Trial Proceedings Transcript Vol. 6
Public Court Documents
August 15, 1983
208 pages
Cite this item
-
Case Files, McCleskey Legal Records. Trial Proceedings Transcript Vol. 6, 1983. f41b19e0-5ba7-ef11-8a69-6045bdd667da. LDF Archives, Thurgood Marshall Institute. https://ldfrecollection.org/archives/archives-search/archives-item/43d1836f-94df-48e7-a977-b212a1b59891/trial-proceedings-transcript-vol-6. Accessed May 14, 2025.
Copied!
Bian
N.U 1)
+
—
IN THE UNITED STATES DISTRICT COURT
FOR THE NORTHERN DISTRICT OF GEORGIA
ATLANTA DIVISION
WARREN MCCLESKEY., ) DOCKET NO. C21-2434A
)
)
PETITIONER, ) ATLANTA, GEORGIA
}
dh 1 ) AUGUST 15, 1983
; )
WALTER D. ZANT. WARDEN, )
}
}
RESPNDANT. )
VOLUME VI
TRANSCRIPT OF PROCEEDINGS
BEFORE THE HONORABLE J. OWEN FORRESTER. UNITED STATES DISTRICT
JUDGE.
APPEARANCES OF COUNSEL:
FOR THE PLAINTIFF: JOHN CHARLES BOGER; TIMOTHY K. FORD
ROBERT H. STROUP.
FOR THE DEFENDANT: MARY BETH WESTMORELAND AND FAULA
K. SMITH.
JIM PUGH
OFFICIAL COURT REPORTER
ROOM 2367. 75 SPRING STREET. S.W.
ATLANTA, GEORGIA 30303
i — iit Soir A —— S———— —————— i ————
O
o
o
W
N
d
N
ee
(W
P
T
E
=
T
T
od
H
Ww
N
e
oO
PY
Nn
HN ITNESSES
WITNESSES FOR THE PLAINTIFF:
BALDUS, DAVID C.
WOODWORTH, GEORGE GORDON
DIRECT CROSS
1036 1090
1193
REDIRECT
11387
1034
RECROSS
O
9
0
N
D
AR
d
W
N
e
P
E
a
[.
%
T
E
PW
R Ww
14
13
DB-114
DB-113
DB-1146
DB-117
DB-118
DB-119
DB-112
DB-110
DB-79A
GW—1
GW-2
GW-3
RESPONDENTS EXHIBITS:
V
O
N
O
C
U
R
W
O
U
N
»
DOCUMENTARY EVIDENCE
MARKED
1037
10352
1033
1062
1049
1071
1073
1089
1193
1210
1228
1110
1111
1112
1128
1173
1173
1175
1175
11735
1176
1176
$3377
1178
1178
1173
1173
1035
RECEIVED
1044
1062
1062
10628
1071
1073
1076
1079
1090
1207
1220
1114
1114
1114
1130
1036
BALDUS - DIRECT
—
(ATLANTA. FULTON COUNTY, GEORGIA: AUGUST 13, 1983,
2 |IN OPEN COURT.)
3 i
4 MS. WESTMORELAND: YOUR HONOR. BEFORE WE BEGIN, I°D
S |JUST LIKE TO INTRODUCT TO THE COURT PAULA SMITH WHO WILL BE AT
a 4 |COUNSEL TABLE WITH ME TODAY.
7 THE COURT: HELLO, MS. SMITH. HOW ARE YOU?
8 MS. WESTMORELAND: A NEW MEMBER OF OUR OFFICE.
9 THE COURT: GLAD TO HAVE HER.
10 ALL RIGHT. MR. BOGER.
11 MR. BOGER: THANK YOU, YOUR HONOR.
12 WE HAVE VERY LITTLE TO FINISH UP ON, BUT I THINK I TOLD
13 YOU WE HAD THIRTY OR FORTY-FIVE MINUTES.
14 IVE GOT A FEW EXHIBITS TO BE ADDED TO THE EXHIBIT
15 BOOK. I HAVE A COPY FOR THE CLERK AND A COPY FOR THE COURT. IF ]
1&4 |CAN HAND THIS OVER AT THIS TIME, AND THAT WILL FACILITATE THE
17 QAUESTIONING OF PROFESSOR BALDUS.
13 THE COURT: ALL RIGHT.
w i7 MR. BOGER: IT PROBABLY WON‘T BE NECESSARY TO PUT THEM
20 IN THE BOOKS RIGHT AT THIS MOMENT. I THINK THEYRE RELATIVELY
£1 TOGETHER.
22 THE COURT: ALL RIGHT.
23 - sw
24 DAVID C. BALDUS,
23 BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
1037
BALDUS - DIRECT
Te
TESTIFIED FURTHER AS FOLLOWS:
2 DIRECT EXAMINATION (CONT IV
3 BY MR. BOGER:
4 @. PROFESSOR BALDUS, WE WERE SPEAKING LAST WEEK ABOUT YOUR
5 |VARIOUS ANALYSES OF FULTON COUNTY HOMICIDE CASES.
a é HAVE YOU DONE ANY ADDITONAL ANALYSES IN FULTON COUNTY
7 WHICH DO CONTROL FOR BACKGROUND FACTORS?
3 A. YES. I HAVE.
? @. LET ME DIRECT YOUR ATTENTION TO WHAT’S BEEN MARKED FOR
10 IDENTIFICATION AS DB-114., I“LL ASK YOU TO IDENTIFY THAT
11 DOCUMENT?
12 A. DB-114 IS A TABLE WHICH PRESENTS THE RESULTS OF A SERIES OF
13 WEIGHTED LEAST SGUARES REGRESSION ANALYSES THAT 1 CONDUCTED AT
14 THE SUCCESSIVE STAGES IN THE FULTON COUNTY CHARGING AND
13 SENTENCING PROCESS.
16 THE PURPOSE OF THE ANALYSIS WAS TO ESTIMATE RACIAL
17 EFFECTS AT THESE STAGES WHILE CONTROLLING FOR A VARIETY OF
18 NON-RACIAL BACKGROUND FACTORS.
w 19 ®R. HOW WERE THOSE FACTORS SELECTED?
20 A. THEY WERE SELECTED IN A STEPWISE REGRESSION PROCEDURE WHICH
21 BRINGS INTO THE ANALYSIS THE NON-RACIAL FACTORS THAT ALSO SHOW A
oi STATISTICALLY SIGNIFICANT RELATIONSHIP WITH THE OUTCOME AT THAT
23 PARTICULAR STAGE IN THE PROCESS.
24 HERE, WE LOOKED AT A SERIES OF STAGES, S00 THE OUTCOME
23 AT ANY ONE STAGE. AT EACH STAGE. RATHER, WAS DIFFERENT.
E
E
E
a
B
S
E
a
B
R
Da
J
ps
Q
—
-
f
y
32
13
1038
BALDUS - DIRECT
THE FIRST ONE WAS THE DECISION AS TO WHETHER THERE
SHOULD BE A MURDER INDICTMENT OR VOLUNTARY MANSLAUGHTER.
THEN AMONG THE CASES INDICTED FOR MURDER. WHETHER THERE
WOULD BE A VOLUNTARY MANSLAUGHTER PLEA ENTERED.
AND AMONG THOSE CASES. THAT DID NOT GO OUT ON VOLUNTARY
MANSLAUGHTER PLEA, WHETHER THERE WOULD BE A MURDER PLEA.
AND THEN IN ADDITION. WE HAD, HAVE A VARIABLE WHICH
ASKS THE MORE GENERAL QUESTION THAT ASKS WHETHER OR NOT THE
PROSECUTOR WAS OFFERED OR AGREED TO ACCEPT A PLEA BARGAIN.
THE NEXT STAGE INVOLVED THE JURY SENTENCING DECISION,
AND THERE WE WERE RESTRICTED BY OUR DATA TO THE QUESTION OF
WHETHER OR NOT THE JURY RETURNED A VERDICT OF GUILT FOR MURDER
OR VOLUNTARY MANSLAUGHTER.
AND FINALLY THE QUESTION WAS WHETHER OR NOT THE CASE
WAS ADVANCED TO A PENALTY TRIAL AFTER A MURDER CONVICTION AT
GUILT TRIAL.
|@. WHAT DOES DB-114 REFLECT ABOUT THE PRESENCE OR ABSENCE OF
RACIAL FACTORS IN THE FULTON COUNTY CHARGING AND SENTENCING
PROCESS?
A. WHAT IT SUGGESTS IS THAT WHEN YOU LOOK AT THE DECISIONS ON
INDICTMENT AND WMEN YOU LOOK AT THE JURY GUILT TRIAL DECISION,
THAT YOU SEE NO RACIAL EFFECTS AT ALL.
HOWEVER. WHEN YOU EXAMINE THE PROSECUTORIAL DECISIONS
TO AGREE TO A PLEA BARGAIN, TO ALLOW A VOLUNTARY MANSLAUGHTER
PLEA TO BE ENTERED, TO ABSTAIN FROM SEEKING A DEATH SENTENCE
re
y
W
N
R
d
W
N
10
24
23
1039
BALDUS - DIRECT
AFTER A MURDER PLEA HAS BEEN ENTERED. AND FINALLY ON THE
GUESTION OF WHETHER THE CASE SHOULD ADVANCE TO A PENALTY TRIAL
AFTER THERE’S BEEN A MURDER CONVICTION. WE SEE STRONG RACE
EFFECTS ESTIMATED IN THESE REGRESSION ANALYSES. THEY’RE
SUBSTANTIALLY STRONGER THAN WHAT WE SEE WHEN WE LOOK AT
COMPARABLE DECISION POINTS STATEWIDE.
R. WHAT'S THE NATURE OF THE RACIAL EFFECT IN THE GUILTY PLEA
PORTION OF YOUR ANALYSIS, FOR EXAMPLE?
A. IN THE GUILTY PLEA PORTION OF THE ANALYSIS, WE FOCUS ON THE
ENTRY OF VOLUNTARY MANSLAUGHTER PLEAS IN CASES WHERE A MURDER
INDICTMENT HAD ORIGINALLY BEEN OBTAINED.
WE FIND THAT DEFENDANTS WITH WHITE VICTIMS HAVE A
NINETEEN PERCENTAGE POINT LOWER ESTIMATED LIKELIHOOD OF HAVING
SUCH A PLEA ENTERED. AND BLACK DEFENDANTS HAVE A SLIGHTLY LOWER
OPPORTUNITY TO DO THAT, BUT IT IS NOT A STATISTICALLY
SIGNIFICANT DIFFERENCE BEYOND THE .0S5 LEVEL. WHICH STANDS IN
CONTRAST TO THE RACE OF THE VICTIM EFFECT. WHICH IS
STATISTICALLY SIGNIFICANT BEYOND THE .05 LEVEL.
WITH RESPECT TO THE PLEA TO MURDER, THE OUTCOME
VARIABLE HERE MEASURES WHETHER A PLEA TO MURDER WAS ENTERED IN A
CIRCUMSTANCE WHERE A DEATH SENTENCE WAS NOT SUBSEQUENTLY SOUGHT.
AND THE RESULT OF THAT ANALYSIS SHOW THAT THE DEFENDANT
WITH A WHITE VICTIM HAS A 29 PERCENTAGE FOINT LOWER LIKELIHOOD
OF ENTERING A PLEA UNDER THOSE CIRCUMSTANCES, AND THAT A BLACK
DEFENDANT HAS A 14 PERCENTAGE POINT LOWER LIKELIHOOD OF ENTERING
Ri NINA. . rt ——— Siren paneer, Sebi. { met, Veen. eee ee. We Stl re. HE a ————,—,—m—_S]e t
1040
BALDUS - DIRECT
pe
A PLEA UNDER THOSE CIRCUMSTANCES.
2 FINALLY, WITH RESPECT TO THE OUTCOME VARIABLE. WHETHER
3 AN OFFER WERE. AGREEMENT WAS MADE AS TO A PLEA BARGAIN. BY A
4 PROSECUTOR. WE SEE RACIAL EFFECTS ALSO.
3S THE QUESTION THAT WAS ADDRESSED IN THE QUESTIONNAIRE
® é HERE, DID NOT ASK WHETHER THERE WAS A PLEA BARGAIN MADE. BUT
YJ WHETHER OR NOT THE PROSECUTOR OFFERED ONE OR EXPRESSED A
8 WILLINGNESS TO AGREE TO ONE. AND YOU CAN SEE IN COLUMNS B AND C
? THAT THE EFFECTS ARE VERY STRONG WITH RESPECT TO BOTH THE RACE
10 OF THE DEFENDANT AND THE RACE OF THE VICTIM.
11 @. SO THE SUBSTANCE OF THAT TESTIMONY IS THAT BLACK DEFENDANT
12 CASES, IN CASES INVOLVING BLACK VICTIM CASES, IN CASES INVOLVING
13 WHITE DEFENDANTS. ARE BEING PERMITTED TO PLEAD OUT AT A
14 DISPROPORTIONATE RATE AT THIS GUILTY PLEA STAGE?
15 A. 1°M SORRY. COUNSEL. WOULD YOU REPEAT THE QUESTION, PLEASE?
16 @. I’M SORRY. YOUR TESTIMONY IN SUMMARY IS THAT BLACK VICTIM
17 CASES AND CASES INVOLVING WHITE DEFENDANTS ARE BEING PERMITTED
is TO PLEAD OUT AT A DISPROPORTIONATE RATE AT THIS GUILTY PLEA
ge 19 STAGE, EVEN CONTROLLING FOR THE BACKGROUND FACTORS?
20 A. YES.
23 QR. AND ARE THESE RESULTS STATISTICALLY SIGNIFICANT?
22 A. WELL, WITH RESPECT TO THESE ANALYSES CONCERNING THE PLEA
23 DECISIONS THERE ARE SIX CO-EFFICIENTS THAT ARE ESTIMATED AND ALL
24 BUT ONE IS STATISTICALLY SIGNIFICANT.
23 THE ONE THAT ISNT IS THIS ONE THAT RELATES TO THE
S
O
S
R
Ti
E
E
E
s
TR
E
T
E
R
E
SE
po
st
oO
Ty
FW
Y
12
13
243
23
1041
BALDUS - DIRECT
VOLUNTARY MANSLAUGHTER PLEA AFTER A MURDER INDICTMENT WITH
RESPECT TO THE RACE OF THE DEFENDANT.
@. IF THESE DISPARITIES EXIST AT THE GUILTY PLEA STAGE. AS YOU
TESTIFIED, WHY DOES YOUR OVERALL FIGURE DOWN IN ROMAN VI, DEATH
SENTENCE GIVEN A MURDER INDICTMENT, WHY IS IT RELATIVELY SMALL?
A. WHEN YOU LOOK AT THE COMBINED EFFECTS OF THE SYSTEM AS
REFLECTED IN ROMAN NUMERAL NUMBER SIX. IT DOESN’T SHOW ANY
RACIAL EFFECT AT ALL.
AND THE, THE LIKELY EXPLANATION FOR THAT IS, I BELIEVE,
THAT WHEN WE’RE ANALYZING THIS ENTIRE POOL OF CASES THAT WE HAVE
A VERY SMALL NUMBER OF DEATH SENTENCES THAT ARE TAKEN OUT OF
THIS ENTIRE POPULATION, THERE ARE ONLY TEN, AND WITH THAT FEW
NUMBER OF DEATH SENTENCES. MULTIPLE REGRESSION ANALYSIS, DOES
NOT PROVIDE A VERY PRECISE ESTIMATE WITH RESPECT TO ANY OF THE
RACIAL CO-EFFICIENTS IN THE ANALYSIS.
THERE'S ANOTHER POSSIBILITY AS WELL, AND THAT IS THAT
THE PENALTY TRIAL DECISION. AS YOU CAN SEE HERE. YOUR HONOR,
THERE WERE ONLY TWENTY CASES AND THAT’S AN INSUFFICIENT NUMBER
OF CASES TO CONDUCT A MULTIPLE REGRESSION ANALYSIS.
SO WE HAD TO LEAVE A BLANK AT THAT STAGE. AND IT MAY BE
THAT THERE IS SOME CORRECTIVE EFFECT AT THAT STAGE IN THE
ANALYSIS.
THE COURT: LOOKS LIKE THERE’S A CORRECTIVE EFFECT AT
LINE, ROMAN NUMERAL FOUR. BUT 1 MAY BE MISINTERPRETING THE
TABLE.
J
re
W
E
Y
S
E
S
R
B
h
SE
R
B
E
E
BU
R
P
O
TE
=
SE
R
N
Vo
M
E
1
S
E
G
BR
oO
re
a
16
17
1042
BALDUS - DIRECT
THE WITNESS: NO. ROMAN NUMERAL FOUR, YOUR HONOR. YOU
SEE A STRONG RACE EFFECT BECAUSE THE CO-EFFICIENTS INDICATE.
CO-EFFICIENT IN "B", ROW 4, SHOWS THAT A DEFENDANT WITH A WHITE
VICTIM HAS A TWENTY PERCENTAGE POINT HIGHER CHANCE, ESTIMATED.
OF HAVING A PENALTY TRIAL THAN SOMEONE WHOSE VICTIM IS NOT
WHITE.
AND THAT A BLACK DEFENDANT, WHEN THESE OTHER FACTORS
ARE CONTROLLED FOR, HAS A HIGHER LIKELIHOOD OF HAVING THEIR CASE
ADVANCED TO A PENALTY TRIAL AS WELL. SO I THINK, YOUR HONCR,
THAT WOULD TEND TO EXACERBATE THE EFFECT. IN FACT, THE, THE
JURY GUILT TRIAL DECISION, SHOWS NO EFFECTS AS WELL AND THAT MAY
TEND VERY WELL TO MINIMIZE THE EFFECT.
S0 THE JURY HERE SEEMS, IT COULD BE THAT THE JURY IS
SERVING A PROPHYLACTIC FUNCTION HERE IN MINIMIZING THESE
EFFECTS, AND THAT, THATS WHY WE DON’T SEE ANY OVERALL EFFECTS.
IT’S FOR THIS VERY REASON THAT WE FOCUSED HEAVILY ON
THE NEAR NEIGHBORS, BECAUSE THE NEAR NEIGHBORS ANALYSIS ENABLES
YOU TO CONCENTRATE ON THE CASES WHERE DISCRETION IS BEING
EXERCISED, AND IT WAS OUR BELIEF THAT YOU COULD GET A BETTER
SENSE OF THE COMBINED EFFECTS OF THE SYSTEM BY LOOKING AT THOSE
NUMBERS THAN YOU COULD BY THE STATISTICAL ANALYSIS.
BUT WITH RESPECT TO THESE DECISIONS IN THE EARLY STAGES
oF THE PROCESS, WE HAVE REASONABLE SAMPLE SIZES. AND WE HAVE A
SUBSTANTIAL NUMBER OF BOTH YES AND NO DECISIONS WHICH ENABLE US
TO HAVE MORE CONFIDENCE IN THE CO-EFFICIENTS THAT ARE ESTIMATED.
1043
BALDUS - DIRECT
ol
e @. SO YOUR LEVEL OF CONFIDENCE WITH RESPECT TO WHAT’S
HAPPENING AT THE GUILTY PLEA STAGE IN THE DECISIONS ADVANCED TO
A PENALTY TRIAL IS, HOW WOULD YOU CHARACTERIZE IT?
A. WELL, I WOULD SAY WE HAVE A FAIR DEGREE OF CONFIDENCE IN THE
RESULTS OF THESE REGRESSIONS BECAUSE WE HAVE REASONABLE SAMPLE
S1ZE, AND THERE IS A GOOD DISTRIBUTION IN TERMS OF THE OUTCOME
VARIABLES, THAT 1S. THERE ARE A GOOD NUMBER OF YES ANSWERS AND
THERE IS A GOOD NUMBER OF NO ANSWERS, WHICH IMPROVES THE
PRECISION OF THE ESTIMATES FOR ALL THE CO-EFFICIENTS IN THE
Pa
y O
O
T
Y
B
o
s
ee
A
N
N
ANALYSIS.
P
y
r
y
I MUST POINT OUT HOWEVER. EVEN WITH RESPECT TO THESE
pb
[3
DECISIONS. THE SAMPLE SIZES ARE NOT MASSIVE. I CAN JUST
13 INDICATE WHAT THEY ARE.
14 WITH RESPECT TO PART 2A, WE HAVE 162 ACTUAL CASES THAT
13 WE WERE WORKING WITH. |
16 BY THE TIME YOU GET DOWN TO THE MURDER PLEA, WE“VE
17 DROPPED TO 77 CASES.
18 WHEN YOU LOOK AT THE COMBINED EFFECTS, THAT IS. WHEN
w 19 YOU LOOK AT PART 2C, WE HAD INFORMATION AS TO WHETHER OR NOT
20 THERE WAS A PLEA OFFERED OR AGREED TO ON 92 CASES.
21 THEN FINALLY WHEN WE GET DOWN TO PART 4, WE HAVE ONLY
22 FORTY-TWO CASES THAT WERE WORKING WITH.
23 SO I MUST NOTE THAT WE DON’T HAVE THE KINDS OF SAMPLE
24 SIZES HERE THAT WE HAVE WHEN WE’RE LOOKING AT THE WHOLE 3TATE,
23 BUT IT“S ENOUGH TO GIVE YOU SOME FLAVOR FOR WHAT IS HAPPENING.
1044
BALDUS - DIRECT
pb
AND THESE CO-EFFICIENTS ARE VERY LARGE. IT“S NOT AS IF WE'RE
2 DEALING WITH SMALL CO-EFFICIENTS, THESE ARE SUBSTANTIAL.
3 SO THAT'S WHAT LEADS ME TO, TO BELIEVE THAT WHAT YOU'RE
4 SEEING HERE IS EVIDENCE OF A REAL EFFECT.
3S MR. BOGER: YOUR HONOR, AT THIS TIME, I MOVE THE
& ADMISSION OF DB-114 INTO EVIDENCE.
7 THE COURT: IT WILL BE ADMITTED.
=] MR. BOGER: THANK YOU. YOUR HONOR.
? THE COURT: I TAKE IT THAT THESE EFFECTS THAT WE SEE IN
10 THE SECTION, ROMAN NUMERAL 2, ARE EFFECTS. WELL. I KNOW THEY
11 ARE, BUT LET ME MAKE SURE, EFFECTS FOR MANY MANY CASES THAT DID
12 NOT END UP IN ADVANCING TO THE PENALTY TRIAL PHASE.
13 THE WITNESS: YES. SIR. THAT’S RIGHT.
14 THE COURT: WHAT WE HAVE, EVEN IF WE DID AWAY WITH
13 CAPITAL PUNISHMENT WE WOULD HAVE THAT SORT OF EFFECT
16 OPERATING IN LOTS OF CASES.
17 THE WITNESS: IF THE SYSTEM CONTINUED AS IT WAS DURING
18 THIS PERIOD. I WOULD THINK THAT WOULD BE CORRECTM YOUR HONOR.
w 19 THE COURT: IF 1 REMEMBER CORRECTLY. YOU HAD ABOUT
20 TWENTY CASES THAT ADVANCED TO THE PENALTY STAGE. IS THAT RIGHT?
23 THE WITNESS: IN THIS CASE. YES, IN FULTON COUNTY.
22 THATS RIGHT. YOUR HONOR.
23 THE COURT: ALL RIGHT. LET ME SEE IF I HAVE ANOTHER
24 GUESTION. |
25 OH, ON THE RACE OF THE VICTIM. YOU SAID THAT YOU DID
Tn
EE
R
B
E
R
A
N
S
l
F
E
So
a
E
E
C
EE
pb
1045
BALDUS - DIRECT
CONTROL FOR SOME BACKGROUND FACTORS.
FROM WHAT I HAVE HEARD PREVIOUSLY, I WAS LED TO BELIEVE
THAT THE STATUS OF THE VICTIM AS A STRANGER OR NON-STRANGER WAS
IMPORTANT IN THE WHOLE DECISION MAKING PRODUCTS.
DO YOU KNOW IF THAT IS ONE YOU CONTROLLED FOR IN THIS
PARTICULAR PROCESS?
THE WITNESS: I HAVE A LIST HERE, YOUR HONOR, OF THE
VARIABLES THAT WERE CONTROLLED FOR IN THOSE ANALYSES.
WITH RESPECT TO THE FIRST ONE. THE VICTIM --
BY MR. BOGER:
@. PROFESSOR BALDUS, WHEN YOU SAY THE FIRST ONE, WHAT DO YOU
MEAN?
A. WITH RESPECT TO THE ANALYSIS CONCERNING 2A, WHETHER THERE
WAS A VOLUNTARY MANSLAUGHTER PLEA ACCEPTED, THE VARIABLE THAT
SPEAKS DIRECTLY TO THAT QUESTION DID NOT ATTAIN A LEVEL OF
STATISTICAL SIGNIFICANCE IN THIS ANALYSIS. HOWEVER, THERE WERE
OTHER VARIABLES THAT. THAT ARE GOOD PROXIES FOR IT. THAT IS,
THAT THE KILLING AROSE FROM A DISPUTE OR FIGHT BETWEEN THE
VICTIM AND THE DEFENDANT. THERE HAD BEEN AN EARLIER ASSAULT ON
THE DEFENDANT BY THE VICTIM AT SOME EARLIER POINT IN TIME,
SUGGESTING A RELATIONSHIP. THAT THE DEFENDANT HAD BEEN ACCUSED
OF SOME SORT OF MISCONDUCT ON THE PART OF THE VICTIM. ALSO
SUGGESTING THAT THERE WAS SOME SORT OF PRIOR RELATIONSHIP.
THOSE ARE THE 3 VARIABLES IN THIS ANALYSIS, YOUR HONOR,
\
THAT I THINK CAPTURE THAT ESSENCE. THAT RELATIONSHIP. BUT THAT
1046
BALDUS - DIRECT
ob
VARIABLE PER SE, STRANGER DID NOT COME INTO THE MODEL, GOES
DIRECTLY TO THAT QUESTION. AND MY BELIEF WOULD BE THAT AFTER
THESE VARIABLES ENTERED THE MODEL. THAT THE VARIABLE THAT SPEAKS
DIRECTLY TO THAT QUESTION DID NOT ADD A SIGNIFICANTLY IMPORTANT
AMOUNT OF ADDITIONAL EXPLANATORY POWER. BECAUSE ITS MOSTLY
RELATED TQ THESE 3 VARIABLES.
@. LET ME JUST CLARIFY YOUR TESTIMONY ON THAT. PROFESSOR
BALDUS.
b
o
m
.
N
N
A
e
.
W
N
UNDER THE STEPWISE REGRESSION PROCEDURE. DID YOU MAKE
po
s o THE SELECTION OF WHICH BACKGROUND VARIABLES WOULD BE HELD
po
o
ro
y CONSTANT?
12 A. NO, I MADE NO JUDGMENT AT ALL. IT WAS STRICTLY WHETHER OR
13 NOT ANY VARIABLE SHOWED IN THE ANALYSIS A LEVEL OF SIGNIFICANCE
14 OF .10 WITH THE OUTCOME VARIABLE. AND THAT“S A COMMON
13 PHENOMENON IN THIS KIND OF REGRESSION ANALYSIS. TWO VARIABLES
16 ARE RELATED TO ONE OTHER. AND HAVE A CORRELATION BETWEEN ONE
17 ANOTHER, AND ONE OF THEM GOES IN, THEN THAT REDUCES
18 SUBSTANTIALLY THE LIKELIHOOD THAT THE OTHER ONE IS GOING TO COME
wo 19 INTO THE ANALYSIS. THAT’S MY OPINION AS TO WHAT HAPPENED
20 HERE WITH RESPECT TO THE VARIABLE "STRANGER."
21 @. HOW MANY BACKGROUND VARIABLES DID YOU USE AT THIS 2A STAGE?
22 A. THERE WERE 23 BACKGROUND CONTROL VARIABLES AT THIS STAGE IN
23 THE ANALYSIS.
24 THE COURT: HOW ABOUT IY, ROMAN NUMERAL FOUR?
23 THE WITNESS: OH, YES, ROMAN NUMERAL FOUR, AT THIS
o
e
B
E
"
Ca
ls
J
N
O
E
E
SH
EE
©
RE
E
S
E
pu
t Oo
a
—-
24
23
1047
BALDUS - DIRECT
STAGE IN THE ANALYSIS. YOUR HONOR, IT WAS NECESSARY TO EXERCISE
A JUDGMENT. IT WAS BASED ON TECHNICAL CONSIDERATIONS.
WITH THE SAMPLE SIZE OF THIS MAGNITUDE. FORTY-TWO
CASES, IF WE ALLOW THE ANALYSIS TO CONTINUE TO PROCEED ITS
NATURAL COURSE, IT WILL INCLUDE A SUBSTANTIAL NUMBER OF
VARIABLES, AND BY VIRTUE OF THE LARGE NUMBER OF VARIABLES
INCLUDED, IT WILL INDICATE A PERFECT FITTING MODEL, GIVE YOU AN .
R-SQUARE OF 1.0 AND SUGGEST THAT EVERY VARIABLE IN THE ANALYSIS
IS STATISTICALLY SIGNIFICANT AT A VERY HIGH LEVEL.
IT BECOMES VERY UNRELIABLE IF YOU ALLOW ALL THOSE
FACTORS TO GO INTO THE ANALYSIS.
SO I CONDUCTED THE ANALYSIS USING A REDUCED SET, THAT
1S, I STOPPED IN THE ANALYSIS, FIRST. AT STAGE 13, STEP NUMBER
13, AND AT THAT STAGE, WE HAD 13 NON-RACIAL BACKGROUND FACTORS
IN THE ANALYSIS.
BECAUSE OF MY CONCERN FOR THIS CHARACTERISTIC OF THE
REGRESSION ANALYSIS, BECAUSE THAT THE PROPERTY SOMETIMES
REFERRED TO AS OVERFITTING, GET TOO MANY VARIABLES IN THERE. IT
REDUCES THE CONFIDENCE YOU HAVE IN THE CO-EFFICIENTS, I WENT
DOWN TO EVEN SMALLER COMBINATIONS. SUB-SETS OF THESE FACTORS.
NINE AND SIX, AND THE RESULTS OF THOSE WERE THE SAME IN EACH
CIRCUMSTANCE.
@. WHAT WERE THE NINE. IF YOU WOULD?
A. OH: YES, THE MINE =
THE COURT: WELL, I UNDERSTAND ROMAN FOUR, IS STILL ON
1048
BALDUS - DIRECT
Ty
13, IS THAT RIGHT?
2 MR. BOGER: I‘M SORRY?
3 THE WITNESS: RIGHT. YOUR HONOR. THAT WAS THE
4 |STATUTORY AGGRAVATING FACTOR Bé, MURDER FOR HIRE.
5 STATUTORY AGGRAVATING FACTOR B-10. WHICH IS TO
R iA 6 |ELIMINATE A WITNESS.
7 MUTILATION.
8 WHETHER THE VICTIM WAS A HOSTAGE.
9 NO VICTIM PROVOCATION.
10 THE MONTHS THE DEFENDANT HAD SPENT IN JAIL.
11 THE PRESENCE OF AN EXECUTION IN THE CASE.
12 EVIDENCE OF AN INCRIMINATING REMARK HAVING BEEN
13 |OVERMEARD BY THE DEFENDANT OR CO-PERPETRATOR.
14 A POLICE WITNESS OF THE DEFENDANT AT OR NEAR THE CRIME.
15 THE NUMBER OF PRIOR FELONIES.
16 THE NUMBER OF NON-STATUTORY MAJOR AGGRAVATING FACTORS
17 IN THE CASE.
18 AND WHETHER OR NOT THE CASE INVOLVED A HATE MOTIVE.
w 19 THOSE ARE THE FACTORS THAT WERE INCLUDED IN THAT
20 ANALYSIS, AS BACKGROUND CONTROLS.
21 BY MR. BOGER:
22 R. AND AT THAT STAGE, CONTROLLING FOR THOSE FACTORS, YOU HAD A
23 TWENTY PERCENT DISPARITY IN THE DECISION OF THE PROSECUTOR TO
24 MOVE A CASE PAST CONVICTION TO THE PENALTY PHASE?
23 A. YES.
un
0
H
N
e
104%
BALDUS - DIRECT
@. ALL RIGHT.
EXCUSE ME, THAT’S RACE OF VICTIM DISPARITY. AND YOU
HAD A NINETEEN. .19 DISPARITY ON RACE OF DEFENDANT?
A. YES, .19 FOR THE RACE OF DEFENDANT CO-EFFICIENT.
?@. NOW. PROFESSOR BALDUS, YOU INDICATED BECAUSE OF THE SMALL
SAMPLE SIZE YOU CONSIDERED ANOTHER ANALYSIS MORE IMPORTANT WITH
RESPECT TO YOUR JUDGMENT OF THE OVERALL PRESENCE OR ABSENCE OF
RACIAL EFFECTS IN THE SYSTEM.
WHAT WAS THAT?
A. YES. THAT WAS WHAT WE HAD REFERRED TO HERETOFORE AS THE
NEAR NEIGHBORS ANALYSIS WHEREBY WE USED THE RESULTS OF A
REGRESSION ANALYSIS TO IDENTIFY A CLASS OF CASES THAT APPEARED
TO BE ROUGHLY COMPARABLE TO WARREN MCCLESKEY IN TERMS OF THE
PREDICTED LIKELIHOOD OF A DEATH SENTENCE.
THEN WE EXERCISED A JUDGMENT AS TO THE LEVELS OF
RELATIVE AGGRAVATION AMONG THOSE CASES. AND WE CREATED THREE
CATEGORIES OF CASES.
AND LAST FRIDAY I TESTIFIED TO DB-109 WHICH SHOWS THE
CONSEQUENCES OF THAT ANALYSIS WHICH SORTED THE CASES, FIRST BY
THE MACHINE. AND THEN ON THE BASIS OF OUR JUDGMENT OF WHAT THE
APPROPRIATE CULPABILITY LEVELS RELATIVE TO MCCLESKEY WERE AMONG
THOSE 32 CASES.
RQ. WITHOUT ASKING YOU TO REPEAT AT LENGTH YOUR ANALYSIS. CAN
YOU TELL US WHAT YOU CONCLUDE FROM THIS COMBINATION OF ANALYSIS
OF DB-114 AND THE NEAR NEIGHBORS ANALYSIS WHICH IS REFLECTED IN
1050
BALDUS ~ DIRECT
po
de
DB-1097?
A. MY CONCLUSION IS THAT WITH RESPECT TO THE CASES WHERE
THERES A REAL RISK OF A DEATH SENTENCE. YOU DO SEE RACIAL
EFFECTS.
R. IN FULTON COUNTY?
A. IN FULTON COUNTY, YOU SEE RACE OF VICTIM EFFECTS.
THERE ARE, FULTON COUNTY HAS A VERY LOW DEATH
SENTENCING RATE. AS A CONSERUENCE., THERE ARE NOT THAT MANY
TH
E
CY
T
E
E
N
a
T
CASES THAT ARE GENUINELY AT RISK FOR A DEATH SENTENCE IN FULTON
10 COUNTY.
11 AND OUR ANALYSIS SUGGESTED IT WAS A POPULATION OF
12 APPROXIMATELY 30-SOME PEOPLE OUT OF MANY MANY HUNDREDS OF
13 HOMICIDE OFFENDERS THERE ARE HERE, AND IT IS MY OPINION THAT
14 THAT TOOK PLACE WHERE YOU SEE THE COMBINED EFFECTS OF THE SYSTEM
13 AS IT RELATES TO PEOPLE WHO ARE AT RISK WITH RESPECT TO CAPITAL
146 PUNISHMENT.
17 WHEN YOU LOOK AT THE, WELL. THAT’S IT.
18 @. PROFESSOR BALDUS. AS YOU KNOW. WARREN MCCLESKEY WAS
db 19 CONVICTED AND SENTENCED TO DEATH FOR HOMICIDE AGAINST A POLICE
20 OFFICER.
21 DID YOU LOOK TO DETERMINE THE NUMBER AND CHARACTER OF
22 THE POLICE OFFICER VICTIM HOMICIDE CASES IN FULTON COUNTY?
23 A. YES. I DID.
24 @. DO YOU RECALL HOW MANY DEFENDANTS YOU LOCATED WHO HAD KILLED
23 POLICE OFFICERS IN FULTON COUNTY?
A —— li AAR SAI STE SABI AAAI ofA. AAI
a
N
O
A
A
L
N
ee
IN
oO
oe
Py
12
13
1031
BALDUS - DIRECT
a.
A. THIS SURVEY OF POLICE OFFICER CASES, IN FULTON COUNTY.
PRODUCED THE FOLLOWING COUNT.
WE FOUND TEN CASES INVOLVING POLICE OFFICER VICTIMS OR
RATHER WE FOUND A TOTAL OF TEN POLICE OFFICER VICTIMS IN FULTON
COUNTY UNDER THE STATUTE THAT WAS ENACTED ON MARCH 28, 1973.
AMONG THOSE CASES WERE 18 OFFENDERS ASSOCIATED WITH
THOSE HOMICIDES IN ONE WAY OR ANOTHER.
OF THOSE 18 OFFENDERS. I MENTION IN PASSING, 18
OFFENDERS WHO WERE NOT KILLED BY THE POLICE IN THE COURSE OF A
GUNFIGHT OR SOMETHING LIKE THAT, THOSE WHO SURVIVED, OF THOSE
28, THERE WERE 17 DISPOSITIONS.
THERE WAS ONE CASE WHERE THE POLICE RECORD THAT WE
LOOKED AT SUGGESTED THAT THE PERSON WAS MENTALLY DERANGED AND
THERE WAS NO DISPOSITION IN THE CASE. I DON‘T KNOW WHAT
HAPPENED. BUT BY VIRTUE OF THAT CHARACTERISTIC OF THE OFFENDER,
IT DOES NOT SEEM APPROPRIATE TO ME TO INCLUDE IN THIS ANALYSIS.
SO THAT LEFT US WITH 17 OFFENDERS WHO WERE POTENTIALLY
COMPARABLE TO MCCLESKEY.
HOWEVER, A -——
R. LET ME ASK YOU BEFORE YOU GO FURTHER. YOU“VE GOT 17
DEFENDANTS, AND YOU SAID YOU HAD 10 POLICE HOMICIDES. JUST TO
BEGIN WITH THE SUMMARY IN EFFECT. HOW MANY DEATH SENTENCES
RESULTED FROM THIS NUMBER?
A. ONE.
R. WHO IS THAT?
1052
BALDUS - DIRECT
1 A. THAT WAS MCCLESKEY’S.
2 @. DID YOU CONDUCT ANY ANALYSIS TO DETERMINE WHY MCCLESKEY’S
3 CASE HAD ADVANCED TO A DEATH SENTENCE, AND OTHERS DROPPED OUT
4 EARLIER?
5 A. WELL, I CAN TELL YOU, COUNSEL, I CAN TELL YOU THE
a 6 DISPOSITION OF THESE CASES.
7 RA. FINE?
8 A. AS THEY WORKED THEIR WAY THROUGH THE SYSTEM.
£4 THE COURT: AM I LOOKING AT DB-1137
10 MR. BOGER: PERHAPS WE COULD FOLLOW THROUGH DB-1135.,
11 YES. YOUR HONOR.
12 BY MR. BOGER:
13 @. IF YOU COULD IDENTIFY THAT DOCUMENT, PROFESSOR BALDUS?
14 A. YES. DB-115 IS A FOOTNOTE FROM OUR REPORT WHICH LISTS THE
135 17 OFFENDERS. AND TEN POLICE OFFICER VICTIMS OF THAT. THAT THIS
16 ANALYSIS RELATES TO.
17 @. AND DOES IT ACCURATELY LIST THOSE?
18 THE COURT: JUST TO CLARIFY THE RECORD, THE VICTIMS
Ww 19 HERE ALL DIED AS A RESULT OF WHATEVER THE OFFENDERS DID?
20 THE WITNESS: YES, YOUR HONOR.
21 THE COURT: OKAY.
22 BY MR. BOGER:
23 @. AND YOU'VE ACCURATELY EXAMINED THIS LIST AND TO THE BEST OF
24 YOUR KNOWLEDGE IT’S TRUE?
23 A. YES.
O
N
F
l
R
e
a
y
N
e
Po
y
pb
Fy
1033
BALDUS - DIRECT
@. IF YOU COULD TAKE US THROUGH THE DISPOSITION OF THESE CASES?
A. YES. OUR ANALYSIS FOCUSED ON THE CASES THAT WE THOUGHT
WOULD BE, WERE MOST COMPARABLE TO WARREN MCCLESKEYS.
THAT MEANT THAT WE WANTED TO LOOK AT CASES WHERE THE
DEFENDANT WAS A TRIGGER MAN, AND THE CASE ALSO INVOLVED A
SERIOUS CONTEMPORANEOUS OFFENSE. BECAUSE ACCORDING TO THE
STANDARD THAT WE’VE BEEN USING BEFORE FOR A, CLASSIFYING CASES,
WE HAD VIEWED AS LESS CULPABLE THAN MCCLESKEY, CASES IN WHICH
THE OFFENDER WAS NOT THE TRIGGER MAN.
SO WE SORTED THE CASES INTO TWO CATEGORIES AND THEY
PRODUCED A POOL OF SEVEN CASES IN WHICH THERE WAS A TRIGGER MAN,
INCLUDING MCCLESKEY, WHO WAS ALSO INVOLVED IN A SERIOUS
CONTEMPORANEOUS OFFENSE.
"AND THOSE SERIOUS CONTEMPORANEOUS OFFENSES WERE
BURGLARY, ARMED ROBBERY. AND IN ONE CASE THE SHOOTING OF ANOTHER
PERSON IN THE COURSE OF THE HOMICIDE.
THE, OF THOSE, LET ME MAKE ONE CORRECTION THERE.
THATS SEVEN CASES IN ADDITION TO MCCLESKEY, BY MY COUNT HERE.
@. PROFESSOR BALDUS, LET ME DIRECT YOUR ATTENTION TO DB-116
MARKED FOR IDENTIFICATION, AND ASK YOU IF YOU CAN IDENTIFY THAT
AS PART OF YOUR DISCUSSION OF THE DISFOSITION OF THESE CASES?
A. THIS IS DB-1167
QA. THATS RIGHT?
A. YES. NUMBER 11é6 LISTS THE OFFENDERS. THE FIRST PAGE OF 116
BREAKS THEM DOWN INTO THE CATEGORIES THAT 1 JUST DESCRIBED.
BALDUS - DIRECT
1 BUT FIRST AT THE TOP IT LISTS ALL 17 OFFENDERS. AND
2 SHOWS THE RACE OF VICTIM AND RACE OF DEFENDANT CHARACTERISTICS
OF THE CASES. AND THE OUTCOME WITH RESPECT TO A LIFE OR DEATH
SENTENCE.
THE SECOND ROW AND THIRD ROW INDICATE THE MORE OR LESS
“
4
5
® 6 AGGRAVATED CASES AS CLASSIFIED IN THE FASHION I JUST DESCRIBED.
7 AND IN THE SUCCEEDING PAGES PRESENT A. WORKING PAPERS
8 THAT SHOW RACE OF VICTIM AND RACE OF DEFENDANT DISPARITIES AMONG
? THESE CASES.
1Q AND THEN FOLLOWING THAT ARE THE POLICE RECORDS THAT
11 WERE USED TO COMPILE THIS TABULATION, AS WELL AS SUMMARIES FROM
: THE BACK. RATHER AT THE BACK OF THIS EXHIBIT, YOUR HONOR, YOU-LL
13 FIND SUMMARIES OF THE TYPE YOU’VE SEEN BEFORE WE GENERATED IN
14 OUR STUDY BECAUSE OF THE 17 DEFENDANTS, THERE WERE FOUR WHO
13 HAPPENED TO BE IN OUR STUDY. SO WE INCLUDED THEIR SUMMARIES
16 HERE AS WELL.
17 SO, TO RETURN, IF I COULD, TO THE DISPOSITION OF THESE
18 SEVEN OFFENDERS, YOU CAN SEE AT THE BEGINNING OF DB—-114, THERE
w» 19 ARE SEVEN PEOPLE, NOT INCLUDING MCCLESKEY. WHO WERE INVOLVED IN
20 A SERIOUS CONTEMPORANEOUS OFFENSE, AND THE TRIGGER MAN.
21 OF THOSE SEVEN OFFENDERS, THREE OF THEM PLED GUILTY TO
22 MURDER, AND HAD NO PENALTY TRIAL FOLLOWING THE GUILTY PLEA.
23 AN ADDITIONAL TWO WENT TO TRIAL ON MURDER CHARGES AND
284 |WERE CONVICTED AND THERE WAS NO PENALTY TRIAL HELD THEREAFTER.
25 TWO OTHER OFFENDERS WENT TO TRIAL ON MURDER CHARGES,
—-
E
E
E
R
T
S
E
L
l
+
L
e
R
E
L
1033
BALDUS - DIRECT
AND THEY WERE CONVICTED AND HAD A PENALTY TRIAL. AND OF THOSE
TWO PENALTY TRIALS. MY RECORDS SHOW THAT ONE RECEIVED THE DEATH
SENTENCE AND ONE RECEIVED A LIFE SENTENCE.
@. AND THE DEATH SENTENCE WAS?
A. MCCLESKEY’S.
@. S0 OUT OF THE SEVEN, FIVE EITHER PLEADED GUILTY TO MURDER OR
WERE CONVICTED AT TRIAL OF MURDER, AND DID NOT ADVANCE TO A
PENALTY PHASE. IS THAT CORRECT, SIR?
A. THAT'S WHAT MY RECORDS SHOW, YES.
Q. AND THE JURY DISPOSED OF ONE CASE AT THE PENALTY PHASE WITH
A LIFE SENTENCE AND ONE WITH A DEATH SENTENCE?
A. YES.
@. WHAT DID YOU CONCLUDE FROM THIS ANALYSIS OF THE POLICE
VICTIM HOMICIDE CASES IN FULTON COUNTY WITH RESPECT TO RACIAL
IMPACT?
A. WELL, FIRST OF ALL, WITH RESPECT TO THE OVERALL RACE EFFECT
IN THESE CASES, IT“S HARD TO DRAW ANY INFERENCE BECAUSE WE ONLY
HAVE ONE DEATH SENTENCE. WE HAVE A LOT OF WHITE VICTIM. OFFICER
VICTIM CASES, WHERE THE DEFENDANT PLED OUT, SO THERE IS NOT AS
SUBSTANTIAL DISPARITY IN TERMS OF THE DEATH SENTENCING RATES IN
THESE TWO POPULATIONS. THERE-“S ONLY ONE DEATH SENTENCE. SO THE
RATE IS LOW IN BOTH SETS OF CASES. SO YOU DON’T SEE A STRONG
OVERALL EFFECT, NUMBER ONE.
THE ONLY HINT ONE MIGHT GET FROM THIS IS THAT IN THE
OTHER PENALTY TRIAL CASE THAT WAS HELD WHERE A LIFE SENTENCE WAS
BALDUS - DIRECT
1 RETURNED, THE VICTIM IN THAT CASE WAS A BLACK VICTIM. BUT WE“RE
2 DEALING THERE WITH A SAMPLE OF TWO CASES AND CERTAINLY ONE CAN’T
3 MAKE VERY MUCH OF THAT, EVEN THOUGH IT IS CONSISTENT WITH THE
4 OTHER EVIDENCE AS TO WHAT WAS GOING ON IN THE EARLIER STAGES.
5S THE, THE PRINCIPAL CONCLUSION THAT ONE IS LEFT WITH IS
6 THAT. THAT THIS DEATH SENTENCE THAT WAS IMPOSED IN MCCLESKEY”’S
7 CASE IS NOT CONSISTENT WITH THE DISPOSITION OF CASES INVOLVING
8 POLICE OFFICER VICTIMS IN THIS COUNTY. THAT THERE WERE A NUMBER
? OF OTHER CASES THAT WERE COMPARABLE IN TERMS OF HAVING,
10 INVOLVING A SERIOUS CONTEMPORANEOUS OFFENSE AND INVOLVING THE
11 DEFENDANT AS A TRIGGER MAN WHERE FOR A VARIETY OF REASONS.
12 PRINCIPALLY DECISIONS BY PROSECUTORS. THE CASE DID NOT RESULT IN
13 A DEATH SENTENCE.
14 S0 I CAN SAY THAT THE PATTERN OF DECISION IN THIS
13 COUNTY IS THAT THERES A VERY LOW DEATH SENTENCING RATE, AND IN
16 OFFICER CASES. IN FACT, THE ONLY ONE IS THE PETITIONER IN THIS
17 PROCEEDING.
18 SO, THIS, BY THE WAY, IS CONSISTENT STATEWIDE. THERES
@ 19 NOT AN IMPORTANT. TERRIBLY HIGH EFFECT ASSOCIATED WITH HAVING A
20 POLICE OFFICER VICTIM STATEWIDE. AS WELL.
21 AND WHEN YOU LOOK AT THAT FACT. THAT KILLING A POLICE
22 OFFICER STATISTICALLY DOES NOT SHOW AN IMPORTANT IMPACT EITHER
23 IN FULTON COUNTY OR STATEWIDE, YOU ALSO EXAMINE THE FACT THAT
24 THE OVERALL DEATH SENTENCING RATE IN THIS COUNTY IS LOW, VERY
23 LOW, IT’S LOW EVEN AMONG THE MORE AGGRAVATED CATEGORIES OF CASES
GO
0
N
O
U
O
N
we
po
t oO
F
y
[i
s
1057
BALDUS - DIRECT
IN THIS JURISDICTION, AND YOU TAKE INTO ACCOUNT THAT EVEN IN B2
CASES, INVOLVING A SERIOUS CONTEMPORANEOUS OFFENSE. THE DEATH
SENTENCING RATE IS LOW IN FULTON COUNTY, EVEN AMONG THE MORE
AGGRAVATED CASES.
WHAT THIS SUGGESTS TO YOU IS THAT MCCLESKEY’S CASE
FALLS IN THIS GREY AREA WHERE LIABILITY AND CULPABILITY ARE NOT
OVERPOWERING, AND THIS IS THE KIND OF CASE WHERE OUR RESEARCH
AND THE LITERATURE CONCERNING THE LIBERATION HYPOTHESIS SUGGESTS
THAT YOU WOULD FIND THE GREATEST LIKELIHOOD THAT SOME
INAPPROPRIATE CONSIDERATION MAY HAVE COME TO BEAR ON THE
DECISION.
IN AN ANALYSIS OF THIS TYPE. OBVIOUSLY ONE CANNDT SAY
THAT WE CAN SAY TO A MORAL CERTAINTY WHAT IT WAS THAT INFLUENCED
THE DECISION. WE CAN‘T DO THAT.
WHAT WE DO KNOW IS WHAT THE OVERALL PATTERN WAS. THE
OVERALL RATES, THE TYPE THAT INVITE A WIDE RANGE IN THE EXERCISE
OF DISCRETION IN THIS KIND OF CASE. AND THERES NOTHING THAT
LOOMS IN MCCLESKEY‘S RECORD THAT WOULD CLEARLY DISTINGUISH HIS
CASE FROM FROM THESE OTHER CASES.
IN A CIRCUMSTANCE LIKE THAT. IT SEEMS, IT IS MY OPINION
THAT A RACIAL FACTOR COULD HAVE BEEN THE CONSIDERATION THAT
TIPPED THE SCALE AGAINST MCCLESKEY IN HIS CASE.
THE COURT: LET ME ASK YOU A QUESTION, IF I MAY.
YOU“VE GOT FIVE CASE CANDIDATES FOR TESTING THAT
THESIS, AND THAT IS THAT RACE IS A MOTOR WHICH IS DRIVING THE
F
o
SEE
|
B
T
N
S
J
E
B
a
BE
E
C
D
h
e
e
s
iW
N
o
r
e
o
O
pant. imi. 5. ten. Abt, eb eet. eel tetas re
1053
BALDUS - DIRECT
SYSTEM.
IS THERE ANYTHING ABOUT THE MACK CASE. THE WALLS CASE.
THE MOHAMED CASE OR THE BOBO CASE FACTUALLY. NOT STATISTICALLY.
BUT FACTUALLY. THAT COULD EXPLAIN WHY WE DID NOT OBSERVE THE
MOTIVE POWER OF RACE IN THOSE CASES? IF YOU KNOW.
THE WITNESS: NO» I DON’T KNOW. IF YOU -- AM I
CORRECT, YOUR HONOR, YOU‘RE SUGGESTING THAT SINCE THEY ARE
INDISTINGUISHABLE IN TERMS OF THE RACIAL DIMENSION OF THE CASE.
RIGHT. WHY WOULD YOU EXPECT TO SEE IT IN THE ONE CONTEXT OR THE
OTHER, THAT, THAT IS SOMETHING I CAN‘T EXPLAIN. THAT'S WHY MY
TESTIMONY IS THAT PRINCIPALLY THIS IS AN ABERRANT SENTENCE. THAT
IS, IT DOES NOT REFLECT THE TREND OF SENTENCE IN THIS
JURISDICTION OR IN THE STATE.
ALL I CAN SAY WITH RESPECT TO THE LIKELIHOOD THERE WAS
A RACIAL FACTOR IN THIS CASE IS THAT IT IS A POSSIBILITY. IT MAY
VERY WELL HAVE BEEN THE FACTOR, BUT THAT’S AS STRONG AN OPINION
AS 1 THINK ONE CAN MAKE ON THE BASIS OF THESE DATA,
MR. BOGER: YOUR HONOR, I MOVE THE ADMISSION OF DB-113
AND 114 INTO EVIDENCE.
THE COURT: MS. WESTMORELAND?
MS. WESTMORELAND: YOUR HONOR, I DONT KNOW THAT I SEE
THE RELEVANCE FOR ALL THE DOCUMENTS INCLUDED IN DB-11é.
THE COURT: WELL, BECAUSE WE'RE GETTING DOWN TO CASES,
I WOULD ASSUME THAT MR. BOGER SUGGESTS THAT WE MIGHT REALLY
START EVALUATING THE FACTS OF THE CASES AS LAWYERS INSTEAD OF AS
1059
BALDUS - DIRECT
STATISTICIANS.
MR. BOGER: I THINK THAT‘S RIGHT, ALTHOUGH PROFESSOR
BALDUS SAYS HE FEELS COMFORTABLE DOING SOME OF THAT AS WELL.
AND I NOTED, WE“RE TRYING TO MOVE QUICKLY. I COULD HAVE TAKEN
PROFESSOR BALDUS THROUGH EACH OF THESE POLICE REPORTS AND EACH
OF THESE SUMMARIES. AND THOUGHT THAT DISCRETION ON THAT POINT
WAS IN ORDER.
THE COURT: I DON‘T, I DON’T KNOW WHAT THEY SAY BECAUSE
I HAVEN’T STUDIED THEM. BU I DON‘T SEE ANY HARM TO THE RECORD
IN ADMITTING THEM. I THINK IT MIGHT FACILITATE AN UNDERSTANDING
OF WHAT MIGHT BE AT WORK, AND UNLESS YOU CAN THINK OF SOME
BETTER REASON. I WILL ADMIT THEM.
MS. WESTMORELAND: NO, YOUR HONOR, IF I COULD JUST ASK
WHICH PARTICULAR CASES THE POLICE OFFICER WAS REFERRING TO. TO
ALL CASES LISTED IN THE POLICE OFFICER’S REPORTS FOR ALL THE
POLICE OFFICER KILLINGS.
