University of Southern California Dissertations and Theses

Death and politics in the United States Courts of Appeals

(USC Thesis Other)

Death and politics in the United States Courts of Appeals

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

DEATH AND POLITICS IN THE
UNITED STATES COURTS OF APPEALS

by

Arthur H. Auerbach

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(POLITICAL SCIENCE)

May 2007

Copyright 2007 Arthur H. Auerbach

ii
Dedication

To my dear wife, Maria, for her abundant love and patience.

To my wonderful children, Kathryn and Christopher, for their constant enthusiasm.

To my mother, Esther, for her unwavering support of my educational dreams.

Some men see things as they are and say why.
I dream dreams that never were and say why not.
George Bernard Shaw

iii
Acknowledgements

This project would not be possible if it were not for the dedication of
several of the finest educators at the University of Southern California. I would
like to personally thank my dissertation chair, Dr. Howard Gillman, for his
constant support in this endeavor. I would also like to thank Dr. John Jeb
Barnes and Dr. Susan Webb Yackee for all of their never-ending assistance.
Moreover, I would like to thank Dr. Jeff Sellers, Dr. Sheldon Kamieniecki and
Law Professor Erwin Chemerinsky, all of whom offered valuable comments and
assisted in strengthening the overall framework of this project. Finally, I would
like to offer a special thank you to Mehdi Majbouri who greatly assisted in the
regression analysis.

iv
Table of Contents

Dedication ii

Acknowledgements iii

List of Tables vi

Abstract viii

Chapter 1: Introduction 1
Literature Review 4
Hypotheses 23
Research Design Methodology 24
Conclusion 30

Chapter 2: History of Capital Punishment in the United States 33
Public Opinion on Capital Punishment 33
Legal History of Capital Punishment 39
Legislative History of Capital Punishment 64
What Does This History Reveal? 72

Chapter 3: The Coding Continuum 78

Chapter 4: Regression Analysis 92
Data, Methods and Variables 94
Findings and Discussion 102
Conclusions 111

Chapter 5: Institutional Analysis 115
Methodology and Finding 115
Findings and Discussion 123
Conclusions 131

Chapter 6: Hierarchical Analysis 136
Methodology and Findings 137
Findings and Discussion 142
Conclusions 156

v
Chapter 7: Signal Theory Analysis 158
Principal of Signal Theory 159
Data, Methods and Signals 161
Findings and Discussion 173
Conclusions 186

Chapter 8: Conclusion 188

Bibliography 200

Appendices 210
Appendix A 210
Appendix B 214
Appendix C 215
Appendix D 216
Appendix E 217
Appendix F 218
Appendix G 219
Appendix H 220
Appendix I 221
Appendix J 222
Appendix K 223

vi
List of Tables

6.0 United States Supreme Court Death Penalty Decision 143
Making from1980-2004

6.1 Fourth Circuit Death Penalty Decision Making from 144
1980-2004

6.2 Fifth Circuit Death Penalty Decision Making from 145
1980-2004

6.3 Sixth Circuit Death Penalty Decision Making from 147
1980-2004

6.4 Ninth Circuit Death Penalty Decision Making from 149
1980-2004

6.5 Fourth Circuit Death Penalty Decision Making Broken 153
Down By Judicial Political Affiliation

6.6 Fifth Circuit Death Penalty Decision Making Broken 154
Down By Judicial Political Affiliation

6.7 Sixth Circuit Death Penalty Decision Making Broken 155
Down By Judicial Political Affiliation

6.8 Ninth Circuit Death Penalty Decision Making Broken 156
Down By Judicial Political Affiliation

7.0 Signal Theory Table, Six Month Lag Harris v. Vasquez 175

7.1 Signal Theory Table, Six Month Lag Federal Death Penalty 176
Act of 1994

7.2 Signal Theory Table, Six Month Lag Anti-Terrorism and 176
Effective Death Penalty Act of 1996

7.3 Signal Theory Table, Six Month Lag Calderon v. Thompson 176

7.4 Signal Theory Table, Six Month Lag Illinois Moratorium 177
on the Death Penalty

7.5 Signal Theory Table, Six Month Lag Atkins v. Virginia 177

vii
7.6 Signal Theory Table, Six Month Lag- Miller-El v. Cockrell 177

7.7 Signal Theory Table, One Year Lag Harris v. Vasquez 180

7.8 Signal Theory Table, One Year Lag Federal Death Penalty 180
Act of 1994

7.9 Signal Theory Table, One Year Lag Anti-Terrorism and 180
Effective Death Penalty Act of 1996

7.10 Signal Theory Table, One Year Lag Calderon v. Thompson 181

7.11 Signal Theory Table, One Year Lag Illinois Moratorium 181
on the Death Penalty

7.12 Signal Theory Table, One Year Lag Atkins v. Virginia 181

7.13 Signal Theory Table, One Year Lag- Miller-El v. Cockrell 182

7.14 Signal Theory Table, Two Year Lag Harris v. Vasquez 184

7.15 Signal Theory Table, Two Year Lag Federal Death Penalty 185
Act of 1994

7.16 Signal Theory Table, Two Year Lag Anti-Terrorism and 185
Effective Death Penalty Act of 1996

7.17 Signal Theory Table, Two Year Lag Calderon v. Thompson 185

7.18 Signal Theory Table, Two Year Lag Illinois Moratorium 186
on the Death Penalty

7.19 Signal Theory Table, Two Year Lag Atkins v. Virginia 186

7.20 Signal Theory Table, Two Year Lag- Miller-El v. Cockrell 186

viii
Abstract
Political scientists have long suggested that the Supreme Court is affected
by both legal and extra-legal variables. This quandary has spread beyond the
Supreme Court as the high court is only able to address a small fraction of the
cases that are presented to it each year. For this reason, political scientists have
begun to scrutinize the decision making of the United States Courts of Appeals as
the true court of last resort. Some in the political science community suggest the
law itself is the primary motivator for judicial decisions while others argue that
judges base their decisions on their own personal policy preferences.
Institutionalists believe the institutional structure of the courts itself drives
decision making whereas some see the hierarchical structure of the courts as
impacting intermediate appellate decisions. A clear consensus among political
scientists has yet to be achieved.
The goal of this study is to gain a better understanding of circuit court
decision making on capital punishment cases through the use of various
quantitative methods. The four models noted above were examined by looking at
capital punishment cases decided in the United States Courts of Appeals from
1980 to 2004 which were derived from the Sixth, Fifth, Fourth, and Ninth Circuits,
respectively representing the northern, southern, eastern and western regions of the
United States. A logistic regression analysis was conducted that incorporated
variables from the various models of judicial decision making. Several alternative
forms of analysis were also conducted to shed further light on the research

ix
question. Institutional influences were examined by looking at the impact of the
ideological composition of circuit panels. The hierarchical model was tested by
comparing decisional trends of the Supreme Court and circuit courts. Time series
analysis was utilized to examine the hierarchical and legal models and determine
whether signals from the Supreme Court and Congress impact circuit court
decision making. In the end, this study offers a greater understanding of judicial
decision making at the circuit court level on the highly controversial issue of
capital punishment.

1
Chapter 1

Introduction
The execution of Robert Alton Harris was scheduled to take place in San
Quentin Prison on April 21, 1992, at 12:01 a.m. Harris journey toward the death
chamber began 14 years earlier when he and his brother, Daniel, kidnapped two
16-year-old boys from a Jack-in-the-Box parking lot in San Diego, California.
After murdering the two boys in cold-blood for their vehicle that would later be
used in a bank robbery, the Harris brothers were arrested, and Daniel confessed to
the murders. Robert ultimately confessed to killing the two boys after they drove
them to a deserted canyon. The case caught the ire of the public eye for many
reasons, including the callousness of the acts in which Harris shot one boy and
then chased down the other, ordering him to stop crying and die like a man
where-upon Harris shot him. Harris thought so little of the acts that he returned to
the car and ate the boys hamburgers while laughing at his brother for not having
the stomach to eat the boys lunch (Cox 1993, 154-60).
The prominence of the Harris case also had to do with the fact that no death
row inmate in California had been executed in over 25 years. The days and weeks
leading up to the execution date demonstrated the relative unease Californians had
with the concept of capital punishment as scores of newspaper articles were
written challenging as well as supporting the use of the death penalty (see
generally Editorial 1992, A30). As it turned out, this quagmire of public opinion
was not only limited to the public at-large but also found its way into the federal

2
judiciary. By many accounts, the American judiciary possesses an apolitical role
in the American political system. The foundation of this wholly institution is that
the courts are entrusted to interpret and apply the rule of law in a neutral fashion
(see Taibi 1990, 956). The importance of applying the rule of law in a non-
political manner is clearly heightened when the discussion turns toward the issue
of capital punishment. Most Americans cringe at the thought that the rule of law
becomes compromised by partisan politics when the issue carries with it
implications of life and death.
After receiving 11 separate reviews from both the state and federal
judiciaries, the time of Harris execution drew near but not before a battery of legal
challenges took place. On April 20, one day before the scheduled execution, the
Ninth Circuit denied Harris final habeas corpus petition as well as vacated a
temporary restraining order that had previously been issued by United States
District Court Judge Marilyn Hall Patel on the issue of whether the execution by
lethal gas was cruel and unusual. At 6:30 p.m. that same day, a single judge from
the Ninth Circuit ordered a stay of execution for 10 days to examine the claim of
new evidence that Harris brother, Daniel, had shot the victims and not Robert.
Four hours later, ten judges of the Ninth Circuit reinstated Judge Patels temporary
restraining order. A third stay was issued by the Ninth Circuit, and, for a time, it
looked as if Harris life would be spared. The State of California, Office of the
Attorney General, immediately filed an appeal to the United States Supreme Court
(Cox 1993, 160-1).

3
At 11:20 p.m., a mere 41 minutes before the scheduled execution, the
Supreme Court lifted the first stay and later lifted the other two stays at 3:00 a.m.
After the Supreme Courts ruling, Harris was taken to the death chamber and was
strapped in place. Shortly before 4:00 a.m., Ninth Circuit Judge Harry Pregerson
issued a fourth stay of execution to address the cruel and unusual punishment
issue. The Supreme Court interceded once again issuing a firm order lifting the
final stay and ordering that no more stays be issued except by the Supreme Court.
Harris was executed 36 minutes later in the San Quentin death chamber (Cox
1993, 161).
The extraordinary judicial haggling that took place between the liberally-
oriented Ninth Circuit Court of Appeals and the conservatively-oriented Supreme
Court gave pause to many within and outside the legal community about the role
of politics in the judiciary. The Ninth Circuits attempt to keep Harris from
execution smacked of anti-death penalty liberalism at its very best. Conversely,
the rapid conservative response from the Supreme Court left little doubt among
many that politics was clearly at-hand. The vast majority of public law
scholarship on judicial decision making focuses on the United States Supreme
Court. However, the Supreme Court is only able to address a small fraction of the
approximately 8,000 petitions for writs of certiorari filed each year or less than
0.5% of the cases in terms of argument and a written opinion (OBrien 2000, 165;
Benesh 2003, 6). The net result of this situation is that the decisions of the United
States Courts of Appeals are left undisturbed in approximately 99% of the cases

4
and have truly become the court of last resort for many litigants in federal court
(Songer, Sheehan, and Haire 2000, 17).
In light of the clear importance of the circuit court within the American
judiciary, a growing number of scholars have turned their interest toward the
federal intermediate court (see e.g., Klein and Hume 2003; Benesh 2003; Klein
2000; Songer, Sheehan, and Haire 2000; Segal, Songer, and Cameron 1995;
Songer and Haire 1992; Songer and Sheehan 1992; Songer and Davis 1989;
Songer 1987; Songer 1982). This body of scholarship has examined circuit court
decision making within various models such as the attitudinal model, institutional
model, hierarchical model and legal model in an attempt to identify which has
greater significance in certain types of cases. Although scholars have begun to
study circuit court decision making, little has been studied with regard to capital
punishment decisions. The lack of circuit court analysis on capital punishment
litigation is disappointing in light of the pragmatic realities that accompany such
decisions. There can be no greater need for an apolitical rule of law when one
considers the taking of a human life. Yet, examples such as the Harris case, and a
multitude of others, begs the question about the extent to which politics plays a
critical role in death penalty decision making at the circuit court level.
Literature Review
Theoretical Debate
A number of theories have been posited with regard to what motivates
judicial decision making by circuit court judges. The legal model asserts that the

5
decisions of courts are based on the facts of a case in light of the plain meaning of
statutes, constitutional provisions, the intent of the framers and precedent. Thus, if
Article I, Section 10 of the Constitution declares that no state shall pass any law
impairing the obligation of contract then the Court will strike down any that does
so (Spaeth 1995, 297). The legal model views judges as constrained
decisionmakers who will base their opinions on precedent and will adhere to the
doctrine of stare decisis (Wasby 1988, 210). Some scholars have gone as far as to
label the legal model as mechanical jurisprudence because of the process by which
judges reach decisions in a highly structured environment (George and Epstein
1992, 324). Murphy (1964) notes that following legal precedent is a well-
entrenched norm of the deliberative process.
According to Gillman (2000), 486, it may be useful to distinguish between
two different kinds of claims made about the nature of legal influence. The legal
positivist claim sees the law as an external constraint on judges resulting in a
certain amount of judicial conformity to an identifiable rule or norm (see Eskridge
1990; Scalia 1989; Schauer 1998, 1991). Thus, impartial judges and rules of law
can by institutional arrangements and role discipline be shown to have less
variance and less correlation to personal factors of judges in decision making
(Gillman 2000, 486; Kalven 1973, 594). A second legal influence may be termed
postpositivist legalists, which is not about the predictive behavior of judges, but
rather their state of mind. The distinguishing factor with post-positivist decision
making is not that the decisionmaker is obedient to conspicuous rules but rather it

6
is a sense of obligation to make the best decision possible in light of ones
general training and sense of professional obligation. On this view, decisions are
considered legally motivated if they represent a judges sincere belief that their
decision represents their best understanding of what the law requires (Gillman
2001, 487).
The attitudinal model was born from the behavioralist tradition.
Behavioralists focus on nonlegal variables and placed their emphasis on judges as
primarily political actors. The attitudinal model holds that judges decide cases
based on their ideological attitude toward policy outcomes. Justices will always
act to advance their preferred policies regardless of the legal factors such as
precedent, text or legislative intent (Whittington 2000, 606). The seminal writing
on the attitudinal model comes from Segal and Spaeth (1993) in which they note
how the attitudinal model differs from the legal model. Simply put, Rehnquist
votes the way he does because he is extremely conservative; Marshall voted the
way he did because he is extremely liberal (Segal and Spaeth 1993, 32-3, 62-5;
Spaeth 1995, 305). In other words, Supreme Court justices vote as they do
because they want their decisions to reflect their individual personal policy
preferences (Segal and Spaeth 1993, 73). The strength of the attitudinal model is
found in its ability to quantify its results. This is not to say that the legal model
cannot equally utilize empirical testing. In fact, Segal and Spaeth (1999) recently
did so. However, the authors concluded precedent rarely influences United
States Supreme Court justices. Thus, although the legal model was capable of

7
empirical testing, the results showed the weakness in the theory, according to
Segal and Spaeth (Segal and Spaeth 1999, 287).
Recently, Segal and Spaeth (2002) reexamined the conflict between the
attitudinal model and legal model. They focused on the influence that stare decisis
has on Supreme Court decisions. The authors found that justices routinely were
able to avoid supporting precedent with which they disagreed. Although support
for stare decisis was minimal, they found an increase in its use with the least
salient Supreme Court cases. They also found that the decision to overturn
precedent was conditioned on the ideological direction of the precedent being
overturned, in accordance with the attitudinal model. Therefore, Segal and Spaeth
found virtually no evidence for concluding that justices decisions were based on
legal factors (Segal and Spaeth 2002, 310-1). The authors note that the legal
model, attitudinal model and others all attempt to provide explanations of what the
Court actually does. However, only the attitudinal models explanation is well
supported by systematic empirical evidence. The fact that the attitudinal model
has been successfully used to predict the Courts decisions further confirms its
status as the best explanation of the Courts decisions (Segal and Spaeth 2002,
351). This being said, it is important to remember Segal and Spaeths analysis
limits the attitudinal model to the Supreme Court and they have yet to apply it to
lower levels of the judiciary. Segal and Spaeth (1993), 70-1, even note that
appellate courts other than the Supreme Court may be affected by extraneous
factors in that they do not control their own docket, lower court judges may be

8
affected by a desire to rise to higher office and lower court judges are subject to
courts superior to their own. Thus, unlike the Supreme Court, other factors than
personal policy preferences may be at work in the intermediate courts.
New institutionalism has been viewed as largely a reaction to the
attitudinalist model. This approach to judicial politics breaks from behavioralist
approaches in rendering decisions interactive, not just reactive as well as shifting
attention from the individual as the central unit of analysis to the institutional
environment (Whittington 2000, 608). [A]t the heart of the new institutionalism
is a challenge to the reductionist and instrumentalist conception of politics that
characterize behavioralism, and a renewed appreciation of the constitutive and
normative conceptions of politics and the role that institutions play in the latter
(Clayton and Gillman 1999, 30). Simply reducing the Supreme Court to its
personnel does not provide a sufficient understanding of the judicial decision
making. Both internal procedures and the norms of the Court and the external
relationship between the Court and its larger political environment affect judicial
outcomes. Thus, judges are not the only relevant actors, and therefore, they do not
decide cases based solely on their personal policy preferences (Whittington 2000,
608).
New institutionalists are divided between rational choice new
institutionalists and historical new institutionalists. As one might expect, rational
choice new institutionalists begin with the assumption about how political actors
behave. Further, judicial decisions may not simply reflect judicial attitudes, but

9
may also reflect the larger strategic environment. Constraints are not only
imposed by other actors but by the institutional context. In this approach,
institutions are understood primarily as a kind of constraint on individual behavior
(Whittington 2000, 611-2). For historical new institutionalists, the law matters to
judicial behavior. Part of the goal is to understand what the law is understood to
be and how it plays into judicial decisions. Generally, historical new
institutionalists tend to be much more concerned with placing the judiciary and the
law within a larger social and intellectual context that both shape the course of the
law and define the legal meaning. Persistent and distinctive features of the Court
as an institution provide[s] important information that must be added to the basic
knowledge of who the justices are (Whittington 2000, 613). Whittington, citing
an earlier contribution from Howard Gillman notes, that if one is interested in
understanding the factors that influence judicial decision making, it is appropriate
to adopt an understanding of institutional politics that is able to take into account
the possibility that some judicial preferences are shaped or constituted within the
normative terrain of their institutional context (Whittington 2000, 623).
A more traditional institutional approach, not to be confused with new
institutional, stems from the socio-psychological theories that emphasize the
influence on judges behavior of personal attributes, political/legal attitudes and
role orientations These varying schools of thought are said to share a common
stimulus-response mechanism, in that, judges behave the way they do because
their social backgrounds, role orientations and ideologies predispose them to do so

10
(Walker (2003), 251). Small group theorists derive from the socio-psychological
approach and suggest that the [c]ollegial courts . . . are small groups in a face-to-
face relationship that intersect under an obligation to solve a specific problem or
set of problems. Reliable theories and perhaps even raw data about human
behavior in small groups may thus be relevant to the study of the judicial process
(Murphy (1966), 1565; see Epstein and Knight (2000), 630, n. 14). Thus, the
small group setting places behavioral constraints upon the individual and subdues
their inclination toward individuality. Specifically, social pressures exist within
the group for a judge to adhere to the dominant value or position expressed in a
decision (Atkins (1973), 43).
Though direct application of small group theory to the study of judicial
decision making would appear limited, the United States Courts of Appeals
provides for an appropriate venue to test such theories (Atkins (1973), 44) The
internal dynamics of a three-judge panel, especially the intrinsic loneliness of a
dissenting vote, may cause judges to withhold their dissent (Songer (1982), 227;
see also Epstein and Knight (2000), 631 [judges occasionally conform to the
behavior of their colleagues but not necessarily to further their goals; rather the
mechanism seems to lie in the desire to retain friendly relations with colleagues, to
get along.]). Cross and Tiller suggest that politically divided federal appellate
panels may be influenced by a whistleblower effect, wherein the presence of a
politically opposed minority representative means that there is someone on a panel
who can identify the majorities disobedience to legal doctrine, thereby influence

11
decision making. Thus, it appears evident that institutional pressures such as
small group influence suggest additional influences on circuit court decision
making (Cross and Tiller (1998), 2173).
Finally, the hierarchical model analyzes strategic circuit court decision
making within a hierarchical setting. Political scientists refer to the theory that
governs this model as the principal-agent theory. The principal-agent theory
recognizes the relationship between a principal, one who has permitted or directed
another to act for his benefit and subject to his direction, and an agent, one who
acts on behalf of the principal (Songer, Segal, and Cameron 1994, 674).
Typically, economists have utilized the principal-agent theory in an attempt to
understand market activity. However, more recently, scholars have employed the
principal-agent theory to increase the understanding of the political world, and
courts in particular (Songer, Segal, and Cameron 1994, 674). The Supreme Court
is viewed as the principal, who subordinates (Circuit Courts of Appeals) are the
agents. If the Courts of Appeals act as faithful agents, they will obediently follow
the policy dictates set down by the Supreme Court (Songer, Segal, and Cameron
1994, 675).
In addition to the principal-agent theory, another unique theory found
within the hierarchical model questions whether the Supreme Court actually sends
signals and indices to lower courts. Cameron, Segal, and Songer (2000) note that
hierarchical control of organizations is found throughout politics. Congress and
the presidents attempt to control agencies, upper level bureaucracies attempt to

12
control lower levels, and higher courts strive to control lower courts. The
authors test this theory by looking at what role the Supreme Court grants of
review play in an attempt to enforce the doctrinal preferences of the Supreme
Court within the judicial hierarchy. The Supreme Courts actions against a lower
court are therefore seen as direct signals to approve or disapprove of the lower
courts rulings (Cameron, Segal, and Songer 2000, 101).
Empirical Studies on Circuit Court Decision Making
Several studies have been conducted on circuit court decision making
examining the various decision making models. That being said, many significant
gaps in the literature exist requiring a more comprehensive study in this area in
order to gain a more thorough understanding of judicial decisions at the circuit
level.
Songer, Sheenhan and Haire (2000) examined circuit court decisions over
six four-year periods from 1925 to 1988, including over 15,000 decisions. The
two main areas of inquiry were the impact of the political party of the appointing
president and the region where the circuit was located. Each circuit judge
involved in decision making during the relevant time period was profiled based on
their appointing president. Songer et al. recognized the historical tie between the
political party of the appointing president and the partisan goals of the appointed
judge (Songer, Sheehan and Haire 2000, 20-1, 24-8). However, decision making
patterns of federal appellate courts only exercise their discretion in about a third of
the cases with the remaining two-thirds of the cases consisting of easy cases in

13
which the legal text are determinative, regardless of extra-legal factors. The data
examined suggested that party affiliation appeared to account for a substantial
portion of variation in judicial voting (Songer, Sheehan and Haire 2000, 104-5,
118).
The authors also examined four separate regions within the United States
(South, Pacific Northwest, Great Plains-Rocky Mountains and Northwest-Great
Lakes) and several types of cases (civil rights/liberties, criminal, labor and
economic). Although regional differences were seen, especially during the first
period studied, the impact of regional difference diminished over time (Songer,
Sheehan and Haire 2000, 121-3). A multivariate model was run to confirm the
effects of both party and region on circuit court decision making. The results
showed that while partisan effects on the circuit courts has increased over time,
regional differences have declined (Songer, Sheehan and Haire 2000, 141-3).
Although the study conducted by Songer et al. offers clear support for the theory
that circuit court judges follow their personal policy preferences, the study is
extremely limited in that it only examined a single model to the exclusion of all
others. Thus, the need for a broader analysis is clear.
David Klein (2000) conducted a study incorporating both quantitative and
qualitative methods to determine how circuit court judges reacted to opportunities
to make an impact on the law. He focused on the announcement and treatment of
new rules by the Circuit Courts of Appeals on cases from 1983 to 1995 in the areas
of anti-trust, search and seizure and environmental law (Klein 2000, 4, 8). Klein

14
selected cases that announced new legal rules within each of the areas yielding 81
instances in 62 cases. Klein coded each rule as either liberal or conservative based
upon which party favored it (Klein 2000, 46-8). He found that of the 13 cases that
the United States Supreme Court took up, the high court agreed with 9 of the new
rules and only found 4 unfavorable. Thus, the Supreme Court substantially agreed
with lower courts when it came to the issuance of new rules (Klein 2000, 51).
Klein conducted a probit analysis of various variables including:
ideological distance between the rule and median judge; the judge who created the
rule; the expertise of the judge who created the rule; whether any dissent was seen
in the case that issued the new rule; support or opposition from other circuits
regarding the new rule; seniority of the judge who created the new rule; support or
opposition within the circuit for the new rule; and several interactive variables
(Klein 2000, 71-3). Klein found that none of the independent variables of interest
were capable of shifting judges from strong support for a rule to strong opposition.
However, almost all of the variables had the power to change the minds of judges
who were not firmly committed to one position with prestige and experience of the
judge having the greatest impact (Klein 2000, 81-3).
Klein likewise examined the impact of circuit courts anticipation of the
Supreme Court granting review of an issue and fear of reversal. He conducted a
logistic regression analysis with various variables but ultimately concluded that
there was no support for the conclusion that circuit court judges were moved by
fear of Supreme Court review or reversal (Klein 2000, 128-30). In the end, circuit

15
court judges agreed with new rules considerably more than they disagree, the
ideology of the judge was an important factor, favorable treatment of new rules
was be based upon how previous courts treated the issues, prestige and expertise of
the opinion writer impacted the acceptance of new rules, a dissent to a new rule
might impact how the rule is received and circuit court judges acted with very little
regard to Supreme Court action on that rule. The factors that appeared to impact
circuit court judges were more profound for less experienced judges that faced
difficulty in reaching decisions. Thus, this study left little doubt that circuit court
judges do think strategically in their decision making but they do so within certain
limitations (Klein 2000, 133-4, 138-142).
Klein is one of the few scholars who chose to analyze institutional factors
that impact judicial decision making. Although Klein extended his research into
an unexplored area with different methods, the fact remains that Klein looked at a
very narrow question of how circuit court judges respond to the issuance of new
rules. This is not what one might categorize as a mainstream political issue.
Further, Klein looked to individual characteristics of judges in order to determine
which characteristics of judges impact their decision making and did not
systematically analyze the major models of judicial decision making noted in some
of the other studies. Although Kleins study advanced the literature in many
respects, it also left some significant variables unexplored.
Segal, Songer and Cameron (1995) examined circuit court decision making
by empirically comparing the legal, attitudinal and hierarchical models. The

16
critical variables for this study included the facts of the case, Supreme Court
doctrine and the policy preferences of circuit judges on search and seizure cases
from 1961 to 1990. The legal and hierarchical models were tested by assessing
how the appellate court doctrine changed over time. An ideological variable for
each judge was incorporated to test the attitudinal model (see Tate and Handberg
1991; Carp and Rowland 1983; Tate 1981). The authors found that the presence
of certain facts during a search had an impact on whether those searches were
viewed as reasonable. Likewise, the hierarchical variable demonstrated that the
circuit judges change their decision making based on shifts in Supreme Court
decision making on this issue. The attitudinal variables were found to be highly
significant, allowing the model to improve its predictability from 66% to 71.2%
(Segal, Songer and Cameron 1995, 237-40).
Segal et al. concluded that neither the legal model nor the attitudinal model
constituted a complete explanation of the circuit court decision making, and the
hierarchical model appeared to dominate over these alternative explanations. The
authors thus felt the hierarchical model provided a simple and useful starting
place for understanding circuit court decision making in this particular area of the
law. However, it was also clear that circuit courts also consistently responded to
case stimuli with regard to search and seizure cases. Although the authors
believed that lower-court judges decision making is clearly value-driven, decision
making was definitely constrained by institutional rules and procedures (Segal,
Songer and Cameron 1995, 244). In light of these findings, the authors found that

17
various types of variables appear to impact circuit court decision making at the
circuit court level.
The study offered by Segal, Songer and Cameron was undoubtedly a step
forward with the goal to conduct a more comprehensive study on circuit court
decision making. However, this study was not based on a highly charged political
issue such as the death penalty but rather on search and seizure cases which do not
always equate to a political-oriented issue. Furthermore, the study omitted key
variables regarding the impact of institutional factors on the circuit judges. To that
extent, this study can be viewed as limited in nature.
In 1992, Songer and Haire utilized logistic regression to analyze circuit
court decisions on obscenity cases from 1957 to 1990. The independent variables
included the ideology of the judges, the impact of Supreme Court precedent, the
facts of the cases and defenses raised by the defendants (Songer and Haire 1992,
963). The non-integrated model showed that the weakest relationship was found
between ideology of the judge and their decision making. However, the integrated
model revealed that most of the variables in the model played a role in circuit court
decision making. Thus, the authors concluded that judicial votes need to be
understood as the resultant of the interplay of a wide variety of forces that include,
at least, their ideology values, their perception of the clarity and relevance of
precedent . . ., the nature of the arguments . . ., the facts of the case and their
attitudes toward the litigants (Songer and Haire 1992, 975-8).

18
The study conducted by Songer and Haire was clearly more
comprehensive than that offered by Songer, Sheehan and Haire. It tested three
models of judicial decision making (attitudinal, legal and hierarchical) over time
and region. That being said, the Songer and Haire study does have pitfalls. The
authors elected to examine obscenity cases which can be difficult to classify as
libel versus conservative. Second, a perhaps most importantly, Songer and Haire
limited their analysis to a single form of quantitative methods. This restriction
impacts the type of data that was available for consideration. Moreover, the study
was so broad in nature that one is left wondering which variable had the greatest
impact on judicial decision making.
Empirical Studies On Circuit Court Decision Using Alternative Methods
Sunstein, Schkade, and Ellman (2003) conducted a study to examine the
institutional impacts on circuit court panel votes.
1
The authors first inquired
whether circuit court judges voting on issues was correlated to the ideology of the
appointing president. Second, the authors look at whether a judges ideology on
circuit court panels would be amplified if the panel had two other judges
appointed by an appointing president of the same political party. Likewise, they
examined whether a judges ideology would be dampened if the panel had no
other judge appointed by an appointing president of the same political party. The
party affiliation of the appointing president was used as a proxy of political
ideology of judges (Sunstein, Schkade, and Ellman 2003, 2-3). Sunstein et al.

1
The exact methodology used by Sunstein, Schkade and Ellman will be set forth later in this study
as the methods are applied to the case sample used herein.

19
examined 4,488 cases and the votes of 12,464 individual judges in the areas of
abortion, affirmative action, campaign finance, capital punishment, commerce
clause challenges, criminal appeals, disability discrimination, industry challenges
to environmental regulation, piercing the corporate veil, race discrimination, sex
discrimination, and claimed takings of private property without just compensation
(Sunstein, Schkade, and Ellman 2003, 7-8).
Sunstein et al. found support for ideological voting, amplification, and
dampening. The authors not only found many extreme effects of ideological
dampening, they also discovered strong amplification effects in which Democratic
appointees, when sitting with two other Democratic appointees were twice as
likely to vote in the stereotypical liberal fashion as were Republican appointees
who were sitting with two Republican appointees. Although the political party of
the appointing president has been viewed as a fairly good predictor of how an
individual will vote, the authors found that the political party of the president who
appointed the other two judges on the panel was at least as good a predictor of how
individual judges will vote. All-in-all, Democratic appointees showed somewhat
greater susceptibility to panel effects than did Republican appointees (Sunstein,
Schkade, and Ellman 2003, 10-11).
The authors found that when a panel was composed of three Democratic
appointees, they issued liberal rulings in 61% of the cases, whereas panels
composed of three Republican appointees issued liberal rulings only 34% of the
time. Panels with two Republican appointees and one Democratic appointee

20
issued liberal rulings 39% and conversely, panels with two Democratic
appointees and one Republican appointee issued 50% liberal decisions (Sunstein,
Schkade, and Ellman 2003, 4, 10). These numbers clearly emphasized the
importance of the panel composition and those plaintiffs chances often depend on
luck of the draw. The authors also noted some counterexamples from their study
(Sunstein, Schkade, and Ellman 2003, 17-19).
Sunstein et al. also discuss how ideology varied by circuit with the Ninth
and the Second Circuits being most liberal and the Fifth and Seventh Circuits
being most conservative. That being said, the results noted above generally
carried across circuits. The authors found that the data showed a pervasiveness of
collegial concurrence, where a concurrence by a judge who signed the panel
opinion either because he was persuaded by the shared opinion of the other two
judges or because it was simply not worthwhile to dissent. Thus, the presence of
such effects raises doubts about one of the most influential models on judicial
decision making -- the attitudinal model (Sunstein, Schkade and Ellman 2003, 4-
6).
Sunsteins model not only conducted an institutional analysis but did so
across time and region. Although the model offered a unique means of analyzing
institutional impacts on circuit court decision making, it did so to exclusion of
other variables such as legal and hierarchical variables that might assist in
explaining the true driving force in this area. Thus, the limited nature of the study
could benefit from a more elaborate examination.

21
Finally, Songer (1987) tested the hierarchical model by looking at the
impact of the Supreme Court on trends in economic policy making on the United
States Courts of Appeals.
2
Specifically, this study focused on labor and anti-trust
decisions of the Supreme Court and Circuit Courts of Appeals from 1950 to 1977
(Songer 1987, 831). During this period there were a total of 160 Supreme Court
decisions and 4454 appeals court labor decisions. The decisions were classified
either as liberal or conservative and computed the percentage of liberal decisions
made by the Supreme Court in each calendar year for each policy area. He then
determined precisely when and to what degree decisional trends changed in the
Supreme Court (Songer 1987, 832-3). After determining when the Supreme Court
policy changed significantly in each policy area, decisional trends on the courts of
appeals were examined to see whether they changed in the same direction in the
period immediately following the change in the Supreme Court policy. The
periods used to analyze appeals court decisions were lagged one year to assure that
appeals court judges had time to become familiar with changes in Supreme Court
policy. Therefore, the basic test was to determine whether the proportion of liberal
decisions on the courts of appeals changed to a statistically significant degree after
the policy change on the Supreme Court (Songer 1987, 834).
Songer found that the labor decisional trends on the courts of appeals
followed the change in Supreme Court policy. After a change in liberal policy
took place in 1958, the courts of appeals liberal decision making increased by

2
The methodology used by Songer will be replicated and applied to the case sample in the instant
case. Thus, greater detail on the Songers methodology will be provided hereafter.

22
9.1%, a difference that was significant at the .001 level. However, the response
by the courts of appeals following the Supreme Courts return to a more
conservative orientation was somewhat ambiguous. The magnitude of change was
much more modest and the level of significance failed to reach .05 (Songer 1987,
834). Following the Warren courts adoption of liberal antitrust policy, the court of
appeals increased their liberal decision making by 8.5%, which was significant at
the .02 level. The response by the courts of appeals following the turn back to the
right was equally decisive (Songer 1987, 835). Songer found that the partisan
make-up of the judges did not play a role in his finding as both Republican and
Democratically appointed judges moved in similar liberal/conservative directions
after the Supreme Court altered its policy in these areas. Thus, Songer concluded
that the Supreme Court exercised considerable impact on the general trends in
economic policy-making in the United States Courts of Appeals (Songer 1987,
837-9).
Songers innovative approach to testing the hierarchical model
demonstrated, to a degree, the relationship between the Supreme Court and its
circuit courts. However, Songer only examined that one model rather than
conducting a comparison to evaluate which model carries greater weight on certain
issues.
The above literature demonstrates that some valuable work has been
accomplished in terms of evaluating the driving forces behind circuit court
decision making. Yet, at the same time, it is clear that a more comprehensive

23
study is needed based on a highly political issue to truly understand the variables
that affect circuit judges decision making processes.
Hypotheses
The existing research on circuit court decision making provides support for
several basic assumptions about the factors that affect circuit court decision
making. Six hypotheses are set forth below in order to test those assumptions
specifically with regard circuit court decision making on capital punishment cases.
H1: The attitudinal model states that judicial decision making is used by judges a
means of maximizing personal policy preferences. Songer, Sheehan, and Haire
(2000) concluded that this goal can be achieved when judges pursue their
ideological preferences. The first hypothesis asserts that a correlation exists
between a circuit court judges ideology based on the political party of the
president that nominated the judge and the judges decision making in capital
punishment cases.

H2: The legal model theorizes that decisions of judges are based on the facts of a
case in light of the plain meaning of statutes, constitutional provisions and the
intent of the framers and precedent. Songer and Haire (1992) and Segal, Songer,
and Cameron (1995) concluded that case characteristics are thus significant in
terms of circuit court decision making. The second hypothesis states that a
correlation exists between certain case characteristics found in death penalty
decisions and the outcome of those decisions.

