IT LEAVES ONE WITH THE IMPRESSION THAT THE RACE OF
7 J THE VICTIM IN THIS SYSTEM IS CLEARLY NOT THE DETERMINANT OF WHAT
8 HAPPENED, BUT RATHER THAT IT IS A FACTOR LIKE A NUMBER OF OTHER
? FACTORS, THAT IT PLAYS A ROLE AND INFLUENCES DECISION MAKING.
10 THE ONE THING THAT“S, THAT STRUCK ME FROM WORKING WITH
11 THESE DATA FOR SOME TIME, THERE IS NO ONE FACTOR THAT DETERMINES
12 WHAT HAPPENS IN THE SYSTEM. IF THERE WERE, YOU COULD MAKE
13 HIGHLY ACCURATE PREDICTIONS OF WHAT“S GOING TO HAPPEN. THIS IS
14 A SYSTEM THAT IS HIGHLY DISCRETIONARY. HIGHLY COMPLEX. MANY
15 FACTORS ARE AT WORK IN INFLUENCING CHOICE. AND NO ONE FACTOR
14 DOMINATES THE SYSTEM. 1IT“S THE RESULT OF A COMBINATION OF MANY
17 DIFFERENT FACTORS THAT PRODUCE THE RESULTS THAT WE SEE, EACH
18 FACTOR CONTRIBUTING MORE OR LESS INFLUENCE.
pS 19 BUT OUT OF ALL THE FACTORS THAT WE STARTED OUT OUR
20 ANALYSIS WITH, THESE ARE THE FACTORS THAT ARE ON THIS PAGE IN MY
21 OPINION THAT GIVE THE BEST PICTURE. IN AN INTERPRETABLE FORMAT
22 OF WHAT DRIVES THE SYSTEM, IN TERMS OF THE SELECTIONS AS TO WHO
23 RECEIVES THE DEATH SENTENCE AND WHO DOES NOT.
24 YOU CAN SEE HERE THAT WITH RESPECT TO RACE OF
23 DEFENDANT. THAT THE EFFECTS ARE MUCH SMALLER IN TERMS OF TOTAL
— — iy Aredia: 3 —, So—— —— —
BALDUS - DIRECT
pb
EFFECT. AND AS I ALSO INDICATED. BECAUSE OF A LACK OF STABILITY
OF RACE OF DEFENDANT EFFECT I WOULD TEND TO DISCOUNT EVEN THE
SORT OF RANKING THAT’S WITHIN THE CONTEXT OF THIS PARTICULAR
TABLE. EACH ONE OF THESE TABLES, EACH ONE OF THESE SETS OF
CO-EFFICIENTS, AND ANALYSIS HAS TO BE EVALUATED IN THE CONTEXT
OF ALL OF THE ANALYSIS THAT HAS BEEN DONE.
BUT THE RACE OF VICTIM EFFECT HERE IS POTENT IN THIS
TABLE AND IN THE TABLE THAT WERE ABOUT TO EXAMINE, I BELIEVE.
vg
0
w
o
a
a
o
p
o
N
BY MR. BOGER?
10 @. BEFORE WE DO, I WOULD LIKE TO ASK, I NOTICE THAT SOME OF THE
11 CO-EFFICIENTS IN COLUMN B SEEM TO HAVE NEGATIVE NUMBERS AND SOME
12 HAVE POSITIVE NUMBERS.
13 WHAT'S THE SIGNIFICANCE OF THAT AND WHY ARE THE
14 NEGATIVE NUMBERS INCLUDED IN HERE AS WELL?
13 A. 0H, YES. THESE ARE VARIABLES THAT HAVE THE EFFECT OF
16 REDUCING AN OFFENDER-S LIKELIHOOD OF RECEIVING A DEATH SENTENCE.
17 THE COURT: I WAS WONDERING IF LINE 5 AND 13 YOU
18 INTENDED TO HAVE THE MINUS?
s 1° THE WITNESS: YES, I DID. YOUR HONOR. WHAT I WAS
20 TRYING TO DO HERE WAS TO INDICATE THEIR RELATIVE MAGNITUDE OF
21 THE IMPACT.
22 THE COURT: SWAYING ONE WAY OR THE OTHER. NOT
23 NECESSARILY EXACERBATING, BUT JUST PUNCH ONE WAY COR THE OTHER?
24 THE WITNESS: PRECISELY. YOUR HONOR. TO SHOW. FOR
23 EXAMPLE, THE FACTOR A OF A PERSON NOT BEING THE TRIGGER MAN IS
-
J
8
A
BB
B
O
N
10
11
BALDUS ~ DIRECT
AN EXTREMELY IMPORTANT FACTOR IN THE SYSTEM, AND I FELT IT
APPROPRIATE TO LIST IT EVEN THOUGH IT WAS IN THE OPPOSITE
DIRECTION. IN THAT ORDER.
BY MR. BOGER:
@. DID YOU PERFORM THIS ANALYSIS USING THE LOGISTIC ANALYSIS.
OR THE LOGISTIC REGRESSION METHOD AS WELL? DID YOU PERFORM
THIS?
A. YES. I DID.
THE COURT: LET ME ASK YOU ONE OTHER QUESTION BEFORE
YOU LEAVE THERE.
WHERE DID YOU GET THESE 227
THE WITNESS: I STARTED OUT WITH A LIST OF 39, YOUR
HONOR, THAT WAS THE RESULT OF, IN MY JUDGMENT. WHAT I THOUGHT
WERE THE VARIABLES THAT SHOWED A PERSISTENT EFFECT ACROSS ONE
AFTER ANOTHER ANALYSIS. THEY WERE ALWAYS THERE, ALWAYS
IMPORTANT WHENEVER I WAS DOING AN ANALYSIS.
BY MR. BOGER:
G0. DOES THIS INCLUDE YOUR ANALYSIS OF THE PROCEDURAL REFORM
STUDY AS AS WELL AS THE CHARGING AND SENTENCING?
A. OH. YES. IT’S BASED ON MY EXPERIENCE IN DEALING WITH THE
CALIFORNIA DATA. ARKANSAS DATA, THE PROCEDURAL REFORM DATA,
THESE DATA. THESE ARE THE CANDIDATES. THAT IS. THE VARIABLES
THAT WERE PERSISTENT IN THEIR EFFECT.
AND ALSO THEY WERE THE FACTORS THAT WERE LINDERSTANDABLE
IN TERMS OF THEIR EFFECT. THE RESULTS WERE CONSISTENT WITH
E
T
RR
R”
Go
d
Sl
e
1
SR
|
T
e
0
RE
E
N
E
ot
a
ha
l
[|
V)
pa
l
oO
pok
e Ww
14
13
814
BALDUS - DIRECT
GENERAL UNDERSTANDING OF HOW THIS SYSTEM. SYSTEMS LIKE IT,
OPERATE. THAT WAS THE SOURCE OF THAT SELECTION, YOUR HONOR.
@. NOW THE QUESTION, PROFESSOR BALDUS, IS HOW DID YOU GET 22
FROM 397?
A. OH, I BEG YOUR PARDON.
THE COURT: NO. I WANT TO KNOW WHERE HE GOT EITHER.
NOW I UNDERSTAND.
THE WITNESS: YES. NOW THE 39 WERE SELECTED THAT WAY
AND THIS 22 WAS SELECTED BY A STATISTICAL DECISION RULE THAT IF
THE AVAILABLE DID NOT SHOW STATISTICAL SIGNIFICANCE AT THE .10
LEVEL, THEN IT WAS DROPPED FROM THE ANALYSIS, BECAUSE IT WOULD
NOT ADD ANY ADDITIONAL INFORMATION AND WOULD CLEARLY NOT REDUCE
THE EFFECTS.
IN FACT, I HAVE THE ANALYSIS, YOUR HONOR. THAT INCLUDE
ALL 39, AND THE RACIAL CO-EFFICIENTS ARE IDENTICAL. THEY“RE
JUST MORE DIFFICULT TO COMPREHEND.
THE COURT: WHAT DID YOU HOLD CONSTANT IN THESE 227
DID YOU JUST LOOK AT THIS ONE FACTOR?
THE WITNESS: NO, YOUR HONOR. WHAT THESE CO-EFFICIENTS
TELL US IS THE IMPACT OF EACH OF THESE VARIABLES, AFTER HOLDING
CONSTANT ALL OF THE OTHER 21. SO IF YOU START OUT HERE --
THE COURT: ALL RIGHT. ALL RIGHT. I GOT YOU.
'BY MR. BOGER:
@. IF YOU WOULD TURN AND IDENTIFY DB-82, PROFESSOR BALDUS.
A. YES. DB-82 IS A LOGISTIC REGRESSION ANALYSIS. IT COMMENCED
o
~
817
BALDUS - DIRECT
WITH THE SAME ORIGINAL LIST OF VARIABLES PLUS ONE ADDITIONAL
ONE. I BELIEVE.
AND IT ALSO BOILED THE VARIABLES DOWN TO THOSE THAT
WERE STATISTICALLY SIGNIFICANT AT THE .10 LEVEL, AND THE
CO-EFFICIENTS WERE ESTIMATED, AND THE DEATH ODDS MULTIPLIERS ARE
CALCULATED AND I RANKED THEM ACCORDING TO THE MAGNITUDE OF THIS
DEATH ODDS MULTIPLIER, WHICH IS THE STATISTIC THAT IS
INTERPRETABLE AS A RESULT OF THIS REGRESSION ANALYSIS.