THE COURT: THAT WAS MY UNDERSTANDING.
IS THAT CORRECT. MR. BOGER?
BY MR. BOGER:
QA. PROFESSOR BALDUS, WOULD YOU CLARIFY THAT?
A. IM SORRY. I DIDN’T GET COUNSEL‘S QUESTION.
@. ARE THESE POLICE OFFICER REPORTS ON ALL THE INDIVIDUAL CASES
OR WHAT SUB-GROUP ARE THEY?
IN OTHER WORDS, DO WE HAVE IN DB-116 ALL THE POLICE
OFFICER REPORTS OR DO WE HAVE REPORTS SUPPLEMENTED BY YOUR
1060
BALDUS - DIRECT
Is
SUMMARIES? WHAT DOES THAT MATERIAL REPRESENT?
2 THE COURT: ARE THE POLICE OFFICERS REPORTS ON ALL 17
3 |CASES?
4 THE WITNESS: NO. THERE ARE NOT POLICE OFFICER REPORTS
= ON ALL 17 CASES, IS MY RECOLLECTION.
é THERE ARE, AS THE. IS NOTED ON DB-115, YOUR HONOR. ON
y THE FOOTNOTE RELATED TO THAT TABULATION. WHICH INDICATES THE
3 SOURCE OF THE DATA, THE DATA CONCERNING THE FIRST SIX VICTIMS
9 WERE OBTAINED FROM THE ATLANTA POLICE DEPARTMENT AND SUPERIOR
10 COURT RECORDS.
11 THOSE REPORTS WITH RESPECT TO THOSE OFFENDERS ARE
12 INCLUDED THERE, TO THE BEST OF MY RECOLLECTION. THAT IS, IN THE
13 DB-116.
14 THE INFORMATION ON BOBO WAS OBTAINED FROM MR. STROUP,
13 COUNSEL FOR MCCLESKEY, AND FROM THE SUPERIOR COURT RECORDS.
16 AND THE DATA ON DEFENDANT MOHAMED WERE OBTAINED FROM
17 THE PROCEDURAL REFORM STUDY. HIS INFORMATION IS NOT INCLUDED
13 THERE, BECAUSE HE IS NOT A POLICE OFFICER. HE WAS A DEPUTY
w 19 SHERIFF. AND THE PAPERS THAT HAD BEEN SUBMITTED IN THIS CASE
20 RELATED ONLY TO POLICE OFFICERS.
21 IN ADDITION. THERE IS A SUMMARY FROM THE, EITHER THE
22 CHARGING AND SENTENCING OR THE PROCEDURAL REFORM STUDY ON JAMES
23 SMITH. DAVID BURNEY, BURNEY, BERNARD DUPREE AND WARREN MCCLESKEY
24 AS WELL AS MOHAMED.
25 THOSE ARE THE SOURCES OF THAT INFORMATION.
1061
BALDUS - DIRECT
p
MR. BOGER: YOUR HONOR. THE FIRST SIX AS TO WHICH
INFORMATION WAS OBTAINED OR DOCUMENTS THAT WERE ADMITTED IN THE
STATE HABEAS CORPUS PROCEEDING IN THIS CASE. THAT'S WHERE MR.
STROUP GOT THEM. AND THATS WHERE WE HAD THEM. AND THATS WHAT
PROFESSOR BALDUS USED FDR HIS ANALYSIS. WE DIDN‘T HAVE ACCESS
TO THE REST OF THEM. AND HIS ANALYSIS PROCEEDED WITH WHAT HE
HAD.
BY MR. BOGER:
@. DO THESE DOCUMENTS. THOUGH, REFLECT SOMETHING ABOUT EACH OF
Fa
y £
0
0
s
R
TL
de
e
h
y
IN
THE TEN HOMICIDES AND THE 17 DEFENDANTS?
11 A. THEY DO REFLECT THE BARE-BONE FACTS. BUT WHAT I DID WITH
12 RESPECT TO THESE OTHER OFFENDERS WAS REQUESTED MR. STROUF TO GO
13 DOWN AND EXAMINE THE FULTON COUNTY COURT RECORDS ON THESE
14 OFFENDERS THAT WERE NOT EMBRACED WITHIN OUR STUDY TO SEE WHAT HE
15 COULD GLEAN BY WAY OF THOSE RECORDS AS TO WHAT HAD HAPPENED IN
16 | THOSE CASES.
17 SO NOT I PERSONALLY. BUT MR. STROUP CONSULTED THOSE
is RECORDS AND GAVE ME THE INFORMATION THAT HE OBTAINED WITH
gr 19 RESPECT TO THEM. AND IT WAS ON THAT BASIS. YOUR HONOR. THAT 1
20 WAS ABLE TO IDENTIFY WHETHER THESE PEOPLE WERE THE TRIGGER MEN,
21 AND WHAT OTHER CONTEMPORANEOUS OFFENSE WAS INVOLVED IN THE CASE.
22 WE DO NOT HAVE A NICE TYPED COMPLETE SUMMARY OF THESE
23 CASES AS WE DID IN OUR OTHER PARTS OF OUR STUDY, BECAUSE THE
24 COURT DOCKETS DO NOT PROVIDE THAT SORT OF RICH INFORMATION YOU
25% [CAN FIND IN EITHER SUPREME COURT RECORDS OR IN THE RECORDS OF
1062
BALDUS - DIRECT
Fo
y THE PARDONS AND PAROLES BOARD.
2 |@. BUT THE DOCUMENTS IN DB-116 ARE THE DOCUMENTS ON THE BASIS
3 |OF WHICH YOU CONDUCTED THE ANALYSIS ABOUT WHICH YOU’VE
4 |TESTIFIED, IS THAT CORRECT?
S |A. YES. IT IS SUPPLEMENTED AS I INDICATED ALSO WITH THE
6 INFORMATION MR. STROUP PROVIDED FOR ME.
2 lg. Ail. RIGHTY, |
8 MR. BOGER: YOUR HONOR. I RENEW THE ADMISSION OF DB-115
9 |AND -116.
10 THE COURT: I“LL ADMIT THEM.
11 MR. BOGER: THANK YOU, YOUR HONOR.
12 |BY MR. BOGER:
13 |@. PROFESSOR BALDUS, BEFORE WE FINISH UP THERE ARE A FEW STRAY
14 |TABLES WE WANT TO LOOK AT VERY QUICKLY, 1 THINK.
15 IF YOU WOULD TURN TO WHAT’S BEEN MARKED AS DB-~117 FOR
16 IDENTIFICATION, I‘LL ASK YOU TO IDENTIFY THAT DOCUMENT?
17 A. YES. DB-117 IS THE, PRESENTS THE RESULTS OF ANALYSIS THAT
1a 1S REPORTED IN TABLE 45 OF QUR REPORT, AND WHAT IT DOES IS SHOW
wo 19 |THE RACE EFFECTS ESTIMATED AT SUCCESSIVE STAGES OF THE CHARGING
20 AND SENTENCING PROCESS. STATEWIDE.
21 BASICALLY, THIS TABLE IS A MIRROR, STATEWIDE, OF DB-114
22 THAT WE JUST EXAMINED WITH RESPECT TO FULTON COUNTY.
23 IT ALSO EMPLOYED A STEPWISE REGRESSION PROCEDURE AND
24 INCLUDED AS BACKGROUND FACTORS THE NON-RACIAL FACTORS THAT
23 SHOWED US A STATISTICALLY SIGNIFICANT RELATIONSHIP WITH THE
1063
BALDUS - DIRECT
pb
OUTCOME VARIABLE AT EACH SUCCESSIVE STAGE OF THE PROCESS.
2 @. BRIEFLY STATED, PROFESSOR BALDUS, WHAT DOES DB-117 REFLECT?
3 A. IT REFLECTS A, SHOWS A STRONG RACE OF VICTIM EFFECT AT THE
4 POINT WHERE PLEAS ARE ACCEPTED TO VOLUNTARY MANSLAUGHTER IN
: 5: CASES INVOLVING A MURDER INDICTMENT.
il b IT SHOWS MILD RACE OF VICTIM AND RACE OF DEFENDANT
7 EFFECTS AT THE POINT WHERE GUILTY PLEAS ARE ENTERED. WITH NO
8 PENALTY TRIAL HELD.
9 IT SHOWS HERE BY WAY OF RECAPULATION THE RACE EFFECTS
10 OBSERVED AT THE POINT WHERE CASES ARE ADVANCED AFTER CONVICTION
1: FOR MURDER AT A GUILT TRIAL TO A PENALTY TRIAL AND THE EFFECTS
12 AT THE PENALTY TRIAL.
13 THESE, HOWEVER. ARE, WERE INDICATED IN EARLIER TABLES.
14 THE PRINCIPAL NEW INFORMATION THAT IS ADDED AS A RESULT
15 OF THIS TABLE ARE THE CO~EFFICIENTS IN ROW 2 AND CO-EFFICIENTS
16 IN ROW 7, WHICH SHOW A RACE OF VICTIM EFFECT WITH RESPECT TO THE
17 LIKELIHOOD THAT A. THE PROSECUTOR WILL ACCEPT OR OFFER A FLEA
18 BARGAIN.
- 12 @. AND IN WHAT DIRECTION IN EFFECT DO THOSE RACIAL EFFECTS RUN?
20 A. YES. EACH OF THESE DECISION, TO ACCEPT A PLEA BARGAIN OR TO
21 ALLOW A PERSON TO PLEAD OUT TO VOLUNTARY MANSLAUGHER OR AT LEAST
22 TO RECOMMEND THAT. AGREE TO IT, THESE ARE DECISIONS THAT TAKE A
23 PERSON OUT OF THE STREAM OF CASES LEADING TO PENALTY TRIAL AND
24 FOR THAT REASON, IF A, AN OFFENDER WITH A WHITE VICTIM HAS A
23 REDUCED CHANCE OF HAVING HIS CASE DISPOSED OF BY VIRTUE OF
1064
BALDUS - DIRECT
pb
EITHER OF THESE DECISIONS. HIS RISK OF RECEIVING DEATH SENTENCE
WILL CONTINUE IN THE PROCESS.
THE COURT: DO YOU SERIOUSLY MEAN TQ SUGGEST THAT HAD
FRANK SCHLATT BEEN BLACK, THIS CASE MIGHT HAVE PLED OUT TO
VOLUNTARY MANSLAUGHTER. EVERYTHING ELSE BEING EQUAL?
THE WITNESS: NO. YOUR HONOR. NOT THIS CASE, NO. THESE
CASES ARE NOT RESTRICTED TO POLICE OFFICER CASES, YOUR HONOR,
1M SORRY IF I MISLED YOU IN THAT RESPECT.
THE COURT: FRANK SCHLATT IS THE VICTIM. HE WAS WHITE.
pt
O
o
p
o
S
E
N
I DIDNT MEAN TO SUGGEST POLICE OFFICER AS A FACTOR, THESE ARE
poe
ts
P
s
STATEWIDE AS I UNDERSTAND IT.
i: THE WITNESS: YES.
13 THE COURT: YOU ARE FOR EXAMPLE MAKING SOME NOTE OF
14 DECISION MAKING EARLY ON IN THE SYSTEM.
15 THE WITNESS: YES.
16 THE COURT: ONE OF WHICH YOU TESTIFIED AT SOME LENGTH
17 ABOUT IS WHETHER OR NOT TO LET THEM PLEAD TO VOLUNTARY
18 MANSLAUGHTER OR NOT.
w 19 MY QUESTION TO YOU IS, BASED ON YOUR IMPRESSION OF THE
20 WAY THE SYSTEM WORKS IN GEORGIA AND IN FULTON COUNTY, WOULD YOU
21 HAVE ANY SERIOUS OPINION OR WOULD YOU HAVE ANY OPINION THAT IF
22 SCHLATT HAD BEEN BLACK INSTEAD OF WHITE. IN THIS CASE, HE WOULD
23 HAVE, MCCLESKEY WOULD HAVE HAD A SUBSTANTIAL CHANCE OF GETTING A
24 VOLUNTARY MANSLAUGHTER PLEA?
23 THE WITNESS: NO, YOUR HONOR, I DON‘T THINK HE WOULD
1045
BALDUS - DIRECT
1 HAVE ANY CHANCE. IN FACT, OUR NEAR NEIGHBORS ANALYSIS WE DID AT
< THAT VERY POINT IN FULTON COUNTY WE HAD A TABLE HERE ON IT, WE
LOOKED AT LAST WEEK, IT WAS IN DB-111, IN FACT. IT SPOKE
PRECISELY TO THAT QUESTION, YOUR HONOR. IT. IT RANKED ALL THE
PEOPLE IN FULTON COUNTY IN TERMS OF THEIR LIKELIHOOD OF HAVING
3
a
S
X 4 6 THEIR CASE GO OUT IN A VOLUNTARY MANSLAUGHTER PLEA AFTER A
7 MURDER INDICTMENT, AND WE LOOKED AT PEOPLE WHO WERE IN NEAR
8 NEIGHBORS OF WARREN MCCLESKEY AND NOT ONE FELL IN THAT CATEGORY.
? I THINK NOT ONLY IS THAT MY IMPRESSION, THAT’S THE
10 BASIS OF THE IMPRESSION IS ON THE EMPIRICAL EVIDENCE GENERATED
11 IN FULTON COUNTY AND THE. I EXAMINED THE SAME BREAKDOWN OF DATA
12 STATEWIDE. AND IT WAS EXACTLY THE SAME. PEOPLE THAT WERE OF
13 MCCLESKEY’S CATEGORY IN TERMS OF SERIOUSNESS OF THE OFFENSE. DO
14 NOT AS A RULE GO OUT ON VOLUNTARY MANSLAUGHTER PLEAS.
15 THE COURT: I DON’T KNOW WHAT YOU DEFINE AS PLEA
16 BARGAINING, BUT LET US ASSUME THAT A PLEA BARGAIN WOULD BY
17 DEFINITION HAVE TO BE EITHER TO A LESSOR OFFENSE OR A SENTENCE
138 SHORTER THAN LIFE.
w iP AM I FAIR 50 FAR?
20 THE WITNESS: YES, YOUR HONOR. THAT'S WHERE A CASE STARTS
21 AS A MURDER INDICTMENT. YES.
22 THE COURT: WOULD YOU CONTEND IF SCHLATT HAD BEEN
23 BLACK, MCCLESKEY WOULD HAVE HAD ANY SUBSTANTIAL CHANCE TO GET A
24 PLEA FOR LESS THAN LIFE?
25 THE WITNESS: NO. NO. BECAUSE THE, THE BASIS OF THAT
WW
)
0
N
h
AR
W
N
e
s
—-
oO
1064
BALDUS - DIRECT
JUDGMENT. YOUR HONOR, IS THAT EVERY, IN FULTON COUNTY, NOW WE“RE
FOCUSING ON FULTON COUNTY, THAT EVERY TRIGGER MAN IN A CASE
INVOLVING A CONTEMPORANEOUS OFFENSE WAS CONVICTED OF MURDER.
NOW THERE ARE SOME OF THE OFFENDERS IN THE OTHER CASES
THAT WERE NOT TRIGGER MEN WHO GOT VOLUNTARY MANSLAUGHTER,
RECEIVED A VOLUNTARY MANSLAUGHTER CONVICTIONS BY PLEA, AND IM
NOT SURE OF THE DETAILS OF ALL OF THOSE BECAUSE THOSE WERE NOT
OUR PRINCIPAL FOCUS. BUT AMONG THIS POPULATION THAT ARE MOST
LIKE MCCLESKEY. ALL THOSE PEOPLE RESULTED IN A MURDER
CONVICTION.
THE COURT: NOW I“M NOT SURE YOU ANSWERED MY QUESTION.
MURDER CONVICTION BASED ON PLEA OR JURY VERDICT?
THE WITNESS: YES, YOUR HONOR.
THE COURT: IS IT POSSIBLE TO GET LESS THAN LIFE FOR
MURDER IN GEORGIA?
THE WITNESS: NOT TO MY KNOWLEDGE. IT ISN“T.
THE COURT: IS LIFE MANDATORY?
MS. WESTMORELAND: YES, YOUR HONOR.
THE COURT: WHERE WE HAVE GOT A MURDER CONVICTION IN
ANY OF THESE FIGURES THAT IS THE EQUIVALENT OF LIFE?
MR. BOGER: AT LEAST AS A SENTENCE, THATS RIGHT. YOUR
HONOR.
THE COURT: AND THAT CAN'T BE MANIPULATED. CAN IT. SO
AS TO SUSPEND THE LIFE SENTENCE AND IMPOSE A TEN-YEAR SENTENCE
OR ANYTHING LIKE THAT?
1067
BALDUS ~ DIRECT
MS. WESTMORELAND: NOT TO MY KNOWLEDGE, YOUR HONOR.
IT’S MY UNDERSTANDING ITS EITHER A LIFE SENTENCE OR A DEATH
SENTENCE.
THE COURT: SO AS IT RELATES TO THIS CASE, THE EXERCISE
OF THE PROSECUTORIAL DISCRETION AT THOSE LOWER LEVELS IS
DIRECTLY IRRELEVANT ALTHOUGH PERHAPS CIRCUMSTANTIAL EVIDENCE OF
A BIAS IN THE SYSTEM.
THE WITNESS: THAT’S CORRECT, YOUR HONOR.
THE COURT: THAT‘S KIND OF WHERE WE ARE.
THE WITNESS: YES. YOUR HONOR.
THE COURT: SO THE NEXT STEP GETS TO BE REALLY WHETHER
TO ADVANCE TO TRIAL. AND THAT WAS FOR MCCLESKEY THE CRITICAL
STAGE. TO ADVANCE IT TO PENALTY PHASE.
THE WITNESS: MY ANSWERS. YOUR HONOR. RELATE TO FULTON
COUNTY. THAT’S THE SUBJECT OF OUR FOCUS HERE.
THE COURT: TO THE EXTENT YOUR FEELING WOULD BE
DIFFERENT ABOUT THE STATEWIDE, I WOULD BE INTERESTED IN HEARING
YOUR OPINION. BUT WHAT I“M MOVING UP ON IS THIS.
TO THE EXTENT THAT THE EXERCISE OF PROSECUTORIAL
DISCRETION IS MOTIVATED BY RACE, AS IT WOULD AFFECT MCCLESKEY.
THAT DECISION BY A PROSECUTOR BECOMES CRITICAL IN MCCLESKEY’S
CASE ONLY AT THE DECISION TO ADVANCE IT TO PENALTY TRIAL.
THE WITNESS: YES. TO ADVANCE IT TO A GUILT TRIAL. AS
OPPOSED TO HAVING A PLEA FOR MURDER. PERHAPS.
THE COURT: WELL. WHAT’S THE DIFFERENCE?
-
vv
O
N
0
G
e
WN
LE
—
=
QO
Py
nN
1063
BALDUS - DIRECT
THE WITNESS: WELL. IF ONE GOES TO --
THE COURT: IF YOU“RE GOING TO LET SOMEBODY PLEAD TO
MURDER OR BE CONVICTED OF MURDER, AND THEY“RE BOTH GOING TO GET
LIFE, WHAT DIFFERENCE DOES IT MAKE? THAT-S NOT A PLEA BARGAIN.
THE WITNESS: THAT’S THE POINT. YOUR HONOR. I THINK IN
MOST OF THESE CASES IF THE CASE, SOMETIMES THE DEFENDANT IS TOLD
THAT WE WILL WAIVE THE. A DEATH PENALTY, AND EVEN IF YOU’RE
CONVICTED. WERE NOT GOING TO SEEK THE DEATH SENTENCE IN THE
CASE. BUT AT LEAST IF YOU ENTER A PLEA AS PART OF DEAL, ONE
KNOWS THERES NO RISK OF A DEATH SENTENCE. WHEREAS IF ONE GOES
TO TRIAL ON THE QUESTION, THERE'S ALWAYS THE CERTAIN POSSIBILITY
THERE MIGHT BE A PENALTY TRIAL.
BUT I AGREE,» THAT COMPARABLY. IN TERMS OF THE RESULT,
IF YOU TAKE THE RISK OF A DEATH SENTENCE OUT OF THE PICTURE,
THEY ARE COMPARABLE. THATS RIGHT.
THE COURT: ALL RIGHT.
MR. BOGER: YOUR HONOR, AT THIS TIME I WOULD MOVE THE
ADMISSION OF DB-117 INTO EVIDENCE.
MS. WESTMORELAND: YOUR HONOR, I THINK THE CONTINUING
OBJECTION THAT 1 NOTED THROUGHOUT THE HEARING WOULD APPLY TO
|THIS EXHIBIT AS WELL.
THE COURT: ALL RIGHT. TI“LL ALLOW IT.
MR. BOGER: THANK YOU. YOUR HONOR.
BY MR. BOGER:
A. PROFESSOR BALDUS, LET ME ASK YOU TO TURN YOUR ATTENTION TO
10469
BALDUS - DIRECT
} WHAT“S BEEN MARKED AS DB-118 FOR IDENTIFICATION.
2 CAN YOU IDENTIFY THAT DOCUMENT?
A. YES. DB-118 IS TABLE 4% FROM OUR REPORT. AND WHAT IT DOES
IS EXPAND THE ANALYSIS THAT IS PRESENTED IN DB-117.
SPECIFICALLY, IT LOOKS AT SUB-GROUPS OF CASES THAT ARE
3
4
he |
RB & DEATH ELIGIBLE UNDER SPECIFIC STATUTORY AGGRAVATING FACTORS.
7 @. IS THIS PART OF THE SERIES OF TABLES WHICH WE DISCUSSED AT
8 SOME LENGTH EARLIER, LOOKING AT THESE STATUTORY AGGRAVATING
2? CIRCUMSTANCE CASES BROKEN DOWN IN THAT FASHION?
10 A. YES. AND THOSE EARLIER ANALYSES FOCUSED ON THE COMBINED
11 EFFECTS OF ALL THE DECISIONS IN THE PROCESS WHEREAS THIS ZEROES
12 IN ON THE SUCCESSIVE STAGES IN THE DISPOSITION OF THE CASES THAT
13 ARE DEATH ELIGIBLE UNDER THESE DIFFERENT STATUTORY AGGRAVATING
14 |FACTORS.
15 . THE COURT: WHAT IS A B4 CASE?
16 THE WITNESS: B4 IS MONETARY MOTIVE. YOUR HONOR.
37 THE COURT: AND B77?
18 THE WITNESS: B7 IS WANTONLY VILE.
19 THE COURT: B10 IS POLICE?
20 THE WITNESS: NO. B10, YOUR HONOR, IS MOTIVE TQ AVOID
21 APPREHENSION, OR ARREST. THAT SORT OF THING.
A BY MR. BOGER:
23 R. WHAT DOES THIS TABLE REFLECT, PROFESSOR BALDUS?
24 A. WELL. THIS TABLE REFLECTS WITHIN THESE SUB-CATEGORIES OF
23 CASES, RACE EFFECTS IN TERMS OF THE DECISION TO ACCEPT A
1070
BALDUS - DIRECT
Pa
r VOLUNTARY MANSLAUGHTER PLEA AFTER MURDER INDICTMENT. THESE ARE
ALL DEATH ELIGIBLE CASES, BUT EVEN AMONG THEM THERE ARE A NUMBER
OF CASES THAT DO RESULT IN VOLUNTARY MANSLAUGHTER CONVICTIONS
FOLLOWING A MURDER INDICTMENT.
AND ROW 1 SHOWS THOSE CO-EFFICIENTS WHICH ARE LISTED
UNDER THE RACE OF THE VICTIM AND YOU CAN SEE THERE THE CHANCES
OF RECEIVING A VOLUNTARY MANSLAUGHTER PLEA AMONG THESE
CATEGORIES OF CASES IS SMARTLY REDUCED IF ONE HAS A WHITE
B
e
SE
ER
SU
E
B
e
Ba
k
B
R
a
VICTIM.
TN
o THE, YOU DO NOT SEE ANY STATISTICALLY SIGNIFICANT
a
| a
md
EFFECTS WITH RESPECT TO THE PLEA TO MURDER.
fy
hb
AND, NOR ARE THERE ANY EFFECTS AT THE GUILT TRIAL.
PY
Ww
UNDER 3A. WITH RESPECT TO B2 AND B4 AND B7. YOU SEE
14 |STRONG RACE OF VICTIM EFFECTS. THAT WHITE VICTIM CASES ARE MUCH
13 LESS LIKELY TO RECEIVE AN OFFER OR AGREEMENT TO A PLEA BARGAIN.
16 AND AT THE NEXT STAGE AFTER A MURDER CONVICTION, THE
17 RACE OF VICTIM EFFECT IS PLAIN IN B2, B4, B7 AND B10 CASES.
13 ALSO. YOU SEE IN THESE CASES STATISTICALLY SIGNIFICANT
es 19 RACE OF DEFENDANT EFFECTS AT THIS STAGE IN THE PROCESS UNDER THE
20 B2, THE B4, AND THE B7 CASES.
21 THOSE ARE THE PRINCIPAL BITS OF INFORMATION THAT ARE
22 RELEVANT, I BELIEVE, TO THIS PROCEEDING. THAT COME FROM ——
23 ®@. OVERALL, PROFESSOR BALDUS, HOW DOES THIS TABLE AND THE
24 PICTURE IT PORTRAYS THROUGH THESE NUMBERS FIT INTO YOUR LARGER
23 TESTIMONY?
O
0
0
N
B
a
e
Pa
y
FN
Py
12
13
1071
BALDUS - DIRECT
A. YES. THEY CONFIRM THE IMPRESSION OF THE OTHER TABLES HAVE
LEFT US. THAT THE RACE EFFECTS ARE STRONG, ESPECIALLY WITH
RESPECT TO PROSECUTORIAL DECISION MAKING, AND THEY ARE
PARTICULARLY IMPORTANT IN THE CATEGORY OF CASES WHERE YOU SEE
MOST OF THE DEATH SENTENCES COMING FROM. THAT IS, THE B7 AND B2
CASES.
@. AND WHAT IF ANY BACKGROUND FACTORS ARE CONTROLLED FOR?
A. THESE ANALYSES ARE CONTROLLED THROUGH STEPWISE REGRESSION
PROCEDURE ALL OF THE VARIABLES THAT ENTERED THESE ANALYSES THAT
THE. AT THE .10 LEVEL OF STATISTICAL SIGNIFICANCE.
MR. BOGER: YOUR HONOR. AT THIS TIME I MOVE THE
ADMISSION OF DB-118 INTO EVIDENCE.
MS. WESTMORELAND: SAME OBJECTION, YOUR HONOR.
THE COURT: 1I‘LL ALLOW THEM.
MR. BOGER: THANK YOU. YOUR HONOR.
THE COURT: I“VE GONE AHEAD TO 119.
BY MR. BOGER:
@. PROFESSOR BALDUS, LAST FRIDAY WE WERE TALKING A LITTLE BIT
ABOUT GEOGRAPHICAL CATEGORIZATIONS, LIKE URBAN AND RURAL
RELATIONSHIPS, AND AT THAT TIME THE JUDGE ASKED YOU BY WHAT
MEANS DID YOU SEPARATE URBAN FROM RURAL JURISDICTIONS WHEN YOU
MADE THE URBAN-RURAL BREAKDOWN.
LET ME ASK YOU TO LOOK AT DB-119 MARKED FOR
IDENTIFICATION. CAN YOU IDENTIFY THAT DOCUMENT?
A. YES. DB-119 IS A LIST OF THE CASES THAT ARE URBAN ACCORDING
1072
BALDUS - DIRECT
re
, TO THE METHOD OF CLASSIFICATION THAT I DESCRIBED TO THE COURT
2 LAST WEEK.
3 THE COURT: BY THE BUREAU OF THE CENSUS?
4 THE WITNESS: WELL, NO, YOUR HONOR. WHAT I DID I
3 TALKED TO SEVERAL PEOPLE HERE IN GEORGIA AND I SAID TO THEM, :
* é WHAT DO YOU CONSIDER TQ BE THE URBAN AREAS OF THIS STATE. AND
7 THEY TICKED OFF THE NAMES OF THE PLACES THEY CONSIDERED
8 TO BE URBAN.
9 THE COURT: THE PEOPLE YOU TALKED TO MUST HAVE MOVED TO
10 TOWN LAST WEEK.
11 : FOR EXAMPLE. ECHOLS COUNTY IS THE LEAST POPULATED
12 COUNTY IN THE ENTIRE STATE OF GEORGIA.
13 THE WITNESS: WHICH COUNTY IS THAT. YOUR HONOR?
14 THE COURT: ECHOLS. WHERE YOUVE GOT IT IN THE SCUTHERN
15 |CIRCUIT.
16 THE WITNESS: WHAT I WAS FOCUSING ON WAS THE ENTIRE
17 |CIRCUIT FOR PURPOSES OF THE ANALYSIS. THERE MIGHT BE A LITTLE
18 |SLIPPAGE HERE. THE NOTION WAS. YOUR HONOR, I GOT A SENSE FROM
w 19 |THE PEOPLE WHAT WERE THE CIRCUITS THAT WERE CONSIDERED URBAN, I
20 |LOOKED AT THOSE CIRCUITS AND FOUND WHAT THE PERCENTAGE OF URBAN
21 |POPULATION THERE WAS IN THEM AND THE CUT POINT SEEMED TO BE MORE
22 |THAN FIFTY PERCENT URBAN. |
23 THE COURT: WELL. I DON’T KNOW THAT IT MAKES ANY
24 |DIFFERENCE TO THE OUTCOME OF THE CASE, BUT IN MY HUMBLE OPINION
23 THIS AIN‘T EVEN CLOSE TO DESCRIBING URBAN.
1073
BALDUS - DIRECT
pt
MR. BOGER: WELL, AS REFLECTING PROFESSOR BALDUS”
THOUGHT PROCESSES ON WHAT WHAT WERE URBAN AND NOT, I MOVE THE
ADMISSION OF -119.
THE COURT: FOR ONCE HE DEPARTED FROM RELIABLE
STATISTICAL SAMPLING TECHNIQUES AND DID A MAN-ON-THE-STREET
POLL. AND HE WAS LED ASTRAY.
MS. WESTMORELAND: FOR THAT PURPOSE I HAVE NO
OBJECTION, YOUR. HONOR.
a
.
0
N
N
d
N
THE COURT: OKAY.
10 BY MR. BOGER:
33 @. NOW. PROFESSOR BALDUS, I“VE GOT A NUMBER OF THINGS I WANT TO
J v RUN BACK THROUGH AND PICK UP BEFORE WE FINISH YOUR TESTIMONY.
13 1D LIKE FIRST TO ASK YOU TO LOOK AT DB-180 MARKED FOR
14 IDENTIFICATION.
13 THE COURT: -1807?
16 MR. BOGER: EXCUSE ME. YOUR HONOR. -80.
17 : THE COURT: WHERE IS THAT?
18 MR. BOGER: IT“S BACK -— IN OTHER WORDS, I HAD A FEW
w 19 THINGS.BASED ON YOUR HONOR‘S RULING ABOUT DB-113, THE REPORT.
20 THAT ID LIKE TO JUST RUN THROUGH. IT“LL TAKE MAYBE TWO OR THREE
21 MINUTES. AND ASK PROFESSOR BALDUS SOME QUESTIONS ABOUT IT.
22 THE COURT: LET ME FIND IT FIRST, IF I MAY. ALL RIGHT.
23 BY MR. BOGER:
24 @. PROFESSOR BALDUS, LOOKING AT DB-80 FOR IDENTIFICATION. ID
235 LIKE TO ASK YOU, LOOKING AT FOOTNOTES 3 AND 4, INDICATE IF YOU
1074
BALDUS - DIRECT
Pa
y WOULD WHAT BACKGROUND VARIABLES YOU USED IN THOSE TWO ANALYSES.
CONTROLLED FOR NINE BACKGROUND VARIABLES. AND CONTROLLED FOR
230-PLUS NON-RACIAL FACTORS.
IF YOU CAN EXPLAIN THAT FOR US.
A. YES. THE FOOTNOTE 3 RELATES TO THE VARIABLES THAT ARE IN
22A, WHICH WAS ADMITTED AS AN ATTACHMENT. I BELIEVE, TO DB-80 IN
THE PROCEEDING LAST WEEK, IF I“M NOT MISTAKEN, COUNSEL.
@. ALL RIGHT, THEN. WHAT ABOUT THE 230-PLUS NON-RACIAL
9
©
MN
Oo
AA
bb
W
N
VARIABLES?
Pt
e oO
A. RIGHT, THOSE ARE THE VARIABLES LISTED AT SCHEDULE 3 OF THE
p
t
p
t
TECHNICAL APPENDIX.
12 @. ALL RIGHT. THE TECHNICAL APPENDIX, I BELIEVE, IS IN
13 EVIDENCE AT THIS POINT AS DB%é4A.
14 NOW. WITH RESPECT TO DB-83., LOOKING AT ROW 1B, EXCUSE
13 ME. |
16 A. YES.
17 a. AT 1K.» STATUTORY AGGRAVATING CIRCUMSTANCES AND 126 FACTORS
18 DERIVED FROM THE ENTIRE FILE BY A FACTOR ANALYSIS.
w 19 WHAT IN FACT ARE THE FACTORS WE’RE TALKING ABOUT HERE?
20 A. YES. THIS ANALYSIS INCLUDED THE TEN STATUTORY AGGRAVATING
23 FACTORS. PLUS AN ADDITIONAL VARIABLE FOR PRIOR FELONY RECORD.
22 AND FOUR SUSPECT FACTORS THAT DETERMINE WHETHER OR NOT AN
23 OFFENDER WAS FROM OUT OF STATE OR A RESIDENT, WHETHER THE —— THE
24 SEX OF THE DEFENDANT. AND WHETHER THE DEFENDANT WAS HIGH OR LOW
23 SOCIO-ECONOMIC STATUS.
pt
N
A
W
T
N
10735
BALDUS - DIRECT
AND THE REMAINING 126 FACTORS THERE WERE, RELATE TO
EVIDENTIARY MATTERS AND NON-STATUTORY AGGRAVATING FACTORS AND
MITIGATING FACTORS THAT WERE DERIVED IN A FACTOR ANALYSIS.
@. ALL RIGHT. LOOKING NOW AT DB-89 MARKED FOR IDENTIFICATION
AND IN EVIDENCE.
WHAT ARE THE FACTORS THAT YOU USED TO CONSTRUCT THE
SCALE FROM ONE TO EIGHT? WHAT VARIABLES.
A. THESE, THIS SCALE WAS BASED ON AN INDEX WHICH WAS
CONSTRUCTED WITH THE VARIABLES AT SCHEDULE 5 OF THE TECHNICAL
APPENDIX.
R. AND THAT'S THE TECHNICAL APPENDIX NOW MARKED AS DB-96A7?
A. YES.
R. ALL RIGHT. LET’S LOOK OVER AT DB-112. NOW YOU“VE NOT
IDENTIFIED THAT DOCUMENT TO DATE.
IF YOU COULD TELL ME WHAT THAT IS. PROFESSOR BALDUS?
A. YES. DB-112 WAS THE FINAL. IT REFLECTS THE FINAL STEP IN
OUR ANALYSIS OF THE NEAR NEIGHBORS IN FULTON COUNTY. IT GIVES A
THUMB NAIL SKETCH OF FULTON COUNTY. BLACK VICTIM CASES INVOLVING
AN EQUAL OR GREATER LEVEL OF CULPABILITY OF WARREN MCCLESKEY
WITH A LIFE SENTENCE IMPOSED.
@. AND DOES IT THEREFORE FIT INTO THE OVERALL ANALYSIS YOU
PREVIOUSLY TESTIFIED TO ABOUT THE SIGNIFICANCE OF THE NEAR
NEIGHBORS?
A. YES. IN THE CONDUCT OF A NEAR NEIGHBORS ANALYSIS OF THE
TYPE WE DID, ONE CAN MAKE COMPARISONS BETWEEN COMPARABLE CASES
BALDUS - DIRECT
i |AND ESTIMATE RATES AT WHICH THEY ARE DISPOSED OF IN A DIFFERENT
2 FASHION.
3 ANOTHER APPROACH IS TO TAKE AN INDIVIDUAL OFFENDER. AND
COMPARE HIM WITH OTHER CASES THAT YOU THINK ARE OF EQUAL OR
GREATER CULPABILITY. IN WHICH A LESSOR SENTENCE WAS IMPOSED.
AND THAT’S WHAT THE PURPOSE OF THIS ANALYSIS IS, WAS,
4
=
&
7 WAS TO IDENTIFY THOSE CASES THAT, IN WHICH ONE PERCEIVES AN
8 OVERALL LEVEL OF CULPABILITY THAT IN MY JUDGMENT WAS COMPARABLE
‘4 TO MCCLESKEY’S, AND SORT THOSE CASES INTO THE CATEGORY OF BLACK
0 VICTIM. THESE ARE THE —
11 @. DOES THIS TABLE REFLECT IN EFFECT THE PRODUCT OF YOUR
12 ANALYSIS, OR ONE OF THE BASES ON WHICH YOU RELIED IN COMPLETING
13 THIS NEAR NEIGHBORS ANALYSIS?
14 A. YES. THIS IS ONE OF THE REASONS THAT I HAVE THE OPINION
15 THAT WITH RESPECT TO THE SERIOUS CASES THAT ARE POTENTIAL
16 CANDIDATES FOR CAPITAL PUNISHMENT IN FULTON COUNTY, THAT YOU SEE
17 THAT THERE ARE IN FACT RACE OF VICTIM EFFECTS BECAUSE IT’S HARD
18 IN MY OPINION TO DISTINGUISH THESE CASES IN TERMS OF THEIR DEATH
Ww 12 WORTHINESS, IF YOU WILL. FROM WARREN MCCLESKEY‘S CASE AND YET
20 THESE ARE ALL BLACK VICTIM CASES THAT RECEIVED A LIFE SENTENCE.
21 MR. BOGER: YOUR HONOR, AT THIS TIME I MOVE THE
22 ADMISSION OF DB-112 INTO EVIDENCE.
23 THE COURT: IT WILL BE ADMITTED.
24 1 BELIEVE, JUST ON AN UNSCIENTIFIC SAMPLING OF EXPOSURE
25 TO THESE KINDS OF CASES. YOU WOULD ALSO FIND AN AWFUL LOT OF
re
y
o
P
RN
E
B
E
R
)
TI
T
f
BU
R
R
e
al
lt
oe ~-t
1077
BALDUS - DIRECT
WHITE VICTIM CASES THAT WERE MORE EXACERBATED OR AGGRAVATED
THAN WARREN MCCLESKEY’S WHERE THE DEATH PENALTY WAS NOT IMPOSED.
WOULDN‘T YOU?
THE WITNESS: I“D HAVE TO LOOK BACK AT THOSE NUMBERS.
YOUR HONOR. BUT I DON’T THINK SO, THAT IN FULTON COUNTY THAT
WAS THE CASE. AMONG THE FULTON COUNTY CASES. MY RECOLLECTION
WAS THAT AMONG THE HIGHLY AGGRAVATED. CERTAINLY AMONG THE MORE
AGGRAVATED CASES THAT WERE WHITE VICTIM CASES DEATH SENTENCES
WERE REGULARLY IMPOSED, YOUR HONOR.
THE COURT: I GOT THIS IMPRESSION FROM, I GUESS FROM
DOCTOR BOWERS WHO WAS BEGINNING TO MENTION SOME CASES.
PROPORTIONALITY REVIEW. WHERE THE DEATH PENALTY WAS NOT IMPOSED
THAT SEEMED MORE EXTREME THAN MCCORQUODALE AND HOUSE. I DON‘T
KNOW WHETHER THEY ARE FULTON COUNTY CASES OR NOT, BUT DURING THE
COURSE OF THAT HEARING I FORMED THE IMPRESSION THAT THERE WERE A
LOT OF FOLKS THAT DIDN‘T GET THE DEATH PENALTY THAT DID BAD
THINGS. VERY BAD THINGS.
MR. BOGER: WELL STIPULATE TO THAT.
BY MR. BOGER:
?. PROFESSOR BALDUS, JUST AGAIN TO CLARIFY THE RECORD, IM
GOING TO SHOW YOU A DOCUMENT THATS PREVIQUSLY BEEN ADMITTED AS
DB-110, AT LEAST A PORTION OF IT’S BEEN ADMITTED.
ILL ASK YOU TO TURN TO THE SECOND PORTION OF THAT
DOCUMENT INVOLVING, ALTHOUGH BEHIND A LABEL THAT HAS TO DO WITH
POLICE CASES, CAN YOU IDENTIFY THOSE DOCUMENTS?
1078
BALDUS - DIRECT
1 A. YES.
2 @. ALL RIGHT. ARE THEY IDENTICAL TO WHAT WE HAVE MARKED IN A
3 LESS CLEAR COPY AS DB-1147
4 A. THEY ARE IDENTICAL. EXCEPT WITH RESPECT TO THE FIRST WORK
3 PAPER THAT IS ON DB-116.
6 , THE LATTER HALF OF DB-110 THAT IS UNDER TAB, FULTON
7 COUNTY POLICE OFFICERS. HAS A WORK PAPER THAT WAS AN INTERIM
3 PAPER AND HAS SINCE BEEN SLIGHTLY MODIFIED.
9 @. SO THE MORE CORRECT COPY IS WHAT WE-VE MARKED AS DB-1167
10 A. YES.
11 Q. BUT OTHERWISE THE POLICE REPORTS AND ITEMS THAT ARE IN
12 THERE, ARE IDENTICAL TO WHAT WE HAVE IN EVIDENCE AS DB-1167
13 A. YES.
14 MR. BOGER: YOUR HONOR. I DON‘T THINK THERE’S ANY
135 PARTICULAR NEED TO MOVE SECOND PART INTO EVIDENCE, I“LL BE GLAD
16 TO DO IT, IT’S SIMPLY. ALTHOUGH I, I GUESS IF THE COURTS GOING
17 TO LOOK AT IT, IT’S A CLEARER COPY THAN WHAT WE“VE GOT IN
18 EVIDENCE AS DB-114. I THINK I REMEMBER MENTIONING TO THE COURT
w 19 AT THE TIME THAT PROFESSOR BALDUS” ORIGINAL DOCUMENT INCLUDED
20 BOTH THE NEAR NEIGHBORS ANALYSIS AND POLICE ANALYSIS IN ONE
21 NOTEBOOK. AND WE’VE GIVEN COPIES OF THAT TO THE STATE OVER THE
22 WEEKEND AS DB-11& AND NOW SIMPLY WANT THE RECORD TO BE CLEAR
23 WE’RE NOT LOOKING AT TWO DIFFERENT THINGS HERE. IF THE COURT
24 WANTS THEM IN EVIDENCE --
23 THE COURT: I DON'T REALLY UNDERSTAND WHAT YOU'RE
ok
0
N
O
R
M
B
N
10
11
107%
BALDUS - DIRECT
SAYING, MR. BOGER. BUT ARE YOU TRYING TO SUBSTITUTE A BETTER
COPY THAN YOU HAVE IN THE RECORD?
MR. BOGER: THAT”S CORRECT. I“VE GOT DB-110 IN THE
RECORD, OR IN EVIDENCE, EXCEPT FOR THE SECOND PORTION OF IT. THE
SECOND PORTION OF THAT ORIGINAL IS WHAT WE-VE PUT INTO EVIDENCE
AS DB-114, WITH THE CORRECTION TO THE INITIAL PAGE OF 116
WHICH IS THE ONE THAT LAID OUT THE HANDWRITTEN FIGURES THAT
PROFESSOR BALDUS HAD DONE, IS MORE UP TO DATE THERE. BUT THE
POLICE REPORTS WHICH ARE RATHER UNCLEAR IN -116 ARE CLEARER IN
-110.
I MOVE ITS ADMISSION FOR THAT PURPOSE. OR I WOULD BE
GLAD TO SUBSTITUTE IT, IF THE COURT WOULD PREFER, WHATEVER -—-
THE COURT: I DON’T THINK THERES AN OBJECTION. BUT I
CAN‘T QUITE --
MS. WESTMORELAND: IM NOT SURE I’M STILL CLEAR WHAT
MR. BOGER IS TRYING TO DO IS THE REASON I“M HAVING A HARD TIME
DECIDING. YOUR HONOR.
I UNDERSTOOD, AND MAYBE I WAS MISTAKEN. DB-110 WAS
ADMITTED. PERHAPS WE ONLY ADMITTED PART OF IT AT THE TIME LAST
WEEK. I DIDN‘T RECALL SPECIFICALLY OVER THAT POINT.
MR. BOGER: THAT’S RIGHT. WHAT WE HAD. YOUR HONOR, WAS
A NOTEBOOK THAT PROFESSOR BALDUS HAD BEEN WORKING WITH. IT WAS
AN ORIGINAL DOCUMENT. I DIDN“T WANT TO BREAK IT UP. BUT IT HAS
BOTH THE NEAR NEIGHBORS ANALYSIS AND THE FULTON COUNTY POLICE
OFFICERS AS VICTIMS ANALYSIS. WHEN WE GOT TO THE NEAR NEIGHBORS
1080
BALDUS — DIRECT
LAST WEEK. I OFFERED THAT PORTION OF IT AS -110.
TODAY, SINCE 1 WANTED FIVE COPIES FOR EVERYBODY, AND
THIS WAS WITH THE COURT OVER THE WEEKEND, I MADE FIVE MORE
COPIES WHICH WERE OFFERED AS -114. AND AS PROFESSOR BALDUS HAS
SAID, THE FIRST PAGE OF —-116 IS THE MORE ACCURATE SUMMARY OF
THAT WORK PAPER. THIS SUMMARY IN -110 IS NOT ACCURATE. IT’S
NOT AS UP TO DATE. BUT THE POLICE REPORTS WHICH FOLLOW IT ARE
THE CLEANEST COPIES AVAILABLE, WHEREAS -116 ARE A LITTLE BIT
LESS CLEAR.
THE COURT: ALL RIGHT. WITH THAT EXPLANATION, I-“LL
ADMIT IT.
MR. BOGER: ALL RIGHT, YOUR HONOR.
AT THIS TIME, I“LL DO TWO OTHER THINGS. OR THREE OTHER
THINGS. VERY QUICKLY.
I WOULD LIKE TO MOVE TO WITHDRAW DB-44 WHICH WAS THE
BOX OF QUESTIONNAIRES THAT WE HAD MARKED. THERE‘S NO REAL NEED
IN OUR JUDGMENT FOR THOSE TO COME INTO EVIDENCE.
ID ALSO LIKE TO MOVE TO WITHDRAW DB-49, WHICH IS THE
COMPUTER TAPE THAT WE SHOWED TO THE COURT. THE STATE HAS COPY
OF THE TAPE LIKE THIS. AND IT IS MY JUDGMENT THAT NOBODY IS
GOING TO SPEND TIME IN CHAMBERS LOOKING OVER THE TAPE ITSELF.
THEN WE HAD MARKED FOR IDENTIFICATION A NOTEBOOK WHICH
INCLUDED SUMMARIES OF THE DB- —— OF THE B2 AND BS FACTORS, AND
IN MY JUDGMENT THERE‘S NO NEED TO BURDEN THE RECORD WITH THIS
DOCUMENT. PROFESSOR BALDUS HAS TESTIFIED WITH RESPECT TO ITS
1081
BALDUS — DIRECT
CONTENTS AND ITS AVAILABLE TO THE STATE IF THEY WANT TO USE IT.
OF COURSE, THOSE MOTIONS ARE MADE SUBJECT TO THE
STATES USE OF THEM FOR ANY PURPOSE THEY WANT IF THEY WOULD LIKE
TO ADMIT THEM.
MS. WESTMORELAND: I HAVE NO OBJECTION, YOUR HONOR. I
DON’T BELIEVE THESE ITEMS WERE ACTUALLY SUBMITTED AS EXHIBITS
LAST WEEK, AT ANY RATE. SO I HAVE NO OBJECTION TO THIS.
THE COURT: ALL RIGHT, IT WILL BE ALLOWED.
MR. BOGER: THANK YOU. YOUR HONOR.
BY MR. BOGER:
@. FINALLY. PROFESSOR BALDUS, I WILL ASK YOU ONE CLOSING
QUESTION.
ARE YOU FAMILIAR WITH A TERM USED IN SOCIAL SCIENCES
CALLED TRIANGULATION?
A. YES.
@. WHAT DOES IT MEAN?
A. THE CONCEPT OF TRIANGULATION RELATES TO THE APPROACH THAT
ONE TAKES OF EMPLOYING A VARIETY OF DIFFERENT METHODS TO ADDRESS
THE SAME QUESTION.
THE CONCEPT HAS THE SAME MEANING THAT IT DOES IN
NAVIGATION, WHERE ONE, DEAD RECKONING IN SAILING FOR EXAMPLE,
ONE TAKES SIGHTS AT DIFFERENT POINTS AND GETS BEARINGS AND LAYS
THE BEARINGS OVER TOP OF ONE ANOTHER AND IF THE BEARINGS ALL
FOCUS ON THE 'SAME POINT, THEY GIVE ONE CONFIDENCE THAT ONE HAS
{PROPERLY IDENTIFIED HIMSELF ON THE MAP.
10382
BALDUS - DIRECT
po
t WE RECOGNIZE IN THAT PROCESS THAT EVERY ONE OF THOSE.
2 EACH OF THOSE BEARINGS HAS SOME ERROR IN THEM. NONE OF THEM IS
3 PERFECT. BUT THAT THE COMBINED MESSAGE OF IT TELLS YOU,
4 ESPECIALLY WHEN THEY PRODUCE THE SAME RESULTS. THAT YOU ARE
3 GETTING A CORRECT PICTURE.
4 -) AND SO IT IS IN STATISTICAL ANALYSIS, OR QUANTITATIVE
0.7 os ANALYSIS OF ALL TYPES. THAT WE TAKE DIFFERENT PROCEDURES AND
3 ADDRESS THE SAME QUESTION. WE KNOW THAT EACH PROCEDURE GIVES US
9 MERELY AN ESTIMATE.
10 BUT WE EXAMINE THE SAME DATA WITH DIFFERENT PROCEDURES,
4 AND SEE THE EXTENT TO WHICH WE GET THE SAME ANSWER, USING
12 ALTERNATIVE PROCEDURES. AND IN THIS CASE. WE ALSO USED
13 QUALITATIVE METHODS. ACTUALLY MAKING CASE COMPARISONS, AND WE
14 FOUND WE GOT THE SAME SORT OF ANSWER.
15 IN ADDITION, THE NOTION OF TRIANGULATION EXTENDS TO THE
16 USE OF DIFFERENT DATA SETS WHICH PROVIDE. WHICH RAISE THE SAME
17 RUESTIONS AND POTENTIALLY PROVIDE THE SAME ANSWERS.
18 IN THIS CASE. WE HAVE TWO COMPLETELY DIFFERENT DATA
19 SETS THAT ADDRESS AT LEAST TWO OF THE CRITICAL POINTS IN
20 GEORGIA“S CHARGING AND SENTENCING PROCESS. THERE, AGAIN. WE
21 USED ALTERNATIVE METHODS TO ADDRESS THE QUESTIONS POSED WITH
22 |RESPECT TO THE RACIAL EFFECTS AT EACH ONE OF THOSE STAGES AND WE
23 FOUND COMPARABLE RESULTS.
24 IT’S THIS WIDESPREAD CONSISTENCY THAT WE SEE IN THE
23 RESULTS OF MANY ANALYSES USING TWO DISTINCT DATA SETS, LOOKING
pb
wv
00
M
N
O
A
»
U
N
I
o
C
PY
nN
13
14
1083
BALDUS - DIRECT
AT DIFFERENT SUB-SETS OF THE DATA SETS. DIFFERENT STAGES IN THE
PROCESS, IT’S THIS TRIANGULATION APPROACH. IF YOU WILL. THAT
PROVIDES THE PRINCIPAL BASIS FOR OUR OPINIONS THAT I HAVE STATED
EARLIER IN THIS PROCEEDING. ABOUT MY BELIEF THAT THERE ARE REAL
RACE EFFECTS OPERATING IN THE CHARGING AND SENTENCING SYSTEM IN
THIS STATE.
MR. BOGER: NO FURTHER. QUESTIONS, YOUR HONOR.
THE COURT: LET ME -— DON’T GO AWAY MR. BOGER. BECAUSE
I HAVE ONE QUESTION.
MR. BOGER: FINE.
THE COURT: TO CLARIFY MY THINKING AT THIS POINT.
THINKING ABOUT DISPARATE TREATMENT CASES, AND THE
DIFFICULTY OF PROVING DISPARATE TREATMENT WITH STATISTICS, ARE
YOU ABLE TO QUANTIFY IN MCCLESKEY”S CASE THE PART THAT RACE OF
THE VICTIM PLAYED IN HIS GETTING THE DEATH PENALTY, UNDER YOUR
VIEW?
THE WITNESS: NO. I CAN GIVE AN OPINION BASED UPON AN
ANALYSIS OF THE DATA, BUT I CAN‘T SAY A PARTICULAR FACTOR
CONSTITUTED FIFTEEN OR TWENTY PERCENT OF THE FORCE.
WE, THE. WE CAN SAY. YOUR HONOR, WHEN WE LOOK AT A, AN
AGGREGATE GROUP OF DECISIONS, WE CAN SAY A PARTICULAR FACTOR
WILL EXPLAIN A CERTAIN AMOUNT OF THE VARIATION THAT WE SEE IN
THE RESULTS OF THE AGGREGATE.
WE CAN ALSO SAY AS WE‘VE TRIED TO DO HERE, THAT WHEN WE
EXAMINE THE IMPACT OF A PARTICULAR SET OF VARIABLES IN A SYSTEM.
ph
M
O
N
A
N
10
1}
1084
BALDUS - DIRECT
THAT YOU CAN ROUGHLY ESTIMATE THE RELATIVE IMPORTANCE OF THEM.
BUT IT’S ONLY AN ESTIMATE.
THAT, I THINK, IS AS FAR AS YOU COULD GO WITH RESPECT
TO QUANTITATIVE METHODS IN TRYING TO IDENTIFY THE IMPACTS OF
FACTORS IN A SYSTEM.
I THINK DUR ESTIMATE. ANOTHER MEASURE WE USED IS THE
IMPACT, PRACTICAL IMPACT THAT THESE EFFECTS HAVE IN THE SYSTEM
WAS THE COMPARISON WE DID BETWEEN THE POPULATION OF DEATH ROW AS
IT IS, AS TO WHAT WE MIGHT EXPECT IN A VERY HYPOTHETICAL
CIRCUMSTANCE. THAT’S A MEASURE OF THE PRACTICAL IMPACT THAT
THESE FACTORS ARE HAVING IN THE SYSTEM.
THE COURT: WELL, THE TROUBLE WITH THAT TABLE, I
THOUGHT AN AWFUL LOT ABOUT IT SINCE I EXCLUDED IT FROM EVIDENCE,
THERE IS NO LOGICAL FOUNDATION FOR THE THESIS THAT IF YOU GOT
PROSECUTORS AND JURIES TO CLEAN UP THEIR ACT. UNDER YOUR VIEW OF
THEIR WRONGDOING. THAT THE DEATH PENALTY RATE WOULD REMAIN AT
ITS PRESENT RATE.
ESSENTIALLY WHAT YOU’RE SAYING, IN THESE FIGURES TO ME,
IS THAT WHAT YOU TESTIFIED TO WITH THE FIRST OF THOSE TWO
TABLES. AND THAT IS, THE SYSTEM IS REACTING LESS PARTIALLY
AGAINST THE HOMICIDE COMMITTED AGAINST A BLACK PERSON THAN IT IS
AGAINST A WHITE PERSON, AND IF THEY APPLIED STANDARD TREATMENT.