H3: The principal-agent theory suggests that lower courts will act as faithful
agents to the higher courts. Songer (1987) found that circuit courts of appeals
followed decisional trends of the Supreme Court with regard to labor and anti-trust
decisions. The third hypothesis maintains that the circuit courts of appeals
decisions on capital punishment cases will follow the decisional trends of the
Supreme Court on this issue.

H4: Institutionalist theory asserts that small group settings place behavioral
constraints upon the individual and subdues their inclination toward individuality.
Sunstein, Schkade, and Ellman (2003) concluded that the ideological make-up of
circuit court panels amplifies or dampen voting on certain issues. The fourth
hypothesis avers that a circuit court judges ideological tendency will be enhanced
if at least one other member of the circuit court panel was appointed from the same
party as that judge. Conversely, the opposite will be found to be true.

24
H5: The principal-agent theory states that lower courts will act on specific signals
from higher courts with regard to judicial decision making. Segal, Songer, and
Cameron (1995) and others have found that a principal-agent relationship exists
between the United States Supreme Court and the United States Courts of Appeals
and that the Supreme Court sends signals to the circuit courts to alter their decision
making. In light of this scholarship, the fifth hypothesis states that circuit judges
will comply with signals from the United States Supreme Court with regard to the
manner that the lower court should resolve capital punishment issues.

H6: The principal-agent theory additionally suggests that courts will act on signals
from other bodies within the American political system. Thus, the sixth
hypothesis asserts that the United States Courts of Appeals will comply with
Congressional signals on death penalty litigation and the partisan support for or
against the death penalty from the governors office of the state where the death
sentence was handed down.

Research Design Methodology
This study examines the attitudinal, institutional, hierarchical, and legal
models with regard to circuit court decisions on capital punishment cases. The
time frame for the study begins in 1980, approximately four years after the
Supreme Court reinstated capital punishment in Gregg v. Georgia , 428 U.S. 153
(1976), and concludes at the end of 2004.
3
Cases were drawn from four circuits.
The selection of circuits was based upon two criteria. First, the circuits were
chosen based upon their location. The circuits include the Sixth, Fifth, Fourth, and
Ninth representing the north, south, east, and west regions of the United States.
Due to the large size of the various circuits, it was not possible to select circuits
that perfectly represent a particular region of the United States. For example, the
Sixth Circuit includes Michigan, a northern state, and Tennessee, a southern state.

3
There is no need to look for cases in the circuits prior to 1980 because few, if any, worked their
way through the state criminal justice system to reach the circuit courts within that time frame.

25
However, the majority of states within each circuit selected for this study
accurately represent that particular region. The second criteria for the selection of
the circuits were that each circuit must contain at least one state that utilizes capital
punishment within their penal systems. As of 2004, the Sixth, Fourth, Fifth, and
Ninth circuits have 341, 375, 625, and 850, inmates on death row, respectively. It
is clear that each circuit possesses at least one state that utilizes the death penalty.
Regression Analysis
4

To test the attitudinal model, logisitic regression was conducted through
the use of STATA. Two dependent variables were utilized in the analysis. The
first dependent variable used the traditional coding practice of classifying circuit
court decisions based on the outcome of the case, either in favor of the state (pro-
death penalty) or in favor of the capital defendant (anti-death penalty).
5
The
second dependent variable used a more nuanced approach, classifying case
outcomes along a continuum running from strong to moderate to minimal anti-

4
A more detailed description of the dependent and independent variables will be discussed in
Chapter 3, The Coding Continuum and Chapter 4, Regression Analysis.
5
The following coding scheme was used:
Pro-death penalty rulings :
All affirmed USDC denials of habeas corpus petitions.
All denials of motions by petitioners for successive petition.
All denials of Certificates of Appeal (COA) from petitioners.
All denials of stays of executions from petitioners.
All denials of motions for reconsideration from petitioners.
All grants of relief for the State (Mandamus).
Anti-death penalty rulings:
All affirmed USDC grants of habeas corpus petitions.
All grants of motions for successive petitions.
All grants of COA from petitioners.
All grants of stays of execution from petitioners.
All grants of motions for reconsideration from petitioners.
All denials of relief for the State (Mandamus).

26
death penalty to minimal to moderate to strong pro-death penalty. The
independent variable used to test the attitudinal model was the personal policy
preferences of the judges. The judges were classified based upon the political
party of the appointing president.
Certain case characteristics were coded in order to test the legal model
through the use of a fact-pattern analysis was done with Segal, Songer, and
Cameron (1995) (search and seizure cases) and Songer and Haire (1992)
(obscenity cases). A total of 8 case characteristics were coded and analyzed. The
characteristics were broken up into crime characteristics and victim characteristics.
Crime characteristics include whether the murders took place during the
commission of a rape, robbery or kidnapping. Brace and Hall (1997) suggest that
crimes of these nature should be viewed as more egregious than other types of
murders and for this reason, one would expect that courts would uphold
convictions for defendants committed of these offenses more often.
Five victim characteristics were also incorporated into the analysis. The
first three of these characteristics involve the vulnerability of the victim. Cases
involving adult female victims, elderly victims (+65 years), and child victims were
coded. Brace and Hall (1993), 17, see the vulnerability of the victims as an
aggravating element to judicial decision making. Likewise, Brace and Hall
(1997), 1218, see the murder of a police officer as another factor that could affect
judges disposition to uphold a capital conviction and sentence. The final victim
characteristic consists of cases in which multiple murder victims occurred. Brace

27
and Hall (1993), 17, identify this characteristic as a feature that should render the
crime more serious than those where this factor is not present and should make
courts more likely to uphold the application of the death penalty.
Independent variables were also be added into the analysis to examine the
institutional and hierarchical influences. The first of these variables involved
coding the political party of the governorship of the state where a death sentence
was handed down. It is theorized that the decision making of the circuit courts
may be affected by the gubernatorial support for or against capital punishment.
Independent variables were also created in order to test the hierarchical model by
regressing the percentages of liberal death penalty decision making by the USSC
versus each circuit from 1980 to 2004 while using various lag periods.
Finally, an independent variable was created to test whether the circuit
courts respond to certain signals sent by Congress to the courts. In 1996, Congress
passed the Anti-Terrorist and Effective Death Penalty Act (AEDPA) to limit the
ability of prisoners to challenge the legality of their convictions. More
specifically, the AEDPA was intended, in part, to limit the ability of federal courts
to use habeas review to delay executions of death row inmates. Under the
AEDPA, stringent deadlines were imposed for the filing of habeas relief in federal
court along with a limitation that only allowing an inmate to file one habeas corpus
petition. Additional habeas petitions could only be filed with the permission of the
federal court (Harvard Law Review 2000). In addition, and perhaps most
importantly, the AEDPA added to the federal habeas corpus statute 2254, subd.

28
(d)(1), language which provided that habeas relief shall could only be granted if
state court decisions were contrary to, or involved an unreasonable application of,
clearly established Federal law, as determined by the Supreme Court of the United
States. According to Circuit Judge Richard Posner, AEDPA reduced federal
judges to potted plants, apparently with the sole job of ensuring that state courts
do not deviate drastically from Supreme Court precedent (Harvard Law Review).
The end result of the passage of the AEDPA should have resulted in a strong
curtailment of circuit court relief on death penalty cases.
Additional Methods Utilized In The Study
The next portion of the study uses additional quantitative methods to distill
what drives circuit court decision making on death penalty cases. Sunstein,
Schkade and Ellman (2003) questioned whether the ideological composition of the
panels impacts judicial decision making on capital punishment cases. After
classifying each circuit judge based upon the appointing president, they looked at
the percentage of liberal decision making by judges individually, by colleagues on
the panel and by the political make-up of the panel. Moreover, the authors
examined what impact, if any, the political make-up of panels had on the decision
making (DDD, DDR, DRR, RRR). The methodology used by Sunstein, Schkade
and Ellman was duplicated for purposes of this study on the case sample selected.
The exact nature of this analysis is set out in Chapter 4 Regression Analysis.
The testing of the hierarchical model was done through two methods.
First, this study looked at decisional trends of the Supreme Court on capital

29
punishment decision decisions and evaluated whether the circuit courts follow the
trends of the Supreme Court. In other words, if the Supreme Court is viewed as
the principal and the Courts of Appeals are seen as its agents, do the agents of the
principal follow the principals decision making trends on the death penalty issue?
Songer (1982) conducted a similar analysis wherein he classified Supreme Court
and circuit court decisions on labor and anti-trust cases from 1950 to 1977 as
either liberal or conservative in nature based upon the outcome of the case. He
then computed the percentage of liberal decisions made by the Supreme Court and
the circuits in each calendar year. Songer then determined when the Supreme
Court policy in both of these areas significantly altered and looked to see if the
circuit courts followed suit. A difference of proportions test (Z) was used to
demonstrate when changes became statistically significant. The same process was
followed in this study to determine whether the circuits have followed shifting
Supreme Court policy on capital punishment decision making. The periods used
to analyze appeals court decisions were be lagged one year to assure that appellate
court judges had time to become familiar with changes in the Supreme Court death
penalty policy.
The second test of the hierarchical model examined whether courts respond
to specific signals from Supreme Court and Congress. As noted in the literature
review, the predominance of research done on the hierarchical model with regard
to Supreme Court-circuit court relations involves the principal-agent theory. The
circuit courts (agents) are permitted or directed to act for the benefit of the

30
Supreme Court (principal). Agents are therefore subject to the direction of the
principal and act on their behalf (Songer, Segal, and Cameron 1995). Cameron,
Segal, and Songer have taken this concept one step farther and suggested that the
Supreme Court sends signals to the lower courts to alter the lower courts decision
making. This study examines whether signal theory can be found in the
relationship between the Supreme Court and circuit courts with respect to death
penalty decisions. This analysis will be used to examine trends in decisions by
circuit courts on death penalty cases for each circuit during the time frame noted
above. Seven specific orders, rulings or legislation from the Supreme Court and
Congress have been identified as signals sent to lower courts. Patterns of decision
making on death penalty cases will be analyzed both before and after the signals to
see if any change in decision making trends has occurred using a difference of
proportions test (Z) to determine whether any changes became statistically
significant.
The alternative methodologies were utilized in the present study to gather
additional data that otherwise might not be obtained through the use of the
regression analysis. In the end, the different methods used will assist in answering
the ultimate question--what forces are driving circuit court judges on decisions that
literally affect life and death.
Conclusion

The contributions of this study are several fold. First, very few studies
have specifically analyzed circuit court decision making on death penalty cases.

31
As noted above, decision making on this issue is extremely valuable for analysis
purposes in that the issue is highly politicized and thus ripe for the infusion of
extra-legal variables impacting judges decisions. It is difficult to imagine a more
political issue for the courts to decide than that of capital punishment. Second,
nearly every study that examines circuit court decisions does so exclusively
through the use of regression analysis. Although this form of methodology does
reveal a vast amount of information, as it has done in this study, regression does
not capture all the data. For this reason, this study has used a triangulation of
methods to assist in gathering additional data that otherwise would not have been
found by exclusively relying regression.
Third, prior scholarship in the area of circuit court decision making that use
regression analysis construct the dependent variable in a dichotomous fashion
based on the outcome of the case. Although such a decision is logical, one must
question whether using a dichotomous dependent variable is simply too broad in
nature, especially with regard to capital punishment litigation. For this reason, this
study utilizes an independent variable continuum for comparison purposes in
which death penalty decisions were placed and analyzed thereafter. It is the hope
of this study that using this more nuanced approach will result in even a greater
understanding of judicial decision making on this crucial issue.
Chapter 2 of this study sets forth a brief overview of the death penalty in
the United States, including the movement of public opinion, significant United
States Supreme Court decisions and the support for capital punishment from the

32
federal and state legislatures over time. Chapter 3 discusses the construction of
the independent continuum variable that was utilized in the regression analysis
for comparison purposes. Chapter 4 sets out the results from the regression
analysis, including the dichotomous outcome analysis, ordered outcome analysis
and the multinomial analysis. Additional quantitative methods are set out in
Chapter 5 (Institutional Analysis), Chapter 6 (Hierarchical Analysis) and Chapter 7
(Signal Theory Analysis). Each of these chapters supplements the findings from
the regression analysis. Finally, Chapter 8 offers the overall conclusions that are
drawn when looking at each of the analysis together. In the end, this study moves
the body of knowledge on circuit court decision making another step closer to fully
understanding what drives the crucial decision makers on matters of life and death.

33
Chapter 2
History of Capital Punishment in the United States

The history of capital punishment in the United States goes far beyond the
basic conflict between those who wish to retain the death penalty and those who
want to abolish it. Rather, the issue of whether the state should have the right to
take a persons life touches many different elements of society. The American
public as a whole appears to have embraced the death penalty since its inception.
However, a closer examination of public opinion over time reveals greater
dissonance than imagined. In addition, the legal history from the perspective of
the United States Supreme Court has likewise experienced a vacillation in terms of
support for this penalty. Lastly, both federal and state legislatures have taken on
much more active roles in terms of support for and against capital punishment. An
examination of public opinion, United States Supreme Court decision making and
the legislative movement on the issue of death penalty might very well hold the
key to discerning the future of this issue.
Public Opinion on Capital Punishment
Attitudes toward the death penalty have been tracked in the United States
through the use of national polls as far back as 1936, some 70 years ago. At first
glance, it might appear that discerning the public support on capital punishment
would be a fairly easy task. Nothing could be farther from the truth. As will be
discussed below, the manner in which the public has been questioned on this topic
has resulted in varying levels of support. Generally, the public has been asked

34
whether they were in favor of the death penalty for someone convicted of murder
(Simon and Blaskovich 2002, 34). It appears that questioning the public in this
manner has created an incomplete picture of public opinion on this issue.
Although the accuracy of past polls has been called into question, one aspect of
public opinion on capital punishment is clear, large portions of society support the
use of this penalty while others find its use nothing less than barbaric.
In 1936, a Gallop Poll reported that 61% of its respondents favored capital
punishment. This number climbed the following year to 65% and to 68% by
1953.
6
(Cullen, Fisher, and Applegate 2000, 13; Vila and Morris 1997, 89-90, 99).
However, beginning in 1956, a steady decline was seen in support for this form of
punishment. The three polls taken by Gallop in the 1950s showed an average fall
of support to 56%. This decline continued in the 1960s until those who favored
capital punishment reached an all-time low of support in 1966 at 42% with 47% of
Americans actually opposing capital punishment and 11% of the respondents
indicating they had no opinion or it depends. The apparent trend against the
death penalty quickly reversed itself beginning in 1967 with support shifting
upward to 54% (Simon and Blaskovich 2002, 34; Cullen, Fisher, and Applegate
2000, 13; Vila and Morris 1997, 127-8). As the 1970s approached, support for
capital punishment remained above the 50% mark. Interestingly, in 1972, the year
the United States Supreme Court ruled that the death penalty, as applied, was

6
According to Simon and Blaskovich, Gallop did not publish poll on support for capital
punishment from 1937 to 1953 (Simon and Blaskovich 2002, 34).

35
unconstitutional, Gallop reported that the public backing in favor of capital
punishment went as high as 60% (Simon and Blaskovich 2002, 34).
Between the moratorium on the death penalty in 1972 and its eventual
reinstatement in 1976, support remained above 60%. The trend in favor of capital
punishment continued from 1976 to 1980. During this period, support for capital
punishment only fell below 66% on two occasions, 1978 was 62% and 1979 was
65% (Simon and Blaskovich 2002, 34). As America shifted to a more
conservative posture in the 1980s with the onset of the Reagan years, so to did
support for capital punishment. With the exception of the first two Gallop Polls
taken in the 1980s (1980 and 1981), every poll taken from 1982 through 1989
found support for the death penalty of at least 70%. Enthusiasm for the death
penalty hit an all-time high in 1994 with an 80% public opinion approval rating.
Since that time, the approval rating for capital punishment has never dropped
below 70% (Simon and Blaskovich 2002, 34; Vila and Morris 1997, 209, 230,
260, 286).
Viewing the above data, one might be inclined to conclude that support for
capital punishment has been and is alive and well in the United States. Evidence
from the polling data clearly shows that the American people appear to have
become significantly more punitive toward criminals beginning the late 1960s.
Several possible explanations have been offered to account for this phenomena,
including: the rising offense rates of the 1960s and the fear of crime it generated;
the politicization of crime and linkage of this issue to a broader concern for a

36
breakdown of law and order; the emergence of racial conflict and the use of
getting tough on crime as a means of appealing to peoples underlying racism
and antipathy toward minorities; the growing lack of confidence in the criminal
justice system; and the movement away from social welfare explanations of crime,
which stress social causes of offending and a lack of offender responsibility, to
individualistic explanations of crime, which places a stress on free choice and just
desserts as a response to breaking the law (Cullen, Fisher, Applegate 2000, 14;
see, e.g., Rankin 1979; Scheingold 1984, 1991; Bohm 1987; Warr 1995a; Beckett
1997).
Although the above data clearly shows strong support for capital
punishment in the United States following the decade of the 1960s, some have
questioned those findings. Cullen (2000) persuasively suggests that one factor that
appears to have affected the legitimacy of the polling data has been the nature of
the question being asked of respondents. As noted above, generally speaking,
respondents were simply asked whether they supported the death penalty for a
person that has committed murder. The problem with a question, as Cullen sees it,
is that it does not give the respondent any sentencing alternative to the death
penalty. In other words, a more properly formed question would inquire whether a
respondent supports capital punishment for someone who commits murder
knowing that an alternative sentence of life without the possibility of parole could
also be imposed. As will be demonstrated below, studies have shown that

37
presenting respondents with alternative sentences does dramatically alter support
for capital punishment.
Bowers, Vandiver, and Dugan (1994) conducted a study to determine the
affect that sentencing options had on support for capital punishment. The results
of these findings were striking. When respondents were given the sentencing
option for a person convicted of murder between the death penalty and life with
the possibility of parole after 25 years, 52.2% of respondent preferred the death
penalty while 38.2% supported the alternative sentence. This level of support for
the death penalty is dramatically lower than any of the traditional polling numbers
between 1973 through the 1990s. When the option of life without the possibility
of parole (LWOP) was given, support for the death penalty fell to 43.1% with
47.7% selecting LWOP as the best option. When the option included the
possibility of parole but added restitution for the victims family, again more
respondents selected the sentencing alternative (49.9%) than the death penalty
(42.8%). Finally, when life without the possibility of parole was combined with
restitution for the victims family, respondent chose this option 60.7% compared to
only 31.6% who preferred capital punishment (Cullen, Fisher, and Applegate
2000, 20). The above data offered by Cullen (2000) clearly demonstrates that
public opinion support for the death penalty is much more complicated than first
imagined. It is not enough to simply ask respondents whether they support capital
punishment for persons who commit murder. Respondents must be given

38
additional information such as sentencing options in order to determine the true
extend of support for the death penalty.
Another issue that appears to impact public opinion support for capital
punishment has to do with the circumstances of the murder. A 1996 poll taken by
Princeton Survey Research Associates/Newsweek reported that 66% of the
respondents who were questions supported capital punishment for a person who
committed murder. However, the respondents were then asked if they endorsed
the death penalty in each case of the following circumstances. When these
circumstances were introduced, the support for the death penalty substantially
declined. Pollsters found that those who favored the death penalty dropped to 56%
when the act of violence was based upon political or ideological believes.
Likewise, support fell to 55% for capital punishment if the convicted murderer was
under the influence of drugs or alcohol at the time of the crime. When respondents
were informed that the murderer was severely abused as a child, the amount of
support for the death penalty dropped 47%. Where the defendant was provoked to
violence by the victim, support for capital punishment fell to 26% and as low as
25% for convicted murderers who were only an accomplice to the person who did
the actually killing (Cullen, Fisher, and Applegate 2000, 16-17).
It is evident that when additional information is given to respondents on the
issue of support for capital punishment, the levels of support can and do shift
significantly. Does this mean that all public opinion polls taken on support for and
against capital punishment since the 1930s should be viewed as somehow skewed

39
or inaccurate? Certainly not. Even with additional information, a significant
trend is clear that the vast majority of Americans do, in fact, support the concept of
capital punishment. However, given certain information, it is also clear that
support for the death penalty is not as overwhelming as is often portrayed. Thus,
in light of the vacillating support for capital punishment, it is evident that
additional research on the issue is justified in order to gain a better understanding
of the role capital punishment plays within society.
Legal History of Capital Punishment
The United States Supreme Court has actively addressed numerous issues
relating to capital punishment. Few would disagree that the most pivotal decision
by the Court with regard to the death penalty took place in 1972 in Furman v.
Georgia,
7
in which the high court ruled that the death penalty, as applied, was
unconstitutional (Latzer 1998, 19). Although Furman is clearly an important case
in death penalty jurisprudence, several key decisions beginning in the 1950s and
1960s need to be discussed in order to properly understand the legal evolution of
capital punishment.
In 1958, the Warren Court decided the case of Trop v. Dulles,
8
a non-
capital decision. The importance of this case was that the court for the first time
defined the standard for determining when a sentence was deemed cruel and
unusual under the Eighth Amendment, a standard that runs through the heart of
nearly every capital punishment decision. Chief Justice Warren, writing for the

7
408 U.S. 238, 92 S. Ct. 2726, 33 L.Ed.2d 346 (1972).
8
356 U.S. 86, 78 S. Ct. 590, 2 L. Ed. 2d 630 (1958).

40
majority, noted that the basic concept underlying Eighth Amendment was nothing
less than the dignity of man. Thus, the Amendment must draw its meaning from
the evolving standards of decency that mark the progress of a maturing society
(Vila and Morris 1997, 102-3).
In the 1960s, the Supreme Court made two significant rulings on capital
punishment cases. These two cases appear to have been a forecasting for what laid
ahead in 1972. In 1963, the Supreme Court denied certiorari in Rudolph v.
Alabama.
9
Interestingly, Rudolph did not actually constitute a decision by the
court. The Supreme Court merely denied certiorari in the case. However, the
Rudolph denial holds significance because it was the first time that several
justices, in a dissent, indicated their desire to consider whether the Eighth and
Fourteenth Amendments of the United States Constitution permitted the
imposition of the death penalty. Justice Goldberg along with Justices Douglas and
Brennan expressed concern whether the death penalty was a proper penalty for a
convicted rapist who had neither taken nor endangered a human life. Prior to the
justices meeting to decide whether to grant certiorari in Rudolph , Justice Goldberg
informed his fellow justices he intended to broaden the issue raised and questioned
whether and under what circumstances the imposition of the death penalty was
proscribed under the Constitution. While Goldbergs dissent and the broader
concern over the death penalty made no difference in that case, the widely
circulated opinion made the legal community well aware that at least three justices

9
375 U.S. 889, 84 S. Ct. 155, 11 L. Ed. 2d 119 (1963).

41
on the Supreme Court where willing to contemplate constitutional challenges to
the death penalty itself (Vila and Morris 1997, 124).
In 1968, the Supreme Court issued a ruling in Witherspoon v. Illinois.
10

The defendant claimed that Illinois law allowed prosecutors to eliminate potential
jurors who opposed or had reservations about the death penalty resulting in juries
that were prosecution-prone and more likely to be biased in favor of the
prosecution in both verdict and sentence (Epstein and Kobylka 1992, 56; Meltsner
1973, 120). In a 6-3 decision, the Supreme Court ruled that the practice of
excluding prospective jurors who had reservations about the death penalty from
capital trials resulted in juries whose sentencing decisions could be considered bias
and was therefore unconstitutional. Although Witherspoon allowed judges to
excuse jurors for cause if those jurors would vote against the death penalty under
any and all circumstances, it also prohibited judges from removing jurors for cause
if they only had a reservation about imposing the death penalty (Latzer 1998, 137-
8; Vila and Morris 1997, 132-3). This ruling appeared to give further support to
the abolitionist cause.
One year before the ruling in Furman, the Supreme Court issued a ruling in
two consolidated death penalty cases, McGautha v. California and Crampton v.
Ohio
11
. Both defendants were convicted of first-degree murder and the sentences
were left up to the absolute discretion of the jury. The defendants suggested the

10
391 U.S. 510, 88 S. Ct. 1770, 20 L. Ed. 2d 776 (1968)
11
The cases were consolidated by the Supreme Court and the opinion issued in 402 U.S. 183, 91 S.
Ct. 1454, 28 L. Ed. 2d 711 (1971).

42
unfettered discretion given to the juries with regard to the sentence allowed for
too much room for whim, caprice, or prejudice to enter into their decisions.
Crampton also challenged the use of a unitary trial in which the same jury
determined the guilt of the defendant and the punishment. The Supreme Court
ruled that the use of standardless juries and unitary trials in capital punishment
cases was not unconstitutional (Epstein and Kobylka 1992, 89). Abolitionist
clearly viewed these decisions as a step backward for their cause, especially after
the progress that appeared to be made in Rudolph and Witherspoon. One member
of the Legal Defense Fund was quoted as saying, After six years of litigation, the
Supreme Court had finally and decisively rejected the two mainstays of the
moratorium strategy (Vila and Morris 1997, 136-7; Meltsner 1973, 244-5).
Although the abolitionist movement appeared to have suffered a set back with the
rulings in McGautha and Crampton, their dreams were finally realized in 1972.
Furman v. Georgia has clearly been the most significant Supreme Court
decision dealing with the death penalty. The 5-4 plurality decision, with each
justice writing a separate opinion, appeared to signal the demise of the death
penalty indicating that the death penalty, as it was applied, was in violation of the
Eighth Amendment. However, the decision did not constitute the death penalty
for capital punishment. Rather, the Court merely indicated that states had to
provide a better structure and more guided discretion to juries for the death penalty
to pass constitutional muster. Only Justices Brennan and Marshall believed that
the death penalty as a penalty was unconstitutional. Justice Douglas took issue

43
with the discriminatory manner it was used against certain segments in society.
Justices White and Stewart found the death penalty was unusual in that it was
wantonly and freakishly imposed. As a result of the Furman decision, more than
600 death row inmates in 32 states had their sentences commuted from death to
life in prison. The Court hinted in its decision that a death penalty statute that
limited jury discretion might be found constitutional (Cochran 2004, 1400-1;
Latzer 1998, 19; Vila and Morris 1997, 141).
The decision in Furman set a multitude of states into action to correct state
death penalty statutes. By the mid-1970s, thirty-four states enacted new statutes
providing for the death penalty (Vila and Morris 1997, 148). The moratorium on
the death penalty was relatively short-lived with the Supreme Court decision in
Gregg v. Georgia,
12
which re-authorized the use of capital punishment. Gregg was
one of five cases that were decided on the same day by the Supreme Court with
regard to the constitutionality of the death penalty. Cases from North Carolina and
Louisiana had mandatory death penalty sentencing schemes. These two statutes
were struck down as unconstitutional. However, Florida, Georgia and Texas
presented three similar death penalty statutes, each of which was upheld by the
Supreme Court. These new statutes offered capital defendants bifurcated trials,
separating the guilt phase from the penalty phase. Each statute also decreased jury
discretion by requiring juries to weigh aggregating versus mitigating
circumstances at sentencing. A jury could not sentence a defendant to death

12
428 U.S. 153, 96 S. Ct. 2909, 49 L. Ed. 2d 859 (1956).

44
without a finding that the aggravating circumstances outweighed the mitigating
circumstances beyond a reasonable doubt. In light of the substantial changes made
to the death penalty statutes of these and other states, the Court ruled in Gregg that
the death penalty was now constitutional under the Eighth Amendment (Cochran
2004, 1400-1; Latzer 1998, 19; Vila and Morris 1997, 161-2).
The year following Gregg, several categorical challenges were made to
various state death penalty statutes. These categorical challenges argued that
under no condition could the death penalty be imposed for certain crimes. One
such categorical challenge took place in Coker v. Georgia.
13
In Coker, the
defendant escaped from a correction institutional, broke into a home, robbed and
raped women in the house. The defendant was arrested and convicted of rape.
Under Georgia law, the jury sentenced him to death. In a 4-3-2 decision, the
Supreme Court agreed with Coker that the death penalty in that situation was
grossly disproportionate and excessive punishment forbidden by the Eighth
Amendment because the rape of the adult female did not involve the taking of the
victims life (Latzer 1998, 61). At that time, Georgia was the only state to allow
for the death penalty in cases of rape. Thus, the impact of the Coker decision did
not affect very many death row inmates. However, Coker was seen as an
important victory for the abolitionists (Schwed 1983, 160) because it appeared
that, based on the Courts ruling, that the death penalty would considered

13
433 U.S. 584, 97 S. Ct. 2861, 53 L. Ed. 2d 982 (1977).

45
unconstitutional for any crime that did not involve the deliberate taking of a
human life (Vila and Morris 1997, 177-8; Paternoster 1991, 101).
In the years following Gregg, the Supreme Court both narrowed the focus
of capital punishment and broadened the range of mitigating circumstances that a
defendant could introduce at trial in an attempt to have his or her life spared. In
the 1978 case of Lockett v. Ohio,
14
Sandra Lockett was the getaway driver during
an armed robbery of a pawnshop in which the store owner was shot and killed by
one of her accomplices. Lockett stood trial, was convicted and sentenced to death
under Ohios aiding and abetting statute. Ohios new death penalty statute
required a death sentence be imposed unless the sentencing judge found one of
three mitigating circumstances, including: that the victims somehow induced the
murder; that the defendant was under duress, coercion, or strong provocation; or
that the defendants offense was the product of psychosis or mental deficiency. As
was the case, none of these mitigating circumstances was found present in
Locketts case (Latzer 1998, 95-9, Vila and Morris 1997, 186; Schwed 1983, 162).
The Supreme Court reversed Locketts death sentence and struck down
Ohios death penalty statute because the Supreme Court believed that capital juries
must be allowed to consider the full spectrum of mitigating circumstances in
deciding the ultimate penalty. This ruling had a profound impact on several newly
created state death penalty statutes. Of the thirty-four states that had reinstituted
the death penalty, only nine state statutes met the Supreme Courts new standards

14
438 U.S. 586, 98 S. Ct. 2954, 57 L. Ed. 2d 973 (1978).

46
for the use of mitigating circumstances. This ruling was again viewed as a
victory for the abolitionists because capital defendants were now allowed to
present all pertinent mitigating factors to the jury creating a greater likelihood that
juries would now vote against the death penalty (Vila and Morris 1997, 186;
Paternoster 1991, 76-7, Schwed 1983, 164).
In 1982, the Supreme Court again issued a ruling based on a categorical
challenge to the death penalty in Enmund v. Florida.
15
As in Lockett, Earl
Edmund drove the getaway car for his two accomplices who shot and killed an
elderly couple during a robbery. Under Floridas felony-murder statute, Enmund
and his accomplices were indicted for first-degree murder and robbery. The
statute provided that any homicide that was committed during the course of
another felony must treat all aiders-and-abettors with equal responsiblity for the
crimes (Vila and Morris 1997, 191; Paternoster 1991, 102; Tushnet 1994, 74). In a
5-4 decision, the high court found the penalty for Enmund was excessive
punishment because of his minor participation in the murder. The Court noted that
only eight states allowed for the death penalty for persons with minor roles in a
felony murder. Thus, the Court concluded that society had generally rejected the
death penalty for this level of participation in a capital crime. It further cited its
language from Coker that the death penalty was generally viewed as excessive for
a participant that does not take a human life (Latzer 1998, 111-9; Vila and Morris
1997, 192).

15
458 U.S. 782, 102 S. Ct. 3368, 73 L. Ed. 2d 1140 (1982).

47
That same year, the Supreme Court issued a ruling in Eddings v.
Oklahoma,
16
in which the Court reaffirmed that the Eighth Amendment required a
capital sentencer to consider any aspect of a defendants character or record or any
of the circumstances of the crime as a basis for a sentence less than death. Eddings
pleaded guilty and was sentenced to death for the shooting of a police officer. The
defendant was only 16 at the time of the crime. During the trial, Eddings
presented evidence that he had a difficult upbringing with his parents getting
divorced, his mother neglected him and his father subjected him to severe physical
punishment. The trial judge refused to consider the mitigating evidence presented
by Eddings regarding his upbringing. Because the sentencing judge refused to
consider the full range of mitigating circumstances, the Supreme Court found
Eddings conviction was in violation of the Eighth Amendment (Vila and Morris
1997, 195).
Not all of the decision making that came out of the Supreme Court
appeared to curb the use of the death penalty. In 1983, United States Supreme
Court Justice Lewis F. Powell gave a speech to a group of appellate judges from
the 11
th
Circuit Court of Appeals. Justice Powell expressed his frustration with the
level of delay in the capital punishment system. He noted that at that time, there
were more than 1,000 convicted murders on death row which was an intolerable
situation. He blamed the system of repetitive review and last minute stays of
execution for the large backlog. Although Justice Powell never classified himself

16
455 U.S. 104, 102 S. Ct. 869, 71 L. Ed. 2d 1 (1982).

48
as strongly in favor or against capital punishment, he gained a reputation by
others as a fervent partisan regarding the constitutionality of the death penalty
(Vila and Morris 1997, 198; Jeffries 1994, 409).
Justice Powells pro-death penalty stance appeared to carry through to the
rest of the court in 1983 with Barefoot v. Estelle.
17
Thomas Barefoot was
sentenced to death in the state of Texas for the murder of a Texas police officer.
The court addressed both procedural and substantive issues. As to the procedural
issue, the high court found that that a federal appeals courts use of a single
summary decision on both the merits of the defendants claim and his request for a
stay of execution was proper, thus clearing the way for widespread use of
expedited federal review. Barefoots substantive claim alleged the psychiatric
testimony admitted on behalf of the prosecution regarding Barefoots strong
likelihood of future dangerousness during the penalty phase was not entirely
unreliable, and thus admissible. The Supreme Court noted while the American
Psychiatric Association had found that psychiatric testimony regarding future
dangerousness was often unreliable, the defense was free to call witnesses to
counter the prosecution. Thus, Barefoot has been seen as a loss for abolitionists on
two counts (Vila and Morris 1997, 199-200; Paternoster 1991, 79).
In 1986, the Supreme Court addressed the critical issue of whether states
were permitted to execute the mentally insane. Alvin Ford showed no signs of
mental illness at the time he committed the murder. However, in 1982, eight years

17
463 U.S. 880, 103 S. Ct. 3383, 77 L. Ed. 2d 1090 (1983).

49
after being placed on Florida death row, Ford developed an apparent mental
disorder characterized as resembling paranoid schizophrenia. Fords attorney
attempted to have his death penalty sentence reversed arguing his client was
incompetent for execution. The Supreme Court in Ford v. Wainwright
18
accepted
certiorari of the issue and held that the execution of the insane was indeed
unconstitutional under the Eight Amendment, based on the nations common law
heritage in which the execution of the insane has been considered savage and
inhuman (Latzer 1998, 213; Vila and Morris 1997, 210).
The following year, the Supreme Court took up the issue of whether the
prosecution could present victim-impact statements to a jury during the penalty
phase in Booth v. Maryland.
19
John Booth was convicted and sentenced to death
in 1983 for the murder of an elderly couple during a break-in robbery of their
home. During the penalty phase, the prosecution was permitted to admit
statements from the victims family describing the severe emotional impact of the
murders on the family along with the family members perceptions of the crimes
and of Booth. In a 5-4 decision, the Supreme Court ruled that the introduction of
victim-impact statements violated the Eight Amendment. According to the Court,
such information was irrelevant to a capital sentencing decision and its admission
created a constitutionally unacceptable risk that a jury would impose the death
penalty in an arbitrary and capricious manner (Vila and Morris 1997, 215-6).

18
477 U.S. 399, 106 S. Ct. 2595, 91 L. Ed. 2d 335 (1986).
19
482 U.S. 496, 107 S. Ct. 2529, 96 L. Ed. 2d 440 (1987).

50
One of the more controversial death penalty rulings by the Supreme Court
came in 1987 with McClesky v. Kemp.
20
The issue in McClesky was whether the
Georgia capital sentencing scheme was administered in a racially discriminatory
manner. In support of this claim, McClesky produced a statistical study of capital
sentencing procedures conducted by Professor David Baldus and his colleagues.
The Baldus Study, as it was known, controlled for 230 variables that might
explain the discrepancies in sentencing on nonracial grounds. Baldus study
offered proof of discrimination against black murder victims. Murderers of black
victims were statistically less likely to get the death penalty than murderers of
white victims (Letzer 1998, 231). Specifically, the study found that defendants
who killed white victims were 4.3 times as likely to receive a death sentence as
those charged with killing blacks victims. The model also showed that black
defendants were 1.1 times more likely to receive the death penalty as other
defendants (Vila and Morris 1997, 218-9).
Although the Court accepted the statistical validity of the study, the Court
rejected McCleskys claim noting the statistics were still insufficient to support an
inference that any of the decisionmakers in his case acted with discriminatory
purpose. Justice Powell, writing for the majority, explained that a successful
Fourteenth Amendment challenge must prove purposeful discrimination against
a defendant. The Baldus study did not accomplish this requirement. Moreover,
the Baldus study did not establish that race had been a factor in McCleskys

20
481 U.S. 279, 107 S. Ct. 1756, 85 L. Ed. 2d 262 (1987).