AND EVEN THOUGH THIS PROCEDURE OPERATES ON THE BASIS OF
DIFFERENT ASSUMPTIONS. ONE CAN SEE THAT THE VARIABLES THAT
APPEAR IN THIS ANALYSIS ARE SUBSTANTIALLY THE SAME AS THE
VARIABLES THAT APPEARED IN THE ANALYSIS REPORTED IN DB-81. AND
MOREOVER YOU SEE THAT THE RACE OF VICTIM EFFECT IS OF
APPROXIMATELY THE SAME MAGNITUDE VIS-A-VIS THE OTHER VARIABLES
THAT IT WAS IN THE EARLIER ANALYSES.
AND SIMILARLY. THE RACE OF DEFENDANT VARIABLE APPEARS
TO HAVE ABOUT THE SAME EFFECT, ALTHOUGH I WILL NOTE HERE AGAIN,
AS IS SEEN THROUGHOUT THESE ANALYSES, IN THE LOGISTIC ANALYSIS,
THE RACE OF DEFENDANT EFFECT DOES NOT ACHIEVE STATISTICAL
SIGNIFICANCE.
THIS ANALYSIS IS FURTHER EVIDENCE. USING A DIFFERENT
PROCEDURE, OF THE PERSISTENCE OF THE RACE OF VICTIM EFFECT IN
THIS CHARGING AND SENTENCING SYSTEM.
MR BOGER: YOUR HONOR, AT THIS TIME I MOVE THE
ADMISSION OF BOTH DB-81 AND “82 INTO EVIDENCE
813
BALDUS - DIRECT
PY
MS. WESTMORELAND: YOUR HONOR, OUTSIDE THE PREVIOUS
OBJECTIONS, OBVIOUSLY. WE WILL NOT REPEAT THEM AGAIN, I DON’T
SEE THE RELEVANCE OF THE PARTICULAR FACTORS THAT HAVE BEEN
LISTED. ASIDE FROM THE FACT THAT PROFESSOR BALDUS HAS SAID THAT
HE USED HIS JUDGEMENT IN DETERMINING IF THEY PERSISTED IN OTHER
SYSTEMS. AND THE STUDY IS OF THIS SYSTEM, AND I DON‘T SEE THE
RELEVANCE TQ THE INSTANT CASE.
MR. BOGER: YOUR HONOR, THE JUDGMENT OF PROFESSOR
vv
©
~N
*
4
+
O
N
BALDUS. OF COURSE, IS ONE OF THE REASONS HE“S AN EXPERT. THEY
10 CAN ATTACK HIS JUDGMENT ON CROSS-EXAMINATION AND I TRUST THEY
11 WILL. BUT IT SEEMS TO ME CLEARLY RELEVANT, IF HE IS A SOCIAL
12 SCIENTIST WHO IS AN EXPERT IN STUDYING THESE SYSTEMS. HAS MADE
13 SOME JUDGMENTS ABOUT BACKGROUND PRACTICE THAT MAY WELL SUPPRESS
14 THE RACIAL EFFECTS. AND HAS IN EFFECT THEREFORE ADVANCED AN
13 AUTHORITATIVE HYPOTHESIS. A RIVAL HYPOTHESIS.
16 THE COURT: I UNDERSTAND THESE TO BE ONES HE SELECTED
17 BECAUSE THEY MIGHT SUPPRESS IT.
18 MR. BOGER: I BELIEVE IT’S HIS TESTIMONY, YOUR HONOR,
". 19 THAT THESE WERE FACTORS THAT IN HIS JUDGMENT BEST EXPLAINED THE
20 SYSTEM OR MIGHT HAVE THE CLEAR -—-
21 THE COURT: IMPACT.
22 ~ MR. BOGER: CLEAR IMPACT.
23 THE COURT: I THINK I UNDERSTAND WHAT TESTIMONY IS, AND
24 I DON‘T THINK IT IS WHAT MR. BOGER SAID IT WAS.
23 1S IT?
BALDUS - DIRECT
THE WITNESS: YOUR HONOR, I WANTED IN THIS ANALYSIS TO
po
e
IDENTIFY THE RIVAL, THE VARIABLES THAT WERE ASSOCIATED WITH. IN
MY OPINION, THE MOST PLAUSIBLE RIVAL HYPOTHESIS. THAT IS,
FACTORS THAT SHOWED A PERSISTENT IMPACT AND EFFECT ON THE DEATH
SENTENCING RATE, A NUMBER OF WHICH ARE HIGHLY CORRELATED WITH
THE RACE OF THE VICTIM. AND IT STRUCK ME THAT THESE ARE THE
IMPORTANT FACTORS. THESE ARE THE IMPORTANT FACTORS THAT WOULD
MOST LIKELY EXPLAIN THE RACE EFFECTS IF THEY WERE GOING TO BE
Ov
Oo
W
A
d
N
EXPLAINED.
10 AND ALSO THEY WOULD PROVIDE A BASIS FOR EVALUATING THE
11 RELATIVE SIGNIFICANCE OF THIS RACIAL VARIABLE IN THIS SYSTEM.
12 THE COURT: WHAT WOULD —--EXCUSE ME. I THINK IT’S TIME
13 FOR ANOTHER BREAK.
14 I WILL ADMIT -81 AND -82.
15 AND WE“LL TAKE A TEN-MINUTE BREAK.
16 MR. BOGER: THANK YOU.
17 -_—-—
is (RECESS TAKEN.)
® ees
20 DAVID C. BALDUS,
i BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
22 TESTIFIED FURTHER AS FOLLOWS:
232 DIRECT EXAMINATION (CONT D)
24 BY BOGER:
23 RA. PROFESSOR BALDUS, LET ME NOW DIRECT YOUR ATTENTION TO DB-33
BALDUS ~ DIRECT
MARKED FOR IDENTIFICATION. I‘LL ASK YOU TO IDENTIFY THAT
DOCUMENT.
A. DB-83 IS A COPY OF TABLE 32 FROM OUR REPORT. IT LISTS THE
RACIAL CO-EFFICIENTS FOR RACE OF DEFENDANT AND RACE OF VICTIM.
WHICH WERE OBTAINED IN A SERIES OF 7 MULTIPLE REGRESSION
ANALYSES THAT WERE CONDUCTED WITH DIFFERENT COMBINATIONS OF
BACKGROUND VARIABLES.
MY EXPERIENCE HAS INDICATED THAT THERE IS NO ONE SET OF
BACKGROUND VARIABLES IN A STUDY LIKE THIS THAT ONE CAN CONTROL
FOR THAT WILL SATISFY EVERY CONCERN THAT AN INVESTIGATOR WILL
HAVE.
IF YOU HAVE A VERY LARGE VARIABLE, THE CONCERN IS THAT
THE VARIABLE HAS TOO MANY VARIABLES IN IT, AND IT“S OVERFITTED.
@. EXCUSE ME. YOU SAID A VERY LARGE VARIABLE?
A. I‘M SORRY. VERY LARGE ANALYSIS, EXCUSE ME, WITH
VARIABLES IN IT. THERE ARE TECHNICAL CONCERNS THAT ARE RAISED
AS A CONSEQUENCE OF INCLUDING A LARGE NUMBER OF VARIABLES IN A
MODEL.
IF ONE EXERCISES AN INDEPENDENT JUDGMENT ABOUT WHAT
VARIABLES SHOULD BE INCLUDED IN THE ANALYSIS, THEN THERES
ALWAYS THE POSSIBILITY BEING RAISED ABOUT THE BASIS OF THAT
JUDGMENT.
ONE USES THE MACHINE TO IDENTIFY THE FACTORS THAT
SHOULD BE INCLUDED BECAUSE OF THE STATISTICAL SIGNIFICANCE.
AND THEN THERE-S CONCERN THAT ONE HAS NOT EXERCISED OR
.
G
U
E
E
E
E
.
T
E
ed
B
E
R
T
BO
E
P
E
BE
T
O
a
E
wr
y nN
BALDUS - DIRECT
APPLIED THE LEARNING AND WISDOM OF THE CHARGING AND SENTENCE
SYSTEM ITSELF TO THE COLLECTION.
SO TO DEAL WITH THESE CONCERNS, AND THEY ARE VALID AND
LEGITIMATE CONCERNS, WE ADOPTED A VARIETY OF DIFFERENT MEASURES,
OR APPROACHES, FOR SELECTING COMBINATIONS OF FACTORS THAT WE
MIGHT CONTROL FOR AS A PRELUDE TO ESTIMATING THE RACE OF VICTIM
AND RACE OF DEFENDANT EFFECTS.
THE FIRST ROW HERE, YOUR HONOR, RESTATES THE RESULTS
THAT WE HAD PRESENTED EARLIER, THAT HAD BEEN ESTIMATED WITH A
SIMULTANEOUS CONTROL FOR MORE THAN TWO HUNDRED AGGRAVATING
FACTORS.