IT WOULD BE THE HIGHER NUMBER.
THAT WAS. IT’S S00 VERY SPECULATIVE. YOU COULD PROBABLY
ARGUE THE OTHER SIDE OF THE COIN AS WELL, EITHER BEING THE
pb
OG.
W
o
O
R
h
N
1085
BALDUS - DIRECT
DEVIL‘S ADVOCATE OR BECAUSE YOU REALLY BELIEVE IT. EITHER ONE.
THAT WAS THE TROUBLE I HAD WITH THAT AS BEING A PRACTICAL
MEASURE.
BUT IN TERMS OF THE PREPONDERANT MOTIVATING FACTOR OR
ANYTHING LIKE THAT, COULD YOU IN FAIRNESS SAY THAT WHAT CAUSED
MCCLESKEY TO GET THE DEATH PENALTY AS OPPOSED TO ANYBODY ELSE,
WAS THE FACT THAT HE MURDERED A WHITE PERSON AS OPPOSED TO A
BLACK PERSON? |
THE WITNESS: NO. I CAN‘T SAY THAT WAS THE FACTOR. NO.
BUT WHAT I CAN SAY, THOUGH, IS WHEN I LOOK AT ALL THE OTHER
LEGITIMATE FACTORS IN HIS CASE, AND I LOOK TO THE MAIN LINE OF
CASES IN THIS JURISDICTION, STATEWIDE, THAT ARE LIKE HIS,
PARTICULARLY THE WAY B2 CASES AND CASES INVOLVING OFFICER
VICTIMS ARE DISPOSED OF IN THIS JURISDICTION, HIS CASE IS
SUBSTANTIALLY OUT OF LINE WITH THE NORMAL TREND OF DECISION ON
SUCH CASES. AND GIVEN THAT IT IS ABERRANT IN THAT REGARD WE ARE
FORCED TO ASK OURSELVES WHAT COULD CAUSE IT.
I CANT SEE ANY FACTORS, LEGITIMATE FACTORS IN HIS CASE
THAT WOULD CLEARLY CALL FOR IT,» THAT WOULD DISTINGUISH IT
CLEARLY FROM THE OTHER CASES. THE CASES ARE NOT IDENTICAL. BUT
THERE‘S NOTHING REALLY CRIES OUT FOR WHY THIS CASE SHOULD BE
TREATED THAT MUCH DIFFERENTLY.
SO YOU’RE LEFT WITH WHAT OTHER FACTOR IT MIGHT BE, AND
WHAT I CAN SAY. AND WHAT I DO SAY IS THAT THE RACIAL FACTOR IS
POSSIBLY THE THING THAT MADE THE DIFFERENCE IN THE CASE. REAL
1026
BALDUS - DIRECT
pt
POSSIBILITY IN MY ESTIMATION, THAT THAT’S WHAT MADE THE
DIFFERENCE. BUT I CANT SAY WITH ANY, I CAN‘T QUANTIFY THE
LIKELIHOOD THAT THAT IS TRUE. THATS AS FAR AS I THINK I CAN GO
IN TERMS OF MAKING RESPONSIBLE JUDGMENT.
THE COURT: IN THE POLICE CASES WE LOOKED AT. ASSUMING
THAT THAT WAS THE FACTOR. IT ONLY SERVED TO PUSH ONE OF THE FIVE
AGGRAVATED WHITE VICTIM CASES OVER THE LINE.
ALL RIGHT. WELL, YOU‘VE DONE ALL YOU CAN DG IN
o
t
h
W
E
"
EA
1
BR
E
©
BE
R
B
R
RESPONSE TO MY QUESTION.
Pa
y o LET ME ASK YOU A QUESTION. MR. BOGER, WHICH OCCURRED TO
11 ME IN LOOKING AT THESE THINGS TODAY.
12 IF I FACTUALLY FOUND WHAT YOU WOULD HAVE ME FIND. TO
13 CONCLUDE AS A MATTER OF LAW THAT BECAUSE THIS EFFECT IS
14 OPERATIVE IN THE SYSTEM SOMEBODY IS BEING DEPRIVED OF EQUAL
15 PROTECTION. WHY WOULD I NOT HAVE TO GRANT THE SAME RELIEF FOR
16 EVERYBODY CONVICTED OF MURDER IN THIS STATE, SERVING A LIFE
17 SENTENCE?
18 MR. BOGER: WELL, I SUPPOSE THE NARROW ANSWER IS
ww 19 BECAUSE WE HAVEN’T ASKED FOR THAT RELIEF, AND WE AREN‘T
20 REPRESENTING THOSE, THOSE PEOPLE.
21 OF COURSE. WE HAVE ALWAYS BASED OUR CLAIM BOTH AS YOU
22 KNOW ON THE EIGHTH AND 14TH AMENDMENT AND UNDER THE EQUAL
23 PROTECTION CLAUSE. AND I SUGGEST TO THE COURT THAT THE STANDARDS
24 MAY BE MORE STRINGENT UNDER THE 3TH AMENDMENT. THE CIRCUIT HAS
25 NOT REALLY HAD TO FACE A FACTUAL CASE LIKE THIS.
oe
L
O
R
B
a
T
E
E
R
M
R
E
Bo
h
RE
E
o
T
I
-
0
Pt
nN
1037
BALDUS - DIRECT
OUR CONTENTION IS THAT WHILE THE EQUAL PROTECTION
CLAUSE IS CERTAINLY SUFFICIENT, THE 8TH AMENDMENT MAY PROVIDE
THE KIND OF DISTINCTION BETWEEN THE CONSTITUTIONAL REQUIREMENT
TO VACATE ALL DEATH SENTENCES IMPOSED UNDER THIS KIND OF SYSTEM.
AND THE ABILITY TO LEAVE STANDING NON-DEATH CASES. EVEN WITH THE
PRESENCE OF SOME RACIAL DISCRIMINATION OF THE SORT OUTLINED.
IN OTHER WORDS. THE SUPREME COURT. FOR EXAMPLE. IN THE
BECK VERSUS ALABAMA CASE SAID THERE ARE PRACTICES GOING ON HERE
WHICH AS A MATTER OF STATE LAW IN NON-CAPITAL CONTEXT DO NOT
REQUIRE THAT THE PRACTICES BE CEASED. BUT IN A CAPITAL CONTEXT,
EVEN THOUGH THE PRACTICES HERE AFFECT ONLY THE GUILTY PHASE. AND
NOT THE PENALTY PHASE, THE LIKELIHOOD THAT DEATH IS A
POSSIBILITY REQUIRES A DIFFERENT OUTCOME. IT’S AS IF THE
SUPREME COURT IS SUGGESTING THAT HIGHER STANDARDS OF FAIRNESS
ARE REQUIRED IN CAPITAL CASES, THE GUILT AND THE PENALTY PHASE.
AND THERE MIGHT BE A BASIS FOR DISTINCTION ON THAT GROUND. EVEN
IF THE JUDGMENT WERE WON ON EQUAL PROTECTION GROUNDS.
THE COURT: I IMAGINE IT DEPENDS ENTIRELY ON YOUR FOINT
OF VIEW. IT YOU‘RE ON DEATH ROW, I“M SURE YOU WOULD FEEL THAT
WAY. BUT IF YOURE DOING LIFE AS OPPOSED TO FIVE OR SIX YEARS
FOR MANSLAUGHTER. YOU’D PROBABLY GET EQUALLY TENSE ABOUT THE
DISPARITY BASED ON THE RACE OF THE VICTIM OR RACE OF THE
DEFENDANT.
MR. BOGER: I“M SURE THE INDIVIDUALS WOULD AND THATS
WHY I’M THANKFUL THAT 1M HERE NOT REPRESENTING ALL OF THEM.
RE —— - BE a EE I - 3 JUICES SUIT Ea ERSSEPEE A Se I
1088
BALDUS - DIRECT
HA
Y THE COURT: ALL RIGHT. LET'S TAKE OUR MORNING RECESS,
2 UNTIL ABOUT 11:30 AND THEN WE WILL COMMENCE CROSS-EXAMINATION.
3 MR. BOGER: IF I COULD INTERRUPT JUST A MOMENT. YOUR
4 HONOR. BECAUSE --—
3 THE COURT: YES. I“M SORRY.
b MR. BOGER: UNFORTUNATELY. WHILE YOU WERE ASKING A FINAL
7 QUESTION, I FOUND ONE FINAL DOCUMENT. THAT NEEDS TO BE IN
8 THE RECORD IN ORDER ™ BE CLEAR. WE HAVE DB-79 ALREADY MARKED
? FOR INDENTIFICATION. I DON’T BELIEVE IT’S BEEN ADMITTED INTO
10 EVIDENCE. I/D LIKE TO SHOW PROFESSOR BALDUS WHAT’S BEEN MARKED
11 AS DB-7%9A.
12 BY MR. BOGER:
13 @. CAN YOU IDENTIFY THAT DOCUMENT. PROFESSOR BALDUS?
14 A. YES. DB-79A IS A DOCUMENT WHICH PRESENTS ESSENTIALLY THE
15 MATERIAL INCLUDED IN DB-79 WITH SOME TECHNICAL CHANGES MADE TO
14 4 f°
17 R. WHAT ARE THOSE CHANGES?
18 A. THE CHANGES GROW OUT OF THE FACT THAT THESE TWO RUNS. THE
» 1% TWO STATISTICAL RUNS THAT. WHOSE RESULTS WERE PRESENTED IN TABLE
20 4 HAD SLIGHTLY DIFFERENT VARIABLES IN THEM. THERE WERE TWO OR
a THREE VARIABLES THAT WERE NOT COMPARABLE IN THE TWO ANALYSES,
22 AND I CONSIDERED IT IMPORTANT THAT. THAT THE TABLE ACTUALLY
23 INDICATE THAT. IT IN NO WAY AFFECTS. THE RESULTS ARE IDENTICAL
24 WITH RESPECT TO THE RACE OF THE VICTIM AND RACE OF THE
23 DEFENDANT.
III sos nes, ——— ————. ———" A IO. A — “A... ABST. FOOD. PASE AORTA
R
e
BE
R"
SU
SE
ES
J
E
R
E
C
C
V
E
WY
oO
rw
1089
BALDUS - DIRECT
THE OTHER, SOME OF THE OTHER CO-EFFICIENTS CHANGE
SLIGHTLY.
Q. SO DB-79A IS A MORE ACCURATE REPORT OF THE RESULTS OF THE
REGRESSIONS ABOUT WHICH YOU PREVIOUSLY TESTIFIED?
A. YES. THAT'S RIGHT.
MR. BOGER: YOUR HONOR, AT THIS TIME. ID MOVE THE
ADMISSION OF DB-79A INTO EVIDENCE. WE‘VE ALREADY GOT TESTIMONY
ON WHAT DB-79 CONTAINS. IT’S SIMPLY TO CLEAR UP THE CHANGE IN
THE CO-EFFICIENTS.
THE COURT: IS -7% ADMITTED?
MR. BOGER: MY IMPRESSION IS -79 IS NOT EVIDENCE, SO WE
DON’T NEED TO CLARIFY OR SUBSTITUTE. BUT IN THE SENSE OF ADD TO
THE RECORD OR CHANGE NUMBERS, BUT SIMPLY ADMIT DB-79A AS THE
MORE ACCURATE REFLECTION OF THE REGRESSION STUDY.
THE COURT: DID YOU OFFER -79 AND I NOT ADMIT IT?
MR. BOGER: MY RECOLLECTION, I CHECKED WITH THE CLERK’S
NOTES THIS MORNING. IS WE NEVER OFFERED -79.
IT DOES CONTROL AS YOU“LL SEE FROM THE TITLE. FOR TEN
STATUTORY AGGRAVATING CIRCUMSTANCES.
. 73 ADDITIONAL MITIGATING FACTORS. ET CETERA.
SO THERE QUITE A FEW CONTROLS IN THIS ONE. WHEN WE
TALKED ABOUT IT, I NEVER GOT AROUND TO MOVING ITS ADMISSION.
THE COURT: DO YOU HAVE ANY SPECIFIC OBJECTION TO THIS?
MS. WESTMORELAND: YOUR HONOR, I WAS CHECKING MY NOTES
AS WELL, AND I SEE AN INDICATION UNDER DB-79 THERE WAS SOME
1090
BALDUS - DIRECT
Fo
y CONCERN EXPRESSED AT THAT TIME, UNTIL IT WAS SHOWN WHAT WAS
2 CONTROLLED FOR. AND I DON‘T THINK THAT HAS BEEN SHOWN AT THIS
3 TIME.
4 OTHER THAN THAT, THERE IS THE CONTINUING OBJECTIONS,
S AND I HAVE NO ADDITIONAL OBJECTION.
il é MR. BOGER: AT THAT POINT, YOUR HONOR, WE HAD NOT
7 OFFERED WHAT BECAME DB-94A, WHICH WAS THE TECHNICAL APPENDIX. NOR
8 |HAD WE OFFERED 22A, TABLE 22.
; BY MR. BOGER:
10 @G. PROFESSOR BALDUS, CAN YOU TELL US BRIEFLY WHAT IS CONTROLLED
11 FOR HERE, BY REFERENCE TO THE OTHER DOCUMENTS THAT ARE IN
12 EVIDENCE, IF YOU HAVE THEM?
13 A. YES, THE BACKGROUND ON RACIAL FACTORS CONTROLLED FOR ABOVE
14 AND BEYOND THE STATUTORY AGGRAVATING FACTORS ARE LISTED IN
15 SCHEDULE 1 OF THE TECHNICAL APPENDIX, WHICH IS --
16 R. DB-946A, I BELIEVE?
17 A. DB-%6A.
18 MR. BOGER: THAT WAS THE DOCUMENT. I BELIEVE, THAT YOUR
pil 19 HONOR ASKED US TO SPELL OUT IN ENGLISH WHAT THOSE FACTORS WERE.
20 THOSE ARE NOW IN ENGLISH IN DB-%6A.
21 THE COURT: ALL RIGHT. ILL ADMIT IT.
22 MR. BOGER: THANK YOU, YDUR HONOR.
23 ~ THE COURT: ALL RIGHT, SIR. LET’S BE IN RECESS UNTIL
24 ABOUT 25 MINUTES TILL TWELVE.
C
a
BE
E"
ST
E
E
:
P
E
L
E
e
S
-
Q
33
12
1091
BALDUS -~ CROSS
(RECESS TAKEN.)
THE COURT: ALL RIGHT, MS. WESTMORELAND.
DAVID C. BALDUS,
|CALLED AS A WITNESS ON BEHALF OF THE PLAINTIFF, BEING FIRST
DULY SWORN, TESTIFIED AS FOLLOWS:
CROSS-EXAMINATION
BY MS. WESTMORELAND:
@. PROFESSOR BALDUS, IN CONSIDERING THE DATA IN BOTH OF YOUR
STUDIES, PROCEDURAL REFORM STUDY AND THE CHARGING AND SENTENCING
STUDY, AS I UNDERSTAND THE NUMBERS CORRECTLY. THEY REFLECT THE
STATUS OF THE PERSON AT THE TIME YOU DID THE STUDY, IS THAT
CORRECT. THE SENTENCING STATUS. THAT IS?
A. NO, THE, THE OBJECT WAS TO ENTER THE SENTENCE THAT WAS
IMPOSED BY THE SENTENCING AUTHORITY AT THE TIME OF THE
DISPOSITION OF THE CASE WITHOUT REGARD TO CHANGES ON APPEAL.
THERE“S ONE EXCEPTION TO THAT. THAT IS, THAT IN THE, IN
THE PENALTY TRIAL PHASE OF THE STUDY, WE TREATED EACH PENALTY
TRIAL AS A SEPARATE EVENT SO THAT IN THAT CASE. THERE WERE
TWELVE OFFENDERS WHO RECEIVED MORE THAN ONE DEATH SENTENCE. SO
THOSE CASES CAME IN AND, NOT COUNTED TWICE. BUT THOSE TWO DEATH
SENTENCE DECISIONS ARE REFLECTED IN THAT POOL OF CASES.
BUT ASIDE FROM THOSE CASES, THE CODING OF THE CASE
TENDED TO REFLECT THE DISPOSITION IN THE FIRST INSTANT.
R. IN REGARD TO, I BELIEVE YOU HAVE, IS IT 128 DEATH SENTENCES
1092
BALDUS - CROSS
p>
IN YOUR CHARGING AND SENTENCING STUDY, IS THAT CORRECT?
A. THAT’S RIGHT.
@. WOULD THOSE REFLECT THE MULTIPLE PENALTY TRIAL CASES. OR DO
NOT?
A. NO. THEY DO NOT.
@. IN REFERRING BACK AND I HATE TO BACKTRACK QUITE SO FAR. BUT
JUST FOR A FEW QUESTIONS BACK TO THE SUPREME COURT QUESTIONNAIRE
WHICH WE DISCUSSED EARLY LAST WEEK, WAS YOUR PRINCIPAL OBJECTIVE
Og
0
N
G
G
N
IN OBTAINING THE INFORMATION FOR THIS QUESTIONNAIRE TO TRY AND
Pa
ws
oO
DEAL WITH THE CASES AS THEY WERE SEEN BY THE GEORGIA SUPREME
-
P
y
COURT?
Fy
tJ
A. THE OBJECT WAS TO REPRESENT THE FILES OF THE GEORGIA SUPREME
13 COURT THAT WE CONSULTED, THAT IS, THE BRIEFS OF COUNSEL IN THE
14 CASE, AS WELL AS THE OPINION OF THE GEORGIA SUPREME COURT, TO
13 PRODUCE A PICTURE OF THE CASE AS IT WAS PERCEIVED BY THE
16 |PRINCIPAL ACTORS IN THE SYSTEM, THAT IS. THE PROSECUTORS AND THE
17 JURIES WHO DISPOSED OF THE CASES.
13 @. IN CODING THAT PARTICULAR QUESTIONNAIRE, WERE SUCH THINGS.
wo 19 FOR INSTANCE, AS WHAT WOULD BE THE RANGE OF RELEVANT INFORMATION
20 TO INCLUDE OR HOW MUCH BACKGROUND INFORMATION TO OBTAIN. WERE
2 THESE THINGS THAT WERE LEFT UP TO THE DISCRETION OF THE CODERS?
22 A. WELL, THEY WERE ASKED TO FILL OUT EVERY QUESTION IN THE
23 QUESTIONNAIRE. THE, THE QUESTIONNAIRE, YOU’RE REFERRING MS.
24 WESTMORELAND. TO THE CODING OF THE SUPREME COURT
25 QUESTIONNAIRES, IS THAT RIGHT?
i
$
0
:
0
N
o
g
b
o
PY
o
24
23
1093
BALDUS - CROSS
®@. WELL, ACTUALLY. I GUESS, BACK TO THE SUMMARIES THAT WERE
DONE. I BELIEVE, IN THOSE CASES, IF I‘M NOT MISTAKEN. WHEN THE
SUMMARIES WERE FILLED OUT IN THE SUPREME COURT?
A. YES. THAT WAS THE WORK THAT WAS DONE BY FRED KYLE AND
FRED CHAIKEN IN THE ARCHIVES. STATE OF GEORGIA. AND THERE THEY
WERE ASKED TO OBTAIN INFORMATION ON ANY FACTORS THAT WERE
TOUCHED ON IN THE QUESTIONNAIRE ITSELF. FRED KYLE. WORKING WITH
ME, HAD, HAD DRAFTED THAT QUESTIONNAIRE. I MEAN WE WORKED ON
THIS FOR OVER A YEAR. SO HE WAS INTIMATELY FAMILIAR WITH THE
INFORMATION THAT WAS GOING TO BE REQUIRED TO COMPLETE THE
QUESTIONNAIRE WHEN THE MATERIAL WAS FINALLY SENT TO IOWA.
SO THAT’S WHAT GUIDED HIM AS TO HIS INQUIRIES AS TO
WHAT INFORMATION TO PUT IN HIS ABSTRACTS.
@. WHEN YOU WERE HIRING THE CODERS. AND PRIMARILY FOR THE
SUPREME COURT QUESTIONNAIRE AT THIS POINT IN TIME. WAS YOUR
PRIMARY CONCERN PERSONAL QUALITIES SUCH AS THOSE YOU DISCUSSED
ON DIRECT EXAMINATION, I BELIEVE YOU MENTIONED ATTENTION TO
DETAIL FOR ONE?
A. YES. THAT WAS THE PRINCIPAL CONCERN. BUT IT WAS. I MEAN IT
WAS NOT THE ONLY CONCERN. I WAS ALSO CONCERNED THEY HAD AN
INTEREST IN CRIMINAL LAW, HAD AN INTEREST IN THIS PROJECT 30
THAT THEY WOULD BE ABLE TO HAVE THE MOTIVATION TO DO THE JOB
PROPERLY. SO IT WAS MORE THAN JUST SIMPLY PLAIN INTELLIGENCE.
Q@. WHEN CO-PERPETRATORS WERE BEING CODED, AND I WOULD LIKE TO
ADDRESS THE QUESTION TQ ALL THREE SETS OF QUESTIONNAIRES THAT
O
N
0
.
N
t
h
a
N
©
2
0
hy
we
PA
Y
1
15
16
1094
BALDUS ~ CROSS
WERE ACTUALLY DONE, WAS ANY ATTEMPT MADE TO HAVE THE SAME CODER
DO THE QUESTIONNAIRES FOR ALL CO-PERPETRATORS?
A. TO THE BEST OF MY RECOLLECTION, THE CODING OF
CO-PERPETRATORS WAS NOT DONE BY THE SAME PERSON UNTIL WE REACHED
THE CHARGING AND SENTENCING STUDY.
THE, THAT’S MY RECOLLECTION OF. WHEN THAT POLICY, THE
POLICY THAT ED GATES TESTIFIED TO EARLIER AS TO THE CODING OF
CO-PERPETRATORS, WAS ADOPTED DURING THE PLANNING FOR THE
CHARGING AND SENTENCING STUDY, TO THE BEST OF MY RECOLLECTION.
@. IN CODING THE SUPREME COURT GUESTIONNAIRES, IF I RECALL YOUR
TESTIMONY CORRECTLY. THE CODERS, WELL. CODERS IS PROBABLY NOT
THE BEST WORD. DATA GATHERERS I SUPPOSE, MERELY DID SUMMARIES AT
THE STATE ARCHIVES. IS THAT CORRECT?
A. THAT WAS THE PRINCIPAL PRODUCT OF THEIR WORK.
THEY DID. THEY DID SUMMARIES IN AN ABSTRACT FORM TO
CAPTURE THE PROCEDURAL POSTURE OF THE CASE. AS WELL AS A SUMMARY
OF THE FACTS AND CIRCUMSTANCES OF THE HOMICIDE.
@. AND THEN THE QUESTIONNAIRES WERE THEN CODED BACK IN IOWA. IS
THAT CORRECT?
A. THAT’S CORRECT.
@. SO WHEN YOU WERE CODING THE QUESTIONNAIRES. YOU DID NOT
ACTUALLY HAVE THE ENTIRE FILE IN FRONT OF YOU. BUT YOU HAD THE
SUMMARIES AVAILABLE, IS THAT CORRECT?
A. THAT“S RIGHT. WITH RESPECT TO THOSE QUESTIONNAIRES.
WITH RESPECT TO THE SOME 230 QUESTIONNAIRES THAT WERE
H
N
A
P
W
N
ee
e
e
O
O
Po
y nN
13
14
15
1093
BALDUS - CROSS
CODED IN THAT WAY, THATS CORRECT.
@. AND IN REGARD TO THE INFORMATION OBTAINED THROUGH THIS
QUESTIONNAIRE. THE 230 CASES, I BELIEVE YOU STATED, --
A. UH HUH.
@. -- WERE YOU UNABLE TO OBTAIN THE RACE OF VICTIM ON ALL THOSE
CASES?
A. I DON’T RECALL THE EXTENT TO WHICH THE SUPREME COURT RECORDS
HAD THE INFORMATION ON THE RACE OF THE VICTIM.
WE, WE TREATED THE RACE OF THE VICTIM PROBLEM AS A, AS
A SINGLE PROBLEM. AT THE END OF THE CHARGING AND SENTENCING
STUDY. THAT IS, WE ENTERED THE INFORMATION WE HAD ON THE RACE
OF THE VICTIM AS WE WENT ALONG AND THAT INFORMATION WAS
AVAILABLE TO US. AS I INDICATED. FREQUENTLY IT WAS NOT
AVAILABLE TO US, SO AT THE END WE DEVISED THE STRATEGY OF HOW TO
GET THAT INFORMATION THROUGH THE BUREAU OF VITAL STATISTICS AND
AT THAT POINT. WE DIDN’T DIFFERENTIATE BETWEEN THE CASES IN THE
PROCEDURAL REFORM STUDY THAT HAD COME IN BY WAY OF THE CHARGING
AND SENTENCING, PARDON ME, THAT WERE IN THE PROCEDURAL REFORM
STUDY AND HAD COME THERE BY WAY OF THE SUPREME COURT
QUESTIONNAIRE, AS DISTINCT FROM THE SUBSEQUENT QUESTIONNAIRE
WHICH WE“VE CALLED THE PROCEDURAL REFORM QUESTIONNAIRE.
@. SO THAT WAS INFORMATION THAT WAS ALL OBTAINED, ESSENTIALLY,
AT ONE POINT IN TIME AFTER THE QUESTIONNAIRES HAD BEEN
COMPLETED?
A. THAT’S CORRECT.
1096
BALDUS - CROSS
ob
@. WERE YOU STILL THEN UNABLE TO OBTAIN RACE OF VICTIM ON ALL
2 THE CASES?
3 A. YES. I THINK IN THE SUPREME COURT CASE. THERE WERE, IN THE,
4 PARDON ME, IN THE, PROCEDURAL REFORM STUDY, IF MY RECOLLECTION
3S IS CORRECT. I THINK THERE WERE FIVE CASES WHERE WE WERE UNABLE
4 & TO ASCERTAIN. BUT I“M NOT CERTAIN OF THAT NUMBER. I HAVE A
7 DOCUMENT THAT COULD PERHAPS HELP ME ANSWER THAT QUESTION. IT’S
8 NOT HERE, I HAVE IT BACK TO MY OFFICE. IT WAS A VERY SMALL
? NUMBER. I THINK IT WAS FIVE, ONE OR TWO: SOMEWHERE IN THAT
10 ZONE. I DONT RECALL EXACTLY.
11 @. HOW WERE THOSE CASES TREATED IN YOUR ANALYSIS OF THE
12 PROCEDURAL. REFORM STUDY?
13 A. THOSE CASES WERE RECODED AND FOR THE PRINCIFAL ANALYSIS.
14 THEY WERE RECODED SO THAT THE RACE OF THE VICTIM IN THOSE CASES
15 WAS THE SAME AS THE RACE OF THE DEFENDANT.
146 WE FOUND IN EXAMINING THE CASES WHERE WE KNEW THE RACE
17 OF THE VICTIM THAT THE RACE OF THE VICTIM AND THE RACE OF THE
18 DEFENDANT WAS THE SAME IN 89 PERCENT OF THE CASES.
w 19 SO, ON THE BASIS OF A POLICY OF IMPUTATION. AS THAT
20 PRACTICE IS CALLED, IS IMPUTING A VALUE TO A CASE, WE FELT THAT
21 WE HAD A VERY HIGH PROBABILITY OF CORRECTING THE, PREDICTING
aR WHAT THE CORRECT RACE OF THE VICTIM WAS.
23 HOWEVER, I WILL MENTION THAT WE CONDUCTED ALTERNATIVE
24 ANALYSES IN WHICH THOSE CASES WERE STRICKEN FROM THE FILES AND
25 THE RESULTS WERE IDENTICAL.
1097
BALDUS - CROSS
@. AND YOU CHANGED SUBSEQUENTLY FROM THE SUPREME COURT
QUESTIONNAIRE TO A PROCEDURAL REFORM STUDY QUESTIONNAIRE.
AND AS I UNDERSTAND ONE OF THE PRIMARY REASONS FOR THIS
WAS THE TIME INVOLVED IN CODING THE SUPREME COURT QUESTIONNAIRE,
IS THAT ACCURATE?
A. IT WAS THE TIME INVOLVED AND WHAT I THOUGHT WE WOULD GET
FROM THE TIME, IN THE SENSE THERE WAS MORE DETAIL, FINE GRAIN
DETAIL IN THAT QUESTIONNAIRE THAN I THOUGHT WOULD BE NEEDED TO
CONDUCT OUR ANALYSIS.
@. SO WHEN YOU CHANGED THE QUESTIONNAIRE, YOU DID NECESSARILY
THEN LOSE SOME OF THE DETAIL THAT WAS AVAILABLE IN THE SUPREME
COURT QUESTIONNAIRE?
A. YES, SOME DETAIL WAS LOST. YES.
@. AND I BELIEVE YOU ALSO STATED. DID YOU NOT, THAT YOU DELETED
QUESTIONS REFERRING TO WHETHER THE SENTENCING JURY HAD CERTAIN
INFORMATION OR NOT?
A. YES. WE FOUND WE WERE UNABLE TO, TO PRECISELY DIFFERENTIATE
ON THE BASIS OF THE RECORDS PRESENTED IN THE GEORGIA SUPREME
COURT THAT WE WERE UNABLE TO IDENTIFY IN PENALTY TRIAL CASES
WITHOUT A MASSIVE INQUIRY PRECISELY THE INFORMATION THAT WAS
AVAILABLE TO THE SENTENCING JURY.
@. SO THOSE QUESTIONS WERE SUBSEGUENTLY JUST OMITTED FROM THE
LATER TWO QUESTIONNAIRES. I GUESS?
A. YES. IN NONE OF OUR ANALYSIS DO WE HAVE INFORMATION ON
WHETHER THE JURY WAS AWARE OF IT AND WHETHER OR NOT THEY. OR HOW
1098
BALDUS - CROSS
pb
THEY BECAME AWARE OF IT. THAT WAS AN APPROACH THAT WAS
CONSIDERED, BUT IT WAS NOT FEASIBLE TO CARRY IT OUT.
Q. WHEN YOU OBTAINED INFORMATION FROM THE PROSECUTORS AND
DEFENSE ATTORNEYS. MY UNDERSTANDING OF THE MUESTIONNAIRES OR
|LETTERS YOU SENT OUT WAS THAT YOU ADDRESSED A FEW SPECIFIC
QUESTIONS ON INFORMATION THAT YOU DID NOT HAVE AT THAT TIME,
IS THAT CORRECT?
A. YES, THAT'S RIGHT.
S
e
FE
RC
”
BE
SE
E
1
S
E
E
R
L
@. YOU DID NOT MAKE ANY ATTEMPT TO CHECK INFORMATION THAT YOU
fb
o HAD ALREADY OBTAINED IN THIS WAY?
[r
y
>
A. NO, I DID NOT.
12 @. IN REGARD TO THE QUESTIONNAIRE IN THE PROCEDURAL REFORM
13 STUDY, DID YOU FIND THAT PROBLEMS AROSE BASED ON THE FOIL METHOD
14 UTILIZED IN THIS QUESTIONNAIRE?
135 A. YES.
16 @. DID YOU FIND THAT YOU WERE UNABLE TO PERHAPS INCLUDE ALL
17 NECESSARY INFORMATION IN THIS METHOD?
13 A. WELL, WE WERE NOT ABLE TO INCLUDE ALL OF THE INFORMATION
Eo 19 THAT WAS AVAILABLE IN THE FILES.
20 WE WERE AWARE OF THIS FROM THE OUTSET. IT WAS DUR
21 CONSIDERED JUDGMENT THAT THE FOIL METHOD LIMITED THE AMOUNT OF
22 INFORMATION THAT WE COULD OBTAIN ON SOME OF THE VARIABLES. WE
23 BELIEVED. HOWEVER, THAT THIS INFORMATION WAS CUMULATIVE. AND
24 WOULD NOT AFFECT THE STATISTICAL ANALYSIS.
23 HOWEVER, SINCE THIS ISSUE HAS BEEN RAISED BY THE
O
W
0
N
B
N
[u
l
1099
BALDUS - CROSS
ATTORNEY GENERAL IN THIS PROCEEDING, WE MODIFIED OUR PROCEDURES
AND CHANGED THE FOILS, EXPANDED THEM ALL IN THE CRITICAL
VARIABLES, RE-ANALYZED EACH CASE TO INCLUDE ALL OF THE
INFORMATION THAT HAD BEEN OMITTED ORIGINALLY. AND RECONDUCTED
ALL THE STATISTICAL ANALYSIS, AND WE FOUND THAT THE RESULTS WERE
IDENTICAL, AND IN FACT, THE RACE EFFECTS BECAME SOMEWHAT
INTENSIFIED WHEN THIS ADDITIONAL INFORMATION WAS INCLUDED IN THE
FILES.
SO IT TENDED TO BEAR OUT OUR ORIGINAL JUDGMENT, AND I
CAN SAY THE JUDGMENT OF EVERY EXPERIENCED SOCIAL SCIENTIST I“VE
SPOKEN TO ABOUT THIS QUESTION IS THAT THIS SORT OF ADDITIONAL
INFORMATION HAS A MARGINAL EFFECT IN TERMS OF HOW CASES ARE
RANKED RELATIVE TO ONE ANOTHER IN TERMS OF THEIR SERIOUSNESS,
AND OUR REANALYSIS OF THE CASES AFTER HAVING INCLUDED ALL THIS
ADDITIONAL INFORMATION IN THE FILES, TENDS TO BEAR OUT THAT
OPINION TO A TEE.
Q@. YOU STATE YOU RE-ANALYZED THE CASES IN ORDER TO INCLUDE THIS
ADDITIONAL INFORMATION.
DID YOU GO BACK TO YOUR ORIGINAL DATA SOURCE TO DO THAT
OR DID YOU JUST UTILIZE THE QUESTIONNAIRES AND THE AMBIGUITY
SHEETS YOU HAVE?
A. WE USED THE SUMMARIES. WE LOOKED AT THE SUMMARIES. YOU
SEE, THE IDEA OF THE SUMMARY WAS TO EMPHASIZE IN THE SUMMARY ALL
OF THE INFORMATION. OR ANY INFORMATION THAT WAS NOT INCLUDED IN
THE CODING OF THE QUESTIONNAIRE. S0 THE CODERS WERE SENSITIVE
Nt, iris paren. x it Sindee Ad {ibid imbibe, tana: setatiaih Shite." sim iat, Sittin meta, sete. ev—— Tr’
1100
BALDUS - CROSS
TO THIS QUESTION, THERE WAS A LIMIT ON THE AMOUNT OF INFORMATION
THAT THE FOILS WOULD CARRY. S0 THEY INCLUDED THAT INFORMATION
IN THE SUMMARIES AND WE WENT BACK AND LOOKED AT THOSE SUMMARIES
FOR EVERY CASE. HERE’S PRECISELY WHAT WE DID.
WE WENT DOWN THE CASES WHERE ALL THE FOILS WERE FULL.
THE CASE HAD NOT. IF THE CODERS HADNT FILLED UP THE FOIL. THEN
THERE WOULD BE NO PROBLEM.
WE FOUND, I THINK, 75, 79 CASES, SOMETHING ROUGHLY IN
THAT CATEGORY. WHERE THE FOILS WERE FILLED.
WE PULLED OUT ALL THOSE CASES AND EXAMINED THE
SUMMARIES AND FOUND INDEED THAT IN ABOUT FIFTY OF THOSE CASES
THERE WAS OVERFLOW INFORMATION, INFORMATION THAT WOULDN'T FIT
INTO THE ORIGINAL FOILS.
SO WE CREATED AN ADDITIONAL SERIES OF FOILS TO
ACCOMMODATE ALL THAT INFORMATION, AND THEN RECODED ALL OF THE
VARIABLES THAT WERE USED IN THE MACHINE ANALYSIS IN THE
PROCEDURAL. REFORM STUDY TO REFLECT THAT ADDITIONAL INFORMATION
AND, AS I HAVE SUGGESTED TO YOU, IT SLIGHTLY ENHANCED THE RACIAL
EFFECTS WE OBSERVED, AND ALL OF THE TABLES THAT WE HAVE TENDERED
TO THE COURT THUS FAR ARE BASED UPON THE ANALYSIS THAT REFLECTS
THIS MODIFICATION OF THE FOILS.
@. IN CONSIDERING THE SAME FOIL METHOD. AND RECOGNIZING THE.
INDICATING THAT YOU RECOGNIZE SOME OF THE DRAWBACKS WITH THAT
FOIL METHOD, DID YOU NOT STILL UTILIZE THAT FOIL METHOD ON AT
LEAST, I BELIEVE, AT LEAST TWO QUESTIONS IN THE CHARGING AND
Oo
NN
OC
QR
O
W
1101
BALDUS - CROSS
SENTENCING STUDY?
A. YES, WE DID. AND THEN WE WENT BACK AND FIXED THEM. TOO. WE
DID THE SAME THING IN THE CHARGING AND SENTENCING STUDY WHERE
THERE WAS AN OVERFLOW. WE WENT BACK AND CREATED NEW VARIABLES.
AND IT MADE NOT ONE FRACTION OF A DIFFERENCE IN ALL OF THE
RESULTS.
@. AND ALL OF THIS RECODING WAS BASED ON THE SUMMARIES THAT YOU
HAD OBTAINED?
A. YES.
@. I BELIEVE YOU STATED EARLIER IN THE PROCEDURAL REFORM STUDY
YOU IMPUTED A VALUE FOR RACE OF VICTIM BASED ON THE RACE OF THE
DEFENDANT.
DID YOU ALSO IMPUTE A VALUE WHEN THE PENALTY TRIAL WAS
WHETHER, WHEN IT WAS UNKNOWN WHETHER A PENALTY TRIAL OCCURRED?
A. YES. THERE WERE 23 OR 24 CASES WHERE WE WERE UNABLE TO
ASCERTAIN BY, BY OUR QUESTIONNAIRES TQ PROSECUTORS AND DEFENSE
ATTORNEYS WHETHER OR NOT THERE HAD BEEN A PENALTY TRIAL IN THE
CASE.
NOW, SOCIAL SCIENCE HAS DEVELOPED A VARIETY OF METHODS
FOR IMPUTING VALUES, AND ONE OF THE MORE RELIABLE ONES IS THE
METHOD OF REGRESSION IMPUTATION.
WE DEVELOPED A MODEL THAT WOULD ENABLE US TO PREDICT A
LIKELIHOOD OF A PENALTY TRIAL FOR ALL THE CASES ON WHICH WE HAD
ACTUAL INFORMATION.
THEN WE PROCEEDED TO MAKE A PREDICTION WITH RESPECT TO
ob
vw
0
N
O
O
p
W
N
10
i
1102
BALDUS — CROSS
THOSE 23 CASES WHERE WE DIDN’T KNOW THE ANSWER, AND ON BASIS OF
THAT WE IMPUTED A VALUE.
I CAN POINT OUT ALSO WE CONDUCTED ALTERNATIVE ANALYSES
WITH ALL THOSE CASES STRICKEN FROM THE FILE, AND THE RESULTS
WERE THE SAME.
THAT IMPUTATION PROCESS HAD NO EFFECT WHATEVER ON THE
SUBSTANTIVE RESULTS.
@. WHEN CO-PERPETRATORS WERE BEING CODED IN THE PROCEDURAL
REFORM STUDY, WERE THEY GENERALLY, OR WERE THE INSTRUCTORS
INSTRUCTED. THE CODERS INSTRUCTED. EXCUSE ME,» TO ATTRIBUTE ALL
OF THE AGGRAVATING AND MITIGATING CIRCUMSTANCES OF THE OFFENSE
TO EACH CO-PERPETRATOR?
A. WELL, AS I INDICATED TO YOU EARLIER. THE CHARGING, PARDON
ME, I’M SORRY. MS. WESTMORELAND. WERE YOU SPEAKING OF THE
CHARGING AND SENTENCING STUDY OR THE --
@. THE PROCEDURAL REFORM STUDY.
A. OH. THE PROCEDURAL REFORM STUDY.
| IN THE PROCEDURAL REFORM STUDY WE DID NOT HAVE A
POLICY, TO THE BEST OF MY RECOLLECTION, OF VIEWING THE
CO-PERPETRATORS AS A UNIT.
WE HIT THE CO-PERPETRATORS AS THEY CAME IN THE LIST OF
CASES TO BE CODED, SO THERE WAS NEVER AN ATTEMPT MADE TO, AS FAR
AS I KNOW. TO COMPARE THE CODING ACROSS THE CO-PERPETRATOR CASES
IN THE PROCEDURAL REFORM STUDY.
SO CONSEQUENTLY. THAT POLICY COULDN’T HAVE PERTAINED.
p-
v
O
O
N
0
A
ep
W
N
—-
oO
—
-
-
1103
BALDUS - CROSS
BECAUSE THEY WERE NOT MAKING THE COMPARISON.
A. TAKING A CASE SEPARATELY, WAS THERE ANY ATTEMPT TO
DISTINGUISH IN TERMS OF CODING, FOR INSTANCE. STATUTORY
AGGRAVATING CIRCUMSTANCES, WAS ANY ATTEMPT MADE TO DISTINGUISH
ANY OF THOSE CASES WHERE THE DEFENDANT WAS A NON-TRIGGER MAN,
FOR INSTANCE. OR WERE THEY CODED AS HAVING ALL THE AGGRAVATED
FEATURES OF THE OFFENSE, EVEN THOUGH THEY MIGHT NOT HAVE
DIRECTLY PARTICIPATED?
A. THERE ARE TWO RESPONSES. FIRST. THERE WAS ATTENTION GIVEN
TO WHETHER OR NOT THE OFFENDER WAS A TRIGGER MAN. THAT'S AN
EXTREMELY IMPORTANT VARIABLE. EVERY ANALYSIS SUGGESTS THAT HAS
A BIG EXPLANATORY EFFECT.
THAT”S A DIFFERENT QUESTION, HOWEVER, BECAUSE THAT IS
NOT, THAT 1S A DIFFERENT QUESTION FROM THE ISSUE OF WHETHER OR
NOT THE STATUTORY AGGRAVATING CIRCUMSTANCES WERE IN THE CASE.
WHAT STATUTORY AGGRAVATING CIRCUMSTANCES WERE IN THE CASE.
BEING A TRIGGER MAN DOES NOT MAKE ONE DEATH ELIGIBLE
UNDER ANY PARTICULAR STATUTORY AGGRAVATING FACTOR.
WHAT MAKES A PERSON DEATH ELIGIBLE IS THE PRESENCE IN
THAT CASE OF THE STATUTORY AGGRAVATING FACTORS. SO THE CODERS
WERE INSTRUCTED WITH RESPECT TO ALL CASES. REGARDLESS OF THE
ROLE THAT THE OFFENDER PLAYED, TO CODE THE VARIABLES THAT. TO
CODE ALL OF THE STATUTORY AGGRAVATING FACTORS THAT PERTAINED TO
THAT CASE. WHETHER THE DEFENDANT WAS THE TRIGGER MAN OR AN
UNDERL ING.
1104
BALDUS - CROSS
pt
BECAUSE IT WAS THOSE FACTORS IN THAT CASE THAT MADE THE
DEFENDANT, EVEN THOUGH HE WASN’T THE TRIGGER MAN. DEATH
ELIGIBLE. AND THAT'S THE, THE INFORMATION WE WANTED THAT FILE
TO CARRY: WAS TO BE ABLE TO IDENTIFY WHO WAS DEATH ELIGIBLE.
@. SO YOU WERE LOOKING AT CIRCUMSTANCES IN THE CASE IN TERMS OF
LIKE STATUTORY AGGRAVATING CIRCUMSTANCES THEN?
A. WELL, WE LOOKED AT THAT. AND THEN WE HAD A WHOLE SERIES OF
QUESTIONS IN EACH QUESTIONNAIRE THAT WENT TO THE ROLE OF THE
PARTICULAR DEFENDANT IN THE MURDER AND IN ANY CONTEMPORANEOUS
O
Y
0
N
p
W
N
F
y
OFFENSE THAT MIGHT HAVE OCCURRED.
SO,» WE WANTED TO CAPTURE THE ESSENCE OF DEATH Ty
[W
's
pt
8 WORTHINESS, BUT WE ALSO WANTED TO DIFFERENTIATE BETWEEN THE ROLE
13 EACH DEFENDANT PLAYED WHEN THEY WERE CO-PERPETRATORS.
14 @. PROFESSOR BALDUS, —-
13 THE COURT: WAIT JUST A SECOND.
14 YOU MEAN IN YOUR SECOND STUDY THERE WERE 23, 24 CASES
17 IN YOUR SAMPLE WHERE YOU DIDN’T KNOW WHETHER OR NOT THERE WAS A
18 FENALTY TRIAL?
w 19 THE WITNESS: IN THE FIRST STUDY. YOUR HONOR.
20 THE COURT: IN THE FIRST STUDY?
21 THE WITNESS: RIGHT.
22 THE COURT: OKAY.
23 HOW DID YOU DECIDE WHICH CASES TO IMPUTE AS BEING THE
24 |ONE WHERE THERE WAS A PENALTY TRIAL.
23 THE WITNESS: WE TOOK THE CASES WHERE WE KNEW THERE WAS
1105
BALDUS - CROSS
1 PENALTY TRIAL. YOUR HONOR, AND DID A REGRESSION ANALYSIS THAT
2 WOULD ENABLE US TO PREDICT HOW LIKELY AN INDIVIDUAL CASE WOULD
BE TO HAVE A PENALTY TRIAL.
THE COURT: YOU THINK THE SYSTEM IS THAT NON-ARBITRARY
THAT YOU COULD MAKE THAT SORT OF IMPUTATION?
3
4
S
e 6 THE WITNESS: WELL, I THOUGHT 50. WE HAD FAIRLY GOOD
7 PREDICTORS. I THINK THE CASES THAT WE HAD, WHERE WE DIDNT HAVE
8 THIS INFORMATION, WERE UNAPPEALED CASES. THAT’S WHY WE COULDNT
J CONSULT THE SUPREME COURT RECORDS AND DEFENSE ATTORNEYS AND
10 PROSECUTORS DID NOT RESPOND TO OUR REQUEST. THAT'S WHY WERE IN
11 THIS POSITION.
12 SO WE FELT THAT AS AN ALTERNATIVE TO DELETING THOSE FROM
13 THE CASES ENTIRELY AND LOSING THE INFORMATION ABOUT THOSE CASES:
14 THAT THEY WERE IMPORTANT TO INCLUDE IN THIS WAY.
135 I WILL SAY WE ADOPTED AN ALTERNATIVE APPROACH, AND THAT
16 WAS TO DELETE THEM FROM THE ANALYSIS AND BY DOING SO, WE LOST
17 CERTAIN INFORMATION ABOUT THOSE CASES, PARTICULARLY WHEN THE
138 ISSUE WAS WHETHER THERE WAS A DEATH PENALTY. I MEAN WE KNEW
w 19 THERE WAS NO DEATH PENALTY IN THE CASE. WE WOULD LOSE A CERTAIN
20 AMOUNT OF INFORMATION WHEN THEY WERE DELETED.
<1 WE DID THE ANALYSIS THAT WAY. WE INCLUDED THEM USING
22 THE IMPUTATIONS THAT WE HAD DEVELOPED AND WE FOUND THERE WAS NO
23 DIFFERENCE IN THE RESULT. THE REASON IS, THAT WHEN YOU“RE
24 DEALING WITH THAT -—-—
23 THE COURT: I DIDN’T ASK BUT ONE QUESTION, AND YOU
11046
BALDUS - CROSS
Fo
s ANSWERED THAT. THANK YOU.
BY MS. WESTMORELAND:
@. PROFESSOR BALDUS, WHEN WAS THE LAST TIME THAT YOU HAVE
UPDATED, I BELIEVE YOU STATED YOU HAD UPDATED SOME OF THE
INFORMATION IN THE PROCEDURAL REFORM STUDY.
WHEN IS THE LAST TIME YOU UPDATED THAT INFORMATION?
A. OH, WITHIN A WEEK, TWO WEEKS, TWO WEEKS OF COMING HERE TO
THIS PROCEEDING. I WOULD SAY THAT ON THE SECOND OR THIRD OF
C
E
E
T
U
BR
B
E
B
a
R
E
S
L
GEORGIA. I MEAN. PARDON ME, THE SECOND OR THIRD OF AUGUST IS
po
te
<Q
WHEN WE MADE THE LAST CHANGES.
i1 SEE, WE UNDERTOOK THIS MASSIVE UPDATING OF THE FILES TO
12 MEET THE OBJECTIONS THAT HAD BEEN RAISED ABOUT THE OMISSIONS OF
13 THESE OVERFLOW CHARACTERISTICS THAT WOULDN’T FIT INTO THE FOILS
14 AND THAT WAS A BIG JOB TO UNDERTAKE THAT. AND WE DIDN‘T GET
15 THAT. THAT JOB COMPLETED UNTIL ABOUT THE FIRST WEEK OF AUGUST
146 AND ONCE IT WAS COMPLETE, WE RERAN ALL THE ANALYSIS. SO THAT'S
17 ABOUT WHEN IT WAS COMPLETED.
18 @. IN QUESTIONNAIRES IN THE PROCEDURAL REFORM STUDY. IF THE
w» 19 CODER DOING THE QUESTIONNAIRE MADE WRITTEN COMMENTS BESIDE A
20 PARTICULAR QUESTION, IF HE FOR INSTANCE WROTE "MAYBE" OR
21 "POSSIBLY" BESIDE A PARTICULAR RESPONSE TO A QUESTION, HOW WOULD
22 THAT BE TREATED WHEN THE QUESTIONNAIRE WAS LATER CODED FOR THE
23 COMPUTER?
24 A. WHEN THE CODERS MADE COMMENTS OR HAD QUESTIONS ABOUT A
23 VARIABLE, THEY WERE INVITED TO MAKE SUCH COMMENTS. THEY WERE
T
R
L
T
I
E
e
Pe
To
>
OO
[o
y nN
13
14
1107
BALDUS - CROSS
INSTRUCTED TO BRING IT TO MY ATTENTION AND I WOULD LOOK AT IT,
LOOK AT THE SUMMARY AND MAKE A DETERMINATION AS TO HOW IT SHOULD
G0. THAT’S THE WAY THE PROCEDURE OPERATED.
@. SO YOU MADE THOSE DETERMINATIONS AFTER THE QUESTIONNAIRES
WERE COMPLETED AND -——
A. YES.
@. -— BASED ON THE QUESTIONNAIRE AND THE SUMMARY?
A. THAT’S RIGHT. AND FREQUENTLY IN CONSULTATION WITH THE CODER
TO GET A SENSE OF WHAT THEY THOUGHT THE ISSUE WAS.
@. IN THE PROCEDURAL REFORM STUDY WHEN CONSIDERING THE
STATUTORY AGGRAVATING CIRCUMSTANCES. DIDNT YOU INCLUDE MULTIPLE
VICTIMS UNDER THE B2 AGGRAVATING CIRCUMSTANCE?
A. YES.
@. DID YOU CONSIDER THAT NO MATTER WHEN THE KILLINGS OCCURRED?
A. WELL. THE REASON FOR THAT WAS THE. THAT THE GEORGIA SUPREME
COURT HELD IN GODFREY, TO MY RECOLLECTION, THAT IS THE SECOND
GODFREY V. GEORGIA CASE AFTER IT WAS REMANDED FROM THE UNITED
STATES SUPREME COURT ON THE GROUND THAT B7 HAD BEEN
INAPPROPRIATELY APPLIED, THE GEORGIA SUPREME COURT VACATED THAT
SENTENCE AND WENT BACK AND GOT A NEW TRIAL. AND GODFREY RECEIVED
A DEATH SENTENCE. THIS WAS BASED ON B2 FINDING.
WHEN THAT CASE WAS APPEALED TO THE GEORGIA SUPREME
COURT, THAT BODY HELD THAT CONTEMPORANEOUS MURDER. YOU KNOW.
ANOTHER KILLING AT THE TIME OF THE PRINCIPAL KILLING. IS
SUFFICIENT TO MAKE YOU DEATH ELIGIBLE UNDER B2, BECAUSE BZ SAYS
TY
H
M
o
N
o
s
B
N
1108
BALDUS - CROSS
YOURE DEATH ELIGIBLE IF YOU ARE. HAVE ALSO COMMITTED
CONTEMPORANEQUSLY ANOTHER CAPITAL CRIME AND THE SECOND MURDER IS
ANOTHER CAPITAL CRIME, CONTEMPORANEOUS TO IT. THAT S WHY WE
CHANGED THE CODING AT THAT TIME.
@. DID YOUR CODING INCLUDE, WAS IT PHRASED SO AS TO INCLUDE
MULTIPLE VICTIMS?
A. YES.
@. OR WAS IT TO INCLUDE A CONTEMPORANEOUS MURDER?
A. NO, IT WAS MULTIPLE VICTIMS. WE DID NOT DIFFERENTIATE A
MURDER CASE. WE VIEWED ALL THE CONTEMPORANEOUS KILLINGS AS
MURDERS.
@. WHEN YOU UPDATED THE FILE ON THE PROCEDURAL REFORM STUDY.
DID YOU ALSO UPDATE YOUR TOTAL VARIABLES?
A. YOU MEAN THE COUNTING, THE VARIABLES THAT COUNT?
@. YES?
A. YES. WHAT WE DID WAS WRITE AN EXTENSIVE PROGRAM THAT WOULD
ALLOW US TO MODIFY EVERY VARIABLE IN THE FILE AS A RESULT OF
THESE UPDATES THAT WERE MADE. IT“S THE SAME TYPE OF SYSTEM THAT
WE HAD DEVELOPED IN THE CHARGING AND SENTENCING STUDY SO THAT
EVERYTIME YOU MAKE AN ENTRY INTO THE UNDERLYING DATA BASE. YOU
RUN A PROCEDURE WHICH WE CALLED TASK, AND THAT HAS THE EFFECT OF
RECODING AUTOMATICALLY, EVERY VARIABLE THAT”3 USED IN THE
STATISTICAL ANALYSIS. AND THAT WAS THE PROCEDURE DEVELOPED AND
APPLIED WITH RESPECT TO THE PROCEDURAL REFORM STUDY.
©. WHEN YOU UPDATED OR WHEN YOU RECODED THE UNKNOWN PENALTY
0
N
O
1109
BALDUS ~ CROSS
TRIALS IN THE PROCEDURAL REFORM STUDY, TO BACKSPACE JUST A
LITTLE BIT. DO YOU HAPPEN TO RECALL WHAT REGRESSION YOU USED TO
RUN THAT PROCEDURE?
A. I THINK IT WAS EITHER X481P OR ONE SUBSEQUENT TO IT. BUT
THEY ARE VERY COMPARABLE. THE RESULTS UNDER THE TWO WOULD BE
VIRTUALLY IDENTICAL. OR RATHER I THINK IT“S P481 I AND D, OR, I
THINK IS THE THE LATEST VERSION OF IT. AND IN THE EARLIER
VERSIONS MAY HAVE HAD ONE OR TWO MORE OR LESS VARIABLES. BUT THE
PREDICTIONS WOULD BE BASICALLY THE SAME FOR THE TWO. I BELIEVE.
@. DID YOU RELY ON ANY PARTICULAR AUTHORITY IN MAKING THE
DECISION TO RECODE PENALTY TRIALS IN THIS MANNER?