51
particular case. Thus, the Supreme Court rejected the challenge and effectively
closed the door on future challenges to the death penalty on the grounds of racial
discrimination short of direct evidence that discrimination was being used in the
prosecution of certain defendants (Letzer 1998, 232; Vila and Morris 1997, 219).
The next year, the Supreme Court first attempted to resolve the issue of
whether it was constitutional to execute juveniles in Thompson v. Oklahoma.
21

William Thompson was 15 years old when he actively participated with three
older friends in the 1983 murder of his former brother-in-law, who had abused
Thompsons sister. Thompson was tried as an adult, convicted and sentenced to
death. In a 5-3-1 opinion, the high court held that the Eighth Amendment did
indeed prohibit the execution of an individual who was under the age of 16 years
of age at the time of the offense. Justice Stevens, writing for the majority, pointed
out (1) that 18 states had set a minimum age of 18 for the death penalty, (2) other
countries of Anglo-American heritage and leading members of the Western
European community appeared to share the opinion that executing persons under
the age of 16 would offend civilized standards of society and (3) modern juries
were typically reluctant to impose the death penalty on persons under the age of
16. Justice Stevens also emphasized the importance of youth as a mitigating factor
in determining a defendants level of culpability. Thus, Justice Stevens concluded
that the standards of decency in society dictated that society could not execute a
minor under the age of 16 (Vilas and Morris 1997, 232-3).

21
487 U.S. 815, 108 S. Ct. 2687, 101 L. Ed. 2d 702 (1987).

52
A year later the Supreme Court addressed the question of whether the
Constitution prohibited the execution of any person under the age of majority. In
Stanford v. Kentucky,
22
Kevin Stanford was 17 years old when he and an
accomplice raped and sodomized a gas station attendant. The victim was then
taken to a secluded area where she was shot in the face and the back of the head.
In a separate case that was consolidated with the Stanford case, 16-year-old Heath
Wilkins stabbed to death a 26-year-old mother of two who was working at a
convenience store. Both Stanford and Wilkins were tried respectively as adults,
convicted and sentenced to die. Justice Scalia noted the question of whether the
Eight Amendment prohibited the execution of 16 and 17 year olds depended on the
evolving standards of decency in society. In other words, had society
abandoned the idea of executing people older than 15 years old but who had not
yet reached the age of majority? The Supreme Court rejected the argument of the
two youths, indicating that neither a historical nor a modern societal consensus
forbade the imposition of capital punishment on individuals who committed
murder at the age of 16 or 17 (Latzer 1998, 203-9; Vila and Morris 1997, 235-6).
Another extremely contentious issue was addressed in 1989, that being,
whether the Constitution allowed for the execution of the mentally retarded. In
Penry v. Lynaugh,
23
the defendant was convicted and sentenced to death for the
1979 rape and murder of a woman in her home. Although Penry was 22 years old
at the time of the crime, pretrial testing established he had the mental age of a 6 ½

22
402 U.S. 361, 109 S. Ct. 2969, 106 L. Ed. 2d 306 (1989).
23
492 U.S. 302, 109 S. Ct. 2934, 106 L. Ed. 2d 256 (1989).

53
year old child with the social maturity of a 9 or 10 year old. His IQ was between
50 and 60, which indicated mild to moderate retardation (Latzer 1998, 219). The
high court reversed Penrys conviction on the ground that the jury had not been
properly instructed that the defendants mental retardation could have been
considered as a mitigating factor. However, the real importance of the decision
was that the Supreme Court rejected Penrys claim that the Eighth Amendment
categorically prohibited the execution of the mentally retarded. Justice OConnor
did indicate that the Eighth Amendment would likely prohibit the execution of
profoundly or severely retarded persons who were unable to understand or
appreciate the wrongfulness of their actions. In the case of Penry, the jury found
that he was competent to stand trial and rejected his insanity defense. The Court
also rejected Penrys claims that a national consensus had been reached in the
United States disallowing the execution of retarded persons (Latzer 1998, 219-29;
Vila and Morris 1997, 241).
The Supreme Court continued to address various issues relating to capital
punishment in the 1990s and, in several cases, changing previous positions decided
by the Court. In 1991, the Supreme Court attempted to address some of the legal
procedural problems that had been created with the death penalty in Coleman v.
Thompson.
24
By the early 1990s, some 10,000 habeas corpus petitions were being
filed with the Supreme Court each year by capital defendants. In order to curb the
growing number of constitutional appeals, the Court began to impose stricter limits

24
501 U.S. 722, 111 S. Ct. 2546, 115 L. Ed. 2d 64 (1991).

54
on their admissibility (Vila and Morris 1997, 260; Regoli and Hewitt 1996, 503).
Roger Coleman had been convicted and sentenced to death in Virginia for the rape
and murder of his sister-in-law. After his appeal was rejected, Coleman raised
several new claims in his habeas corpus petition. The County Circuit Court in
Virginia rejected his claims and Coleman appealed the decision to the Virginia
Supreme Court. However, Colemans appeal was three days beyond the 30 day
period in which to file such an appeal. As a result, the Virginia Supreme Court
rejected Colemans habeas petition on procedural grounds (Vila and Morris 1997,
260-1).
The Supreme Court granted certiorari to determine the constitutionality of
Virginias procedural rule. In a 6-3 decision, the high court held that the Virginia
Supreme Court did not err in rejecting Colemans claims. The federal courts are
required to follow the procedural guidelines laid out by independent and adequate
state procedural grounds. Colemans only hope in the case was to show that his
attorneys were ineffective by failing to file the appeal on time. Coleman did not
make such as a claim. Thus, Colemans execution was significant because he was
never allowed to present his constitutional claims in court (Vila and Morris 1997,
260-1; Tushnet 1994, 106). Some have claimed that Coleman was literally hung
on a technicality (Vila and Morris 1997, 261; Kaplan and Cohn 1992, 56).
In 1991, the Supreme Court readdressed an issue it had decided in the
1980s, whether it is constitutional for the prosecution to admit victim impact
statement during the penalty phase of a capital trial. Pervis Payne had been

55
convicted and sentenced to death for the murders of a young woman and her two-
year-old daughter. In Booth v. Maryland (1987), the Supreme Court had already
ruled that it was unconstitutionally permissible to admit victim impact statements
at the penalty phase of a capital trial due to the possibility that such evidence
would result arbitrary discrimination against a defendant. Four years later, the
Court in Payne v. Tennessee (1991) reversed its prior decision holding that the
Eighth Amendment did not, after all, prohibit the presentation of victim impact
statements during a capital trial. Chief Justice Rehnquist explained that a capital
defendant was allowed to present all relevant mitigating evidence during the
penalty phase, thus prohibiting the prosecution from offering victim impact
evidence to counter the mitigating evidence gave the defense an unfair advantage
over the prosecution (Latzer 1998, 169-71; Vila and Morris 1997, 263-4).
One of the most controversial decisions of the 1990s, if not from any
decade, came in the decision of Herrera v. Collins,
25
, when the Supreme Court in a
6-3 decision held that a claim of innocence based on newly discovered evidence
was not a ground for federal habeas corpus relief. Many people around the
country were outraged by such a ruling in which was perceived by some that the
Supreme Court was sanctioning the possible execution of an innocent person.
Leonel Herrera had been convicted for the murders of two Texas police officers
and sentenced to death in 1982. The evidence against Herrera was overwhelming
in both cases, including the fact his Social Security card was found at the scene of

25
506 U.S. 390, 113 S. Ct. 853, 122 L. Ed. 2d 203 (1993).

56
the first murder, blood from the first victim was on his jeans, his wallet and the
car that he had been driving, a handwritten note implying his guilt had been found
on him when he was arrested, and he had been identified by the second victim
before the victims death. Herrera pleaded guilty to the first murder after being
convicted and sentenced to death in the second murder (Vila and Morris 1997,
274).
Herrera unsuccessfully challenged his conviction and sentence on appeal.
Ten years later, he filed a second habeas corpus petition alleging his innocence and
that his now dead brother had instead committed the crimes. Herrera offered
affidavits from his brothers lawyer and one of his brothers former cellmates, both
of whom indicted Herreras brother told them he committed the murders. An
affidavit was also offered from the son of Herreras brother who indicated he
witnessed his father shoot the two police officers. Chief Justice Rehnquist wrote
that the purpose of a federal habeas corpus was to ensure that individuals were not
imprisoned in violation of the Constitution but not to correct errors of fact. Justice
Rehnquist noted that Herrera was not without a forum to resolve his factual
discrepancy regarding his claim of innocence. Executive clemency provides the
ultimate fail safe mechanism in the American criminal justice system. In
addition to this portion of the ruling, Justice Rehnquist also noted that it did not
find Herreras new claim of innocence to be credible (Vila and Morris 1997, 274-
275).

57
Although the ruling in Herrera may have been legally sound, many in the
United States considered the decision unconsciousable and began questioning the
credibility of the capital punishment system as a whole. One of the most notable
critiques of this system sat on the Supreme Court itself. In Callins v. Collins,
26

Justice Harry Blackmun gave some insight into his concerns over the
implementation of the death penalty in the United States. In the dissenting opinion
to the denial of certiorari, Justice Blackmun espoused that he would no longer
tinker with the machinery of death. He noted that for 20 years he had worked to
develop the procedural and substantive rules that would lend more than the mere
appearance of fairness to the death penalty. Rather than continue to coddle the
Courts delusion, Justice Blackmun felt morally and intellectually obligated to
concede that the death penalty experiment had failed. The problem, according the
Justice Blackmun, was the inevitability of factual, legal and moral error in a
system that would wrongfully kill some defendants. Moreover, the system failed
to deliver the fair, consistent, and reliable sentences of death as required by the
Constitution (Vila and Morris 1997, 280-1).
Shortly after Justice Blackmuns dissent, the Court issued a ruling in
Simmons v. South Carolina.
27
Jonathon Simmons was sentenced to death for
brutally beating an elderly woman to death. Simmons had two previous violent
assault convictions which, had he not been sentenced to death for the murder, he
would not have been eligible for parole. During the penalty phase, the prosecutor

26
510 U.S. 1141, 114 S. Ct. 1127, 127 L. Ed. 2d 435 (1994).
27
512 U.S. 154, 114 S. Ct. 2187, 129 L. Ed. 2d 133 (1994).

58
argued that Simmons was too dangerous to ever be set free. However, the jury
was not informed that Simmons was ineligible for parole. In a 7-2 opinion,
Simmons conviction was reversed because the jury was improperly informed that
a sentence of life imprisonment carried with it no possibility of parole. Thus,
based upon the prosecutors argument, the jury may have opted for death because
they improperly believed that Simmons might be paroled at some time in the
future (Vila and Morris 1997, 281-2).
In 1996, the Supreme Court addressed the constitutionality of the newly
passed Antiterrorism and Effective Death Penalty Act of 1996. The AEDPA made
fundamental restrictions on federal habeas corpus relief. Habeas corpus has been a
powerful weapon against the unlawful detention in the United States. However,
many have argued that this weapon has been extensively abused by capital
defendants to ensure delay in the carrying out of death sentences. In Felkner v.
Turpin,
28
Ellis Wayne Felkner was convicted in 1981 of murder and rape of a 19-
year-old college student in Georgia. He was denied relief on his state appeal,
twice denied state collateral relief, and likewise denied relief with his first habeas
corpus petition. Five days after President Clinton signed the AEDPA into law,
Felkner filed another habeas corpus petition in an attempt to delay his execution.
In a unanimous decision, the Supreme Court denied the habeas relief as the
petition did not meet the necessary requirements for review of a subsequent habeas

28
518 U.S. 651, 116 S. Ct. 2333, 135 L. Ed. 2d 827 (1996).

59
corpus petition. Thus, the AEDPA withstood its first constitutional challenge
(Vila and Morris 1997, 295-9).
In Ring v. Arizona,
29
the Supreme Court addressed who had the decision
making power for the penalty of death. On November 28, 1994, a Wells Fargo
armored van pulled up to a department store in Glendale, Arizona. One of the
armed guards entered the store but found the van missing upon his return. Five
hours later, the van was discovered in the parking lot of a church. The driver of
the van was found inside with a single gunshot wound to his head. A police
informant led the police to the girlfriend of a suspect. The suspect implicated two
of his friends in the robbery/murder, one of which was Timothy Ring. A jury
convicted Ring of first-degree felony murder. Under Arizona law, Ring could not
be sentenced to death unless further findings of fact were made by the judge. In
his verdict, the judge determined that Ring was the person who killed the victim, a
fact not proved during trial but in a separate hearing after the verdict. The judge
also found Ring two additional aggravating circumstances that made Ring eligible
for the death penalty which was ultimately imposed (Gaffney 2003, 770-2).
The question in Ring was whether a judge could determine an aggravating
factor which exposed a defendant to the death penalty or whether the Sixth
Amendment jury trial guarantee required that such a determination be entrusted to
the jury. Justice Ginsburg, writing for the majority, noted that the Sixth
Amendment does not depend on the relative rationality, fairness, or efficiency of

29
536 U.S. 584, 122 S. Ct. 2428, 153 L. Ed. 2d 556 (2002).

60
potential factfinders. The Court noted that of the 38 states that utilize capital
punishment and 29 of them require juries, not judges, to make sentencing
decisions regarding the imposition of the death penalty. Thus, Justice Ginsburg
announced a new rule regarding the sentencing of a capital defendant that the Sixth
Amendment required a jury, rather than a judge, to determine all factors that are
necessary to impose the death penalty. The Court extended this holding not only
to capital punishment cases but also all criminal cases where a defendants
sentence might be enhanced because of aggravating circumstances (Gaffney 2003,
782-4).
As the Supreme Court entered the new millennium, the high court opted to
reconsider an issue previously decided. As noted above, in 1989, the Supreme
Court upheld the death penalty for mentally retarded offenders in Penry v.
Lynaugh. In 2002, the Supreme Court recognized a change in national consensus
that occurred on the issue since Penry, reexamining the death penalty for mentally
retarded persons in Atkins v. Virginia.
30
Daryl Atkins was convicted of abduction,
robbery, use of a firearm, and murder during the commission of an armed robbery.
At the penalty phase, a forensic psychologist testified that Atkins had an IQ of 59
and was mildly mentally retarded. The Supreme Court noted in its opinion that
much had changed in the United States since Penry. The Court stated that
following the Penry decision, at least 17 states enacted legislation that prohibited
the practice of executing mentally retarded offenders. In light of this change, the

30
536 U.S. 304, 122 S. Ct. 2242, 153 L. Ed. 2d 335 (2002).

61
Court indicated that such a prohibition provided powerful evidence that today
American society views mentally retarded offenders as categorically less culpable
than the average criminal. In addition, the justices found that of those states that
still imposed the death penalty, making no concessions for the mentally retarded,
only five executions had taken place since Penry in 1989 (Heisler 2004, 39-41).
Justice Stevens, writing for the majority, accepted the assumption that the
mentally retarded do indeed have significant developmental and psychological
limitations. Specifically, he noted limited adaptive skills such as communication,
self-care, and self-direction that became manifest before age 18. The Atkins
Court acknowledged that such mental limitations could result in less degree of
culpability for those offenders who possess them. The majority noted that such
individuals have diminished capacities to understand and process information, to
communicate, to abstract from mistakes and learn from experience, to engage in
logical reasoning, to control impulses and to understand reactions of others.
Consequently, while such mental problems do not constitute an exemption from
criminal sanctions for mentally retarded, they do diminish their personal
culpability. Moreover, the dual purposes for the death penalty, retribution and
deterrence, are rendered virtually a purposeless and needless imposition of pain
and suffering. Justice Stevens noted, A moral and civilized society diminishes
itself if its system of justice does not afford recognition and consideration of
mental limitations in a meaningful way (Heisler 2004, 41-2).

62
The final Supreme Court death penalty decision for this review was issued
on March 1, 2005, in which the high court opted to reconsider the constitutionality
of executing 16 and 17 year old juveniles. In Roper v. Simmons,
31
17 year old
Christopher Simmons planned and carried out capital murder. The most chilling
aspect of the case was the callousness of the crime itself. Not only did Simmons
tell a number of friends about how he was going the commit the crime but also
assured his friends that they could get away with it because they were minors.
On the night of the crime, Simmons and two others entered the home of the female
victim. They duct taped the victims eyes, mouth and hands and drove to a state
park. The three boys walked the victim to a railroad trestle where they hogtied her
hands and feet with electrical wire. She was then thrown from the bridge and
drowned in the water below. Simmons was arrested the next day and ultimately
confessed to the murder. Simmons was tried as an adult and convicted of
burglary, kidnapping, stealing and first degree murder. The jury recommended a
death sentence and the judge concurred. Following the Supreme Courts decision
in Atkins, Simmons filed for post-conviction relief in the Missouri Supreme Court
who struck down the death sentence (125 S. Ct. 1183).
The Supreme Court concurred with the ruling of the Missouri Supreme
Court and held that the Eighth and Fourteenth Amendments forbid the imposition
of the death penalty on offenders who were under the age of 18 when their crimes
were committed, thereby overruling the prior Supreme Court precedent laid out in

31
__ U.S. ___, 125 S. Ct. 1183, 161 L. Ed. 2d 1 (2004).

63
Stanford v. Kentucky. The Court based its decision on the objective indicia of
national consensus against execution of juveniles and the Courts own
determination in the exercise of its independent judgment, demonstrating that the
death penalty was a disproportionate punishment for juveniles. As for the national
consensus, the Court noted that the majority of states had rejected execution of
juveniles, the death penalty was infrequently used in those states that allowed for
it, and a trend was seen toward abolition in this regard. The Court further noted
that the death penalty must be reserved for the worst offenders with the most
serious crimes whose extreme culpability makes them the most deserving of
execution. Juveniles do not fall into this category as they are susceptible to
immaturity and irresponsible behavior, they are vulnerable and possess a
comparative lack of control over the immediate surrounds allowing for a level of
forgiveness, and it is less supportable to classify juveniles as irretrievably
deprived character. Finally, the high court noted that international opinion,
although not controlling, had uniformly rejected the execution of juveniles. In
light of the foregoing, the Supreme Court ruled that the Constitution barred the
imposition of the death penalty for juvenile offenders (125 S. Ct. 1183).
The Supreme Courts legal decisions on the death penalty demonstrate that
the Court, much like public opinion, has vacillated both for and against this form
of punishment. However, in order to gain a complete understanding of capital
punishment it is necessary to also examine the legislative history of the death
penalty.

64
Legislative History of Capital Punishment
The legislative history of capital punishment has experienced great
fluctuation over time both from federal and state perspectives. Throughout certain
time periods, state legislatures have taken the initiative on the death penalty while
at other times the federal legislature has been more prominent.
The concept of the death penalty came to the American colonies from
England itself. However, many of the colonists were not as ardent supporters of
this form of punishment as the English Crown. Following the revolution, many
colonists began to question the right of a government to take a human life. The
relative ambivalence toward capital punishment must have been in the minds of
the Founding Fathers, as no explicit mention of the death penalty can be found
within the four corners of the original document (Culter 2000, 1193). In 1790,
the newly formed Congress opted to pass the Act of the Punishment of Certain
Crimes Against the United States. This Act sanctioned the use of capital
punishment but only for a limited number of offenses, including treason, willful
murder on foreign property, forgery, piracy, counterfeiting and several crimes on
committed on the high sea. In the 1800s, the United States experienced a time of
social change and again many called for the complete abolition of the death
penalty (Culter 2000, 1193-4; Little 1999, 360).
One of the unique features of the death penalty debate going back as far as
the late 1800s was the great movement for and against the death penalty as a result
of historical events. Just as the abolition movement appeared to be making

65
headway in the early to mid-1800s, Civil War broke out in the United States
temporarily derailing the movement. In the 1890s, the Progressive Movement
swept over the United States and social reformers were concerned about
government corruption and focused on areas such as poverty, house, social
injustice, and crime (Nevans and Commager 1986, 286-97). Not surprisingly,
abolition rose once again. Evidence of the progressive influence was seen with an
important modification in state death penalty statutes away from the mandatory
imposition of the death penalty on certain crimes to permissive sentencing.
However, the onset of the World War I again placed abolitionist concerns on the
backburner for most Americans. This ebb and flow continued with the Great
Depression and then Americas entrance into World War II (Kirchmeier 2002, 9-
11).
Perhaps Americas greatest opportunity for abolition of the death penalty
came in the 1960s. Abolitionists renewed their movement during this era which
was lead by the NAACP. The number of executions in the United States during
this time came to a virtual standstill. Several states actually discontinued the use
of the death penalty all together. The abolition effort proved so successful that
between 1967 and 1972 not one execution took place in the United States at either
the federal or state level. Some have suggested that the unparalleled quiet in the
execution chamber spoke loudly of the discontent in society. Following the
Supreme Court ruling in Furman, a total moratorium on the death penalty was put

66
in place until the ruling in Gregg v. Georgia, four years later (Kirchmeier 2002,
15-7; Culter 2000, 1195-8).
It would be logical to conclude that following the 1960s era of social
consciousness, members of the state legislators would embrace the ruling in
Furman. To the contrary, Furman set off a firestorm of state legislative activity to
reinstate the death penalty. In the year following Furman , bills to restore capital
punishment were introduced in more than half of the state legislatures nationwide.
By the mid-1970s, thirty-four states had signed new capital statutes and more than
600 prisoners had been sentenced to death under the new laws (Vila and Morris
1997, 148; Schwed 1983, 145). State legislative bodies were not only active in
developing new capital statutes but also revising existing statutes by adding
factors, such as drug-related murder, drug trafficking, and multiple murder, to the
list of aggravating factors to be considered in the sentencing phase of first-degree
murder trials (Vila and Morris 1997, 172).
Following the Furman decision, a number of bills were introduced in the
United States Congress to revise the federal death penalty procedures. However,
the legislature ultimately deferred on the legislation until the Gregg group of cases
was decided by the Supreme Court in 1976. Even after Gregg was decided, the
federal legislature was very hesitant to embrace the death penalty. The first
reinstatement of the death penalty at the federal level came in the Antihijacking
Act of 1974. Although this legislation provided for the use of the death penalty,
the type of crime involved was obviously very narrow in nature. Further federal

67
death penalty legislation was not put in place for ten years. Some have surmised
the reason for the delay was the relative frequency in which the Supreme Court
was issuing death penalty rulings and adjusting the constitutional requirements.
Finally, in 1986, Congress amended a federal statute (18 U.S.C. 1512) to add a
possible death penalty for the killing of witnesses. Although this federal statute
was hardly evidence of Congresss firm commitment to the re-institution of the
federal death penalty, the amendment was seen as positive movement for pro-
death penalty advocates (Little 1999, 377-80).
In 1988, Congress was able to successfully pass the Continuing Criminal
Enterprise Death Penalty bill. The law, also known as the Drug Kingpin Act
(hereafter DKA), was part of Americas War on Drugs and allowed for the use
of the death penalty for certain crimes associated with the drug trade (Cutler 2000,
1200-1; Little 1999, 381-4). Specifically, the DKA allowed for the death penalty
for anyone who intentionally kills or counsels, commands, induces, procures, or
causes the intentional killing of an individual and such killing results while
importing a controlled substance, possessing a controlled substance on a vessel,
aircraft, or vehicle, or manufacturing or possessing with the intent to distribute.
The DKA also provided for the death penalty for those who kill during the
commission of, or in furtherance of, or while attempting to avoid apprehension,
prosecution, or service of a prison sentence involving a controlled substance.
Congresss concern over the proper use of the death penalty was evident by the
fact that the legislation not only incorporated the requirements set out in Gregg but

68
also additional protective measures above and beyond the Gregg mandates,
including reasonable advance notice that the government will seek the death
penalty, notification of the aggravating factors the government will seek to prove,
and the appointment of competent lawyers for federal criminal appeals and any
post-conviction habeas proceedings (Culter 2000, 1202; Little 1999, 382-3).
By 1994, the DKA had been upheld consistently against multiple
constitutional attacks (Little, 384). The Supreme Court had fine-tuned many
constitutional guidelines on the implementation of the death penalty. Thus,
Congress felt it was time to make a bold move regarding capital punishment and it
did so with the passage of the Federal Death Penalty Act of 1994. Congress
repeated much of the procedure from the DKA. However, the FDPA was far
broader in focus and application than the DKA. The FDPA federalized the death
penalty for nearly 60 crimes. The FDPA also allows for the imposition of the
death penalty for crimes that did not involve homicide such as espionage and
treason (Culter 2000, 1209-10, 1214; Little 1999, 385). One of the more
interesting aspects of the FDPA involves an omission. During the Congressional
debate on the Act, an amendment was offered which provided for an
antidiscrimination provision that would guard against racial bigotry in capital
sentencing. The amendment would have allowed an inference of discrimination to
be drawn by a showing of disparity in the number of members of a particular race
who were executed compared to the race of the defendant. A showing of racial
disparity would have shifted the burden to the government to show nonracial

69
factors to explain the disparity. Due to the controversial nature of the
amendment, it did not pass (Culter 2000, 1215-6).
Perhaps the pinnacle of pro-death penalty federal legislation came about in
1996 with the passage of the Anti-Terrorism and Effective Death Penalty Act. The
focus of the AEDPA imposed stringent filing deadlines for habeas petitions and
placed limits on the ability of defendants to file successive petitions. It also
required greater deference be paid to state court decisions by the federal court
(Smith 1997, 115). It was speculated that the strict filing deadlines would increase
the likelihood that offenders would default their claims, especially because they
often do not have lawyers who can help them prepare their petitions to make
effective presentation of their constitutional claims. In addition, the mandated
deference of federal courts to state courts rulings also imposed a difficult burden
on prisoners seeking to gain federal court review. Prisoners must show by clear
and convincing evidence that, except for a constitutional error at trial, they would
not have been found guilty. This is an extraordinary difficult standard to meet and
is one that can sacrifice defendants constitutional rights by failing to correct errors
made in state court. In the end, the AEDPA placed significant procedural and
substantive changes on habeas petitioners, making it more difficult for them to
obtain federal habeas corpus relief (Aarons 1998, 31).
Just when it appeared that both federal and state legislatures were strongly
supporting the concept of capital punishment, an upwelling for a nationwide
moratorium on the death penalty took center stage in the death penalty debate in

70
the mid-1990s. In 1997, the American Bar Association took a stand and called
upon each death penalty jurisdiction to impose a moratorium on executions until
those jurisdictions complied with ABA policies in this regard, including: (1)
ensure that death penalty cases were administered fairly and impartially in
accordance with due process and (2) minimize the risk that innocent persons may
be executed. The resolution did not take a position on whether the death penalty
should be abolished but focused on the proper implementation of the punishment
(Kirchmeier 2002, 36).
It now appears that the ABAs call for a moratorium was the tip of the
iceberg on the issue of the propriety in the application of the death penalty. In
1999, the state of Nebraska became the first state to vote for a moratorium since
Gregg was decided. The Nebraska legislature voted 27 to 21 for a bill that would
have imposed a two-year death penalty moratorium and required a study on the
fairness of the capital punishment system. Republic Senator Kermit Brashear was
the sponsor of the bill. Although the senator favored the death penalty, he had
concerns over the fairness of its application. The apparent victory for the
abolitionists was short-lived as Nebraska Governor Mike Johanns vetoed the
proposed bill. The single most significant event that established the Moratorium
Movement in the United States took place in 2000 when Illinois Republican
Governor George Ryan imposed a moratorium on executions in his state three
days before his term ended (Lanier and Acker 2004, 579). Governor Ryans
position was not based on moral opposition to the death penalty as much as his

71
concerns over the systematic problems with the Illinois capital punishment
system. A majority of the commission established by Governor Ryan to study the
capital punishment system in Illinois favored abolition but acknowledged that
support for the death penalty would like this option difficult. The commission
therefore offered a number of reforms (Lanier and Acker 2004, 579-80;
Kirchmeier 2002, 44-5).
Following Illinois moratorium, many other state legislatures actively
pursued moratoriums and/or investigations into their capital punishment schemes.
Alabama, Kentucky, New Jersey, and Ohio all introduced bills for moratoriums
and investigations. In 2000, Indiana Governor Frank OBannon requested the
Criminal Law Study Commission investigate the application of Indianas capital
punishment laws. Likewise, commissions were set up in Arizona, Connecticut,
Nebraska, Nevada, North Carolina, Pennsylvania and Tennessee to study the
fairness of capital punishment in those states. In May, 2002, the Governor of
Maryland ordered a two-year moratorium on executions pending a study that
focused on the racial bias in the implementation of the death penalty (Lanier and
Acker 2004, 579-83).
Calls for a moratorium and investigation into the fairness of the application
of the death penalty at the federal level have not been as eagerly pursued as seen
with the states. In 1990 and 2001, Democratic Senator Russ Feingold introduced
legislation to abolish the death penalty altogether at the federal level (Ogloff 2004,
388; Feingold, 2000, 18-23). Following Governor Ryans commutation of all

72
Illinois death row inmates, Senator Feingold introduced the National Penalty
Moratorium Act of 2000, in which he called for a moratorium at both the state and
federal levels pending a national review of its administration. Similar bills were
introduced in the House. None of these bills has ever gotten out of committee
(Ogloff 2004, 388; Lanier and Acker 2004, 583-4). In 2001, Senator Feingold
offered Congressional testimony regarding the drop in public support for the death
penalty when alternative sentences are provided, social science evidence refuting
the deterrent effect of the death penalty and the racial bias in the death penalty
sentencing as well as the cases of innocent individuals being sentenced to death
(Ogloff 2004, 388). Although some may see Feingolds attempts to bring about a
national moratorium as a personal crusade, it is clear that there is at least some
support for a moratorium movement at the federal level. However, at the present,
its time has not come.
What Does This History Reveal?

Several conclusions can be derived from the above discussion. First, on its
face, the public opinion numbers indicated that the public appears to heavily
support the use of the death penalty. At no time since public opinion has been
recorded has support ever fallen below 42%. However, a closer look at the type of
questions being asked of respondents indicates that those numbers may be seen as
misleading. Many Americans, in the neighborhood of 40% to 50% (possibly even
lower) might not opt to support capital punishment given satisfactory alternative
sentencing and/or mitigating circumstantial evidence. Thus, it is reasonable to

73
conclude that public opinion may very well support the death penalty but not at
the levels suggested by the generic polls offered beginning in the 1930s.
Second, it is likewise clear that the Supreme Court has been and will
continue to fine-tune both procedural and substantive issues on the death penalty.
Interestingly, the Supreme Courts attempts to refine the death penalty have not
been one-sided. Many recent decisions such as Atkins v. Virginia and Roper v.
Simmons appear to show a court that is leaning toward abolition. However, only a
few years yearly, the Supreme Court upheld the AEDPA which was clearly seen as
pro-death penalty legislation. The only proper conclusion that can be drawn from
the Courts seesaw decision making on capital punishment is that the Supreme
Court recognizes that several important issues remain with regard to the
appropriate application of the death penalty. The high court has elected not to run
and hide from these difficult issues. To the contrary, it appears the Court is
embracing the opportunity to resolve these questions. One would anticipate that
the Supreme Court will continue to address various capital punishment issues but
its future support for or against the death penalty remains unclear.
Third, it is evident that state legislatures, and the federal legislatures to a
lesser degree, are uncomfortable with the current state of the death penalty
systems. The Moratorium Movement appears to be gaining momentum
nationwide at the present time. This is not to say that states or the federal
government will begin to dismantle the capital punishment system, only that it is
apparent many in state and federal legislatures are ill-at-ease with the current state

74
of the death penalty. For this reason, one can anticipate further state
commissions being appointed to investigate the use of capital punishment. Some
states may go as far as implementing moratoriums on the death penalty until
legislators are satisfied with the fairness of its application. Only time will tell
whether this phenomena will lead to further death penalty reform or its complete
abolition.
At first glance, it would be easy to view public opinion, Supreme Court
decision making, and the legislative movement on the death penalty as
independent of one another. However, it does appear that at least some connection
is found between the three. Several examples can be identified. Beginning in the
1960s, public support for capital punishment began to wane reaching its lowest
level of support in 1966 at only 42%. Although support did rise following that
year, it remained just above 50% when the Supreme Court issued the ruling in
Furman. Some might suggest that any connection between the publics lack of
support for capital punishment and Supreme Court decision making is tenuous at
best. However, following the Furman decision, public support for the death
penalty immediately climbed well above 60%. With this high level of support for
capital punishment, many state legislatures promptly introduced revised death
penalty statutes in order to reinstitute capital punishment and the Supreme Court
reinstituted the death penalty during a time when public support for capital
punishment was over 60%.

75
During the 1980s, public opinion in support of the death penalty
continued to rise well above 70%. During that time, the federal legislature
reintroduced and passed legislation for the use of the federal death penalty. As the
1990s approached, support continued to climb reaching its zenith of 80%. Only
two years later, the federal legislature passed the FDPA and two years after that
the AEDPA became law. The Supreme Court found that both pieces of legislation
passed constitutional muster. Finally, beginning in 2000, public opinion remained
high but had fallen off from the levels seen in 1994. Accusations that innocent
persons were sitting on death rows throughout the country appeared to instigate the
call for the Moratorium Movement (Ogloff 2004, 386). State legislatures began to
investigate the fairness of the application of the death penalty through state
sponsored committees. During this same time frame, the Supreme Court made
two categorical rulings in Atkins and Roper that it is unconstitutional to execute
the mentally retarded and anyone under the age of 18.
This is not to say that a direct link exists between the public, the judiciary
and legislatures on the issue of capital punishment. However, that being said,
common sense demonstrates at least some connection is present. This conclusion
can be drawn if only based upon the following factors. First, members of both the
state and federal legislature rely upon their constituency to gain election. If those
in political power are out of step too long from the electorate, it is reasonable to
assume that those persons will not remain in power very long. This does not
suggest that politicians have to follow public opinion polls in every decision but it

76
is clear that a politicians career does rely on the public for their position and if a
politician to too far out of step with the public on a controversial issue such as the
death penalty for too long, that politician might find himself out of public service.
The same argument cannot be made of the judiciary as the federal judiciary
is appointed by a president, confirmed by the Senate and serve for life (Const., Art.
II, Sec. 2). If members of the Supreme Court, or any federal judge for that matter,
do not want to listen to the public opinion on a particular issue, they have little
incentive to do so. However, the one key factor that brings public opinion into
federal judicial decision making on the death penalty is the nature of the test that
the courts apply when analyzing the constitutionality of capital punishment. In
Trop v. Dulles, the Supreme Court noted the test used by courts in analyzing
violations of the Eighth Amendment is that a court must draw its meaning [of
cruel and unusual punishment] from the evolving standards of decency that mark
the progress of a maturing society (Vila and Morris 1997, 102-3). By the
Supreme Courts own definition, any court that addresses issues of Eighth
Amendment violations must contemplate public opinion in order to evaluate what
society views as cruel and unusual treatment. Once again, this does not require
courts to necessary follow the exact dictates of the public but some level of
consideration is clearly at-hand.
The exact future of the death penalty is unclear. However, it can be
assured that any major changes to this very challenging issue will involve the

77
citizens of the United States, the justices of the Supreme Court and members of
both the federal and state legislatures.