WE START FROM THAT POINT AND TRY AND DEVELOP METHODS TO
LIMIT THE NUMBER OF VARIABLES. THE BASIC PRINCIPLE IN DATA
ANALYSIS IS THAT IT IS DESIRABLE TO TRY AND DEVELOP A MORE
PARSIMONIQUS MODEL THAT CAPTURES THE MAIN CHARACTERISTICS OF THE
SYSTEM BUT DOES NOT OVERBURDEN WITH TOO MANY VARIABLES WHICH ARE
CARRYING BASICALLY REDUNDANT INFORMATION.
AND WHAT I“M GOING TO DO HERE IS GIVE YOU AN IDEA OF
SOME OF THE TECHNIQUES THAT ARE USED AND THAT WE USED TO TRY TO
PRUNE DOWN THE NUMBER OF VARIABLES WE HAD TO A MORE PARSIMONIOUS
SCALE.
THE FIRST, PARDON ME.» THE SECOND ROW. LABELED "B",
PRESENTS THE RESULTS WE OBTAINED AFTER PRUNING THE FILE OF
VARIABLES WITH A PROCEDURE KNOWN AS FACTOR ANALYSIS. AND TQ PUT
IT VERY SIMPLY. THE IDEA OF FACTOR ANALYSIS IS TQ IDENTIFY
| BALDUS —~ DIRECT
|
1 CLUSTERS OF VARIABLES THAT ARE CARRYING THE SAME INFORMATION.
2 THERE ARE A NUMBER OF VARIABLES IN THIS FILE THAT ARE VERY
RELATED IN TERMS OF THE INFORMATION THAT THEY'RE CARRYING. AND
WHAT A FACTOR ANALYSIS DOES FOR A STATISTICAL PROCEDURE IS
IDENTIFY CLUSTERS OF THOSE VARIABLES AND THEN CREATES A FACTOR.
3
4
5
& A NEW VARIABLE. WHICH CARRIES A SUBSTANTIAL PROPORTION OF THE
7 INFORMATION THAT'S EMBODIED IN THOSE INDIVIDUAL VARIABLES.
8 50 BY DOING THIS: YOU CAN IDENTIFY SERIES OF FACTORS
9 WHICH CARRY THE INFORMATION OF A SUBSTANTIALLY LARGER NUMBER OF
10 VARIABLES. AND THIS IS ONE WAY YOU CAN CUT DOWN ON THE NUMBER
11 OF VARIABLES THAT YOU INCLUDE IN YOUR ANALYSIS.
12
AND IT’S A WIDELY USED PROCEDURE IN THE SOCIAL
13 SCIENCES.
14
WE EMPLOYED SUCH A TECHNIQUE AND CUT THE GROUP OF |
15 VARIABLES THAT WE WOULD CONTROL FOR BY WAY OF BACKGROUND
|
16 ADJUSTMENT TO 126 VARIABLES. THAT WAS THE FIRST PROCEDURE THAT |
17 WE USED.
|
% 18
THE NEXT PROCEDURE THAT WE USED WAS TO USE A VERY
19 COMMONLY USED PROCEDURE KNOWN AS STEPWISE REGRESSION ANALYSIS.
—
—
—
—
—
—
20 AND BASICALLY WHAT A STEPWISE REGRESSION ANALYSIS DOES 1s
|
21
THE COURT: IS THAT WISE?
22
THE WITNESS: YES, YOUR HONOR.
23
THE COURT: W-I1-S-E, WISE?
24
MR. FORD: YES» YOUR HONOR. AND THE REASON THAT ITS
|
|
25 CALLED STEPWISE. 1 BELIEVE. IT STARTS OUT WITH A LARGE FILE OF |
i
D
v
B
N
d
D
O
N
pot
e
2823
BALDUS - DIRECT
VARIABLES, THE MACHINE CALCULATES WHICH OF ALL THOSE VARIABLES
WILL HAVE THE BIGGEST EFFECT IN TERMS OF EXPLAINING THE RESULTS.
THAT VARIABLE IS STEPPED INTO THE ANALYSIS 30 TO SPEAK, STEPPED
INTO THE FINAL RESULTS.
(AND YDU-’LL HAVE AN ANALYSIS. THE FIRST STEP IN THE
ANALYSIS WILL HAVE ONE VARIABLE, AND YOU-LL SEE THE REGRESSION
CO-EFFICIENT FOR THAT ONE VARIABLE. THEN THE MACHINE EXAMINES
THE OTHER VARIABLES THAT ARE NOT INCLUDED AND IT FINDS THE ONE
THAT ADDS THE MOST ADDITIONAL INFORMATION IN TERMS OF EXPLAINING
THE RESULTS.
NOW YOU HAVE A TWO-VARIABLE MODEL. WE‘LL GO OUT AND DO
THE SAME THING AGAIN. IT DOES IT AGAIN, AGAIN AND AGAIN UNTIL
IT HAS IDENTIFIED A GROUP OF VARIABLES WHICH ARE ALL
STATISTICALLY SIGNIFICANT AT SOME LEVEL WHICH YOU CAN PRESCRIBE
IN THE ANALYSIS. AND THEN THE MACHINE WILL STOP SELECTING WHEN
NONE OF THE OTHER VARIABLES CAN ENTER AT A STATISTICAL LEVEL
THAT YOU HAVE PRESCRIBED.
SO FOR THE PURPOSES OF THIS ANALYSIS. I STARTED OUT
WITH THE 230-0DD VARIABLES, AND CONDUCTED SUCH AN ANALYSIS, AND
IN THE END, BOILED THE FILE DOWN TO 43 NON-RACIAL VARIABLES.
THE COURT: THESE WERE IN COMPOSITES. THESE WERE
INDIVIDUAL VARIABLES THAT I COULD GO AND FIGURE OUT PRECISELY
WHICH ONE IN THE STUDY YOURE REFERRING TO IN THE QUESTIONNAIRE.
THE WITNESS: RIGHT.
AND THERE WERE 44 OF THOSE. AND THAT IS ONE PROCEDURE.
824
BALDUS - DIRECT
Ty
THERE I EXERCISED NO JUDGMENT WHATEVER. I LET THE MACHINE
EXERCISE THE JUDGMENT STRICTLY ON THE BASIS OF STATISTICAL
SIGNIFICANCE. THAT'S WHAT BROUGHT THE VARIABLES INTO THE MODEL.
AND ALONG WITH THE RACE OF THE VICTIM AND RACE OF THE
DEFENDANT VARIABLES. THEY WERE INCLUDED IN THE ANALYSIS AS WELL.
ANOTHER APPROACH THAT WE TOOK WAS TO IDENTIFY A GROUP
OF VARIABLES THAT WERE WHAT WE CALL LEGITIMATE NON-ARBITRARY
VARIABLES. THE TEACHING OF GODFREY V. GEORGIA IS, IN PART, BY
WAY OF A FOOTNOTE. THAT A JURY OR PROSECUTOR, I TAKE IT IS
OO
.
0
0
Nd
W
B
B
N
Pr
y IMPLICIT IN IT. CANNOT RATIONALLY CONSIDER CERTAIN FACTORS. IN
-
pb
THE SPECIFIC FOOTNOTE THE COURT SUGGESTED THAT IT WOULD BE
Te
y [8
IRRATIONAL TO CONSIDER THE GRUESOMENESS OF THE SCENE AS A FACTOR
[
Ww
IN EVALUATING THE SERIOUSNESS OF THE OFFENSE.
14 AS A CONSEQUENCE. WE CONDUCTED AN ANALYSIS THAT
15 EXCLUDED FACTORS RELATED TO THE SOCIAL STATUS OR EDUCATIONAL
1&6 LEVEL OF THE DEFENDANT. THOSE WOULD BE INAPPROPRIATE FACTORS TO
17 CONSIDER UNDER GODFREY.
13 WE ALSO LIMITED THE ANALYSIS TO AGGRAVATING FACTORS.
% 12 WHICH IN THE ANALYSIS HAD CO-EFFICIENTS THAT WERE IN THE
20 AGGRAVATED DIRECTION. THE THEORY WAS THAT IF IT WAS AN
21 AGGRAVATING FACTOR FROM A CONCEPTUAL STANDPOINT AND IT HAD A
22 MITIGATING EFFECT. THAT THAT WOULD BE AN IRRATIONAL APPLICATION
23 OF DECISION TO TREAT SOMETHING THAT WAS CONSIDERED UNDER LAW AS
24 AN AGGRAVATING FACTOR, TREAT IT AS A MITIGATING FACTOR, OR TO
23 TREAT SOMETHING THAT WAS A MITIGATING FACTOR AS AN AGGRAVATING
825
BALDUS - DIRECT
1 FACTOR.
2 SO WE MADE THESE, THIS IS WHERE WE INTERVENED IN TERMS
3 OF WHAT THE MACHINE HAD PRODUCED. WE TRIED TO PRODUCE A SERIES
4 OF VARIABLES LIMITED TO AGGRAVATING FACTORS THAT HAD AN
3 AGGRAVATING EFFECT IN TERMS OF THE LIKELIHOOD OF A DEATH
SENTENCE. AND MITIGATING FACTORS THAT HAD A MITIGATING EFFECT.