A. YES. I CONSULTED WITH PROFESSOR WOODWORTH ABOUT THE METHOD
OF DEALING WITH MISSING VALUES OF THIS TYPE. AND I HAVE. I
RELIED ON HIS PROFESSIONAL JUDGMENT ABOUT IT.
SINCE THEN I“VE CONSULTED THE LITERATURE MYSELF AND
THERE“S A CITATION TO THE PRINCIPAL AUTHORITY THAT I LOOKED To,
THAT’S CITED AS A FOOTNOTE IN THE REPORT.
@. WOULD THAT BE THE ARTICLE BY KALTON, I BELIEVE, KOLTON?
A. WOULD YOU LIKE ME TO CHECK? I“VE GOT TQ CHECK THE REPORT TO
SEE WHAT THAT ACTUAL --
®@. IF YOU WOULD, PLEASE.
A. YES, THE, FOOTNOTE 1 ON PAGE 21 OF THE REPORT.
@. AND WHO WROTE THAT ARTICLE, PLEASE?
A. G. KALTON, K-A-L-T-0-N.
@. THANK YOU.
1110
BALDUS - CROSS
PROFESSOR BALDUS, IN THE PROCEDURAL REFORM STUDY AND IN
CODING THE PROCEDURAL REFORM STUDY, WOULD IT BE POSSIBLE THAT
ONE CO-PERPETRATDOR WOULD RECEIVE OR BE CODED AS HAVING
AGGRAVATING CIRCUMSTANCES WHICH WERE NOT ATTRIBUTED TO ANOTHER
CO-PERPETRATOR?
A. YES, THAT WOULD BE A POSSIBILITY.
@. WOULD, COULD THAT COME ABOUT JUST DUE TO DIFFERENT CODERS
CODING THE TWO CASES?
A. WELL, THEY COULD COME ABOUT FROM THAT. IT COULD COME ABOUT
FROM THE FACT THAT SOME OF THE CASES IN THE STUDY WERE CODED IN
THE SUPREME COURT RECORDS, AND THE SOME OF THEM WERE CODED IN
THE PAROLE BOARD. SO YOU HAVE DIFFERENT INFORMATION ABOUT THE
CASES. AND ALSO YOU HAVE DIFFERENT CODERS.
MS. WESTMORELAND: I WOULD LIKE AT THIS TIME, TO GET A
DOCUMENT MARKED FOR IDENTIFICATION AS RESPONDENTS EXHIBIT
NUMBER 1. AND I HAVE A COPY FOR THE USE OF THE COURT AS WELL.
UNFORTUNATELY. YOUR HONOR, I DON’T HAVE THESE IN
NOTEBOOKS. I DO APOLOGIZE FOR THE LACK OF CONVENIENCE.
THE COURT: SPEAK TO THE GENERAL ASSEMBLY ABOUT YDUR
APPROPRIATIONS.
MS. WESTMORELAND: I WISH YOU WOULD.
BY MS. WESTMORELAND:
@. PROFESSOR BALDUS, I HAVE HANDED YOU WHAT’S BEEN MARKED AS
RESPONDENTS NUMBER 1.
COULD YOU IDENTIFY THAT DOCUMENT FOR ME, PLEASE?
1111
BALDUS - CROSS
rw
A. YES. THIS APPEARS TO BE A XEROX COPY OF THE PROCEDURAL
2 REFORM STUDY QUESTIONNAIRE THAT WAS CODED BY KATHY CHRISTIAN.
3 @. AND WHAT’S THE OFFENDER’S NAME ON THAT?
4 A. DAVID BURNEY.
S @. AND DO YOU KNOW WHERE THE ORIGINAL OF THIS QUESTIONNAIRE
i & WOULD BE AT THIS TIME?
7 A. I BELIEVE IT WOULD BE IN MY POSSESSION.
8 @. AND DO YOU RECALL PROVIDING COPIES OF SEVERAL GUESTIONNAIRES
2? YO UD w=
10 A. YES.
11 @. -- IN THIS CASE?
12 A. YES, I DO.
13 @. DO YOU RECALL IF THIS WAS ONE OF THE CASES?
14 A. I DON’T RECALL PRECISELY, BUT I MEAN, I WOULD ASSUME THATS
135 THE CASE. I DON’T RECALL EXACTLY BY NAME.
14 ms. WESTMORELAND: I WOULD LIKE TO GET TWO DOCUMENTS
17 MARKED AS RESPONDENT’ S EXHIBITS 2 AND 3.
18 I HAVE COPIES FOR THE COURT AS WELL.
w» 1? MAY I APPROACH THE WITNESS. YOUR HONOR?
20 THE COURT: YES.
21 BY MS. WESTMORELAND:
22 |@. PROFESSOR BALDUS, COULD YOU LOOK FIRST AT WHAT HAS BEEN
23 |LABELED AS RESPONDENT-S NUMBER 2 AND IDENTIFY THAT DOCUMENT FOR
24 US, PLEASE?
25 A. YES. THIS APPEARS TO BE A PHOTOCOPY OF DEFENDANTS EXHIBIT
L
B
R
E
SE
S
O
SE
B
R
De
EE
a
W
O
T
E
S
E
F
R
E
D
E
E
N
(@
)
Pa
y a
16
17
1112
BALDUS - CROSS
NUMBER 2 OF THE QUESTIONNAIRE, COMPLETED BY ED GATES ON WARREN
MCCLESKEY IN THE SUPREME COURT.
@. WOULD THAT BE, WOULD THIS BE THE PROCEDURAL REFORM STUDY
QUESTIONNAIRE OR THE SUPREME COURT QUESTIONNAIRE?
A. WELL, THESE ARE THE, BOTH OF THESE QUESTIONNAIRES ARE WHAT
WE CALL THE PROCEDURAL REFORM STUDY QUESTIONNAIRE.
@. AND WOULD YOU LOOK AT RESPONDENT’S EXHIBIT NUMBER 3 AND
IDENTIFY THAT DOCUMENT. PLEASE?
A. THIS APPEARS. NUMBER 3, DEFENDANT’S EXHIBIT NUMBER 3.
APPEARS TO BE THE SUPPLEMENTAL QUESTIONNAIRE THAT WAS COMPLETED
FOR DEFENDANTS WHOSE RECORDS HAD NOT YET APPEARED IN THE
DEPARTMENT OF CORRECTIONS FILES.
@. AND AS TO BOTH OF THESE DOCUMENTS. WOULD THE ORIGINALS BE IN
YOUR POSSESSION?
A. I BELIEVE SO.
@. AND ONCE AGAIN, DO YOU RECALL IF THESE ARE COFIES OF
DOCUMENTS PROVIDED TO RESPONDENT IN THIS CASE?
A. THAT IS MY RECOLLECTION.
@. IF I COULD ASK YOU TO LOOK AT EXHIBIT NUMBER 1. AND EXHIBIT
INUMBER 2, AND TURN TO PAGE 9 OF BOTH OF THESE EXHIBITS. THIS
WOULD BE QUESTION NUMBER 30.
~~ DO YOU NOTE A DIFFERENCE IN THE CODING OF THESE TWO
QUESTIONNAIRES ON THIS PARTICULAR QUESTION?
A. YES. THE, THEY HAVE ONE FOIL CODED IN COMMON, THAT IS. THE
DEFENDANT PLANNED A CONTEMPORANECQUS OFFENSE FOR MORE THAN FIVE
UIUUIORIERUIGIGTMGFUTTL WUC
1113
BALDUS - CROSS
1 MINUTES. THAT”S NUMBER 23.
2 @. LET ME BACKTRACK JUST ONE MOMENT.
DO YOU HAPPEN TO KNOW IF THESE ARE CO-PERPETRATORS.
DAVID BURNEY AND WARREN MCCLESKEY?
A. 1 BELIEVE THART“S TRUE.
3
4
S
: . oi @. IF YOU WOULD COMPLETE YOUR ANSWER ON THE COMPARING OF THE
7 TWO QUEST IONS, PLEASE? |
8 A. CERTAINLY. THE CODERS CODED THE FOIL NUMBER 28 OR ENTERED
9 THE ANSWER TO NUMBER 28 IN THE FOILS, IN EACH QUESTIONNAIRE.
10 HOWEVER, KATHY CHRISTIAN, WHO CODED HER DOCUMENT IN THE PAROLE
11 BOARD FILES, CODED THAT THE CASE INVOLVED A CONTEMPORANEOUS
12 FELONY AND WAS UNNECESSARY TO COMPLETE THE CRIME. THAT WAS THE
13 EXTENT OF HER CODING.
14 WHEN ED GATES CODED HIS DOCUMENT IN THE SUPREME COURT
15 FILES vv
146 @. WHICH WOULD BE WHICH QUESTIONNAIRE?
27 A. THAT IS DB -- I’M SORRY -~ DEFENDANT’S EXHIBIT 2, HE DID NOT
138 CODE NUMBER 13 BUT INSTEAD HE CODED BLOODY, NUMBER 7. AND HE
Ww 19 CODED THAT THE DEFENDANT ACTIVELY RESISTED ARREST.
20 Q. IN LIGHT OF THE FACT THAT THESE TWO PERSONS WERE
22 CQO~-PERPETRATORS?
22 A. UH HUH.
23 @. DOES THIS REFLECT INCONSISTENCIES IN THE CODING?
24 A. THESE TWO QUESTIONNAIRES ARE NOT CODED CONSISTENTLY.
25 QR. THANK YOU.
1114
BALDUS - CROSS
ot
MS. WESTMORELAND: YOUR HONOR, I INTEND TO SUBMIT THESE
MN
DOCUMENTS AS EXHIBITS. BUT I WILL WAIT UNTIL THE PRESENTATION OF
THE RESPONDENTS CASE. IF THE COURT WOULD PREFER.
THE COURT: NQ.
MS. WESTMORELAND: IN THAT CASE. I WOULD LIKE TO SUBMIT
AT THIS TIME RESPONDENT“S EXHIBITS NUMBER 1, 2 AND 3.
MR. BOGER: NO OBJECTION, YOUR HONOR.
THE COURT: WHAT’S 3 FOR?
MS. WESTMORELAND: THREE. YOUR HONOR. IS MERELY THE
Fo
y O
W
M
e
R
y
C
D
COMPLETION OF THE QUESTIONNAIRE ON WARREN MCCLESKEY.
THE COURT: DOES IT SHOW SOMETHING IN ISSUE? fr
y
po
te
Te
y Mn
MS. WESTMORELAND: OTHER THAN HOW THE PARTICULAR CASE
13 IN THE INSTANT CASE WAS CODED. AND I THINK IT WOULD BE RELEVANT
14 |FOR THAT PURPOSE. YOUR HONOR.
15 THE COURT: ALL RIGHT.
16 BY MS. WESTMORELAND:
17 @. PROFESSOR BALDUS, IF WE COULD MOVE INTO THE CHARGING AND
18 SENTENCING STUDY?
w 19 A. CERTAINLY.
20 |@. WHEN YOU WERE FIRST CONTACTED ABOUT THIS STUDY, I BELIEVE
21 YOU STATED YOU WERE CONTACTED BY MRS. DOWLING, IS THAT CORRECT.
22 |A. YES. ANNE DOWLING, I BELIEVE HER NAME IS.
23 |@. DID YOU INDICATE AT THE TIME EITHER TO HER OR THE OTHER
24 PERSONS YOU TALKED WITH WHAT YOU THOUGHT THE OUTCOME OF THE
aw STUDY MIGHT BE?
CL ———— AT——— CUI, SNC, AGT. STS AE AL. AAAI IMD, A. ATPASE. SHAOGO UNSAID. SIA
Sah
n
BEL
B
E
NC
Ee
H
i
E
o
E
R
a
nN
I.
oO
Fo
rs
w
14
13
1115
BALDUS - CROSS
A. NO. AT THAT TIME, WHEN I SPOKE TO HER, I DIDN’T KNOW WHAT
THE OUTCOME MIGHT BE, BECAUSE I HAD, WE HAD NO DATA REALLY THAT
WOULD SUPPORT ANY PREDICTION. THAT WAS EARLY IN THE SPRING OR
EARLY WINTER OF 1980, TO THE BEST OF MY RECOLLECTION.
AND WE DIDN‘T DISCUSS ANYTHING EXCEPT WHAT THE
QUESTIONS MIGHT BE THAT WOULD BE APPROPRIATE FOR THEM TO SEEK
FUNDING FOR.
@. WHEN YOU ACCUMULATED INFORMATION FOR THE CHARGING AND
SENTENCING STUDY. YOU ADDED IN THE VOLUNTARY MANSLAUGHTER
CONVICTIONS, AS I RECALL?
A. YES.
@. BUT. FROM MY RECOLLECTION OF YOUR TESTIMONY YOU DID NOT HAVE
ANY INFORMATION ON ANYONE CONVICTED OF A LESSOR OFFENSE THAN
VOLUNTARY MANSLAUGHTER. DID YOU?
A. NO, WE DID NOT.
@. AND YOU DID NOT HAVE ANY INFORMATION ON ANY PERSONS WHO
MIGHT HAVE BEEN INDICTED BUT LATER ACQUITTED OF THOSE OFFENSES?
A. WE HAD NO SUCH INFORMATION.
@. IN THIS PARTICULAR STUDY. YOU ALSO DID A STRATIFIED SAMPLING
BASED ON JUDICIAL CIRCUIT. IS THAT CORRECT?
A. YES.
0. WHAT WAS YOUR REASONING FOR STRATIFYING BASED ON JUDICIAL
CIRCUIT IN THIS SAMPLE?
A. WE WANTED TO INSURE THAT WE HAD GOOD COVERAGE OF THE ENTIRE
STATE.
O
H
A
R
P
W
N
I
oO
| Tr
y
12
13
1116
BALDUS - CROSS
ON THE BASIS OF AN ARTICLE I READ ABOUT DEATH
SENTENCING IN GEORGIA, IT SUGGESTED THERE WAS A SUBSTANTIAL
AMOUNT OF VARIATION IN DEATH SENTENCING PRACTICES FROM PLACE TO
PLACE IN THE STATE, AND I WANTED TO INSURE GOOD COVERAGE. AND
ONE OF THE PRINCIPAL REASONS FOR THAT WAS TO ALLOW US TO MAKE
GEOGRAPHIC COMPARISONS BETWEEN DIFFERENT PARTS OF THE STATE. AND
|UNLESS WE STRATIFIED ON A GEOGRAPHIC BASIS. AS WE DID, BY
CHOOSING THE DISTRICTS AS THE FOCUS FOR THE
PURPOSES OF OUR STRATIFICATION I WAS CONCERNED THAT WE WOULD NOT
BE ABLE TO MAKE THESE SORTS OF GEOGRAPHIC COMPARISONS.
@. SO ONE OF THE MAIN PURPOSES FOR THIS STRATIFICATION WAS TO
BE ABLE TO MAKE GEOGRAPHIC COMPARISONS, THEN?
A. YES. AND ALSO TO BE SURE THAT OUR SAMPLE WAS REPRESENTATIVE
OF THE ENTIRE STATE. I MEAN RANDOM SAMPLING FROCEDURES ALWAYS
RUN THE RISK THAT AN AREA, JUST LUCK OF THE DRAW. IS GOING TO BE
OMITTED. AND WE DIDNT, WE THOUGHT IT IMPORTANT TO REDUCE THE
RISK OF THAT OCCURRING.
@. DOES THIS TYPE OF STRATIFIED SAMPLING BASED ON JUDICIAL
CIRCUIT COMPLY WITH THE GENERALLY ACCEPTED STATISTICAL
PROCEDURES FOR STRATIFIED SAMPLING?
A. THIS IS A QUESTION THAT“S BEYOND MY COMPETENCE. I WAS
ADVISED BY PROFESSOR WOODWORTH AND BY THE SAMPLING EXPERT THAT
WE CONSULTED AT THE UNIVERSITY OF IOWA THAT THIS WAS CONSISTENT
WITH GENERALLY ACCEPTED PRINCIPLES, GIVEN WHAT OUR PURPOSES
WERE. I MEAN. SAMPLING PLANS ARE SHAPED BY THE GOALS OF THE
1117
BALDUS - CROSS
1 SEARCHERS. AND OUR GOAL WAS TO BE ABLE TO MAKE COHERENT
2 STATEMENTS ABOUT THE ENTIRE STATE AND THAT’S WHAT ANIMATED THE
CONSTRUCTION OF THIS SAMPLING PLAN.
@. IN EXAMINING THE DATA IN THE CHARGING AND SENTENCING STUDY.
ONE OF THE QUESTIONS THAT IT LOOKED FOR WAS THE CREDIBILITY OF
3
4
5
& é& WITNESSES. AND I BELIEVE IT WAS STATED EARLIER, THIS
7 INFORMATION WOULD HAVE BEEN OBTAINED FROM EITHER THE POLICE
8 REPORTS OR THE PAROLE OFFICER REPORTS. IS THAT CORRECT?
4 A. WELL, ALL OF THE INFORMATION CONCERNING STRENGTH OF EVIDENCE
10 IN THE CHARGING AND SENTENCING STUDY WAS OBTAINED FROM THE
11 ~~ |PAROLE BOARD FILES. AND THOSE FILES REFLECTED THE WITNESS
12 REPORTS THAT WERE IN THE FILES OF THE DISTRICT ATTORNEY. SO
13 THATS WHERE THAT INFORMATION WOULD HAVE COME FROM, WAS IN THE
14 FILES IN THE PAROLE BOARD THAT WERE ABSTRACTS OF WHAT THE FILES
13 IN THE DISTRICT ATTORNEY’S OFFICE WERE.
146 @. SO YOU HAD NO ACTUAL INFORMATION THEN ON WHAT A PARTICULAR
17 PROSECUTOR MIGHT HAVE FELT ABOUT THE CREDIBILITY OF A PARTICULAR
18 WITNESS, THEN?
Ww 19 A. UNLESS IT WAS REFLECTED IN THE FILE.
20 6. AND THE POLICE OFFICER OR PAROLE OFFICER REPORT OR THE
21 PAROLE BOARD FILE?
22 A. WELL» MY UNDERSTANDING OF THE WAY THAT THE PAROLE BOARD
23 |COLLECTS INFORMATION, THEY, THEY PREPARE A FILE ON EACH OFFENDER
249 WHEN THEY COME IN THE SYSTEM. IN THE COURSE OF DOING SO. THEY
25 INTERVIEW THE DISTRICT ATTORNEY, THEY LOOK AT THE DISTRICT
1113
BALDUS - CROSS
| o
d ATTORNEY'S FILES. THEY TRY TO OBTAIN ALL THE INFORMATION THAT
2 WAS AVAILABLE TO THE DISTRICT ATTORNEY AND THEY COLLECT
3 INFORMATION FROM THE CDURT-S RECORDS THAT MAY EXIST IN THE CASE.
4 SO ANY INFORMATION WE HAD ABOUT THE CREDIBILITY OF A WITNESS. IN
3 A FILE, WOULD HAVE BEEN OBTAINED FROM THOSE SOURCES.
® & WE HAD NO INDEPENDENT INTERVIEWS WITH THE DISTRICT
7 ATTORNEYS IN OUR CODING. WE RELIED STRICTLY ON THE FILES OF THE
8 BOARD OF PARDONS AND PAROLES, WITH A FEW EXCEPTIONS WHERE WE
9 CONSULTED THE RECORDS IN THE SUPREME COURT, IF THERE WASN‘T
10 |A FILE IN THE GEORGIA SUPREME, PARDON ME, IN THE PARDONS AND
23 PAROLES BOARD.
12 @. IF IM NOT MISTAKEN. IT SEEMS THAT I RECALL THAT IT WAS
13 INDICATED EITHER DURING YOUR EARLIER TESTIMONY OR THE TESTIMONY
14 OF ED GATES THAT THIS DETERMINATION WAS ONLY MADE UNLESS, IF
15 THERE WAS A STATEMENT IN THE FILE SPECIFICALLY THAT THE WITNESS
16 WAS NOT CREDIBLE.
17 IS THAT ACCURATE?
18 A. AT THIS POINT I CAN‘T RECALL WHAT THE PRECISE CODING RULE
Ww 19 WAS. IT MAY HAVE BEEN THAT THERE HAD TO BE AN EXPLICIT RULE.
20 IF ED GATES TESTIFIED TO THAT EFFECT. THEN THAT’S THE CORRECT
23 ANSWER. I WAS JUST TESTIFYING AS TO WHAT THE SOURCE OF OUR
22 INFORMATION WOULD BE: THAT IS, THE FILES OF THE BOARD OF PARDONS
23 AND PAROLES.
24 @. AGAIN, CHARGING AND SENTENCING STUDY, DID YOU END UP WITH
25 |CASES IN WHICH IT WAS UNKNOWN AS TO WHETHER THERE WAS A PENALTY
pr
$v
©
~N
O0
6
OO
»
O
N
10
11
12
1119
BALDUS - CROSS
TRIAL OR NOT?
A. YES.
?. DO YOU HAPPEN TO KNOW HOW MANY?
A. AMONG THE CASES THAT RESULTED IN A MURDER CONVICTION, I
BELIEVE THERE WERE SOME 40 CASES. BUT AMONG THE CASES WHERE THE
INFORMATION WAS RELEVANT, THAT IS. THE CASES THAT WENT TO A
PENALTY TRIAL. PARDON ME. CASES THAT RESULTED IN A MURDER
CONVICTION AT TRIAL. AND WHETHER OR NOT THOSE CASES WERE
ADVANCED TO PENALTY TRIAL. THAT WAS THE UNIVERSE OF CASES THAT
WE WERE INTERESTED IN, THOSE CASES THAT HAD ENDED UP IN A MURDER
CONVICTION AT TRIAL. I CANNOT RECALL PRECISELY HOW MANY THOSE
ARE. THAT“S SOMETHING I COULD FIND OUT FOR YOU. I THINK THERE
WERE TWENTY. MAYBE THIRTY CASES.
THE COURT: TWENTY OR THIRTY WHAT?
THE WITNESS: CASES WHERE WE DIDN‘T KNOW THAT
INFORMATION, YOUR HONOR.
THE COURT: WELL, WHATS THE NUMBER 607
THE WITNESS: THAT’S ALL CASES INVOLVING A MURDER
CONVICTION, WHETHER BY PLEA OR AS THE RESULT OF CONVICTION AT
TRIAL. AND FOR THE PURPOSES OF OUR ANALYSIS WE WERE FOCUSING
STRICTLY ON THE CASES THAT HAD RESULTED IN A CONVICTION AT
TRIAL.
SO THERE WERE, I THINK, SOME 40 CASES OF ALL TYPES OF
MURDER CONVICTIONS, BUT OUR VARIABLE THAT WE USING --
THE COURT: SIXTY CASES OF ALL TYPES OF MURDER
1120
BALDUS - CROSS
1 CONVICTIONS WHICH WHAT?
hr
THE WITNESS: IN WHICH WE DIDN’T KNOW IF THERE WAS A
3 PENALTY TRIAL. YOUR HONCR.
4 BUT THE POPULATION OF CASES THAT WE’RE, WHERE THIS
a CREATED ANY PROBLEM WAS IN THE CATEGORY WHERE THERE HAD BEEN A
CONVICTION AT TRIAL. THAT WAS THE DECISION WE WERE TRYING TO
MODEL, THE PROSECUTORIAL DECISIONS TO SEEK A DEATH SENTENCE
AFTER THERE HAD BEEN A CONVICTION OBTAINED AT TRIAL.
ww
0
9
BY MS. WESTMORELAND:
10 R. RATHER THAN THE PROSECUTORIAL DECISION TO SEEK A DEATH
11 SENTENCE AFTER THE ENTRY OF A GUILTY PLEA?
12 A. THAT’S RIGHT. THAT WAS THE SAME TACT WE HAD TAKEN IN THE
13 PROCEDURAL REFORM STUDY, BECAUSE THE REASON WAS THAT THE. THE
14 VERY FEW. PARDON ME. VERY FEW MURDER PLEAS OR CASES ARE ADVANCED
135 TO A PENALTY TRIAL. I MEAN, THERE AREN‘T ENOUGH DECISIONS MADE
16 IN THAT CONTEXT OF CASES TO BE ABLE TO SAY ANYTHING
17 STATISTICALLY ABOUT THAT DECISION.
13 BUT, AMONG CASES WHERE THERE WAS A CONVICTION AT TRIAL,
& 19 A THIRD OF THOSE CASES GO FORWARD TO A» GO FORWARD TO A PENALTY
20 TRIAL. S50 THAT'S WHAT WE FOCUSED ON.
23 THE COURT: LET ME MAKE SURE I UNDERSTAND YOU.
22 THE WITNESS: YES.
23 THE COURT: HOW BIG WAS YOUR SAMPLE?
24 THE WITNESS: IT WAS 10460, APPROXIMATELY.
23 THE COURT: OUT OF 2400 YOU HAD 1060 SAMPLES?
fe
y
of
)
C
0
a
o
fe
O
R
P
I
N
10
11
1121
BALDUS - CROSS
THE WITNESS: YES, YOUR HONOR.
THE COURT: STRATIFIED?
THE WITNESS: YES.
THE COURT: WEIGHTED OR UNWEIGHTED?
THE WITNESS: UNWEIGHTED, THAT’S -— THOSE ARE THE RAW
COUNTS, YOUR HONOR, OF THE NUMBER OF CASES WHOSE FILES WE
EXAMINED. WE EXAMINED A THOUSAND.
THE COURT: IN THE SECOND STUDY?
THE WITNESS: YES. ACTUALLY I THINK IT WAS A 1082.
SOMETHING OF THAT ORDER.
THE COURT: AND TWENTY OF THOSE YOU DIDN‘T KNOW WHETHER
THERE WAS PENALTY TRIAL, TWENTY OF THOSE WHERE A MURDER
CONVICTION OCCURRED?
THE WITNESS: THAT IS MY RECOLLECTION. I DO NOT HAVE
THAT BIT OF INFORMATION RIGHT ON MY FINGERTIPS, BUT I CAN GET IT
VERY EASILY.
BY MS. WESTMORELAND:
Q@. YOU INDICATED JUST A MINUTE AGO YOU EXAMINED THE FILE IN
APPROXIMATELY, I GUESS, 1084 CASES?
R. DID YOU NOT EXAMINE THE FILES IN THE REMAINDER OF THE CASES
IN THE UNIVERSE OF 2400 IN ANY FASHION?
A. YES, WE DID EXAMINE THEM. WE EXAMINED THE CASES THAT HAD
INVOLVED MURDER CONVICTIONS TQ DETERMINE WHETHER OR NOT THEY
WERE CASES THAT ALSO HAD ADVANCED TO A PENALTY TRIAL, BECAUSE IF
1122
BALDUS - CROSS
po
te
THEY WERE CASES THAT HAD ADVANCED TO A PENALTY TRIAL. WE WANTED
2 |To PICK THEM UP AND INCLUDE THEM IN OUR STUDY, EVEN THOUGH THEY
3 |HAD NOT BEEN PICKED IN THE FORMAL SAMPLING PROCEDURE.
4 |@. DID YOU CONSIDER IT IMPORTANT TO OBTAIN A GOOD MEASURE OF
5 |THE STRENGTH OF THE EVIDENCE IN EACH CASE?
2 & |A. YES.
7 |@. AND WAS THIS INFORMATION ALSO IN THE CHARGING AND
8 |SENTENCING STUDY, OBTAINED FROM THE PARDONS AND PAROLES BOARD
9 |FILES? |
10 IA. ves.
11 |@. AND DID YOU, BY USING THESE FILES. MAKE ASSUMPTIONS AS TO
12 |WHAT MIGHT BE AVAILABLE TO PROSECUTORS?
13 |A. WE MADE THE ASSUMPTION THAT THE INFORMATION THAT WAS PRESENT
14 IN THESE FILES WAS INFORMATION THAT WAS KNOWN TO THE PROSECUTOR,
15 YES.
146 @. DID YOU ALSO ASSUME THAT IT WAS INFORMATION THAT WAS KNOWN
17 TC THE JURY?
18 A. IN CASES THAT ADVANCED TO A JURY TRIAL. YES.
@ 19 @. AND BY THE SAME TOKEN, DID YOU ASSUME THAT IT WAS ALSO KNOWN
20 TO THE SENTENCING JURY AS WELL?
21 A. YES.
22 @. DID YOU EVER MAKE ANY ATTEMPT TO BREAK DOWN THE EVIDENCE AS
23 IT MIGHT BE KNOWN TO A PARTICULAR DECISION MAKER AT THE TIME OF
24 THE DECISION?
235 : MR. BOGER: YOUR HONOR, I“LL OBJECT TO THAT QUESTION.
B
a
S
Y
RH
EE
PR
E
|
SE
F
E
b
a
R
R
A
a
a
MN
poh
oO
[e
ry
W
14
15
1123
BALDUS —- CROSS
AS UNCLEAR.
THE COURT: I UNDERSTAND IT.
MS. WESTMORELAND: ILL REPHRASE IT.
THE COURT: REPHRASE IT.
MS. WESTMORELAND: I-LL REPHRASE IT IF THAT WOULD HELP,
YOUR HONOR.
BY MS. WESTMORELAND:
@. IN CONSIDERING THE STRENGTH OF EVIDENCE THAT YOU, THE
INFORMATION THAT YOU OBTAINED, AND IN DOING YOUR SUBSEQUENT
ANALYSES, DID YOU EVER CONSIDER THE EVIDENCE AS IT WAS KNOWN TO
THE PARTICULAR PERSON MAKING THE DECISION, BEING THE JURY,
JUDGE, PROSECUTDR. AT THE TIME THAT DECISION WAS MADE?
A. WELL, THERE ARE TWO POSSIBLE ANSWERS TO THAT QUESTIOM. IM
NOT SURE WHICH YOURE ADDRESSING.
IF YOURE ASKING WHETHER WE INTERVIEWED PROSECUTORS,
WHEN THEY WERE MAKING THEIR DECISIONS IN EACH CASE. THE ANSWER
IS NO, WE DID NOT INTERVIEW PROSECUTORS AND WE DID NOT
| INTERVIEW JURORS. WE JUST RELIED ON THE INFORMATION THAT WAS
AVAILABLE IN THE PAROLE BOARD FILES.
HOWEVER. WE DID EXTENSIVE ANALYSIS OF EACH DECISION
POINT IN THE PROCESS TO SEE THE EXTENT TO WHICH OUR MEASURES OF
THE STRENGTH OF EVIDENCE HELPED EXPLAIN THOSE DECISIONS AND THEY
HAD A VERY SUBSTANTIAL STATISTICAL EFFECT ON THOSE SENTENCING --
PARDON ME —— ON THOSE DECISIONS AT EACH STAGE. THEY AFFECTED.
THEY HELPED EXPLAIN STATISTICALLY THE INDICTMENT DECISION, PLEA
BALDUS - CROSS
1 BARGAINING DECISIONS, MURDER CONVICTIONS AT TRIAL. THE EVIDENCE
2 IS OVERWHELMING I BELIEVE FROM OUR STUDY THAT THE STRENGTH OF
THE EVIDENCE IN A CASE AFFECTS THE OUTCOME OF THE PROCESS AT
EACH STAGE ALONG THE WAY, UP TO AND INCLUDING THE DEATH
SENTENCING DECISION AT PENALTY TRIAL.
3
a
=
® 5 SO TO THE EXTENT THAT. THAT”S. THATS THE KIND OF
7 |ANALYSIS WE DID ABOUT THE EFFECT OF THOSE ON THE PEOPLE THAT
& |WERE MAKING THE DECISIONS AT THE TIME. BUT WE DIDN‘T INTERVIEW
9 |THE PEOPLE THAT WERE MAKING THE DECISIONS AT THE TIME THEY MADE
10 |THEM, NO.
£1 |@. IN DOING ANY OF YOUR STUDIES, DID YOU EVER INTERVIEW OR TALK
12 |WITH ANY PROSECUTORS IN THE STATE?
13 |A. NO» NO» NOT ON A SUBSTANTIVE MATTER, NO. I DID NOT.
14 |@. IN THE CHARGING AND SENTENCING STUDY, IN CONSIDERING THE
5 |CO-PERPETRATOR CASES. 1 BELIEVE YOU INDICATED THERE WAS AN
16 |ATTEMPT TO HAVE THE SAME CODER DO ALL CO-PERPETRATORS, IS THAT
17 |CORRECT?
12 |A. YES. THAT WAS THE POLICY THAT WE DEVELOPED IN GEORGIA, IN
@ 19 |THE CHARGING AND SENTENCING STUDY. RIGHT.
20 |@. WHAT WAS THE PURPOSE FOR THAT POLICY?
21 |A. THE PURPOSE. THE PRINCIPAL PURPOSE OF IT WAS TO, AS I RECALL
22 |1T. AT LEAST FROM MY PERSPECTIVE, THE PRINCIPAL PURPOSE OF IT
23 |WAS TO MAKE THE DATA COLLECTION PROCESS MORE EFFICIENT. THAT
24 |1S. IF SOMEONE KNEW SOMETHING ABOUT A CASE, THEY WOULD BE IN A
23 BETTER POSITION TO CODE ANOTHER CASE THAT WAS FACTUALLY RELATED
20
BALDUS - CRO3S
TO IT.
IT TAKES A LONG TIME TO GET A HANDLE ON THE CASE.
ESFECIALLY A COMPLICATED ONE INVOLVING CO-PERPETRATOR CASES.
THEY ARE THE MOST DIFFICULT CASES TO CODE.
SO WHEN IT WAS SUGGESTED THAT WE OUGHT TO SORT THESE
CO~-PERPETRATOR CASES S50 ONE PERSON COULD DO THEM. THAT SEEMED TO
ME AN EFFICIENT WAY OF PROMOTING THE PROCESS.
ALSO IT WOULD ALLOW US TO IDENTIFY ISSUES OF
AMBIGUITIES WITH RESPECT TO ALL THE FILES. IF THEY WERE ALL DONE
BY THE SAME PERSON. IF ONE PERSON DOING ONE FILE WOULD LOOK AT
ANOTHER FILE, THEY COULD SEE AMBIGUITIES IN THE FILE AND WOULD
KNOW MORE ABOUT THE FILE THAN SOMEONE WHO DIDNT HAVE KNOWLEDGE
OF THE OTHER CASES.
2. WERE THE CODERS ALLOWED TO USE .INFORMATION FROM ONE
CO-PERFETRATOR’S FILE TO CODE A QUESTIONNAIRE IN THE OTHER
CO-PERPETRATOR’S CASE?
A. 1 DON’T KNOW THE. I CANNOT RECALL THE ANSWER TO THAT
QUESTION. IT WAS MY GOAL THAT THE, THAT EACH FILE BE CODED ON
ITS OWN MERITS. THAT IS, WHAT WAS IN THE FILES. IS WHAT SHOULD
GO INTO THE GUESTIONNAIRE.
BUT IM SURE THAT WHEN AMBIGUITIES AROSE. THEY WOULD
CROSS CHECK THE FILES, IT WAS INEVITABLE THEY WOULD TRY AND DO
THAT. AND TRY TO COME UP WITH SOME JUDGMENT AS TO WHAT PEER --
THE INFORMATION THAT WAS AVAILABLE IN EACH FILE. 350 I DON'T, I
DON-T RECALL SPECIFICALLY THE ANSWER TO THAT QUESTION, BUT IT
BALDUS - CROSS
SEEMS TO ME GIVEN MY KNOWLEDGE OF THE GENERAL PRACTICES THAT THE
CODERS USED, THEY DID NO DOUBT, CONSULT. YOU KNOW, DIFFERENT
FILES TO RESOLVE AMBIGUITIES.
@. IN LOOKING AT THE PARTICULAR FACTORS THAT WERE CONSIDERED IN
THE CHARGING AND SENTENCING STUDY, I BELIEVE ONE OF THE
MITIGATING CIRCUMSTANCES YOU LISTED WAS THE EFFECTS OF ALCOHOL
ON THE DEFENDANT. WAS THAT CORRECT?
A. 1 DON’T RECALL THAT PRECISELY, BUT IT COULD VERY WELL BE
THAT I CHARACTERIZED THE USE OF ALCOHOL BY DEFENDANT AS A
MITIGATING FACTOR AND THE REASON FOR THAT WAS, IT IS IN MANY
CASES, APPELLATE. SUPREME COURT CASES. CONSIDERED DEATH
WORTHINESS. I PERCEIVED IT TO BE A MITIGATING FACTOR. AND ALSO
QUR STATISTICAL ANALYSES IN THIS STUDY AND IN OUR EARLIER
EMPIRICAL STUDIES OF DEATH SENTENCING INDICATED IT HAD A
MITIGATING EFFECT BY AND LARGE.
ALTHOUGH I WILL SAY IT WAS NOT CONSISTENT. SOMETIMES
IN CERTAIN SUB-GROUPS OF CASES. BEING A DRUNK DEFENDANT DOES NOT
AUALIFY AS MITIGATING. IT DEPENDS ON OTHER CIRCUMSTANCES. BUT
I WOULD SAY ON THE WHOLE. DRUNKENNESS ON THE PART OF A DEFENDANT
REFLECTING LACK OF JUDGMENT TENDS TO BE A, HAVE A MITIGATING
EFFECT.
THE COURT: ALL RIGHT. M3. WESTMORELAND. LET”S BREAK
FOR LUNCH AT THIS TIME.
WE“LL BE IN RECESS UNTIL QUARTER DF TWO.
Se
rr
a L
N
O
w
1127
BALDUS - CROSS
(COURT RECESSED FOR LUNCH.)
THE COURT: ALL RIGHT, MS. WESTMORELAND.
DAVID C. BALDUS.,
BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
TESTIFIED FURTHER AS FOLLOWS:
CROSS-EXAMINATION (CONT D)
BY MS. WESTMORELAND:
@. PROFESSOR BALDUS, I BELIEVE WHEN WE STOPPED FOR LUNCH WE
WERE DISCUSSING A VARIABLE CONCERNING THE EFFECTS OF ALCOHOL ON
A DEFENDANT. AND THE CHARACTERIZATION OF THAT PARTICULAR
VARIABLE.
IN THE CHARGING AND SENTENCING STUDY, I BELIEVE YOU
ALSO HAVE A VARIABLE CONSIDERING. OR QUESTIONNAIRE ITEM,
CONSIDERING THE EFFECTS OF ALCOHOL ON THE VICTIM. IS THAT
CORRECT?
A. YES.
QR. AND HOW IS THAT CHARACTERIZED? AS AGGRAVATING OR
MITIGATING?
A. THE ONLY PLACE THAT I WOULD CHARACTERIZE IT AS AGGRAVATING
NR MITIGATING WOULD HAVE BEEN IN SCHEDULE 1 OF TECHNICAL
APPENDIX, AND I DON-T, ON LOOKING OVER SCHEDULE 1, WHICH
EMBRACES THE. THE MITIGATING VARIABLES THAT WE SO CHARACTERIZED,
I DON’T SEE IT ANYWHERE HERE, ALTHOUGH IT MAY BE IN THIS LIST.
IT IS NOT APPARENT TO ME.
5
(Q)
BALDUS - CROSS
R. IS, IM SORRY GO AHEAD?
A. IM SORRY. THE EVIDENCE ON THE AMOUNT OF EVIDENCE WE HAVE
ON VICTIM DRINKING VARIABLE IS LESS CLEAR THAN IS THE ONE ABOUT
THE DEFENDANT BEING INTOXICATED.
@. IS IT A VARIABLE THAT COULD PERHAPS GO EITHER WAY, BE
AGGRAVATING OR MITIGATING?
A. THAT WOULD NOT SURPRISE ME AT ALL THAT IT COULD POSSIBLY GO
ONE WAY OR ANOTHER, DEPENDING ON THE CIRCUMSTANCES.
AND THE LAW ON THE MATTER IS VERY VAGUE.
@. WHEN YOU RECODED THE VARIABLES IN THE CHARGING AND
SENTENCING STUDY IN ORDER TO GET THEM READY FOR COMPUTER
ANALYSIS, IF I UNDERSTAND CORRECTLY. I BELIEVE YOU RECODED THE 1
AND THE 2 VALUES FROM THE QUESTIONNAIRE AS IF THEY BOTH EXISTED
IN THE FILE, IS THAT CORRECT?
A. YES. CORRECT.
@. AND IN THAT INSTANCE. THE "2", I BELIEVE. WOULD REPRESENT
"SUGGESTED BY THE FILE BUT NOT EXPRESSLY STATED" OR LANGUAGE TO
THAT EFFECT. IS THAT CORRECT?
A. YES, THATS RIGHT, BECAUSE MANY TIMES THE FILE WILL NOT
STATE EXPLICIT SOMETHING OCCURRED OR DID NOT OCCUR, BUT RATHER
THE CODER MUST EXERCISE AN INFERENCE FROM ALL THE FACTS IN THE
RECORD.
RA. IM GOING TO HAND YOU WHAT S BEEN MARKED AS RESPONDENT “3
EXHIBIT NUMBER 4, AND ASK IF YOU CAN IDENTIFY THAT DOCUMENT FOR
ME. PLEASE?
[
oi
(0
BALDUS - CROSS
A. YES. THIS APPEARS TO BE THE RUESTIONNAIRE CODED FROM WARREN
MCCLESKEY’S CASE IN THE CHARGING AND SENTENCING 3TUDY.
MS. WESTMORELAND: I HAVE A COPY FOR THE COURT AS WELL.
THE COURT: DELORES?
BY MS. WESTMORELAND:
@. WOULD YOU REFER BACK TO PAGE. AND I DON’T BELIEVE YOU HAVE
RESPONDENTS EXHIBIT NUMBER 2, WOULD You REFER TO PAGE 9 ON
RESPONDENTS EXHIBIT NUMBER 2, PLEASE, WHICH WAS THE PROCEDURAL
REFORM STUDY QUESTIONNAIRE?
A. J’VE (OT 17.
@. AND LOOKING AT THE CODING ON THAT PAGE. IS MY INTERPRETATION
CORRECT IN THAT IT IS CODED IN THAT CASE THAT THE DEFENDANT
ACTIVELY RESISTED ARREST?
THE COURT: ON PAGE 27
MS. WESTMORELAND: I HAVE RESPONDENTS EXHIBIT Z, YOUR
HONOR.
THE COURT: OH.
THE WITNESS: THAT'S RIGHT.
BY MS. WESTMORELAND:
R. THEN, ON PAGE. REFERRING TO RESPONDENTS EXHIBIT NUMBER 4 ON
PAGE 1&4, WHICH WOULD BE PART C OF QUESTION 47, IS IT CORRECT TO
ISTATE THAT TO BE ACTIVELY RESISTING OR AVOIDING ARREST IS NOT
CODED IN THIS PARTICULAR GQUESTIONNAIRE?Y
A. THAT 5S CORRECT.
2. AND DO THESE TWD QUESTIONNAIRES REFER TO THE SAME DEFENDANTY
-
hy
BALDUS - CROSS
A. THEY DQ.
@. THANK YOU.
MS. WESTMORELAND: YOUR HONOR, I WOULD LIKE TO MOVE THE
ADMISSION OF RESPONDENTS EXHIBIT NUMBER 4 INTO EVIDENCE. THAT
BEING THE QUESTIONNAIRE RELATING TO THE PETITIONER IN THE
INSTANT CASE.
MR. BOGER: NO OBJECTION. YOQUR HONOR.
THE COURT: IT WILL BE ADMITTED.
BY MS. WESTMORELAND:
#2. PROFESSOR BALDUS, IN YOUR UPDATING OF THE CHARGING AND
SENTENCING STUDY INFORMATION. HAVE YOU ADDED A VARIABLE IN THE
STUDY OR IN YOUR ANALYSIS TO ACCOUNT FOR MISSING ITEMS?
A. THIS IS IN THE CHARGING AND SENTENCING?
2. YES, UH HUH?
A. WELL, IN EACH VARIABLE IN THE CHARGING AND SENTENCING STUDY,
THERE“S A PROVISION FOR CODING FACTORS "UNKNOWN.
@. HAVE YOU ADDED. IS THERE A VARIABLE IN THE. I BELIEVE IT"S
THE "2" TEST FILE? WOULD THAT BE THE CHARGING AND SENTENCING
FILE?
A. YES.
RA. IS THERE A VARIABLE IN THAT FILE TO COUNT THE NUMBER OF
MISSING OR UNKNOWN ITEMS?
A. YES.
R. DO YOU HAPPEN TO KNOW WHAT THE COUNT IS FOR THAT ITEM IN THE
CHARGING AND SENTENCING STUDY, FOR THE TOTAL NUMBER OF UNKNOWNS?
1131
BALDUS - CROSS
A. THE TOTAL NUMBER OF UNKNOWNS?
CG. YE3?
A. FOR ALL CASES?
8. YES?
A. NO, I DON‘T. I KNOW WHAT THE AVERAGE IS.
2. WHAT WOULD BE THE AVERAGE?
A. THE AVERAGE IS ABOUT 33 UNKNOWN OBSERVATIONS ACROSS ALL oF
THE FILES. OUT OF FIVE HUNDRED VARIABLES IN EACH ONE, THERE'S
AN AVERAGE OF ABOUT 33 WHERE "UNKNOWN" WAS CODED.
2. AN AVERAGE OF ABOUT 33 PER CASE?
A. YES.
3. PER FLLE?
A. RIGHT.
@. IN CONSIDERING THE STATISTICAL METHODS AND PROCEDURES THAT
WERE UTILIZED. AND IN PARTICULAR, THE REGRESSIONS. WHO
DETERMINED WHICH REGRESSION PROCEDURES TO RUN? DID. I'M
ASSUMING YOU DID MAKE THAT DETERMINATION, OR DID SOMEONE ELSE?
A. IF, IF BY THAT YOU MEAN. WELL. LET ME TELL YQU PRECISELY HOW
|THE DECISIONS WERE MADE.
PROFESSOR WOODWORTH ADVISED ME THAT WE SHOULD USE
WEIGHTED LEAST SGUARES AND LOGISTIC ANALYSIS IN THE CHARGING AND
SENTENCING STUDY. HE ADVISED ME THAT WE SHOULD USE ORDINARY
LEAST SRUARES IN THE PROCEDURAL REFORM STUDY AND ALSO LOGISTIC
REGRESSION IN THAT STUDY.
ON THE BASIS OF THAT GUIDANCE. THEN I WOULD SPECIFY THE
pot
e
c
E
T
l
DE
RE
f
e
D
E
10
11
1132
BALDUS - CROSS
COMBINATION OF VARIABLES THAT WOULD BE INCLUDED IN THE ANALYSES.
@. AND WHO DETERMINED WHICH REGRESSIONS WOULD BE UTILIZED FOR
YOUR REFORT. YOUR LATER WORK?
A. I MADE THE INITIAL. THAT INITIAL DETERMINATION AND WOULD
DRAFT A REPORT AND I WOULD PRESENT IT TO PROFESSOR WOODWORTH TO
REVIEW. AND ON THE BASIS OF HIS TECHNICAL EVALUATION OF THE RUNS
THAT WERE DONE, WE WOULD SOMETIMES MAKE MODIFICATIONS. BUT
PRINCIPALLY. THE PRIOR DETERMINATIONS THAT THESE WERE THE
AFPROPRIATE METHODS WOULD STAND AND THOSE WOULD BE THE
REGRESSIONS THAT WOULD GO INTO THE REPORT.
@. IN DOING YOUR ANALYSES. YOU INITIALLY. I BELIEVE, CONTROLLED
FOR THE FELONY MURDER CIRCUMSTANCES, I BELIEVE ONE OF THE WAYS
YOU“VE REFERRED TQ IT. IS THAT CORRECT?
A. YES. IF THERE WERE FELONY CIRCUMSTANCE IN THE CASE, THAT'S
CORRECT.
@. THAT’S ONE OF THE PRELIMINARY ANALYSES THAT YOU DID?
A. RIGHT.
3B, ISNT IT IRUE THAT THAT CIRCUMSTANCE AT LEAST EXPLAINED A
PORTION OF THE DISPARITY THAT EXISTED?
A. YES» IT DiD.
@. DID YOU ALSO FIND THAT THE STRANGER VICTIM EXPLAINED A
PORTION OF THE DISPARITY?
A. YES, A LITTLE BIT MORE. IT EXPLAINED DURESS.
@. AND THEN ALSO THE PROVISION FOR FAMILY-LOVER-LIGUOR-OR BAR
ROOM GQUARREL., DID THAT ALSO EXPLAIN A PORTION OF THE DISPARITY
Fed
SL
1133
BALDUS - CROSS
AS WELL?
A. YES. A LITTLE BIT.
@. YOU ALSO MADE A BREAKDOWN OR MADE AN ANALYSIS CONTROLLING
FOR PRIOR RECORD, IS THAT CORRECT?
A. YES, THAT'S CORRECT.
R. DID YOUR ANALYSIS IN THIS REGARD EVER TAKE INTO
CONSIDERATION WHETHER THAT PRIOR RECORD WAS KNOWN TO THE FERSON
MAKING THE DECISION?
A. WELL. AS I INDICATED TO YOU BEFORE. THE INFORMATION THAT WE
OBTAINED WITH RESPECT TO THE PRIOR RECORD WAS INFORMATION THAT
WAS IN THE FILE OF THE PAROLE BOARD. THAT IS, IT WOULD BE DRAWN
FROM THE F.B.I. RAP SHEET. THAT WAS THE PRINCIPAL SOURCE OF THE
INFORMATION. AND WE PROCEEDED ON THE ASSUMPTION THAT THE F.B.I.
RAP SHEETS WERE THE TYPE OF INFORMATION THAT WAS GENERALLY
AVAILABLE TO THE PROSECUTORS.
@. DID YOU ALSO THEN PRESUME IT WAS PRESENTED TO THE JURY AS
WELL?
A. IF THE CASE REACHED A PENALTY TRIAL AND THAT WOULD BE THE
ONLY CONTEXT IN WHICH WE WERE FOCUSING HARD ON JURY DECISION
MAKING WAS AT THE PENALTY TRIAL STAGE. THERE, IT"S MY
UNDERSTANDING OF THE LAW THAT PRIOR RECORD IS A NON-AGGRAVATING.
PARDON ME. A NON-STATUTORY AGGRAVATING FACTOR OF THE TYPE THAT
IS PRESENTED ROUTINELY TO JURIES IN THIS STATE.
@. YOU DON’T KNOW WHETHER THAT INFORMATION WAS PRESENTED, DO
You?
SD ern Tse ren mam arta arma ie ao a)
rr ep — Fry SE — et | NTN RE, or er, ope, at pe Fray, | pm, Nr | TTS Teeny | ge—— pay
Mn
Ww
BALDUS - CROSS
A. NO, NOT IN ANY ONE CASE. I JUST KNOW THE INCENTIVES THAT
ARE IN THE SYSTEM TO INTRODUCE THAT SORT OF EVIDENCE AT THE
TRIAL. THATS ALL.
RR. WHEN YOU DID YOUR CROSS TABULATIONS, I BELIEVE DB-467 WAS A
PORTION OF THAT CROSS TABULATION. IF YOU-D LIKE TO REFER TQ IT.
IN DOING -- GO AHEAD. I-31 SORRY.
A. I HAVE IT NOW, MS. WESTMORELAND.
@. UNDER THE CIRCUMSTANCES OF THE PRESENT STUDY THAT YQU’RE
DOING, ISN‘T IT BASICALLY TRUE THAT CROSS TABULATIONS OF THIS
TYPE LOSE THE EFFECTIVENESS DUE TO THE FACT THAT YOUR SAMPLE
S1ZE DECREASES RATHER RAFIDLY?
A. YES. THE, WELL, COULD YOU RESTATE THE QUESTION AGAIN, MS.
WESTMORELAND
@. YES, LET ME REPHRASE IT AND PERHAPS THAT LL HELP.
IN UTILIZING CROSS TABULATIONS IN THE TYPE OF STUDY
THAT YOURE DOING. DOES NOT THE SAMPLE SIZE DECREASE RATHER
RAPIDLY, THUS LIMITING THE USEFULNESS OF THE CROSS TABULATION IN
THIS TYPE OF STunyY?
A. YES, IT LIMITS IT.
@. ONE OF THE THINGS YOU“VE DISCUSSED WAS YOUR LIBERATION
HYPOTHESIS WHICH WOULD REFER TO THE MID- RANGE CASES, I WOULD
SUPPOSE, THE ONES THAT WERE NOT NECESSARILY THE MOST AGGRAVATING
OR MOST MITIGATING CASES IN EXISTENCE.
AND I BELIEVE YQU SAID THE LIBERATION HYPOTHESIS CAME
INTO PLAY IN THESE MID-RANGE CASES, IS THAT CORRECT?
A
BALDUS - CROSS
A. YES, THAT’S WHERE YOU SEE THE EFFECTS OF RACIAL FACTORS AND
OTHER FACTORS AS WELL. THESE FACTORS THAT INFLUENCE THINGS. ANY
ONE INDIVIDUAL THING. HAS ITS MOST PROMINENT EFFECT IN THAT
MID-RANGE.
@. S0 CERTAIN LESS OBVIOUS FACTORS WOULD HAVE THE MOST
INFLUENCE IN THE MID-RANGE CASES THAT WERE NOT THE MOST
AGGRAVATED OR MITIGATED?
A. I THINK SO.
@. SO FACTORS THAT MIGHT ONLY APPEAR IN THIS ONE CASE MIGHT
ACTUALLY HAVE AN EFFECT IN THAT CASE AND NEVER SHOW UP IN OTHER
CASES, WOULDN’T THAT BE CORRECT?
A. THEY MIGHT.
MS. WESTMORELAND: PARDON ME, YOUR HONOR, WHILE I FIND
AN EXHIBIT.
BY MS. WESTMORELAND:
R. WOULD YOU REFER TO DB-84, PROFESSOR BALDUS?
A. I HAVE IT.
@. DOES THIS PARTICULAR TABLE, WHICH WAS TABLE 33, I BELIEVE,
OUT OF YOUR REPORT. DOES THIS PARTICULAR TABLE REFER TO ANY
PARTICULAR DECISION MAKER IN THE SYSTEM, OR THE OVERALL SYSTEM?
A. NO, THIS FOCUSES ON THE FROSECUTORIAL DECISION TO REQUEST A
CHARGE AND OBTAIN A CHARGE, REGUEST AND OBTAIM. ACTUALLY IT
REPRESENTS THE JOINT DECISIONS OF THE PROSECUTOR AND THE JUDGE
AT THE PENALTY TRIAL. THAT'S COLUMN C.
THEN COLUMN D REFLECTS THE ACTIONS OF THE DEATH
NH
6
C
p
“0
BALDUS - CROSS
SENTENCING JURIES AT THOSE PENALTIES TRIALS.
@. SO DO YOU HAVE INFORMATION ON INSTANCES WHERE A CHARGE ON A
PARTICULAR AGGRAVATING CIRCUMSTANCE WAS REQUESTED BUT NOT GIVEN?
A. NO.
@. S0 THIS WOULD JUST MERELY REFLECT THOSE AGGRAVATING
CIRCUMSTANCES THAT WERE CHARGED BY THE COURT?
|A. THAT’S RIGHT.
@. AND WOULD THAT INFORMATION HAVE BEEN OBTAINED FROM THE TRIAL
TRANSCRIPT OR OTHER SOURCES?