78
Chapter 3

The Coding Continuum

The coding of legal cases by political scientists within the judicial decision
making realm has traditionally been conducted in a simple, if not rudimentary,
method. Segal and Spaeth (1993), 218-19, the principal authority in analyzing the
relationship between judicial attitudes and the votes on the Supreme Court, began
by classifying decisions of the United States Supreme Court on search and seizure
cases into two categories, liberal versus conservative. The use of a dichotomous
dependent variable has been repeatedly used by other scholars conducting similar
analysis both in the Supreme Courts and lower courts as well (see, e.g., Klein
2002; Sheehan and Haire 2000). This process has been carried out repeatedly by
scholars who analyze capital punishment decisions. Typically, death penalty cases
are coded based upon the outcome of the case (state versus defendant) (see, e.g.,
George and Epstein 1992; Tauber 1998; Brace and Hall 1997; Traut and Emmert
1998; Brace and Hall 1995; Brace and Hall 1993).
Coding death penalty cases based on the outcome of court decisions is
without question a straight-forward process. However, the problem with utilizing
such a process is that dichotomous coding treats each decision equal to the other
and does not allow for the possibility that certain judicial decisions possess a
greater bias for or against the death penalty than other decisions. The bottom line
is that all death penalty decisions for or against capital punishment should not be
treated equally in terms of the true implications of the court decisions. By

79
simplifying the process to a dichotomous measure, scholars miss out on a bevy of
information regarding judicial decisions in this area of the law. Death penalty
litigation, like nearly every area of the law, is extremely complex and generally
involves multiple legal issues and layers of analysis. Due to the complex nature of
the cases, a more sophisticated coding process of the cases is desirable and will
assist in gaining a better understanding of death penalty decision-making. Rather
than relying solely on a dichotomous measure, this study will use both a
dichotomous measure and the coding continuum set forth below. The purpose for
coding the dependent variable through the traditional method and the new
continuum is to allow for a comparison of the two methods at the conclusion of the
analysis. The coding continuum that will be used in this study will be broken out
into six categories: (1) strong anti-death penalty; (2) moderate anti-death penalty;
(3) minimal anti-death penalty; (4) minimal pro-death penalty; (5) moderate pro-
death penalty; (6) strong pro-death penalty.
32
Each category in the coding process
will first be defined and then the applicable case decisions that fall within that
category will be noted.
The strong anti-death penalty category contains court decisions that
demonstrate a direct bias by a judge against the use of capital punishment or
components within death penalty statutes. This category possesses four types of

32
The following is a visual representation of the coding continuum.

__(1)___________(2)_____________(3)____________(4)____________(5)____________(6)__
Strong Moderate Minimal Minimal Moderate Strong
Anti-DP Anti-DP Anti-DP Pro-DP Pro-DP Pro-DP

80
court decisions. Court decisions that result in a reversal or remand based upon a
direct challenge to the constitutionality or application of death penalty statutes are
included in the strong anti-death penalty category. Likewise, decisions that
challenge the aggravating or mitigating circumstances applied in a particular case
fall within this category. Whether a court reverses based upon a constitutional
deficiency in the death penalty statute or the application of the aggravating or
mitigating circumstances, the net result is that the death penalty will not be
imposed which creates a strong implication against capital punishment. This
category also includes circuit court decisions that uphold the ruling of the United
States District Court which reversed a death penalty conviction or sentence.
Additionally, decisions finding sentencing error during the penalty phase of a
capital trial are also included in the strong anti-death penalty category. In both the
reversal of a death penalty conviction or sentence, the death penalty will not be
imposed unless the state opts to retry the defendant, a very lengthy and costly
process. In each of these four types of court decisions, a strong anti-death penalty
bias is found based on the direct challenge to either the capital punishment
legislation or the penalty itself.
The moderate anti-death penalty category is defined as those court
decisions that, like the strong anti-death penalty category, demonstrate a bias
against the application of the death penalty. However, the bias may also result
from some non-ideological factors. For this reason, court decisions that fall
within this category are classified as moderate anti-death penalty. Four types of

81
court decisions are also included within this category. Cases resulting in a
reversal or remand by a circuit court based upon an error made during jury
selection, also known as Witherspoon error, fall within this category. Witherspoon
error takes place when a perspective juror is improperly excused for cause from a
capital jury because the juror stated an opposition to the death penalty. The only
jurors who cannot legally sit on a capital punishment case are those who indicate
that under no circumstances could they impose the death penalty (Witherspoon v.
Illinois, 391 U.S. 510, 522-24 (1968)). Circuit court rulings that Witherspoon
error took place in a particular case reflect bias against the implementation of
capital punishment. Such reversals require retrial with a new jury. However, non-
ideological factors can also be said to affect such rulings. A circuit court might
simply have felt that a prosecutor erred by not following the requirements laid out
in Witherspoon for jury selection in a capital case. Thus, such rulings are viewed
as only moderately anti-death penalty.
The second type of court decisions within the moderate anti-death penalty
category are all cases in which an en banc hearing is granted for the defendant in a
capital punishment case. En banc hearings take place when the full panel of the
circuit court of appeals grants the defendant a rehearing of a three member panel
decision from that circuit (Taha 2005, 1249). The granting of such a hearing for a
capital defendant clearly reflects a bias in favor the defendant and against the state
as the defendant is given another chance to prevail on appeal. However, at the
same time, the granting of an en banc hearing may simply be based upon a legal

82
error by the three judge panel or the need to reconsider the decision in light of
new legal precedent that was issued after the decision by the three judge panel.
Either way, some non-ideological factors may come into play within the decision
to grant an en banc hearing and for this reason these decisions are placed within
the moderate anti-death penalty category.
The third type of court ruling within this category is the court of appeals
decision to grant a defendant the right to file a successive petition. In 1996,
Congress passed the Anti-Terrorism and Effective Death Penalty Act which,
among other things, set out a provision that barred capital defendant from filing
multiple petitions for writs of habeas corpus. Capital defendants are only
permitted to file successive petitions after it is determined that (1) that applicant
shows that the claim relies on a new rule of constitutional law, made retroactive to
cases on collateral review by the Supreme Court, that was previously unavailable;
or the factual predicate for the claim could not have been discovered previously
through the exercise of due diligence; and the facts underlying the same, if proven
and viewed in light of the evidence as whole, would be sufficient to establish by
clear and convincing evidence that, but for the constitutional error, no reasonable
fact-finder would have found the applicant guilty of the underlying offense (28
U.S.C. 2244(b)(1)-(2)(B)). The standard set out by Congress presents a significant
barrier for capital defendant to meet in order to warrant the granting of a
successive habeas corpus petition. Due to the difficulty in obtaining a successive
petition, the granting of such a petition must be seen as bias in favor of a

83
defendant. At the same time, as with other types of case decisions in this
category, some non-ideological reason may justify the courts decision to allow the
capital defendant to file a successive petition. Thus, the granting of a successive
petition is viewed as possessing a moderate bias against the death penalty.
The fourth and final type of circuit court decision that falls within the
moderate anti-death penalty category is cases where defense counsel is ruled to
have acted ineffectively. Ineffective assistance of counsel takes place when the
competence of a defendants trial attorney falls below what is expected of a
reasonably competent attorney and that such actions prejudice the outcome of the
trial (Strickland v. Washington, 466 U.S. 668, 687 (1984)). Unlike the other type
of court decisions that are found within this category which specifically implicate
the death penalty (Witherspoon) or grant a capital defendant specific relief (grant
of en banc hearing or a successive petition), ineffective assistance of counsel
claims are placed within this category due to the extreme discretion given to court
on this particular issue. Criminal defendants are guaranteed the right to effective
assistance of counsel under the Sixth Amendment. The Supreme Court's decision
in Strickland gave appellate courts overly broad discretion to determine exactly
what constitutes ineffective assistance of counsel. As a result, there is little
consistency within judicial districts or across districts (Murphy 2000, 180). It is
this very broad discretion given to the courts, and circuit courts in particular, that
allows those courts to issue rulings based upon a particular bias rather than
following the letter of the law. Thus, circuit court decisions that find a defense

84
attorney ineffective in a capital punishment case will be viewed as moderately
anti-death penalty due to the significant discretion given to the courts on this
particular issue of the law.
The minimal anti-death penalty category contains court decisions that
appear neutral on their face in terms of bias for or against the death penalty other
than the fact that the ruling is in favor of the capital defendant and against the
state. The legal issues found within this category include cases reversed or
remanded due to: prosecutorial misconduct, the erroneous admission of evidence
at the guilt or penalty phases of the trial, sufficiency of the evidence at the guilt
phase, jury misconduct, the need for an evidentiary hearing to resolve a factual
question, instructional error by the trial court, new evidence discovered after trail,
the states failure to exhaust remedies, Brady error,
33
Double Jeopardy violations,
the United States District Court failure to review the record on an issue presented,
the United States District Court improperly applied abuse of the writ doctrine.
This category also includes circuit court decisions to grant a certificate of
probability/certificate of appealability or a stay pending resolution of an issue.
Likewise, court rulings denying the state rehearing or en banc hearings fall within
this category. The grant of a CPC/COA, stay and the denial of rehearing or en
banc by the state should generally be seen as minimal victories by a capital
defendant.

33
See Brady v. Maryland, 373 U.S. 83, 86 (1963).

85
Following the continuum from minimally anti-death penalty to the pro-
death penalty side of the equation, the minimal pro-death penalty category
contains two types of court decisions. As with minimal anti-death penalty
category, minimal pro-death penalty court decisions are those that are seen as
neutral in terms of bias but minimally pro-death penalty as the decisions
themselves favor the state and go against the capital defendant. The first of these
court decisions are those in which the defendants claims are rejected by the circuit
court due to the defendants failure to exhaust all state court remedies. The
doctrine of exhaustion of remedies states that a federal court may not generally
hear challenges that a habeas petitioner failed to present to the state courts during a
direct appeal or state habeas proceedings. The federal court reviewing the habeas
petition will dismiss any such unexhausted claims (Rose v. Lundy, 455 U.S. 509,
518-19 (1982)). Although a habeas petitioner can return to state court to file a
state habeas petition seeking to exhaust the unexhausted claims, that petition is
likely to be rejected for noncompliance with the state's procedural rules that
prohibit piecemeal challenges to convictions (Coleman v. Thompson, 501 U.S.
722, 735, n.1 (1991). The state court's refusal to entertain the claim would satisfy
the federal habeas exhaustion requirement, as "there are no state remedies any
longer "available' to [the petitioner] (Engle v. Issac, 456 U.S. 107, 125 n. 28)
(Hoffstadt 2005, 1161-2).
Rulings in favor of the state based upon unexhausted claims, does at first
glance, appear to be a clear victory for the state. The claims presented to the

86
federal court are dismissed and the merits of the petition are not addressed.
However, it is important to note that when claims are rejected by the federal court
based upon the failure to exhaust state court remedies, those claims are not
rejected with prejudice. Capital defendants are permitted to present those claims
to the state court and, if rejected, then those claims can be asserted in the federal
courts as exhausted claims. In essence, these procedural complications, while
ensuring state deference, effectively delay review of the substantive issues:
determining the constitutionality of each state's death penalty procedures are
federal questions that cannot be raised until exhaustion of state remedies and
appeal to the federal courts (see Zimring 2003, 7). Thus, when a federal court
rejects a capital defendants claim based on the failure to exhaust state court
remedies, what appears to be a strong pro-death penalty ruling is in fact a partial
victory for the capital defendant by dragging out the litigation and prolonging the
defendants life. The mixed result from these types of court rulings requires that
such decisions be classified as only minimally pro-death penalty.
The second type of court decision that falls within the minimal pro-death
penalty category is when capital defendants are denied motions for either
rehearing by the panel that heard the case or by the full court en banc.
Conventional wisdom dictates that motions for rehearing in circuit courts are
almost never granted and en banc rehearings occur about as often as the planets
align (Riback 2003, 134). Due the frequent nature with which rehearing motions
are denied both by the original panel that ruled on the case or en banc, little is

87
construed by such denials. For this reason, these types of denials are given only
minimal pro-death value.
Two circuit court case decisions are included in the moderate pro-death
penalty category. The moderate pro-death penalty category is defined as those
cases in which a greater likelihood of bias in favor of the death penalty can be
drawn by the nature of the decision itself. However, as with the moderate anti-
death penalty category, some non-ideological factors may come into play within
this category requiring a moderate pro-death penalty label. The first type of court
decision is when a motion made by the state for an en banc hearing is granted. As
noted above, a circuit courts decision to grant a motion to hear a case en banc is an
infrequent proposition and the granting of such a motion is clearly a positive sign
for the party that had the motion granted (see Riback 2003, 134). Systematic
examinations of en banc review have found that the decision to grant en banc
review can be explained by looking beyond the federal rules of appellate
procedure. Scholars have gone as far to hypothesize that political rather than legal
reasons accounted for en banc practice (George and Solimine 2003, 179). Due to
the limited practice of granting en banc motions combined with the possibility that
such an occurrence my involve consideration of political bias rather than legal,
such decisions must be categorized as higher than minimally pro-death penalty. At
the same time, the granting of a motion for en banc could be based upon nothing
more than the three-judge panel misapplying the law in its decision or that a new
decision had been issued since the time of the three-judge opinion. In light of

88
these possibilities, a circuit courts grant of a motion for en banc review is placed
within the moderate pro-death penalty category.
The second type of court decision that falls within the moderate pro-death
penalty class of decisions is when a capital defendants claims are rejected based
upon the doctrine of abuse of the writ. Federal courts have the discretion to
dismiss a habeas petition or certain claims within a habeas petition under the
doctrine of abuse of the writ when the petition contains repetitive claims or when
the petitioner brings several habeas petitions. This doctrine is administered by
adhering to the principles of equity (Nkama 2001, 195). When claims are denied
based upon abuse of the writ, the merits of the defendants or states claims are not
addressed. Rather, the court opts to disregard the claims because they have been
presented to the court before and rejected or brought before the court multiple
times in multiple habeas corpus petitions. The severity of such a doctrine is clear
as the issue is disregarded out-of-hand. However, just because a claim has been
rejected based upon abuse of the writ does not mean that the merits of the claims
were not previously addressed in another petition. Although a courts use of abuse
of the writ is a victory of the state, the defendant was still able to have a court
resolve the merits of the claim previously. For this reason, decisions based on
abuse of the writ are not a clear cut victory for the state and must therefore be
classified as a moderate pro-death penalty decision.
Finally, three types of circuit court decisions are found within the strong
pro-death penalty category. Each type of court decision in this grouping solidly

89
favors the state in its quest to uphold the death penalty conviction and sentence of
capital defendants with the defendant being the clear loser in the legal battle and
one step closer to the defendants demise. The first of these court rulings is when
a defendants claim or claims are procedurally defaulted from consideration. If a
petitioner has failed to follow prescribed state procedural rules and denied the
claim on such grounds, federal courts will generally refuse to consider such claims
on habeas review. A petitioner can overcome the procedural default only by
demonstrating either (1) cause for his procedural default and actual prejudice as a
result of the alleged violation of federal law; or (2) that failure to review the claims
will result in a fundamental miscarriage of justice (Hoffstadt 2005, 1162).
A petitioner can satisfy the cause requirement by showing, for example,
that some objective factor external to the defense impeded counsel's efforts to
comply with the State's procedural rule, that the factual or legal basis for a claim
was not reasonably available to counsel, or that governmental interference
rendered procedural compliance impracticable. To establish prejudice, a petitioner
must show not merely that the errors at his trial created a possibility of prejudice,
but that they worked to his actual and substantial disadvantage, infecting his entire
trial with error of constitutional dimensions (Hoffstadt 2005, 1162-3). Circuit
court rulings in which a capital defendant has been determined to have
procedurally defaulted on a claim or claims which precludes review is a severe
penalty and must be treated as such. When a court finds that a capital defendant
has procedurally defaulted on a claim, the merits of that claim(s) are never

90
addressed. For this reason, cases that base their rulings, even in part, on
procedural default are classified as strongly pro-death penalty.
The second type of court decisions that falls within the strong pro-death
penalty category are those in which the circuit court reverses a grant of a habeas
corpus petition by the United States District Court in favor of the state. Whether
the case is presented on appeal or through a petition for writ of habeas corpus,
reversals for the state reflect a position by the circuit court that the lower court
erred in granting relief to the capital defendant but, more importantly, that the
conviction and death sentence will stand unless the full court of appeal opts to
review and overturn the three panel judge or the United States Supreme Court
accept review of the decision and reverses it. As noted above, en banc review is
very rare (see Riback 2003, 134) and review by the Supreme Court is even less
likely accepting only 1% of the petitions for writ of certiorari presented to it each
year (OBrien 2000, 165). Barring en banc or Supreme Court review, the ruling of
the three-judge panel will stand and ultimately the death penalty sentence will be
carried notwithstanding clemency on the part of a governor or president. In light
of the potential finality of a circuit court reversal in favor of the state, such rulings
must fall within the strong death penalty category.
The final group of circuit court decisions that come within the strong pro-
death penalty classification are all circuit court rulings that uphold the district
court rulings in favor of the state. Decisions by the circuit court possess a level of
finality that is not seen with other court. Circuit courts have the ability and

91
obligation to review the decisions of the district court. Moreover, circuit courts
generally also have the ability to review state court decisions as long as a federal
constitutional right is implicated which is often the case (OBrien 2000, 169).
Thus, absent en banc review by the full court of appeal or the United States
Supreme Court, the circuit court decisions are truly the court of last resort in the
United States. For this reason, when the circuit court opts to uphold the decision
of the United States District Court in a capital case, the circuit court for all intents
and purposes issues the final opinion in that capital case. A capital defendants
only alternative once the courts have issued their final ruling is to hope for
clemency by the state or federal executive. Clemency in the United States is
indeed a rarity (Carter 2005, 268: Since the reinstatement of capital punishment
in the United States in 1976, clemency has been granted in 228 cases, 167 of
which were the blanket grant in Illinois. During the same period of time, 960
executions have taken place, and there are presently over 3,400 persons on death
row in the United States (footnotes omitted)). Unless a capital defendant is able
to obtain clemency, the ruling of circuit court will result in the death sentence
being carried out. Such rulings must therefore be classified as strongly pro-death
penalty.
The coding continuum is a unique approach to classifying death penalty
cases that goes beyond the mere dichotomous outcome-based method. It will be
through this coding process that one will gain a better understanding of circuit
court bias both for and against capital punishment.

92
Chapter 4

Regression Analysis

The federal circuit courts of appeals play an intriguing role in the American
judiciary. While the circuit courts are not the highest court in the land, the
importance of these courts has increased dramatically in recent years as the United
States Supreme Court is only able to address but a small fraction of the cases that
are presented to it each year. For this reason, the circuit courts really have become
the court of last resort for the vast majority of litigates that come before it (Tobias
2005, 744). It is for this very reason that political scientists have begun to focus
on these courts and question the driving force(s) behind judicial decision making
of this body. While some suggest the law itself is the primary motivator for
judicial decisions, others argue that judges base their decisions on their own
personal policy preferences. Institutionalists believe that the institutional structure
of the courts itself drives the decision making whereas some see the hierarchical
structure of the courts as impacting intermediate appellate decisions (see Epstein
1995, 230-235). A clear consensus among political scientists has yet to be
achieved.
The divergence between these theories becomes even more evident when
examining circuit court decision making on volatile political issues such as capital
punishment. American society has long struggled over the propriety of the death
penalty. Many in society demand justice for those who are murdered in the form
of retribution whereby capital defendants must give up their life for the taking of

93
another (see Bowman 2005, 1350, n.21). Death penalty opponents contend that
the state undermines its moral authority and ultimately denies the value of each life
when it executes killers in an effort to decree that murder is wrong (Preston 2004,
241). This controversy has not only found its way into society but has also
permeated into the American judiciary. Judicial nominees for the Supreme Court,
and lower court judges for that matter, are routinely questioned during
confirmation hearings about their personal beliefs on the death penalty (see
Williams 2003, 332).
The controversial nature of capital punishment both outside and within the
American judiciary is the very reason that death penalty cases were chosen for this
study. In order to examine whether legal and extra-legal influences are affecting
circuit court decision making, it was necessary to select a legal issue that possesses
a great likelihood of ideological influence on the judges deciding those cases to the
exclusion of all other forms of influence such as legal, institutional, hierarchical,
etc. Thus, if non-ideological variables are found to be significant influences in
capital punishment decision making in conjunction with or to the exclusion of
judicial ideology, then such results can be generalized to judicial decision making
on less politically charged issues (see generally Eckstein 1975). In other words,
the selection of death penalty cases in this study is akin to a most-likely
observation wherein ideological variables will be found to play a role in circuit
court decision making on this issue to the exclusion of all others (see King,
Keohane, and Verba 1994, 209). Thus, based upon the importance of capital

94
punishment within American society as well as the judiciary, the evaluation of
judicial decision making on this particular issue will create a gateway to better
understanding judicial decision making overall.
Data, Methods and Variables
The data source for this study consists of 918 state death penalty
decisions
34
that reached the federal circuit courts of appeals via petitions for writ
of habeas corpus. Federal death penalty cases were excluded from the sample as
federal appellate standards of review differ from the standards of review applied in
habeas corpus cases. The cases were drawn from four circuits, including the 4
th
,
5
th
, 6
th
and 9
th
circuits over a 25 year period between 1980 and 2004.
35
These
specific circuits were selected in order to gain a regional, diverse representation of
death penalty cases decided in the northern, southern, eastern and western regions
of the United States. Case decisions included published and unpublished written
opinions as well as miscellaneous motions filed in state death cases.
Three types of regressions were run through the use of STATA, including
logistic regression, ordered logistic regression and multinomial logistic regression.
Two dependent variables were tested in the study. The logic regression utilized
the traditional dependent variable used by a multitude of scholars when evaluating
judicial decision making (see, e.g., Traut and Emmert 1998, 1170; Tauber 1998,
196; Brace and Hall 1995, 15; Segal and Spaeth 1993, 218). The outcomes of

34
Initially, en banc decisions were included in the case sample along with panel decisions.
However, due to the variation in the number of judges that participated in en banc panels across
circuits, the en banc rulings were excluded from the analysis.
35
The cases were obtained through the use of the Lexis-Nexis search engine. Searches were
conducted by year using the terms death penalty and the name of each of the four circuits.

95
capital cases by circuit panels were coded based upon whether the panels voted to
uphold the death sentences in the favor of the state or overturn the death sentence.
A dummy variable was created for this purpose in which cases that upheld death
sentence were coded 0 while decisions that overturned death sentences were
coded 1. The ordered and multinomial regressions used an alternative
dependent variable known as the coding continuum. Rather than merely
classifying the outcome of death cases based upon a dichotomous variable, the
continuum categorized case outcomes on a more sophisticated scale ranging from
1 to 6. Chapter 3 - The Coding Continuum - sets out the exact nature of each
category and the coding.
Eleven independent variables, noted below, were created to evaluate the
various theories of judicial decision making. In addition, both the logistic
regression and the ordered regression included dummy variables for three of the
four circuits to capture differences across circuits. Likewise, time dummy
variables were included to account for the possibility that time itself or a
maturation effect on the court may impact decision making. For this reason, 24 of
the 25 years studied to assist in reducing any omitted variable bias.
Panel Ideology Variable
The first independent variable that was analyzed was the impact of judicial
personal policy preferences on judicial decision making. Brace and Hall, 1995,
11, note that an important approach to the study of judicial behavior, developed
largely from studies dealing with the United States Supreme Court and born of the

96
behavioralist movement, is the attitudinal perspective. The attitudinal model
holds that judges decide disputes in light of the facts, given their ideological
attitudes and values (see generally Segal, Songer and Cameron 1995; Segal and
Spaeth; Schubert 1965, 1974). Most basically, attitudinal theory conceptionalizes
judges as acting to achieve case outcomes most proximate to their individual
policy preferences (Brace and Hall 1995, 11). Therefore, while most lawyers,
judges, and legal scholars presume that judges make decisions by examining
relevant legal doctrine, most social scientists adhere to the precept that judges
make decisions based on their ideological preferences. This perspective is best
captured by the attitudinal model. The attitudinal theory assumes that judges are
rational actors who, when making decisions, seek to maximize their sincere policy
preferences or attitudes. Nearly all judges, in contrast to elected policymakers,
have not publicly disclosed their policy preferences and thus a proxy for attitudes
is necessary. Attitudinal studies have demonstrated that the ideological direction
("liberal" or "conservative") of the party of a judge's appointing President is a
strong predictor of how votes will be cast by justices of the Supreme Court as well
as the Court of Appeals judges (George 1999, 237-38).
The coding for this variable required that each circuit panel be assigned a
number ranging from 0 to 3. Panels that were composed of judges who were all
appointed by Democratic presidents were coded 0. Panels with two Democratic
appointed judges and one Republican appointed judge were coded 1. Panels
with two Republican appointees and one Democratic appointee were coded 2

97
and panels where all three judges were appointed by Republican presidents were
coded 3. Although this variable was designed to analyze the impact of judicial
ideology on decision making, the variable will inevitably also pick up on panel
(institutional) effects on decision making. Institutional theorists suggest that
judicial decision making is impacted by the very nature of the judicial institution
itself. A complete understanding of what affects judges decision making must take
into account the possibility that some judicial preferences are shaped and
constituted within their institutional context (Whittington 2000, 623). Judges
understand that they face a number of constraints including the institutional
context in which they act (Burton 2003, 577). In other words, the panel
composition itself may affect the outcome of a decision based upon the political
make-up of the panel members (see Whittington 2000, 612).
Thus, the decision making by a panel composed of three judges appointed
by a Democratic president may be affected differently than a panel of two judges
appointed by a Democratic president and one from a Republican president or two
judges appointed by a Republican president and one from a Democratic president
or all three judges appointed by a Republican president. The construction of this
variable may therefore mask the institutional effects which cannot be parsed out
given the manner in which it is measured. That being said, the actual impact of
institutional effects on circuit judge decision making will be addressed through a
separate analysis in Chapter 5 Institutional Analysis.

98
Case Characteristics Variables
The next set of independent variables were case characteristics of each
case. The legal model postulates that decisions of the courts are based upon the
facts of the case in light of the plain meaning of statutes and the United States
Constitution, the intent of the framers, and precedent (Segal, Songer, and Cameron
1995, 230). This traditional or jurisprudential approach emphases the rule of law
in dictating judicial outcomes. Traditional legal theory views judicial decision
making as the application of existing sources of law to cases appearing before
courts, with the discretion of individual judicial actors rigorously constrained by
legal president. However, some judicial behavior scholars have recognized the
importance of case characteristics and have incorporated the variables into models
of voting behavior (Brace and Hall 1995, 10). Thus, it has been suggested that the
inclusion of certain aggravating case characteristics into capital cases will result in
less successful appeals. In other words, the inclusion of some case characteristics
will impact judicial decision making (see Traut and Emmert 1998, 1172).
Certain case characteristics have been found to be theoretically significant
in terms of impacting judicial decision making. Each case within the dataset was
coded for the inclusion of seven aggravating case characteristics. The
characteristic was coded 1 if a case involved robbery, rape, kidnapping, murder
of an elderly victim, murder of a minor, murder of a police officer and/or the case

99
involved multiple murder.
36
Conversely, characteristics were coded 0 if the
case did not involve that particular characteristic. Most of the appellate decisions
included sufficient facts to determine whether the above characteristics were part
of the case. In those cases where the facts were not adequately set forth, the
published state appellate opinion was utilized to determine the characteristics of
the case.
Signal Theory Variables
Two independent variables were created to examine the extent that circuit
courts react to signals from other bodies of the government, including Congress
and the political party of the governor in state where the death penalty trial was
held. The question thus becomes whether the circuits courts are truly independent
when it comes to decision making on capital punishment cases or do circuits listen
to extra-judicial influences. The first of these signals coded the application of the
Anti-Terrorist and Effective Death Penalty Act (1996) to the death penalty cases.
AEDPA, as it is known, was a significant piece of pro-death legislation signed into
law in 1996 that should have had a profound impact on the circuit courts ability to
overturn capital punishment cases. The AEDPA not only set out a more stringent,
pro-death penalty standard of review that the circuits were required to follow, it
also created a significant hurdle that capital defendants must meet in order to allow
for the issuance of an evidentiary hearing in order to prove their claims (Harvard

36
The seven case characteristics is by no means an exhaustive list of aggravating case
characteristics in death penalty cases. For example, California has a total of 22 separate special
circumstance findings that justify the imposition of the death penalty (Cal. Pen. Code, 190.2).

100
Law Review 2000). This pro-death penalty signal from Congress should have
resulted in fewer death penalty cases being overturned and a much higher
percentage of death penalty cases being upheld for the state. In order to test this
theory, a dummy variable was created in which all death penalty cases that applied
the legislation were coded 1 and those that did not were coded 0
37

The second signal theory variable examined whether signals sent from the
governor in the state where the death penalty conviction was handed down had any
impact on circuit court decision making. This theory is based on the assumption
that Republican governors are highly supportive of capital punishment where
Democratic governors are less supportive. If the theory holds true, a stronger
correlation should be seen between circuits with cases coming from states with
Republican governors versus circuits with cases coming from states with
Democratic governors. A dummy variable was used for the political party of the
governor with Republican governors being coded 1 and Democratic governors
being coded 0.
Hierarchical Variable
An independent variable was created to test the hierarchical theory of
judicial decision making wherein federal circuit courts are guided by the policy
changes of the Supreme Court. According to scholars, the world of law has long
recognized the relationship between a principal or one who has permits or directs

37
Although AEDPA was signed into law on March 24, 1996, the USSC ruled that the application
of the AEDPA only applied to habeas corpus cases that were filed after the date of enactment (see
Lindh v. Murphy, 518 U.S. 651 (1996)).

101
another to act for his benefit and subject to his direction or control and an agent,
one who acts in the behalf of the principal (Songer 1987, 674). This principal-
agent theory, as it is known, explicitly questions the degree to which agents act for
their principals versus the extent to which they act on their own behalf. In the
context of the Supreme Court and the circuits courts, the question becomes to what
extent the circuit courts listen or are guided by policy shifts of the Supreme Court.
The regression that tested the hierarchical model was run separate and apart
from the other regressions. In this analysis, the independent variable in the
regression represented the single percentage of liberal death penalty decisions
issued by the Supreme Court in every year from 1980 to 2004. The dependent
variables were the percentages of liberal death penalty decisions calculated by year
for all of the circuits combined and each circuit independently. A total of three
analyses were run. The first analysis examined correlations between the Supreme
Court and the circuits within the same year. A second analysis looked at the
correlations between the Supreme Court and circuits with a one-year time lag for
the circuit decisions being used to ensure that appellate courts judges had time to
become familiar with changes in Supreme Court policy (Songer, 1987, 834). The
third analysis incorporated a two-year time lag. The ultimate question for the
hierarchical analysis was to determine whether a correlation existed between
United States Supreme Court and circuit court decision making on capital
punishment cases.

102
Findings and Discussion:
Bivariate Model - Logistic Regression Analysis

In order to gain a full understanding of which independent variables play
an integral role in judicial decision making of circuit courts on death penalty cases
for the case sample studied, it was necessary to first examine the bivariate model.
A total of 11 individual dummy variables were coded for each independent
variable noted above (Appendix B.) Each of these variables represented one of the
theories of judicial decision making being tested in this study.
38
For purposes of
the bivariate model, each of the 11 independent variables was regressed against the
traditional outcome dependent variable where each case was coded 0 for a pro-
death penalty decision and 1 for a anti-death penalty decision.
39

The bivariate model revealed that 3 of the 11 independent variables were
significant at the .05 level. These significant variables included: panel ideology,
governor and the post-AEDPA variable. The police officer variable fell just
outside the .05 level of significance at .099. Although the reduction of variables
from 11 to 3 would appear to assist in a better understanding of which variables
play a role in circuit court decision making, such really is not the case as each of
the variables represent each of the three judicial decision making theories being
tested: panel ideology (attitudinal theory), governor and post-AEDPA (signal

38
The independent variable that tested the hierarchical theory of judicial decision making was
omitted from the fully specified model. In order to properly test the hierarchical theory, it was
necessary to regress different dependent and independent variables than those used to test the other
three theories (legal, attitudinal and signal theories). Thus, the hierarchical regression, as will be
discussed later, was done separate and apart from the fully specified models.
39
The Coding Continuum dependent variable was only used in the ordered outcome analysis
and the multinomial outcome analysis.

103
theory) and police officer (legal theory). That being said, the data does clearly
indicate that the bulk of the legal theory variables, seven of eight to be exact, were
not statistically significant and thus do not appear to play a role in circuit court
decision making within the four circuits studied.
The next inquiry must determine whether the observed effect of the three
independent variables that were found to be significant had the correct sign. The
expected effect of each of the independent variables was negative as the standard
dependent variable was coded 0 for pro-death penalty decisions and 1 for anti-
death penalty decisions. Each of the independent variables was coded 1 for
cases in which a particular pro-death penalty variable was found to be present in a
case and 0 if not. In light of the construction of the dependent and independent
variables, the expected effect was negative for each independent variable. Of the
three independent variables that were found to be within the .05 level of
significance, all three variables had the proper sign. It must therefore be
concluded that each of these variables played a least some role in terms of judicial
decision making at the circuit court level on death penalty cases. Conversely, it
can also be said that the remaining variables that were either not statistically
significant and/or their observed effect did not meet the expected effect. Thus,
these variables did not impact judicial decision making in this area of the law.
Findings and Discussion:
Fully Specified Model - Logistic Regression Analysis

The results from the fully specified model reveal additional insight into the
decision making processes of circuit court judges. Two of eleven independent

104
variables were statistically significant within the .05 level, including: panel
ideology (-.555) and the post-AEDPA (-1.143) variables with the adult female
variable falling just outside the level of statistical significance (Appendix C). The
signs of the coefficients for the panel ideology variable and the post-AEDPA
variable were negative and thus the observed effect matched the expected effect.
These results mirrored the findings of the bivariate model to a large degree. Two
of the three variables found significant in the bivariate model (panel ideology and
post-AEDPA) were found significant in the fully specified model. Perhaps equally
important was the finding that all of the remaining independent variables that were
not statistically significant in the bivariate model where likewise no significant in
the fully specified model.
The lstat table calculated the predictability of the fully specified model for
both pro and anti-death penalty decisions. The fully specified model accurately
predicted pro-death penalty decisions in 92.21% of pro-death penalty decisions, a
very high percentage indeed. However, the model only accurately predicted anti-
death penalty decisions in 37% of the cases.
40
Thus, the lstat table clearly
indicates the fully specified model does an exceptional job at predicting pro-death
penalty decisions but not anti-death penalty cases. The pseudo-R
2
for this model
was 22.14%. However, it must be noted that there is debate over the usefulness of
this statistic given the problems with the R
2
measures in general and the pseudo-R
2

specifically. Thus, most scholars rely on alternative statistics for determining the

40
The model accurately predicted 639 out of 693 pro-death penalty decisions. Conversely,
the model only accurately predicted 82 of 225 anti-death penalty decisions.

105
utility of their models (see Benesh 2003; see also Hagle and Mitchell 1992, King
1986, Achen 1982).
Findings and Discussion:
Magnitude of Effect for Each Independent Variable

The regression analysis conducted above demonstrates that two main
variables (panel ideology and post-AEDPA) consistently showed statistical
significance with the correct sign.
41
However, these analyses are limited in that
they do not allow for the determination of the actual impact of each individual
variable. Political Scientist Gary King created a calculation, known as First
Differences, in order to gain a better idea of the actual impact of variables in non-
linear regression.
42
The calculation distills the magnitude of effect of each
independent variable given a particular change in a single explanatory variable,
while statistically holding constant the other variables in the equation (King
(1998), 107). In other words, a coefficient X in the modal case has Y effect
on the percentage of cases being pro-death penalty.
The results were quite revealing. As with the regression analysis, two
variables showed significant magnitudes of effects with the remaining variables
having little impact. The panel ideology variable had the largest effect wherein a
change in the variable from 0 (all Democratic panel) to 3 (all Republican
panel), resulted in the likelihood of a pro-death penalty outcome increase by an

41
The magnitude of effect can only be calculated for regressions using the dichotomous outcome
dependent variable and not the ordered dependent variable.
42
The calculation is as follows: D
logit
= [1+ exp(-X
j
(b)
b
j
X
*

b
*
)]
-1

- [1+ exp(-X
j
(a)
b
j
X
*

b
*
)].
-1

106
astounding 37.6%. Likewise, the post-AEDPA variable a showed sizable
magnitude of effects at 18.9%. All of the other variables that were tested for their
magnitudes of effect had results of less than 9% effect. Thus, the calculation for
the magnitudes of effect demonstrates that the panel ideology and post-AEDPA
variables had the largest effect, far and above the effects of the remaining nine
variables.
Findings and Discussion:
Fully Specified Model - Ordered Logistic Regression

As noted earlier, two dependent variables were utilized in this study to
determine the impact of various independent variables on circuit court decision
making on capital punishment cases. The first dependent variable used the
traditional coding method of creating a dummy variable of 0 for cases that were
pro-death penalty and 1 for cases whose outcome was anti-death penalty. The
second dependent variable used the Coding Continuum that was previously set
out in Chapter 3. Rather than merely coding cases based upon a dichotomous
variable, the coding continuum used a more nuanced approach placing capital case
outcomes along a continuum (1-6) ranging from a strong to moderate to minimal
anti-death penalty cases to minimal to moderate to strong pro-death penalty cases.
In order to properly analyze this dependent variable, ordered logistic regression
was utilized.
43

43
The ordered logistic regression can be considered a parallel regression in order to compare the
results with the prior analysis conducted with the traditional dependent variable.