AND WE CONDUCTED THAT ANALYSIS, LIMITING THE VARIABLES ALSO IN
6
7
3 TERMS OF THOSE THAT SHOWED A STATISTICALLY SIGNIFICANT EFFECT.
9 WE DID THAT FIRST USING THE LEAST SQUARES REGRESSION
10 APPROACH. THATS WHAT’S SPECIFIED IN ROW D OF TABLE 32. WE
11 ENDED UP WITH A POOL OF 14 VARIABLES THAT HAD, THOSE 14
12 VARIABLES EXPLAINED A VERY SUBSTANTIAL PART OF WHAT WAS GOING ON
13 IN THE SYSTEM.
14 WE THEN REPEATED THAT PROCEDURE USING THE LOGISTIC
135 REGRESSION PROCEDURE WHICH I DESCRIBED EARLIER, AND PRODUCED A
16 SET OF 13 VARIABLES.
17 HAVING IDENTIFIED THESE DIFFERENT SETS OF VARIABLES.
18 ALL COLLECTED ON THE BASIS OF A DIFFERENT SET OF PRINCIPLES. WE
» 19 THEN CONDUCTED A SERIES OF LEAST SQUARES REGRESSION ANALYSES AND
20 THE RESULTS OF THEM ARE PRESENTED HERE IN TABLE 32.
21 AND IF ONE READS DOWN COLUMN B, TABLE 32, --
22 BY MR. BOGER?:
23 QR. THIS IS DB-83 YOU'RE REFERRING TO?
24 A. YES. ONE CAN SEE A HIGH LEVEL OF CONSISTENCY IN TERMS OF
25 THE MAGNITUDE OF THE REGRESSION CO-EFFICIENTS ESTIMATED FOR THE
WY
H
N
C
B
N
=
=
TE
be
h
3
0
N
=
0
T
N
a
16
17
BALDUS ~ DIRECT
RACE OF VICTIM, AS WELL AS CONSISTENTLY HIGH LEVELS OF
STATISTICAL SIGNIFICANCE ESTIMATED FOR THOSE REGRESSION
CO-EFFICIENTS.
IT’S THIS SORT OF CONSISTENCY IN THE PATTERN THAT GIVES
ONE CONFIDENCE THAT WHAT YOU“RE SAYING HERE IS A REAL EFFECT IN
THE SYSTEM: THAT NO MATTER HOW YOU PICK A GROUP OF VARIABLES. NO
MATTER WHAT THEY ARE, BASICALLY YOU SEE THE SAME PATTERN.
THE RESULTS WITH RESPECT TO THE RACE OF DEFENDANT
VARIABLE SHOW A SIMILAR PATTERN OF CONSISTENCY ACROSS THE
DIFFERENT METHODS ALTHOUGH IN ONE INSTANCE. THE LAST INSTANCE
THERE, ROW E, THE CO-EFFICIENT IS NOT AS LARGE.
MY CONCLUSION IS FROM LOOKING AT THIS IS THAT IF YOU
LOOK STRICTLY AT THE LEAST SQUARES ANALYSES THAT ARE REPORTED IN
HERE. YOU SEE BOTH STRONG AND PERSISTENT RACE OF THE VICTIM
EFFECTS AND RACE OF DEFENDANT EFFECTS.
HOWEVER, WE WERE WELL AWARE OF THE FACT THAT WE
SOMETIMES GOT DIFFERENT RESULTS WHEN WE USED THE LOGISTIC
REGRESSION APPROACH AND WE CONSIDERED IT APPROPRIATE TO REPEAT
AS MUCH OF THIS ANALYSIS AS POSSIBLE USING THE LOGISTIC
REGRESSION METHODS. AND THOSE RESULTS ARE FRESENTED ON THE
SECOND PAGE OF DB-83.
FOR THE REASONS I STATED TO YOU EARLIER. YOUR HONOR.
THE SOFTWARE AVAILABLE FOR CONDUCTING LOGISTICS THAT IS
AVAILABLE AT THE UNIVERSITY OF IQWA AND AT SYRACUSE UNIVERSITY
WOULD NOT ACCOMMODATE THESE LARGE ANALYSES. YOU CAN'T DO A
827
BALIUS ~ DIRECT
p>
STEPWISE REGRESSION WITH TWO HUNDRED VARIABLES IN THE SAME WAY
YOU CAN IN A LEAST SQUARES. SO WE WERE REDUCED TO USING IN THIS
APPROACH A SMALLER NUMBER OF BACKGROUND VARIABLES THAT CONTROLS.
AND YOU CAN SEE THAT IN EACH CASE, THE RACE OF VICTIM
CO-EFFICIENT IS SUBSTANTIAL AND STATISTICALLY SIGNIFICANT.
HOWEVER, AGAIN WE SEE THE PATTERN WITH RESPECT TO THE RACE OF
DEFENDANT, THAT IT’S NOT EVEN CONSISTENT BETWEEN THE TWO
ANALYSES REPORTED IN A AND B, AND IN NEITHER CASE IS IT
6
N
O
s
W
N
STATISTICALLY SIGNIFICANT BEYOND THE .035 LEVEL.
Fe
y oO
AGAIN, WE SEE HERE THERE-S A LACK OF STABILITY IN THE
RESULTS ACHIEVED USING THE TWO DIFFERENT REGRESSION PROCEDURES. —-
Fy
12 WHICH AGAIN REDUCES OUR CONFIDENCE THAT IN THE SYSTEM AS A WHOLE
13 THERE“S A REAL EFFECT FOR THAT VARIABLE.
14 THE COURT: ALL RIGHT. LET ME ASK YOU A QUESTION.
15 BACK YONDER SOMEWHERE WHEN WE STARTED. WE STARTED WITH SOME
16 FIGURE OF .0S AS BEING THE PROBABILITY OF RECEIVING THE DEATH
17 PENALTY IF YOU WERE INDICTED FOR HOMICIDE OR MURDER IN GEORGIA.
18 NOW TO UNDERSTAND THE EFFECT OF WHAT YOU SEE HERE IN
1 1% COLUMN B, AM I TO ADD THE .06 TO THE .057?
20 THE WITNESS: IN TERMS OF THE RACE OF DEFENDANT?
21 THE COURT: NO, RACE OF VICTIM.
22 THE WITNESS: THIS IS ON THE FIRST PAGE OF DB-83, YOUR
23 HONOR?
24 THE COURT: IM LOOKING AT DB-83, PICK A LINE. I DON’T
23 CARE WHICH ONE, LET'S JUST TAKE A AS AN EXAMPLE. YOUVE GOT
OO
1
©
M
N
O
U
B
3
M
re
po
et
Ty
T
W
ee ib pr — Sb Sitesi seein _ (tidbit eevee 4 - BE Ba
BALDUS - DIRECT
YOUR CO-EFFICIENT THERE OF .06. WE TALKED ABOUT THAT EARLIER AS
BEING SIX PERCENTAGE POINTS.
THE WITNESS: YES.
THE COURT: AM I TO UNDERSTAND THAT THAT, THAT IF THE
VICTIM IS WHITE, YOU HAVE A .06 PERCENT CHANCE OF GETTING THE
DEATH PENALTY OR AM I TO UNDERSTAND THAT YOU HAVE A .0S PLUS A
.06, OR AM I TO UNDERSTAND NEITHER?
THE WITNESS: NO, I WOULD SAY NEITHER, YOUR HONOR.
WHAT THIS TELLS IS -- LET ME GO BY WAY OF BACKGROUND.
THE CHANCES THAT ANYBODY IS GOING TO RECEIVE A DEATH
SENTENCE IS GOING TO DEPEND ON WHAT THE OTHER AGGRAVATING AND
MITIGATING CIRCUMSTANCES ARE IN THEIR CASE. AS WE’LL SEE LATER
ON IN SOME TABLES. AMONG THE VAST MAJORITY OF THE CASES IN OUR
UNIVERSE. THERE‘S NO CHANCE OF RECEIVING THE DEATH SENTENCE
BECAUSE THE CASES ARE JUST SIMPLY NOT AGGRAVATED ENOUGH.
BUT AMONG THE CASES WHERE THERES ANY CHANCE OF A DEATH
SENTENCE. EVEN WITHIN THAT GROUP, THERE ARE SIGNIFICANT
DIFFERENCES IN TERMS OF THE RISK OF A DEATH SENTENCE, THE
PROPORTION THAT GET DEATH SENTENCES WITHIN THESE SUB-GROUPS,
DEPENDING ON THE HOW AGGRAVATED THE CASES ARE, SO IF YOU TAKE
ANY ONE OF THOSE GROUPS AND COMPARE THE RATES AT WHICH WHITE AND
BLACK VICTIM CASES ARE BEING SENTENCED. YOU MIGHT FIND IN A
HIGHLY AGGRAVATED CASE THAT THE AVERAGE CLASS OF CASES, THAT THE
AVERAGE RATE IS,» SAY, 65 PERCENT, .65, THIS .0&6 WE SUGGEST THAT
IN THAT CLASS OF CASES THAT THE RATE IS GOING TO BE, 3IX
82%
BALDUS ~ DIRECT
PERCENTAGE PDINTS HIGHER, NOT SIX PERCENT OF THAT UNDERLYING
SIX.
MOREOVER, THIS REPRESENTS AN AVERAGE EFFECT ACROSS ALL
THE CATEGORIES OF CASES.