A. NO, THE SOURCE OF THIS IS ALWAYS THE GEORGIA SUPREME COURT
|REPORT. WHICH SYSTEMATICALLY REPORTS IN EVERY DEATH CASE WHAT
WAS CHARGED AND WHAT WAS FOUND.
@. SO YOU RELIED EXCLUSIVELY ON THE OPINIONS OF THE SUPREME
COURT IN DETERMINING THIS INFORMATION?
A. I DON’T KNOW THAT IT WAS EXCLUSIVELY THAT. SOMETIMES THE.
SOMETIMES WE WOULD CONSULT THE. THE TRANSCRIPT IN THE GEORGIA
SUPREME COURT. BUT OUR ULTIMATE CHECK IN EACH CASE. IF I
REMEMBER CORRECTLY. WAS THE GEORGIA SUPREME COURT REPORT.
@. IF YOu-‘D REFER TO DB-90, PROFESSOR BALDUS?
A. UH HUH. YES, I HAVE IT.
§. 1 BELIEVE YOU NOTE OR YOU MAY HAVE NOTED ON DIRECT
EXAMINATION, IM NOT CERTAIN, THAT THE OVERALL RACE OF VICTIM
DISPARITY WAS STATISTICALLY SIGNIFICANT AT THE .01 LEVEL. DO
YOU RECALL THAT?
A. YES.
prt
ed
C
R
E
E
E
R
TE
ER
«
E
E
BE
R
l
E
L
TE
Fy
Oo
11
14
BALDUS - CROSS
R. IN MAKING THAT STATEMENT, DID THAT STATEMENT INDICATE THAT
YOUU DID NOT HAVE THE BREAKDOWNS FOR THE INDIVIDUAL CATEGORIES
THAT ARE LISTED IN THIS TABLE?
A. NO, I DID NOT CALCULATE THE STATISTICAL SIGNIFICANCE OF
DISPARITIES AT EACH LEVEL OF THIS TABLE. I JUST ESTIMATED THE
OVERALL EFFECT WITHIN THIS SUB-GROUP OF CASES. o
@. WOULD IT HAVE BEEN DIFFICULT TO DO SO BECAUSE OF THE SAMPLE
SIZES?
A. NO, IT WOULDN’T HAVE BEEN DIFFICULT TO MAKE THE CALCULATION,
BUT I THOUGHT THAT THE OVERALL IMPACT TOLD THE TALE OF WHAT THE
OVERALL EFFECT WITHIN THIS CLASS OF CASES WAS.
NO, IT’S NOT DIFFICULT, IT WOULD CONSIST OF EIGHT
ADDITIONAL STATISTICAL ANALYSES TO CALCULATE THE IMPACT WITHIN
|ANY ONE OF THOSE SUB-CATEGORIES.
@. WOULD YOU REFER TO DB-92. PLEASE, THE SECOND PAGE OF THAT
EXHIBIT?
A. YES.
G. ON THE SECOND LINE OF THE EXHIBIT ON THIS PAGE, 1 NOTE THAT
THE VARIABLE MURDERED TO GET MONEY VALUE FOR SELF OR OTHER. HAS
A NEGATIVE REGRESSION CO-EFFICIENT FOR WHITE VICTIM CASES, IS
THAT CORRECT. AND FOR BLACK VICTIM CASES AS WELL?
A, YES.
@. WOULD THIS BE AN UNEXPECTED OUTCOME FOR THE PARTICULAR
VARIABLE?
A. IF THAT WERE THE ONLY VARIABLE IN THE MODEL. YES. BUT IT’S
BALDUS - CROSS
NOT THE ONLY VARIABLE IN THE ANALYSIS. THERE ARE OTHER
VARIABLES IN HERE THAT ARE HIGHLY CORRELATED WITH THIS VARIABLE.
AND IT. THAT WOULD BE MY OPINION AS TO THE. WHAT THE SOURCE OF
THIS REVERSE SIGN WOULD BE, THAT IS, THIS IS AN AGGRAVATING
FACTOR APPEARING TO HAVE A MITIGATING EFFECT. AND THE REASON
IS, I THINK, THAT THIS FACTOR IS HIGHLY CORRELATED WITH LDF B7D
AND THAT. WHEN THAT IS TAKEN, WHEN THAT IS CONTROLLED FOR FIRST,
WHEN THAT I5 CONTROLLED FOR IN THE MODEL. THAT HAS THE EFFECT OF
PRODUCING THIS APPARENTLY PERVERSE SIGN ON THIS VARIABLE. AND
THATS WHY IT-S IMPORTANT TO LOOK AT THE OTHER VARIABLES THAT
ARE IN THE ANALYSIS TO SEE WHY IT IS THAT YOU MIGHT HAVE A
VARIABLE RUNNING IN THIS SORT OF DIRECTION. IT MAY BE THAT THIS
SYSTEM IS OPERATING IN AN ARBITRARY FASHION. BUT MY JUDGMENT
HERE WOULD BE THAT’S NOT WHAT IS PRODUCING THIS. THAT THIS IS
HIGHLY CORRELATED WITH THE OTHER VARIABLE.
@. SO THE INCLUSION OF CORRELATED VARIABLES COULD HAVE THE
EFFECT, COULD AFFECT THE REGRESSION CO-EFFICIENT. EITHER MAKING
THE SIGN CHANGE OR HAVE SOME EFFECT LIKE THAT?
A. YES. THAT" 3 RIGHT.
@. LDOK AT DB-%4. YOU REFER TO A PROSECUTORS DECISION TO
ADVANCE TO A PENALTY TRIAL?
A. YES.
la. IS THAT CORRECT?
A. CORRECT.
RQ. WHEN MAKING THESE DETERMINATIONS, DID YoU FEEL THAT YOU HAD
[]
a
»
ww
N
BALDUS ~ CROSS
AVAILABLE TO YOU ALL OR MOST OF THE INFORMATION THAT THE
FROSECUTOR HAD IN MAKING THESE DECISIONS?
A. YES, I THINK THE FILE APPROXIMATES THE INFORMATION THAT WAS
AVAILABLE TO THE PROSECUTORS, AT THIS STAGE IN THE PROCESS.
THE FILE DOES NOT PERFECTLY REFLECT IT, I“M SURE. BUT
THE IMPORTANT THING IS THAT IN SOME CASES THERE MAY BE MORE THAN
THE PROSECUTOR HAD AVAILABLE; IN OTHERS, LESS. IN THE END. IT
INTRODUCES NO BIAS, AS FAR AS I KNOW, THAT WOULD AFFECT ANY OF
THE CRITICAL RACE CO-EFFICIENTS.
@. DO YOU HAVE ANY WAY OF CONSIDERING JUDGMENTAL CALLS THAT MAY
BE MADE BY A PROSECUTOR AT THIS STAGE?
FOR EXAMPLE. TO CLARIFY THE QUESTION SOMEWHAT. A
PROSECUTORS DECISION THAT THIS PARTICULAR JURY IS JUST NOT
GOING TO RETURN A DEATH PENALTY?
A. NO, THE, THE ANALYSIS IS BASED STRICTLY ON THE FACTS OF THE
CASES, AND THE STRENGTH OF THE EVIDENCE IN THE CASE. AND
PRESUMABLY THE STRENGTH OF THE EVIDENCE IN THE CASE IS GOING TO
REFLECT IN PART A LIKELIHOOD THAT A JURY IS GOING TO RETURN A
DEATH SENTENCE.
SO EVEN THOUGH WE DON‘T HAVE ANY QUESTION TO GO
DIRECTLY TO WHAT IS IN THE MIND OF THE PROSECUTOR. I THINK WE
HAVE A REASONABLE APPROXIMATION OF HOW SOMEONE WOULD EVALUATE A
CASE. THEY WOULD EVALUATE A CASE, IF IT WASN“T, IF THE EVIDENCE
WEREN“T VERY STRONG AND OUR MEASURES REFLECT THAT, AND EVEN WITH
THAT REFLECTED IN THE ANALYSIS, IT DOESN’T. THAT DOES NOT IN ANY
C
E
H
R
EE
+
ER
|
N
E
R
1140
BALDUS - CROSS
WAY AFFECT THESE RALE EFFECTS.
@. SO YOU THINK. YOURE SAYING THAT YOU HAVE THE STRENGTH OF
EVIDENCE VARIABLES AVAILABLE AND THOSE FARTICULAR VARIABLES
SHOULD ACCOUNT FOR THE DECISIONS BEING MADE, OR SHOULD ACCOUNT
FOR A PORTION OF THE DECISION?
A. WELL. WHAT I“M SAYING IS THAT THE LAW FROVIDES THAT IF
CLEARLY, THE UNDERSTANDING, CERTAINLY, THE UNITED STATES SUPREME
COURT REPORTS. THE WAY THEY VIEWED THIS SYSTEM HERE, THAT THE
PROSECUTOR IS GOING TO BE INFLUENCED BY THE PRESENCE OF
AGGRAVATING FACTORS IN THE CASE AND BY THE LIKELIHOOD THAT THE
| JURY WILL IMPOSE THE DEATH SENTENCE. AND THOSE ARE CLEARLY
APPROPRIATE THINGS FOR A PROSECUTOR TO CONSIDER.
I THINK THAT OUR DATA ON THESE CASES EMBODIES THOSE
DIMENSIONS OF THE CASES, BECAUSE IT GOES TO THE FRESENCE OR
ABSENCE OF STATUTORY, NON-STATUTORY. AGGRAVATING FACTORS.
MITIGATING FACTORS, THOSE ARE ALL THE FACTORS THAT WOULD INCLINE
A JURY TO IMPOSE A DEATH SENTENCE.
ALSO IN THE CHARGING AND SENTENCING STUDY WE HAVE GOOD
MEASUREMENTS ON THE STRENGTH OF THE EVIDENCE. I THINK THOSE
DIMENSIONS OF THE CASES ARE REFLECTED IN THE ANALYSIS THAT
CONCERNS THE DECISION OF A PROSECUTOR TO ADVANCE THE CASE TO THE
PENALTY TRIAL.
S50 THAT IS MY ANSWER WITH RESPECT TO THAT DIMENSION OF
THE CASE. I THINK IT IS CAPTURED IN THESE DATA.
R. IN THIS TABLE. AND IN THE PREVIOUS ONE. WHICH WAS DB-93,
[]
1141
BALDUS - CROSS
WOULD IT BE FAIR TO SAY THAT THE REGRESSION REALLY HAS NO WAY OF
KNOWING WHAT PARTICULAR FACTORS CARRY THE MOST WEIGHT WITH THE
DECISION MAKER AT ANY PARTICULAR STAGE?
A. IN ANY ONE CASE, I WOULD AGREE. THE REGRESSION DOESNT
SHED MUCH LIGHT ON ANY ONE DECISION, BUT AT ANY ONE STAGE. I
THINK THE REGRESSION ANALYSIS DOES SHED IMPORTANT LIGHT ON THE
FACTORS THAT APFEAR TO BE DETERMINING DECISIONS AT THAT POINT.
IT DOESNT TELL YOU PRECISELY. IT GIVES YOU A GOOD ESTIMATE,
ESPECIALLY IF THE CO-EFFICIENTS ARE STRONG AND THEY RE HIGHLY
SIGNIFICANT STATISTICALLY.
I THINK THAT’S WHAT THE REGRESSSION ANALYSIS DOES
DEMONSTRATE TO YOU, IS WHAT ARE THE IMPORTANT FACTORS OPERATING
WITHIN A STRATEM OF DECISION WITHIN THE SYSTEM.
RR. WOULD YOU LOOK ON EXHIBIT DB-?87 UNDER COLUMN A, YOU LIST
THE HEADING FOR THE COLUMN AS NON-RACIAL BACKGROUND VARIABLES
CONTROLLED FOR, AND YOU ALSO LIST DOWN IN NUMBER s, 1460
NON-RACIAL AGGRAVATING AND MITIGATING FACTORS.
IN USING NON-RACIAL, EXCUSE ME, IN THAT SENSE. ARE YOU
REFERRING TO VARIABLES THAT ARE NOT CORRELATED WITH THE RACE OF
A VICTIM?
A. OH NO. LOT OF THEM ARE CORRELATED. A GREAT NUMBER -—— IVE
NOT ACTUALLY DONE A CORRELATION ANALYSIS ON THIS PARTICULAR SET
OF VARIABLES. BUT I HAVE ON A COMPARABLE SET IN THE CHARGING AND
SENTENCING STUDY, AND A VERY SUBSTANTIAL PROPORTION OF THOSE
VARIABLES ARE CORRELATED WITH THE RACE OF THE VICTIM. CORRELATED
1142
BALDUS - CROSS
BOTH WITH THE RACE OF THE VICTIM AND THE DEATH SENTENCING
RESULT.
THE COURT: FOR EXAMPLE?
THE WITNESS: I’M SORRY, YOUR HONOR?
THE COURT: FOR EXAMPLE?
THE WITNESS: FOR EXAMPLE, THE PRESENCE OF A
CONTEMPORANEOUS OFFENSE. THE B7 FACTOR, TORTURE, EXECUTION-STYLE
MURDER. ALL OF THE BIG PROPORTION OF THE MAJOR NON-STATUTORY
AGGRAVATING FACTORS AND STATUTORY AGGRAVATING FACTORS SHOW A
POSITIVE CORRELATION WITH BOTH THE DEATH SENTENCING RESULT AND
THE RACE OF THE VICTIM.
OF THE TEN STATUTORY AGGRAVATING CIRCUMSTANCES, THERE
ARE AT LEAST SIX. Bi, B2, B&, B7, I THINK B10, THEY SHOW THAT
PROPERTY, OF BEING CORRELATED WITH EOTH THE RACE OF THE VICTIM
AND OUTCOME VARIABLE.
BY M3. WESTMORELAND:
@. DO YOU FIND OTHER AGGRAVATING AS WELL AS MITIGATING FACTORS
THAT ARE HIGHLY CORRELATED THE WITH THE RACE OF THE VICTIM AS
WELL?
A. DOH, YES. YES. THERE’S A TABLE THAT I PREFARED TO
DEMONSTRATE EXACTLY THAT PQINT IN THE CHARGING AND SENTENCING
STUDY, THAT THERE ARE OVER MORE THAN A HUNDRED VARIABLES THAT
SHOW STATISTICALLY SIGNIFICANT RELATIONSHIP WITH BOTH THE DEATH
SENTENCING RESULT AND THE RACE OF THE VICTIM.
@. IN MAKING YOUR DETERMINATION IN THE TABLES IN WHICH YOU MADE
H
ra
A]
1143
BALDUS -— CROSS
YOUR BREAKDOWNS ON URBAN AND RURAL JUDICIAL CIRCUITS. DID YOU
PRESUME THAT ALL COUNTIES IN ONE CIRCUIT HAD THE SAME PROPERTY
OF BEING URBAN OR RURAL. NO MATTER WHAT THE POPULATION OF THAT
PARTICULAR COUNTY WAS?
A. YES. I PROCEEDED ON THE ASSUMPTION THAT THE CRITICAL
ADMINISTRATIVE UNIT WAS THE CIRCUIT AND I HAD TO TREAT THEM ALL
ALIKE IF I THOUGHT THAT WAS GOING TO BE THE UNIT. SO THE ANSWER
13, CORRECT.
@. IN DOING THE ANALYSES. PROFESSOR BALDUS. WERE ANY
INTERACTION VARIABLES DEFINED?
A. YES.
2. DO YOU RECALL HOW MANY OR OF WHAT -—-
A. CERTAINLY. I CAN TELL YOU THE INTERACTION ANALYSIS THAT I
DID, WHICH WAS PRELIMINARY. AND THEN PROFESSOR WOODWORTH DID AN
EXTENSIVE AMOUNT OF INTERACTION ANALYSIS AS WELL.
WHAT I DID WAS TAKE THE FIVE-VARIABLE MODEL THAT
INCLUDED OUR FELONY CIRCUMSTANCES. PRIOR RECORD, MULTIPLE
VICTIMS, DRUNKEN. LOVER RUARREL, AND ONE OTHER VARIABLE THAT
ESCAPES ME AT THE MOMENT. AND I TOOK THOSE VARIABLES AND CREATED
A SERIES OF INTERACTION TERMS. I CREATED AN INTERACTION TERM
BETWEEN EACH OF THOSE VARIABLES AND THE RACE OF THE DEFENDANT.
EACH OF THOSE VARIABLES AND THE RACE OF THE VICTIM AND EACH OF
THOSE VARIABLES WITH EACH OTHER. S50 THAT YOU HAD A SERIES OF
INTERACTIONS. EVERY COMBINATION OF VARIABLES THAT WAS IN THAT
BASICALLY 7-VARIABLE ANALYSIS. THAT IS, WHEN YOU INCLUDED THE
[4
a
1144
BALDUS -~ CROSS
FIVE NON-RACIAL FACTORS AND THE RACIAL FACTOR, THERE WAS A
COMBINATION FOR EACH OF THOSE VARIABLES IN THAT ANALYSIS.
AND THEN I CONDUCTED THE REGRESSION ANALYSIS WITH THOSE
INTERACTION TERMS.
@. IN THE CHARGING AND SENTENCING STUDY. AND IN THE PROCEDURAL
REFORM STUDY AS WELL. WHAT PROCEDURE WAS FOLLOWED WHEN THE I.G.
OF THE DEFENDANT WAS LUNKNOWN?
A. I DID NOT USE THE I.Q. OF THE DEFENDANT IN THIS STUDY.
THE. I FOUND. AS WE GOT GOING IN THE STUDY TO
COMPLETION, THAT THERE WERE A NUMBER OF VARIABLES, THERE WERE
VARIABLES OF THE TYPE THAT ONE COULD NOT RELY ON THE INFORMATION
ON THE FILE TO MAKE A JUDGMENT ABOUT WHETHER OR NOT IT WAS
PROBABLY KNOWN TO THE DECISION MAKER.
OUR PRINCIPAL ASSUMPTION WAS THAT INFORMATION
CONCERNING A LEGITIMATE AGGRAVATING AND MITIGATING FACTOR IN
THE CASE WAS THE KIND OF INFORMATION THAT IS PRESERVED IN THE
RECORDS OF THESE FILES. AND THAT IT APPROXIMATES WHAT WAS KNOWN.
HOWEVER, WITH RELATION TO SUSPECT FACTORS, RACIAL
FACTORS, I.Q@., SEX. NOT SEX, THAT APPEARS. BUT EDUCATION. I.&.,
THOSE ARE FACTORS THAT THE SYSTEM DOES NOT AUTOMATICALLY INCLUDE
IN A FILE. SO TO GET INFORMATION ON THOSE FACTORS, WE HAD TO GO
TO THE DEPARTMENT OF CORRECTIONS. BUT IT TURNED QUT THAT THE
INFORMATION IN THE DEPARTMENT OF CORRECTIONS HAS SERIOUS
PROBLEMS WITH IT. AND THE PRINCIPAL PROBLEM IS THAT THE
DEPARTMENT OF CORRECTIONS DOES NOT COLLECT INFORMATION ON
1145
BALDUS - CROSS
OFFENDERS WHO HAD BEEN SENTENCED TO DEATH IN THE SAME WAY OTHER
OFFENDERS. AND AS A CONSEQUENCE, THERES A VERY HIGH PROFORTION
OF MISSING INFORMATION WITH RESPECT TO THESE VARIABLES AS IT
RELATES TO THE DEATH SENTENCE CASES.
AND PROFESSOR WOODWORTH DID EXTENSIVE WORK ON THIS. AND
CONCLUDED THAT BECAUSE OF THIS CORRELATION BETWEEN THE MISSING
VALUES ON THIS VARIABLE. AND THE OUTCOME VARIABLE. THE DEATH
SENTENCING RESULT. THAT IT WAS NOT PRUDENT TO USE THOSE
VARIABLES. S0 THEY WERE NOT USED.
2. IN SITUATIONS IN WHICH THE EXISTENCE OF PRIOR CONVICTIONS
WAS UNKNOWN, WHAT CODING PROCEDURES WERE UTILIZED?
A. THE, WE RELIED ON THE RECORD, AND IF THE RECORD DID NOT
INDICATE THERE WAS A PRIOR RECORD. THEN THE CASE WAS TREATED AS
IF THERE WERE NO PRIOR RECORD, AGAIN ON THE ASSUMPTION THAT THE
RECORD REFLECTED WHAT PEOPLE SAW, AND HOW FEOPLE PERCEIVED THE
CASE, AND IF THEY DID NOT PERCEIVE A RECORD THEY WOULD NOT TREAT
A PERSON AS HAVING A RECORD.
2. IN CASES IN WHICH THERE WERE MULTIPLE VICTIMS INVOLVED, WERE
THESE MULTIPLE VICTIMS ACCOUNTED FOR IN THE PROCEDURAL REFORM
STUDY AND IN THE CHARGING AND SENTENCING STUDY?
@. DID YOU PROVIDE FOR MORE THAN TWO VICTIMS OR THREE VICTIMS
IN EACH STUDY OR HOW MANY VICTIMS?
A. WELL, ILL EXPLAIN TO YOU WHAT WE DID.
AS IS THE CASE WITH VIRTUALLY ALL VARIABLES THAT
1146
BALDUS - CROSS
INVOLVE COUNTS OF ONE SORT OR ANOTHER, HOW MANY PRISON SENTENCES
YOU HAVE. HOW MANY VICTIMS, HOW MANY MITIGATING FACTORS, WE
WOULD CREATE A VARIABLE THAT WOULD INCLUDE THE COUNT, AND THEN
CREATE A VARIABLE THAT WOULD SAY ONE OR MORE EXISTS.
WITH RESPECT TO THE VARIABLE OF MULTIPLE VICTIMS WE
FOUND THAT THE DIMENSION THAT WAS CRITICAL IN TERMS OF EXPLAINING
DECISIONS WAS WHETHER ONE OR MORE VICTIMS EXISTED IN THE CASE.
WE DID ANALYSES WHERE YOU DID THE COUNTING. IT DIDNT
ADD ANY ADDITIONAL LIGHT. IT DIDN‘T EXPLAIN THE DECISIONS ANY
BETTER.
S00 SINCE THE OTHER VARIABLE IS EASIER TO INTERPRET.
THAT”S WHY WE USED THAT VARIABLE.
@. DID YOU PROVIDE INFORMATION ON THE QUESTIONNAIRE TO CODE
INFORMATION FOR ALL THE VICTIMS?
A. YES.
@. HOW WAS THAT DONE, BY A MULTIPLE VICTIM SHEET OR --
A. YES, THERE WAS A SUPPLEMENTAL SHEET FOR EACH VICTIM.
@. AND WAS THERE TO BE A SUPPLEMENTAL SHEET ADDER FOR EACH
ADDITIONAL VICTIM IN THE CASE?
A. YES.
R. IN DEFINING THE VARIABLES TO UTILIZE IN THE ANALYSIS. DID
YOU CONSIDER ANY INFORMATION FROM THE MULTIPLE VICTIMS, THE
SECOND VICTIM OR THIRD VICTIM, ON DID YOU ONLY CONSIDER THE
FIRST?
A. NO. WE CONSIDERED THE ADDITIONAL INFORMATION AND THE
ro
1147
BALDUS - CROSS
PRINCIPAL ADDITIONAL INFORMATION WE WERE CONCERNED WITH WAS THE
RACE OF THE SUPPLEMENTAL VICTIMS SO THAT WE ASKED THE QUESTION
WHETHER OR NOT, FOR THE PURPOSES OF DEFINING WHETHER THE CASE
INVOLVED A WHITE VICTIM, THE MEASURE WAS, WAS, IS THERE ONE OR
MORE WHITE VICTIM IN THE: CASE? WE EXAMINED THAT INFORMATION.
WE DID EXAMINE THE OTHER INFORMATION CONCERNING THE
VICTIM. WE FOUND THAT BY AND LARGE THAT. THAT ADDITIONAL
INFORMATION ABOUT THOSE VICTIMS WAS NOT TERRIBLY POWERFUL IN
EXPLAINING THE RESULTS, AND IT WAS NOT INCLUDED.
THE ONLY INFORMATION FROM THE SUPPLEMENTAL VICTIM
INCLUDED IN THE MACHINE ANALYSIS IS THE QUESTION ABOUT WHETHER
OR NOT THERE WERE MORE THAN ONE WHITE VICTIM.
RQ. IN INSTANCES IN WHICH YOU WERE DETERMINING MITIGATING
CIRCUMSTANCES, OR CIRCUMSTANCES PERTAINING TO THE VICTIM, WHICH
VICTIM WOULD BE UTILIZED WHEN YOU HAD MULTIPLE VICTIMS?
A. THE FIRST VICTIM.
TO THE BEST OF MY KNOWLEDGE THAT 3S WHAT WE VE
DONE.
RR. WOULD THAT BE THE FIRST VICTIM CODED OR THE FIRST VICTIM
KILLED?
A. I DON’T RECALL THE —-- I DONT, I DON’T RECALL THAT WE HAD
ANY PARTICULAR RULE ABOUT THAT. IT WAS THE WAY IN WHICH THEY
WERE PICKED BY THE CODERS. THE CODER WOULD IDENTIFY THE FIRST
VICTIM.
THE INSTRUCTIONS WERE TO PICK THE PERSON THAT SEEMED TO
1142
BALDUS - CROSS
BE THE PRINCIPAL VICTIM IN THE CASE, THE PERSON THE ANIMUS WAS
PRINCIPALLY DIRECTED FOR. AGAINST. FOR EXAMPLE IN A CASE LIKE
GODFREY, THE FIRST VICTIM WOULD BE THE WIFE, THE VICTIM WHOSE
RELATIONSHIP TENDED TO EXPLAIN WHAT WAS GOING ON BETTER THAN THE
SUPPLEMENTAL VICTIMS.
R. IS THAT THEN PRESUMING THAT ONE VICTIM WOULD HAVE A
RELATIONSHIP WHICH WOULD EXFLAIN THE KILLING MORE THAN ANOTHER
VICTIM WOULD?
A. WELL, IF THEY DIDs THEN IF THEY DIDN-“T, IT WOULDN‘T MAKE ANY
DIFFERENCE WHICH ONE YOU CODED. THEY‘RE ALL EQUAL IN TERMS OF
THEIR RELATIONSHIPS. IT WOLILDNT MAKE ANY DIFFERENCE. SO THAT
WAS THE RATIONALE BEHIND IT, IT WOULDN‘T MAKE ANY DIFFERENCE.
SO WE WANTED TO PRESENT THE INFORMATION THAT WOULD EXPLAIN THE
DYNAMICS OF THE RELATIONSHIP BETWEEN THE OFFENDER AND THE
VICTIM.
THE COURT: STRANGER BREAKS INTO A HOUSE. AND HE SHOOTS
A FATHER AND MOTHER AND A CHILD. WHO DO YOU PICK?
THE WITNESS: WELL, THAT. I. FRANKLY I CANNOT TELL YOU
WHAT THE RULE WAS THEY USED IN GEORGIA TO MAKE THAT
{DETERMINATION, BUT THAT WOULD BE AN EXAMPLE.
I PRESUME IT WOULD BE THE MALE. THE HEAD OF THE
HOUSEHOLD WOULD BE THE PERSON THAT WOULD BE PICKED. BUT I WOULD
THINK, THAT WOULD BE AN EXAMPLE, THAT THEY WERE ALL STRANGERS.
AND AS FOR THE RELATIONSHIPS BETWEEN THEM. IT WOULDN’T MAKE ANY
DIFFERENCE WHICH PERSON YOU CODED, YOUR HONOR.
| L
L]
THE PROSECUTOR TO ENTER INTO PLEA NEGOTIATIONS?
1149
BALDUS - CROSS
WE CONSIDERED LOTS OF THESE OPTIONS AS PERHAPS CREATING
WEIGHTS.
WE STUDIED, WHEN YOURE TALKING ABOUT MULTIPLE
VICTIMS, YOU’RE NOT TALKING ABOUT AN ENORMOUS AMOUNT OF
INFORMATION. YOURE TALKING ABOUT 40 CASES.
THE COURT: FORTY?
THE WITNESS: FORTY CASES APPROXIMATELY, MULTIPLE
VICTIM CASES, AND 1 PERSONALLY EXAMINED THE DISTRIBUTION OF THE
INFORMATION ON ALL THOSE VARIABLES AND WITH THE IDEA OF TRYING
TO CAPTURE THIS DYNAMIC. AND I CONCLUDED ON LOOKING AT IT THAT
I COULD COME UP WITH NO INTERESTING INDEX OR WAY OF COMEINING
THIS INFORMATION INTO NEW VARIABLES THAT WOULD CREATE BETTER
UNDERSTANDING OF WHAT WAS GOING ON.
I WAS SURPRISED WHEN THAT HAPFENED., BECAUSE I HAD HOPED
TO BE ABLE TO USE ALL THAT INFORMATION. BUT I WAS STYMIED IN TERMS
OF DEVELOPING A PROBATIVE MEASURE THAT WOULD CAFTURE THOSE
DIMENSIONS OF THE CASE.
R. WHEN YOU WERE LOOKING AT SITUATIONS INVOLVING GUILTY PLEAS.
AND YQU DID SOME ANALYSES BREAKING DOWN ON THAT PARTICULAR
DECISION MAKING STAGE, WHAT FACTORS DID YOU HAVE AVAILABLE TO
YOU THAT YOU CONSIDERED CONSIDERED RELEVANT IN THE DECISION FOR
A. I CONSIDERED ALL THE VARIABLES IN THE FILE RELEVANT,
PARTICULARLY THOSE THAT HAD TO DO WITH THE, THE TWO CRITERIA
THAT I MENTIONED TO YOU EARLIER THAT WERE OUTLINED AS THE SORT
od
hd
1130
BALDUS - CROSS
OF GUIDELINES AS CONTEMPLATED BY THE UNITED STATES SUPREME
COURT, THAT IS, THE SERIOUSNESS OF THE OFFENSE. LIKELIHOOD THAT
A JURY WOULD IMPOSE THE DEATH SENTENCE AND WE HAVE A LOT OF
INFORMATION IN THE FILE THAT DEALS WITH THE. THOSE FACTORS. THAT
IMPACT ON JURY DEATH SENTENCING DECISIONS.
FURTHERMORE, THE STRENGTH OF THE EVIDENCE, THE
LIKELIHOOD OF GETTING A CONVICTION. THAT WAS THE OTHER PRINCIFAL
FOCUS OR ANALYSIS, AND THOSE INDEED WERE THE. UNDER THOSE TWO ;
HEADINGS. YOU CAN ORGANIZE EVERY VARIABLE IN OUR FILE. THEY
ALL, THAT I5, THE SUBSTANTIVE VARIABLES, THEY ALL RELATE TO ONE
OR BOTH OF THOSE QUESTIONS.
©. WOULD YOU SAY OR WERE THERE FACTORS RELATING TO THE
BARGAINING DECISION THAT YOU DID NOT HAVE INFORMATION ABOUT?
A. YES. 1I.THINK THAT THERE ARE PERHAPS FACTORS THAT INFLUENCE
DECISION MAKING BY PROSECUTORS THAT WE DIDN’T HAVE INFORMATION
ABOUT.
@. FOR EXAMPLE. WOULD YOU HAVE INFORMATION ABOUT A PROSECUTORS
DESIRE TO TRY TWO CO-FERPETRATORS TOGETHER?
A. NO. I DON‘T THINK THE FILE REFLECTS THAT DESIRE ON THE PART
OF A PROSECUTOR. NO. |
R. WOULD YOU HAVE INFORMATION AS TO THE FROSECUTOR”S3
DETERMINATION THAT HIS MAIN WITNESS WOULD NOT STAND UF TO
CROSS-EXAMINAT ION?
A. WELL, I“D HAVE TO LOOK AT THE RUESTIONNAIRE ABOUT THAT. WE
DO HAVE A QUESTION OR TWO THAT RELATE TO THE. THE, THE FROBLEMS
—
c
o
E
Y
E
E
S
R
f
T
M
e
10
11
|EVIDENCE, BUT I DON’T RECALL ANY MEASURE OF WHETHER THERE WERE
1151
BALDUS - CROSS
WITH A PRINCIPAL WITNESS. BUT THAT INFORMATION DID NOT CONTAIN A
LOT OF SUBTLETY AND NUANCES. I MEAN IT INDICATES. DOES THE FILE
INDICATE THAT THERE ARE PROBLEMS WITH THE PRIMARY WITNESS? IF
IT WEREN’T REFLECTED IN THE FILE, WE WOULD HAVE NO INFORMATION
ON IT.
THE COURT: YOU MEASURED THE AVAILABILITY OF PHYSICAL
ANY FDURTH AMENDMENT PROBLEMS WITH GETTING IT IN.
THE WITNESS: THERE IS A QUESTION THAT DEALS WITH THAT,
YOUR HONDR. BUT FRANKLY WE DIDN“T CONSIDER THE FILES THAT WE
WERE CONSULTING VERY RELIABLE SOURCES OF INFORMATION ABOUT ANY
LEGAL ISSUES HERE. THE FOURTH AMENDMENT. AND BASICALLY. I FELT
UNCOMFORTABLE ABOUT USING THAT VARIABLE FOR THAT REASON. THAT
FILE DID NOT, I MEAN THE INFORMATION, IT“S ONE THING TO
TALK ABOUT HAVING SOME CONFIDENCE ON INFORMATION RELATING TO THE
FACTS. SORT OF THE BRUTE FACTS IN THE CASE. IT“S ANOTHER THING
TO HAVE VERY MUCH CONFIDENCE IN THE TRANSCRIPTION OF INFORMATION
HAVING ANYTHING TO DO WITH THE SUBTLETIES OF THE FOURTH AMENDMENT.
50 I DID NOT CONSIDER THAT A TERRIBLY RELIABLE VARIABLE.
I WOULD SAY. JUST IN PASSING, THAT I THINK IT IS TRUE
WE DID NOT HAVE INFORMATION ON SOME OF THE NUANCES AND SUBTLETIES |
OF THIS PROCESS, BUT DON THE BASIS OF MY KNOWLEDGE OF HOW THIS
SYSTEM OPERATES AND OTHER SYSTEMS, I DONT SEE THAT WOULD HAVE
ANYTHING EXCEPT A RANDOM EFFECT ON THE ANALYSIS THAT WE DID.
@. WOULD YOU HAVE INFORMATION ON INSTANCES IN WHICH THE
1152
BALDUS - CROSS
PROSECUTOR OFFERED A PLEA BARGAIN AND THE DEFENDANT REFUSED TO
ACCEPT IT?
A. IN 4&0 PERCENT OF THE CASES WE HAVE THAT INFORMATION. YES.
@. DO YOU HAVE INFORMATION THAT WOULD PERHAPS BE A SOMEWHAT
MORE SUBTLE DERIVATION OF THAT, THE PROSECUTOR WOULD SAY. "YES,
I‘LL TAKE A GUILTY PLEA, BUT I“M NOT GOING TQ MAKE A
RECOMMENDATION AS TO THE SENTENCE?"
A. YES. WE, OUR QUESTIONNAIRE WOULD INDICATE THAT. OUR
QUESTIONNAIRE ADDRESSED THAT VERY QUESTION, WAS THERE AN
AGREEMENT TO A PLEA, WAS THERE AN AGREEMENT AS TO A CERTAIN
RECOMMENDED SENTENCE. AND IN A GREAT MANY OF THE CASES THERE
WAS AN AGREEMENT TQ THE PLEA SUGGESTED, BUT NO INDICATION THERE
WAS ANY PARTICULAR PROMISE AS TO WHAT WOULD BE RECOMMENDED.
Rl. AND I BELIEVE YOU STATED EARLIER THIS IS INFORMATION YOU
OBTAINED FROM THE PROSECUTORS AND DEFENSE COUNSEL?
A. THAT'S RIGHT.
R. WOULD THAT BE CORRECT?
THEN ON WHAT PERCENTAGE DF THE CASES DID YOU HAVE THAT
TYFE OF INFORMATION?
A. 60 PERCENT.
THE COURT: &0O PERCENT YOU KNEW WHAT?
WHAT WAS YOUR QUESTION, MS. WESTMORELAND?
MS. WESTMORELAND: AT THIS POINT, THE INFORMATION ON
WHETHER ESSENTIALLY IT WAS OFFERED, A FLEA WAS OFFERED WITHOUT A
RECOMMENDATION OF A PLEA BARGAIN, WAS THAT SPECIFIC QUESTION.
th
B
E
1153
BALDUS - CROSS
BY MS. WESTMORELAND:
@. DID YOUR 60 PERCENT RELATE TO INFORMATION ON PLEA BARGAINS
IN GENERAL?
A. YES. AND WHEN WE WOULD GET AN ANSWER TO THAT QUESTION. THE
QUESTION WAS USUALLY. THE QUESTIONNAIRE WOULD USUALLY ADDRESS
THE OTHER QUESTION. IT WOULD SAY, "WAS THERE A PLEA BARGAIN
OFFERED OR ACCEPTED?"
AND THEN IF THE ANSWER WAS "YES" TO THAT. THEN WE GOT
ANSWERS ON THAT FIRST QUESTIONNAIRE ON ABOUT 60 FERCENT
OF THE CASES. THEN IF IT WAS, THE NEXT QUESTION WOULD BE, "WHAT
WAS OFFERED, WHAT WAS THE PROPOSAL." AND IN A LOT OF THOSE
CASES, THERE WOULD BE NO SPECIFIC OFFER MADE. IT WAS JUST A
TENDER OF A FLEA BARGAIN. OR A MANIFESTED WILLINGNESS TO
AGREE TO ONE, BUT THE QUESTIONNAIRE WOULD BE SILENT AS TO WHAT
IT WAS THAT WAS ACTUALLY OFFERED.
IF ANYTHING WAS OFFERED, AND I ASSUME THAT 3S THE KIND
OF CASE YOU’RE REFERRING TO, NOTHING WAS OFFERED, EXCEPT THE
NOTION OF A PLEA IN GENERAL.
@. DID YOU UTILIZE IN THE ANALYSES THAT WERE DONE COMPOSITE
VARIABLES?
A. YES. IF YOU MEAN BY COMPOSITE VARIABLES COUNTS OF THINGS
OR, ILL PUT IT THIS WAY, WHEN WE MADE THE RECODES, WE WOULD
COMBINE INFORMATION IN THE QUESTIONNAIRE S50 IT WOULD BE
REFLECTED IN A SINGLE VARIABLE. MANY OF THE VARIABLES CARRY
VERY CLOSE TO THE SAME INFORMATION, OR THE INFORMATION IS
BALDUS - CROSS
RELATED. SO VERY OFTEN WE WOULD WANT TO COMBINE THAT
INFORMATION INTO ONE VARIABLE. THAT IDENTIFIED A PARTICULAR
DIMENSION OF THE CASE AT A SLIGHTLY HIGHER LEVEL OF ABSTRACTION.
FOR EXAMPLE, THE VARIABLE YOU REFERRED TQ EARLIER ABOUT LOVER,
DRINKING. BAR ROOM QUARRELS OR SOMETHING, IT WAS A VARIABLE THAT
PICKS UP A LOT OF DIFFERENT CASES FALLING IN THAT CATEGORY. BUT
IT REFLECTS A DIMENSION OF THE CASE CONSIDERED TO BE IMPORTANT
IN TERMS OF ESTIMATING THE DEATH WORTHINESS OF A CASE. SO WE
MADE COMPOSITES IN THAT SENSE.
@. IN MAKING THESE COMPOSITE VARIABLES. ISN‘T IT POSSIBLE THAT
THE USE OF THESE TYPES OF VARIABLES CAN INCREASE THE EFFECT OF
LINKNOWNS ON THE OUTCOME?
A. I DON’T THINK SO. BUT THAT‘S A TECHNICAL QUESTION THAT I
CONSIDER BEYOND MY COMPETENCE.
THE COURT: WHAT WAS THE —-
THE WITNESS: IT’S NOT CLEAR TO ME WHAT EFFECT YOU MEAN
EY "UNKNOWNS." I“M HAVING TROUBLE UNDERSTANDING WHAT YOU! MEAN
BY EFFECTS OF UNKNOWNS. I DON‘T KNOW WHAT, QUITE WHAT YOU MEAN
BY THAT.
BY MS. WESTMORELAND:
Qn. WELL. IN THE SENSE OF AFFECTING THE OUTCOMES OF REGRESSIONS,
FOR INSTANCE. OR YOUR SUBSEQUENT STATISTICAL ANALYSEST
A. NO. I DON’T THINK SO. IN TERMS OF THE WAY WE HANDLED THE
DATA AND TREATED UNKNOWNS, WE TREATED AN UNKNOWN AS INDICATING
THE DECISION MAKER DID NOT KNOW WHETHER A FACT EXISTED. IF IT
BALDUS - CROSS
WAS NOT KNOWN WHETHER THE FACT EXISTED. THEN WE TREATED THAT
VARIABLE AS NOT BEING A FACTOR IN THE DECISION OF THE DECISION
MAKER. AND ON THE BASIS OF THAT PERSPECTIVE IT HAS NO EFFECT,
WHATEVER. THE PRESENCE OF THE UNKNOWN GIVES US A CLEAR PICTURE
OF WHAT WAS ACTUALLY PERCEIVED IN THAT CASE. AND THE UNKNOWNS
UNDER THOSE CIRCUMSTANCES HAVE NO EFFECT ON THE REGRESSION
ANALYSIS.
THIS IS NOT & -— SEE, WHEN YOURE ADDRESSING THE
QUESTION AT THIS LEVEL, IT“S NOT A TECHNICAL STATISTICAL
QUESTION. THE WAY I VIEW THIS FILE. WE DID NOT HAVE MISSING
DATA ON THE LEGITIMATE AGGRAVATING AND MITIGATING FACTORS OF
THESE CASES. WE HAD A CORRECT REFRESENTATION OF WHAT WAS IN
THAT FILE. THOSE WERE NOT MISSING DATA AS STATISTICIANS VIEW
MISSING DATA PROBLEMS.
S0 CORRESPONDINGLY. WE DID NOT HAVE ANY PROBLEMS WITH
RESPECT TO THE REGRESSIONS FROM THAT PERSPECTIVE. WHICH IS THE
UNDERLYING PERSPECTIVE OF QUR STUDY.
@. I THINK YOU JUST ANSWERED THIS NEXT QUESTION.
SO YOU PRESUMED, THEN, IF YOU HAD UNKNOWNS, THEN THOSE
ITEMS WERE NOT KNOWN TO THE FERSON MAKING A DECISIDN, IS THAT
CORRECT?
A. THATS RIGHT. THAT’S OUR PRINCIPAL ASSUMPTION IN THE
ANALYSIS.
THE COURT: IS THAT YOUR ASSUMPTION ON BLANKS OR ON
UNKNOWNS?
BALDUS - CROST
THE WITNESS: OH. ON UNKNOWNS, YOUR HONOR. BLANKS,
YOUR HONOR. VERY OFTEN MEAN THE FACT WAS NOT THERE.
THE COURT: THE FACT IT WASN‘T IN THE FILE INDICATED TD
YOU THE DECISION MAKER HAD NO INFORMATION ON THAT SUBJECT?
I THOUGHT I REMEMBERED MR. GATES” TESTIMONY BEING IF
THE ,FILE WAS AMBIVALENT, AND THERE WAS SOME INDICATION MAYBE S50,
MAYBE NOT, THAT THEY CODED IT LNKNOWN.
AM 1 DISREMEMBERING THE TESTIMONY, MR. BOGERY
MR. BOGER: YOUR HONOR, I WASN“T IN THE COURTROOM
THROUGHOUT THE TESTIMONY OF MR. GATES.
I THINK MR. FORD WAS.
MR. FORD: IF I UNDERSTOOD YOUR HONOR’S STATEMENTS, I
THINK IT WAS CORRECT.
ON MOST OF THE QUESTIONS, OF COURSE. THERE WAS A
VARIATION. BUT MY RECOLLECTION IS THAT ON MOST OF THE QUESTIONS
WHERE THERE WAS A 1, 2, "U", BLANK FORMAT. THE "U" WAS LISED
WHERE IT WAS UNABLE TO STATE WHETHER IT WAS THAT --
THE COURT: WE HAD A CASE, I DON’T BELIEVE YOU WERE IN
THE COURTROOM EITHER, PROFESSOR BALDUS. WHERE THE DEFENDANT
MENTIONED THE CRIME TO SEVERAL OTHER PEOPLE. AND THERE WAS SOME
QUESTION HAVING TO DO WITH WHETHER HE WAS BOASTFUL ABOUT IT, TWO
OR THREE OTHER RUESTIONS THAT WOULD RELATE TO REMORSEFUL.
BOASTFUL, OTHER ATTITUDINAL THINGS. AND IF 1 REMEMBER MR. GATES”
TESTIMONY, HE TESTIFIED THAT SINCE IN THAT CASE THERE WAS SOME
INDICATION THAT THOSE FACTORS MIGHT HAVE BEEN FRESENT. BUT YOU
Ty
B
k
1
n
t
S
E
«
L
E
E
4
R
R
R
E
BR
R
3
-
BALDUS - CROSS
COULDNT ASCERTAIN WHETHER THEY CUT FRO OR CON, HE CODED THE
"UNKNOWN" AND THAT WAS THE RULE OF DECISION. WHEREAS IF THERE
WAS NO INDICATION THAT HE EVER MENTIONED IT TO ANYBODY. THEN HE
WOULD CODE IT AS A BLANK, MEANING THAT THE DECISION MAKER DIDN’T
KNOW IT.
KNOWING THAT ABOUT THE TESTIMONY. DOES THAT CHANGE YOUR
VIEW OF WHAT YOUVE JUST SAID?
THE WITNESS: NO. BECAUSE I THINK THAT UNDER BOTH
CIRCUMSTANCES. THE EFFECTS WOULD BE LIKELY TO BE THE SAME ON A
DECISION MAKER. THAT IS, WE WERE INTERESTED IN IDENTIFYING. YOUR
HONOR, THE FACTS THAT WERE CLEARLY INVOLVED IN THE CASES THAT
WOULD BE LIKELY TO DETERMINE THE OUTCOME.
IF A PERSON WAS UNCERTAIN ABOUT SOMETHING, REALLY
DIDN‘T KNOW ONE WAY OR ANOTHER WHETHER IT EXISTED. OUR VIEW IS,
ON THE BASIS OF MY KNOWLEDGE OF HOW PECPLE PROCESS INFORMATION,
THAT THEY-D TREAT THAT THE SAME WAY THEY WOULD IF THEY KNEW THAT
THE FACT DIDN-T EXIST. THAT IS, IT IS NOT AN AFFIRMATIVE
FACTOR. EACH ONE OF THESE IS A POSITIVE AGGRAVATOR OR MITIGATOR
AND THE QUESTION IS DID THE EVIDENCE SUPPORT AN INFERENCE. THE
EVIDENCE IN THE FILE, DOES IT SUPPORT AN INFERENCE THAT THE
FERSON BELIEVED THAT TO EXIST.
UNLESS THE RECORD WOULD SUGGEST TO US THAT THE FERSON
WHO KNEW WHAT WAS IN THAT FILE BELIEVED THAT IT EXISTED, THEN WE
CONSIDERED IT APPROPRIATE TO TREAT IT AS NOT A FACTOR IN THE
ANALYSIS, AND THEREFORE TO TREAT IT AS NOT PRESENT IN THE CASE.
BALIUS - CROSS
THE COURT: S0 YOUR "U“3" AND BLANKS WERE THE SAME
THING.
THE WITNESS: NOT ENTIRELY. YOUR HONOR, BECAUSE A BLANK
-— OH, IN TERMS OF THE ULTIMATE WASH. YES, IN TERMS OF THE
ULTIMATE EFFECT. YES. THAT IS, IF YOU, WE WOULD TREAT SOMETHING
UNKNOWN IN THE SAME WAY AS WE WOULD TREAT THAT AGGRAVATOR OR
MITIGATOR IF IT WERE CODED AS NOT PRESENT IN THE CASE. THAT IS,
THAT YOU COULD MAKE AN INFERENCE THAT IT DID NOT OCCUR, THAT ON
THE BASIS OF OUR THEORY OF HOW PEOPLE MAKE DECISIONS THAT THE
BELIEF THAT SOMETHING DIDN’T EXIST WOULD HAVE THE SAME EFFECTS
ON THE DECISION AS A STATE OF IGNORANCE ABOUT WHETHER IT
EXISTED.
THE COURT: YOU SEEM ABOUT TO LEAVE PLEA BARGAINING,
AND I HAVE A COUPLE QUESTIONS.
THE WITNESS: IM SORRY, YOUR HONOR. I MISSED THAT.
THE COURT: MS. WESTMORELAND SEEMS TO BE ABOUT TO LEAVE
THE SUBJECT OF FLEA BARGAINING WITH YOU.
WERE YOU?
MS. WESTMORELAND: JUST ABOUT, YOUR HONOR. FLEASE GO
AHEAD.
THE COURT: THERE ARE A WHOLE LOT OF THEORIES ON WHY
PLEA BARGAINING HAFPENS, ONE OF WHICH IS CLASSICALLY CALLED CASE
PRESSURE.
THE WITNESS: IM SORRY. YOUR HONOR. I DIDN'T GET
THAT.
BALIUS - CROSS
THE COURT: I SAID THERE ARE A LOT OF THEORIES ABOUT
WHY PLEA BARGAINING OCCURS, ONE OF WHICH IS CALLED THE CASE
PRESSURE THEORY. OR WHATEVER YOU WANT TO CALL IT.
DID YOU MAKE ANY STUDIES, GOING BACK TO YOUR EXAMFLE
ABOUT DISCRIMINATION AGAINST FEMALES AT BERKELEY AND THAT HAD A
NAME?
THE WITNESS: OH, YES. SIMFSON“S PARADOX.
THE COURT: DID YQU SATISFY YOURSELF THAT THE OFFERING
OF PLEA BARGAINS OR NOT OFFERING OF PLEA BARGAINS WAS NOT BEING
EXFLAINED BECAUSE OF ANY SORT OF CASE PRESSURE EXPLANATION IN
SOME JURISDICTIONS.
THE WITNESS: NO, I DO HAVE EVIDENCE THAT. I MEAN IM
FAMILIAR WITH THE LITERATURE THAT INDICATES CASE PRESSURE IS
IMPORTANT IN DETERMINING WHEN FLEA BARGAINS ARE ENTERED.
I DON“T HAVE ANY EMPIRICAL EVIDENCE ON WHAT THE CASE
PRESSURE IS IN JURISDICTIONS WITHIN THE STATE OF GEORGIA. I
MEAN I HAVE SEEN THE WORK OF OTHERS WHO HAVE INTERVIEWED
PROSECUTORS IN THIS JURISDICTION AND THAT IS A PROMINENTLY
FEATURED CONSIDERATION IN THEIR DETERMINATIONS AS TO WHETHER TO
ADVANCE CASES TO PENALTY ‘TRIAL. WHAT THE COST OF IT WOULD BE.
THE PRESSURES OF RESOURCES AND TIME. BUT I DON'T HAVE AN
INDEPENDENT MEASURE ON THAT. YOUR HONOR.
THE COURT: A FELLOW NAME HEMANN, I THINK IT’S
H-E-M-A-N-N-, WROTE A BOOK CALLED “PLEA BARGAINING." ARE YOU
FAMILIAR WITH IT? THE SIMPLE TITLE IS "PLEA BARGAINING."
rn)
11460
BALDUS ~ CROSS
THE WITNESS: I DON'T KNOW THAT I“M FAMILIAR WITH THAT
ONE, YOUR HONOR. NO.
THE COURT: HIS THESIS IS THAT PLEA BARGAINING IS NOT
EXFLAINED BY THE CASE PRESSURE METHOD, BUT INSTEAD IS EXPLAINED
BY AN ALTERNATIVE MECHANISM FOR DELIVERING OF CERTAINTY.
THE BASIC NOTION. I GUESS, IS NOBODY REALLY WANTS TO GO
THROUGH THE ORDEAL OF DUE PROCESS IF THEY KNOW WHERE THEYRE
GOING TO COME OUT ON THE OTHER END OF PIPE.
I DON’T PRETEND TO TAKE SIDES BETWEEN THE TWO THEORIES.
BUT IN HIS WRITINGS, HE SUGGESTS THAT, FOR INSTANCE. LOWER
SOCIO-ECONOMIC PEOPLE TEND TO PLEA BARGAIN MORE READILY BECAUSE
WHAT THEY WANT IS CERTAINTY AND NOT VINDICATION OF THEIR GOOD
NAME OR AVOIDANCE OF CONSEQUENCES WHICH HIGHER SOCIO-ECONOMIC
PEOPLE MIGHT WISH TO AVOID, INCLUDING JAIL.
THATS ONE OBSERVATION.
I THINK ANOTHER OBSERVATION IS THAT YOUNGER LAWYERS,
INEXPERIENCED LAWYERS, TENDED NOT TO PLEA BARGAIN, BECAUSE THEY
DID NOT HAVE, IN HIS VIEW A REALISTIC ATTITUDE ABOUT WHAT WAS
GOING ON WITH THEIR CLIENT, AND WHAT WAS LIKELY TO HAFFEN IN THE
SYSTEM, ALL THAT SORT OF THING.
IM WONDERING IF YOU DID ANY SORTS ON THE FLEA
BARGAINING THAT WOULD RELATE TO SOCIO-ECONOMIC STATUS TO
AFPOINTED, RETAINED LAWYER, ANYTHING LIKE THAT THAT WOULD SHED
ANY LIGHT ON WHETHER THOSE FACTORS MIGHT EXPLAIN, FOR EXAMPLE.
THE PLEA BARGAINING IN EARLY DISPOSITION DISPARITIES THAT YOU
11461
BALDUS - CROSS
NOTICED.
THE WITNESS: YOUR HONOR. WE HAD A MEASURE OF
SOCIO-ECONOMIC STATUS THAT WAS INCLUDED IN OUR ANALYSIS, AND IT
WAS BASED ON FOUR FACTORS. PRINCIPAL ONE OF WHICH WAS WHETHER OR
NOT YOU HAD AN APPOINTED LAWYER.
THE QUESTION OF DEVELOPING A MEASURE FOR SOCIO-ECONOMIC
STATUS WAS COMPLICATED BY THE REASON I MENTIONED EARLIER ABOUT
1.02. WE HAD A LOT OF MISSING INFORMATION ON THE QUESTION OF
SOCIO-ECONOMIC STATUS.
AND THIS IS A BIT OF INFORMATION THAT YOU CAN'T JUST
RELY ON WHAT”S IN THE FILE. IN MY OPINION. THIS IS ILLEGAL OR
SUSPECT IN ANY EVENT, INFORMATION. AND THE FROSECUTORS AND
FECPLE WHO COLLECT THE INFORMATION, DO NOT HAVE A STRONG
INCENTIVE TO PUT THAT INFORMATION IN THE FILE. S50 WE HAD TQ BE
SURE WE HAD THE CORRECT INFORMATION ON THOSE VARIABLES.
30 THE WAY WE PROCEEDED WAS TO TAKE THE FOUR DIFFERENT
VARIABLES THAT WE HAD, THAT WERE STRONG INDICATIONS OF LOW
SOCIO-ECONOMIC STATUS AND THE FIRST ONE WAS WHETHER OR NOT THE
OFFENDER HAD APPOINTED COUNSEL AND WE HAD THAT INFORMATION ON
SOME 60 OR 70 PERCENT OF THE CASES. S0 THAT INFORMATION AS
TO APPOINTED COUNSEL WAS EMBODIED IN THIS MEASURE.