107
The results from the regression were similar in nature and assist in
confirming the prior findings. Two of eleven variables were found to be
significant at the .05 level (Appendix D). The two variables were the post-
AEDPA variable with a coefficient of .732 and the panel ideology variable with a
coefficient of .488. The expected effect for these coefficients was positive due to
the construction of the coding continuum dependent variable.
44
In this case, the
observed effect of both significant variables matched the expected effect. The
governor variable fell just outside the level of statistical significance with a
coefficient of .283 while none of the remaining variables showed statistical
significance. The pseudo-R
2
was reported at 13.66%, a little more than half of the
previous models using the traditional dependent variable.
When examining the results from the ordered logistic regression, the
findings are largely consistent with what was reported in the logistic regression.
The panel ideology and post-AEDPA variables clearly play a significant role in
circuit court decision making in capital cases. Although the coefficients for the
governor variable and adult female variable fell just outside the .05 level of
significance, by-and-large all of the remaining variables do not appear to be
playing a role in circuit court death penalty decision making. In the end, the

44
Each of the independent dummy variables was coded in a dichotomous fashion wherein the
variables were coded 1 if the presence of a pro-death penalty characteristic was found and 0 if
not. The coding continuum dependent variable was coded using a range from 1 to 6. Cases were
coded 1 for strong anti-death penalty cases, 2 for moderate anti-death penalty cases, 3 for
minimal anti-death penalty cases, 4 for minimal pro-death penalty cases, 5 for moderate pro-
death penalty cases, and 6 for strong pro-death penalty cases. Since the independent variables
went from 0 to 1 (anti-death penalty to pro-death penalty) and the coding continuum went from
1 to 6 (anti-death penalty to pro-death penalty), the expected effect of the coefficient is positive

108
usefulness of recoding the dependent variable from a standard dichotomous
variable into a continuum gave further confirmation about which variables take a
prominent role circuit court decision making.
Findings and Discussion:
Fully Specified Model Using Multinomial Analysis

Multinominal regression was performed for comparison purposes. The
fully specified model was found 2 of 11 variables were statistically significant
with the proper sign in which the observed effect matched the expected effect.
The variables were the robbery variable listed under the minimal anti-death
penalty category with a coefficient of .945 and the post-AEDPA variable listed
under the moderate pro-death penalty category had a coefficient of 1.63. These
results clearly support the findings from the prior models wherein the post-
AEDPA variable was relevant in the death penalty decision making process.
However, other variables such as case characteristics (robbery variable) appear to
play at least some role when using this alternative form of analysis as evidenced
by these results. No other variables were found to be significant in any of the
other categories. The pseudo R
2
for this model was approximately the same as
seen with the logic regression when incorporating the dichotomous dependent
variable at 25.95%.
Findings and Discussion:
Hierarchical Analysis

In order to test the hierarchical theory of judicial decision making, a
separate analysis was run using Ordinary Least Squares estimates (OLS) in order

109
to determine whether a relationship could be found between the Supreme Courts
decisions and circuit court decisions on capital punishment cases (see Heise 1998,
603). The percentage of yearly pro-death penalty decisions for the each circuit
along with a combined variable of all the circuits acted as the dependent variables.
Similarly, the percentage of yearly pro-death penalty decisions by the Supreme
Court acted as the independent variable.
The results from the analysis showed a very little evidence to support the
hierarchical theory. The first analysis incorporated no lag time between Supreme
Court decision making and the circuit court decisions on death penalty cases.
None of the circuits showed statistical significance at the .05 level. Both the Fifth
and Sixth Circuits were relatively close with the Fifth Circuit reporting .116 and
the Sixth Circuit at .139 but the fact remains that none of the circuits, whether
combined or viewed separately demonstrated statistical significance. The data
from the second analysis incorporated a one-year time lag and the results showed
the Fifth Circuit fell within the .05 level of significance at .049. However, the
coefficient was negative at -.271 which indicates an inverse relationship between
the decision making of the Supreme Court and the Fifth Circuit, thus going
contrary to the hierarchical theory. None of the other circuit variables showed
statistical significance. Finally, the third analysis incorporated a 2 year time lag
but again did not show the existence of a hierarchical relationship. None of the
regressions that incorporated the two year time lag fell within the .05 level of
significance. The all-circuits variable did come close to being statistically relevant

110
with the all circuits variable reporting .106 level of significance. This single
example supporting hierarchical theory is weak, at best.
The data from each of the analysis clearly leads to the conclusion that
very little relationship exists between the Supreme Court and the four circuits
studied herein with regard to capital punishment decision making. This conclusion
is not terribly unexpected in light of two key facts that surround the Supreme
Court and circuit courts. First, as noted earlier, the Supreme Court is only able to
address a very small faction of the cases that are presented to it each year. OBrien
(2000), 161-2, notes the high court hears argument and issues opinions in less than
one percent of the 10,000 petitions that are filed with the court each year. Such a
low level of responsiveness by the Supreme Court to the lower courts, and the
circuit court in particular, lessens the nature of the principal-agent relationship. It
is reasonable to suggest that when agents do not feel a strong sense of relationship
to their principal, then agents are more likely to conduct their business as they see
fit rather than at the bidding of their principal.
Second, the issue of capital punishment in the United States has been and
will continue to be controversial within society. Due to the contentious nature of
the issue, judges who rule on such cases in lower courts may be more inclined to
follow their personal beliefs on the issue and not the policies of higher courts.
Thus, it appears that as long as the circuits are given free reign to decide cases with
only minimal ramifications from the Supreme Court on controversial issues such
as capital punishment, the hierarchical theory of judicial decision making will

111
continue to play only a modest role, at best, in explaining why judges on the
federal circuit decide cases in the manner that they do.
Conclusions
Several conclusions can be drawn from each of the above analysis when
taken as a whole. First, the data demonstrates that only a limited number of
variables are involved in circuit court decision making on capital punishment
cases. When examining the results from each model with the possible exception
of the multinomial model,
45
it is evident that two main variables are responsible
for the vast majority of circuit court decision making in death penalty cases within
this case sample. The panel ideology variable that represented the attitudinal
theory and the post-AEDPA variable, which stood as a proxy for the signal theory,
were both found to be significant with the observed effect matching the expected
effect. It is thus clear that circuit judges are clearly driven by their political
ideology on death penalty decisions as well as institutional limitations on their
decision making from other political branches such as Congress. The importance
of these two main variables was also confirmed by their magnitudes of effects in
which these variables showed sizable effects for each variable.
The importance of the above analyses should not only be looked at to
determine which variables were significant but also which variables were found
not to play a large role in judicial decision making. According to the logistic and

45
The results from the multinomial analysis showed that at least one case characteristic (robbery)
was significant with the proper sign. Thus, it would appear that under certain types of analysis,
variables such as the case characteristics may show a higher level of importance.

112
ordered logistic regressions, the political party of the governor and the case
characteristics showed no statistical significance on judicial decision making at the
circuit court level. That being said, the multinomial analysis did demonstrate that
certain type of analysis can result in case characteristics taking a more prominent
role. Although such evidence shows that case characteristics can play a role in
certain cases and should not be totally disregard in terms of impact, it is likewise
clear that the affect of such variables are far less important than the two main
variables.
The fully specified model using the traditional dependent variable in
logistic regression also set out the accuracy of the models predictability. As noted
earlier, this model accurately predicted pro-death penalty cases 92.21%. This
number shows a very high level of predictability for pro-death penalty cases.
However, that conclusion must be tempered by the fact that the model only
accurately predicted the outcomes of anti-death penalty cases from 30%. Thus, it
is clear that the strength of this models is its ability to accurately predict pro-death
penalty outcomes but not anti-death penalty cases.
The predictability of pro-death penalty cases in this model is in-line with
numbers published by other scholars in the field. Klein (2002) conducted a probit
analysis on court of appeals decisions from 1980 to 1990 in order to determine the
impact of court of appeals decisions on other court of appeals judges. Klein used
predicted values and found that his model correctly predicted 78.3% of cases.
Benesh (2002) likewise used a probit analysis to examine the attitudinal, strategic,

113
legal and institutional influences on the Court of Appeals decision making.
Benesh did report that the pseudo R2 was .71 and her model correctly predicted all
cases 94% of the time. Songer, Segal and Cameron (1994) conducted a multivariate
analysis using logit regression. Like Klein, Songer et al. looked at the sign of the
coefficient, the expected effect and whether the model correctly predicted the
outcome. In this case, the model correctly predicted 88% of the cases. Songer and
Haire (1992) conducted a logistic regression using 18 variables of possible influence
on circuit court judges decision making. The authors found their model correctly
predicted 79.2% of the votes.
The final element of this analysis examined the potential impact of
hierarchy by the Supreme Court on the circuit court with regard to capital
punishment policy. The data from the hierarchical analysis in this study is clear
that little, if any, hierarchal effect is taking place between the Supreme Court and
the circuit courts on the issue of capital punishment. The hierarchical model
showed only one instance of statistical significance which occurred between the
Supreme Court and the Fifth Circuit while incorporating a one-year time lag. Such
a finding offers little evidence to support the conclusion that principal-agent theory
is an important component of circuit court decision making.
The above models are unmistakably clear that only a few variables appear
to guide judicial decision making on death penalty cases at the circuit court level.
That being said, this conclusion is based on a case sample from 1980 to 2004 that
involved only four circuits. It is likely, if not probable, that different circuits

114
around the United States will continue to deal with the capital punishment
decision making differently based upon a multitude of reasons. Moreover, simply
because certain factors affect how certain circuits decide death penalty cases in the
present does not mean that other factors will not become relevant in the future. As
support for and against the death penalty continues to change throughout society
so to will the factors that impact judges who rule on such controversial issues. The
law has often been described as living organism that changes to reflect the values
of society (Dickson 2000, 388). It is inevitable that judges working within such a
system will be affected by various factors throughout time. However,
understanding what impacts judges during one specific time and in specific places
does allow for a greater understanding of decision making as a whole.

115
Chapter 5

Institutional Analysis
The institutional theory of judicial decision making takes a starkly different
approach to understanding what affects judges in particular cases. As noted earlier,
institutional theorists suggest that more is at play than merely looking at the
political ideology of the judges (attitudinal model), the law itself (legal model) or
decisions of higher courts (hierarchical model). Rather, to truly understand the
decision making process, the institutional theory asserts that scholars must
examine the impact of the institution itself on decision making and not just the
players within the institution. In other words, collegial courts, acting as small
groups, interact amongst themselves in such a way as to affect the behavior of the
judges in their decision making (Murphy 1966, 1565). Following this approach,
institutions act as a primary type of constraint on the individual behavior of the
judges (Whittington 2000, 611-12). To shed further light on this theory, Sunstein,
Schkade, and Ellman (2004) conducted a unique study to further examine this
theory. The methodology, as set forth below, will be applied to the case sample in
the present study to determine the impact of institutional variables on judicial
decision making on capital punishment cases.
Methodology and Findings
Sunstein examined a number of legal issues that seemed likely to reveal
divisions between Republican and Democratic judicial appointees from a purely
political preference standpoint. The authors looked at cases involving abortion,

116
affirmative action, campaign finance, capital punishment, Commerce Clause
challenges to congressional enactments, the Contracts Clause, criminal appeals,
disability discrimination, industry challenges to environmental regulation, piercing
the corporate veil, race discrimination, sex discrimination and claimed takings of
private property without just compensation. Sunstein presented three hypotheses
for testing which he identified as ideological voting, ideological dampening, and
ideological amplification (Sunstein, Schkade, and Ellman 2004, 304).
Ideological voting refers to a judges ideological tendency to vote on a
particular issue based upon the preference of the party of the appointing president.
Capital punishment is a prime example of such a tendency with the Republican
Party strongly supporting the death penalty and the Democratic Party possessing a
lower level of support. The basis of ideological voting comes from the attitudinal
theory which asserts that judges merely vote according their personal policy
preferences or, stated differently, decision making is based upon the judges
political posture (Segal and Spaeth 1993, 73). The second hypothesis presented by
Sunstein is referred to as ideological dampening. According to Sunstein, a
judges ideological tendency is likely to be dampened if he/she is sitting with
two judges of a different political party. For example, a Democratic appointee
might be less likely to vote in a stereotypically liberal fashion if accompanied by
two Republican appointees, and conversely, a Republican appointee might be less
likely to vote in a stereotypically conservative fashion if accompanied by two
Democratic appointees (Sunstein, Schkade, and Ellman 2004, 304).

117
Sunstein referred to his third hypothesis as ideological amplification.
Under this hypothesis, a judges ideological tendency is likely to be amplified if
he/she is sitting with two judges from the same political party. Sunstein asserts
that a Democratic appointee will show an increased tendency to vote in a
stereotypical fashion if accompanied by two Democratic appointees, and a
Republican appointee will be more likely to vote in a stereotypically conservative
fashion if accompanied by two Republican appointees (Sunstein, Schkade, and
Ellman 2004, 304-5).
Sunsteins dataset examined 4,959 panel decisions, and the 4,874
associated individual judges votes. Cases were drawn from 12 of the 13 circuit
court of appeals. The results from the study showed that all three hypotheses were
strongly confirmed in several different types of cases, including campaign finance,
affirmative action, sex discrimination, disability discrimination, Contracts Clause
violations, and review of environmental regulations. The ideological dampening
appeared to cause a leveling effect, wherein party differences were wiped out by
the influence of panel composition. Democratic appointees, when sitting with two
Republican appointees, were as likely to vote in the stereotypically conservative
fashion as were Republican appointees, when sitting with two Democratic
appointees. Likewise, evidence of strong amplification was found such that party
differences played a lesser role in decision making due to the influence of panel
composition (Sunstein, Schkade, and Ellman 2004, 305).

118
Taken as a whole, Sunstein found Democratic appointees, sitting with two
Democratic appointees, were about twice as likely to vote in the stereotypical
liberal fashion then Republican appointees sitting with two Republican appointees.
This proved to be a far larger disparity than the disparity between Democratic and
Republican votes when either was sitting with one Democratic appointee and one
Republican appointee. When looking at the aforementioned areas of the law,
Sunstein concluded that the political party of the president who appoints the other
two judges on the penal was at least as good a predictor of how individual judges
will vote. All in all, Democratic appointees show[ed] somewhat greater
susceptibility to panel effects than do Republican appointees (Sunstein, Schkade,
and Ellman 2004, 305-6).
Counterexamples in the dataset where also found, including, criminal
appeals, taking clause cases, and the Commerce Clause challenges to
congressional enactments. Cases involving these areas of the law refuted each of
the three hypotheses. Lastly and most importantly to this study, two types of case
decisions--capital punishment and abortion--showed a mixed result with the first
hypothesis (ideological voting) being confirmed and the second (ideological
dampening) and third (ideological amplification) hypotheses being refuted.
Sunstein surmised that these results were indicative of the political divisiveness of
these two issues such that judges apparently vote their convictions and are not
affected by panel composition (Sunstein, Schkade, and Ellman 2004, 306, 312).

119
Sunstein reported the percentage of liberal votes in each of the 13 areas of
the law. As to the capital punishment cases, the authors found that 20% of judges
appointed by a Republican president voted in favor of the capital defendant while
42% of judges appointed by a Democratic president voted in that fashion.
Sunstein measured ideological voting by subtracting the liberal Republican votes
from the percentage of liberal Democratic votes. The differential was 22%.
Sunstein next measured the percentage of liberal decisions (pro-capital defendant)
with two judges on the panel that were appointed by a Republican president versus
the percentage of liberal decisions with two judges on the panel that were
appointed by a Democratic president. Sunstein found that panels with two
Republican appointee judges voted liberal in death penalty cases 29% of the time
while panels with two Democratic appointee judges voted liberal 30% of time.
The differential in that case was a mere 1% and thus was a negligible difference.
This finding was significantly different from the average differential from all the
cases examined. The overall average differential was 14%, which Sunstein noted
was as large as the basic difference between the parties. Thus, the panel colleague
impact was found to occur but not with any degree of significance in capital
punishment cases (Sunstein, Schkade, and Ellman 2004, 315).
Sunstein next examined the pro-liberal voting percentages with all-
Democratic panels, Democratic majority panels, Republican majority panels and
all-Republican panels. The data reflected that 33% of all-Democratic panels
issued liberal votes on capital punishment cases. Democratic majority panels

120
issued a greater percentage of liberal votes in death penalty cases at 38%. This
finding would appear to be counter-intuitive as one might think that the most
liberal panels would be the all-Democratic panels. The majority Republican
panels voted liberally 22% of the time in capital punishment cases. Lastly, the all-
Republican panels voted in support of capital litigants only 18% of the time, nearly
one-half less than the all-Democratic panels. The panel composition influence was
measured by subtracting the percentage of liberal votes by the all-Democratic
panels from the percentage of liberal votes by the all-Republican panels. Sunstein
found the difference to be 15% (Sunstein, Schkade, and Ellman 2004, 314).
Based upon the aforementioned data, Sunstein drew several conclusions
regarding the ideological voting of Republican and Democratic judges on death
penalty cases as well as the panel effects on those judges. Overall, Sunstein found
that capital punishment cases demonstrated a large party difference when dealing
with individual judges votes between those judges that were appointed by a
Republican president versus those judges who were appointed by a Democratic
president. The 22% difference was significantly higher than the average
differential involving all the different types of cases in the sample which was 14%.
Sunstein viewed a 13% average difference in ideological voting as not huge but
substantial.
46
Since the party effect differential for capital punishment cases was
22%, some 9% above the average difference, it is clear that strong ideological
voting does take place with regard to capital punishment cases. However,

46
The average differential in ideological voting across all of the areas of law examined was 13%
(Sunstein, Schade, and Ellman 2004, 314).

121
according to Sunstein, there was practically no difference between panel
colleague votes for Republican and Democratic appointees with only a 1%
differential between the percentage of liberal votes for panels with two or more
Republican appointees versus panels with two or Democratic appointees (Sunstein,
Schkade, and Ellman 2004, 314).
The panel composition numbers appeared to be more revealing with regard
to voting dampening and amplification. A consistent difference between all-
Republican panels (RRR) and Republican majority panels (RRD) versus
Democratic majority panels (DDR) and all Democratic panels (DDD) was reported
by Sunstein. As noted above, an all-Republican appointee panel voted liberally
18%. Republican majority panels reported a 4% increase in liberal voting over the
all-Republican appointee panel at 22%. Democratic majority panels voted
liberally 38% of the time or 20% above the all-Republican appointee panel and all-
Democratic panels voted liberally 33%. Although the differential from all-
Democratic appointee panels from all-Republican appointee panels was 15%, this
number was significantly lower than the average difference across all cases which
was found to be 27% (Sunstein, Schkade, and Ellman 2004, 314).
Sunstein draws the conclusion that very little is occurring in capital
punishment decisions with regard to the impact of the panel other than ideological
voting. He bases this conclusion by first calculating the statistical probability of
support for the death penalty with each panel composition based upon the sheer
number of people leaning in a certain direction, or in other words, by putting

122
aside the possibility that the composition of the panel effects the decision making
of the panel. For example, Republican appointees were likely to rule in favor of a
particular type of program 40% of the time, whereas Democratic appointees were
likely to rule in favor 70% of the time. Using a simple statistical analysis, it
follows that an all-Democratic panel would rule in favor 78% where an all-
Republican panel would rule in favor 66%
47
(Sunstein, Schkade, and Ellman 2004,
314). Applying this statistical formula to Sunsteins death penalty sample, the
predicted probability of liberal rulings was as follows: DDD = 38%, RDD = 27%,
RRD = 17% and RRR = 10%. Again, Sunsteins actual findings were reported as
follows: DDD 33% (-5%), RDD=38% (+11%), RRD= 22% (+5%), and RRR =
18% (+8%). Thus, according to Sunstein, these differences were not sufficient to
conclude that panel composition was affecting the decision making process
(Sunstein, Schkade, and Ellman 2004, 327).
Finally, Sunstein focused specifically on the liberal voting of Republican
appointees on all-Republican panels, Republican appointees on Republican
majority panels, a Republican appointee on Democratic majority panels, a
Democratic appointee on a Republican majority panels, Democratic appointees on
Democratic majority panels and Democratic appointees on all-Democratic panels

47
These figures come from the multinomial probabilities of getting at least two votes to uphold (a
Yes vote), given the panel composition. For a three-judge panel, there are four ways to get a
decision to uphold votes of YYY, YYN, YNY, and NYY, from judges 1, 2, and 3, respectively.
For example, for an all-Democratic panel (DDD), the probability of a vote to uphold the program is
P(YYY) + P(YYN) + P(YNY) + P(NYY) = .7*.7*.7 + .7*.7 (1-.7) + .7*(1-.7)*.7 + (1-.7) * .7*.7 =
.343 +.147 + .147 + .147 = .784, which rounds to 78%. For one Republican and two Democrats
(RDD), the calculation is .4*.7*.7 + .4*.7 (1-.7) + .4*(1-.7)*.7 + (1-.4) * .7*.7 = .196 + .084 + .084
+ .294 = .658 or 66% (Sunstein, Schkade and Ellman 2004, 329).

123
to determine the possible impact of majority and minority status on panels.
Republican appointees voted for capital defendants 19% of the time on all-
Republican appointee panels. Likewise, Republican majorities on Republican
majority panels voted also voted liberally 19%. Republican appointed judges who
sat on Democratic majority appointees on majority panels voted liberal 24% of the
time. Democratic appointees voted liberal 37% when sitting on all-Democratic
panels. Democratic majorities on Democratic majority panels voted for capital
defendants 44% and Democratic appointed judges sitting with majority Republican
panels voted liberal 40% of the time (Sunstein, Schkade and Ellman 2004, 328).
Sunstein suggests since the all-Republican panels, the Republican majority panels
and the Republican appointees who sat with Democratic majority panels were all
within 5% of each other (RRR: 19%, RRD: 19%, RDD: 24%) that such a finding
indicates very little panel composition impact on judicial decision making.
Likewise, similar findings were found with liberal voting on all-Democratic
panels, Democratic majority panels and Democratic appointees sitting on
Republican majority panels (DRR: 40%, DDR: 44%, DDD: 37%). Since the
percentages of support were within 7% of each other, no significant panel
composition impact was found to have taken place. Thus, according to Sunstein,
these differences did not demonstrate any significant level of panel effects.
Findings and Discussion
The present study examined a total of 918 circuit panels that exclusively
addressed death penalty decisions from across four circuits (4
th
, 5
th
, 6
th
, 9
th
) and the

124
accompanying 2,795 votes of individual judges. Following the format laid out by
Sunstein, each judge was classified based upon the political party of the appointing
president. It was found that 16.3% of the Republican-appointed judges voted in a
liberal fashion in favor of capital defendants. Democratic-appointed judges voted
liberally nearly twice as often at a rate of 32.3% (Appendices E & F). At face
value, these numbers indicate that party voting, as suggested by Sunstein, is
clearly at play in term judicial decision making on death penalty cases. It is
interesting to note that the percentage of liberal voting by both Republican and
Democratic appointees was down from what was found in Sunsteins study
(Republican-appointees: 20%, Democratic-appointees: 42%). This difference may
be explained by the region in which the cases were drawn. Sunstein drew his case
sample from 12 circuits while this study took its cases from four circuits,
representing northern, southern, eastern, and western regions. It is possible that
these four circuits taken as whole tend to be more conservatively-oriented than
some of the circuits included in the Sunsteins study.
The degree of ideological voting was derived by subtracting the percentage
of liberal Republican judicial votes from the percentage of liberal Democratic
judicial votes. In this case, the differential was 16%. This finding is 6% lower
than that seen in Sunsteins study. However, Sunstein noted that the average
differential across all the various types of cases was 13% and this finding was not
huge but substantial. Since Sunstein found the 13% differential to be substantial

125
then it is only reasonable to conclude that the 16% differential found in this study
to be substantial evidence of ideological voting as well.
Panel colleagues were also studied to determine the extent of influence by
party affiliation, as done by Sunstein. The percentage of liberal voting in death
penalty cases was calculated with panels that contained Republican majority
appointed judges versus Democratic majority appointed judges. The results were
similar to that seen with the votes of individual judges. Panels with a Republican
majority appointed judges voted liberally15.6% of the time while Democratic
majority panels voted liberally 37.2% (Appendices E & G). Once again,
Democratic majority panels voted in favor of capital defendants over twice as
often as their Republican counterparts and at rates very close to what were seen
with voting of individual judges. Thus, the very same ideological voting taking
place with individual judges was seen with panel colleagues.
Sunstein drew the conclusion that very little party affiliation influence was
occurring when looking at Republican majority panels versus Democratic majority
panels. He based this conclusion by subtracting the percentage of liberal votes
from panels with Democratic majorities from panels with Republican majorities.
As noted above, when Sunstein followed this process, the differential came to only
1% as Democratic majority panels voted liberally 30% of the time and Republican
majority panels voted liberally 29%. A similar finding in the present case was not
made. The differential in the present study was found to be 21.6%, with
Democratic majority panels voting liberally 37.2% and Republican majority panels

126
voting liberally 15.6%. It is evident that a significant portion of the difference
can be attributed to the very low percentage of Republican majority panels voting
liberal. Thus, based on the 21.6% differential, it can be concluded that a
significant amount party influence is impacting both Republican and Democratic
majority panels as Sunstein viewed a 13% differential as a substantial party
difference.
Perhaps the most enlightening element of study was the examination of the
panel compositions. As with Sunstein, the percentage of liberal voting was
calculated for all-Republican panels, Republican majority panels, Democratic
majority panels and all-Democratic panels. The numbers revealed that 10.9% of
all-Republican panels (RRR) voted liberal in death penalty cases. Republican
majority panels (RRD) voted liberal 18.1% of the time. Democratic majority
panels (RDD) voted liberally 33.8% and all-Democrat panels (DDD) voted in
favor of capital defendants 52.9% of the time (Appendices E & H). Taking these
numbers at face value, it appears as the number of Democratic appointed judges
entered onto three judge panels, the support for the traditional liberal position
increased. This increase was consistent through each of the four combinations of
panels with the least support for the liberal position coming from all-Republican
panels, then Republican majority panels, then Democratic majority panels and
finally the greatest support for liberal voting in death penalty case being seen with
all-Democratic panels.

127
The numbers found in the instant study differ from those found in
Sunsteins study in several respects. First, the all-Republican, Republican majority
and Democratic majority findings made by Sunstein (RRR: 18%, RRD: 22%,
DDR: 38%) were all higher in terms of liberal voting than what was found in the
present study (RRR: 10.9%, RRD: 18.1%, DDR: 33.8%). However, the all-
Democratic panels voted more liberally in this study (DDD: 52.9%) as opposed to
the all-Democratic panels in Sunsteins study (DDD: 33%). A second difference
was seen with the panel composition that reflected the highest percentage of liberal
voting. Sunstein reported that the Democratic majority panels voted liberally 38%
of the time which was 5% higher than the liberal voting percentage for all-
Democratic panels. This finding would appear to be counter-intuitive as one
would expect the panel composition with the greatest liberal voting to be the all-
Democratic panel. This was the very finding in the present study as the panel
composition with the highest liberal voting percentage was the all-Democratic
panel at 52.9%. Sunstein does not attempt explain this deviation. Thus, the liberal
votes from the various panel composition in the present study follows the
anticipated pattern with greater liberal voting taking place on panels with a greater
number of Democratic appointees where such was not the case in Sunsteins study.
Sunstein notes that analyzing the numbers themselves does not accurately
reflect the potential impact of panel composition on judicial decision making. To
do so, it is first necessary to determine the probability that certain panels will vote
liberally based upon the political appointment of the judges. Once the

128
probabilities are determined then one can look to the numbers to see if more is at
work in terms of decision making other than the political posture of the judges.
As noted above, Sunstein used a statistical formula to determine the liberal
percentage of support in capital punishment cases which did not account for
influence from judicial colleagues (see footnote 47). Using this formula, the
predicted probability for a liberal vote in a death penalty cases in the instant study
was as follows: all-Republican panels: 8%, Republican majority panels: 11%,
Democratic majority panels: 16% and all-Democratic panels: 22%. Thus, any
increase over the predicted probabilities would suggest that something more was at
play in terms of judicial decision making on death penalty cases beyond mere
voting based on political stereotypes.
The percentages of actual liberal voting was found to be as follows: all-
Republican voting: 10.9% (2.9% over the probability), Republican majority
voting: 18.1% (7.1% over the probability), Democratic majority voting: 33.8%
(17.8% over the probability) and all-Democratic panels: 52.9% (30.9% over the
probability). These numbers reveal several important points. First, unlike
Sunstein, the liberal voting in each of the four panel compositions had a number
that was above its predicted probability.
48
The lowest of these numbers was the
all-Republican panels and the highest was the all-Democratic panels. These
findings support the voting amplification theory espoused by Sunstein that when
judges sit with a panel that is composed of members from the same party, there

48
Only 3 of 4 panel combinations were above their predicted probabilities in Sunsteins study
(RRR, RRD, DRR).

129
voting will be amplified in the direction of that party. Second, although the all-
Republican and Republican majority panels only showed a modest increase over
their respective predicted probabilities (RRR: +2.9%, RRD: +7.1%), the increases
over the Democratic majority and all-Democratic panels were quite substantial.
Democratic majority panels voted liberally in death penalty cases 17.8% above its
predicted probability and all-Democratic panels voted 30.9% over its predicted
probability. At a very minimum, the percentage of liberal votes in Democratic
majority and all-Democratic panels clearly supports the theory that more is at hand
than judges simply voting their personal policy preferences and, in great
likelihood, the composition of at least the Democratic majority and all-Democratic
panels are being affected by the composition of those particular panels. This
finding is consistent with Sunsteins conclusion that Democratic appointed judges
appear to be more susceptible to the influence of its colleagues than their
Republican counter-parts (Sunstein, Schkade, and Ellman 2004, 306).
Lastly, the present study examined the potential impact of judicial decision
making based on the composition of the panel by focusing on the percentage of
liberal voting by Republican appointees on all-Republican panels, the Republican
appointees on Republican majority panels, the Republican appointee on
Democratic majority panels, the Democratic appointees on Republican majority
panels, the Democratic appointees on Democratic majority panels and the
Democratic appointees on all-Democratic panels, as was done by Sunstein.
Republican appointees voted liberally on all-Republican panels 10.3% of the

130
time.
49
It was found that Republican appointees on Republican majority panels
voted liberally 17.2% while Republican appointees sitting on a Democratic
majority panels voted liberally 22% (Appendices E & I). The largest differential
from these numbers was found between the all-Republican panels and the
Republican appointee sitting on a Democratic majority panels at 11.7%. This data
appears to support both voting dampening and voting amplification. As the
numbers of Democratic appointees were added to three judge panels, the
percentage of liberal votes increased consistently from RRR: 10.3% to RRD:
17.2% to RDD: 22%. Thus, conservative votes were amplified with an all-
Republican panels and Republican majority panels. However, when a Republican
appointee was placed with two Democratic appointees, the Republicans
conservative voting was dampened resulting in a higher percentage of liberal
voting on those panel types.
The same phenomena were seen when looking at all-Democratic panels,
Democratic majority panels and Democratic appointees with Republican
majorities. Democratic appointees on all-Democratic panels voted liberally
46.6% of the time. Democratic appointees on a Democratic majority panel voted
liberally 35.4%, and Democratic appointees sitting on Republican majority panels
voted liberally 23.9% (Appendices E & I). Thus, a Democratic appointees vote

49
This number is different from the percentage of liberal voting by the all-Republican panels
because the percentage of liberal votes by all-Republican panels does not count dissenting
Republican votes, just the panel votes. The percentage of liberal voting by all-Republican judges
on all-Republican panels does incorporate dissenting votes, thus resulting in a difference in the two
calculations. The same applies to the all-Democratic panel percentages versus the all-Democratic
judge percentages.

131
when sitting with a Republican majority panel was clearly dampened. Once
again we find that the more Democratic appointees on a panel resulted in greater
liberal voting which was consistent along with three different combinations of
panels. More importantly, the differential between the all-Democratic appointee
panel and the Democratic appointee with a Republican majority panel was 22.7%.
This finding is far greater than the 7% differential found by Sunstein. Therefore,
these findings support the conclusion both Republican judicial appointees and that
Democratic judicial appointees are affected in their voting on capital punishment
cases based upon the nature of the panel composition with the Democratic
appointees experiencing greater affect by panel composition.
Conclusions
Several conclusions can be drawn from the data noted above. First and
foremost, ideological voting is without question taking place in death penalty cases
at the circuit court level. Capital punishment is a highly politicized issue in society
that appears to have crossed over into the realm of the judiciary. Both Sunsteins
study and the current study support this conclusion as Democratic appointee
judges voted liberally in favor of capital defendants over twice as much as
Republican appointee judges. This finding was also confirmed by comparing the
percentage of liberal votes when two Republican appointee judges were on a panel
versus two Democratic appointee judges sat on a panel. Moreover, in both the
individual judge data and the panel colleague data, the differential between
Democratic appointee judges and Republican appointee judges was substantial

132
allowing for the conclusion that at a minimum, the political posture of the judges
sitting on capital punishment cases is impacting the decision making.
Perhaps the more interesting question to be discussed is not whether
ideological voting is taking place at the circuit court level on capital punishment
cases but rather what other effects are taking place in these particular cases.
Sunstein suggests that nothing beyond ideological voting is seen from the data and
therefore ideological voting is the dominant theory on these particular types of
cases. This finding was not confirmed in this study. Examining the four different
panel combinations revealed that ideological amplification was taking place. Each
of the four panel combinations was above its predicted probability with the
Democratic majority panel and all-Democratic panel being well above their
predicted probability at 17.8% and 30.9%, respectively. Although the all-
Republican and Republican majority panels were only marginally above the
probabilities, the fact remains that they were above. It is also worth noting that
voting amplification appears to be taking place with Democratic appointed judges
and not the Republican appointed judges. Sunstein acknowledged this was the
case when viewing various case types but not capital punishment cases. For
whatever reason, Democratic appointed judges appear to be more susceptible to
voting amplification than that of Republican appointed judges. Thus, ideological
amplification appeared to affect these two panel types, at a minimum.
Ideological amplification was also found to be present when examining the
percentage of liberal votes from Democratic appointees on all-Democratic panels,

133
Democratic appointees on Democratic majority panels, and Democratic
appointees on Republican majority panels. Again, the data showed the differential
from all-Democratic panels to Democratic appointees with Republican majorities
was 22.7%, a large number indeed. Thus, as the number of Democratic appointees
increases, so too did the liberal tendency of those judges to vote in a liberal fashion
on capital punishment cases, thereby, amplifying their liberal voting. Republican
appointees on all-Republican panels and Republican appointees on Democratic
majority panels had a differential of 11.3%, much small than seen with the
Democratic appointees and relatively similar to what was found in Sunsteins
study. Thus, based upon these numbers, it appears that it is only the Democratic
appointee judges that are affected by voting amplification.
In addition, ideological dampening appeared to be playing a role in
decision making when comparing Democratic appointees on Republican majority
panels with the Democratic appointees on Democratic majority panel and all-
Democratic panels. The fewer Democratic appointees on a particular panel, the
lower rate of liberal voting took place. It therefore appears that panels with one or
more Republican appointees were dampened in their liberal voting. The converse
was equally true with lower liberal voting taking place when more Republican
appointees were put on panels. Although the differential between all-Republican
appointee panels and a Republican appointee panel with a Democratic majority
was lower at only 11.3%, the fact remains the more Republican appointees placed

134
on a panel, the more their votes were amplified and the fewer Republican
appointees put on a panel, the more their votes were dampened.
Although any suggestion as to why amplification is seen more with
Democratic appointees and not as much with Republican appointees is
supposition, it is reasonable to suggest that judges may feel a sense of solidarity
with other judges that possess similar ideological positions. Assuming this is the
case, few can argue against the claim that capital punishment is one of the most
contentious issues found in society both past, present and future. That being so, a
highly contentious issue such as capital punishment could very well have the effect
of bringing Democratic appointed judges together to strongly push for the liberal
position in favor of capital defendants. Conversely, capital punishment cases may
bring Republican appointed judges together to support their stereotypical
conservative, pro-death penalty position. This theory of solidarity is supported by
all of the data discussed including both individual judges, panel colleagues and
panel composition.
Both Sunsteins study and the current study showed numerous similarities
in terms of the voting of individual judges as well as the affects of panel
composition. However, some differences were revealed between the two studies
which require at least some explanation. The main difference is found with the
findings on panel composition. Although the all-Republican and Republican
majority liberal voting were relevantly consistent between the two studies, the
present study found much higher liberal voting with Democratic majority panels

135
and all-Democratic panels than that found in Sunsteins study. As suggested
earlier, one possible explanation is based on the pool of cases that each study
selected. Sunstein took cases from 12 circuits around the United States whereas
the instant study focus on four regional circuits. One of the circuits in particular
used in this study is the Ninth Circuit which has been notoriously liberal by just
about any standard. Thus, the use of this particular circuit may have resulted in a
higher liberal voting record by Democratic majority and all-Democratic panels.
The evidence seems clear that the issue is not whether a judicial institution
itself affects the decision making process in capital punishment cases but rather to
what degree. At a minimum, it can be concluded that when dealing with highly
politicized issues such as capital punishment, judges are not only affected by their
political policy preferences but also by their fellow judges on the panel. Capital
punishment litigation is but one area of legal decision making where institutional
theory has found relevance. It is only logical to assume that this theory possesses
relevance across the legal spectrum.