THE COURT: WELL. I THINK YOU AND I ARE TALKING ABOUT
THE SAME THING.
MR. BOGER. WHERE. WHAT“S THE BASE DOCUMENT WHERE WE
STARTED QUT WITH —
MR. BOGER: 1 BELIEVE WE“RE BACK IN ABOUT DB- EARLY
60°8, YOUR HONOR. LET“S SEE. DB-359 OR DB-38.
THE COURT: OF THE, YOU SEE THAT ON -59, DO YOU HAVE
THAT?
THE WITNESS: YES, YOUR HONOR. I DO.
THE COURT: YOU SEE THAT .057?
THE WITNESS: RIGHT.
THE COURT: NOW, CAN I ADD THE .0& TO THAT?
THE WITNESS: I DONT THINK THAT THATS A MEANINGFUL
CALCULATION, YOUR HONOR, IN THE SENSE THAT, WELL, I GUESS IF
YOU’RE TALKING ABOUT THE AVERAGE EFFECT, THAT MIGHT MAKE SOME
SENSE. BUT I THINK IT“S MORE UNDERSTANDABLE TO THINK OF IT IN
TERMS OF DIFFERENT CATEGORIES OF OFFENDERS WHO ARE IN DIFFERENT
RISK CATEGORIES. AT LEAST THE WAY I CONCEPTUALIZE IT. TQ TRY TO
UNDERSTAND HOW THE SYSTEM OPERATES. THAT ON THE --
THE COURT: LETS GU TO LINE 2 AND EVERYBODY IS AT RISK
THERE, HAVING BEEN INDICTED IN THE FIRST CASE FOR A CAFITAL
S
N
6
a
v
©
10
11
12
BALDUS - DIRECT
OFFENSE.
BY VIRTUE OF YOUR FINDINGS COULD I THEN SAY THAT IF THE
VICTIM HAPPENED TO BE WHITE, HOLDING CONSTANT FOR 230 OTHER
POSSIBLE VARIABLES. THAT HE WOULD THEN JUMP TO A .11 CHANCE OF
GETTING THE DEATH PENALTY OR IS IT HE GOES FROM A .05 TO A .067
THE WITNESS: OH, NO. IT’S AN INCREMENT. IT”S AN
ADDITIONAL SIX PERCENTAGE POINTS THAT YOU INCREASE YOUR RISK.
BUT IT’S, THAT NUMBER, YOUR HONOR. IS AN AVERAGE OF THE
INCREASED RISK ACROSS ALL THESE CATEGORIES OF CASES. THAT'S THE
POINT I WANT TO EMPHASIZE.
THE COURT: LET ME TELL YOU THE CONTEXT IN WHICH I“M
ASKING THE QUESTION.
HAVING NOW COME TO THE POINT WHERE YOU“VE GOT 30ME
EVIDENCE THAT TENDS TO DEMONSTRATE YOUR CONTENTION, PETITIONER’S
CONTENTION, THE COURT IS GOING TO HAVE TO BE CONCERNED ABOUT
LEGAL SIGNIFICANCE AS WELL AS STATISTICAL SIGNIFICANCE.
AND IN THINKING ABOUT THE COMPARISONS THAT ARE
AVAILABLE, SUCH AS UNDERREPRESENTATION IN JURIES. ALL THAT SORT
OF THING. WE GET FINAGLE FACTOR THAT COURTS HAVE ALWAYS
RECOGNIZED OF FOUR PERCENT UNDERREPRESENTATION IS OKAY, THAT
SORT OF THING, AM I LOOKING AT A FIGURE THATS GOING TO TELL ME
BASICALLY WE“RE TALKING ABOUT A DOUBLING OF PROBABILITIES OR A
HALVING OF PROBABILITIES, DEPENDING ON WHICH WAY YOU LOOK AT
IT. OR AM I TALKING ABOUT SIX PERCENT INCREASE. AND I THINK THE
ANSWER IS I’M TALKING ABOUT A MUCH LARGER EFFECT.
orn print poses, § S——— — FE
O
O
N
d
d
N
>
o
Po
y
pe
s
831
BALDUS ~ DIRECT
IS THAT TRUE, HOLDING EVERYTHING ELSE CONSTANT, WE-RE
NOT TALKING ABOUT A RACIALLY IMPACTED VARIABLE WHICH HAS ONLY A
SMALL EFFECT.
THE WITNESS: WHEN YOU LOOK AT ALL THE CASES, YOUR
HONOR, THE AVERAGE EFFECT IS SMALL. THE EFFECT WHICH I HOPE WE
CAN DEMONSTRATE EVENTUALLY TO YOU HERE IN THE CASES WHERE
EXERCISE IS REALLY DISCRETION, OR DISCRETION IS REALLY
EXERCISED. IT’S SUBSTANTIALLY LARGER.
THE COURT: WELL. LET ME JUST WAIT FOR THE LAST CHAPTER
THEN, OKAY?
GO AHEAD.
ALL RIGHT, I‘LL ADMIT 83.
MR. BOGER: THANK YOU, YOUR HONOR.
BY MR. BOGER:
0. NOW, PROFESSOR BALDUS, ALL OF THE ANALYSES YOU HAVE SHOWN US
TO THIS POINT WERE CONDUCTED IN THE CHARGING AND SENTENCING
STUDY ON YOUR ENTIRE UNIVERSE OF CASES. USING YOUR SAMPLE
PROCEDURE, IS THAT CORRECT?
A. WELL. SOME OF THE ANALYSES WE JUST LOOKED AT. IN FACT. THE
ANALYSIS IN TABLE 32, RESTRICTED THE UNIVERSE SLIGHTLY. IT WAS
RESTRICTED TO PEOPLE WHO WERE INDICTED FOR MURDER, WHEREAS MANY
OF THE OTHER ONES INCLUDED THOSE ADDITIONAL 141 ESTIMATED CASES.
THE RESULTS ARE IDENTICAL IN THE TWO ANALYSES.
BUT THAT“S ESSENTIALLY CORRECT: IT INCLUDES ALL THE
PEOPLE THAT WERE INDICTED FOR MURDER.
i. ————
pb
N
E
:
N
n
W
N
pb
Qo
(a
y
Fur
y
12
132
832
BALDUS - DIRECT
@. DID YOU DO ANY ANALYSIS THAT LOCKED AT SELECTED SUB-GROUPS
OF YOUR UNIVERSE, OR YOUR SAMPLE?
A. YES.
@. FOR WHAT REASONS?
A. ONE OF THE OBJECTIVES OF THAT RESEARCH ENTERPRISE WAS TO TRY
IDENTIFY AND DETERMINE If THERE WERE CLASSES OF CASES WHERE
|RACIAL EFFECTS WERE MORE PROMINENT.
IT’S WIDELY RECOGNIZED IN THE, IN THE LITERATURE AND
COMMON SENSE WOULD INDICATE THAT THESE SORTS OF EFFECTS ARE NOT
LIKELY TO BE UNIFORM THROUGHOUT THE SYSTEM. THEY'RE LIKELY TO
BE CONCENTRATED IN CERTAIN CATEGORIES.
ALSO YOU ARE VERY MUCH INTERESTED IN IDENTIFYING THE
STAGES IN THE PROCESS THAT WERE A LIKELY SOURCE OF THESE
DISPARITIES.
WHEN ONE LOOKS AT THE COMBINED EFFECTS OF THE SYSTEM.
AND THE RACIAL EFFECTS THAT ARE ATTENDANT TO SUCH EXAMINATION,
ITS IMPOSSIBLE TO TELL WHERE THE DISPARITIES ARE COMING FROM.
ARE THEY COMING FROM THE JURY DEATH SENTENCING DECISIONS? ARE
THEY COMING FROM PLEA BARGAIN DECISIONS? WHERE ARE THEY COMING
FROM? YOU CAN“T TELL FROM THAT ANALYSIS ALONE.
THE FIRST STAGE OF THE ANALYSIS WAS TO FOCUS IN ON THE
TYPES OF CASES DEFINED BY STATUTORY AGGRAVATING CIRCUMSTANCES.
@. LET ME DIRECT YOUR ATTENTION THEN TO DB-84, AND ILL ASK IF
YOU CAN IDENTIFY THAT DOCUMENT?
A. YES. DB-84 IS TABLE 33 FROM OUR REPORT. IT PRESENTS A
o
u
0
N
e
(8)
YH
oO
N
B
10
11
BALDUS - DIRECT
SERIES OF MEASURES OF THE IMPORTANCE OF TEN STATUTORY
AGGRAVATING FACTORS IN THE CHARGING AND SENTENCING PROCESS.
I WAS VERY MUCH INTERESTED IN THE PRACTICAL IMPACT OF
THESE STATUTORY AGGRAVATING FACTORS, AND I THOUGHT IT WAS
IMPORTANT TO FOCUS ON THE CASES INVOLVING THE STATUTORY
AGGRAVATING FACTORS THAT ARE MOST IMPORTANT IN THE SYSTEM. AND
IN THE COURSE OF EXAMINING THE EVIDENCE ON THAT. I PRODUCED THE
RESULTS THAT ARE PRESENTED IN DB-84.