IN ADDITION, THE DEPARTMENT OF CORRECTIONS HAD A
VARIABLE CALLED SOCIO-ECONOMIC STATUS THAT WENT DIRECTLY TO THE
AUESTION, WASNT AMBIGUOUS AT ALL. THE PROBLEM WAS WE HAD A LOT
OF MISSING INFORMATION.
wh a.
[\)
1162
BALDUS - CROSS
AND FRANKLY I BELIEVE THE FACT OF HAVING APPOINTED
COUNSEL WAS PROBABLY A BETTER MEASURE OF INDIGENCY THAN THE
ESTIMATE OF SOMEBODY IN THE DEPARTMENT OF CORRECTIONS.
THE COURT: BY THE TIME THEY. GET TO PRISON, MOST FOLKS
ARE INDIGENT, EVEN IF THEY DIDN'T START OUT THAT WAY.
THE WITNESS: s0 I TOOK THAT AS A SECOND OPTION. IF
YOU DIDN‘T GET AN ANSWER ON WHETHER THERE WAS APPOINTED COUNSEL.
THEN WE WOULD GO TO THIS QUESTION.
STILL, THERE WERE A FEW PEOPLE THAT WE DID NOT HAVE AN
ANSWER, BECAUSE OF THE MISSING VALUES.
SO WE WOULD ASK THE NEXT QUESTIONS. DID THE PERSON HAVE
AN UNSKILLED .JOB, ANOTHER PROXY FOR LOW SOCIO-ECONOMIC STATUS.
AND THEN FOR THE FEW REMAINING WHERE WE DIDNT HAVE AN
ANSWER, WE ADDRESSED THE QUESTION. DID THE PERSON HAVE AN STH
GRADE DR LESS EDUCATION, ON THE ASSUMPTION MOST PEOPLE IN
TODAYS SOCIETY IN THAT CAPACITY ARE PROBABLY OF LOW
SOCIO-ECONOMIC STATUS.
AND ON THE BASIS OF THAT COMBINATION OF FACTORS. WE HAD
A VARIABLE WHICH WE INCLUDED IN ALL OF THESE ANALYSES, WHEN WE
WOULD DO THE LARGE SCALE REGRESSION RUNS. AND THAT DID NOT
EXFLAIN AWAY THE RACIAL EFFECT. THAT’S WHAT I CAN TESTIFY, YOUR
HONOR.
THE COURTS IT IN COMBINATION WITH ALL THE OTHER
VARIATIONS WHICH WERE CUTTING ONE WAY OR THE OTHER. BUT I WAS
SPECIFICALLY TRYING TQ TQ SEE IF YOU TESTED SPECIFICALLY AT THE
Py
2
3
4
=
&
7
8
>
10
1163
BALDUS - CROSS
PLEA BARGAINING AND EARLIER DISPOSITION LEVEL, TO SEE WHETHER OR
NOT EITHER SOCIO-ECONOMIC STATUS, CASE PRESSURE. APPOINTED
COUNSEL, ANY OF THOSE THINGS EXPLAINED IT.
THE WITNESS: WELL, THE TWO VARIABLES THAT I HAVE. YOUR
HONOR, RELATE TO THE APPOINTED COUNSEL AND THE SOCIO-ECONOMIC
MEASURE. WHEN YOU TOOK ALL THE OTHER FACTORS INTQ ACCOUNT, THEY
DIDNT TEND TO HAVE MUCH EFFECT. MY RECOLLECTION.
THE COURT: YOU DIDN‘T DO IT BY ITSELF, JUST TO SEE
WHAT WOULD HAPPEN?
THE WITNESS: THATS RIGHT. THE ONLY PLACE I DID IT
WAS WITH RESPECT TO THE OVERALL EFFECTS OF THE SYSTEM, AND WHEN
YOU LOOK AT THE OVERALL EFFECTS OF THE SYSTEM AND YOU ONLY
CONTROL FOR THAT SOCIO-ECONOMIC STATUS. YOU DO SEE DEATH
SENTENCING RATES THAT ARE SUBSTANTIALLY HIGHER FOR THE LOWER
SOCIO-ECONOMIC STATUS PEOPLE. BUT AGAIN, THAT ISN“T CONTROLLING
FOR ALL SORTS OF OTHER BACKGROUND FACTORS.
THE COURT: I WOULD IMAGINE. AND IM CURIOUS TO SEE IF
YOU DID IT, AS TO WHETHER OR NOT AS BETWEEN A PLEA TO VOLUNTARY
MANSLAUGHTER OR PLEA TO MURDER THAT THERE WOULD BE THAT
CORRELATION AND PROBABLY BE THAT SORT OF CORRELATION ON PLEAS.
THE WITNESS: I CAN’T ANSWER THAT QUESTION, YOUR HONOR.
I DID NOT DQ THAT ANALYSIS.
BY MS. WESTMORELAND:
@. ONE QUESTION, PROFESSOR BALDUS. RELATING TO SOMETHING YOU
STATED JUST A FEW MINUTES AGO, I BELIEVE.
11464
BALDUS - CROSS
pb
IN THE CHARGING AND SENTENCING STUDY. ON THE
2 AUESTIONNAIRE, IF I RECALL. THERE WERE, I BELIEVE. FOUR OPTIONS
3 FOR THE CODER. WOULD THAT BE CORRECT?
4 A. ON THE AVERAGE. THAT-S RIGHT. THE METHOD OF CODING VARIED
3S SOMEWHAT FROM PLACE TO PLACE, BUT THAT WAS THE MAIN METHOD.
a & @. BUT THE 1, 2, BLANK OR "U" FOR UNKNOWN?
7 A. YES.
8 @R. NOT ON ALL QUESTIONS. BUT -——
? A. RIGHT. ON THE BULK OF THE QUESTIONS. THAT WAS THE METHOD.
10 QR. WHEN THOSE ANSWERS WERE SUBSEQUENTLY CODED FOR COMPUTER
11 USAGE, DID YOU COME OUT WITH JUST TWO VALUES THEN, ESSENTIALLY
12 EVERYTHING CAME OUT TO EITHER BE PRESENT OR NOT PRESENT IN THOSE
13 SITUATIONS?
14 A.. THAT’S RIGHT. FOR THE REASON THAT I STATED. IF IT WAS
15 UNKNOWN, IF A PERSON IS IN A STATE OF IGNORANCE. THEY ARE IN THE
146 SAME STATE AS BELIEVING THAT SOMETHING DOESNT EXIST. YOU HAVE
17 TO BELIEVE THAT AN AGGRAVATOR OR A MITIGATOR EXISTS IN A CASE
18 BEFORE IT-S GOING TO INFLUENCE YQU.
- 19 @. I BELIEVE YOU MENTIONED THIS MORNING A FACTOR ANALYSIS THAT
20 YOU DID. WAS THAT CORRECT?
21 A. YES.
22 RB. WAS THAT DONE?
23 A. UH HUH.
24 QR. WAS THE PURPOSE IN DOING THAT FACTOR ANALYSIS TO PRODUCE
25 UNCORRELATED VARIABLES?
1165
BALDUS - CROSS
1 A. WELL, TO REDUCE THE AMOUNT OF MULTI-COLLINEARITY WHICH IS
2 THE CORRELATION TERMINOLOGY BETWEEN THE VARIABLES, YES. TO GET
A MORE PARSIMONIOUS MODEL, AND WE START WITH A VERY LARGE MODEL
AND WE WANTED TO TRY TO REDUCE IT DOWN IN STAGES TQ SEE IF JUST
THE SHEER MAGNITUDE OF THE ANALYSIS MIGHT BE BIASING THE
3
4
S
* é CO~-EFFICIENTS IN ONE WAY OR ANOTHER.
7 SO THE METHOD OF FACTOR ANALYSIS WAS ONE WAY TO CUT THE
8 |FILE IN HALF. HOPEFULLY RETAINING THE SAME INFORMATION IN IT.
9 THAT WAS MORE OF A CONCERN, THAT WAS MORE OF THE OBJECTIVE THAN
10 REDUCING THE MULTI-COLLINEARITY, ALTHOUGH THAT IS ONE OF THE
11 ADVANTAGES OF FACTOR ANALYSIS IS THAT IT DOES JUST THAT. BUT IT
12 DOESNT ELIMINATE IT.
13 @. IN RUNNING YOUR LARGE-SCALE REGRESSIONS, I BELIEVE WAS THE
14 TERMINOLOGY YOU’VE USED FREQUENTLY FOR SOME OF THE REGRESSIONS
15 YOU“VE USED?
14 A. YES.
17 @. AND THIS MAY HAVE BEEN YOUR PREVIOUS TESTIMONY. I“M NOT
18 CERTAIN?
19 A. UH HUH.
20 R. IS IT NOT REALLY INEVITABLE THAT SOME OF THE SIGNS OF THE
3 CO-EFFICIENTS WILL RUN IN A WRONG OR AN UNEXPECTED DIRECTION FOR
22 THAT PARTICULAR VARIABLE?
23 A. YES, BECAUSE OF THE AMOUNT OF MULTI-COLLINEARITY CORRELATION
24 BETWEEN ALL THESE VARIABLES. SOME OF THEM ARE GOING TO RUN IN
25 PECULIAR DIRECTIONS AND MAKE IT VERY HARD TO INTERPRET THE
p
r
vv
O
O
O
N
C
W
N
10
11
11466
BALDUS - CROSS
RESULTS.
THE COURT? MULTI- WHAT?
THE WITNESS: IT’S CALLED MULTI-COLLINEARITY. YOUR
HONOR. THAT MEANS THAT THE VARIABLES THAT ARE IN THE ANALYSIS
ARE THEMSELVES CORRELATED WITH ONE ANOTHER. THAT IS, IF YOU
HAVE, IF YOU INCLUDE THE TEN STATUTORY AGGRAVATING FACTORS IN A
SINGLE ANALYSIS TQ TRY TO ESTIMATE THE ‘IMPACT QF EACH, THOSE
RESULTS ARE GOING TO BE AFFECTED BY THE FACT THAT MANY OF THE
STATUTORY AGGRAVATING FACTORS ARE THEMSELVES CORRELATED WITH ONE
ANOTHER.
WE SAW THE OTHER DAY THAT BS AND B10 ARE HIGHLY
CORRELATED. B2 AND B4 ARE HIGHLY CORRELATED. AND WHEN YOU
INCLUDE A VERY LARGE-SCALE ANALYSIS, THIS MULTI-COLLINEARITY OR
INTER-CORRELATION CHARACTERISTIC OF THESE DATA IS GOING TO
AFFECT CO-EFFICIENTS. AND MAKE THEM DIFFICULT TO INTERPRET.
THE COURT: I WOULD LIKE TO TELL COUNSEL THAT I DONT
UNDERSTAND WHY AND BEFORE THIS CASE IS OVER, I WOULD LIKE
SOMEBODY TO EXPLAIN THAT TO ME, BUT I DON’T KNOW WHETHER THIS IS
THE WITNESS OR WHETHER WE ASK THE OTHER PROFESSOR.
THE WITNESS: WELL, I COULD JUST TELL YOU IN A NUTSHELL
IF YOU WOULD LIKE ME TO. YOUR HONOR.
THE COURT: I WAS TRYING TQ BE DELICATE. IF YOU KNOW.
PLEASE TELL ME. IF YOU DON’T, ——
THE WITNESS: AS I UNDERSTAND IT. YOUR HONOR, WHAT THE
MULTI-COLLINEARITY DOES IS IT REDUCES THE PRECISION OF THE
1
oO
YW
oO
N
°
4
pp
O
N
(
=
T
E
=
a
in
b
h
o
N
[o
y o~
37
18
11467
BALDUS - CROSS
STANDARDAIRE ESTIMATES WITH RESPECT TO THE VARIABLES THAT ARE
INTERRELATED TO ONE ANOTHER AND TENDS TO REDUCE THE STATISTICAL
SIGNIFICANCE OF THE VARIABLES THAT ARE HIGHLY CORRELATED TO ONE
ANOTHER.
IT ALSO MAKES IT DIFFICULT TO DETERMINE WHICH FACTOR IS
PRODUCING THE EFFECT IN AN ANALYSIS. IF YOU HAVE TWO OR THREE
VARIABLES IN AN ANALYSIS, AND THEY ARE CORRELATED WITH ONE
ANOTHER. IT“S HARD TO TELL WHICH ONE YOU SHOULD ATTRIBUTE EFFECT
TO WHEN THEY BOTH HAVE CO-EFFICIENTS OF SOME GIVEN SIZE.
THOSE ARE THE, THATS WHAT MAKES IT DIFFICULT TO
INTERPRET THE RESULTS, THE EFFECT IT HAS ON IT, IT BIASES OR
MAKES LESS RELIABLE THE STATISTICAL ESTIMATES, AND ALSO MAKES IT
HARD TO INTERPRET THE CO-EFFICIENTS IN TERMS OF HOW MUCH IMPACT
THOSE CO-EFFICIENTS ARE HAVING.
THATS WHY I MENTIONED THE OTHER DAY, WHEN WE HAD A
LIST OF VARIABLES, AND WE HAD THE RACE OF THE VICTIM AND
RACE OF THE DEFENDANT IN RANK ORDER. I SAID YOU HAVE TO BE
CAUTIOUS ABOUT MAKING ANY ESTIMATE AS TO HOW BIG AND RELATIVE
EFFECT ANY ONE IS HAVING VIS-A-VIS THE OTHER. BECAUSE OF THIS
PROPERTY OF MULTI-COLLINEARITY. IT MAKES IT DIFFICULT TO MAKE
THAT SORT OF ESTIMATE.
SO THAT I3 MY UNDERSTANDING OF IT, OF WHAT THE
PRINCIPAL PROBLEMS THAT ARE PRODUCED BY THIS MULTI-COLLINEARITY.
SO THAT WHAT YOU DO KNOW, THOUGH. WHEN YOU HAVE A
SERIES OF VARIABLES THAT ARE CORRELATED TO ONE ANOTHER AND ONE
11463
BALDUS - CROSS
[
A
OF THEM EMERGES AS VERY STRONG AND STATISTICALLY SIGNIFICANT.
YOU CAN BE FAIRLY CONFIDENT THAT THAT VARIABLE. GIVEN ALL THE
COMPETITION IT HAS, IS REFLECTING A REAL EFFECT IN THE SYSTEM.
SO THAT IT’S, WHEN A CO-EFFICIENT LOSES ITS SIGNIFICANCE AND
LOSES ITS EFFECT, YOU DON’T KNOW WHETHER THAT’S A RESULT OF
STATISTICAL INTERACTION GOING ON, USED IN AN INFORMAL SENSE. OR
WHETHER THAT MEANS THAT THAT VARIABLE JUST DOESN’T HAVE ANY
EFFECT.
C
R
E
R
E
R
E
e
B
E
E
T
Li
gh
BUT YOU CAN HAVE A LOT OF CONFIDENCE. I BELIEVE,
r
OQ
CORRECT, I“M CORRECT IN THIS, THAT WHEN YOU DO HAVE THESE
oe
pt
e VARIABLES THAT ARE INTERRELATED TO ONE ANOTHER. AND ONE OR TWO
12 OF THEM STAY IN THERE AND SHOW A STRONG STATISTICALLY
13 SIGNIFICANT EFFECT, GIVEN THE FACT THAT MULTI-COLLINEARITY TENDS
14 |TO SUPPRESS THE STATISTICAL EFFECT OF THESE VARIABLES, THEN THAT
15 GIVES YOU MORE CONFIDENCE THAT THOSE VARIABLES ARE IMPORTANT
16 IN THE SYSTEM. BUT IT MAKES IT HARD TO MAKE RELATIVE
17 [COMPARISONS WITH ONE ANOTHER. THAT’S WHAT MAKES IT TRICKY.
18 BY MS. WESTMORELAND:
» 19 R. IN RUNNING YOUR REGRESSIONS, ALONG THIS SAME LINE OF
20 THINKING, DID YOU END UP WITH VARIABLES WITH SIGNS IN THE WRONG
21 DIRECTION, WITH STATISTICAL SIGNIFICANCE?
22 A. YES. YES, WE DID.
23 Rl. WOULD THAT BE CAUSED BY THE SAME CORRELATION PROBLEM?
24 A. YES. IT COULD BE. IT COULD REFLECT THAT.
235 IT COULD, COULD REFLECT CHANCE RESULTS. THERE’S A
116%
BALDUS - CROSS
PY
CERTAIN AMOUNT OF CHANCE IN ANY REGRESSION, THAT YOULL GET A
CO~-EFFICIENT THAT IS STATISTICALLY SIGNIFICANT, JUST AS A CHANCE
PHENOMENON.
ALSO IT COULD REFLECT THE FACT THAT THE SYSTEM DOESN’T
OPERATE ACCORDING TO THE WAY THE LAW THINKS IT SHOULD. THAT ON
SOME VARIABLES THE SYSTEM COULD WELL JUST BE OPERATING IN A
PERVERSE WAY, THAT IS, IF YOU HAVE A MITIGATING FACTOR AND IT
0
~~
Oo
44
+»
L
N
TENDS TO PRODUCE, IT SUGGESTS THAT THERES AN AGGRAVATING EFFECT
0 ASSOCIATED WITH THAT FACTOR. IT COULD BE THAT THAT’S A CHANCE
-
QO
RESULT AS IT RELATES TO THE INTER-CORRELATION OF THE VARIABLES.
Fo
y
[W
Y BUT IT ALSO COULD RESULT FROM THE FACT THAT THE PEOFLE
12 THAT ARE RUNNING THE SYSTEM DONT EVALUATE THAT FACTOR IN THE
13 WAY THAT WE THINK THEY SHOULD FROM A, YOU KNOW, A CONCEPTUAL OF
14 THE LEGAL STANDPOINT.
13 SO THAT IN FACT WAS THE PRINCIPAL CONCERN THAT LED US
16 TO PRODUCE THESE PARED DOWN MODELS THAT CLEANED ALL THOSE
37 VARIABLES OUT, BECAUSE THEY RAISED LEGAL CONCEPTUAL PROBLEMS.
13 THE IDEA OF HAVING VARIABLES IN AN ANALYSIS THAT HAVE A
w 12 PERVERSE AND LOOK LIKE A PERVERSE EFFECT IS TROUBLESOME.
20 AND I MENTIONED THAT GODFREY V. GEORGIA SUGGESTED THAT
21 VARIABLES THAT ARE USED, FACTORS THAT ARE USED TO DETERMINE
22 COMPARABLE CASES SHOULD BE RATIONAL. IF A FACTOR IS HAVING WHAT
23 LOOKS LIKE A PERVERSE EFFECT, IT WAS MY JUDGMENT THAT WE SHOULD
24 TRY AN AFPROACH THAT PRUNES THOSE OUT AND LIMIT IT JUST TO THE
25 ONES THAT HAVE A RATIONAL APPARENT EFFECT IN THE SYSTEM. AND
8)
O
o
t
y
N
a
g
y
R
T
td
1170
BALDUS - CROSS
THAT WAS THE WHOLE THEORY UNDERLYING ONE OF THESE ALTERNATIVE
APPROACHES THAT WE USED:
CONTROLLING FOR BACKGROUND FACTORS. TO DEAL WITH THIS
PROBLEM THAT YOU’RE REFERRING TO.
BY MS. WESTMORELAND:
&. OUT OF YOUR, I BELIEVE IT WAS 128 DEATH PENALTY CASES IN THE
CHARGING AND SENTENCING STUDY. IS THAT CORRECT?
A. YES.
@. DO YOU HAVE AN ESTIMATE OF HOW MANY OF THOSE CASES WERE WHITE
VICTIM CASES?
A. I CAN TELL YOU, I THINK, =-— OH, YES, I THINK IT WAS 83
PERCENT, 1 BELIEVE. WERE WHITE VICTIM CASES. THAT WOULD BE, I
THINK THERE WERE 18, MAYBE 20 BLACK VICTIM CASES. GREAT BULK
WERE WHITE VICTIM CASES.
Q. AROUND 108, 110, SOMEWHERE IN THAT NEIGHBORHOOD?
A. ILL SAY ILL GIVE YOU YOU AN UNAMBIGUOUS ANSWER AT 85
PERCENT, WHATEVER THAT IS.
@. CONSIDERING THAT HIGH PERCENTAGE OF WHITE VICTIM CASES IN
THE DEATH SENTENCE CASES, WOULD YOU THEN EXPECT THAT IN YOUR
SIMPLE REGRESSIONS THE CO-EFFICIENT OF YOUR, I GUESS IT’S A
RECODED RACE OF VICTIM VARIABLE YOU GOT, WOULD BE POSITIVE,
CONSIDERING THAT HIGH PROPORTION OF WHITE VICTIM CASES?
A. I’“M AFRAID I -- COULD YOU RESTATE THAT QUESTION, PLEASE?
THE COURT: I DON’T UNDERSTAND THAT QUESTION. AT ALL,
MS. WESTMORELAND.
1171
BALDUS — CROSS
BY MS. WESTMORELAND:
@. DUE TO THE, AND MAYBE I“M NOT EXACTLY SURE HOW TO EXPRESS
IT. MAY BE THE PROBLEM, THE HIGH PROPORTION OF WHITE VICTIM
CASES. YOU HAVE A. LET ME START AT THE BEGINNING.
YOU HAVE A RECODED RACE OF VICTIM VARIABLE INCLUDED IN
YOUR ANALYSIS. I“M NOT SURE OF THE SYMBOL FOR IT, WHI, SOMETHING
TO THAT EFFECT. :
A. YES, WHVICRC IS THE VARIABLE.
@. THATS THE ONE I‘M TRYING TO REFER TO, YES.
A. RIGHT.
@. GIVEN THE HIGH PERCENTAGE, HIGH PROPORTION OF WHITE VICTIM
CASES AMONG THE DEATH SENTENCE CASES, WOULDN‘T IT BE LOGICAL TO
EXPECT IN A SIMPLE REGRESSION THAT THE CO-EFFICIENT OF THAT
VARIABLE WOULD HAVE A POSITIVE SIGN?
A. YOU MEAN BY A SIMPLE VARIABLE ONE WITH ONLY. A BI-VARIABLE
REGRESSION WITH ONLY THE RACE OF VICTIM IN THE ANALYSIS. YOU
MEAN.
@. YES?
A. YES. IF YOU HAVE NO BACKGROUND CONTROLS THATS PLAIN.
THERE IS A STRONG CORRELATION, UNADJUSTED CORRELATION BETWEEN
THE RACE OF THE VICTIM AND DEATH SENTENCING OUTCOME.
@. WOULD IT THEN BE POSSIBLE IN USING A LARGER REGRESSION FOR,
OR AT LEAST WOULD IT BE A LOGICAL ASSUMPTION THAT IN USING THE
LARGER REGRESSION, THE REGRESSION COULD ELIMINATE EFFECTS OF
BACKGROUND VARIABLES AND WOULD STILL HAVE A POSITIVE
23172
BALDUS - CROSS
.
CO-EFFICIENT FOR THAT RECODED VARIABLE?
2 A. BY INTRODUCING THE BACKGROUND CONTROLS. WHAT YOU DO IS YOU
3 IDENTIFY THE RACE EFFECTS AMONG COMPARABLE GROUPS OF CASES AS
4 DEFINED BY THE BACKGROUND FACTORS SO THAT YOU ARE GOING TO SEE
3 THE RACE EFFECTS IF AMONG SIMILARLY SITUATED CASES WITH RESPECT
i é TO THOSE BACKGROUND CONTROLS THERE ARE DIFFERENCES IN THE
7 TREATMENT OF BLACK AND WHITE VICTIM CASES. YOULL SEE THE RACE
8 EFFECTS.
9 FRANKLY, I WAS AMAZED WHEN WE INCLUDED ALL THESE
10 VARIABLES IN THESE ANALYSES THAT WE STILL HAD RACE EFFECTS LEFT.
31 I NEVER EXPECTED TQ SEE SUCH A RESULT.
12 @. LET ME MOVE INTO THE FULTON COUNTY STUDY AT THIS TIME.
13 THE COURT: LET’S TAKE A SHORT BREAK BEFORE YOU DO
14 THAT, ABOUT TEN MINUTES.
15 ---
16 (RECESS TAKEN.)
17 ---
18 DAVID C. BALDUS,
Re 19 |BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
20 TESTIFIED FURTHER AS FOLLOWS:
21 CROSS-EXAMINATION (CONT D)
22 BY MS. WESTMORELAND:
23 . PROFESSOR BALDUS, BEFORE WE PROCEED INTO THE FULTON COUNTY,
24 2IRy
235 MS. WESTMORELAND: -- AND YOUR HONOR, AS WELL. I HAVE
L
o
v
i
e
W
I
R
Y
BR
E
S
R
C
B
E
E
B
E
e
h
U
E
Pr
y oO
fo
b
Fe
y
17
18
19
1173
BALDUS - CROSS
CERTAIN EXHIBITS THAT I BELIEVE ARE ALL INFORMATION WE HAD
OBTAINED FROM PROFESSOR BALDUS AND JUST FOR THE PURPOSES OF THE
RECORD, AT THIS POINT I WOULD LIKE TO HAVE HIM MERELY GO THROUGH
AND IDENTIFY THEM, NOT TO BE INTRODUCED AT THIS TIME, BUT
PERHAPS TQ BE REFERRED TO BY DOCTOR KATZ WHEN HE TESTIFIES FOR
THE STATE. BUT AS THEY WERE EXHIBITS OBTAINED FROM PROFESSOR
BALDUS, I WOULD LIKE TO HAVE HIM IDENTIFY THEM FOR THE RECORD.
THE COURT: ALL RIGHT.
BY MS. WESTMORELAND:
@. PROFESSOR BALDUS. LET ME HAND YOU WHAT HAVE BEEN MARKED THUS
FAR, AND I“LL ASK YOU IF YOU CAN IDENTIFY THESE.
MS. WESTMORELAND: YOUR HONOR, I ALSO HAVE A COPY FOR
THE COURT AS WELL.
BY MS. WESTMORELAND:
R. IF YOULL BEGIN WITH WHAT HAS BEEN MARKED AS RESPONDENTS
EXHIBIT 5 AND I ASK YOU IF YOU CAN IDENTIFY THAT DOCUMENT FOR
Us. FLEASE?
A. YES. THIS IS A DOCUMENT. TABLE 5 FROM OUR REPORT, AND IT
LISTS APPROXIMATELY ONE HUNDRED VARIABLES. NON-RACIAL VARIABLES,
THAT HAVE A STATISTICALLY SIGNIFICANT ASSOCIATION WITH THE DEATH
SENTENCING RESULT, AND THE VICTIMS RACE.
R. WHAT TYPE OF PROCEDURE WAS UTILIZED IN PREPARING THIS TABLE,
PROFESSOR BALDUS?
A. COLUMN B REFLECTS FOR EACH OF THESE VARIAELES A BI-VARIATE
REGRESSION. THERES A CO-EFFICIENT HERE WHICH SHOWS THE
~.
5
A
Hp
0
PO
9
10
1174
BALDUS - CROSS
UNADJUSTED IMPACT. STATISTICALLY, FOR EACH OF THESE VARIABLES ON
THE DEATH SENTENCING RATE.
COLUMN C PRESENTS THE CORRELATION BETWEEN THE VARIABLE.
WHICH IS IN COLUMN A, NON-RACIAL VARIABLE, WITH THE VARIABLES
FOR THE VICTIM’S RACE.
AND INCLUDED IN THIS TABLE ARE ONLY VARIABLES THAT SHOW
A STATISTICALLY SIGNIFICANT RELATIONSHIP BETWEEN THE VICTIM’S
RACE. AND THE NON-RACIAL FACTOR, AND THE NON-RACIAL FACTOR ALSO
HAS A STATISTICALLY SIGNIFICANT RELATIONSHIP WITH THE DEATH
SENTENCING RESULT.
THESE ARE THE VARIABLES THAT WOULD. ONE WOULD EXPECT
WOULD BE THE BEST CANDIDATES TO EXPLAIN AWAY RACE OF VICTIM
EFFECT IN THE ANALYSIS.
MS. WESTMORELAND: THAT SHOULD BE NUMBER 11, I BELIEVE.
EXCUSE ME FOR A MOMENT, YOUR HONOR.
THE CLERK: LET ME GIVE YOU THIS BACK. I HAVE A COPY.
MS. WESTMORELAND: 1 APOLOGIZE TO THE COURT. MY
NUMBERING SYSTEM MAY HAVE GOTTEN SLIGHTLY ASKEW. WELL TRY TO
CLARIFY THE EXHIBITS AS WE GO THROUGH.
BY MS. WESTMORELAND:
R. PROFESSOR BALDUS, LET ME DO THESE ONE AT A TIME SO WE DONT
GET TOO CONFUSED.
A. CERTAINLY.
2. IN THE DOCUMENTS I HAVE JUST HANDED YOU, COULD YOU PLEASE
TAKE THEM ONE BY ONE AND IDENTIFY BY EXHIBIT NUMBER AND THEN
1175
BALDUES - CROSS
1 IDENTIFY THE DOCUMENT ITSELF. PLEASE?
2 A. YES. THIS WOULD BE RESPONDENT”S EXHIBIT & WHICH LISTS PART
3 OF THE RESULTS OF THE FACTOR ANALYSIS THAT I CONDUCTED TO LIMIT
4 THE SIZE OF THE FILE. AND WHAT IT SHOWS. I BELIEVE, ARE THE
3 CORRELATION CO-EFFICIENTS BETWEEN THE VARIOUS FACTORS. AND EACH
® é OF THE UNDERLYING VARIABLES IN THE FILE THAT WERE USED TO CREATE
7 THE FACTORS.
8 THIS ONE --
2? @. DOES THAT DOCUMENT READ FACTOR ANALYSIS OF THE AGGRAVATING
10 VARIABLES?
11 A. YES. YES, IT DOES. I THINK I REPRESENTED IT CORRECTLY.
12 |@. COULD YOU REFER TO THE NEXT DOCUMENT AND IDENTIFY IT,
13 PLEASE?
14 A. YES. RESPONDENT’S EXHIBIT 7 IS LABELED "FACTOR ANALYSIS OF
13 THE EVIDENTIARY VARIABLES." AND IT SHOWS THE SIMPLE CORRELATION
16 BETWEEN EACH OF THE FACTORS ESTIMATED IN THE FACTOR ANALYSIS
17 WITH EACH OF THE UNDERLYING EVIDENTIARY FACTORS.
13 @. COULD YOU IDENTIFY THE NEXT EXHIBIT?
w 19 A. YES. CERTAINLY. THE NEXT ONE DOES THE SAME THING WITH
20 RESPECT TO THE MITIGATING FACTORS USED IN THE ANALYSIS TO CREATE
z1 [FACTORS FOR MITIGATING CIRCUMSTANCES. IT SHOWS THE ——
22 QQ. THAT WOULD BE RESPONDENTS EXHIBIT NUMBER 87?
23 A. YES. I’M SORRY, IT WOULD BE RESPONDENT-S EXHIBIT NUMBER 8,
24 CALLED FACTOR ANALYSIS FOR THE MITIGATING VARIABLES.
23 : AND FINALLY. OR NEXT, RESPONDENT“S EXHIBIT 9? SHOWS THE
Pe
10
11
Uo
MO
O
N
6
R
e
W
N
1176
BALDUS ~ CROSS
CORRELATIONS IN A DOCUMENT ENTITLED "FACTOR ANALYSIS OF THE
AGCRAVATING VARIABLES."
@. DOES THAT HAVE THE WORD "JURY" HANDWRITTEN ON THE FRONT OF
37
A. RIGHT. THIS INCLUDES THE FACTORS THAT WERE ESTIMATED
FOR THE JURY DECISION MAKING POINT. IT SHOWS THE CORRELATIONS
BETWEEN THE UNDERLYING VARIABLES AND THE FACTORS THAT WERE
ESTIMATED FOR THE AGGRAVATING FACTORS IN THE JURY ANALYSIS.
RESPONDENT-S EXHIBIT 10 DOES THE SAME THING FOR THE
DECISION MAKING OF THE JURIES REPORTING THE SIMPLE CORRELATIONS
BETWEEN THE FACTORS ESTIMATED IN THE FACTOR ANALYSIS WITH THE
UNDERLYING FACTORS USED TO ESTIMATE THOSE FACTORS. IM SORRY,
THE UNDERLYING VARIABLES USED TO ESTIMATE THE FACTORS.
MR. AND THIS DOCUMENT READS "FACTOR ANALYSIS OF THE MITIGATING
VARIABLES," AND THEN IT HAS "ROTATED FACTOR PATTERN?"
A. YES.
@. THATS RESPONDENTS EXHIBIT 107
A. THAT'S RIGHT.
MS. WESTMORELAND: I ASK THE COURT REPORTER TO MARK THIS
DOCUMENT AS RESPONDENTS EXHIBIT 10A.
YOUR HONOR. THIS MAY HAVE BEEN INADVERTENTLY LABELED 11
IN YOUR COPY.
BY MS. WESTMORELAND:
Re CAN YOU IDENTIFY THIS DOCUMENT. RESPONDENTS EXHIBIT 10A,
PLEASE?
wv
o
N
R
d
O
N
-
a
o
Ty
Pr
y
12
313
1177
BALDUS - CROSS
A. CERTAINLY. IT“S RESPONDENT-S EXHIBIT 10A, AND IT IS THE
PRESENTATION UNDER THE HEADING OF "FACTOR ANALYSIS
OF THE EVIDENTIARY VARIABLES: JURY."
OF THE SIMPLE COERCION CO-EFFICIENTS BETWEEN THE FACTORS
ESTIMATED FOR THE EVIDENTIARY AND DIMENSIONS OF THE CASES. AND
PRESENTS THE CORRELATION CO-EFFICIENTS BETWEEN THOSE FACTORS AND
UNDERLYING EVIDENTIARY VARIABLES.
@. PROFESSOR BALDUS, I“VE HANDED YOU ANOTHER SERIES OF
DOCUMENTS.
WOULD YOU PLEASE IDENTIFY THESE AS WELL?
1A. RESPONDENT“S EXHIBIT, I‘M SORRY, --
|@. I BELIEVE IT BEGINS WITH RESPONDENTS EXHIBIT NUMBER 117
A. YES. RESPONDENTS EXHIBIT NUMBER 11 PRESENTS A SERIES OF
REGRESSIONS FOR DIFFERENT DECISION POINTS IN THIS CHARGING AND
SENTENCING SYSTEM.
YOU WANT ME TO SAY ANYMORE THAN THESE ARE REGRESSIONS
THAT YOU FOUND AND HAD COPIED OR YOU WANT ME TO GQ INTO DETAIL
ABOUT EACH ONE? IM NOT SURE I CAN GIVE YOU TOO MUCH DETAIL
ABOUT EACH ONE.
QR. IT’S NOT NECESSARY TO GO INTO THAT MUCH DETAIL. AS LONG AS
WE GET A GENERAL IDEA OF WHAT THE DOCUMENT CONTAINS.
A. YES. THE WEIGHTED LEAST SQUARES REGRESSION AND ALSO
LOGISTIC REGRESSIONS, ALL OF WHICH I CONDUCTED EARLIER THIS
YEAR.
THE COURT: THAT'S 127?
rw
y
v
0
NN
OC
a
bd
O
N
10
1178
BALDUS - CROSS
MS. WESTMORELAND: YOUR HONOR, IT SHOULD BE LABELED 11.
ONE EXHIBIT GOT LEFT OUT. I BELIEVE YOUR COPY MAY BE LABELED
12.
BY MS. WESTMORELAND:
@. WOULD YOU EXAMINE THE NEXT EXHIBIT. PLEASE?
A. YES. DEFENDANTS EXHIBIT 12 ALSO INCLUDES THE REGRESSIONS
THAT UNDERLIE THE, SOME OF THE FIGURES IN TABLE 32 OF OUR WORKING
DRAFT OF JUNE 15, I BELIEVE, 1983. THESE WERE ALSO GIVEN TO THE
ATTORNEY GENERAL ON. IN THE DISCOVERY PROCESS.
@. DOES THAT INDICATE THAT THIS IS A STEPWISE REGRESSION?
A. YES. THIS ONE I‘M LOOKING AT HERE IS A STEPWISE PROCEDURE.
R. WOULD YOU EXAMINE THE NEXT. I“M SORRY, GO AHEAD?
A. NO, THAT'S FINE.
@. WOULD YOU EXAMINE THE NEXT DOCUMENT. PLEASE?
A. YES. RESPONDENT’S EXHIBIT 13 PRESENTS THE RESULTS OF A
FACTOR ANALYSIS USING THE FACTORS WHICH I ESTIMATED. AND
INCLUDED AN ANALYSIS FOR THE STATUTORY AGGRAVATING FACTORS FOR
THE JURY DECISION MAKING AT PENALTY TRIAL.
2. DOES THAT HAVE HANDWRITTEN. "FACTOR, JURY" ON THE DOCUMENT?
A. YES. |
2. WOULD YOU EXAMINE RESPONDENTS EXHIBIT NUMBER. I BELIEVE YOU
HAVE IT LABELED 147?
A. YES. THIS IS A. THIS IS THE RESULT OF A CROCREG,
C-R-0-C-R-E-D REGRESSION ANALYSIS OF THE DEATH PENALTY GIVEN A
MURDER INDICTMENT. EMPLOYING THE FACTORS ESTIMATED IN FACTOR
S
E
E
R
S
E
Y
BE
C
S
T
E
E
S
E
EE
E
E
[W
N oO
11
- 12
1179
BALDUS - CROSS
ANALYSIS WITH STATUTORY AGGRAVATING FACTORS INCLUDED AS WELL.
@. I BELIEVE THERE‘’S SOME HANDWRITING ON HERE. IT‘S SOMEWHAT
FAINT, I SEE THE WORD "FACTORY -—-
THE COURT: SCORES?
MS. WESTMORELAND: SCORES. PERHAPS?
THE WITNESS: YES.
BY MS. WESTMORELAND: :
@. AND WOULD YOU EXAMINE RESPONDENT‘S EXHIBIT NUMBER 15,
PLEASE?
A. YES, RESPONDENT“S 13 CONSISTS OF, OF A FEW MISCELLANEOUS
CROSS TABULATIONS PLUS A LARGE-SCALE REGRESSION ANALYSIS WITH AN
OUTCOME VARIABLE, DEATH SENTENCE GIVEN A MURDER INDICTMENT.
@. PROFESSOR BALDUS, ALL THESE DOCUMENTS THAT YOUYVE JUST
IDENTIFIED, RESFONDENT-S EXHIBIT 5, I BELIEVE, WAS A TABLE FROM
YOUR REPORT BUT THE REMAINDER OF THESE DOCUMENTS. DO YOU HAVE
THE ORIGINAL FOR THESE DOCUMENTS?
A. YES.
@. AND WOULD THESE BE DOCUMENTS THAT YOU COPIED ALL THESE AND
PROVIDED TO RESPONDENT?
A. YES, I DID.
fl. THANK YOLl,
AND NOW, PROFESSOR BALDUS, IF WE COULD. I WOULD LIKE TO
LOOK A LITTLE MORE SPECIFICALLY AT THE FULTON COUNTY AREA AND
THE INFORMATION THAT YOU HAVE COVERED ON FULTON COUNTY.
IN DOING YOUR ANALYSES, I BELIEVE YOU INDICATED TEN
118Q
BALDUS - CROSS
PERSONS THAT RECEIVED DEATH SENTENCES IN FULTON COUNTY. WAS THAT
CORRECT?
A. YES. ON THE BASIS OF MY INFORMATION, THAT“S CORRECT.
@. DO YOU HAPPEN TO KNOW HOW MANY OF THOSE PEOPLE ARE STILL
UNDER DEATH SENTENCE AT THIS TIME?
A. NO, I DO NOT KNOW THAT. I COULD FIND THAT OUT. BUT I DONT
KNOW. IT WAS NOT, WE DID NOT ADJUST THIS ANALYSIS FOR
SUBSEQUENT DECISIONS.
RA. IN EXAMINING THE FULTON COUNTY DATA. AND IN SOME OF THE
LATER ANALYSES THAT YOU DID, IN PARTICULAR, PERHAPS THE NEAR
NEIGHBOR ANALYSIS AND POLICE OFFICER COMPARISONS, YOURE
COMPARING ON A FAIRLY SMALL SAMPLE, ARE YOU NOT?
2. IN THAT TYPE OF COMPARISON, WOULD YOU SAY THAT IT WAS
IMPORTANT TO HAVE AS ACCURATE A DATA BASE AS POSSIBLE?
A. WELL, I THINK IT“S ALWAYS IMPORTANT TO HAVE ACCURATE DATA.
GQ. WOULD IT BE MORE IMPORTANT IN THIS TYPE OF SITUATION,
INVOLVING SUCH A SMALL SAMPLE SIZE. TO HAVE DATA OBTAINED FROM A
PRIMARY SOURCE, FOR INSTANCE, FROM THE PROSECUTOR HIMSELF?
A. WELL, IF WE WERE DOING A DIFFERENT STUDY AND WE WERE DOING
FULTON COUNTY OVER AGAIN, IT MIGHT BE AN INTERESTING APPROACH TO
50 AND TALK TO THE PROSECUTOR. BUT THAT WAS NOT THE DESIGN OF
THIS STUDY. THE DESIGN OF THIS STUDY WAS TO DRAW ON THE
RECORDS. PUBLIC RECORDS.
BUT THAT WOULD BE AN INTERESTING APPROACH TO GO AND
1181
BALDUS — CROSS
TALK TO PROSECUTORS. THAT MIGHT GIVE ONE A DIFFERENT PERCEPTION
OF WHAT WENT ON.
@. THE QUESTION IS, IF YOU HAVE, IN A SMALL SAMPLE SIZE, DOES
THAT BECOME MORE IMPORTANT THAN WITH A LARGER SAMPLE SIZE?
THE COURT: ISN‘T THAT REALLY EVIDENT?
MS. WESTMORELAND: PERHAPS IT IS, YOUR HONOR.
BY MS. WESTMORELAND: | |
Q@. 1 BELIEVE YOU INDICATED IN DOING YOUR POLICE OFFICER
ANALYSIS IN FULTON COUNTY THERE WAS ONLY THE INSTANT CASE IN
WHICH THERE HAD BEEN A DEATH SENTENCE GIVEN, IS THAT CORRECT?
A. YES.
R. AND BASED ON THAT SMALL NUMBER, IT MAKES IT DIFFICULT IF
NOT, VERY DIFFICULT TO DO ANY TYPE OF STATISTICAL ANALYSIS ON
THAT BASIS. WOULD THAT BE ACCURATE?
A. YES, IT’S DIFFICULT. I WOULD NOT CONSIDER IT IMPOSSIBLE TO
DO AN ANALYSIS THAT RESULTS IN A NUMERICAL ESTIMATE ABOUT WHATS
HAPPENING.
Q. WOULD YOU REFER TO DB-109, PLEASE?
IN THIS PARTICULAR TABLE. THE CASES, IF I RECALL, HAVE
BEEN BROKEN DOWN ON AGGRAVATION LEVELS, IS THAT CORRECT?
A. YES. THAT’S RIGHT.
Q. AND HOW MANY FACTORS DID YOU UTILIZE IN MAKING THAT
BREAKDOWN?
A. FIRST I USED 1S FACTORS. IN A. THAT WERE ESTIMATED IMPORTANT
IN A REGRESSION ANALYSIS STATEWIDE TO IDENTIFY A PRESUMPTIVELY
+
A
OR
R
1
[5
UN
TEV
I
(
1182
BALDUS - CROSS
COMPARABLE POOL OF CASES TO MCCLESKEY. AND THOSE ARE THE
VARIABLES LISTED IN, I BELIEVE IT’S SCHEDULE 5 OF, MY
RECOLLECTION, OF THE TECHNICAL APPENDIX.
THEN HAVING IDENTIFIED THAT GROUP OF 32 CASES, ACTUALLY
IT WAS 27 SUPPLEMENTED BY THE FIVE PENALTY TRIAL CASES, THEN WE
USED THE BREAKDOWN OF FIVE CATEGORIES, WHICH WOULD BUMP A CASE
UP INTO A HIGHER LEVEL OF AGGRAVATION: 3 FACTORS THAT WOULD BUMP
A CASE DOWN TO A LOWER LEVEL OF AGGRAVATION.
1. WHAT DECISION, OR WHAT JUSTIFICATION WAS UTILIZED FOR
DECIDING TO MAKE THE BREAKDOWN ON 135 FACTORS, RATHER THAN SOME
DTHER NUMBER?
A. OH. THOSE ARE THE FACTORS THAT SHOWED IN A STATEWIDE
ANALYSIS. STATISTICAL SIGNIFICANT RELATIONSHIP AND PRODUCED
CO-EFFICIENTS FOR AGGRAVATING FACTORS THAT RAN IN AN AGGRAVATING
DIRECTION, AND CO-EFFICIENTS FOR MITIGATING FACTORS THAT RAN IN
A MITIGATING DIRECTION. THAT WAS THE FRINCIPLE THAT WAS USED TO
IDENTIFY THOSE FACTORS.
2. DOES THAT INVOLVE, THEN, A PRESUMPTION THAT THE FACTORS
WHICH ARE MOST IMPORTANT IN THE REGRESSION ANALYSIS ON A
STATEWIDE BASIS, ARE ALSO GOING TO BE THE MOST IMPORTANT IN
FULTON COUNTY?
A. YES. IT USED THOSE FACTORS, THAT-S RIGHT. FOR A THRESHOLD.
THIS WAS TO, IT WAS A PRINCIPLE OF ECONOMY TO TRY TO FIGURE OUT
A WAY TO IDENTIFY WHAT WERE LIKELY THE MOST SERIOUS CASES HERE.
THAT WAS THE FIRST STEP IN THE ANALYSIS.
1182
BALDUS - CROSS
1 Q. WAS THIS ONE OF THE INSTANCES IN WHICH YOU INCLUDED. AND I
2 DON’T RECALL EXACTLY, THE. A ROLE BY A DISTRICT ATTORNEY OR A
3 JUDICIAL OFFICER TYPE FACTOR, WAS THAT INCLUDED IN THIS
4 |ANALYSIS?
5 |a. I DD NOT BELIEVE IT WAS. NO.
® 6 |@. DO YOU RECALL WHETHER IT WAS CONSIDERED IN ANY OF YOUR
7 [FULTON COUNTY ANALYSES?
8 |A. NOT TO MY KNOWLEDGE.
? ©. I BELIEVE YOU TESTIFIED YESTERDAY. FRIDAY ACTUALLY. THAT A
10 |STEPWISE, THAT THE STEPWISE REGRESSION ANALYSIS RUN ON FULTON
1 4 COUNTY BASICALLY DID NOT SHOW STATISTICALLY SIGNIFICANT
12 CO-EFFICIENTS FOR RACE OF VICTIM VARIABLES BEYOND THE .03 LEVEL.
13 WAS THAT CORRECT?
14 A. YES. THAT’S WITH REGARD TO THE COMBINED EFFECTS OF ALL THE
13 DECISIONS. THAT IS, ON THE DEPENDENT MEASURE. A DEATH PENALTY
1& GIVEN A MURDER INDICTMENT. WE DID NOT OBTAIN A STATISTICALLY
17 SIGNIFICANT RACE CO-EFFICIENT FOR THE RACE OF THE DEFENDANT OR
13 THE RACE OF THE VICTIM.
w 19 R. IN GENERAL, PROFESSOR BALDUS, IN MAKING ALL OF YOUR ANALYSES
20 I TAKE IT THAT YOU DO NOT ASSERT THAT RACIAL DISCRIMINATION IS
21 THE ONLY POSSIBLE EXPLANATION OF WHY THERE MAY BE A DISPARITY IN
22 WHITE VICTIM AND BLACK VICTIM CASES, IS THAT CORRECT?
23 A. WELL, THE TERM RACIAL DISCRIMINATION IS AS I PERCEIVE IT. IS
24 A LEGAL CONCLUSION.
25 WHAT I HAVE TRIED TO DEMONSTRATE IS THAT. I TRIED TO
BALDUS ~- CROSS
EXPRESS MY OPINION THAT CASES THAT ARE SIMILARLY SITUATED IN
THIS JURISDICTION ARE EVALUATED DIFFERENTLY BY THE PEOPLE WHO
PROCESS THEM IN THE SYSTEM. THATS THE BASIS OF MY OPINION.
THATS MY ULTIMATE FACTUAL INFERENCE , BECAUSE THAT”S THE ONLY
INFERENCE, THAT’S AS FAR AS MY INFERENCE. PARDON ME. THAT’S AS
FAR AS MY DATA CAN CARRY ME BY WAY OF MAKING INFERENCES ABOUT
FULTON COUNTY OR ABOUT THE STATE OF GEORGIA AS A WHOLE.
@. IN EXAMINING OR IN CONSIDERING THE DATA OBTAINED IN BOTH
STUDIES ——
A. I’M SORRY. I MISSED THE BEGINNING.
2. IN CONSIDERING THE DATA OBTAINED IN BOTH STUDIES, --
A. YES. |
BR. -— IS IT FAIR TO SAY, 1 BELIEVE YOU MAY HAVE ALREADY
INDICATED THIS, THAT YOU DON‘T HAVE ANY INDICATION OF THE
COMPOSITION OF THE JURY OF ANY PARTICULAR CASE?
A. NO, I DO NOT.
R. AND YOU ALSO HAD NO WAY OR DID NOT OBTAIN INFORMATION
CONCERNING ATITUDES OF JURORS ABOUT THE DEATH PENALTY?
A. NO, I HAD NO SUCH INFORMATION.
R. WOULD YOU HAVE ANY INFORMATION CONCERNING SYMPATHY FACTORS
THAT MAY BE PRESENTED TO A JURY DURING A SENTENCING PHASE OF A
TRIAL?
A. WELL, I THINK A LOT OF THE MITIGATING FACTORS. ON WHICH WE
DO HAVE DATA. WOULD PROVIDE THE BASIS OF SUCH AN APPEAL TQ A
JURY
1135
BALDUS - CROSS
TI
N BUT IF YOU MEAN INFORMATION BEYOND THAT. I HAVE NO.
NOTHING MORE THAN THAT. IN A DATA SENSE.
Q. INFORMATION SUCH AS NEIGHBORS TESTIFYING THAT HE WAS A NICE
PERSON ALL OF HIS LIFE AND THAT KIND OF A SYMPATHY TYPE FACTOR?
A. NO.
@. AND I BELIEVE YOU ALSO INDICATED EARLIER THAT THE RACE OF
THE VICTIM OR OF THE DEFENDANT WAS NOT THE DETERMINANT FACTOR IN
THE SENTENCING QUTCOME. WOULD THAT BE CORRECT?
,
B
E
S
W
A
R
Y
B
E
r
d
SE
E
BO
RE
Da
l
A. NOT THE ONLY FACTOR IN THE PROCESS, THAT’S QUITE RIGHT.
a
oO
MS. WESTMORELAND: MAY I HAVE JUST A MOMENT, YOUR HONOR?
11 BY MS. WESTMORELAND:
12 @. PROFESSOR BALDUS, IN CONSIDERING REGRESSIONS AND THE USE OF
13 REGRESSIONS, WOULD IT BE FAIR TO SAY THAT THE RESULTS OF A
14 REGRESSION CANNOT REALLY BE CONSIDERED TRUSTWORTHY UNTIL YOU
13 HAVE TESTED THE UNDERLYING ASSUMPTIONS OF REGRESSION PROCEDURES
16 IN GENERAL IN A CASE?
17 A. THAT QUESTION APPROACHES THE LIMITS OF MY COMPETENCE, BUT I
18 SAY I DO RECOGNIZE REGRESSION ANALYSES RESTS ON CERTAIN
XM 19 lASSUMPT IONS AND THE EXTENT TO WHICH THOSE ASSUMPTIONS HOLD IS
20 RELEVANT TO THE INTERPRETATION ONE PLACES OM THE RESULTS OF SUCH
21 AN ANALYSIS.
22 : M5. WESTMORELAND: EXCUSE ME. YOUR HONOR.
23 BY MS. WESTMORELAND:
24 2. PROFESSOR BALDUS, IF I COULD ASK YOU TO LOOK ONE MORE TIME
25 AT RESPONDENTS EXHIBIT NUMBER 37%
N
o
a
n
W
w
3
?
10
1186
BALDUS - CROSS
A. CERTAINLY.
@. JUST FOR FURTHER IDENTIFICATION OF THIS DOCUMENT, ARE THESE
WEIGHTED VALUES IN THIS PARTICULAR TABLE?
A. YES. THESE CO-EFFICIENTS ARE ALL ESTIMATED WITH A WEIGHTED
ANALYSIS.
@. AND IN DEVELOPING THE WEIGHTS TO BE UTILIZED IN THE CHARGING
AND SENTENCING STUDY. DID YOU, I GUESS, PRESUME THAT YOU HAD ALL
PENALTY TRIAL CASES IN DEVELOPING THOSE WEIGHTS?
A. NO. WE DIDNT MAKE ANY SUCH PRESUMPTION. WE EXAMINED ALL
THE CASES TO FIND OUT WHICH ONES HAD PENALTY TRIALS. WE
ATTEMPTED TO DETERMINE WHICH ONES HAD PENALTY TRIALS AND WHICH
ONES DIDNT.
2. DID YOU DEVELOP THE WEIGHTS. THEN, BASED ON HAVING ALL
PENALTY TRIAL CASES PRESENT OR HAVING SOME OF THE PENALTY TRIAL
CASES PRESENT?
A. WE DEVELOPED THEM ON THE BASIS OF THE PENALTY TRIALS THAT WE
FOUND. THAT’S HOW WE MADE THE DETERMINATION.
WE CONDUCTED AN EXTENSIVE SEARCH FOR PENALTY TRIAL
CASES. AND WE MAY HAVE MISSED A FEW. BUT I THINK WE HAVE
SUBSTANTIALLY THE PENALTY TRIAL CASES.
AND THEY, THEY WERE USED. THAT INFORMATION THEN
PROVIDED THE BASIS FOR CALCULATING THE WEIGHT.
MS. WESTMORELAND: THAT-S ALL THE QUESTIONS I HAVE.
THE COURT: ALL RIGHT. DO YOU HAVE ANY REDIRECT?
MR. BOGER: YES. YOUR HONOR.
1187
BALDUS - REDIRECT
1 RE-DIRECT EXAMINATION
2 BY MR. BOGER:
@. PROFESSOR BALDUS, YOU TESTIFIED THAT IN THE FROCEDURAL
REFORM STUDY YOU IMPUTED SENTENCING GUTCOMES TO THAT SMALL
NUMBER OF CASES WHERE THERE WERE UNKNOWN SENTENCING OUTCOMES. IS
3
4
5
® b THAT CORRECT?
7 A. WE KNEW THE SENTENCING OUTCOME IN ALL CASES. WHAT WE DID
8 WAS WE IMPUTED IN THE PROCEDURAL REFORM STUDY THE OUTCOME
? WHETHER OR NOT THE PROSECUTOR SOUGHT A DEATH SENTENCE AFTER A
10 CONVICTION AT A GUILT TRIAL.
11 2. I’M SORRY?