136
Chapter 6

Hierarchical Analysis

The hierarchical approach to understanding judicial decision making
presents its own unique theory to explain why judges vote in the manner that they
do. As noted earlier, the principal-agent (hierarchical) theory questions the degree
to which agents act on behalf of their principals versus the extent to which they act
on their own behalf (Songer, Segal, and Cameron 1994, 673). An agency
relationship is established when a principal delegates some rights to an agent who
is bound by a formal or informal contract to represent the principals interests in
return for payment of some kind (Benesh 2002, 7). The principal-agent theory is
particularly relevant when discussing circuit court decision making. Circuit courts
are beholden to only one court, that being, the United States Supreme Court. Due
to the nature of this relationship, principal-agent theorists have argued that the
circuit courts may act as an agent to their principal, the Supreme Court (see Klein
2003; Benesh 2002; Songer, Segal and Cameron 1994; Epstein 1995).
Songer (1987) set out to examine the nature of the relationship between
circuit courts and the Supreme Court. Specifically, Songer sought to determine
whether circuit courts of appeals react to changes in decision trends by the
Supreme Court on specific policy issues. Songer used a unique methodology in
which he identified specific periods of decisional trends by the Supreme Court on
labor and anti-trust decisions and then looked to see if the circuit courts changed
their decision making on those issues following the shift by the Supreme Court

137
(Songer 1987, 832). The methodology used by Songer, as set forth below, will be
applied to the case sample in the present study in order to see if circuit courts of
appeals decision making on capital punishment flows from decisional trends of the
Supreme Court.
Methodology and Findings
The focus of Songers 1987 study examined labor and anti-trust decisions
of the courts of appeals and the Supreme Court that were published with full
opinions from 1950 to 1977.
50
During this period, 160 anti-trust decisions were
handed down by the Supreme Court and 1,153 decisions by the courts of appeals.
The Supreme Court ruled on 221 labor decisions while the courts of appeals
decided 4,454 such decisions. Songer classified each case either as liberal or
conservative based upon the widely used definition of liberalism as described by
Goldman (1966) (Songer 1987, 832).
The initial step in conducting the analysis was to compute the percentage
of liberal decisions by the Supreme Court for each year within each policy area. It
was next necessary to determine when and to what extent the decisional trends of
the Supreme Court changed from liberal to conservative or conservative to liberal
in each policy area during the relevant time period. Songer identified decisional
changes by using a difference of proportions or (Z) test in which changes were
deemed statistically significant. Songer reported that the between 1950 to 1958,
the Supreme court announced 54.3% liberal decisions on labor issues. From 1959

50
The courts of appeals anti-trust decisions were extended backward to 1947 as there were too few
cases in the early period studied to permit adequate analysis (Songer 1987, 832).

138
to 1969, the Supreme Court increased its liberal decision making to 79.8%.
Finally, from 1970 to 1977, the high court took a conservative turn issuing a mere
43% of liberal labor opinions. Each decision making shift was statistically
significant to the .001 level (Songer 1987, 833).
The exact same process was followed to determine the decisional trend
changes of the Supreme Court for anti-trust decisions. Songer found that
decisional trends by the Supreme Court were decidedly more liberal after 1956 and
turned more conservative after 1973. Thus, from 1950 to 1956, the Supreme Court
issued 40.6% liberal decisions. Liberal decisions on anti-trust cases jumped to
79.5% from 1957 to 1973 and then fell back to just 37% from 1974 to 1977. Both
of these changes were found to be statistically significant at the .001 level (Songer
1987, 833).
Songer noted that the Supreme Court shifts in labor and anti-trust cases
were not only supported through quantitative methods but also through scholarly
opinion. St. Antoine stated that the Warren Courts main achievement in the labor
field involved a restructuring of intergovernmental relations thereby nationalizing
the regulation of labor relations with industries affecting interstate commerce. St.
Antoine indicated the shift back to a more conservative labor policy began in 1970
when the Burger Court issued two rulings that expressly overruled two of the
Warren Courts pro-union decisions (Songer 1987, 833; St. Antoine 1983, 126,
166). Similar scholarly support was found for Supreme Court shifts on anti-trust
decisions. Kauper wrote that no one could quarrel with the simple assertion that

139
the Warren Court had a significant, if not extraordinary, liberal impact on the
development of anti-trust law (Kauper 1968, 134). This liberalization however
was relatively short-lived as Markovits argued that the Burger Court issued two
major anti-trust decisions that swung the political posture of the Supreme Court
back to the right (Songer 1987, 834; Markovits 1983, 180, 186).
After Songer determined the specific years in which Supreme Court
decisional trends shifted in both policy areas, he then turned his attention toward
labor and anti-trust cases decided by the circuit courts of appeals by classifying
those decisions as liberal or conservative. The percentage of liberal decision
making for those cases was then calculated. Songer recognized that any impact of
Supreme Court on the federal courts of appeals would not take place immediately
and for this reason he lagged the periods used to analyze the appeals courts
decisions by one year to ensure that the appeals court judges had time to become
familiar with the changes in Supreme Court policy. Now that both Supreme Court
and courts of appeals cases were classified, the basic test was to determine whether
the proportion of liberal decisions on the courts of appeals changed to a
statistically significant degree after a policy change by the Supreme Court. Songer
opted to utilize one-tailed difference of proportions since it was predicted that the
Supreme Court and the courts of appeals would change in the same direction
(Songer 1987, 834).
Songer found the courts of appeals followed the Supreme Courts dramatic
increase in liberal labor decisions from the first measured period (1950 to 1958) to

140
the second measured period (1959 to 1969). The difference in liberal decisions
by the courts of appeals increased 9.1%, a difference that was statistically
significant at the .001 level. However, the response by the courts of appeals
following the Supreme Courts return to more conservative labor decisions in the
third measured period (1970 to 1977) was viewed by Songer as somewhat
ambiguous with only a 2% conservative shift, which failed to reach the .05 level.
The anti-trust decisions offered even more support for Songers hypotheses.
Following the Warren Courts adoption of a decidedly more liberal anti-trust
policy from the first measured period to the second, the courts of appeals likewise
became more liberal, a difference of 8.5% that was statistically significant at the
.02 level. Interestingly, when the Supreme Court moved into a conservative
posture during the third measured period, the courts of appeals followed suit at a
difference of 8.6% from the previous level, a difference that is significant at the
.02 level (Songer 1987, 835).
Songer notes the shifting of the courts of appeals votes might be due to
changes in the partisan composition of the lower courts rather than to any Supreme
Court influence. In order to assess this alternative explanation, Songer calculated
the votes of individual judges for each time period with a control for the party of
the appointing president. Songer found the changes in voting patterns of
Republican judges was consistent with his hypothesis of Supreme Court impact.
For both labor and anti-trust decisions, the proportion of liberal Republican votes
increased to a statistically significant degree after the Warren Court policies and

141
then decreased to a statistically significant degree after the Burger Court began to
rule in these areas. Songer also found that Democratic judges followed the
decisional trends of the Supreme Court in both labor and anti-trust decisions, and
the magnitude of the changes was significant at the .01 level in three of three of the
four cases. Although Democratic judges liberal voting was higher than the
corresponding figure for Republican judges in each of the three time periods,
Songer found it significant that the trends over time for both parties followed the
direction of the changes enacted by the Supreme Court (Songer 1987, 836-837).
The results from Songers study supported his hypothesis. For both labor
and anti-trust cases, the proportion of liberal decisions by the courts of appeals
consistently moved in the direction of the Supreme Court trend. For anti-trust
cases, both changes (conservative to liberal and liberal to conservative) were
statistically significant. Shifts on labor cases were only statistically significant to
.05 level when the Supreme Court shifted from conservative to liberal decision
making. In light of these results along with the findings that such results were not
based on changes in the partisan composition of the courts of appeals, Songer was
able to report that the Supreme Court does exercise a considerable hierarchical
impact on the general trends of judicial decision making with regard to labor and
anti-trust policy decisions in the United States Courts of Appeals (Songer 1987,
838-9).

142
Findings and Discussion
Songers methodology was applied to the case sample in the instant study.
First, each death penalty decision issued by the Supreme Court between 1980 and
2004 was classified either as liberal or conservative based upon the outcome of the
decision (Appendix J). All decisions that issued a ruling in favor of the
government were identified as conservative while all decisions that issued a ruling
in favor of the defendant were classified as liberal. The use of this classification
method has been commonly used by scholars when analyzing capital punishment
decisions (see, e.g., George and Epstein 1992, 327; Traut and Emmert 1998, 1170;
Tauber 1988, 196).
An examination of the 25 year period used in the present study showed six
time periods that demonstrated a shift in death penalty decision making by the
United States Supreme Court. Table 6.0 shows that from 1980 to 1986, 8 of 33
death penalty decisions or 24.2% were classified as liberal. During the next six
years (1987 to 1992), the Supreme Court ruled in capital defendants favor in 29 of
56 cases or 51.7%, a significant swing toward liberal voting. From 1993 to 1994,
the Supreme Court took on a conservative direction and only ruled liberally in 3 of
15 cases or 20%. Another shift in trend occurred from 1995 to 1996 in which the
high court ruled in favor of capital defendants in 5 of 9 cases or 55%. From 1997
to 1999, the Supreme Court moved back to a conservative posture by ruling for
capital defendants in only 2 of 11 cases 18.1%. Finally, from 2000 to 2004, the

143
Supreme Court issued liberal rulings in 13 of 25 cases or 52% of the time.
51
The
differences in liberal voting by the Supreme Court between each of the time
periods (1980-1986, 1987-1992, 1993-1994, 1995-1996, 1997-1999, 2000-2004)
was found to be statistically significant within the .01 level while analyzing a one-
tailed difference of proportions test.
52

Table 6.0

United States Supreme Court Death Penalty Decision Making from 1980 to 2004:

Period % Liberal Voting N (Z) Differential
1980-1986 24.2% 33 ----
1987-1992 51.7% 56 .014
1993-1994 20.0% 15 .032
1995-1996 55.5% 9 .086
1997-1999 14.2% 14 .047
2000-2004 52.0% 25 .023

The same classification process was followed for each of the four circuits
examined in the present study while incorporating a one-year time lag as was done
by Songer to ensure that appeals court judges had time to become familiar with
changes in Supreme Court policy (Songer 1987, 834). Table 6.1 shows that the
Fourth Circuit decided 20 death penalty cases from 1981 to 1987 and ruled
liberally 3 times or 15% of the time. From 1988 to 1993, 37 death penalty cases
were decided in the Fourth Circuit and 4 of those cases were decided liberally or
10.8%. The Fourth Circuit decided 18 death penalty cases from 1994 to 1995 and
ruled liberally in only 2 decisions or 11.1%. From 1996 to 1997, the Fourth

52
The Z-Test for 2 Proportions Calculator located at http://www.dimensionresearch.com was
used to determine whether the independent groups of cases were significantly different from one
another.

144
Circuit voted in a liberal fashion in 2 of 27 death penalty cases or 7.4%. The
conservative trend continued from 1998 to 2000 when the Fourth Circuit issued
only 2 of 66 liberal decisions in death penalty cases or 3%. Finally, the Fourth
Circuit issued 4 liberal rulings out of 39 cases or 10.2% from 2001 to 2004.
Table 6.1

Fourth Circuit Death Penalty Decision Making from 1980 to 2004:

Period % Liberal Voting N (Z) Differential
1981-1987 15.0% 20 ----
1988-1993 10.8% 27 n/s
1994-1995 11.2% 18 n/s
1996-1997 7.4% 27 n/s
1998-2000 3.0% 66 n/s
2001-2004 10.2% 39 n/s

The above data indicates that Fourth Circuit decision making on capital
punishment cases during the 25 year period examined was very consistent. There
was no statistical difference at the .01 level between any of the six periods in terms
of liberal decision making. Rather than following the decisional trends of the
Supreme Court in which the Supreme Court moved from conservative to liberal in
three distinct periods of time (1987-1992, 1995-1996, 2000-2004), the Fourth
Circuit started from a conservative posture and the decision making remained
conservative throughout. Thus, it appears that decision making by the Fourth
Circuit with regard to death penalty cases was not affected by Supreme Court
shifts in capital punishment policy.
The Fifth Circuit showed a similar conservative voting pattern on death
penalty decisions as seen from Table 6.2. From 1981 to 1987, the Fifth Circuit
addressed 124 death penalty cases and voted liberally in 38 cases or 30.6%. The

145
Fifth Circuit decided 102 capital cases from 1988 to 1993 and voted in a liberal
fashion only 15 times or 14.7%. From 1994 to 1995, the Fifth Circuit continued to
rule conservatively when it resolved 26 capital cases and voted liberal just twice or
7.6%. The Fifth Circuit ruled on 42 cases from 1996 to 1997 and issued 3 liberal
rulings or 7.1%. From 1998 to 2000, the Fifth Circuit decided 67 capital cases, 3
of which were liberal or 4.4%. Lastly, from 2001 to 2004, the Fifth Circuit
decided 122 death penalty cases with 24 of those decisions being classified as
liberal or 19.6%.
Table 6.2

Fifth Circuit Death Penalty Decision Making from 1980 to 2004:

Period % Liberal Voting N (Z) Differential
1981-1987 30.0% 124 .004
1988-1993 14.7% 102 n/s
1994-1995 7.6% 26 n/s
1996-1997 7.1% 42 n/s
1998-2000 4.4% 67 n/s
2001-2004 19.6% 122 .004

The death penalty decision making by the Fifth Circuit on capital
punishment cases showed a bit more vacillation than what was seen with the
Fourth Circuit. However, the changes observed with the Fifth Circuit voting do
not appear to be related to the decisional trends of the Supreme Court. The
difference in decision making by the Fifth Circuit from the first time period (1981-
1987) to the second (1988-1993) was 15.9%. This difference was statistically
significant at the .01 level. That being said, the movement of the Fifth Circuit
from a conservative position to an even more conservative position did not follow
the trend of the Supreme Court. Where the Fifth Circuit went from conservative

146
voting to even more conservative voting, the Supreme Court went from
conservative voting (1980-1986) to a much more liberal position (1987-1992).
The conservative voting for the Fifth Circuit continued during the third
time period (1994-1995) and the percentage of liberal voting during that time was
not significantly different from the prior period. The Supreme Court voting on
capital cases did move back to a conservative posture during this time frame but
one cannot conclude that the Fifth Circuit was reacting the Supreme Court shift in
voting as the Fifth Circuit case decisions were not statistically different from its
voting in the prior period. In other words, the Firth Circuit continued to vote in a
conservative fashion regardless of the Supreme Court trends. During the fourth
time period, the Fifth Circuit voted in an even more conservative manner from
1996 to 1997 (7.1% liberal votes) while Supreme Court decisions became
distinctly liberal. This conservative trend continued in the fifth time frame in
which the Fifth Circuit became even more conservative (only 4.4% liberal) while
the Supreme Court voted liberal. This pattern of decision making by the Fifth
Circuit indicates a strongly conservative orientation that was not affected by the
policy changes of the Supreme Court.
An interesting finding was discovered with regard to the sixth time frame
(2001-2004). During that time frame, the Supreme Court moved in a statistically
significant liberal manner. Likewise, the Fifth Circuit moved from an extremely
conservative voting position in the fifth time period (4.4% liberal voting) to a more
liberal position (19.6% liberal voting). The difference with Fifth Circuit voting

147
between the fifth and six time frames was statistically significant at a .004 level.
Thus, the liberal movement of the Supreme Court and the Fifth Circuit during the
last time frame allows for the possible conclusion that the Fifth Circuits decision
making on capital cases may have been attributed to a shift in Supreme Court
policy on death penalty cases.
Sixth Circuit decision making on death penalty cases during the 25 year
period was starkly different from what took place in the Fourth and Fifth Circuits.
Table 6.3 indicates that from 1981 to 1987, the Sixth Circuit only ruled on one
capital cases and that decision went in favor of the state. The Sixth Circuit
addressed five death penalty cases from 1988 to 1993 in which the court ruled
liberally in two of those cases or 40%. From 1994 to 1995, the Sixth Circuit
decided on three death cases and ruled in favor of the defendants in two of those
cases or 66% of the time. This high level of liberal voting continued in 1996 and
1997 in which the court ruled liberally in 4 of 8 cases or 50%. From 1998 to 2000,
the Sixth Circuit resolved 17 cases and issued liberal rulings in 6 cases or 35.2%.
Finally, from 2001 to 2004, the Sixth Circuit ruled on 53 capital cases in which 21
or 39.6% were decided in a liberal manner.
Table 6.3

Sixth Circuit Death Penalty Decision Making from 1980 to 2004:

Period % Liberal Voting N (Z) Differential
1981-1987 0.0% 1 ----
1988-1993 40.0% 5 ----
1994-1995 66.6% 3 n/s
1996-1997 50.0% 8 n/s
1998-2000 35.2% 17 n/s
2001-2004 42.8% 53 n/s

148
The Sixth Circuit, like the Fourth and Fifth Circuits, showed little sign of
being affected by the Supreme Court decision making on capital cases. There was
no statistical difference within the .01 level between any of the time periods. Sixth
Circuit decision making on death penalty cases remained high for liberal decisions
in each of the time frames and never dropped below 35%, except for the first time
period in which the Sixth Circuit only decided one death penalty case. Since there
was no vacillation from one time period to the next and none of the differences
between time periods were statistically significant, it must be concluded that the
Supreme Court had no impact on death penalty decision making in the Sixth
Circuit during the time of the study on death penalty cases. Regardless of the
Supreme Courts conservative tone during the third and fifth time frames, the Sixth
Circuit remained true to its liberal position.
Finally, the Ninth Circuit decision making on capital cases showed an even
greater liberal orientation than the Sixth Circuit. Table 6.4 indicates that from
1981 to 1987, the Ninth Circuit decided 11 cases and ruled liberally in 5 or 45.4%
of the time. The Ninth Circuit decided 52 death cases from 1988 to 1994 and ruled
in favor of defendants in 26 of those cases or 50%. The Ninth Circuits liberal
tendency continued from 1994 to 1995 when the court ruled liberally in 13 of 24
capital cases or 54.1%. From 1995 to 1996, the Ninth Circuit support for liberal
decisions turned a bit more conservative with the court only ruling liberally in 16
of 39 cases or 41% of the time. A similar level of support continued from 1998 to
2000 in which the Ninth Circuit ruled in favor of capital defendants in 21 of 49

149
cases or 42.8%. Lastly, Ninth Circuit decision making made a distinctively
liberal turn from 2001 to 2004 when the court issued 33 liberal rulings out of 51
death cases or 64.7% of the time.
Table 6.4

Ninth Circuit Death Penalty Decision Making from 1980 to 2004:

Period % Liberal Voting N (Z) Differential
1981-1987 45.4% 11 ----
1988-1993 50.0% 54 n/s
1994-1995 54.1% 24 n/s
1996-1997 41.0% 39 n/s
1998-2000 42.8% 49 n/s
2001-2004 64.7% 51 .023

Like each of the other circuits, the Ninth Circuits did not appear to be
affected by Supreme Court decision making on death penalty cases. None of the
differences between the first five time frames were statistically significant at the
.01 level. The Ninth Circuit issued a high degree of liberal decision making
regardless of any conservative policy changes that took place by the Supreme
Court. The only time period that showed a difference that was statistically
significant at the .01 level occurred between the fifth time period (1998-2000) and
the sixth (2001-2004). Between these time periods, the Ninth Circuit moved from
a liberal position (42.8% liberal voting) to an even greater liberal position (64.7%
liberal voting). The shift did follow a change by the Supreme Court, however, it is
difficult to conclude that the Ninth Circuit was reacting to Supreme decision
making as the Ninth Circuit went from a liberal position (42.8% liberal voting) to a
more extreme liberal position (64.7% liberal voting) whereas the Supreme Court
went from a conservative position (14.2% liberal voting) to a liberal position (52%

150
liberal voting). What can be said about Ninth Circuit decision making is that
throughout the 25 year period studied, the Ninth Circuit was clearly liberally
oriented and, at best, may have increased its liberal posture based upon a liberal
shift by the Supreme Court in one of five instances.
The data above allows for several conclusions to be drawn regarding the
impact the Supreme Court had on the four circuits studied with regard to decision
making on capital punishment cases during the 25 years studied. First and
foremost, the Supreme Court had little or no impact on the decision making of the
Fourth, Fifth, Sixth and Ninth Circuit with regard to capital punishment cases.
When looking at the combined 20 time period differences from all of the circuits,
only 3 of the 20 periods showed statistical difference. Moreover, of the three time
period differences, only two of those moved in the same direction as the Supreme
Court. This evidence alone suggests the four circuits were not looking to the
Supreme Court for guidance in terms of capital punishment decision making.
This conclusion is also bolstered by the political nature of the issue itself.
As noted earlier, the death penalty is seen in the United States as one of the most
divisive issues found within American society (see Kahan 1999, 448), and the
judicial branch is not immune (see Segal and Spaeth 1993, 73). For this reason,
judges on the circuit court may be more inclined to follow their own personal
political beliefs on the issue of the death penalty rather than a policy change from
the Supreme Court. This theory also supports the reason that Songer was able to
conclude that circuits generally do follow policy changes by the Supreme Court

151
with regard to labor and anti-trust cases. Although both labor and anti-trust cases
clearly contain political elements, it is not unreasonable to conclude that capital
punishment possesses greater dissonant opinions throughout the United States than
either labor or anti-trust issues. Therefore, the actual amount of Supreme Court
impact on circuit courts may very well be dependant on the political nature of the
policy issue.
A second conclusion that can be drawn is that the independence seen
within the circuits in terms of judicial decision making is not likely to change with
time. One of the few checks that are placed on circuit courts in terms of their
decision making is through Supreme Court review with the threat of reversal.
Such a threat is no longer viable as the Supreme Court grants a full hearing to
fewer than one hundred of the nine thousand cases on the docket each term or
less than one percent of the petitions for writ of certiorari that are filed with it each
year (OBrien (2005), 160, 231). In light of the lack of judicial control conducted
by the Supreme Court on lower courts, circuit courts do not have to be overly
concerned with any real ramifications from the Supreme Court.
Songer notes that even where decisional trends of the courts of appeals are
consistent with his hypothesis of Supreme Court impact, alternative explanations
are possible. He states that there are frequently partisan differences in the voting
behavior of appellate court judges. In his study of labor cases, 57% of the votes
cast by Democratic judges were liberal compared to 52% for Republican judges, a
difference which is statistically significant. In anti-trust cases, Democratic judges

152
cast liberal votes 42.7% of the time where their Republican counter-parts issued
liberal votes in 38.5% of the cases. This difference was found to be statistically
significant at the .01 level. The existence of partisan differences raises the
possibility that the changes in the decisional trends by the courts of appeals may be
due to changes in the partisan composition of the lower courts rather than the
Supreme Court influence. In order to analyze this alternative explanation, the
votes of individual judges were analyzed in each time period with a control for
party of the appointing president. Songer found that for both labor and anti-trust
decisions, the proportion of liberal Republican votes increased to a statistically
significant degree after the Warren Court policies became established and then
decreased to a statistically significant extent after the policy changes adopted by
the Burger Court.
The same process was followed in the instant case. Overall, the data
showed that in only 6 of 40 time frames or 15% of the time did the four circuits
shift in voting in a statistically significant manner and in the same direction as the
United States Supreme Court. This is a contrary finding to what Songer found in
his study in which the trends over time for both parties followed the direction of
changes enacted by the Supreme Court (Songer 1987, 837). Table 6.5 shows the
liberal voting by both Republican and Democratic judges on the Fourth Circuit
showed a conservative voting trend. No time period for Republican judges
eclipsed 14.8% of liberal decision making with the average liberal voting only
4.6% over the five time periods. Only two of the five time periods (1988-1993,

153
2001-2004) for Republican judges made statistically significant changes and only
one of those periods (2001-2004) did so in the same direction as the Supreme
Court. Although Democratic judges showed a greater propensity for liberal
voting, no time period raised above 25.9% and the average liberal voting was only
13.2%. None of the time periods for Democratic judges were found to be
statistically different from one another. Thus, these findings are consistent with
the earlier conclusion that circuits by-and-large do not look to the Supreme Court
for decision making on capital punishment cases.
Table 6.5

Fourth Circuit Death Penalty Decision Making Broken Down by Judicial Political
Affiliation:

Republican Period % Liberal Voting N (Z) Differential
1981-1987 14.8% 27 .037
1988-1993 2.7% 73 n/s
1994-1995 3.1% 32 n/s
1996-1997 2.4% 41 n/s
1998-2000 1.8% 107 n/s
2001-2004 9.8% 61 .025

Democratic 1981-1987 25.9% 27 ----
1988-1993 15.3% 39 n/s
1994-1995 10.5% 19 n/s
1996-1997 19.3% 31 n/s
1998-2000 9.0% 88 n/s
2001-2004 8.1% 61 n/s

Table 6.6 sets out the findings from the Fifth Circuit showed similar
conservative voting with the average liberal vote for Republican and Democratic
judges at 17.2%. When solely looking at Republican judges, no time period had
greater than 20% liberal voting. Three time periods (1988, 1993, 1994-1995,
2001-2004) showed statistically significant differences but only two of these

154
periods (1994-1995, 2001-2004) moved in the same direction as the Supreme
Court. Democratic judges on the Fifth Circuit tended to vote in a greater liberal
manner than their Republican counter-parts but liberal voting for any of the time
periods never went beyond 33.5%. Only one of the five time periods (2001-2004)
demonstrated a shift in a statistically significant manner that was in the same
direction as the Supreme Court. Thus, when examining the Republican and
Democratic judges for the 10 time periods, only four of those periods supported
Songers hypothesis.
Table 6.6

Fifth Circuit Death Penalty Decision Making Broken Down by Judicial Political
Affiliation:

Republican Period % Liberal Voting N (Z) Differential
1981-1987 20.0% 145 ----
1988-1993 13.7% 160 .096
1994-1995 6.1% 65 .083
1996-1997 8.3% 96 n/s
1998-2000 4.7% 126 n/s
2001-2004 15.1% 224 .003

Democratic 1981-1987 33.4% 215 ----
1988-1993 26.5% 94 n/s
1994-1995 15.3% 13 n/s
1996-1997 8.8% 34 n/s
1998-2000 8.0% 75 n/s
2001-2004 22.7% 136 .006

Evidence from the Sixth Circuit was even less convincing on the impact of
the Supreme Court on circuit courts. Contrary to the Fourth and Fifth Circuits, the
Sixth Circuit showed a great propensity for liberal voting on death penalty cases
with a combined liberal vote of Republican and Democratic judges at 32.3%.
Liberal voting for Republican judges went as high as 50% (1996-1997) and as low

155
as 7.4% (1998-2000). The time period from 1998 to 2000 showed a statistically
significant difference from the prior time period and the Sixth Circuit did move in
the same direction as the Supreme Court. That being said, none of the other nine
time periods showed a significant difference once again leading to the conclusion
that the Supreme Court had little, if any, impact on the this circuit.
Table 6.7

Sixth Circuit Death Penalty Decision Making Broken Down by Judicial Political
Affiliation:

Republican Period % Liberal Voting N (Z) Differential
1981-1987 0.0% 0 ----
1988-1993 36.3% 11 n/s
1994-1995 28.5% 7 n/s
1996-1997 50.0% 8 n/s
1998-2000 7.4% 20 .012
2001-2004 13.5% 74 n/s

Democratic 1981-1987 0.0% 1 ----
1988-1993 28.5% 7 n/s
1994-1995 50.0% 2 n/s
1996-1997 58.3% 12 n/s
1998-2000 35.5% 20 n/s
2001-2004 54.1% 72 n/s

Finally, the Ninth Circuit showed the greatest liberal voting over the 25
year period with the average voting of Republican and Democratic judges at
46.1%, nearly half of all cases that went before it. None of the five time periods
for the Republican judges showed statistically significant change with the highest
level of liberal voting taking place in 2001-2004 at 43.1% and the lowest taking
place in 1996-1997 at 25.8%. Democratic judges showed three statistically
significant differences (1994-1995, 1996-1997, 1998-2000), however, the decision
making in all three time frames went in a counter direction to the Supreme Court

156
trend. Thus, none of the statistical different time frames followed follow the
decision making trend of the Supreme Court.
Table 6.8

Ninth Circuit Death Penalty Decision Making Broken Down by Judicial Political
Affiliation:

Republican Period % Liberal Voting N (Z) Differential
1981-1987 29.1% 24 ----
1988-1993 33.8% 65 n/s
1994-1995 31.4% 35 n/s
1996-1997 25.8% 58 n/s
1998-2000 30.5% 59 n/s
2001-2004 43.1% 44 n/s

Democratic 1981-1987 38.0% 21 ----
1988-1993 52.5% 59 .075
1994-1995 71.4% 28 .005
1996-1997 36.5% 41 .011
1998-2000 61.1% 67 n/s
2001-2004 71.1% 97 n/s

Conclusions
The above data does not support the finding made by Songer that the
circuits do follow the policy trends of the Supreme Court when controlling for the
political party of the judges. Out of the 40 time periods (20: Republican and 20:
Democratic), only 10 showed statistically differences and only five of those
changes went in the same direction as the Supreme Court. When considering this
fact along with the initial finding that only two instances out of 20 or 10% of the
time did the circuits appear to follow the policy trend of the Supreme Court, it is
clear that the 4
th
, 5
th
, 6
th
and 9
th
circuit decision making on capital punishment is
not impacted trends from a higher court. Generally speaking, each circuit was
consistent with its decision making on this particular area of the law. Where

157
Fourth and Fifth Circuits showed a strong conservative tendency on death penalty
voting, the Sixth Circuit showed a much more moderate decision making trend
with the Ninth Circuit evidencing a clear liberal trend. This conclusion is also
supported by the recognition that circuit courts have become increasingly more
independent as the Supreme Court is not in a realistic position to police the
lower courts when the high court can only address but a small faction of the cases
that are presented to it each year. Although the Supreme Court does have the final
word in terms of the American judiciary, it is evident that its impact on lower
federal appellate courts with regard to capital punishment decision making is de
minimis, at best.

158
Chapter 7

Signal Theory Analysis

The American judiciary is governed by the basic principal of stare
decisis, which mandates that courts are to abide by, or adhere to, settled points of
law (Shapiro 1972, 125). As applied in hierarchical system of courts, the duty of
a subordinate court to follow the laws as announced by superior courts is
theoretically absolute (Moore, Lucas, and Currier 1993, 1-2). This obligation to
follow precedent begins with necessity, and a contrary necessity marks its outer
limit. Justice Cardozo recognized that no judicial system could do society's work
if it eyed each issue afresh in every case that raised it (see Cardozo, 1921, 149).
Indeed, the very concept of the rule of law underlying our own Constitution
requires such continuity over time that a respect for precedent is, by definition,
indispensable (see Powell 1991, 16.) At the other extreme, a different necessity
would make itself felt if a prior judicial ruling should come to be seen so clearly as
error that its enforcement was for that very reason doomed (Planned Parenthood
v. Casey, 505 U.S. 833, 854 (1992)). However, not every inferior judge accepts
such a passive role (Caminker 1994, 818) and it is from this conflict that so many
theories have been born as to the true driving force behind judicial decision
making at the circuit court level.
This chapter explores whether circuit courts respond to direct and/or
indirect signals from the Supreme Court with regard to death penalty decision
making. It also examines the extent that circuit courts react to signals from

159
Congress or gubernatorial directives on capital punishment issues. Thus, the
issue is whether circuit courts are responsive to signals from higher courts and/or
extra-judicial influences versus alternative influences such as personal policy
preferences, the law itself and the like.
Principal of Signal Theory
The hierarchical control of organizations is found throughout the political
realm. Congress and the president attempt to control agencies, upper levels of
bureaucracies attempt to control lower levels, and higher courts strive to control
lower courts. The nature of this control is based upon the concept of hierarchy
with superiors in rule-based hierarchies often employing some form of auditing for
control (Cameron, Segal, and Songer 2000, 101). As noted earlier, the theoretical
background that supports such a position is known as the principal-agent theory.
Under this doctrine, the agent is the body that acts on behalf of its principal where
the agent has a primary duty to act for the benefit of the principal (Songer, Segal
and Cameron 1994, 673-674). Thus, in the federal court system, a court must
follow the precedents established by the court(s) directly above it. District courts
must follow both Supreme Court decisions and those issued by whichever court of
appeals has revisory jurisdiction over its decision, and the courts of appeals must
heed Supreme Court decisions (Caminker 1994, 824).
Thus, in the context of the federal courts, the circuit courts of appeal act as
agents on behalf of their principal, the Supreme Court, such that Supreme Court
policymaking is carried out by the decision making of the circuit courts. Scholars

160
have explored the principal-agent theory and questioned whether certain courts
react to specific signals between the various courts to assist in guiding the lower
courts decision making on particular issues. Cameron, Segal and Songer
conducted one such study examining how the Supreme Court used signals and
indices from lower courts to select a relative handful of cases from a plethora of
potential cases that filed certiorari with the Supreme Court. The original sample of
more than 1,100 published court of appeals cases yielded only 18 cases reviewed
by the Supreme Court (Cameron, Segal and Songer 2000, 108-9).
Cameron, Segal and Songer ran a game theory model by coding a choice-
based sample for various characteristics.
53
The authors found that a conservative
Supreme Court declines to review conservative decisions from lower courts
regardless of the facts of the cases or the relative ideology of the judges. However,
a conservative higher court probabilistically reviews liberal decisions, with the
audit rate being tied to observable facts and the ideology of the lower court. The
evidence examined broadly supported the model that suggested the role of
certiorari was used for enforcing the doctrinal preferences of the Supreme Court
(Cameron, Segal and Songer 2000, 108-9, 113-4).

53
The dependent variable was whether certiorari was granted. Four independent variables were
coded, including: whether the lower court admitted the evidence; an index, the publicly observable
intrusiveness of the search; another index, the ideology of the lower court; and the ideology of the
Supreme Court (Cameron, Segal and Songer 2000, 109).

161
Data, Methods and Signals
54

The present study is an extension of the work conducted by Cameron,
Segal and Songer. While Cameron, Segal and Songer focused on the role of
judicial review by the Supreme Court for the enforcement of its preferences, this
study is concerned with whether the Supreme Court is able to change death penalty
decision making of the circuit court judges through specific signals sent from the
high court. Similarly, questions are raised whether signals from Congress in the
form of capital punishment legislation and a gubernatorial moratorium on
executions in a state has any impact on the circuit courts decisions in death penalty
cases.
A total of seven signals have been identified in the form of orders or
opinions from the Supreme Court, capital punishment legislation from Congress
and a state moratorium on executions issued by a former governor of Illinois, all of
which took place between 1992 and 2004. Each of these signals sets out a clear
preference with regard to the death penalty, four pro-death penalty and three anti-
death penalty. After the signals are set out below, the percentage of pro-death
penalty decisions for each circuit individually as well as all of the circuits
cumulatively was calculated six months prior to the signal and six months after,
one year before and one year after and two years before and two years after, for
comparison purposes. A difference of proportions (Z) test was used to determine
whether any statistical difference took place between the circuit court decision

54
The signals are presented in chronological order.

162
making before the signals versus after the signals. From this analysis, inferences
can be drawn as to the possible impact of signals from both the Supreme Court and
Congress with regard to capital punishment.
The first signal is a pro-death penalty order from that Supreme Court to the
Ninth Circuit in Vasquez v. Harris, 503 U.S. 1000 (1992). This order is perhaps
the strongest example of a Supreme Court signal to a specific circuit court. Harris
was convicted of the murder of two boys in Mira Mesa, California in 1978 and
sentenced to die (Cox 1993, 158-60). After 11 separate reviews of his conviction
and sentence in both state and federal courts, Harris execution was set for April
21, 1992, at 12:01 a.m. On April 17, Harris filed three new lawsuits, including a
ninth habeas petition with the California Supreme Court, a fourth habeas petition
filed in federal court and a federal class action lawsuit claiming that execution by
lethal gas constituted cruel and unusual punishment. On April 20, the Ninth
Circuit unanimously denied Harris federal petition for writ of habeas corpus. The
court also vacated a temporary restraining order that had been issued by United
States District Court Judge Marilyn Hall Patel on the cruel and unusual
punishment question (Cox 1993, 160).
That same day, at 6:30 p.m., a single Ninth Circuit judge issued an order
staying Harris execution for 10 days in order to examine Harris claim that new
evidence pointed to his brother as the person who shot the boys. At 10:20 p.m.,
ten Ninth Circuit judges reinstated District Court Judge Patels temporary
restraining order. Less than one hour later, a single judge of the Ninth Circuit

163
issued a separate stay on the cruel and unusual punishment issue. The State
appealed to the United States Supreme Court. At 11:20 p.m., the Supreme Court
lifted the first stay and at 3:00 a.m., on a vote of 7 to 2, the Ninth Circuit lifted the
other two stays. Shortly before 4:00 a.m., Judge Harry Pregerson of the Ninth
Circuit granted a fourth stay of execution in order to examine the cruel and
unusual punishment issue. Harris had already been strapped to the chair in the
death chamber at the time of this final stay (Cox 1993, 160-1). At 5:45 a.m., the
Supreme Court lifted the final stay and issued the following order to the Ninth
Circuit:
The application to vacate the stay of execution of sentence
of death presented to Justice O'Connor and by her referred
to the Court is granted, and it is ordered that the order
staying the execution entered by the United States Court
of Appeals for the Ninth Circuit on April 21, 1992 is
vacated. No further stays of Robert Alton Harris'
execution shall be entered by the federal courts except
upon order of this Court. Vasquez v. Harris, 503 U.S.
1000 (1992).