ONE WAY YOU CAN ASSESS THE IMPORTANCE OF A STATUTORY
AGGRAVATING FACTOR IS TO EXAMINE THE PROPORTION OF CASES THAT
ARE DEATH ELIGIBLE AS A RESULT OF THE PRESENCE OF THAT
FACTOR IN THE CASE. AND WHAT I MEAN BY THAT, YOUR HONOR, IS
ACCORDING TO OUR RECORDS. THE FACTOR THAT DEFINES DEATH
ELIGIBILITY IN THE STATUTE WAS PRESENT IN THE CASE. REGARDLESS
OF WHAT HAPPENED TO THE CASE. MANY OF THESE CASES PLEAD QUT TO
MANSLAUGHTER, MURDER, VARIETY OF OTHER DISPOSITIONS,
AND WHEN YOU LOOK DOWN COLUMN B OF DB-84, YOU CAN SEE
THAT B2 STATUTORY AGGRAVATING FACTOR APPEARS FAIRLY IMPORTANT.
BECAUSE SOME TWENTY PERCENT OF ALL THE CASES APPEAR TO HAVE THAT
STATUTORY FACTOR PRESENT IN THE UNIVERSE. THAT’S WHAT THIS
COLUMN INDICATES, THE PERCENTAGE OF ALL THE CASES IN THE TOTAL
UNIVERSE THAT HAVE THIS STATUTORY FACTOR PRESENT.
WE CAN SEE THAT B2, B3 APPEARS TO BE REASONABLY
IMPORTANT IN THE TERMS OF THIS DIMENSION. SO DOES B4. B4 IS IN
MANY WAYS A MIRROR OF B2, BECAUSE BZ IS DOMINATED BY ARMED
$
NN
Oo
u
rn
0
BALDUS - DIRECT
ROBBERY.
THEN AS YOU LOOK DOWN THE REST OF THESE. YOU CAN SEE
THAT OF THE REMAINING B2 FACTORS, THE ONLY ONE THAT LOOKS TO BE
IMPORTANT IS B7. FIFTY-ONE PERCENT OF THE CASES IN OUR UNIVERSE
APPEAR TO HAVE ONE OR MORE, PARDON ME. APPEAR TO HAVE A FACTOR
THAT WOULD MAKE THEM DEATH ELIGIBLE UNDER THE B7 STATUTORY
AGGRAVATING CIRCUMSTANCE.
NOW, ONE OF THE CHALLENGES THAT WE HAD IN DOING THIS
ANALYSIS WAS TO DETERMINE WHEN THE CASE IS DEATH ELIGIBLE UNDER
B7, BECAUSE THAT IS A VERY GENERAL STANDARD. SO, WE UNDERTOOK
AN EXTENSIVE RESEARCH PROJECT IN WHICH WE READ ALL OF THE CASES
OF THE GEORGIA SUPREME COURT WHICH DEFINED THE FACTUAL LIMITS OF
THE B7 FACTOR AND THEN FOR EACH OF THOSE 3 CASES WE PRODUCED A
COMPUTER PRINTOUT WHICH IDENTIFIED THE PRECISE FACTS THAT WERE
IN THOSE CASES IN TERMS OF THE COMPUTER CODING THAT APPEARED TO
SATISFY THE CONDITIONS SPECIFIED BY THE GEORGIA SUPREME COURT
FOR MAKING A CASE DEATH ELIGIBLE UNDER B7.
AND AS A RESULT OF THAT ANALYSIS, WE PRODUCED FIVE.
FIVE-FACTOR CRITERIA, SET OF CRITERIA, THAT WOULD THEN BE USED,
WOULD BE APPLIED TO THE CASES. SPECIFIED DEATH ELIGIBLE CASES
UNDER THIS FACTOR.
SO ON THE BASIS OF THIS, YOU CAN SEE THAT IT LOOKS LIKE
THERE ARE NUMBER OF CANDIDATES FOR IMPORTANCE IN THE SYSTEM. WE
WERE PARTICULARLY INTERESTED IN THE PRACTICAL IMPORTANCE OF
THESE PARTICULAR FACTORS.
833
BALDUS -~ DIRECT
THE NEXT COLUMN, BC, BEGINS TO HIGHLIGHT WHAT THE
FACTORS ARE THAT DOMINATE THE CHARGING AND SENTENCING SYSTEM IN
GEORGIA.
AS YOU CAN SEE, AS YOU DO DOWN COLUMN C. YOU SEE THE
PERCENTAGE AND NUMBER OF CASES IN OUR UNIVERSE OF CASES THAT
WENT TO A PENALTY TRIAL WHERE THE PROSECUTOR REQUESTED AND
OBTAINED A CHARGE WITH RESPECT TO THAT STATUTORY AGGRAVATING
FACTOR. SO THAT YOU CAN SEE SIXTY PERCENT OF ALL THE CASES THAT
WENT TO A PENALTY TRIAL HAD A CHARGE ON B2 FACTOR.
AND YOU LOOK DOWN OVER THE REST OF THESE FACTORS. THE
ONLY OTHER ONE THAT EVEN COMES CLOSE TO IT IS B7 FACTOR. 356
PERCENTAGE, S56 PERCENT OF ALL THE CASES THAT WENT TO PENALTY
TRIALS HAD A CHARGE PRESENTED TO THE JURY ON THE B7 FACTOR.
THEN WHEN YOU LOOK OVER ON COLUMN D, YOU SEE THE
EFFECTS OF THOSE PRACTICES BY PROSECUTORS BECAUSE YOU SEE THE
PROPORTION AND NUMBER OF DEATH SENTENCING DECISIONS IN WHICH
THESE INDIVIDUAL STATUTORY AGGRAVATING FACTORS WERE FOUND.
AGAIN YOU SEE THAT B2 AND B7 ARE THE FACTORS THAT
REALLY DETERMINE WHO GETS DEATH IN THE SYSTEM, BECAUSE SIXTY-SIX
PERCENT OF THE CASES INVOLVED A FINDING OF B2 AND SOME 5%
PERCENT INVOLVED A B7 FINDING.
THE OTHERS. YOU CAN SEE, IN TERMS OF THEIR PRACTICAL
IMPACT ON THE NUMBERS OF PEOPLE THAT ARE SENTENCED TO DEATH. ARE
OF RELATIVELY MINOR IMPORTANCE WHEN CONTRASTED TO THE B2 AND B7
FACTORS.
836
BALDUS ~ DIRECT
1 AND THEN ONE FINAL MEASURE OF THIS. YOUR HONOR. IS TO
2 LOOK AT THE CASES WHERE THERE WAS ONLY ONE STATUTORY AGGRAVATING
3 FACTORS THAT SUPPORTED THE DEATH SENTENCE. WE FOUND THAT IN
MANY CASES THERE ARE MULTIFLE STATUTORY AGGRAVATING FACTORS THAT
ARE CHARGED. BECAUSE THESE FACTORS ARE FREQUENTLY -- MORE THAN
ONE OF THESE FACTORS IS FREQUENTLY PRESENT IN A CASE.
a
3
1)
7 WE LOOKED FOR THIS FINAL BIT OF EVIDENCE ON THE
8 IMPORTANCE OF THIS FACTOR AT THE CASES IN WHICH THE DEATH
2? SENTENCE THAT WAS ACTUALLY IMPOSED WAS SUPPORTED BY ONLY ONE
10 STATUTORY AGGRAVATING FACTOR AND WE FOUND, AGAIN, THAT IT WAS
11 THE B2 FACTOR AND THE B7 FACTOR THAT DOMINATED THE FIELD. THOSE
12 WERE THE FACTORS THAT STANDING ALONE SUPFORTED MOST OF THE DEATH
13 SENTENCES WHERE THERE HAD BEEN A FINDING OF ONLY ONE STATUTORY
14 AGGRAVATING FACTOR.
15 THEN OUR FINAL MEASURE WAS THE RESULT OF A REGRESSION
146 ANALYSIS, LEAST SQUARES AND LOGISTIC REGRESSION ANALYSES, THAT
37 SHOWED THE STATISTICAL EFFECT OF THESE VARIABLES IN TERMS OF
18 EXPLAINING THE OUTCOMES. AND YOU CAN SEE THERE ALSO THAT THE B2
4 12 FACTOR IS EXTREMELY IMPORTANT.
20 : THE B7 FACTOR. FROM A STATISTICAL STANDPOINT, DOES NOT
21 APPEAR THAT IMPORTANT. IT ONLY HAD A CO-EFFICIENT OF 3 POINTS
22 ASSOCIATED WITH IT. WHEN YOU LOOK AT THE LOGISTIC RESULTS,
23 HOWEVER, WHEN YOU LOOK AT THE, PARDON ME. WHEN YOU LOOK AT THE
24 LEAST SQUARES RESULTS, WHEN YOU LOOK LOGISTIC RESULTS OVER IN
25 THE LAST COLUMN, G. OF DB-84, YOU CAN SEE THAT THAT MEASURE
O
f
IB
NN
0
OU
H
W
o
N
Fo
y
e37
BALDUS - DIRECT
SUGGESTS THAT IT HAS A LARGER IMPACT THAN IS SUGGESTED BY THE
LEAST SQUARES RESULT.