12 A. 23 CASES, 1 BELIEVE.
13 R. DID YOU MAKE A SIMILAR IMPUTATION IN THE CHARGING AND
14 SENTENCING STUDY WITH ANY CASES THAT HAD SIMILAR PROBLEMS?
15 A. NO, WE DIDN‘T. IN THE OTHER STUDY, THE SECOND STUDY, WE HAD
16 THE SAME PROBLEMS TO A CERTAIN EXTENT. THERE WERE CERTAIN CASES
17 WHERE WE DIDN’T KNOW WHETHER THERE WAS A PENALTY TRIAL. WHETHER
18 IT HAD GONE TO A MURDER-GUILT TRIAL. AND IN THAT STUDY WE
w 1? DROPPED ALL THE CASES WHERE WE DIDN’T KNOW WHETHER THERE WAS A
20 PENALTY TRIAL.
21 AGAIN, THE IDEA WAS THE EXTENSION OF THE TRIANGULATION
22 APPROACH. WE TOOK THE APPROACH OF IMPUTATION IN THE ONE STUDY:
23 THEN WE TOOK AN ALTERNATIVE APPROACH OF JUST DROPPING ALL THE
24 CASES WHERE WE DIDN‘T HAVE THE INFORMATION IN THE OTHER STUDY.
23 2. ON CROSS-EXAMINATION YOU WERE ASKED ABOUT DB-98, IN WHICH
fo
e
P
A
Oo
¥v
0
N
Oo
U
B
»
W
O
N
re
y
pb
Ju
de
r
BALDUS - REDIRECT
THERE WAS NOTED THERE WERE A GREAT NUMBER OF VARIABLES
CORRELATED WITH THE RACE OF THE VICTIM IN THE DEATH SENTENCING
DUTCOME.
AND I BELIEVE YOU GAVE TESTIMONY THAT THERE WERE MORE
THAN A HUNDRED VARIABLES WHICH WERE SO CORRELATED WITH THE RACE
OF THE VICTIM AND THE SENTENCING OUTCOME IN THE CHARGING AND
SENTENCING STUDY.
WHAT IF ANYTHING IS THE SIGNIFICANCE OF THAT HIGH
NUMBER OF CORRELATED VARIABLES?
ia. I THINK THE SIGNIFICANCE OF IT IS THAT IN THE FACE OF THIS
TREMENDOUS LIST OF VARIABLES THAT ARE CORRELATED WITH BOTH THE
DEATH SENTENCING QUTCOME AND THE RACE OF THE VICTIM, EVEN AFTER
CONTROLLING SIMULTANEOUSLY FOR ALL THOSE FACTORS, WE STILL GOT
SUBSTANTIAL RACE OF VICTIM, RACE OF DEFENDANT CO-EFFICIENTS THAT
WERE HIGHLY SIGNIFICANT FROM A STATISTICAL STANDPOINT.
WHAT ONE WOULD EXPECT IN THE FACE OF THE INTRODUCTION
OF BACKGROUND CONTROLS FOR THAT MANY VARIABLES THAT WERE HIGHLY
CORRELATED ON THESE TWO DIMENSIONS ONE WOULD EXPECT THAT TO WASH
OUT COMPLETELY THE RACE OF THE DEFENDANT AND THE RACE OF THE
VICTIM EFFECTS. UNLESS THERE WAS IN FACT A REAL EFFECT OUT IN
THE SYSTEM THAT WE WERE EXAMINING.
R. THE STATE ASKED YOU ON CROSS-EXAMINATION ABOUT FOSSIBLE
INTERACTION TERMS. I BELIEVE YOUR TESTIMONY WAS YOU HAD DONE
SOME PRELIMINARY INTERACTION ANALYSES.
DID THOSE ANALYSES YOU CONDUCTED ELIMINATE THE RACIAL
1139
BALDUS - REDIRECT
1 EFFECTS THAT YOU HAD FOUND?
2 A. NO, THEY HAD NO EFFECT WHATEVER ON THE RACIAL EFFECTS THAT
3 WE OBSERVED. WHAT THEY DID DO WAS FRESENT THE SAME SORT OF
4 PICTURE THAT I PRESENTED YESTERDAY IN TERMS OF THE FACTORS THAT
5 TENDED TO BE WEIGHTED MORE HEAVILY IN THE WHITE VICTIM CASES AS
kid b OPPOSED TO THE BLACK VICTIM CASES. ONE OF THE THINGS THAT
z INTERACTION ANALYSIS DOES BETWEEN. SAY, FELONY CIRCUMSTANCES AND
8 RACE OF VICTIM, THAT INTERACTION TERM, CO-EFFICIENT THAT ONE
? GETS WITH THAT INTERACTION TERM IN THE REGRESSION ANALYSIS TELLS
10 YOU THE EXTENT TO WHICH THE FELONY CIRCUMSTANCES VARIABLE HAS A
11 LARGER IMPACT AMONG WHITE VICTIM CASES THAN IT DOES AMONG BLACK
12 VICTIM CASES.
3 ANOTHER WAY OF GETTING THE SAME PICTURE WE GOT
14 YESTERDAY WHEN WE COMPARED THE TWO REGRESSION, SETS OF
13 REGRESSION CO-EFFICIENTS THAT WERE ESTIMATED, ONE FOR THE WHITE
14 VICTIM CASES AND SECONDLY FOR THE BLACK VICTIM CASES, TO
17 COMPARE THE RELEVANT IMPACT OF THE LEGITIMATE VARIABLES IN THOSE
18 TWO POPULATIONS.
w 37 AN INTERACTION ANALYSIS GIVES YOU THE SAME SORT OF
20 INFORMATION. AND THAT ZS THE INFORMATION THAT CAME THROUGH. IT
7.2 1 SIMPLY GAVE US A MORE REFINED SENSE OF WHERE, WHAT THE MECHANISM
22 WAS THAT WAS PRODUCING THE OVERALL RACE OF VICTIM EFFECTS.
23 R. PROFESSOR BALDUS, THE STATE ASKED YOU ABOUT DB-10%9. I
24 BELIEVE ONE OF ITS QUESTIONS WAS WHERE YOU DREW THE 13 VARIABLES
25 THAT YOU USED TO BEGIN YOUR ANALYSIS. AND I BELIEVE YOUR
Pa
rs
L
E
A
R
TE
RR
GR
E
e
E
E
E
E
E
C
R
10
11
1120
BALDUS - REDIRECT
TESTIMONY IN RESPONSE WAS THAT YOU WEREN‘T CERTAIN, BUT YOU
THOUGHT THAT TABLE S OF THE TECHNICAL APPENDIX REFLECTED THOSE
VARIABLES.
I WANTED TO GIVE YOU THE OPPORTUNITY TO LO3K AT THAT
SCHEDULE S IN THE TECHNICAL APPENDIX, WHICH IS DB-94A AND SEE
WHETHER UPON LOOKING AT THAT DOCUMENT YOU CAN CONFIRM OR DENY
THAT THOSE ARE THE VARIABLES.
A. VERY WELL.
YES. THERE“S 14 VARIABLES IN SCHEDULE S. OF THE
TECHNICAL APPENDIX.
NO, I“LL HAVE TO. I GUESS I JUST RECALL, REFRESH MY
MEMORY THAT I‘M, THE. I, THE, THE MODEL THAT WAS USED TO
ESTIMATE THE SCORES THAT PRODUCED THIS ORIGINAL SORT WERE A
FORERUNNER OF THOSE 14. THEY WEREN‘T THE EXACT 14.
I CAN GIVE YOU) THE TECHNICAL NAME. THE EXACT NAME OF
THAT, IT’S DPMURIDX WAS THE NAME OF THE INDEX THAT WAS USED TO
PRODUCE THAT.
THERE WERE SLIGHT MODIFICATIONS MADE AT A LATER TIME,
S0 THAT THE 14 VARIABLES OF THAT SCHEDULE 5S OF THE TECHNICAL
APPENDIX. ARE BASICALLY REPRESENTATIVE OF THOSE. AND IF YOU
CREATED AN INDEX WITH THOSE TWO SETS OF VARIABLES, THEY WOULD
BE, I“M CERTAIN, VERY HIGHLY CORRELATED. S00 THAT IT’S
ESSENTIALLY THAT GROUP OF VARIABLES.
2. BUT THAT. WAS THAT SERIES OF LETTERS YOU JUST READ OFF NOW.
THE MODEL. I TAKE IT, WHICH YOU DID USE TO CONSTRUCT THAT FIRST
BALDUS - REDIRECT
1 RUN ON THE, ON THE INDEX METHOD. WAS THAT MADE AVAILABLE TO THE
2 STATE?
3 A. YES. AND ALSO I JUST RECALL NOW, YESTERDAY I READ THOSE
VARIABLES INTO THE RECORD. WHEN WE WENT OVER THIS, COUNSEL.
YESTERDAY, YOU ASKED ME THAT YESTERDAY AND I READ THE VARIABLES
FROM THE FOOTNOTE OF THE FULTON COUNTY REPORT. SO NOW THAT I
4
5
é&
7 RECALL THIS. I READ THOSE YESTERDAY INTO THE RECORD. THOSE
8 VARIABLES.
9 QR. LET ME NOTE FINALLY THE STATE ASKED YOU WITH RESPECT TO
0 THOSE VARIABLES WHETHER ONE OF THEM WAS WHAT THEY CALL THE
11 AVENGE, OR MOTIVE TO AVENGE ROLE OF D.A., JUDICIAL OFFICER AND
12 LAWYER.
13 LET ME DIRECT YOU AGAIN TQ SCHEDULE S AND SEE IF THAT
14 REFRESHES YOUR MEMORY WHETHER IN THAT SERIES OF 15 THAT WAS A
i= VARIABLE EMPLOYED OR NOT?
14 A. WHERE 1 HAVE TO LOOK IS IN THE FOOTNOTES TO THE FULTON
17 COUNTY REPORT, AH, I WOULD LIKE TO AMEND MY REMARKS.
18 I HAD FORGOTTEN THAT 1 HAD GIVEN TESTIMONY YESTERDAY
iw 9 CONCERNING THESE AND INDEED, AVENGE WAS ONE OF THOSE VARIABLES.
20 fl. THAT INFORMATION OTHERWISE IS IN THE RECORD FROM YOUR
21 PREVIOUS TESTIMONY?
22 A. YES. I READ THOSE IN THE RECORD YESTERDAY AFTERNOON.
23 MR. BOGER: I HAVE NO FURTHER QUESTIONS, YOUR HONOR.
24 THE COURT: ALL RIGHT, THANK YOU, PROFFESSOR BALDUS.
23 GO DOWN.
~
O
a
+»
“
oO
10
ob
Ww
1172
BALDUS - REDIRECT
THE WITNESS: THANK YOU.
(WITNESS EXCUSED)
THE COURT: ALL RIGHT, MR. BOGER, CALL YOUR NEXT
WITNESS.
MR. BOGER: YOUR HONOR, BEFORE WE DO. I WANT TO ASK
ABOUT SCHEDULING.
| AS YOUR HONOR MAY BE AWARE. THERE WAS A PARDONS AND
PAROLES OFFICIAL THAT WE HAD TALKED ABOUT DEFERRING UNTIL
TUESDAY.
THERES ALSO THE QUESTION OF WHETHER WERE GOING TO
REACH THE PORTION OF OUR CASE INVOLVING INEFFECTIVE ASSISTANCE
BEFORE WE G0 INTO THE STATE’S CASE ON THE STATISTICAL EVIDENCE.
OR WHETHER THE COURT INTENDS TO DEFER THAT PORTION OF THE CASE
UNTIL LATER.
THE COURT: I DO.
MR. BOGER: VYDU“RE GOING TO DEFER IT. ALL RIGHT.
WE CALL DOCTOR GEORGE WOODWROTH TO THE STAND.
THE COURT: MR. WOODWORTH, COME UP.
THE CLERK: IF YOU WILL RAISE YOUR RIGHT HAND, SIR.
TRIAL NOW BEFORE THIS COURT SHALL BE THE TRUTH, THE WHOLE TRUTH,
AND NOTHING BUT THE TRUTH. SO HELP YOU GOD?
THE WITNESS: I DO.
THE CLERK: HAVE A SEAT ON THE WITNESS STAND. SIR,
AND STATE YOUR FULL NAME FOR THE RECORD.
DO YOU SOLEMNLY SWEAR THE EVIDENCE YOU SHALL GIVE IN THE
11293
WOODWORTH — DIRECT
Pr
y THE WITNESS: MY NAME IS GEORGE GORDON WOODWORTH. MY
< ADDRESS IS 14 WESTVIEW ACRES, ROUTE &, IOWA CITY, IOWA, 52240.
3 a Co
4 GEORGE GORDON WOODWORTH
3 CALLED AS A WITNESS ON BEHALF OF THE PETITIONER. BEING FIRST
Ki &. DULY SWORN, TESTIFIED AS FOLLOWS:
7 DIRECT EXAMINATION
8 BY MR. BOGER:
2? @. PROFESSOR WOODWORTH, WOULD YOU PLEASE GIVE YOUR NAME AND
10 ADDRESS ONE MORE TIME FOR THE RECORD?
31 A. GEDRGE GORDON WOODWORTH. 14 WESTVIEW ACRES. ROUTE 4. IOWA
12 CITY, IOWA, 52240.
13 MR. BOGER: .
14 @. AND DOCTOR WOODWORTH, WHAT IS YOUR OCCUPATION?
13 A. I’M AN ASSOCIATE PROFESSOR OF STATISTICS AT THE UNIVERSITY
146 OF IOWA.
i7 MR. BOGER: YOUR HONOR, I INTEND TO QUALIFY PROFESSOR
18 WOODWORTH AS AN EXPERT IN THE THEORY AND APFLICATION OF
19 STATISTICS AND IN STATISTICAL COMPUTATIONS, ESPECIALLY ANALYSIS
20 OF DISCREET OUTCOME DATA.
21 BY MR. BOGER:
22 @. PROFESSOR WOODWORTH. LET ME FIRST ASK YOU, IF YOU WOULD. TO
23 LOOK AT WHAT HAS BEEN MARKED FOR IDENTIFICATION AS GW-1.
24 IDENTIFY THAT DOCUMENT. IF YOU CAN?
25 A. THAT'S MY RESUME.
a
HH
»
W
N
1174
WOODWORTH - DIRECT
R. IS THAT RESUME CURRENT?
A. IN MOST RESPECTS. THERE IS THE OMISSION OF MY APPOINTMENT
TO THE COMMITTEE ON LAW AND JUSTICE STATISTICS OF THE AMERICAN
STATISTICAL ASSOCIATION.
@. OTHER THAN THAT APPOINTMENT. IS IT UP TO DATE?
A. YES. 1T Is.
@. IM GOING TO ASK YOU A FEW QUESTIONS ABOUT YOUR RESUME AND
CREDENTIALS, PROFESSOR WOODWORTH.
FIRST LET'S BEGIN WITH YOUR UNDERGRADUATE EDUCATION.
WHERE DID YOU RECEIVE YOUR UNDERGRAUATE AGREE?
THE COURT: MR. BOGER?
MR. BOGER: YES, SIR.
THE COURT: THE RESUME LOOKS FAIRLY COMPLETE AND I
THINK THE OUTCOME OF YOUR PROFFER OF HIM AS AN EXPERT IS
PROBABLY A FOREGONE CONCLUSION, ISN'T IT?
MR. BOGER: IT IS. YOUR HONOR. I THINK THERE MAY BE
SOME BATTLE OF EXPERTS ON STATISTICAL QUESTIONS AND IT MIGHT BE
IMPORTANT TO GET SOME SENSE. SO I CAN BE VERY RUICK.
THE COURT: WELL, ANYTHING YOU THINK IS IMPORTANT
BEYOND THE CURRICULUM VITAE THAT WE HAVE HERE, BRING IT OUT.
LETS MOVE ON.
MR. BOGER: I-LL TRY TO MOVE QUICKLY ON THAT.
BY MR. BOGER:
ii. PROFESSOR WOODWORTH, IN YOUR UNDERGRADUATE CAREER DID YOU
TAKE ANY COURSES IN STATISTICS?
1193
WOODWORTH — DIRECT
1 A. I TOOK ONE COURSE IN STATISTICS.
2 @. AND UNDER WHAT CIRCUMSTANCES DID YOU TAKE IT?
3 A. THIS COURSE FOLLOWED AN EXPERIENCE IN THE SUMMER OF MY
4 JUNIOR YEAR IN WHICH I WORKED AS AN UNDERGRADUATE IN AN
3 UNDERGRADUATE PARTICIPATION PROGRAM IN STATISTICS AT THE
® & UNIVERSITY OF MINNESOTA.
7 RX. DID YOU RECEIVE ANY SCHOLARSHIPS WHILE AN UNDERGRADUATE?
8 A. 1 HELD THE ALFRED P. SLOAN SCHOLARSHIP FOR FOUR YEARS.
? @. AND DID YOU RECEIVE ANY AWARDS OR HONORS AS A UNDERGRADUATE?
10 A. GRADUATED CUM LAUDE.
11 I WAS ELECTED TO PHI BETA KAPPA AND SIGMA XI.
12 @. WHAT IS SIGMA XI?
13 A. SIGMA XI IS A SCIENCE HONORARY. SIMILAR TO PHI BETA KAPPA.
14 @. I NOTE FROM YOUR RESUME THAT YOU RECEIVED YOUR F.H.D. DEGREE
15 FROM UNIVERSITY OF MINNESOTA, IS THAT CORRECT?
146 A. THATS CORRECT.
17 QR. IN WHAT DEPARTMENT?
18 A. IN THE DEPARTMENT OF MATHEMATICAL STATISTICS.
w 19 R}. DOES THE STATISTICS DEPARTMENT AT MINNESOTA. HAVE ANY
20 PARTICULAR BENT OR EMPHASIS?
21 A. YES, IT WAS A THEORETICAL OR MATHEMATICAL STATISTICS
22 DEPARTMENT.
23 RQ. AND WHAT ARE THE PARTICULAR STRENGTHS OF THAT KIND OF
24 AFPROACH TO STATISTICS?
23 A. THE STRENGTHS OF THAT APPROACH ARE THAT IT PROVIDES ONE WITH
1196
WOODWORTH - DIRECT
TRAINING AND EDUCATION IN THE FOUNDATIONS OF STATISTICS, WHICH
ARE ULTIMATELY BASED ON PROBABILITY THEORY. IT ENABLES ONE TO,
TO MAKE JUDGMENTS AS TO WHEN STATISTICAL TECHNIQUES CAN BE
APPROPRIATELY APPLIED, AND IT ALSO GIVES ONE THE TOOLS WITH
WHICH TO MAKE NECESSARY MODIFICATIONS TO STATISTICAL TECHNIQUES
WHEN REQUIRED.
@. ALL RIGHT. DID YOU TAKE ANY COURSES WHILE A GRADUATE
STUDENT AT MINNESOTA THAT ARE PARTICULARLY RELEVANT TO THE
STUDIES BEFORE THIS COURT?
A. WELL. ASIDE FROM THE STANDARD SEQUENCE OF COURSES THAT ANY
PHD STUDENT WOULD TAKE, I HAD COURSES IN THEORY --
QR. IF YOU WOULD WAIT JUST A MINUTE. PROFESSOR WOODWORTH.
THE COURT: WELL TAKE A RECESS FOR AT LEAST 10 OR 15
MINUTES. BE AVAILABLE, COUNSEL.
(RECESS TAKEN.)
GEORGE GORDON WOODWORTH.
BEING PREVIOUSLY DULY SWORN, RESUMED THE WITNESS STAND AND
TESTIFIED FURTHER AS FOLLOWS:
DIRECT EXAMINATION (CONT D)
BY MR. BOGER:
3. PROFESSOR WOODWORTH, I WAS ASKING YOU DURING YOUR PHD COURSE
WORK WHETHER YOU TOOK ANY COURSES THAT WERE PARTICULARLY
RELEVANT TO THE STUDIES THAT ARE BEFORE THE COURT NOW?
1127
WOODWORTH — DIRECT
A. PROBABLY THE MOST DIRECTLY RELEVANT COURSE WAS IN
NONPARAMETRIC ANALYSIS.
RQ. WHAT IS NONPARAMETRIC ANALYSIS?
A. NONPARAMETRIC IS A GROUP OF STATISTICAL THEORY WHICH
DEVELOPS ANALYSES WHICH ARE NOT STRONGLY. WHICH DO NOT REQUIRE
ASSUMPTIONS OF NORMALITY AND OTHER ASSUMPTIONS ABOUT THE DATA.
THE COURT: DO NOT REQUIRE ASSUMPTIONS OF WHAT?
THE WITNESS: ASSUMPTIONS THAT THE DATA HAVE A NORMAL
DISTRIBUTION, YOUR HONOR.
RY MR. BOGER:
@. AND HOW IS THAT RELATED TO THESE STUDIES?
A. WELL, THAT. OF COURSE, IS A SITUATION WERE IN IN THESE
STUDIES.
ONE OF THE COMPONENTS IN NONPARAMETTRIC ANALYSIS OR AT
LEAST ONE BRANCH IS STATISTICS THAT INTERSECTS WITH
NONPARAMETRIC ANALYSIS IS WHAT-S CALLED ANALYSIS OF CATEGORICAL
DATA.
@. WHAT DD YOU MEAN BY THAT?
A. CATEGORICAL DATA ARE DATA THAT COME IN DISCREET PACKETS LIKE
LIFE OR DEATH, FOR EXAMFLE. THERE-S NO CONTINUUM BETWEEN LIFE
AND DEATH. OR HAVING A BLACK VICTIM COR A WHITE VICTIM.
2. S00 THAT WAS ONE OF YOUR SPECIALTIES IN COURSE WORK.
DID YOu DO ANY FURTHER WORK IN NONPARAMETRICS?
A. IT’S BEEN A STEADY THEME IN MY RESEARCH. 1-VE STUCK WITH
THIS PARTICULAR AREA, AMONG OTHERS. FOR MOST OF MY CAREER.
1198
WOODWORTH - DIRECT
R. WHAT WAS YOUR PHD DISSERTATION IN?
A. MY PHD DISSERTATION WAS IN THE AREA OF NONPARAMETRIC, IT’S
OTHER COMMUNICATION NUMBER ONE. IT“S ON TESTS OF INDEPENDENCE.
NONPARAMETRIC TESTS OF INDEPENDENCE.
QR. YOUR OTHER COMMUNICATION NUMBER ONE, YOU MEAN IN YOUR
RESUME?
A. RESUME, YES.
R. DID YOU TAKE ANY COURSES IN MULTIVARIATE ANALYSIS IN YOUR
PHD WORK?
A. YES, I DID. I HAD A 3-TERM SEQUENCE IN THE THEORY OF
MULTIVARIATE ANALYSIS.
@. WHAT, VERY BRIEFLY, IS MULTIVARIATE ANALYSIS?
A. MULTIVARIATE ANALYSIS IS A GROUP OF STATISTICAL TECHNIQUES
WHICH ARE CONCERNED WITH ANALYZING MORE THAN ONE VARIABLE AT A
TIME.
R. IS THAT WHAT TECHNIQUES ARE EMPLOYED IN THESE STUDIES?
A. YES. INDEED.
QR. NOW, WHO WAS ON YOUR THESIS COMMITTEE?
A. WELL, AMONG MY ADVISERS AT ONE TIME OR ANOTHER WERE RICHARD
SAVAGE WHO IS NOW AT YALES AND MILTON SOBEL, WHO IS A
DISTINGUISHED PROFESSOR AT SANTA BARBARA.
RN. WHATS THEIR REPUTATION WITHIN THE STATISTICAL COMMUNITY?
A. BOTH OF THESE MEN ARE FELLOWS IN THE AMERICAN STATISTICAL
ASSOCIATION AND HAVE NATIONAL RECOGNITION IN THE PROFESSION.
i. DID YOU DO ANY CO-AUTHORSHIPS WITH THEM?
1199
WOODWORTH - DIRECT
1 A. YES, I DID. I HAVE, MY FIRST PUBLICATION I3 CO-AUTHORED
2 WITH BOTH OF THOSE GENTLEMEN AND MY THIRD ONE IS CO-AUTHORED
3 WITH PROFESSOR SOBEL.
4 @. I NOTICE FROM YOUR RESUME THAT UPON RECEIPT OF YOUR PHD,
3S YOU OBTAINED ACADEMIC EMPLOYMENT AT STANFORD, IS THAT CORRECT?
6 A. THAT’S CORRECT.
7 QR. IN WHAT DEPARTMENT?
3 A. DEPARTMENT OF STATISTICS.
? @. AND WHAT“S THE REPUTATION OF STANFORD’S STATISTICAL
10 DEPARTMENT. OR STATISTICS DEPARTMENT NATIONALLY?
11 A. 1 BELIEVE IT’S CONSIDERED TO BE ONE OF THE FIRST TIER OF
32 STATISTICS DEPARTMENTS.
13 QR. IN WHAT AREAS DID YOU TEACH AT STANFORD?
14 A. WELL, I TAUGHT IN MY AREA OF SPECIALIZATION WHICH IS
13 NONPARAMETRIC.
1&6 I ALS0 TAUGHT VARIOUS GENERAL SERVICE COURSES.
17 THE COURT: I TAUGHT WHAT?
18 THE WITNESS: GENERAL SERVICE COURSES. YOUR HONOR. THATS A
w 19 COURSE YOU WOULD TEACH TO PEOPLE IM OTHER DEPARTMENTS,
20 LIKE PSYCHOLOGY OR ENGINEERING.
21 THE COURT: WHAT WE USED TO CALL SURVEY COURZES.
22 THE WITNESS: SURVEY. RIGHT.
23 BY MR. BOGER:
24 RN. WHAT WERE YOUR RESEARCH INTERESTS DURING THIS TIME AT
23 STANFORD?
83
0
.
0
a
WOODWORTH — DIRECT
A. MY RESEARCH INTERESTS WERE IN, AGAIN IN NONPARAMETRIC
ANALYSIS. AND CATEGORICAL DATA ANALYSIS.
MY HARDEST RESEARCH EFFORT WHILE AT STANFORD IS
REFLECTED BY A FUBLICATION NUMBER 4.
@. VERY BRIEFLY, WHAT DOES THAT PUBLICATION TALK ABOUT?
A. IT TALKS ABOUT ANALYSIS OF A CERTAIN GROUP OF, THE
PROPERTIES RATHER OF A CERTAIN GROUP OF STATISTICAL ANALYSES FOR
DEALING WITH, COULD BE CATEGORICAL DATA, BUT IT’S SOMEWHAT MORE
GENERAL THAN THAT.
2. NOW PROFESSOR WOODWORTH, YOU-VE TESTIFIED. I BELIEVE, THAT
THE TRAINING THAT YOU HAD IN STATISTICS WAS LARGELY THEORETICAL
TRAINING.
DID AT ANY POINT DID YOU BEGIN TO CONDUCT ANY APPLIED
STATISTICAL RESEARCH?
A. YES. TOWARDS THE END OF MY TIME AT STANFORD, I. I BECAME
INTERESTED IN APPLIED STATISTICS. SO I BEGAN, I EXFRESSED THIS
INTEREST TO PEOPLE IN THE DEPARTMENT OF BIOSTATISTICS AT THE
STANFORD MEDICAL SCHOOL AND WAS INVITED TO COME AND PARTICIPATE
IN SOME NF THEIR TEACHING AND THEIR PROJECTS.
IN PARTICULAR AT THAT POINT I BECAME INVOLVED IN THE
FINAL STAGES OF THE HALOTHANE STUDY.
2. IS THAT THE NATIONAL HALOTHANE STUDY WHICH PROFESSOR BALDUS
EARLIER TESTIFIED —-
A. YES.
2. —— CONCERNING?
L
S
S
1
T
E
IE
J
1
EE
WOODWORTH —- DIRECT
WHAT WAS YOUR RESPONSIBILITY WITH RESPECT TO THAT
STUDY?
A. WELL, MY RESPONSIBILITY WAS TO DO A LITERATURE REVIEW AND AN
INDEPENDENT ANALYSIS OF SOME OF THE METHODOLOGY THAT HAD BEEN
DEVELOPED BY HALOTHANE STATISTICIANS AND THIS 1S REFLECTED IN
OTHER COMMUNICATION NUMBER 2, WHICH IS A REPORT TO THE NATIONAL
RESEARCH COUNCIL.
QR. WHO INVITED YOU TO CONDUCT THIS REVIEW?
A. FREDERICK MOSTELLER INVITED ME TO CONDUCT THAT REVIEW.
@. AND WHO IS FREDERICK MOSTELLER?
A. FREDERICK MOSTELLER IS A WELL KNOWN STATISTICIAN. HE’S ONE
OF THE CHIEF STATISTICIANS IN THE HALOTHANE STUDY.
@. S0, WHAT. YOU REVIEWED THE LITERATURE AND THE METHODS
INVOLVED IN THE NATIONAL HALOTHANE STUDY: IT EVENTUALLY
RESULTED, I THINK YOU TESTIFIED, IN A REPORT TO THE NATIONAL
RESEARCH COUNCIL.
WHO ARE THEY?
A. THE NATIONAL RESEARCH COUNCIL WAS THE FUNDING BODY FOR THE
[HALOTHANE STUDY.
@. ALL RIGHT. DID YOU AT ANY TIME HAVE ANY OTHER OPPORTUNITIES
TO TAKE ON AN APPLIED STATISTICAL POSITION RATHER THAN A
THEORETICAL POSITION WHICH YOU EARLIER INDICATED YOU HAD
OCCUPIED?
A. IN MY LAST YEAR AT STANFORD, NEXT TO THE LAST YEAR AT
STANFORD, I WAS OFFERED A JOINT APPOINTMENT WITH, BETWEEN
[a
nN
oO
B
M
N
O
a
s
PA
s
1202
WOODWORTH - DIRECT
STATISTICS AND BIOSTATISTICS.
@. WOULD THAT HAVE INVOLVED APPLIED APPLICATION?
A. YES. IT WOULD.
@. AND DID YOU ACCEPT THAT EMPLOYMENT?
A. NO. INSTEAD I TOOK A YEAR’S LEAVE.
BG. WHAT DID YQU Do DURING THAT YEAR OF LEAVE?
A. WELL, ONE THING I DID WAS PREPARE THE REVIEW. THE
LITERATURE. THE LITERATURE REVIEW AND ANALYSIS OF METHODS THAT
WAS OTHER COMMUNICATION NUMBER 2.
AND THE OTHER HALF OF MY TIME WAS SPENT TEACHING
STATISTICS IN LUND, SWEDEN. IN THE INSTITUTE OF TECHNOLOGY
THERE.
2. AND WHAT DID YOU TEACH AT THAT TIME? |
A. I TAUGHT THEORY OF PROBABILITY AND MATHEMATICAL STATISTICS. .
2. ALL RIGHT. WHERE DID YOU NEXT ACCEPT ACADEMIC EMPLOYMENT?
A. I WENT TO THE UNIVERSITY OF IOWA NEXT.
(. AND AT IOWA. WHAT COURSES WERE YOU ASKED TO TEACH?
A. WELL, I WAS. I WAS HIRED WITH THE UNDERSTANDING THAT I WOULD
DEVELOP, PARTICIPATE IN THE DEVELOPMENT OF AN APPLIED STATISTICS
SECTION IN THE MATHEMATICS STATISTICS PROGRAM.
S00, IN THE BEGINNING I TAUGHT MANY OF THE AFFLIED
COURSES. THIS WOULD INCLUDE DESIGN AND ANALYSIS OF EXPERIMENTS.
APPLIED MULTIVARIATE ANALYSIS, STATISTICAL COMPUTATION. AND THEN |
VARIOWS SERVICE COURSES ALSO.
R. WHAT COURSES HAVE YOU TAUGHT RECENTLY?
WOODWORTH — DIRECT
1 A. RECENTLY. WELL, NOT RECENTLY. OVER THE LAST SEVERAL YEARS,
2 THE DEPARTMENT HAS HIRED A SOLID GROUP OF APPLIED STATISTICIANS.
AND IT HASN’T BEEN NECESSARY FOR ME TO TEACH SO MANY APPLIED
COURSES. RECENTLY IVE HAD THE PLEASURE OF BEING ABLE TO TEACH
THEORY COURSES AGAIN. AND I TAUGHT THE THEORY OF MULTIVARIATE.
3
4
3
® & THEORY OF REGRESSION, AND THEORY OF STATISTICS.
7 ae: 1 NOTICE IN YOUR RESUME IT STATES THAT YOU HAVE SERVED WITH
8 THE STATISTICAL CONSULTING CENTER AT IOWA.
9 WHAT IS THAT?
10 A. THE STATISTICAL CONSULTING CENTER IS A SERVICE THAT THE
11 STATISTICS DEPARTMENT PROVIDES FOR THE REST OF THE UNIVERSITY.
12 ITS STAFFED BY ADVANCED GRADUATE STUDENTS UNDER THE SUPERVISION
13 OF THE DIRECTOR. AND IT PROVIDES STATISTICAL SERVICES TO
14 STUDENTS, FACULTY, STAFF.
1% Gl. WHAT WAS YOUR TITLE AT THE CENTER?
16 A. I FOUNDED THE CENTER, AND SERVED AS DIRECTOR FOR SEVERAL
17 YEARS,
18 RB. WHAT IS YOUR TITLE NOW?
w 1% A. LAST YEAR, I RESIGNED AS DIRECTOR, AND -— YEAR BEFORE LAST 1
20 RESIGNED. PROFESSOR RUSSELL LANTH WAS AFFOINTED DIRECTOR.
21 RUSS WENT ON LEAVE LAST YEAR, SO I-M NOW ACTING DIRECTOR.
22 2. YOUR TESTIMONY WAS THAT YOU HAVE GIVEN THROUGH THIS CENTER
23 STATISTICAL ADVICE TO VARIOUS PEOPLE. SINCE YOUVE BEEN
24 INVOLVED WITH THE STATISTICAL CONSULTING CENTER, HOW MANY
23 EMPIRICAL STUDIES WOULD YOU ESTIMATE YOUVE GIVEN ADVICE TO OR
1204
WOODWORTH — DIRECT
CONSULTATION TO?
A. IN ONE WAY. PARDON ME. IN ONE WAY OR ANOTHER. I“VE GIVEN
ADVICE TO TEN OR TWENTY A YEAR FOR EIGHT YEARS, PERHAPS.
@. SO EIGHTY TO A HUNDRED AND FIFTY. THAT KIND OF THING?
A. UH HUH.
Q. HAVE YOU DONE ANY PARTICULAR CONSULTATIONS THAT ARE RELEVANT
TO THE STUDY BEFORE THE COURT NOW?
A. WELL, YES. I HAVE COLLABORATED WITH PROFESSOR BALDUS FOR
SEVERAL YEARS, BEGINNING WITH OUR ANALYSIS OF THE DATA GATHERED
BY THE STANFORD LAW REVIEW. THE, THE CONSULTATION WHICH IS MOST
RELEVANT TO THIS MATTER AT ISSUE HERE IS THE WORK THAT I BEGAN
DOING WITH PROFESSOR BALDUS REGARDING THE DATA GATHERED BY THE
STANFORD LAW REVIEW STUDENTS.
@. ALL RIGHT. DURING YOUR CONSULTING WORK, DID YOU DEVELOP ANY
KNOWLEDGE OR EXPERTISE IN THE AREA OF STATISTICAL COMPUTATION?
A. YES. I HAVE BEEN IN CHARGE OF THE COURSE IN STATISTICAL
COMPUTATION. I TEACH THEORY OF STATISTICAL COMPUTATION. IN
ADDITION TO WHICH, OF COURSE, ANYONE WHO RUNS A CONSULTING
CENTER IS GOING TO LEARN HOW TO COMPUTE, AND I AM QUITE. I AM
SKILLED IN THE USE OF THE STANDARD STATISTICAL COMPUTING
FACKAGES. |
RF. SO YOUVE WORKED EXTENSIVELY WITH COMPUTERS?
A. I USE THEM DAILY.
fl. ALL RIGHT. LET ME ASK YOU QUICKLY ABOUT YOUR CURRENT
RESEARCH INTERESTS?
10
1205
WOODWORTH ~~ DIRECT
A. MY CURRENT RESEARCH INTERESTS ARE. AROSE OUT OF CONTACTS
THAT 1 MADE AS DIRECTOR OF THE STATISTICAL CONSULTING CENTER.
MY RESEARCH INTERESTS ARE IN THE AREA OF LAW AND JUSTICE
STATISTICS. IN STATISTICAL APPLICATIONS IN BIOMECHANICS, AND THE
THEORY OF MULTIVARIATE ANALYSIS. PARTICULARLY DISCREET
MULTIVARIATE ANALYSIS. AND IN STATISTICAL COMPUTING.
@. A DISCREET MULTIVARIATE ANALYSIS IS WHAT? IS THAT RELATED
TO THIS STUDY AT ALL?
A. YES, FOR EXAMPLE, LOGISTIC REGRESSION WOULD FALL UNDER THAT
HEADING.
R. ALL RIGHT. NOW, LET ME QUICKLY LOOK AT ONE OR TWO OF YOUR
PUBLICATIONS AND ASK YOU A QUESTION ABOUT THEM.
I NOTE ON PAGE 2 OF WHATS BEEN MARKED GW-1 THAT YOUR
FIRST PUBLICATION LISTED AS SOMETHING CALLED THE "ANNALS OF
MATHEMATICAL CALIFORNIA STATISTICS." WHAT IS THAT JOURNAL?
A. THATS THE, THAT’S A PUBLICATION OF THE INSTITUTE OF
MATHEMATICAL STATISTICS. IT7S THE PREMIERE JOURNAL IN
MATHEMATICAL STATISTICS IN THIS COUNTRY.
®. IS IT A REFEREED JOURNAL IN THAT SENSE. PEER REVIEW?
A. YESs IT IR.
. HOW MANY PUBLICATIONS HAVE YOU HAD IN THAT JOURNAL?
A. WELL, IN THE DAYS WHEN I WAS DOING MATHEMATICAL STATISTICS,
LOOKS LIKE FIVE.
XR. OKAY. I NOTICE YOUR SIXTH PUBLICATION IS IN SOMETHING
CALLED "THE AMERICAN STATISTICIAN."
oc
(A
R
&
LO
WN
WOODWORTH — DIRECT
WHAT IS THAT JOURNAL?
A. THAT’S A PUBLICATION THAT GOES TO ALL MEMBERS OF THE
AMERICAN STATISTICAL ASSOCIATION.
GQ. AND IS THAT ALSO A REFEREED READ JOURNAL?
A. THAT IS REFEREED. YES.
CG. ARE YOU A MEMBER OF ANY PROFESSIONAL ASSOCIATIONS?
A. 1‘M A MEMBER OF THE AMERICAN STATISTICAL ASSOCIATION.
fn. AND HAVE YOU EVER BEEN INVITED TO PRESENT PAPERS TO THAT
ASSOCIATION?
A. YES. ON SEVERAL OCCASIONS. THE MOST RECENT WAS THIS PAST
SATURDAY. WHEN I PRESENTED A PAPER IN TORONTO.
R. WHAT WAS THE SUBJECT OF THAT PAPER?
A. THE SUBJECT OF THAT PAPER WAS THE ANALYSIS OF THE TYPE OF
STRATIFIED DATA THAT WE/RE DEALING WITH IN THIS STUDY.
MR. BOGER: YOUR HONOR, I SUBMIT THAT PROFESSOR
WOODWORTH IS AN EXPERT GUALIFIED IN THE THEORY AND APPLICATION
DF STATISTICS. AND IN THE STATISTICAL COMPUTATIONS, ESPECIALLY
ANALYSIS OF DISCREET OUTCOME DATA.
THE COURT: MS. WESTMORELAND, DO YOU WISH TO VOIR DIRE
HIM ON HIS QUALIFICATIONS?
MS. WESTMORELAND: NO, I DO NOT.
THE COURT: ALL RIGHT, HE WILL BE ACCEPTED AND MAY
EXPRESS HIS OPINION.
MR. BOGER: THANK YOU. YOUR HONOR.
I MOVE THE ADMISSION OF GW-1 INTO EVIDENCE.
1207
WOODWORTH — DIRECT
MS. WESTMORELAND: I HAVE NO OBJECTION. Pa
THE COURT: IT WILL BE ADMITTED.
BY MR. BOGER:
Q. PROFESSOR WOODWORTH, THERE’S BEEN TESTIMONY BY PROFESSOR
BALDUS THAT YOU“VE BEEN INVOLVED IN BOTH THE PROCEDURAL REFORM
STUDY AND THE CHARGING AND SENTENCING STUDY.
COULD YOU GIVE US A BRIEF OUTLINE OF YOUR
RESPONSIBILITIES IN THOSE STUDIES?
vg
O
N
Oa
3
W
N
A. THE FIRST RESPONSIBILITY THAT I HAD IN THE PROCEDURAL REFORM
10 STUDY WAS TO GIVE SOME ADVICE AS TO HOW FRED KYLE WAS TO
11 | ALLOCATE HIS EFFORTS IN THE FIRST SUMMARY HE WAS ATTEMPTING TO
12 USE THE SUPREME COURT QUESTIONNAIRE.
13 @. IN WHAT SENSE DO YOU MEAN ALLOCATE HIS EFFORTS?
14 A. WELL, AS PROFESSOR BALDUS TESTIFIED, IT WAS QUICKLY APPARENT
15 THAT FRED WOULD NOT BE ABLE TO GATHER DATA ON ALL CASES THAT
16 SUMMER AND THERE WAS CONCERN WHATEVER HE WOUND UP WITH WOULD BE
17 A RANDOM SAMPLE FOR STATISTICAL ANALYSIS.
13 2. WHAT DID YOU DO?
w 19 A. I GAVE HIM A SCHEDULE OF RANDOM NUMBERS FOR HIM TO USE IN
20 SELECTING WHICH CASE TO, TD CODE NEXT.
21 RQ. AND WAS THE ADVICE THAT YOU GAVE HIM A VALID AND
Se STATISTICCALLY SOUND PROCEDURAL —-
23 A. YES, STANDARD WAY OF SELECTING A RANDOM SAMPLE.
24 @. WHAT WAS YOUR NEXT CONTRIBUTION TO THE PROCEDURAL REFORM
23 STUDY?
Pa
y
4
3
d
W
O
N
1208
WOODWORTH ~ DIRECT
A. MY NEXT CONTRIBUTION WAS TO. TQ TAKE THE DATA FILE THAT HAD
BEEN ENTERED INTO IT TO THE STATISTICS DEPARTMENT AT THE
UNIVERSITY OF IOWA AND CONVERT IT INTO A FILE SUITABLE FOR
STATISTICAL ANALYSIS.
THIS INVOLVED MERGING THE DATA FROM THE SUPREME COURT
AND PROCEDURAL REFORM QUESTIONNAIRES WITH THE DATA FROM THE
DEPARTMENT oF OFFENDER REHABILITATION TAPE.
A. DID YOU, YOURSELF, CONDUCT THAT MERGER?
A. YES, 1 DID.
@. AND WAS THAT CONDUCTED ACCORDING TO ACCEPTABLE STATISTICAL
TECHNIQUES AND PROCEDURES?
A. YES. I USED THE STATISTICAL ANALYSIS SYSTEM. IT’S A
STATISTICAL PACKAGE CALLED SAS FOR SHORT. IT HAS A VERY
POWERFUL SET OF DATA MANAGEMENT PROCEDURES AND I USED THE MERGE
PROCEDURE AND THE SORT PROCEDURES IN VARIOUS WAYS TO ACHIEVE
THIS MERGER.
@. DID YOU PERFORM ANY CHECKS ON THE MERGER TO MAKE SURE IT WAS
DONE APPROPRIATELY?
A. YES. I PERFORMED A VARIETY OF CHECKS AND THESE TYPICALLY
WOULD CONSIST OF AFTER THE MERGER. PRINTING OUT VARIOUS KEY
ITEMS OF INFORMATION. FOR EXAMPLE, THE CASE NUMBER THAT HAD
BEEN ASSIGNED TO THE CASE, THE, FOR EXAMPLE, DATE OF OFFENSE,
DATE OF ARREST. AND SO ON. I PRINTED OUT THOSE VARIABLES WHICH
WERE BOTH ON THE DEPARTMENT OF OFFENDER REHABILITATION DATA FILE
AND ON OUR QUESTIONNAIRE. AND THEN THESE WERE HAND COMPARED BY
WOODWORTH — DIRECT
1 CLERKS AND STUDENTS.
2 @. ALL RIGHT. WHAT WAS YDUR NEXT CONTRIBUTION AFTER THIS
MERGER HAD BEEN COMPLETED TO THE PROCEDURAL REFORM STUDY?
A. AFTER THIS MERGER HAD BEEN COMPLETED WITH REGARD TO THE
PROCEDURAL REFORM STUDY, I PRODUCED VARIOUS INITIAL TABULATIONS
3
4
3
& AND SIMPLE STATISTICAL ANALYSES TO BEGIN TO GAIN AN
7 UNDERSTANDING OF THE RELATIONSHIPS AMONG THE VARIABLES ON THE
8 DATA SET.
4 @. AND FOR WHAT PURPOSE WAS THIS?
10 A. VARIOUS PURPOSES, ONE OF WHICH WAS TO CHECK FOR THE
11 CONSISTENCY OF THE DATA TO SEE THAT THERE WAS NOTHING
12 EGREGIQUSLY WRONG WITH IT, AND SECOND, TO BEGIN TO SCREEN
13 VARIABLES TO DISCOVER WHICH VARIABLES WERE INFLUENCING THE
14 OUTCOME OF THE, OF THE SENTENCING PROCESS AND WHICH WERE NOT.
13 G. DID YOU FIND ANYTHING EGREGIOUSLY WRONG?
16 A. OH NO. NOTHING SERIOUS. WE WOULD FIND MINOR
17 INCONSISTENCIES SUCH AS CODES THAT WEREN‘T PERMITTED.
13 ®. DID YOU CORRECT ALL THOSE YOU FOUND?
ww 1% A. YES, THESE WERE CORRECTED ON AN ONGOING BASIS.
20 Q. PROFESSOR BALDUS TESTIFIED THAT YOU WERE RESPONSIBLE FOR
23 DEVELOPING A SAMPLING PLAN FOR THE CHARGING AND SENTENCING
22 STUDY. IS THAT CORRECT?
23 A. THAT’S CORRECT.
24 RQ. WHAT DID THAT INVOLVE?
23 A. A SAMPLING PLAN IS A SET OF INSTRUCTIONS TO THE CODERS WHO
tit tet. itp S—— ——
WOODWORTH ~ DIRECT
ARE GOING TO BE ON THE SCENE IN GEORGIA AS TO WHICH CASES TO
SELECT FROM THE FILES OF THE PAROLE BOARD AND TO REDUCE TO THE
FORM OF A CODED QUESTIONNAIRE.
@. WHAT’S THE PURPOSE OF THE SAMPLING PLAN?
A. THE PURPOSE OF THE SAMPLING PLAN IS TO INSURE THAT THE DATA
ONCE GATHERED CAN BE VALIDLY ANALYZED BY STATISTICAL PROCEDURES.
@. LET ME DIRECT YOUR ATTENTION TO WHATS BEEN MARKED GW-2 FOR
IDENTIFICATION. CAN YOU IDENTIFY THAT DOCUMENT?
A. YES. THIS IS. THIS DOCUMENT PRESENTS MY SAMPLING PLAN IN
GREAT DETAIL. IT INCLUDES EVERYTHING FROM A SORT OF GENERAL
SUMMARY TO THE PLAN, DOWN TO THE COMPUTER FROGRAMS THAT I WROTE
TO GENERATE THE RANDOM NUMBERS.
@. ALL RIGHT. I. AT THIS POINT, WE THOUGHT WE MIGHT GO THROUGH
IT IN SOME DETAIL, BUT LET ME JUST ASK YOU A SUMMARY QUESTION
ABOUT IT. AND ANY OTHER TESTIMONY THATS NECESSARY CAN BE PICKED
UP ON REDIRECT.
WAS THIS SAMPLING PLAN DEVELOPED BY YOU IN ACCORD WITH
ACCEPTABLE STATISTICAL TECHNIGUES.
AR. YES, IT WAS.
@. AND DID YOU CONSULT WITH ANYONE ABOUT THE SAMPLING FLAN?
A. YES, 1 DID. IN THE DESIGN STAGES OF THIS PLAN I DID CONSULT
WITH PROFESSOR LEON BURMEISTER OF THE UNIVERSITY OF IQWA.
A. WHO IS HE?
A. HE'S A SPECIALIST IN SAMPLING DESIGN.
Gl. WHAT WAS HIS OPINION, IF YOU RECALL, OF YDUR SAMPLING PLANT
WONDWORTH — DIRECT
1 A. HIS OPINION AS I RECALL IT WAS THAT THE PLAN WAS
STATISTICALLY VALID, ALTHOUGH HE DID POINT OUT, WHICH I WELL
KNEW AT THAT TIME, THAT IT WOULD PRESENT SOME INTERESTING
2
W
N
CHALLENGES IN THE ANALYSIS PHASE.
@. DID HE FIND IT UNACCEPTABLE, GIVEN THE PURPOSES OF YOUR AH
5 STUDY?
7 A. NO. HE DID NOT FIND IT UNACCEPTABLE.
8 @. DID HE FIND IT VALID?
9 A. OH. YES, HE FOUND IT VALID.
0 MR. BOGERt YOUR HONOR, AT THIS TIME, I MOVE THE
11 ADMISSION OF COW-2 AS REFLECTING THE SAMPLING PLAN IN THIS STUDY.
12 MS. WESTMORELAND: YOUR HONGR, ILL OBJECT TO THE
13 ADMISSION OF THIS DOCUMENT. WE HAVE HEARD BRIEF TESTIMONY THAT
14 THIS IS THE SAMPLING PLAN. THIS WOULD BE THE SAME CATEGORY AS
15 THE REPORT OF PROFESSOR BALDUS. APPEARS TO BE HEARSAY TESTIMONY
16 AND UNSWORN TESTIMONY IN ANY EVENT.
pe THE COURT: SUSTAIN THE OBJECTION.
15 BY MR. BOGER:
Ww 19 Q. ALL RIGHT, PROFESSOR WOODWORTH. LETS TALK A LITTLE BIT
20 ABOUT THE SAMPLING PLAN. WHAT WAS THE OUTLINE THAT YOU PROPOSED
21 TO FOLLOW OF A PLAN FOR SAMPLING?
22 A. THE OUTLINE WAS, WAS AS FOLLOWS:
23 KNOWING THAT THE PRECISION WITH WHICH ONE CAN ANALYZE A
24 RARE EVENT DEFENDS ON HOW MANY TIMES YOU OBSERVE THAT RARE
23 EVENT. AND THE RARE EVENT IN THIS CASE IS THE DEATH SENTENCE.
WOODWORTH — DIRECT
WE KNEW THAT WE WERE GOING TO HAVE TO OBSERVE ALL OR
SUBSTANTIALLY ALL OF THE DEATH SENTENCES IN ORDER TO GET
ACCURATE RESULTS WITH REGARD TO THAT DECISION. THE SAME IS TRUE
OF DECISIONS TO SEEK A PENALTY TRIAL. BECAUSE THAT IS AGAIN A
COMPARATIVELY RARE EVENT IN THE CONTEXT OF THE LNIVERSE THAT
WERE TALKING ABOUT HERE.
FOR THAT REASON WE DECIDED TO TAKE A HUNDRED PERCENT
SAMPLE OF THOSE TWO STRATA OF THE UNIVERSE,
NOW, THAT LEFT US WITH A QUESTION OF WHAT TO DO WITH
THE OTHER TWO GROUPS, WHICH IS TO SAY MURDER CASES WHICH
RECEIVED A LIFE SENTENCE WITHOUT GOING TO PENALTY TRIAL AND
VOLUNTARY MANSLAUGHTER CASES.
WE DECIDED. WE DECIDED IN THOSE TWO CASES TO TAKE A
RANDOM SAMPLE, AND FOR REASONS OUTLINED BY PROFESSOR BALDUS., WE
DECIDED IT WAS IMPORTANT TO STRATIFY THIS SAMPLE BY JUDICIAL
CIRCUIT 30 AS NOT TO PRODUCE AN UNREPRESENTATIVE SAMPLE BY THE
ACCIDENTS OF RANDOM SAMPLING.
SO WHAT I DID WAS EACH CIRCUIT PRODUCE A SCHEDULE QF
RANDOM NUMBERS, TQ DIRECT THE, TQ DIRECT THE CODERS, THE DATA
LOGGERS, AS TD WHICH CASES TO PICK UP.
@. LET“S ASK YOU TO GO THROUGH A COUPLE OF DOCUMENTS IN
ILLUSTRATING YOUR TESTIMONY.
WHAT IS APPENDIX 1A. PROFESSOR WOODWORTH. WHICH I3
ANNEXED TO THIS?
A. APPENDIX 1A IS SIMPLY A LIST FROM THE DEFARTMENT OF OFFENDER
WOODWORTH ~ DIRECT
1 REHABILITATION SHOWING THE CASES. ALL CASES IN THE, WITHIN THE
2 TIME FRAME OF THE STUDY WHO WERE SERVING TIME FOR MURDER.
3 THAT“S CRIME CODE 1101 THAT I SEE THERE.
@. THIS IS A DOCUMENT YOU OBTAINED FROM THE DEPARTMENT OF
OFFENDER REHABILITATION?
A. THAT’$ CORRECT. AND AS YOU CAN SEE, IT“S BROKEN DOWN BY
CIRCUITS.
GB. WHAT’S APPENDIX 1B?
SARE
s
R
"
Sl
¢
A. EXCUSE ME, I MISSPOKE. THIS ONE IS NOT BROKEN DOWN BY
10 CIRCUITS.
11 AFPENDIX B IS THE SIMILAR DOCUMENT FOR VOLUNTARY
12 MANSLAUGHTER CASES.
13 Q. ALSO, OBTAINED FROM THE DEPARTMENT OF OFFENDER
14 REHABILITATION?
15 A. THAT IS CORRECT.
16 THE COURT: MAKE SURE I UNDERSTAND YOUR TESTIMONY.
17 THIS MULTI-PAGE DOCUMENT WHICH IS RIGHT THICK, WAS EVERYBODY
18 THEN IN THE CUSTODY OF THE DEPARTMENT OF REHABILITATION SERVING
o 19 A SENTENCE FOR MURDER?
gn THE WITNESS: IM NOT SURE WHETHER THEY WERE ALL IN
21 CUSTODY. YOUR HONOR. THEY WERE ON THE TAPES OF DEPARTMENT OF
22 OFFENDER REMABILITATION.
23 | THE COURT: LOT OF THEM.
24 BY MR. BOGER:
29 @2. LETS THEN MOVE TO APPENDIX 3A. CAN YOU IDENTIFY THAT
nN
~
o>
48
0
.
2
0
WOODWORTH ~ DIRECT
DOCUMENT?
A. APPENDIX 3A, CAN YOU GIVE ME THE TITLE?
@. THE RANDOM SAMPLE AND SUPPLEMENTATION SCHEDULE?
A. NOW THIS DOCUMENT IS. IS BASED ON A, ON THE SAME LIST AS
WE“VE SEEN BEFORE IN APPENDIX 1A, FOR EXAMPLE. EXCEPT THIS ONE
HAS BEEN SORTED BY CIRCUIT. YOU CAN SEE THE COLUMN MARKED CKT.
YOU SEE A SERIES OF 17S, THEN THEY TURN INTO 2-S, THEN THEY TURN
INTO 3“S, WHICH, OF COURSE. IS THE ATLANTA CIRCUIT. THIS WOULD Bi
A LIST OF ALL PEOPLE SERVING, ALL PEOPLE ON THE DEPARTMENT OF
REMABILITATION TAPE WHOSE MOST SERIOUS CRIME IS MURDER. THAT'S
THE 1101 CODE YOU SEE STEADILY DOWN THAT LAST COLLIMN.