This order from the Supreme Court to the Ninth Circuit is classified as a clear, pro-
death penalty signal to the Ninth Circuit to stop with its anti-death penalty
strategy.
Two years after the Harris debacle, Congress issued a strong, pro-death
penalty signal with the passage of the Federal Death Penalty Act. In 1994,
Congress enacted the FDPA as part of an omnibus crime bill, the Violent Crime
Control and Law Enforcement Act of 1994. Prior to the passage of the FDPA,
federal statutes did not permit for the broad imposition of the death penalty for

164
federal crimes.
55
The purpose of the FDPA was to increase the number of federal
capital crimes and to provide an expanded federal system to impose the death
penalty. The FDPA expanded the death penalty to sixty different crimes, but this
did not mean that the FDPA created sixty crimes. One way to consider the effect
the FDPA had on existing substantive criminal law was that it "merely applie[d] its
procedural provisions to [those] crimes, making them death eligible." For
example, the FDPA transformed carjacking into a capital crime by adding the
death penalty as the maximum punishment but not by changing the elements of
carjacking. The statutory scheme of the FDPA expanded the scope of the federal
death penalty and provided the procedure for death penalty cases. The FDPA not
only allowed for the imposition of the death penalty in homicide cases but also for
a few non-homicide crimes such as espionage and reason (Culter 2000, 1209-10;
Little 1999, 385). In light of the broad nature of the FDPA and its pro-death
penalty implications, the passage of the FDPA is considered a pro-death penalty
signal from Congress to the federal courts.
The third signal is another federal, pro-death penalty piece of legislation
that was signed into law by President Bill Clinton. President Clinton signed the
Anti-Terrorism and Effective Death Penalty Act into law on April 24, 1996,
concluding decades of debate in Congress about if and how it should reform
habeas procedures. In the end, the final impetus needed to enact the habeas reform

55
In 1988, Congress passed law federal law that allowed for the imposition of the death penalty for
a limited class of crimes that were associated with the drug trade (Cutler 2000, 1200-1; Little 1999,
381-4).

165
legislation came from the Oklahoma City bombing. Republicans reacted to the
attack by pushing anti-terrorism legislation through Congress. Believing that
Timothy McVeigh would face the death penalty for the bombing, Congress
attached the habeas reform measures to the antiterrorism bill to ensure efficient
resolution of his likely challenges (McCollough 2005, 374-5; Liebman 2001, 427).
After so many years of redrafting and amending various habeas reform bills, the
final legislation ended up unwieldy, awkward, and confusing leaving courts to
grapple with the language and procedures of the AEDPA. The general response to
the legislation has been less than enthusiastic (McCollough 2005, 375).
Congress's primary goal in passing the habeas reform provisions of the
AEDPA was to "to reduce the abuse of habeas corpus that results from delayed
and repetitive filings." Frustrated with the long delays in implementing state
capital sentences created by successive federal habeas petitions, Congress set out
to expedite the federal review process and to create finality in judgments in order
to preserve efficacious state sentencing. To ensure that efficiency and finality
would not completely undermine the value of federal habeas corpus for capital
defendants, however, Congress retained 21 U.S.C. § 848(q)(4)(B), which entitles
every indigent capital defendant to appointed counsel during federal habeas
procedures (McCollough 2005, 375).
The AEDPA brought comprehensive changes to the treatment of federal
capital and non-capital claims by state prisoners without changing the fundamental
framework of preexisting habeas law. The AEDPA's primary changes to habeas

166
law involve the new statute of limitations, increased deference to state court
findings, strict limitations on successive habeas petitions, and special procedures
for capital habeas petitioners (McCollough 2005, 375). The end result of the
passage of the AEDPA is that many federal habeas litigates are not able to file
federal habeas corpus petitions due to missed deadlines, federal courts are
obligated to defer to state court findings, habeas litigants are unable to bring
multiple petitions, all of which leads to less opportunity for federal capital habeas
litigants to gain relief in federal court. Due to the significant changes that AEDPA
placed on capital litigation, the AEDPA was selected as a pro-death penalty signal
from Congress.
The fourth signal came from the Supreme Court in Calderon v. Thompson,
523 U.S. 538 (1998). Thompson was convicted and sentenced to death in a
California Superior Court for the murder and rape of a woman. The death
sentence was affirmed on direct appeal by the California Supreme Court. Before
exhausting his state habeas options, Thompson filed a federal habeas petition with
a principal claim of ineffective assistance of counsel. The United States District
Court for the Central District of California granted relief as to the rape conviction
and death sentence. The Court of Appeals for the Ninth Circuit reversed the
district court's granting of the writ and the United States Supreme Court denied
Thompson's petition for writ of certiorari. In a successive state habeas petition,
Thompson alleged new evidence invalidating the rape conviction, claiming

167
consensual sex with the victim. The California Supreme Court denied this
petition on July 16, 1997 (Mahoney 1998, 48).
Thereafter, Thompson moved the Ninth Circuit to recall its mandate
denying habeas relief. After advising the full court of the motion, and hearing no
requests for en banc consultation, the three judge panel denied the motion on
behalf of the Ninth Circuit on July 28, 1997. On July 30, 1997, the Ninth Circuit
voted to consider en banc whether to recall the mandate it earlier issued denying
habeas relief and scheduled oral arguments for August 1. Meanwhile, California
had conducted a clemency review, denied clemency, and set an execution date.
Two days before Thompsons execution, the Ninth Circuit recalled its mandate
and granted habeas relief to Thompson, citing en banc procedural difficulties in its
internal notification process (Mahoney 1998, 48-49).
The State of California filed a petition for writ of certiorari with the United
States Supreme Court requesting reversal of the Ninth Circuit ruling. Although it
was the state that pursued relief in the Supreme Court, it was Thompson who faced
an uphill battle. The Rehnquist Court placed great emphasis on the expansion of
non-constitutional doctrines such as mootness, ripeness, standing, procedural
default, non-retroactivity, independent state grounds and abuse of the writ. It also
emphasizes at every opportunity legal doctrine such as comity and finality. Under
the Rehnquist Court's jurisprudence, these rules regularly prove decisive in
limiting the ability of lower federal courts to redress constitutional violations and

168
shutting the doors of the courthouse to many litigants. Thompson was no
exception (Reinhardt 1999, 317).
On April 29, 1998, in a five-to-four ruling, the United States Supreme
Court held that a United States Court of Appeals had an inherent power to recall its
mandates, subject only to review for an abuse of discretion. The Supreme Court
held that the Ninth Circuit committed an abuse of discretion when it recalled its
mandate (Mahoney 1998, 48). A recall of the mandate could only take place to
avoid a miscarriage of justice. The standard for a miscarriage is only concerned
with actual innocence as compared with legal innocence. Thus, a petitioner under
this standard must show it is more likely than not that no reasonable juror would
have convicted him in light of the new evidence, a very high procedural hurdle to
overcome. The Supreme Court concluded Thompson presented little evidence to
undermine the trial evidence of his guilt in the rape and murder of the victim (see
Reinhardt 1999, 343-4). In light of the extremely high burden the Supreme Court
placed on Thompson to gain relief from his death sentence and its emphasis on
procedural barriers that Thompson could not meet, the ruling in Calderon v.
Thompson is viewed as a pro-death penalty signal from the Supreme Court to the
circuit courts.
Beginning in the new millennium, a shift in support for the death penalty
appears to have taken place throughout the country. Many began to see "crisis of
confidence" in the death penalty. The crisis was initially precipitated by the
shocking revelations that at least 13 persons on Illinois' Death Row, and many

169
more nationwide, were innocent of the crimes for which they were sentenced to
die. This fact was exacerbated by a major academic study at Columbia University,
revealing that more than two-thirds of all death sentences imposed since 1972
were eventually reversed, either on appeal or in post-conviction hearings. The
conclusions of the Columbia study resonated with the Illinois experience, which
contributed to a growing national concern that the system of capital punishment in
America was not producing, and may even be incapable of producing, acceptably
reliable substantive results (Hoffmann 2005, 561-2).
Following this apparent change in death penalty support, Illinois Governor
George Ryan's announced a moratorium on executions in Illinois, effective
January 31, 2000 (Illinois Government News Network, 2000). This action was
one of the most dramatically significant events in a generation in helping shape the
political climate surrounding the death penalty (Kirchmeier 2002, 44-5). The
governor's action came on the heels of intense adverse publicity about the state's
death penalty (Armstrong & Possley 1999), including the stark revelation of the 13
death-sentenced individuals that were found to be innocent of their alleged crimes,
compared with 12 who had been executed. Shortly after imposing the moratorium,
Governor Ryan signed Executive Order Number 4 (2000), creating a 14-member
Commission on Capital Punishment to "study and review" Illinois' death penalty
and make appropriate corresponding recommendations (Ryan, 2000; Hoffmann
2005, 579).
After laboring for 2 years, the Illinois Governor's Commission on Capital

170
Punishment produced 85 recommendations regarding the administration of
capital punishment in that state. Among its recommendations were changes in
police practices to help safeguard the reliability of evidence produced during
pretrial investigations; the introduction of guidelines and centralized review to
help regulate prosecutors' capital-charging decisions; significant restrictions in the
range of death penalty eligible murders; periodic training for judges and attorneys
in capital cases; and the commitment of substantial additional resources. The
Commission also recommended that better data on capital cases be collected (e.g.,
details about the trial, background of the defendant, and basis for the sentence
imposed) (Commission on Capital Punishment, 2002; Hoffmann 2005, 579).
In January 2003, just 3 days before his term ended, Governor Ryan gave a
speech at Northwestern University College of Law during which he announced his
decision to grant a "blanket commutation" for all death-sentenced prisoners in
Illinois. During his remarks, the Governor observed:
Our capital system is haunted by the demon of error, error in
determining guilt, and error in determining who among the guilty
deserves to die. Because of all of these reasons today I am
commuting the sentences of all death row inmates (Death Penalty
Information Center, 2004e).

Governor Ryan thus granted clemency to all remaining 156 prisoners on
death row, as well as to another 11 persons convicted of capital crimes,
who were waiting to be sentence or resentencing (Hoffmann 2005, 579).
Following the anti-death penalty signal from Governor Ryan, a similar
signal took place from the Supreme Court. On June 20, 2002, the Supreme Court

171
reversed its previous pro-death penalty position on the execution of the mentally
retarded and concluded that the execution of such persons was now
unconstitutional in Atkins v. Virginia, 536 U.S. 304 (2002). In re-evaluating its
prior position, the Atkins Court held that capital punishment was an
unconstitutionally cruel and unusual punishment for the mentally retarded for two
reasons. First, the Court concluded that the impairments associated with mental
retardation both reduced the culpability of the mentally retarded, making death a
disproportionate punishment for them, and created a "special risk of wrongful
execution." Second, the Court found that a national consensus had emerged that
death is an excessive punishment for the mentally retarded (Fagan and West 2005,
427-28).
The Atkins decision overturned Penry v. Lynaugh with the majority citing
the "evolving standards of decency" as dictating this dramatic departure from
precedent, observing "much has changed since [the Penry decision]." The
majority also expressed its lack of confidence in the machinations of procedure to
protect the rights of the mentally retarded who face capital trials and the death
penalty. Analogizing the needs of the mentally retarded to the needs of the insane,
the Atkins Court borrowed from its Ford v. Wainwright decision and ruled: "[The
death penalty] is excessive and . . . the Constitution places a substantive restriction
on the State's power to take the life' of a mentally retarded offender." The ruling in
Atkins is a clear shift of Supreme Court policy from a pro-death penalty position

172
to an anti-death penalty position and thus this decision is viewed as an anti-death
penalty signal from the Supreme Court.
The final signal came from the Supreme Court in Miller-El v. Cockrell,
537 U.S. 322 (2003). Thomas Joe Miller-El, an African-American, was tried and
convicted in Texas for murder and sentenced to death. During the voir dire stage
of the trial, two assistant district attorneys used peremptory challenges to exclude
10 of 11 African-American perspective jurors. In contrast, the prosecution only
struck 4 of 31 non-African-American perspective jurors. On review in the state
and federal judiciaries, Miller-El claimed that the prosecution had engaged in
racial discrimination during the jury selection process. As evidence of the
improper challenges, Miller-El offered both the statistical disparity in peremptory
challenges and the differing questions that were posed to African-American
perspective jurors versus non-African American (Williams 2005, 645-6).
The Texas Court of Criminal Appeals denied Miller-Els claim. Miller-El
then filed a petition for writ of habeas corpus with the federal district court which
likewise denied the claim. Miller-El filed a request for a certificate of
appealability so that the Fifth Circuit could review the issue. The Fifth Circuit
ruled that Miller-Els claim of racial discrimination in jury selection was not even
debatable among jurists of reason and thus denied the request for the certificate of
appealability. Miller-El appealed the Fifth Circuit denial to the Supreme Court.
In an opinion dated February 25, 2003, the Supreme Court reversed the Fifth
Circuit, holding that the merits of Miller-Els claim should have been considered

173
as reasonable jurists could have debated whether the prosecutors, who struck 91%
of eligible African-American venire members but only 13% of non-black venire
members, had engaged in racially discriminatory jury selection. In a clear
statement from the high court, the opinion stated that happenstance is unlikely to
produce this disparity (Williams 2005, 647).
The strongly worded opinion issued by the Supreme Court stands as a
signal from the Supreme Court that it supports a more moderate death penalty
stance which the circuits should take heed. This conclusion is bolstered by final
result in the Miller-El case. Following the issuance of the certificate of
appealability by the Fifth Circuit, the Fifth Circuit denied Miller-Els claim. On
review from the denial, the Supreme Court granted Miller-Els habeas petition
once again disagreeing with the Fifth Circuit position (Williams 2005, n.125).
Findings and Discussion
Six Month Lag

The evidence drawn from an examination of circuit court decisions six
months before the signals and six months after the signals showed that these
particular signals had little, if any, impact on circuit court decision making on
capital punishment cases. As can be seen from Tables 7.0 through 7.6, no
statistical difference greater than or equal to a confidence level of 90% was found
when examining the Supreme Court signals in Harris v. Vasquez, Calderon v.
Thompson or Miller-El v. Cockrell. Likewise, neither of the death penalty statutes
enacted by Congress nor the moratorium of executions issued by Governor Ryan
had any statistical affect on the four circuit courts studied. In fact, the only signal

174
that resulted in a statistical difference at or above 90% involved the Supreme
Court decision of Atkins v. Virginia. That being said, the result in which the Ninth
Circuit decision making on death penalty cases went from 9% pro-death penalty
before the signal to 66% pro-death penalty after the signal was counter-intuitive as
the signal was an anti-death penalty message, yet the Ninth Circuit increased its
pro-death penalty decision making six fold after the signal (see Table 7.5). Thus,
when examining signals six months on either side of a signal at a level of greater
than or equal to 90%, these particular signals appeared to have no impact on circuit
court decision making.
If the desired confidence level is reduced from 90% to 80%, the signals
appear to have a greater affect. Four instances of statistical difference were found,
each of which came from Supreme Court decisions. However, only two of the
four instances followed the direction of the signals. The cumulative percentage of
pro-death penalty decisions from all of the circuits before and after Calderon v.
Thompson resulted in a statistical difference of 86.8%. Likewise, the percentage
of anti-death penalty decision making by the Fifth Circuit before and after Atkins
v. Virginia registered a statistical different at 83.1%. Equally important, the
circuit court decision making following each of these signals went in the proper
direction. Calderon v. Thompson was a pro-death penalty decision and the circuits
appeared to have responded to this signal with a cumulative increase in pro-death
penalty decisions. Atkins v. Virginia was an anti-death penalty decision and the

175
Fifth Circuit responded accordingly by reducing its pro-death penalty decision
making to a statistically significant degree.
Although the above instances appear to support the position that circuits do
respond to higher court signals, the fact remains that the other two instances of
statistically significance differences involved decision making that went counter to
the direction of the signal, thus not supporting the hypothesis that circuit courts
respond to signals from the Supreme Court. Moreover, the overall results from
this analysis show that no instances of statistical difference that went in the same
direction as the signal at or equal to 90% and only 2 instances out of a possible 35
demonstrated a circuit court response to Supreme Court at the 80% level. Thus,
when examining circuit court decisions six months prior to the signals and six
months after, little evidence supports the conclusion that circuits are listening to
the Supreme Court, Congress or gubernatorial signals.
Table 7.0

Signal Theory Table, Six Month Lag Harris v. Vasquez

Date of Signal: 4-21-92

Pre-Signal
10-21-1991 to
4-21-1992
Post-Signal
4-22-1992 to
10-21-1992
Statistical
Difference
Found
4
th
Circuit 2/3 = 66% 5/5 = 100% No
5
th
Circuit 12/13 = 92.5% 9/10 = 90% No
6
th
Circuit 0/0 = 0% 0/1 = 0% No
9
th
Circuit 6/10 = 60% 2/6 = 33.3% No
Cumulative 20/26 = 76.9% 16/22 = 72.7% No

176
Table 7.1

Signal Theory Table, Six Month Lag Federal Death Penalty Act

Date of Signal: 9-13-94 Pre-Signal
3-13-94 to
9-13-94
Post-Signal
9-14-1994 to
3-13-1995
Statistical
Difference
Found
4
th
Circuit 2/2 = 100% 2/2 = 100% No
5
th
Circuit 9/9 = 100% 3/3 = 100% No
6
th
Circuit 0/0 = 0% 1/1 = 100% n/a
9
th
Circuit 3/6 = 50% 2/3 = 66.6% No
Cumulative 14/17 = 82.3% 8/9 = 88% No

Table 7.2

Signal Theory Table, Six Month Lag Anti-Terrorism and Effective Death
Penalty Act

Date of Signal: 4-24-96 Pre-Signal
10-24-1995 to
4-24-1996
Post-Signal
4-25-1996 to
10-25-1996
Statistical
Difference
Found
4
th
Circuit 5/5 = 100% 7/7 = 100% No
5
th
Circuit 7/7 = 100% 8/9 = 88% No
6
th
Circuit 0/1 = 0% 1/1 = 100% No
9
th
Circuit ¾ = 75% 5/8 = 62.5% No
Cumulative 15/17 = 88.2% 21/25 = 84% No

Table 7.3

Signal Theory Table, Six Month Lag Calderon v. Thompson

Date of Signal: 4-29-98 Pre-Signal
10-29-1997 to
4-29-1998
Post-Signal
4-30-1998 to
10-30-1998
Statistical
Difference
Found
4
th
Circuit 15/16 = 93% 14/14 = 100% No
5
th
Circuit 12/13 = 92.3% 12/12 = 100% No
6
th
Circuit 0/2 = 0% 1/1 = 100% No
9
th
Circuit 7/13 = 53.8% 6/10 = 60% No
Cumulative 34/44 = 77% 33/37 = 89.1% No (.132)

177
Table 7.4

Signal Theory Table, Six Month Lag Illinois Governor Ryan Moratorium

Date of Signal: 1-31-00 Pre-Signal
7-31-1999 to
1-31-2000
Post-Signal
2-1-2000 to
8-1-2000
Statistical
Difference
Found
4
th
Circuit 5/5 = 100% 11/11 = 100% No
5
th
Circuit 7/8 = 87.5% 9/9 = 100% No
6
th
Circuit 1/1 = 100% 4/6 = 66% No
9
th
Circuit 2/4 = 50% 3/6 = 50% No
Cumulative 15/18 = 83% 27/32 = 84% No

Table 7.5

Signal Theory Table, Six Month Lag Atkins v. Virginia

Date of Signal: 6-20-02

Pre-Signal
12-20-2001 to
6-20-2002
Post-Signal
6-21-2002 to
12-21-2002
Statistical
Difference
Found
4
th
Circuit 6/8 = 75% 1/1 = 100% No
5
th
Circuit 10/10 = 100% 13/17 = 76.4% No (.135)
6
th
Circuit 4/6 = 66% 6/9 = 66.6% No
9
th
Circuit 1/11 = 9% 2/3 = 66.6% Yes (.087)
Cumulative 21/35 = 60% 22/30 = 73.3% No (.192)

Table 7.6

Signal Theory Table, Six Month Lag Miller v. Cockrell

Date of Signal: 2-25-03

Pre-Signal
8-25-2002 to
2-25-2003
Post-Signal
2-26-2003 to
8-26-2003
Statistical
Difference
Found
4
th
Circuit 4/5 = 80% 11/11 = 100% No
5
th
Circuit 8/9 = 88% 21/23 = 91% No
6
th
Circuit 7/12 = 58.3% 2/4 = 50% No
9
th
Circuit 2/4 = 50% 3/7 = 42.8% No
Cumulative 21/30 = 70% 37/45 = 82.2% No (.169)

One-Year Lag
The evidence altered only slightly when examining circuit court decisions
one-year prior to the signals and one-year after as set out in Tables 7.7 through

178
7.13. Once again, only one instance was found in which a particular circuit
responded to a Supreme Court signal resulting in a statistical difference of pro-
death penalty decision making at or above the 90% range. Table 7.13 indicates
that a statistical difference was found in pro-death penalty decision making by the
Fourth Circuit at 90.2%. However, the signal in this instance involved the
Supreme Court decision in Miller-El v. Cockrell, which is classified as an anti-
death penalty decision. Following Miller-El , the Fourth Circuit increased its pro-
death penalty decision making which is counter to what should have occurred had
the Fourth Circuit been responding to the Supreme Court signal. No evidence
therefore was found at or above the 90% level to support the conclusion circuits
respond to certain signals.
As was seen with the six month interval, a larger number of occurrences
were found at or above the 80% confidence level. Both the Fifth Circuit and the
cumulative pro-death penalty decisions following the passage of the Federal Death
Penalty Act resulted in statistical differences of 80.3% and 81.4%, respectively.
However, the direction of circuit court decision making resulted in a decrease of
support for the death penalty when the signal was pro-death penalty. Similar
findings occurred when the Ninth Circuit decision making shifted at a statistical
difference following Atkins v. Virginia, an anti-death penalty decision, at 83.3%.
However, the Ninth Circuit increased its pro-death penalty decision making
following the Supreme Courts anti-death penalty decision which was counter to
the signal. Moreover, the cumulative pro-death penalty decision making following

179
Miller-El v. Cockrell was statistically different at 84.8% but as with the Ninth
Circuit, the percentage of pro-death penalty decision making increased which was
counter to the signal.
Of the six instances where a statistical difference was found at or above
80% after a signal was issued, only two of those occurrences went in the same
direction as the signal. The first was seen with the cumulative pro-death penalty
voting after the Supreme Court decision in Calderon v. Thompson at 84.2% and
the cumulative circuit voting after Governor Ryan of Illinois issued a moratorium
on all state executions at 83.9%. After both of the signals, the cumulative circuit
court decision making followed the direction of the two signals, supporting the
conclusion that circuit courts not only follow certain signals from the Supreme
Court but also signals from extra-judicial sources such as a governors moratorium
on state executions. However, this conclusion must be tempered by the fact that,
as was seen with the analysis involving the six month lag, only two instances out
of 35 were found in which circuit court decision making shifted to a statistically
significant degree and in the same direction of the signal. Although one cannot
completely disregard the theory of signal theory, the fact remains that less than
convincing evidence has been found to support this theory.

180
Table 7.7

Signal Theory Table, One Year Lag Harris v. Vasquez

Table 7.8

Signal Theory Table, One Year Lag Federal Death Penalty Act

Date of Signal: 9-13-94 Pre-Signal
9-13-93 to
9-13-94
Post-Signal
9-14-1994 to
9-14-1995
Statistical
Difference
Found
4
th
Circuit 6/6 = 100% 8/8 = 100% No
5
th
Circuit 14/14 = 100% 10/12 = 83.3% No (.197)
6
th
Circuit 2/2 = 100% ½ = 50% No
9
th
Circuit 6/13 = 46.1% 5/12 = 41.6% No
Cumulative 28/35 = 71.4% 23/34 = 70.5% No (.186)

Table 7.9

Signal Theory Table, One Year Lag Anti-Terrorism and Effective Death Penalty
Act

Date of Signal: 4-24-96 Pre-Signal
4-24-1995 to
4-24-1996
Post-Signal
4-25-1996 to
4-25-1997
Statistical
Difference
Found
4
th
Circuit 12/13 = 92.3% 12/13 = 92.3% No
5
th
Circuit 14/16 = 87.5% 19/20 = 95% No
6
th
Circuit 0/1 = 0% 2/3 = 66% No
9
th
Circuit 5/12 = 41.6% 12/19 = 63.1% No
Cumulative 31/42 = 73.8% 45/55 = 81.8% No

Date of Signal: 4-21-92

Pre-Signal
4-21-1991 to
4-21-1992
Post-Signal
4-22-1992 to
4-22-1993
Statistical
Difference
Found
4
th
Circuit 7/8 = 87.5% 8/8 = 100% No
5
th
Circuit 16/17 = 94.1% 17/20 = 85% No
6
th
Circuit 0/0 = 0% 0/1 = 0% n/a
9
th
Circuit 9/14 = 64.2% 5/10 = 50% No
Cumulative 32/39 = 82% 30/39 =76.9% No

181
Table 7.10

Signal Theory Table, One Year Lag Calderon v. Thompson

Date of Signal: 4-29-98 Pre-Signal
4-29-1997 to
4-29-1998
Post-Signal
4-30-1998 to
4-30-1999
Statistical
Difference
Found
4
th
Circuit 21/22 = 95.4% 26/27 = 96.2% No
5
th
Circuit 24/26 = 92.3% 22/22 = 100% No
6
th
Circuit 2/5 = 40% ¾ = 75% No
9
th
Circuit 13/25 = 52% 10/19 = 52.6% No
Cumulative 60/78 = 76.9% 61/72 = 84.7% No (.158)

Table 7.11

Signal Theory Table, One Year Lag Illinois Governor Ryan Moratorium

Date of Signal: 1-31-00 Pre-Signal
1-31-1999 to
1-31-2000
Post-Signal
2-1-2000 to
2-1-2001
Statistical
Difference
Found
4
th
Circuit 14/16 = 87.5% 16/16 = 100% No
5
th
Circuit 19/20 = 95% 20/22 = 90.9% No
6
th
Circuit ¾ = 75% 6/11 = 54.5% No
9
th
Circuit 7/11 = 63.6% 6/15 = 40% No
Cumulative 43/51 = 84.3% 48/64 = 75% No (.161)

Table 7.12

Signal Theory Table, One Year Lag Atkins v. Virginia

Date of Signal: 6-20-02 Pre-Signal
6-20-2001 to
6-20-2002
Post-Signal
6-21-2002 to
6-21-2003
Statistical
Difference
Found
4
th
Circuit 8/10 = 80% 7/9 = 77% No
5
th
Circuit 18/19 = 94.7% 20/24 = 83.3% No
6
th
Circuit 12/17 = 70.5% 9/15 = 60% No
9
th
Circuit 5/20 = 25% 3/5 = 60% No (.167)
Cumulative 43/66 = 43% 39/53 = 73.5% No

182
Table 7.13

Signal Theory Table, One Year Lag Miller v. Cockrell

Date of Signal: 2-25-03 Pre-Signal
2-25-2002 to
2-25-2003
Post-Signal
2-26-2003 to
2-26-2004
Statistical
Difference
Found
4
th
Circuit 10/13 = 76.9% 14/14 = 100% Yes (.098)
5
th
Circuit 21/25 = 84% 35/41 = 85.3% No
6
th
Circuit 10/17 = 58.5% 6/10 = 60% No
9
th
Circuit 2/6 = 33.3% 7/13 = 53.8% No
Cumulative 43/61 = 70.4% 62/78 = 79.4% No (.152)

Two-Year Lag
Tables 7.14 through 7.20 set out the data from the final component of this
analysis that looked at circuit court decisions two years before signals and two
years after. Four instances were found in which a statistical difference occurred at
or above 90% level of confidence, including the cumulative circuit voting
following Governor Ryans moratorium (98.8%), the Ninth Circuit following
Atkins v. Virginia (97.7%), the Ninth Circuit after Miller-E l v. Cockrell (92.5%)
and the cumulative circuit voting following Miller-El case (100%) (see Tables
7.18, 7.19, 7.20). Of these four instances, only the cumulative circuit voting after
Illinois Governor Ryan issued a moratorium on executions resulted in circuit court
decisions that reduced support for the death penalty. As seen with the prior two
analysis, only one instance out of 35 took place where the circuit courts shifted
their decision making to a statistically significant degree at or above 90% and in
the same direction as the signal. Thus, at the 90% confidence level, little evidence
supports the conclusion that circuit courts respond to signals from the Supreme
Court or otherwise.

183
When considering the statistical differences at the 80% level or above, a
total of nine instances fell within this category. Three of these occurrences
resulted in shifts after the signals that went counter to the expected result,
including the Fifth Circuit after Harris v. Vasquez (81.8%); the cumulative circuit
voting after the enactment of the Federal Death Penalty Act (82.7); and cumulative
circuit voting after Atkins v. Virginia (89.6%) (see Tables 7.14, 7.15, 7.19). The
other six remaining instances possessed a confidence level at least 80% and the
shift was in the expected direction. These instances included: the cumulative
circuit voting after Calderon v. Thompson (89.8%); the Fifth and Ninth Circuits
after Governor Ryans moratorium (82.2% and 80.6%, respectively); the Fifth
Circuit after Atkins v. Virginia (81.7%); and the Fifth and Sixth Circuits after
Miller-El v. Cockrell (83.5% and 80.5%, respectively) (see Tables 7.17, 7.18, 7.19,
7.20).
This evidence is clearly the strongest support for the signal theory when
looking at the results involving the six month, one-year and two-year lags. That
being said, 6 instances of 35 are a mere 17.1% and not an overwhelming
confirmation that circuit courts respond to specific signals. This is not to say that
signal theory does play any role in circuit court decision making on death penalty
cases, clearly it appears to play at least some role, but the extent of that role
remains in question. One reasonable suggestion would be that circuits respond to
specific signals from the Supreme Court, in other words, when the Supreme Court
issues a signal from a case from a particular circuit. In the instant study, three

184
Supreme Court signals were issued in cases that stemmed from the Ninth Circuit
(Harris v. Vasquez, Calderon v. Thompson) and the Fifth Circuit (Miller-El v.
Cockrell). The Ninth Circuit did not register a statistical difference in its decision
making in either the Harris case or the Thompson case. The Fifth Circuit, on the
other hand, did show a statistical difference following the ruling in the Miller-El
case. Moreover, the Fifth Circuit did so following the moratorium imposed by
Governor Ryan and in the Atkins case. Does this mean that the Fifth Circuit is
more reactive to signals from the Supreme Court or extra-judicial influences?
Such a conclusion would be pure supposition but the evidence noted above does
give some credence that signals from the courts and elsewhere does play a role
with certain circuit courts on capital punishment issues.
Table 7.14

Signal Theory Table, Two Year Lag Harris v. Vasquez

Date of Signal: 4-21-92 Pre-Signal
4-21-90 to
4-21-92
Post-Signal
4-22-1992 to
4-22-1994
Statistical
Difference
Found
4
th
Circuit 13/14 = 92.8% 15/16 = 93.7% No
5
th
Circuit 20/22 = 90.9% 20/26 = 83.3% No (.182)
6
th
Circuit 0/0 = 0% 2/3 = 66.6% N/A
9
th
Circuit 12/21 = 57.1% 15/28 = 53.5% No
Cumulative 45/57 = 78.9% 62/83 = 74.6% No

185
Table 7.15

Signal Theory Table, Two Year Lag Federal Death Penalty Act

Date of Signal: 9-13-94 Pre-Signal
9-13-92 to
9-13-94
Post-Signal
9-14-1994 to
9-14-1996
Statistical
Difference
Found
4
th
Circuit 13/14 = 92.8% 20/21 = 95.2% No
5
th
Circuit 27/32 = 84.3% 24/27 = 88.8% No
6
th
Circuit 2/2 = 100% 2/4 = 50% No
9
th
Circuit 14/24 = 58.3% 10/20 = 50% No
Cumulative 56/72 = 77.7% 56/72 = 77.7% No (.173)

Table 7.16

Signal Theory Table, Two Year Lag Anti-Terrorism and Effective Death Penalty
Act

Date of Signal: 4-24-96 Pre-Signal
4-24-1994 to
4-24-1996
Post-Signal
4-25-1996 to
4-25-1998
Statistical
Difference
Found
4
th
Circuit 16/17 = 94.1% 33/35 = 94.2% No
5
th
Circuit 23/25 = 92% 43/46 = 94.4% No
6
th
Circuit 1/3 = 33.3% 4/8 = 50% No
9
th
Circuit 9/19 = 47.3% 25/44 = 56.8% No
Cumulative 49/64 = 76.5% 105/133 = 78.9% No

Table 7.17

Signal Theory Table, Two Year Lag Calderon v. Thompson

Date of Signal: 4-29-98 Pre-Signal
4-29-1996 to
4-29-1998
Post-Signal
4-30-1998 to
4-30-2000
Statistical
Difference
Found
4
th
Circuit 34/36 = 94.4% 40/42 = 95.2% No
5
th
Circuit 43/46 = 93.4% 41/42 = 97.6% No
6
th
Circuit 4/8 = 50% 8/10 = 80% No
9
th
Circuit 25/44 = 56.8% 15/27 = 55.5% No
Cumulative 106/134 = 79.1% 104/121 = 85.9% No (.102)

186
Table 7.18

Signal Theory Table, Two Year Lag Illinois Governor Ryan Moratorium

Date of Signal: 1-31-00 Pre-Signal
1-31-1998 to
1-31-2000
Post-Signal
2-1-2000 to
2-1-2002
Statistical
Difference
Found
4
th
Circuit 43/45 = 95.5% 21/21 = 100% No
5
th
Circuit 40/41 = 97.5% 37/41 = 90.2% No (.178)
6
th
Circuit 5/6 = 83.3% 17/26 = 65.3% No
9
th
Circuit 14/28 = 50% 11/31 = 35.4% No (.194)
Cumulative 102/120 = 85% 86/119 = 72.2% Yes (.012)

Table 7.19

Signal Theory Table, Two Year Lag Atkins v. Virginia

Date of Signal: 6-20-02

Pre-Signal
6-20-2000 to
6-20-2002
Post-Signal
6-21-2002 to
6-21-2004
Statistical
Difference
Found
4
th
Circuit 20/22 = 90.9% 20/22 = 90.0% No
5
th
Circuit 38/42 = 90.4% 61/74 = 82.4% No (.183)
6
th
Circuit 17/27 = 62.9% 16/28 = 57.1% No
9
th
Circuit 8/33 = 24.2% 12/22 = 54.5% Yes (.023)
Cumulative 83/124 = 66.9% 109/146 = 74.6% No (.104)

Table 7.20

Signal Theory Table, Two Year Lag Miller v. Cockrell

Date of Signal: 2-25-03

Pre-Signal
2-25-2001 to
2-25-2003
Post-Signal
2-26-2003 to
12-31-2004
Statistical
Difference
Found
4
th
Circuit 14/17 = 82.3% 20/21 = 95.2% No
5
th
Circuit 36/42 = 85.7% 58/76 = 76.3% No (.165)
6
th
Circuit 23/33 = 59.2% 12/22 = 54.5% No (.195)
9
th
Circuit 7/27 = 25.9% 11/22 = 50% Yes (.075)
Cumulative 80/199 = 67.2% 111/141 = 78.7% Yes (.000)

Conclusions
In the end, it appears that signal theory had a minimal impact on circuit
court decision making on death penalty cases based on the signals examined in this

187
study on four circuits. Of the 10 instances where a statistical difference was
found at or above 80% with the circuit court response matching the expected affect
from the signal when considering all three lag periods, the Supreme Court signal
from Calderon v. Thompson and Governor Ryans Moratorium showed three
instances, the Supreme Court signal in Atkins showed two instances and the
Supreme Court signal in Miller-El v. Cockrell showed two. Neither of the two
legislative signals demonstrated any relationship to circuit court decision making
nor the Supreme Courts rebuke in Harris v. Vasquez. Ironically, these three
signals can easily be considered the strongest signals from the Supreme Court and
Congress with regard to their pro-death postures. Thus, it appears that the issue
comes down the circuit courts willingness to consider various influence in it
decision making on a highly political issue. Where some circuits, such as the Fifth
Circuit, appear amenable to direction from sources such as the Supreme Court,
others such as the Ninth Circuit appear to march to the beat of their own drum
no matter the suggestion from their principals or extra-judicial influences.