THE KEY, THOUGH, I THINK, TO UNDERSTANDING THE
IMPORTANCE OF THESE FACTORS IS THE SIGNIFICANCE THEY HAVE IN
ACCOUNTING FOR THE DEATH SENTENCES THAT ARE IMPOSED IN THIS
SYSTEM. IT WAS THE RESULT OF THIS ANALYSIS, YOUR HONOR. THAT
SUGGESTED THAT IT MIGHT BE USEFUL TO FOCUS ON THE DISPOSITION OF
CASES THAT FALL UNDER THESE CATEGORIES.
MR. BOGER: YOUR HONOR, AT THIS POINT I WOULD THEN MOVE
THE ADMISSION OF DB-84 INTO EVIDENCE.
MS. WESTMORELAND: YOUR HONOR, ASIDE FROM THE STANDARD
OBJECTION, I WOULD ALSO OBJECT TO DB-84 BECAUSE THIS DOCUMENT
DOES NOT APPEAR TO HAVE ANY RELEVANCE, AT LEAST ON THE FACE OF
IT, AS TO THE ISSUES PRESENT IN THIS CASE, ASIDE FROM THE FACT
THAT IT BASICALLY SHOWS, I GUESS. OUTCOMES BASED ON CHARGING A
CERTAIN STATUTORY AGGRAVATING CIRCUMSTANCE, AND I DON‘T THINK
THAT’S PARTICULARLY RELEVANT TO RACIAL EFFECTS AT THIS TIME.
I FURTHER HAVE SOME QUESTION AS TO STATEMENTS THAT MAY
HAVE BEEN MADE, AND I DON’T KNOW IF THIS IS INTENDED TO REFLECT
PROSECUTORIAL DECISION ON THE CHARGING OF THE PARTICULAR
AGGRAVATING CIRCUMSTANCES. I BELIEVE UNDER THE STATUTE THAT THE
JUDGE COULD DECLINE TO CHARGE CERTAIN AGGRAVATING CIRCUMSTANCES.
IF NOT SUPPORTED BY THE EVIDENCE OR COULD PERHAPS EVEN CHARGE
HIS OWN IF HE THOUGHT IT WAS SUPPORTED BY THE EVIDENCE. IM NOT
SURE IT WOULD REFLECT A PROSECUTORIAL DECISION NECESSARILY.
10
24
235
838
BALDUS - DIRECT
MR. BOGER: YOUR HONOR. I DON’T HEAR ANY OBJECTION
THERE THAT REALLY GOES TOWARD MORE THAN THE WEIGHT OF THIS
EVIDENCE. PROFESSOR BALDUS HAS TESTIFIED HE WANTED TO FIND THE
PLACES IN THE SYSTEM WHERE THERE MIGHT BE SIGNIFICANT RACIAL
EFFECTS, AND TO DO FOR, HE WANTED TO LOOK IN PART AT THOSE
STATUTORY AGGRAVATING FACTORS AND IN CASES THAT CONTAINED THEM.
IT DID APPEAR TO BE MORE DEATH WORTHY OR ELIGIBLE FOR DEATH AND
THAT THIS ANALYSIS THAT HE’S CONDUCTED IS THE FIRST STEP IN HIS
APPROACH TO IDENTIFYING THOSE SUB-GROUPS OF CASES ON WHICH HE
WANTED TO DO FURTHER ANALYSIS.
AND SO IT’S HEADED DIRECTLY TOWARDS HIS ANALYSIS BASED
ON RACE, AND THEREFORE IT SEEMS TO ME IS THE FOUNDATION FOR THE
FURTHER TESTIMONY HE WOULD GIVE ON THE RACIAL EFFECTS. PERHAPS
ITS NOT A NECESSARY FOUNDATION. HE COULD SIMPLY ARBITRARILY
HAVE SAID I“M GOING TO LOOK AT B2 AND B7 CASES. BUT HE EXPLAINS
HERE WHY HE PICKED THOSE TWO AGGRAVATING CIRCUMSTANCES FOR
FURTHER ANALYSIS.
THE COURT: IT SEEMS TO ME THIS FALLS IN THE CATEGORY
OF A LOT OF DOCUMENTS THAT IVE ADMITTED AS DEMONSTRATIVE AND I
WILL DO THE SAME WITH THAT ONE, UNDERSTANDING THAT THAT IS JUST
A PREDICATE FOR THE NEXT STEP THAT -——
MR. BOGER: ALL RIGHT. YOUR HONOR.
BY MR. BOGER:
@. PROFESSOR BALDUS, ONCE YOU HAD IDENTIFIED THESE TWO
CATEGORIES, OR SUB-CATEGORIES OF CASES, WHERE DID YOUR ANALYSIS
pa
Wo
08
N
O
0
s
W
N
10
11
BALDUS ~ DIRECT
LEAD YOU NEXT?
A. THE NEXT STEP WAS TO PULL OUT THOSE CASES AND SUBJECT THEM
TO SEPARATE ANALYSIS. AND I DID THAT USING A VARIETY OF
COMBINATIONS OF BACKGROUND FACTORS TO ADJUST FOR THE NON-RACIAL
FACTORS THAT MIGHT POSSIBLY EXPLAIN ANY RACIAL DISPARITIES THAT
MIGHT BE PERCEIVED IN THESE ANALYSES.
Q. HOW DID YOU, HOW DID YOU DO THAT?
A. WELL, YOU SIMPLY REQUEST THE COMPUTER TO SELECT THOSE CASES
AND THEN THOSE CASES CAN BE SUBJECTED TO SEPARATE ANALYSIS, THE
SORT OF -
@. LET ME ASK YOU TO TURN YOUR ATTENTION TO DB-835. I“M ASKING
YOU TO IDENTIFY THAT DOCUMENT?
A. YES, DB-85 IS TABLE 34 FROM OUR REPORT, AND IT PRESENTS THE
RESULTS OF THOSE ANALYSES.
@. WHAT DO THOSE RESULTS REFLECT?
A. THEY PRESENT THE RACE OF VICTIM AND RACE OF DEFENDANT
EFFECTS ESTIMATED WITHIN THESE TWO SUB-GROUPS OF CASES WHEN EACH
WAS ANALYZED SEPARATELY AND THEN WHEN THE TWO GROUPS OF CASES
WERE ANALYZED TOGETHER. AND THE CO-EFFICIENTS IN ROW 1 ARE
THOSE THAT WERE ESTIMATED BEFORE ADJUSTMENT FOR ANY NON-RACIAL
BACKGROUND FACTORS. THOSE ARE UNADJUSTED MEASURES EFFECT: AND I
PUT THEM IN THERE AS SORT OF A POINT OF REFERENCE, POINT OF
‘BEGINNING FOR THE ANALYSIS.
THE NEXT PANEL OF CO-EFFICIENTS IN COLUMN B, PARDON ME,
ROW 2, SHOWS THE EFFECTS AFTER ADJUSTMENT FOR THE STATUTORY
340
BALDUS - DIRECT
AGGRAVATING FACTORS.
@. IS THAT SIMULTANEOUS ADJUSTMENT?
A. YES. ROW 3 SHOWS THE RACIAL EFFECTS AFTER ADJUSTMENT FOR
THE SAME 230 BACKGROUND FACTORS THAT WE USED TO CONTROL FOR
NON-RACIAL FACTORS IN OUR EARLIER ANALYSIS THAT WAS REPORTED.
AND IN FACT, IN ROW 5 OF DB-85. I HAVE THOSE RESULTS THERE,
PRESENTED AGAIN AS A POINT OF COMPARISON.
I THINK THE KEY RESULTS IN THIS ANALYSIS ARE FOUND IN
THE THIRD ROW WHERE YOU SEE THE RESULTS FOR THE B2 CASES. AND
THE RESULTS FOR THE B7 CASES. AND THE B2 AND B7 CASES TOOGETHER.
WHEN YOU COMBINE THE B2 AND B7 CASES TOGETHER. IT SHOWS
STRONG RACE OF VICTIM EFFECTS. AS WELL AS STRONG RACE OF
DEFENDANT EFFECTS.
AND WHAT IS PARTICULARLY INTERESTING IS THE RACE OF
DEFENDANT EFFECTS THAT YOU SEE IN THE B7 CASES.
THERE IS A POCKET WHERE YOU SEE STRONG RACE OF
DEFENDANT EFFECTS.
@. IS THAT THE FIGURE .13 --
A. YES.
@. =-- IN THE SECOND COLUMN?
A. RIGHT. IT SUGGESTED BLACK OFFENDERS IN THIS CATEGORY OF
CASES HAVE A 13 PERCENTAGE POINTS HIGHER CHANCE OF RECEIVING
DEATH SENTENCE THAN WHITE OFFENDERS. AND I MIGHT ADD THAT
THAT’S A RESULT THAT'S CONSISTENT WITH THE EXPECTATION THAT IN
AGGRAVATED CASES, BS CASES, THESE CHARACTERISTICS THAT WE
E
E
B
E
R
WE
E
E
BL
Se
C
a
e
P
R
P
O
=
3
.
N
e
DO
Pr
y Ho
B41
BALDUS - DIRECT
IDENTIFIED ARE QUITE OUTRAGEOUS AND THAT THESE ARE THE KINDS OF
CASES THE LITERATURE SUGGESTS ONE MIGHT SEE THOSE EFFECTS, AND
THATS WHAT WE WERE TESTING FOR HERE IN THIS ANALYSIS.