WHAT 1 DID WAS PRODUCE, WHAT I DID WAS WRITE A COMPUTER
PROGRAM THAT PRODUCED THE COLUMN MARKED SELECT, YOU SEE THAT?
THAT IS THE NEXT TO THE LAST COLUMN.
@. AND WHAT, WHAT DOES SELECT MEAN?
A. SELECT INDICATES TO THE CODERS, EXCUSE ME. SELECT INDICATES
TO THE DATA GATHERERS WHICH CASES THEY SHOULD SELECT FROM THAT
CIRCUIT.
NOW. 25 PERCENT OF THE NUMBERS IN THAT COLUMN ARE 175.
AND THAT WAS THE BASIS FOR OUR 25 PERCENT RANDOM SAMPLE.
THE REMAINING NUMBER IN THAT COLUMN ARE THE DIGITS, 2.
3, 4, 5, ET CETERA.
THE CODERS WERE INSTRUCTED TO, IN CERTAIN CIRCUITS,
TAKE ADDITIONAL CASES BEYOND THE 235 PERCENT RANDOM SAMPLE. IN
THE SERUENCE INDICATED BY THOSE DIGITS.
1215
WOODWORTH — DIRECT
Y
k
| @. WHAT DO YOU MEAN BY THAT EXACTLY? |
A. WELL, PERHAPS I COULD MAKE IT A LITTLE CLEARER BY LOOKING AT
ANOTHER APPENDIX.
IF YOU) LOOK OVER IN APPENDIX ROMAN NUMERAL V, ORDER OF
SELECTION FOR CASE SUPPLEMENTATION, IT’S THREE PAGES, FOUR PAGES
BEFORE THE END OF THIS DOCUMENT.
THE, THESE ARE THE INSTRUCTIONS FOR CASE
SUPPLEMENTATION.
NOW THE PURPOSE OF CASE SUPPLEMENTATION WAS TO GO
P
Y
oo
E
R
SN
E
I
SR
4
SE
-
SE
1
BEY
)
BEYOND THE 2% PERCENT RANDOM SAMPLE TO AUGMENT THE SAMPLE FROM
P
y
FA
S THOSE CIRUITS, FROM WHICH WE DIDNT HAVE A VERY GREAT NUMBER OF
12 OBSERVATIONS. SO THE INSTRUCTIONS HERE IN CIRCUIT ONE WAS TO
13 TAKE CASES TWO THROUGH SEVEN.
14 IF WE TURN BACK TO APPENDIX 3A. WE’LL SEE THE CODERS
13 HAVE BEEN INSTRUCTED TO TAKE JOE COOK BECAUSE HE HAS THE NUMBER
16 non FOLLOWING HIS MAME IN THE SELECT COLUMN. TO PICK UP KENNETH
17 THOMAS, BECAUSE HE HAS THE NUMBER "3", AND 30 ON.
18 @. SO THESE DOCUMENTS WERE PREPARED IN 1980 OR “81 TO INSTRUCT
w 19 THE CODERS?
20 A. THAT’S CORRECT. THESE WERE PREPARED BEFORE ED GATES WENT
21 DOWN TO SUPERVISE THE DATA CODING.
22 @. LET ME BRIEFLY ASK YOU TO IDENTIFY WHAT’S BEEN MARKED AS
23 APFENDIX 3B?
24 A. APPENDIX 3B IS A CORRESPONDING SET OF INSTRUCTIONS REGARDING
25 THE VOLUNTARY MANSLAUGHTER CASES.
1214
WOODWORTH ~ DIRECT
i @. AND WHAT IS AFFENDIX 4A?
2 A. APPENDIX 3A. THAT“S A COMPUTER PROGRAM WRITTEN IN THE BASIC
3 COMPUTER LANGUAGE FOR. SORRY, IT’S IN FORTRAN, WRITTEN FOR THE
PURPOSE OF GENERATING RANDOM NUMBERS TO GUIDE THE SELECTION OF
DEATH SENTENCE CASES FOR THE 23 PERCENT RANDOM SAMPLE.
@. ALL RIGHT. AND 4B, WHICH IS, WHICH HAS TWO PARTS. IF YOU
4
-
&
7 COULD TELL US WHAT BOTH THOSE PARTS REPRESENT? .
8 A. 4B 1S, OKAY, 4B IS THE OUTFUT OF A SIMILAR COMPUTER PROGRAM
J WHOSE PURPOSE WAS TO PRODUCE THE NUMBERS IN THE COLUMN MARKED
8) SELECT IN THE PREVIOUS EXHIBITS. THIS IS THE ONE THAT PRODUCED
11 THE RANDOM SAMPLE OF MURDER-~LIFE CASES.
12 @. SO THESE ARE THE ACTUAL DOCUMENTS THAT PRODUCED THE NUMBERS
13 THAT GENERATED THE CASES?
14 A. YES. WHAT THE PROGRAM DID WAS PRINT QUT EXACTLY WHAT YOU
13 SEE HERE. THEN I TOOK IT AND CUT IT INTO STRIPS AND PASTED IT
16 ON TO THE COMPUTER PAGES SUPPLIED BY THE DEPARTMENT OF OFFENDER
17 REHABILITATION. THAT’S HOW WE PRODUCED APPENDIX 3A. FOR
18 EXAMPLE.
Ww 17 SEE THAT COLUMN MARKED CIRCUIT AND SELECT. THAT
20 ACTUALLY IS A STRIP OF PAPER THATS BEEN PASTED IN THERE, AND IN
21 APPENDIX 4B, IS, IS ALMOST THE PROGRAM THAT PRODUCED THOSE
22 STRIPS.
23 Q. NOW YOU“VE IDENTIFIED APPENDIX 5 FOR US. LET’S 00 AND LODK
og AT APPENDIX 4. THE LAST THREE PAGES OF GW-2,
23 A. APPENDIX &. LET-5 SEE. I“M HAVING TROUBLE LOCATING IT.
4
5
4
8
23
WOODWORTH — DIRECT
RQ. LAST THREE PAGES OF GW-27?
A. OH. APPENDIX &. PARDON ME. RIGHT. APPENDIX & IS. WAS
ACTUALLY PRODUCED BY PROFESSOR BALDUS UNDER MY INSTRUCTIONS.
THIS REPRESENTS THE COMPUTATION OF THE WEIGHTS THAT WERE USED IN
THE ANALYSIS OF THE DATA.
@. DID YOU ANALYZE THE PROCEDURES THAT WERE USED HERE BY
PROFESSOR BALDUS UNDER YOUR DIRECTION?
A. 1 CHECKED THEM LINE BY LINE. HE FOLLOWED MY INSTRUCTIONS TO
THE LETTER.
@. ARE THEY STATISTICALLY CORRECT AND ACCEPTED METHODS OF
WEIGHT ING?
A. THEY ARE STATISTICALLY CORRECT AND ACCEPTABLE.
@. AND IS IT YOUR PROFESSIONAL OPINION THAT THEY RE APPROPRIATE
FOR USE IN THE KINDS OF STUDY THAT ARE BEFORE THE COURT?
A. THE PRINCIPLE OF WEIGHTING BY THE USE OF THE RATIO OF THE
SAMPLE SIZE TO THE SIZE OF THE UNIVERSE IS WIDELY ACCEPTED. YES.
Gl. NOW, LET ME ASK YOU, IN SUM, THIS FINAL SAMPLING PLAN, IS IT
CONVENTION WITHIN THE STATISTICAL PROFESSION WHEN ONE IS TO DO
EMPIRICAL RESEARCH TO DEVELOP SUCH A PLAN?
A. PARDON ME? WOULD YOU REPEAT THE QUESTION?
@. IM SORRY. IS IT CONVENTION WITHIN THE STATISTICAL
PROFESSION WHEN ONE IS TO DO EMPIRICAL RESEARCH INVOLVING
SAMPLING TO DEVELOP THE KIND OF SAMPLING PLAN WE HAVE HERE IN
CW-27
A. YES, CERTAINLY.
fo
e
0
O
N
WOODWORTH — DIRECT
@. AND WAS THIS DEVELOPED FOR LITIGATION OR FOR ACTUAL CONDUCT
OF THE RESEARCH?
A. THE DESIGN OF THIS PLAN WAS NOT INSPIRED BY THE NEEDS OF
LITIGATION. IT WAS INSPIRED BY OUR LIMITED RESOURCES IN DATA
GATHERING. WE HAD TO MAXIMIZE THE AMOUNT OF PRECISION WE COULD
GET FROM THE NUMBER OF HOURS WE HAD AVAILABLE IN GEORGIA.
@. BUT THE ASSEMBLY OF THIS PLAN ITSELF, THE DOCUMENTS THAT ARE
REFLECTED HERE, WHEN DID THAT TAKE PLACE?
A. THE ACTUAL ASSEMBLY OF THIS DOCUMENT TOOK PLACE FOR THE
PURPOSE OF THIS TRIAL, YES.
QA. BUT THE UNDERLYING DOCUMENTS INCLUDED, WERE THEY IN
EXISTENCE AT THE TIME THAT THE SAMPLE WAS DRAWN IN THE CASES?
A. YES. SPECIFICALLY THE DOCUMENT WHICH WAS PREPARED IS MY
FINAL REPORT TO PROFESSOR BALDUS AND MR. GATES IN APPENDIX 2,
TECHNICAL MEMO ON SAMPLING PROCEDURE. THAT IS ESSENTIALLY THE
DOCUMENT WHICH I PRESENTED TO PROFESSOR BALDUS, IT WOULD HAVE
BEEN IN THE SPRING OF “81, I BELIEVE, WITH THE SOLE EXCEPTION OF
FOOTNOTE DONE, WHICH I ADDED FOR CLARITY.
MR. BOGER: YOUR HONOR, AT THIS TIME, I WOULD RENEW THE
OFFER OF GW-2. IT SEEMS THAT PROFESSOR WOODWORTH HAS TESTIFIED
THAT THE GREAT PREPONDERANCE, IF NOT ALL OF THESE DOCUMENTS,
WERE GENERATED FOR THE STUDY3 THEY REFLECT THE PLAN THAT WAS
ACTUALLY EMPLOYED IN THE WAY THE CASES WERE DRAWN.
FOR THAT PURPOSE, IT SEEMS TO ME THEY SHOULD BE
ADMISSIBLE.
Pr
W
w
om
1219
WOODWORTH — DIRECT
MS. WESTMORELAND: YOUR HONOR, I7LL RENEW MY OBJECTIONS
AS PREVIOUSLY STATED AND NOTE THAT PERHAPS CERTAIN PARTS OF THE
DOCUMENT MIGHT BE ILLUSTRATIVE OF THE TESTIMONY GIVEN BUT IN
PARTICULAR THE FIRST PORTION OF APPENDIX 2, SEEMS TO CLEARLY
FALL WITHIN MY PREVIOUS OBJECTION TO THIS WHOLE DOCUMENT.
THE COURT: ILL SUSTAIN THE OBJECTION.
MR. BODER: YOUR HONOR. THEN I MOVE THE ADMISSION OF
ALL PORTIONS OF GW-2, EXCEPT FOR THE PORTIONS THAT I UNDERSTOOD
THE STATE TO OBJECT TO.» WHICH WERE THE INITIAL SAMPLING AND
WEIGHTING PROCEDURE DOCUMENT. DOCUMENT OF FOUR PAGES IN LENGTH,
AND APPENDIX 2,» TECHNICAL MEMO ON SAMPLING. THE OTHER DOCUMENTS
HAVE NO TESTIMONIAL QUALITY AT ALL. THEY ARE THE
COMPUTATIONAL AND RANDOM NUMBERS THAT ARE GENERATED IN THE
CIRCUITS, AND DOCUMENTS WHICH INDICATE WHICH CASES WERE
SELECTED BY THOSE PROCEDURES. I DIDN’T HEAR THE STATE TO HAVE
ANY OBJECTIONS TO THOSE PARTS OF GW-2.
THE COURT: ALL RIGHT.
MS. WESTMORELAND: MY INDICATION MERELY WAS THAT I
OBJECTED PARTICULARLY TO THOSE PARTICULAR PARTS THAT I OBJECTED
TO PREVIOUSLY. YOUR HONGCR.
AND 1 ALSO WOULD QUESTION. I“M NOT CERTAIN, BUT SEEMS
LIKE, 1 DON’T KNOW IF MANY OF THESE DOCUMENTS HAVE PREVIOUSLY
BEEN PRESENTED OR NOT. I“M UNCLEAR AS TO PARTICULARLY THE LAST
PAGE OF THE LAST --
THE COURT: AT LEAST ONE OF THEM MAY BE REDUNDANT,
1220
WNODWORTH —~ DIRECT
NUMBER & MAY BE.
MR. BOGER: THAT MAY BE CORRECT. YOUR HONOR, AND I
DON’T PARTICULARLY PRESS THAT DOCUMENT ON THE COURT.
MS. WESTMORELAND: THAT WAS THE ONE I WAS REFERRING TO.
THE COURT: WHAT EVIDENTIARY VALUE DO THEY HAVE?
MR. BOGER: WELL, YOUR HONOR, I THINK THEY DO SHOW
WHICH CASES WERE SELECTED, ACCORDING TO WHAT PRINCIPLES. WHETHER
THERE WAS CARE TAKEN IN THAT SELECTION. WHETHER IT WAS DONE IN
ACCORD WITH THE VALID AND ACCEPTABLE STATISTICAL TECHNIQUES. AND
I SUPPOSE THEY GIVE ANY FACT FINDER, THE COURT, THE STATE ON
CROSS, THE OPPORTUNITY TO COME IN AND SUGGEST THAT THE SAMPLE
WAS NOT APPROPRIATELY DRAWN.
IF WE HAVE NO QUARREL WITH THE STATE ON THIS MATTER.
OBVIOUSLY WE DON-T NEED TO PUT IT IN. IF THEY’LL STIPULATE THE
VALIDITY AND ACCEPTABLENESS OF THE SAMPLE. WE CAN GO ON.
M3. WESTMORELAND: NQ SUCH STIPULATION, I THINK, IS
PLANNED FOR THIS POINT. YOUR HONOR.
THE COURT: I ADMIT IT.
MR. BOGER: THANK YOU. YOUR HONOR.
BY MR. BOGER:
Mm. PROFESSOR WOODWORTH. YOU INDICATED THAT YOU HELPED AND DREW
THE SAMPLING DESIGN AND THE SAMPLING PLAN. WHAT WAS YOUR NEXT
CONTRIBUTION TO THE CHARGING AND SENTENCING STUDIES?
A. MY NEXT CONTRIBUTION WAS TO TAKE THE DATA AS ENTERED FROM
QUESTIONNAIRES BY THE LABORATORY FOR POLITICAL RESEARCH AT THE
WODDWORTH — DIRECT
UNIVERSITY OF IOWA TO MERGE IT WITH THE DATA PROVIDED BY THE
Py
DEPARTMENT OF REHABILITATION, AND TO ASSEMBLE THOSE DATA IN A
MN
FORM THAT COULD BE PROCESSED BY THE STATISTICAL ANALYSIS SYSTEM
PACKAGE.
@. AND WAS THAT DONE ROUGHLY IN THE SAME WAY YOU TESTIFIED THAT
YOU MERGED THE DATA FILES FROM THE PROCEDURAL REFORM STUDY AND
THE SIMILAR FILE FROM THE DEPARTMENT OF OFFENDER REHABILITATION?
A. YES.
@. AND WERE BOTH OF THOSE MERGERS ACCOMPLISHED ACCORDING TO
oo
SE
E
T
n
C
L
)
T
U
B
E
T
E
Fo
y VALID AND ACCEPTABLE STATISTICAL TECHNIGUES AND PROCEDURES?
P
y
Fa
y A. YES, THEY WERE.
12 MN. AFTER YOU’VE DONE THE MERGER, WHAT WAS YOUR NEXT TASK WITH
13 THE CHARGING AND SENTENCING STUDY?
14 'A. ONCE AGAIN. I WAS IN CHARGE OF PROCESSING ANY UPDATES TO
13 THIS STUDY THAT. PROVIDED BY PROFESSOR BALDUS, AND I WOULD
16 ACCOMPLISH THIS BY. BY WRITING A COMPUTER PROGRAM WHICH
17 IDENTIFIES THE CASE FOR WHICH DATA IS TO BE CORRECTED OR
13 UPDATED, AND ENTERS THE NEW VALLE OF ANY PARTICULAR DATA ITEM.
Ww 1? MN. ARE THERE STANDARD PROCEDURES TO BE USED TQ ACCOMPLISH THAT
20 UFDATING?
21 A. FROM THE POINT OF VIEW OF COMPUTATION. YES, THERE’S A
pa STANDARD PROCEDURE WITHIN THE SAS PACKAGE. AND THATS CALLED THE
23 UPDATE PROCEDURE.
24 @. DID YOU EMPLOY THOSE PROCEDURES IN YOUR UPDATING OF THE
23 CHARGING AND SENTENCING STUDY?
Pr
y »J
Pr
PJ
WOODWORTH — DIRECT
1 A. vES, I PID. AND IN ADDITION, IT“S PRUDENT FOR ANY SITUATION
ps LIKE THIS TO DO WHAT“S CALLED AN ECHO CHECK, THAT IS TO DAY, TO
3 HAVE THE COMPUTER PRINT OUT THE CHANGES THAT IT HAS JUST DONE.
4 1 DID THIS EVERYTIME I MADE AN UPDATE ON THE VARIABLE. 1 HAD
3 THAT UPDATE PRINTED OUT ALONG WITH THE CASE NUMBER. AND THESE I
& VISUALLY CHECKED AGAINST PROFESSOR BALDUS” INSTRUCTIONS.
7 @. IF YOU FOUND ERRORS THERE. WHAT DID YOU DO?
2 A. 1 ALWAYS MAINTAINED A BACKUP FILE. BEFORE I RAN ANY
2? UPDATES, I RAN A COPY OF THE PREVIOUS FILE. IF I DISCOVERED I
10 HAD MADE AN ERROR, I CORRECTED THE ERROR, AND THEN RERAN THE
11 PROGRAM USING THE BACKUP FILE.
12 @. AT SOME POINT DID YOU AND PROFESSOR BALDUS BEGIN TO IDENTIFY
13 THE ANALYTICAL METHOD YOU WOULD USE TO ANALYZE THESE DATA?
14 A. YES, WE DID. SOME OF THESE METHODS WE DEVELOPED EARLIER IN
135 OUR COLLABORATION. OTHERS HAD TO BE DEVELOPED TO DEAL WITH THE
156 ADDED COMPLEXITIES OF THE STRATIFIED SAMPLE.
17 Q. WHAT METHODS DID YQU SETTLE ON FOR USE IN THE STUDY?
1a A. WE SETTLED ON WEIGHTED REGRESSION, ON LOGISTIC REGRESSION.
1 | 1% AND ON CROSS TABULATION METHODS THAT PROFESSOR BALDUS HAS TALKED
20 ABOUT, UNDER THE HEADING OF INDEX METHODS.
21 @. LET ME ASK YOU BRIEFLY TO ADDRESS ALL THREE OF THOSE. WHAT
oe IS WEIGHTED LEAST SQUARES REGRESSION?
23 A. WEIGHTED LEAST SQUARES REGRESSION IS A METHOD FOR ESTIMATING
24 THE AVERAGE, IN THIS CASE. RATE OF 30ME OUTCOME. LETS SAY,
25 FOR THE SAKE OF CONCRETENESS, THE RATE OF DEATH SENTENCING. LEAST
1223
WOODWORTH ~ DIRECT
i SQUARES REGRESSION WOULD HAVE AS ITS PURPOSE TO PRODUCE AN
2 ESTIMATE OF THE AVERAGE DEATH SENTENCING RATE FOR DEFENDANTS
3 HAVING CERTAIN COMBINATION OF INDEPENDENT VARIABLES.
@. ALL RIGHT. ARE THERE ANY STATISTICAL ASSUMPTIONS OF THE
LEAST SQUARES REGRESSION METHOD?
A. THERE ARE A VARIETY OF ASSUMPTIONS.
@. WHAT ARE THEY?
A. THE CHIEF ASSUMPTION IS THAT THE EFFECT OF SAY RACE OF
O
0
~N
0
AA
ob
»
DEFENDANT OR PRESENCE OF A B2 FACTOR IS CONSTANT ACROSS ALL
10 LEVELS OF AGGRAVATION.
11 @. DID YOU JUDGE THAT THAT ASSUMPTION WAS WARRANTED WITH
12 RESPECT TO THESE DATA?
13 A. WE BELIEVE IT’S NOT WARRANTED WITH RESPECT TO THESE DATA.
14 HOWEVER, WE BELIEVE THAT NEVERTHELESS THE REGRESSION
15 CO-EFFICIENTS PRODUCED BY THIS TECHNIQUE ARE INTERPRETABLE.
1&6 THE COURT: ARE WHAT?
37 THE WITNESS: ARE INTERPRETABLE.
ia THE COURT: INTERPRETABLE?
w 1% THE WITNESS: YES, THAT MEANS THEY HAVE STATISTICAL
20 MEANING AND VALIDITY.
21 THE COURT: I UNDERSTAND THE WORD, BUT YOU SWALLOWED
22 THE “ABLE" PART OF THAT WORD.
23 THE WITNESS: EXCUSE ME. BAD HABIT.
24 BY MR. BOGER:
23 @. WHAT DID YOU DO TO COMPENSATE FOR THE PROBLEMS THAT YOU SAW
14
12249
WOODWORTH ~ DIRECT
WITH THE ASSUMPTIONS OF THE WEIGHTED LEAST SQUARES METHOD OF
REGRESSION WITH RESPECT TO THESE DATA?
A. WELL, WHAT ONE DOES WHEN ONE SUSPECTS THE ASSUMPTIONS ARE
NOT TRUE IS TO MAKE SURE THAT THE FAILURE OF THESE ASSUMPTIONS
IS NOT MISLEADING.
NOW WHAT WE DID. AND WHAT I DID, TO A CERTAIN EXTENT ON
MY OWN, WAS TO CONDUCT. WELL, FIRST OF ALL WHAT WE DID, WAS WHAT
PROFESSOR BALDUS CALLS THE TRIANGULATION METHOD. THAT IS TO SAY,
WE USED TECHNIQUES WHICH HAVE SOMEWHAT DIFFERENT SETS OF
ASSUMPTIONS. FOR EXAMPLE, LOGISTIC REGRESSION DOES NOT HAVE
THIS ASSUMPTION OF CONSTANT EFFECT ACROSS ALL LEVELS OF
AGGRAVATION,
NOR DOES THE INDEX METHOD. THE INDEX METHOD INVOLVES
EVEN FEWER ASSUMPTIONS THAN THE METHOD OF LOGISTIC REGRESSION.
IN ADDITION, BOTH PROFESSOR BALDUS AND I CONDUCTED
VARIOUS TESTS TO SEE IF, IF VIOLATIONS OF THE ASSUMPTIONS OF
THESE TECHNIQUES WERE IN THEMSELVES RESPONSIBLE FOR OUR FINDING
OF SIGNIFICANT RACIAL EFFECTS.
©. WHAT DID YOUR TESTS OF THESE QUESTIONS DEMONSTRATE?
lA. THE TESTS THAT I CONDUCTED DEMONSTRATED TO MY SATISFACTION
RACIAL EFFECTS CANNOT BE ACCOUNTED FOR BY ANY VIOLATION OF THE
ASSUMPTIONS,
@. LET ME ASK YOU JUST FOR CLARITY ON THE RECORD, WHY DID YOU
USE THE WEIGHTED LEAST SQUARES REGRESSION IF IT HAD SOME
PROBLEMS, EVEN PROBLEMS THAT YOU SAY YOUVE ACCOUNTED FOR BY
1225
WOODWORTH — DIRECT
1 DIAGNOSTIC TESTS?
“ A. THE REASON WE USED WEIGHTED LEAST SQUARES. AS I SUGGESTED
3 PREVIOUSLY, IS THAT THE REGRESSION CO-EFFICIENT DOES HAVE USEFUL
4 INTERPRETATION. NAMELY. IT IS THE AVERAGE DISPARITY OF, AGAIN,
a LETTER FOCUS ON LET’S SAY THE RACE OF VICTIM REGRESSION
& CO-EFFICIENT, THAT CO-EFFICIENT IS INTERPRETABLE AS THE AVERAGE
7 NUMBER OF FERCENTAGE POINTS DISPARITY BETWEEN WHITE VICTIM AND
8 BLACK VICTIM CASES.
? NOW. WE. WE KNOW OR WERE QUITE CERTAIN THAT IN POINT
10 OF FACT THAT DISPARITY I3 NOT CONSTANT ACROSS ALL LEVELS OF
11 AGGRAVATION, NEVERTHELESS, THE AVERAGE SIZE OF THAT DISPARITY
12 IS A USEFUL PIECE OF INFORMATION.
13 Ql. SO IS THERE ANY DISTORTION IN THE, AS LONG AS ONE KNOWS THE
14 DEFINITION OF WHAT THE WEIGHTED LEAST SQUARES REGRESSION
13 CO-EFFICIENT MEANS, IS THERE ANY DISTORTION IN THE
16 ANALYSIS?
17 A. THERE WOULD BE A DISTORTION IF, IF THE. IF THE EFFECT WOULD
1a NOT BE THERE IF WE USED A MORE CAREFULLY SPECIFIED REGRESSION.
wn 17 @. YOU TESTIFIED THAT YOU USED A SECOND METHOD?
20 THE COURT: ARE YOU ABQUT TO GO INTO SOME QTHER METHOD?
23 MR. BOGER: I AM. YOUR HONOR,
22 THE COURT: EXPLAIN TO ME CONCEPTUALLY IN MORE DETAIL
23 THAN YOU HAVE DONE, WHAT YOU DO IN LEAST SQUARES ASSUMPTION.
24 WHAT MATHEMATICAL STEPS OR LOGICAL STEPS DO YOU GO THROUGH?
23 THE WITNESS: THE STEPS THAT YOU GO THROUGH. YOUR HONOR,
ol
D
s
N
e
rs
pt
_—
_
|
Ld
oO
[Y
s
14
15
WOODWORTH ~ DIRECT
IS FIRST OF ALL TO BEGIN WITH A SET OF INDEPENDENT VARIABLES
THAT YOU WANT TO CONTROL ON.
THEN WHAT HAPPENS MATHEMATICALLY IS THAT THE COMPUTER
WILL SEARCH FOR A SERIES OF WEIGHTS, THE REGRESSION
COEFFICIENTS. TO APPLY TO EACH OF THESE VARIABLES.
THE PRINCIPLE BY WHICH IT DECIDES ON THE SET OF WEIGHTS
THAT IT CONSIDERS MOST APPROPRIATE IS CALLED LEAST SQUARES.
THATS WHY WE CALL IT LEAST SQUARES REGRESSION.
THE PRINCIPLE IS THIS: ONE TAKES. ONE TAKES A WEIGHTED
COMPOSITE OF ALL THE INDEPENDENT VARIABLES AND COMPUTES THE
VALUE OF THAT COMPOSITE IN EACH CASE. AND THEN TAKES THE
DIFFERENCE BETWEEN THAT COMPOSITE AND THE DEPENDENT VARIABLE.
THEN ONE MANIPULATES THESE REGRESSION CO-EFFICIENTS
ALGEBRAICALLY UNTIL YOU MINIMIZE THE SUM OF SQUARES OF THOSE
DEVIATIONS FROM THE COMPOSITE VARIABLE. THE PREDICTOR FART OF
THE MODEL SO TO SPEAK. AND THE ACTUAL DATA.
$0, FOR EXAMPLE. IF WE HAD TWO VARIABLES, LET“S SAY,
LEVEL OF, THAT“S NOT A GOOD ONE, WE'RE BEGGING THE QUESTION
THERE. LET’S SAY WE HAD PRESENCE OF A PRIOR RECORD, AND RACE OF
THE VICTIM. AND LET/S SAY THOSE WERE THE ONLY VARIABLES IN THE
MODEL .
THEN WHAT THE REGRESSION WOULD DO IS TO SEARCH FOR A
WEIGHT TO ATTACH TO PRESENCE OF A PRIOR RECORD AND A WEIGHT TO
ATTACH TD RACE OF VICTIM.
AND LET'S SAY THE WEIGHTS ARE .3 AND .7.
WOODWORTH — DIRECT
1 THE COURT: HOW DOES IT CONDUCT ITS SEARCH?
2 THE WITNESS: IT CONDUCTS IT BY WHATS CALLED SOLVING
THE NORMAL EQUATIONS. I1T-5 A CALCULUS PROBLEM, YOUR HONOR. THE
UNKNOWNS IN THIS EQUATION ARE THE WEIGHTS THAT ARE TO BE
ATTACHED TO THE INDEPENDENT VARIABLES.
3
4
3
e& b WHAT WERE TRYING TO DO IS MINIMIZE A FUNCTION OF THOSE
7 WEIGHTS, SO WHAT ONE DOES TECHNICALLY IS TO TAKE THE DERIVATIVE
8 WITH RESPECT TO THESE UNKNOWNS. SET THOSE DERIVATIVES EQUAL TO
9 ZERO,
10 BY MR. BOGER:
11 Q. PROFESSOR WOODWORTH. IS THERE A WAY OF DESCRIBING WHAT THIS
12 LEAST SQUARES REGRESSION DOES INVOLVING X - Y COORDINATES AS
13 OPPOSED TO THE EQUATIONS YOU'VE TALKED ABOUT NOW?
14 A. WHAT IT DOES WITH RESPECT TO X - Y COORDINATES? WOULD YO)
i= CLARIFY THE QUESTION?
14 @. WELL, MAYBE, IN A VERY LIMITED WAY IN WHICH I UNDERSTAND
17 |REGRESSION, IS SOMETHING YOU AND I HAVE NOT TALKED ABOUT IN
18 TERMS OF X -~ Y COORDINATES. BUT IF ONE SAW A SET OF X ~- Y
wn 19 COORDINATES WITH THE BACKGROUND FACTORS WHICH YOU IDENTIFIED AND
20 LITTLE DOTS IN THE FIELD THERE REPRESENTING THE INSTANCES OR
23 OCCASIONS THE CASES IF YOU WOULD. WHAT IN EFFECT. IN THAT
22 SETTING, DOES LEAST SQUARES REGRESSION TRY TO DO?
23 A. WELL , TO START WITH THE EXAMPLE OF THE SINGLE INDEPENDENT
24 VARIABLE CALLED X, AND A SINGLE DEPENDENT VARIABLE. CALL IT VY.
23 ONE COULD MAKE A GRAPH, AND THE HORIZONTAL AXIS WOULD BE THE
P
y
P
y
fo
e
a
O
e
N
R
B
N
WOODWORTH - DIRECT
VALUES OF X3 ON THE VERTICAL AXIS WOULD BE THE VALUES OF Y. AND
THE DATA POINTS WOULD BE SCATTERED THROUGH SPACE.
NOW WHAT LEAST SQUARES DOES IS ATTEMPT TO FASS A LINE
THROUGH THIS CLOUD OF POINTS. AND THE CRITERION IT USES TO
DECIDE WHERE TO LOCATE THIS LINE IS THAT POSITION SUCH THAT THE
DEVIATIONS OF THE POINTS FROM THIS LINE ARE IN THE AGGREGATE
MINIMIZED.
NOW THE PRINCIPLE OR THE OBJECTIVE FUNCTION AS THEY SAY
THAT’S BEING MINIMIZED HERE IS THE TOTAL OF THE SQUARES OF THOSE
DEVIATIONS.
ALL RIGHT. WHEN WE GET INTO THE BIVARIATE SITUATION. IF
1 COULD BE PERMITTED TO USE A VISUAL AID --
@. OF COURSE, IF YOU‘VE GOT ONE YOU COULD USE?
A. IT’S SITTING BEHIND YOU, COUNSELLOR.
MR. BOGER: LET ME HAVE THIS MARKED FOR IDENTIFICATION
AS GW-3A.
THE COURT: NOW I KNOW WHAT YOU ALL DID OVER THE
WEEKEND.
MR. BOGER: FOR OUR CHILDREN.
THE WITNESS: I HOPE THIS DOESN‘T HAVE TO BE INTO THE
RECORD.
THE COURT: BUT OF COURSE.
THE WITNESS: THIS MODEL THAT IVE MADE HERE REPRESENTS
REGRESSION IN WHICH WE HAVE TWO INDEPENDENT VARIABLES. THIS
PERHAPS IS THE X AXIS, AND THE Z AXIS, RUNNING IN THESE
122%
WOODWORTH - DIRECT
1 DIRECTIONS. AND THE VERTICAL DIRECTION WE HAVE THE Y VALUES. THE
MN
DEPENDENT VARIABLES.
3 NOW. WHAT REGRESSION ATTEMPTS TO [0 IS TO PASS A PLANE
4 THROUGH THESE POINTS IN SUCH A WAY THAT THE DEVIATIONS OF THE
J POINTS FROM THE PLANE IS AS. SMALL AS CAN POSSIBLY BE MADE IN THE
& 5 AGGREGATE SENSE.
7 THE REGRESSION CO-EFFICIENTS ARE THE SLOPES OF THIS
a |PLANE IN THE TWO DIRECTIONS. THE REGRESSION CO-EFFICIENT OF Z
$ |VARIABLE IS THE SLOPE MOVING PARALLEL TQ THE Z AXIS,» AND
10 |THE CO-EFFICIENT OF THE X VARIABLE IS THE SLOPE PARALLEL TO THE
11 |x Axis.
12 NOW. THE WAY IN WHICH THE COMPUTER ACTUALLY DETERMINES
13 |HOW TO LOCATE THIS PLANE 1S DETERMINED BY CALCULUS.
14 |BY MR. BOGER:
15 |@. LET ME ASK YOU, IF YOU WOULD, —-
15 MR. BOGER: WELL. IF YOUR HONOR HAS MORE GUESTIONS, I
17 |WAS GOING TO USE THAT MODEL TO CONTRAST LOGISTIC REGRESSION WITH
18 |LINEAR REGRESSION.
» 15 THE COURT: I DON’T KNOW WHETHER 1 DO OR NOT. LET ME
20 |TRY TO ABSORB WHAT HE SAID. YOU SAY THAT THE CO-EFFICIENT IS
21 |THE SLOPE OF THE PLANE.
22 THE WITNESS: CORRECT. NOW THE PLANE HAS TWO SLOPES.
23 |ONE IN THE NORTH-SOUTH AND ONE IN THE EAST-WEST.
24 THE COURT: I GUESS I°M NOT IMMEDIATELY UNDERSTANDING
23 HOW THAT, IS THAT THE RATIO OF THE SLOPE OR WHAT IS IT? HOW
a
[$8
B
O
1
E
n
RE
10
31.
WOODWORTH ~ DIRECT
DOES IT BECOME A PERCENTAGE?
THE WITNESS: WELL, IN OUR APPLICATION THE VERTICAL
DIMENSION HERE, EACH STRAW HERE WOULD REFRESENT A CASE, A CLASS
OF SIMILARLY SITUATED DEFENDANTS, AND THE VERTICAL LENGTH OF
THIS STRAW WOULD THE PROBABILITY OF THEIR RECEIVING A DEATH
SENTENCE OR THE PROPORTION OF THEM RECEIVING THE DEATH SENTENCE.
THE COURT: ALL RIGHT.
THE WITNESS: AND THEN THE X AXIS MIGHT REPRESENT SOME,
IVE SHOWN IT AS CONTINUOUS HERE, AND, OF COURSE, OURS ARE
DISCRETE. BUT THE X AXIS, THIS MIGHT REPRESENT THE ACCUMULATION
OF ONE SET OF FACTORS, PERHAPS A COMPOSITE INDEX OF AGGRAVATION
AND THE OTHER AXIS MIGHT REPRESENT ANOTHER. ANOTHER VARIABLE.
SO AS MORE AND MORE. AS WE CET TO HIGHER VALUES ON THE
X AXIS, THE SURFACE SLOPES UP. AS WE GET TO HIGHER VALUES ON
THE Z AXIS, THE SURFACE SLOPES UP,
WHAT THAT MEANS IS THAT THE FROPORTION OF CASES
RECEIVING THE DEATH SENTENCE GETS HIGHER AND HIGHER.
THE COURT: ALL RIGHT.
BY MR. BOGER:
@. NOW HOW WOULD YOU CONTRAST LOGISTIC REGRESSION ANALYSIS WITH
ORDINARY LEAST SQUARES, OR WEIGHTED LEAST SQUARES REGRESSION
ANALYSIS?
A. THE PRINCIPAL CONTRAST IS IN WHAT IS ASSUMED BY THE SHAFE OF
THIS SURFACE. IN LOGISTIC REGRESSION ANALYSIS ITS ASSUMED THAT
THE SURFACE CAN‘T CET ABOVE ONE, IN OTHER WORDS. THE PROPORTION
WOODWORTH - DIRECT
OF PEOPLE RECEIVING THE DEATH SENTENCE CAN’T RISE ABOVE A
HUNDRED PERCENT. NOR CAN IT FALL BELOW ZERO PERCENT.
IF THIS WERE A LOGISTIC REGRESSION. THE SURFACE WOULD
LOOK SOMETHING LIKE A SKI JUMP, IT WOULD FLATTEN OUT AND RISE
RATHER STEEPLY AND THEN FLATTEN OUT AGAIN.
NOW THE REASON THAT 1 SAY WE FIND THE LOGISTIC
REGRESSION NEVERTHELESS, SORRY, THE LINEAR REGRESSION
NEVERTHELESS USEFUL IS THE FACT THAT MOST OF THE SKI JUMP DOES
LIKE AN INCLINED PLANE, IT’S JUST GOT A LITTLE CURLY FOOT
OUT HERE AND CURVES UP AHEAD, S60 THAT IN A SENSE, THE LINEAR
MODEL IS A REASONABLE APPROXIMATION OF WHAT THIS SKI JUMP LOOKS
LIKE FOR MOST OF ITS PATH.
@. IN TERMS OF THE DATA IN THESE STUDIES. ARE THE ASSUMPTIONS
OF LOGISTIC REGRESSION APPROPRIATE OR INAPPROPRIATE?
A. WELL, I HAVEN’T GOTTEN INTO THE, WHATS CALLED THE RESIDUAL
STRUCTURE IN THE MODEL. AND THIS IS THE, THIS IS ANOTHER AREA IN
WHICH LOGISTIC IS APPROPRIATE.
NOW WHAT IS MEANT BY THE RESIDUAL IN A REGRESSION IS
THE AMOUNT BY WHICH EACH DATA POINT ACTUALLY DEVIATES FROM THE
SURFACE THAT WE’RE PASSING THROUGH THE DATA POINTS.
NOW THE, THE LEAST SQUARES REGRESSION DOES ASSUME THAT
THE PATTERN OF DEVIATIONS. AS YOU CAN SEE, I DIDNT CUT THESE
STRAWS OFF VERY NICELY AND THERE ARE 30ME DEVIATIONS UP AND DOWN
FROM A SMOOTH, FLAT PLANE HERE, AND THOSE DEVIATIONS ARE
REFERRED TO AS THE RESIDUAL VARIATION.
—~
8)
Io
nN
WOODWORTH ~ DIRECT
NOW, WHAT LEAST SQUARES REGRESSION ASSUMES IS THAT THAT
RESIDUAL VARIATION IS PRETTY MUCH THE SAME OVER ALL THE
SURFACE, WHEREAS A LOGISTIC REGRESSION MAKES THE ASSUMPTION THAT
THAT RESIDUAL VARIATION IS SMALLER, WHEN THE PROPORTION OF DEATH
SENTENCING IS SMALL. LARGER IN THE MIDDLE RANGE, AND. AGAIN
SMALLER WHEN THE PROPORTION OF DEATH SENTENCE IS LARGE.
Q. HOW DO THOSE --
THE COURT: WAIT JUST A SECOND, LET ME MAKE A NOTE OR
TWO.
WHAT WOULD BE AN EXAMPLE OF RESIDUAL VARIATION?
THE WITNESS: WELL. YOUR HONOR, FOR EXAMPLE, IF WE HAD
A GROUP OF CASES WHICH WERE VERY HIGHLY AGGRAVATED. THEN PERHAPS
$9 PERCENT OF THEM WOULD RECEIVE THE DEATH SENTENCE. THERES
SIMPLY NOT MUCH ROOM FOR VARIATION THERE. WHEREAS IF WE HAD A
SET OF MODERATELY AGGRAVATED CASES, 40 PERCENT OF THEM RECEIVED
THE DEATH SENTENCE, THEN THERE’S QUITE A BIT OF ROOM FOR
VARIATION.
ONE COULD HAVE A SAMPLE IN WHICH THE DEATH SENTENCING
RATE MIGHT GO UP TO FORTY-FIVE PERCENT. ANOTHER SAMPLE WHERE
MIGHT GO DOWN TO 3% PERCENT IN THAT CATEGORY, JUST THROUGH
NATURAL STATISTICAL VARIATION. WHEREAS AT THE HIGH RATES,
THERES JUST NOT ANY ROOM FOR VARIATION. IF YOU TOSS A COIN
THATS GOT HEADS ON BOTH SIDES, THERES GOING TO BE NO VARIATION
IN THE PROPORTION OF HEADS YOU’RE GOING TO GET.
SIMILARLY, IF YOU GO DOWN TO THE VERY HIGHLY MITIGATED
®
W
o
W
a
s
B
N
[=
ok
oO
Fa
s
Jo
ab
WOODWORTH —- DIRECT
| CASES. AGAIN THERE'S NOT MUCH ROOM FOR VARIATION. THE RESIDUUAL
VARIATION AROUND THIS SURFACE WOULD AGAIN BE VERY SMALL.
THE COURT: ALL RIGHT. RESIDUAL VARIATION IS THE TERM
YOU-RE USING TO DESCRIBE THE DEGREE OF DEVIATION FROM THE PLANE?
THE WITNESS: YEAH. FROM THE MODEL IN GENERAL.
THE COURT: WE OBSERVED EMPIRICAL, IF YOU WILL,
DEVIATION FROM THE PLANE.
| THE WITNESS: YES, SIR. THE RESIDUAL VARIATION IS A
COMPONENT OF ANY STATISTICAL MODEL. THERE ARE TWO COMPONENTS TO
THE TIME MODEL. THERE’S THE SYSTEMATIC PART, WHICH DESCRIBES
THE RELATIONSHIP. AVERAGE RELATIONSHIP BETWEEN THE INDEPENDENT
VARIABLES AND THE DEPENDENT VARIABLES.
THE OTHER PART IS THE RESIDUAL PART WHICH, YOU KNOW, IN
A STATISTICAL SENSE DESCRIBES HOW THE TYPICAL CASE DEVIATES FROM
THE AVERAGE CASE. HOW INDIVIDUAL CASES DEVIATE FROM THE AVERAGE
CASE.
THE COURT: IF I UNDERSTAND WHAT YOU JUST TESTIFIED TO,
THIS MODEL ASSUMES THAT THE RESIDUAL VARIATION OCCURS AT THE
MID-POINT OF A LINE.
THE WITNESS: THAT THE GREATEST RESIDUAL VARIATION
OCCURS AT THE MID-POINT.
THE COURT: ALL RIGHT. GO AHEAD.
BY MR. BOGER:
2. ARE THERE ANY REASONS WHY THE LOGISTIC REGRESSION MODEL
MIGHT BE APPROPRIATE FOR THE DATA SETS THAT YOU'RE ANALYZING?
WOODWORTH ~ DIRECT
1 A. THE REASON. CHIEF REASON IS THAT IT SEEMS TO REFLECT IN MANY
2 RESPECTS THE FEATURES WE OBSERVE IN THE DATA, WHAT PROFESSOR
3 BALDUS CALLS THE LIBERATION HYPOTHESIS. IS A BUILT-IN FEATURE OF
THE LOGISTIC MODEL.
THE COURT: YOU ASSUMED THE LIBERATION HYPOTHESIS OR
DID YOU PROVE IT.
4
S
&
7 : THE WITNESS: ITS BUILT IN THE LOGISTIC REGRESSION
8 MODEL. YOUR HONOR, BUT WE HAVE ACTUALLY DEMONSTRATED IT IN THE
? INDEX METHOD. WHICH DOES NOT BUILD IN THAT ASSUMPTION.
Q BY MR. BOGER:
it Q. LET ME ASK YOU TO TURN BRIEFLY TO THE INDEX METHOD AND TELL
12 ME WHAT THAT IS?
13 A. WELL, IN THE INDEX METHOD, IN THE INDEX METHOD, ONE WAY
14 OR ANOTHER CONSTRUCTS A COMPOSITE VARIABLE MEASURING THE LEVEL
13 OF AGGRAVATION OF A CASE. WE HAD VARIOUS WAYS OF CONSTRUCTING
14 THESE COMPOSITE VARIABLES, BUT TYPICALLY THESE METHODS INVOLVE
17 EITHER LOGISTIC OR LEAST SQUARES REGRESSION, TO IDENTIFY THE
13 WEIGHTS TO BE ATTACHED TQ EACH VARIABLE IN THE COMPOSITE.
* 19 THEN WHAT YOU DO IS. WHAT YOU WOULD LIKE TO DO 1S TO
20 MAKE A GRAPH SHOWING THE AVERAGE DEATH SENTENCING RATE FOR
21 INCREASING LEVELS OF AGGRAVATION FOR DIFFERENT KINDS OF CASES,
22 LIKE BLACK VICTIM CASES AND WHITE VICTIM CASES.
23 SO WHAT ONE WOULD LIKE TO DO IS TO BE ABLE TO HAVE A GRAPH
24 WITH LEVEL OF AGGRAVATION ON THE HORIZONTAL AXIS, AVERAGE DEATH
3 SENTENCING RATE ON THE VERTICAL AXIS, AND ON THAT GRAFH YOU
WOODWORTH — DIRECT
1 WOULD LIKE TO PLOT HOW THE DEATH SENTENCING RATE RISES AS THE
2 LEVEL OF AGGRAVATION GOES UP.
3 NOW, ONE CAN‘T DO THAT DIRECTLY BECAUSE OUR DEPENDENT
VARIABLE 1S EITHER ZERO. FOR LIFE; OR 1, FOR DEATH. SO IF YOU
MAKE SUCH A GRAPH, IT WONT SHOW ANY SMOOTH TREND.
NOW THE REASON FOR THAT IS THAT THE MODEL. THESE
REGRESSION ESTIMATES THAT WE“RE PRODUCING IN THESE REGRESSION
MODELS ARE NOT INTENDED TO MODEL DECISIONS IN THE INDIVIDUAL
4g
0
N
e
a
ob
CASES. THEY’RE INTENDED TO MODEL THE TREATMENT OF AGGREGATES.
10 FOR EXAMPLE, AN AGGREGATE OF SIMILARLY SITUATED CASES.
11 SO WHAT WE DO IN THE INDEX METHOD IS TO USE A TRICK
12 WHICH IS IN THE STATISTICIAN‘S TOOL KIT. AND THATS WHAT'S
13 CALLED DISCREETIZATION.
14 GIVE YOU AN EXAMPLE OF WHAT DISCREETIZATION MEANS, IT’S
15 WHAT THE CENSUS BUREAU DOES WHEN THEY HAVE GIVE YOU A TABLE
16 BROKEN DOWN BY AGE. FOR EXAMPLE, THEY-LL GIVE YOU THE
17 FOPULATION OF ATLANTA IN FIVE-YEAR AGE INTERVALS LIKE ZERO TO
13 FIVE —— THAT WOULDN’T WORK -- ONE TO FIVE, SIX TO TEN AND SO ON.
R i THEY DON’T GIVE A TABLE OF ALL ONE-YEAR OLDS, ALL TWO-YEAR OLDS
20 AND SO ON. AND THE, THERE ARE TWO REASONS FOR NOT DOING IT.
£1 ONE. THE TABLE WOULD BE TOO LONG, BUT THE OTHER REASON IS WHEN
22 YOU HAVE A MODERATE AMOUNT OF DATA. YOU’RE JUST NOT GOING TO
232 HAVE VERY MANY DNE-YHEAR OLDS OR TWO-YEAR CLDS, SO YOU WON'T GET
24 STABLE STATISTICS.
otal SO WHAT WE DID IS TAKE THIS, THIS LEVEL OF AGGRAVATION
po
S
W
N
T
R
e
e
W
N
iN
RS
pt
J
fe
y G
14
1%
20
WOODWORTH ~ DIRECT
COMPOSITE VARIABLE AND WE CHOPPED IT UP THE WAY THE CENSUS
BUREAU CHOPS UP AGE INTO INTERVALS, AND THEN WITHIN EACH
INTERVAL. WE CAN PRODUCE STATISTICS, BECAUSE WITHIN EACH
INTERVAL WE MIGHT HAVE A HUNDRED CASES. SO DOWN AT THE LOW END
OF THIS SCALE WE‘RE DOING TO SEE MAYBE A HUNDRED CASES AND NO
DEATH SENTENCES IN THEM, SO WE’LL AVERAGE THE DEATH SENTENCING
RATE IN THAT GROUP OF CASES AND PLOT IT ON THE GRAPH.
AND THAT IN ESSENCE IS THE INDEX METHOD.
8. HAVE YOU HAD ANY PRIOR PROFESSIONAL EXPERIENCE IN EMPLOYING
INDEX METHODS?
A. WELL. I CAN ANSWER THAT IN TWO WAYS. ONE WAY, ITS A
ROUTINE SORT OF KNEEJERK RESPONSE FOR PRACTICING STATISTICIANS,
TO DISCRETIZE A VARIABLE.
@. IN OTHER WORDS. THMAT‘S USED A GREAT DEAL?
A. IT’S USED A GREAT DEAL. AND FOR THAT REASON IT“S DIFFICULT
TO DOCUMENT. BECAUSE ITS LIKE DOCUMENTING HOW MANY TIMES
SOMEBODY USES A HAMMER.
ON THE OTHER HAND. THERES ONE SPECIFIC STUDY IN WHICH
A TEAM OF STATISTICIANS ACTUALLY VERY CAREFULLY CONSIDERED THE
USE OF THIS TOOL AND THAT IS IN THE HALOTHANE STUDY IN WHICH I.
PARTICIPATED. SO THE TOOL HAS SO TO SPEAK HAS BEEN STUDIED, AND
ITS PROPERTIES HAVE BEEN STUDIED, AND TO A CERTAIN EXTENT
ENDORSED BY WHAT WE COULD CALL A BLUE RIBEON PANEL OF
STATISTICIANS IN THAT ONE INSTANCE.
@. LET ME ASK YOU, PROFESSOR WOODWORTH, IN SUM, DO YOU HAVE AN
WONDWORTH — DIRECT
EXPERT OPINION ON WHETHER THE METHODS USED FOR ANALYSIS AND
DEVELOPMENT OF CO-EFFICIENTS IN THE PROCEDURAL REFORM STUDY AND
THE CHARGING AND SENTENCING STUDY METHODS YOUVE TESTIFIED TO.
LEAST-WISE REGRESSION, LOGISTIC REGRESSION AND THE INDEX METHOD
ARE VALID AND ACCEPTABLE STATISTICAL METHODS TO ANALYZE THESE
DATA?
A. YES.
Q@. I ASKED YOU WHETHER YOU HAD AN OPINION,
A. YES, 1 HAVE AN OPINION.
@. WHAT IS THAT OPINION?
A. MY OPINION IS ON THE BASIS OF VARIOUS DIAGNOSTIC TESTS THAT
I HAVE RUN THAT THE METHODS EMPLOYED BY PROFESSOR BALDUS ARE
STATISTICALLY VALID AND MOREOVER THAT THE RACIAL EFFECTS HAVE
NOT BEEN PRODUCED BY INCORRECT, BY INVALID STATISTICAL METHODS
OR INCORRECT SPECIFICATION OF THE MODELS.
@. WHAT DO YOU MEAN BY THE WAY, BY INCORRECT SPECIFICAION OF
THE MODEL?
A. WELL. FOR EXAMPLE THE LEAST SQUARES REGRESSION IS
MISSPECIFIED IN THE SENSE THAT IT DOES NOT ALLOW FOR A VARIATION
IN THE SIZE OF THE RACE OF VICTIM EFFECT WITH LEVEL OF
AGGRAVATION.
@. BUT YOUR TESTIMONY IS THAT THAT DOES NOT FRODUCE --
A. MY TESTIMONY IS THAT DOES NOT PRODUCE THE RACE OF VICTIM
EFFECT.
MR. BOGER: ALL RIGHT, YDUR HONOR, AT THIS POINT IM
e]
, B
O
E
R
1238
WOODWORTH ~ DIRECT
GOING TO BE GETTING INTO ANOTHER AREA OF STATISTICAL
SIGNIFICANCE. I NOTE THE TIME. I JUST WANTED TO FIND OUT WHAT
THE COURTS SCHEDULE IS.
THE COURT: YES. WE WILL ADJOURN FOR THE DAY. THE
REASON 1M BEING QUIET FOR A MINUTE. I“M TRYING TO LET WHAT I7VE
HEARD SINK IN TO MAKE SURE I DON‘T HAVE A GUESTION WHILE IT’S
FRESH ON MY MIND BEFORE WE LEAVE THIS SUBJECT.
WHAT ELSE ARE YOU GOING TO COVER WITH HIM?
MR. BOGER: WE/RE GOING TO TALK BRIEFLY ABOUT THE
STATISTICAL SIGNIFICANCE AND THE TESTS THAT PROFESSOR WOODWORTH
EMPLOYED FOR THAT.
WERE GOING TO CO THROUGH BRIEFLY SOME DIAGNOSTIC
TESTS HE RAN TO SEE WHETHER THESE METHODS, THERES ONE OR TWO
TABLES, I THINK, WHICH WILL SHOW THE RESULTS OF THE DIAGNOSTIC
TESTS HE RAN TO SEE IF THESE METHODS DIDN‘T PRODUCE FALSE
DUTCOMES OR IMPRESSIONS.
WERE GOING TO TALK BRIEFLY ABOUT SOMETHING CALLED
R-SGUARE AND WHAT IT MEANS FOR THE STUDY. |
THEN GO THROUGH THE TWO GRAPHS THAT WILL BE BEFORE THE
COURT, OR TABLES, BEFORE, FIGURES I GUESS THEY RE CALLED BEFORE
THE COURTS, AND THATS IT. I SUSPECT WE‘VE GOT AN HOURS WORTH
OF TESTIMONY TOMORROW MORNING ON DIRECT.
THE COURT: IS HE GOING TO TRY TO QUANTIFY THE DATA IN
THOSE TWO CHARTS. OR THE EFFECT?
MR. BOGER: I-“M NOT SURE EXACTLY WHETHER IT CAN BE
1 CALLED QUANTIFICATION. I MEAN. I, AS AN OLD METHODIST I TAKE A
2 LOT OF WHAT HE DOES ON FAITH. BUT I THINK WE WILL TALK IN THOSE
3 FIGURES WHICH ARE INCLUDED AS GW-S.
3 THE COURT: ILL WAIT UNTIL YOU GET INTO THAT, THEN, TO
5 ASK HIM.
6 ALL RIGHT. WE-“LL BE IN RECESS. THEN, UNTIL 9130 IN THE
7 MORNING.
8 MR. BOGERS THANK YOU, YOUR HONOR.
= -
10 (COURT ADJOURNED FOR THE DAY.)
11
12
13
14
15
16
17
19