@. WHAT DO ROWS 4 AND J REFLECT?
A. OM, ROW 4 WAS DONE TO AVOID THE FROBLEM OF HAVING TOO LARGE
A MODEL. TOO LARGE AN ANALYSIS. WE PRUNED THE FILE OF CASES,
USING THE STEPWISE REGRESSION, THE PROCEDURE I DESCRIBED
EARLIER, WHICH LIMITS THE VARIABLES IN THE ANALYSIS TO THOSE
THAT SHOW STATISTICAL SIGNIFICANCE AND IN HERE I PICKED THE
LEVEL .0S TO LIMIT THEM. AND YOU CAN SEE THAT WHEN THAT IS
DONE, YOU CAN SEE THE RACE EFFECTS DECLINE SOMEWHAT BUT STILL
THEY MAINTAIN A SIGNIFICANCE. THE OVERALL EFFECT OF THIS IS
THAT THE TWO CLASSES OF CASES WHERE WE SEE THE MOST PRACTICAL
EFFECT IN TERMS OF WHO IS SENTENCED TO DEATH.
QR. BASED ON YOUR DB-847
A. THAT’S RIGHT, BASED ON THE ANALYSIS OF THE PRACTICAL
CONSEQUENCES OF THE DIFFERENT AGGRAVATING FACTORS. WE SEE WITHIN
THIS CLASS OF CASES, THESE TWO CLASSES OF CASES WHERE WE FIND
APPROXIMATELY NINETY PERCENT OF ALL THE DEATH SENTENCES. THAT
THE RACIAL EFFECTS ARE SOMEWHAT STRONGER WITHIN THIS CATEGORY OF
CASES THAN THEY ARE OVERALL.
Q. WHATS THE SIGNIFICANCE OF THAT FOR YOUR OPINIONS THAT
YOU’VE EXPRESSED EARLIER?
A. WELL, IT SUGGESTS AGAIN THAT THE SYSTEM DOES NOT OPERATE
UNIFORMLY WITH RESPECT TO THESE MATTERS. THE B2 CASES, THE B7
842
BALDUS ~ DIRECT
Ty
CASES ARE IMPORTANT, WHEN THOSE FACTORS EXIST. THE RISK OF A
DEATH SENTENCE GOES UP. THAT’S THE AREA WHERE EXERCISE, WHERE
DISCRETION IS EXERCISED. AND THAT'S THE AREA WHERE YOU SEE THE
RACIAL EFFECTS MORE PROMINENTLY THAN YOU DO IN THE OTHER AREAS.
THE CENTRAL MESSAGE THAT COMES THROUGH IS THE RACE EFFECTS ARE
CONCENTRATED IN CATEGORIES OF CASES WHERE THERE’S AN ELEVATED
RISK OF A DEATH SENTENCE. THERES NO SUGGESTION IN THIS
RESEARCH THAT THERE-‘S A UNIFORM. INSTITUTIONAL BIAS THAT
ADVERSELY AFFECTS DEFENDANTS IN WHITE VICTIM CASES IN ALL
O
0
8
o
N
B
h
W
N
Po
CIRCUMSTANCES, OR A BLACK DEFENDANT IN ALL CASES. THERE’S
fb
Fo
y NOTHING TO SUPPORT THAT CONCLUSION. ITS A VERY COMPLICATED
12 SYSTEM.
13 MR. BOGER: YOUR HONOR. AT THIS POINT I MOVE THE
14 ADMISSION OF DB-8S INTO EVIDENCE.
15 MS. WESTMORELAND: YOUR HONOR, I JUST HAVE THE SAME
1% OBJECTION AS TO THE TWO PREVIOUS EXHIBITS.
17 THE COURT: ALL RIGHT. IT WILL BE ADMITTED IN
18 EVIDENCE.
: 4 19 MR. BOGER: THANK YOU, YOUR HONOR.
20 BY MR. BOGER:
21 Rf. PROFESSOR BALDUS, DID YOU DO ANY OTHER ANALYSES INVOLVING
22 THE B2 NOR B7 AGGRAVATING CIRCUMSTANCES, TO DETERMINE THE
23 RELATIONSHIP BETWEEN DEATH SENTENCING RATES THERE AND RACIAL
24 FACTORS?
843
BALDUS - DIRECT
@. LET ME DIRECT YOU TO DB-846 FOR IDENTIFICATION AN ASK YOU IF
YOU CAN IDENTIFY THAT DOCUMENT?
A. YES. DB-86 PRESENTS THE RESULTS OF A COHORT ANALYSIS. OR
QUALITATIVE ANALYSIS OF THE DATA IN THIS CASE. YOU MAY RECALL
THAT WHEN WE COLLECTED THE DATA ON THESE CASES. WE COMPLETED FOR
EACH CASE A, A NARRATIVE SUMMARY. AND I CONSIDERED IT IMPORTANT
TO DO AN ALTERNATIVE TYPE OF ANALYSIS THAT DID NOT REST ON
STATISTICAL MANIPULATIONS. THAT IT WAS IMPORTANT FOR US TO. TO
LOOK AT THE CASES AND CATEGORIZE THE CASES ACCORDING TO THE
CHARACTERISTICS THAT APPEARED IN THOSE SUMMARIES AND TO TRY AND
SORT THEM IN TERMS OF LEVELS OF COMPARABILITY. HOW TO TRY AND
GROUP THE MORE SERIOUS CASES TOGETHER. THE CASES OF MEDIUM
SERIOUSNESS AND THOSE OF LEAST SERIOUSNESS. AND ESTIMATE RACIAL
EFFECTS WITHIN THOSE SUB-GROUPS OF CASES.
BECAUSE OF THE IMPORTANCE OF THE B2 STATUTORY
AGGRAVATING FACTORS, WE CONDUCTED SUCH AN ANALYSIS ON THE CASES
FROM BOTH THE PROCEDURAL REFORM STURDY AND THE CHARGING AND
SENTENCING STUDY. WE WENT THROUGH THOSE CASES, IDENTIFIED THOSE
CASES THAT HAD A B2 STATUTORY AGGRAVATING FACTOR. AND WE FOUND
SOME 438 CASES ON WHICH WE WERE ABLE TO LOCATE A SUMMARY. THERE
WERE A HANDFUL OF SUMMARIES THAT WE WERE UNABLE TO LOCATE.
WE THEN ADDRESSED THE QUESTION OF HOW TO SORT THESE
CASES INTO COMPARABLE GROUPS, EVEN THOUGH WE HAD A LARGE
POPULATION OF CASES, FOR ALL ELIGIBLE UNDER ONE STATUTORY .
AGGRAVATING FACTOR FOR A DEATH PENALTY. WE RECOGNIZED THAT THERE
a44
BALDUS —- DIRECT
1 WERE IMPORTANT DIFFERENCES BETWEEN THESE CASES, AND DB-84, IF
2 ONE SCANS DOWN THE LEFT-HAND MARGIN, PRESENTS THE SYSTEM THAT WE
USED TO SORT THE CASES INTO SUB-CATEGORIES WHERE THERE WERE
SIMILAR FACTS.
BY MR. BOGER:
3
4
S
A é @. HOW DID YOU DEVELOP YOUR SORTING SYSTEM?
7 A. WE LOOKED AT THE CHARACTERISTICS OF THE CASES THAT WE
3 THOUGHT WERE IMPORTANT IN DETERMINING DEATH ELIGIBILITY. FIRST,
9 WE SORTED ALL THE ARMED ROBBERY CASES, WHICH ARE A DISTINCT AND
10 IMPORTANT CLASS, AND, HOWEVER. WE FELT —-— HERE WE HAD 273 CASES,
11 AND OBVIOUSLY THERE WERE DISTINCTIONS BETWEEN THEM. SO WE
iz FURTHER DIVIDED THOSE CASES. AND A NUMBER OF ARMED ROBBERIES
13 HAD OTHER CONTEMPORANEOUS OFFENSES; A NUMBER HAD MULTIPLE
14 VICTIMS: THOSE WERE OBVIOUSLY FACTORS THAT WOULD TEND TO
1S DISTINGUISH THEM FROM THE RUN OF THE MIND, ARMED ROBBERY,
16 HOMICIDE WITH NO OTHER CONTEMPORANEOUS OFFENSE.
37 SO AS YOU CAN SEE ON TABLE 38B. UNDER SECTION 2H, WE
18 SORTED OUT THE CASES. THOSE WITH KIDNAP, ARSON OR BURGLARY, MORE
R 4 19 THAN ONE VICTIM. WITH OTHER CONTEMPORANEOUS OFFENSES.
20 AND WE GOT GROUPS OF CASES THAT WERE LARGE ENOUGH TOD
21. MAKE SOME REASONABLE COMPARISON. BECAUSE THE NUMBERS OF CASES
22 WERE NOT ANY LARGER IN THESE CATEGORIES, WE COULDN‘T FURTHER
2 SUBDIVIDE THESE CASES ACCORDING TO ADDITIONAL AGGRAVATING AND
24 MITIGATING FACTORS.
23 HOWEVER, WITH THE CATEGORY OF CASES INVOLVING ARMED