188
Chapter 8
Conclusion
The rule of law has always played an essential role in the American
judiciary. Although there is uniform agreement that the rule of law must exist in
order for a legitimate democracy to thrive, political scientists throughout time have
questioned the role that politics plays in the judicial system, and in particular, the
federal system. At a minimum, the infusion of politics into the decision making
processes of judges within the American court system would seem to erode the
very concept of the rule of law leaving the possibility that American court system
may be viewed as nothing more than a mere shell of what it stands for. The debate
over politics within the judicial system is clearly heightened when the issue before
the courts involves capital punishment. Few issues, if any, take precedent over the
right of the state to take the life of a person. The above analysis set out to examine
the extent to which legal and extra-legal factors play a role in circuit court decision
making on issue of the death penalty. The study addressed this task by using
multiple methods as well as an alternative coding scheme in order to confirm that
the dichotomous coding mechanism traditionally used by scholars accurately
classifies cases as either pro-death penalty versus anti-death penalty.
The first hypothesis asserted that a strong correlation exists between circuit
court panels ideology based upon the political party of the president that
nominated the judge and their support for capital punishment. Songer, Sheehan
and Haire (2000) as well as other scholars have concluded that a circuit court

189
panels ideology has a direct impact on how judges make decisions. Applying
this generalized statement to capital punishment in particular, one would conclude
that Republican appointments are stronger supporters of capital punishment while
Democrats are less supportive. The data from this study bears out this finding.
The logistic regression revealed that the panel ideology independent
variable was found to be statistically significant with the correct sign. Likewise,
similar findings were made when the continuum dependent variable was
substituted for the dichotomous outcome dependent variable. This evidence alone
allows for the conclusion that the ideological background of circuit court judges
definitely affects their decision making processes. Moreover, this evidence was
also confirmed through the use of the alternative methodology offered by Sunstein,
Schkade and Ellman (2003).
Sunstein et al. initially examined the liberal voting of Republican and
Democratic-appointed circuit judges on a variety of issues, including capital
punishment. The authors discovered that a 22% differential existed between
Republican-appointed judges and Democratic-appointed judges on the issue of the
death penalty. Applying their method to the instant sample, a 16% differential was
found with Republican-appointed judges voting liberal in capital punishment cases
16.3% while Democratic-appointed judges did so in 32.3%. Such a finding was
found to be substantial evidence of ideological voting. This evidence clearly
supports the results from the regression analysis that circuit judges are clearly
relying on their ideology for their decision making in, at the very minimum, death

190
penalty cases. Moreover, this conclusion is elevated in importance when one
considers the magnitude of effect that this one variable has on judicial decision
making. It was found that the inclusion of this variable into the decision making
equation while holding all other variables constant increased the likelihood of a
pro-death penalty decision by 37%. This result can lead to but one conclusion that
circuit court judges are definitely affected by their political ideology with regard to
decisions on capital punishment cases and that such a fact dramatically impacts the
outcome of cases.
The second hypothesis stated that a correlation exists between certain case
characteristics found in death penalty cases which affected the outcome of those
decisions on appeal. Songer and Haire (1992) and Segal, Songer and Cameron
(1995) concluded that some case characteristics can affect circuit court decision
making. Various case characteristics were coded as independent variables in the
regression analysis, including death penalty cases that involved robbery, rape,
kidnapping, adult female, elderly, minor, police officer and multiple murders.
These characteristics were tested against both the dichotomous outcome and
ordered dependent variables as well as in the multinomial analysis. Neither the
dichotomous outcome nor ordered dependent variables revealed any case
characteristics as statistically significant at the .05 level.
Lastly, the multinomial analysis confirmed the suggestion from the logistic
and ordered regressions that case characteristics play a minimal role in decision
making, at best. Although little can be taken away from the multinomial analysis

191
other than the fact that case characteristics appear to play some part in circuit
court decision making in capital punishment litigation, this fact in-and-of-itself
demonstrates that no one variable can be looked upon as an answer to the enduring
question of what drives judicial decision making. The answer is clear that a
multitude of factors impact circuit judges, some more such as judicial ideology and
panel composition and some less such as case characteristics.
The third hypothesis maintained that circuit courts will follow the
decisional trends of the Supreme Court on the issue of capital punishment decision
making. Songer (1987) found that circuit courts follow the decisional trends of the
Supreme Court with regard to labor and anti-trust decision, supporting the
hierarchical model of judicial decision making. This hypothesis was tested
through two means. First, a series of hierarchical regressions were run to
determine whether a correlation existed between Supreme Court decisional trends
on capital punishment cases and decisions of the four circuits in this study. The
first regression examined death penalty cases within the same year. A second
regression incorporated a one-year lag between the Supreme Court decisions and
the appellate courts and the third regression utilized a two year lag. These
analyses showed no support for the theory that circuit courts follow the decisional
trends of the Supreme Court in death cases as none of the regressions were found
to be statistically significant at the .05 level where the expected effect matched the
observed effect.

192
The alternative methodology performed by Songer (1987) was also
employed in the present study to confirm the findings from the regression analysis.
Of the 20 time periods examined between the four circuits, only two demonstrated
a statistical difference that moved in the same direction of the Supreme Court. A
similar finding was found when the data was controlled for political appointment
on the circuit courts. In both instances, there was very little evidence that circuit
courts were looking to the Supreme Court for guidance with regard to how the
lower courts should decide death penalty cases, thus confirming the findings from
the regression analysis. The third hypothesis therefore was found not to be true
wherein circuit court decisions on capital punishment cases did not follow the
decisional trends of the Supreme Court during the period of this study for the four
circuits examined.
The fourth hypothesis avered that circuit court judges ideological tendency
will be enhanced if at least one other member of the circuit court panel was
appointed from the same party as that judge. Conversely, the opposite will be
found to be true. Sunstein, Schkade and Ellman (2003) found that the ideological
make-up of circuit court panels both amplifies and dampens voting on various
issues. This hypothesis was tested through the alternative methodology that was
utilized by Sunstein et al. The alternative methodology offered by Sunstein
confirmed the hypothesis. The raw numbers showed how liberal voting on circuit
court panels from the four circuits studied shifted based upon the make-up of the
panels (RRR: 10.9%, RRD: 18.1%, DDR: 33.8%, DDD: 52.9%). These numbers

193
clearly demonstrate that as the number of Democrats entered onto circuit court
panels, liberal voting increased to the tune of over 50% when a panel had three
Democratic appointments. This conclusion did not change even after calculating
the predicted probabilities. In this case, the percentage of liberal voting for each
combination of panels exceeded their probability based upon the political
appointment of the judges. Thus, based upon these numbers, one is left with the
conclusion that more is taking place within circuit panel votes on death penalty
cases beyond mere voting based on political stereotypes of the judges.
Sunsteins analysis did more than merely demonstrate that institutional
factors are involved in circuit court decision making. Sunsteins methodology also
allowed for the determination of the effect that certain panel make-ups had on
members of the panels. What became clear was that when a panel was made up of
two members of one party and one member of the other party, the minority panel
members liberal voting was affected whether the majority members were
Democratic or Republican appointments. However, the evidence established that
Democratically-appointed judges were more apt to be affected by the make-up of
the panels than their Republican-appointed counterparts. Although the explanation
for this difference is unclear, what is clear is that the make-up of circuit court
panels does affect the judicial voting in capital punishment decision making.
Thus, hypothesis four was found to be true.
The fifth hypothesis states that circuit judges will comply with signals sent
to them from the United States Supreme Court with respect to the manner in which

194
the lower court should resolve a death penalty cases. Segal, Songer and Cameron
(1995) found that a principal-agent relationship exists between the Supreme Court
and the circuit courts and that the Supreme Court sends direct and indirect signals
to the circuits to alter their decision making practices. The testing of this theory
examined four signals sent from the United States Supreme Court to the circuits in
the form of specific orders directed at a particular circuit or written opinions on
death penalty cases. Circuit court voting on death penalty cases was calculated six
months, one year and two years before the signal and then after the signals to
determine whether any statistical difference was seen based upon the signal. No
statistical difference was detected at the .05 level at any of the three intervals. Five
instances did show changes in circuit court voting following Supreme Court
signals, however, in each instance the shift in circuit voting pattern was counter to
the signal sent from the Supreme Court. Thus, those five instances did not lend
support that circuits respond to signals from the Supreme Court. In light of these
findings, the fifth hypothesis must be found not true.
The final hypothesis suggests the circuit courts will comply with
Congressional signals concerning death penalty litigation. A post-AEDPA
independent variable was included in both sets of regressions (dichotomous
outcome and ordered dependent variables). This variable was found to be
statistically significant within the .05 level of significance and with the correct sign
in each regression. The importance of this particular variable is also seen with the
magnitude of effect of 18.9%, while holding all other variables constant. Thus,

195
one can classify the impact of the post-AEDPA signal as a clearly important
variable on death penalty decision making.
A corollary hypothesis suggests that partisan signals in support of or
against the death penalty from a governors office of the state where the death
sentence was handed down impacts circuit court decision making in capital
punishment cases. A governor variable was included in the regression analyses;
however, the results were starkly different from what was seen with the panel
ideology and post-AEDPA variables. The governor variable was not found to be
statistically significant at the .05 level. Moreover, a low magnitude of effect was
found at 1.9% further confirming the de minimus impact of this particular variable
on circuit court decision making. Thus, any signal from a governors office is not
felt and/or followed by the circuit courts on this issue.
Some contrary findings were found when examining the alternative
methods used in the signal theory analysis. The method used looked at specific
Congressional and Gubernatorial signals and determined whether any statistical
difference at the .05 level was seen in death penalty decision making six months,
one year and two years before and after the signals. Neither the post-AEDPA nor
the Federal Death Penalty Act signals from Congress showed any statistical
difference in death penalty decision making at the .05 level for any of the three
time periods. Likewise, the death penalty moratorium issued by Illinois Governor
Ryan in 2000 did not show any effect by any one circuit during the three time
periods but the cumulative voting of all circuits two years before the moratorium

196
and two years after did demonstrate a statistical difference within the .05 level.
Whether this shift in decision making was attributable to Governor Ryans
announcement is questionable in light of the two year time lag for the shift to take
place. That being said, this data does offer some evidence that the circuit courts do
look toward certain extra-legal signals with regard to the death penalty. Although
the alternative methods used appear to conflict with the regression analysis, the
end result is clear that courts appear to listen to signals from bodies outside that of
the judiciary.
The ultimate conclusions to be drawn from the testing of these hypotheses
are clear. First and foremost, support was found for the attitudinal, institutional
and legal models of circuit court decision making on the issue of capital
punishment. Perhaps the most important aspect of this finding was that each
model was found was found to play a role in circuit court decision making. Some
scholars have suggested there is but one dominant paradigm of judicial decision
making, namely, the attitudinal model (see George 1998, 1642; Cross 2005, 326,
n.4). The results from this study clearly show otherwise. It is undoubtedly true
that personal policy preferences play an important role in terms of circuit court
decisions on capital punishment cases. However, that being said, the importance
of this one variable cannot be heralded to the exclusion of all others. Clearly
internal (panel influence) and external (Congress) institutional pressures on circuit
courts influence decision making on death penalty issues. Thus, there is no one

197
all-dominant variable when examining capital punishment cases but rather a
select few that appear to control judicial decisions.
Second, some evidence was found supporting the idea that certain case
characteristics may affect decision making. However, evidence of such an impact
was minimal, at best. Likewise, gubernatorial signals to the circuit courts likewise
did not appear to affect judicial decisions. Very little support was also found to
support the hierarchical model. It appears the circuits are simply not willing to
comply with the decisional trends of the Supreme Court with regard to capital
punishment decisions. Thus, the testing of each model through several methods
resulted in a finding that only a few factors are impacting circuit court judges in
death penalty decision making.
The real issue therefore to be addressed is not whether extra-legal factors
play a part in circuit court decision making but rather to what degree. Although
many may place judges above the fray in terms of influential forces, the reality is
that judges are a mere reflection of society. Since society is affected by various
forces, it should not be surprising that judges are likewise affected. Moreover,
politics has been part of our judiciary since its very inceptions, as evidenced by
Justice John Marshalls most eloquent opinion in Marbury v. Madison (1803).
Thus, scholars need to stop placing the American judiciary on a non-political
pedestal and realize that politics always has and always will be part of the courts.
Perhaps a more constructive approach would be to think of the American judiciary,
as some suggest (see Perretti (1999)), as nothing more than another political

198
branch of the American government that perpetuates democracy through the
nomination process and is kept in check through various internal and external
constraints as is the case with the executive and legislative branches.
That being said, not all issues presented before courts possess equal levels
of political influence. Certain issues are simply more politically-oriented than
others. Capital punishment is one of those highly charged political issues that
brings politics to the forefront in the courts whether one is discussing the circuit
courts of appeals or other courts, for that matter. Therefore, the intriguing
question and one that may be taken up in future research is at what point is politics
infused into the decision making process. Capital punishment, abortion and
political elections are all prime examples of issues that one would expect politics
to be involved in judicial decisions but many issues presented before the courts
possess lower levels of political influence. By understanding where this dividing
line takes place, a better understanding of the American judiciary can be had. A
similar question that merits future examination is how clear of a signal must a
Congressional be in order for the courts to take notice. In this case, the precise
language of the AEDPA sent an unmistakable signal to the courts and this study
demonstrated the relevance of this signal in circuit court decisions. However,
what is not clear is the degree of specificity necessary from Congress to the courts
in order for a palpable change of judicial decision making to take place.
The study of judicial decision making at the circuit court level on the issue
of capital punishment has demonstrated that no one decision making model fully

199
explains all aspects of this process. American courts are a fluid concept that
change over time and, for this reason, what may affect courts today may not
necessarily do so in the future. For this reason, scholars must continue to utilize a
variety of methods in analyzing various legal doctrine in order to gain greater
insight into what drives the judicial decision making process in the United States
court system.

200
Bibliography
Aarons, Dwight. 1998. Getting Out of this Mess: Steps Toward Addressing and
Avoiding Inordinate Delay in Capital Cases. Journal of Criminal Law and
Criminality 89:1.

Achen, Christopher H. 1982. Interpreting and Using Regression. Newbury Park:
Sage Distribution.

Armstrong, K. and M. Possley. (1999, January 10-14). Trial and Error: How
Prosecutors Sacrifice Justice to Win. Chicago Tribune[5-part series].

Atkins, Burton M. 1973. Judicial Behavior and Tendencies Towards Conformity
in a Three Member Small Group: A Case Study of Dissent Behavior on the
U.S. Court of Appeals. Soc. Sci. Q. 54:41.

Beckett, Katherine. 1997. Making Crime Pay: Law and Order in Contemporary
American Politics. New York: Oxford University Press.

Benesh, Sara C. 2002. The U.S. Court of Appeals and the Law of Confessions
Perspectives on the Hierarchy of Justice . New York: LFB Scholarly
Publishing LLC.

Bohm, Robert M. 1987. American Death Penalty Attitudes: A Critical
Examination of Recent Evidence. Criminal Justice & Behavior14:380.

Bowers, William J., Margaret Vandiver, and Patricia H. Dugan. 1984. A New
Look at Public Opinion on Capital Punishment-What Citizens and Legislatures
Prefer. American Journal of Criminal Law 22:77.

Bowman, Frank O. III. (2005). Symposium: Panel Four: The Institutional
Concerns Inherent In Sentencing Regrimes: The Failure of the Federal
Sentencing Guidelines: A Structural Analysis. Colum. L. Rev. 105:1315.

Brace, Paul and Melinda Gann Hall. 1997. The Interplay of Preferences, Case
Facts, Context, and Rules in the Politics of Judicial Choice. Journal of
Politics 59:1206.

Brace, Paul and Melinda Gann Hall. 1995. Studying Courts Comparatively: The
Views from the American States. Pol. Res. Q. 48:5.

Brace, Paul and Melinda Gann Hall. 1993. Integrated Models of Judicial
Dissent. Journal of Politics 55:914.

201
Burton, Shawn. 2003. Justice Scalias Methodological Approach to Judicial
Decision Making: Political Actor or Stategic Institutionalists? U. Tol. L. Rev.
34:575.

Cameron, Charles M., Jeffrey A. Segal and Donald R. Songer. 2000. Strategic
Auditing in a Political Hierarchy: An Informational Model of the Supreme
Courts Certiorari Decisions American Political Science Review 94:101.

Caminker, Evan H. 1994. Why Must Inferior Courts Obey Superior Court
Presents? Stanford Law Review 46:817.

Cardozo, Benjamin. 1921. The Nature of the Judicial Process. New Haven: Yale
University Press.

Carp, Robert A., and C.K. Rowland. 1983. Policymaking and Politics in the
Federal District Court. Knoxville: University of Tennessee Press.

Carter, Linda E. (2005). Lessons From Avena: The Inadequacy of Clemency and
Judicial Proceedings for Violations of the Vienna Conventions on Consular
Relations. Duke J. Comp. & Intl 15:259.

Clayton, Cornell W. and Howard Gillman, eds. 1999. Supreme Court Decision-
Making: New Institutionalist Approaches. Chicago: University of Chicago
Press.

Cochran, Jill M. 2004. Counting Death: 30 Years Since Furman, Is The Death
Penalty Any Less Discriminatory? Looking At The Problem of Jury
Discretion In Capital Sentencing. Val. U.L. Rev. 38:1399.

Cox, Deirdre J. 1993. Criminal Procedure: The Robert Alton Harris Decision:
Federalism, Comity and Judicial Civil Disobedience. Golden Gate U.L. Rev.
23:155.

Cross, Frank B. 1997. Political Science and the New Legal Realism: A Case of
Unfortunate Interdisciplinary Ignorance. Nw. U.L. Rev. 92:251.

Cross, Frank B. and Emerson H. Tiller. 1998. Judicial Partianship and Obedience
to Legal Doctrine: Whistleblowing on the Federal Courts of Appeals. Yale
L.J. 107:2155.

Cullen, Francis T., Bonnie S. Fisher and Brandon K. Applegate. 2000. Public
Opinion about Punishment and Corrections. Crime & Just. 27:1.

202
Culter, Christopher Q. 2000. Death Resurrected: The Reimplementation of the
Federal Death Penalty. Seattle Univ. L. R. 23:1189.

Dickson, Brian. 2000. A Life In The Law: The Process of Judging. Sask. L. Rev.
63:373.

Eckstein, Harry. 1992. Regarding Politics Essays on Political Theory, Stability,
and Change. Berkeley, CA: University of California Press.

Editorial, San Francisco Chronicle, 24 April 1992, final edition.

Espstein, Lee and Jack Knight. 2000. Toward a Strategic Revolution in Judicial
Politics: A Look Back, A Look Ahead. Political Research Quarterly
53(3):625

Epstein, Lee and Joseph F. Kobylka. 1992. The Supreme Court and Legal Cause.
Chapel Hill, NC: The University of North Carolina Press.

Eskridge, William N. 1990. The Next Textualism. UCLA L. Rev. 37:621.

Fagan, Jeffrey and Valerie West. 2005. Innocence in Capital Sentencing: Article:
The Decline of the Juvenile Death Penalty: Scientific Evidence of Evolving
Norms. J. Crim. L. & Criminology 95:427.

Feingold, R. 2000. The Death Penalty Under Attach. Criminal Justice 16:18.

Gaffney, Matt. 2003. Criminal LawThe Sixth Amendment Clash: Judges vs.
Juries. Ring v. Arizona, 122 S.Ct. 2428 (2002). Wyo. L. Rev. 3:769

George, Tracey E. 1999. The Dynamics and Determinants of the Decision to
Grant En Banc Review. Wash. L. Rev 74:213.

George, Tracey E. 1998. Developing a Positivist Theory of Decisionmaking on
the United States Courts of Appeals. Ohio St. L.J. 58:1635.

George, Tracey E. and Lee Epstein. 1992. On the Nature of Supreme Court
Decision Making. American Political Science Review 86:323.

George, Tracey E. and Michael E. Solimine. 2001. Supreme Court Monitoring of
the United States Courts of Appeals En Banc. S. Ct. Econ. Rev. 9:171.

Gillman, Howard. 2001. What's Law Got to Do with It? Judicial Behavioralists
Test the "Legal Model" of Judicial Decision Making. Law & Soc. Inquiry
26:465.

203
Gillman, Howard. 1993 . The Constitution Besieged: The Rise and Demise of
Lochner Era Police Powers Jurisprudence . Durham, N.C.: Duke University
Press.

Hagle, Timothy M. and Glenn E. Mitchell II. 1992. Goodness-of-Fit Measures for
Probit and Logit. American Journal of Political Science 36:762.

Harvard Law Review. 2000. The Prison Reform Act and the Antiterrorism and
Effective Death Penalty Act: Implications for Federal District Judges. Harv.
L. Rev. 115:1846.

Heise, Michael. 1998. Equal Educational Opportunity, Hollow Victories, and The
Demise of School Finance Equity Theory: An Empirical Perspective and
Alternative Explanation. Ga. L. Rev. 32:543.

Herman, Joshua. 2004. Death Denies Due Process: Evaluating Due Process
Challenges To The Federal Death Penalty Act. DePaul L. Rev. 53:1777.

Heisler, Richard. 2004. The Kids Are Already: Roper v. Simmons and the
Juvenile Death Penalty After Atkins v. Virginia. Sw. U. L. Rev. 34:25.

Hoffman, Joseph L. 2005. Protecting the Innocent: The Massachusetts
Governors Counsel Report. J. Crim. L. & Criminology 95:561.

Hoffstadt, Brian M. 2005. The Deconstruction and Reconstruction of Habeas. S.
Cal. L. Rev. 78:1125.

Kahan, Dan M. 1999. The Secret Ambition of Detterence. Harv. L. Rev. 113:413.

Kauper, Thomas E. 1968. The Warren Court and the Antitrust Laws: Of
Economics, Populism, and Cynicism. In Richard H. Sayler, Barry B. Boyer,
and Robert E. Goodling, Jr.. The Warren Court: A Critical Analysis. New
York: Chelsea House.

Kalven, Harry, Jr. 1973. Toward a Science of Impartial Judicial Behavior. U.
Cin. L. Rev. 42:591.

King, Gary. 1998. Unifying Political Methodology: The Likelihood Theory of
Statistical Influence. Ann Arbor, Michigan: University of Michigan Press.

King, Gary. 1986. How Not to Lie with Statistics: Avoid Common Mistakes in
Quantitative Political Science. American Journal of Political Science 30:666.

204
King, Gary, Robert O. Keohane and Sidney Verba. 1994. Designing Social
Inquiry. Princeton, New Jersey: Princeton University Press.

Klein, David K. 2000. Making Law in the United States Courts of Appeals.
Cambridge, England: Cambridge University Press.

Klein, David K. and Robert J. Hume. 2003. Fear of Reversal as an Explanation
of Lower Court Compliance. Law & Soc. Rev. 37:579.

Jeffries, John C., Jr. 1994. Justice Lewis F. Powell, Jr. New York: Scribners.

Kaplan, David A. and Bob Cohn. Hung on a Technicality. Newsweek. April 6,
1992.

Kirchmeier, Jeffrey L. 2002. Another Place Beyond Here: The Death Penalty
Moratorium Movement In The United States. U. Colo. L. Rev. 73:1.

Lanier, Charles S. and James R. Acker. 2004. Capital Punishment, the
Moratorium Movement, and Empirical Questions: Looking Beyond Innocence,
Race and Bad Lawyering in Death Penalty Cases. Psych. Pub. Pol. And L.
10:577.

Latzer, Barry. 1998. Death Penalty CasesLeading U.S. Supreme Court Cases
On Capital Punishment. Maine: ButterworthHeinemann.

Liebman, James S. 2001. An Effective Death Penalty? AEDPA and Error
Detection in Capital Cases. Brook. L. Rev. 67: 411.

Little, Rory K. 1999. The Federal Death Penalty: History and Some Thoughts
About the Department of Justices Role. Fordham Urb. L. J. 26:347.

Mahoney, Matthew. 1988. United States Supreme Court: Calderon v. Thompson
118 S. Ct. 1489 (1998). Cal. Def. J. 11:47.

Markovits, R.S. 1983. The Burger Court, Antitrust, and Economic Analysis. In
Vincent Blasi, The Burger Court: The Counter Revolution that Wasnt . New
Haven: Yale University Press.

McCollough, Arron G. 2005. The Washington and Lee Law Alumni Association
Student Notes Colloquium: For Whom the Court Tolls: Equitable Tolling of
the AEDPA Statute of Limitations in Capital Habeas Case. Wash. & Lee L.
Rev. 62:365.

205
Melsner, Michael. 1973. Cruel and Unusual: The Supreme Court and Capital
Punishment. New York: Random House.

Moore, James William, Jo Desha Lucas and Thomas S. Currier. 1993. Moores
Federal Practice. California: Matthew Bender.

Murphy, Amy R. 2000. Further Developments: The Constitutional Failure of the
Strickland Standard in Capital Cases Under The Eighth Amendment. Law &
Contemp. Prob. 63:179.

Murphy, Walter. 1966. Court as Small Groups. Harv. L. Rev. 79:1551.

Murphy, Walter. 1964. Elements of Judicial Strategy. Chicago: The University of
Chicago Press.

Nkama, Jude Obasi. 2001. The Great Writ Encumbered By Great Limitations: Is
The Third Circuits Notice Requirement For Habeas Relief A Structural Bias
against Persons in Cusotyd?" Seton Hall Legis. J. 26:181.

Nevins, Allen and Henry Steele Commager. 1986. A Pocket History of the United
States (8
th
Edition). New York: Pocket Books.

OBrien, David M. 2000. Storm Center: The Supreme Court in American
Politics, 5
th
Edition, New York: W.W. Norton and Co.

Ogloff, James R. P. 2004. Struck in the Dark Ages: Supreme Court Decision
Making and Legal Developments. Psych. Pub. Pol. And L. 10:379.

Paternoster, Raymond. 1991. Capital Punishment in America. New York:
Lexington Books.

Peretti, Terri Jennings. 1999. In Defense of a Political Court. Princeton: Princeton
University Press.

Powell. 1991. Stare Decisis and Judicial Restraint. Journal of Supreme Court
History 13:16.

Preston, K. Lynn. 2004. Honorable James J. Gilbary Symposium on Law,
Religion, and Social Justice: Evolving Standards of Decency in 2003-Is The
Death Penalty On Life Support?: The Death Penalty Experiment: The Facts
Behind The Conclusions. Dayton L. R. 29:223.

Rankin, Joseph. 1979. Changing Attitudes Toward Capital Punishment. Social
Forces 58: 194.

206
Regoli, Robert M. and John D. Hewitt. 1996. Criminal Justice. Englewood Cliffs,
New Jersey: Prentice-Hall.

Reinhardt, Stephen. 1999. The Anatomy of an Execution: Fairness vs. Process.
N.Y.U.L. Rev. 74:313.

Riback, Stuart M. 2003. First Arguments at the Supreme Court of the United
States: First Argument Impressions of the Supreme Court. J. App. Prac. &
Process 5:134.

Ryan, G. H. 2000. Executive Order creating the Governor's Commission on
Capital Punishment. Retrieved June 14, 2002, from
http://www.idoc.state.il.us/ccp/ccp/executive_order/executive_order.html

Scalia, Antonin. 1989. The Rule of Law as a Law of Rules. U. Chi. L. Rev.
56:1175.

Schauer, Frederick. 1991. Rules and the Rule of Law. Harv. J. L. & Pub. Poly
14:645.

Schauer, Frederick . 1988. Formalism. Yale L. J. 97:509.

Scheck, Barry, Peter Neufeld, and Jim Dwyer. 2000. Actual InnocenceFive Days
To Execution, and Other Dispatches from the Wrongly Convicted . New York:
Random House, Inc.

Scheingold, Stuart A. 1991. The Politics of Law and Order: Criminal Process and
Cultural Obsession. Philadelphia: Temple University Press.

Scheingold, Stuart A. 1984. The Politics of Law and Order: Streetcrime and
Public Policy. New York: Longman.

Schubert, Glendon. 1974. The Judicial Mind Revisited: A Psychometric Analysis
of Supreme Court Ideology. New York: Free Press.

Schubert, Glendon. 1965. The Judicial Mind: The Attitudes and Ideologies of
Supreme Court Justices, 1946-1963. Evanston, Il: Northwestern University
Press.

Schwed, Roger E. 1983. Abolition and Capital Punishment. New York: AMS
Press.

Segal, Jeffrey A. and Harold J. Spaeth. 2002. The Supreme Court and the
Attitudinal Model Revised. Cambridge: Cambridge University Press.

207
Segal, Jeffrey A. and Harold J. Spaeth. 1999. Majority Rule or Minority Will.
Cambridge: Cambridge University Press.

Segal, Jeffrey A. and Harold J. Spaeth. 1993. The Supreme Court and the
Attitudinal Model. Cambridge: Cambridge University Press.

Segal, Jeffrey A., Donald R. Songer, and Charles M. Cameron. 1995. Decision
Making on the U.S. Courts of Appeals. In Contemporary Courts, edited by
Lee Epstein, 227-45. Washington D.C.: CQ Press.

Shapiro, Martin. 1972. Toward a Theory of Stare Decisis. Journal of Legal
Studies 1:125.

Simon, Rita J. and Dangy A. Blaskovich. 2002. A Comparative Analysis of
Capital Punishment. Maryland: Lexington Books.

Smith, Christopher E. 1997. The Rehnquist Court and Criminal Punishment. New
York: Garland Publishing, Inc.

Songer, Donald R. 1987. The Impact of the Supreme Court on Trends in
Economic Policy Making in the United States Courts of Appeals. The
Journal of Politics 49:830.

Songer, Donald R. 1982. Consensual and Nonconsensual Decisions in
Unanimous Opinions of the United States Courts of Appeals. American
Journal of Political Science 26:225.

Songer, Donald R., Jeffrey A. Segal, and Charles M. Cameron. 1994. The
Hierarchy of Justice: Testing a Principal-Agent Model of Supreme Court
Circuit Court Interactions. American Journal of Political Science 38:673.

Songer, Donald R., Reginald S. Sheehan, and Susan B. Haire. 2000. Continuity
and Change on the United States Courts of Appeals . Ann Arbor, Michigan:
The University of Michigan Press.

Songer, Donald R. and Sue Davis. 1989. The Impact of Party and Region on
Voting Decisions in the United States Courts of Appeals, 1955-1986. The
Western Political Quarterly 43:317.

Songer, Donald R. and Susan B. Haire. 1992. Integrating Alternative
Approaches to the Study of Judicial Voting: Obscenity Cases in the U.S.
Courts of Appeals. American Journal of Political Science 36:963.

208
Songer, Donald R. and Reginald S. Sheehan. 1992. Who Wins on Appeal?
Underdogs and Underdogs in the United States Courts of Appeals. American
Journal of Political Science 36:235.

Spaeth, Harold J. 1995. The Attitudinal Model. In Contemporary Courts,
edited by Lee Epstein, 296-346. Washington D.C.: CQ Press.

St. Antoine, Theodore J. 1968. Judicial Valour and the Warren Courts Labor
Decisions. In Richard H. Sayler, Barry B. Boyer, and Robert E. Goodling, Jr ..
The Warren Court: A Critical Analysis. New York: Chelsea House.

Steiker, Carol S. 2002. Wayne Morse Center for Law and Politics Symposium:
The Law and Politics of the Death Penalty: Abolition, Moratorium, or
Reform?: Capital Punishment and American Exceptionalism. Or. L. Rev.
81:97.

Sunstein, Cass R., David Schkade, and Lisa Michelle Ellman. 2003. Ideological
Voting on Federal Courts of Appeals: A Preliminary Investigation. VA. L.
Rev. 90:301.

Taibi, Anthony. 1990. Frontier of Legal Thought III: Note: Politics and Due
Process: The Rhetoric of Social Security Disability Law. Duke L.J. 1990:913.

Tate, Neal C. 1981. Personal Attribute Model of Voting Behavior of U.S.
Supreme Court Justices: Liberalism in Civil Liberties and Economic
Decisions, 1946-88. American Political Science Review 75:355.

Tate, Neal C. and Roger Handberg. 1991. Time Binding and Theory Building in
Personal Attribute Models of Supreme Court Voting Behavior, 1916-88.
American Journal of Political Science 35:460.

Taha, Ahmed E. 2005. Shows Panels Affect Judges: Evidence From The United
States District Court. U. Rich. L. Rev. 39:1235.

Tauber, Steven. 1998. On Behalf of the Condemned? The Impact of the NAACP
Legal Defense Fund on Capital Punishment Decision Making in the U.S.
Courts of Appeals. Political Research Quarterly 51:191.

Tobias, Carl. 2005. The Federal Appellate Court Appointments Conundrum.
Utah L. Rev. 2005:743.

Traut, Carol Ann and Craig F. Emmert. 1998. Expanding the Integrated Model of
Judicial Decision Making: The California Justices and Capital Punishment.
Journal of Politics 60:1166.

209
Tushnet, Mark. 1994. The Death Penalty: Constitutional Issues. New York: Facts
on File.

Vila, Bryan and Cynthia Morris. 1997. Capital Punishment In The United States
A Documentary History. Connecticut: Greenwood Press.

Walker, Thomas. 2003. David J. Danelski: Social Psychology and Group
Choice. In The Pioneers of Judicial Behavior, edited by Nancy Maveety,
248-269. Michigan: The University of Michigan Press.

Warr, Mark. 1995a. The PollsPoll Trends: Public Opinion on Crime and
Punishment. Public Opinion Quarterly 59:296.

Wasby, Stephen L. 1988. The Supreme Court in the Federal Judicial System .
Nelson Hall.

Williams, Kenneth. 2003. Should Judges Who Oppose Capital Punishment
Resign? A Reply To Justice Scalia. VA. J. Soc. Poly & L 10:317.

Whittington, Keith. 2000. Once More Unto the Breach: Post-Behavior
Approaches to Judicial Politics. Law & Soc. Inquiry 25:601.

Zimring, Franklin E. 2003. The Contradiction of American Capital Punishment.
Oxford: Oxford University Press.

210
Appendix A

Variable Codebook

standard = Dichtomous Outcome
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 4 / 1012

tabulation: Freq. Value Description
745 0 Pro Death Penalty
263 1 Anti Death Penalty

contrk = Continuous Outcome
type: numeric (byte)

range: [1,6] units: 1
unique values: 6 coded missing: 4 / 1012

tabulation: Freq. Value Description
67 1 Strong Anti-Death Penalty
41 2 Moderate Anti-Death Penalty
167 3 Minimal Anti-Death Penalty
33 4 Lowly Pro-Death Penalty
5 5 Moderate Pro-Death Penalty
690 6 Strong Pro-Death Penalty

panel id. = Ideology of the Panel
type: numeric (byte)

range: [0,3] units: 1
unique values: 4 coded missing: 70 / 1012

tabulation: Freq. Value Description
44 0 No Republican (All Democrats)
323 1 One Republican, Two Democrats
398 2 Two Republican, One Democrats
177 3 All Republican (No Democrats)

211
govr = Republican Governor
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 4 / 1012

tabulation: Freq. Value Description
356 0 Democrat
652 1 Republican

ptaedpa = Post Passage of AEDPA
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 15 / 1012

tabulation: Freq. Value Description
649 0 Before AEDPA
348 1 After AEDPA

rape = Rape Occurred During the Murder
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
778 0 Rape Not In Case
217 1 Rape In Case

kidnap = Murder Victim was Kidnapped
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
852 0 Kidnapping Not In Case
143 1 Kidnapping In Case

212
adultfem = Murder Victim was an Adult Female
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
483 0 Victim Adult Female
512 1 Victim Not Adult Female

elderly = Murder Victim is Elderly (65+)
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
929 0 Victim Not Elderly (+65)
66 1 Victim Elderly

minor = Murder Victim is a Minor (<18 years old)
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
866 0 Victim Not A Minor (<18)
129 1 Victim A Minor (<18)

policeoff = Murder Victim was a Police Officer
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
931 0 Victim Not A Police Officer
64 1 Victim A Police Officer

213
multmrd = Multiple Murders Took Place
type: numeric (byte)

range: [0,1] units: 1
unique values: 2 coded missing: 17 / 1012

tabulation: Freq. Value Description
737 0 Single Murder Case
258 1 Multiple Murder Case

214
Appendix B

Bivariate Model

Note: p< .05; N: 918 (1981-2004)
Note: The dichotomous dependent variable and each independent variable were coded
using dummy variables.
Note: Results from 1980 were dropped by STATA due to the limited number of
observations.
Variables Ex. Effect Ob. Effect Coefficient Significance
Penal Ideology - - - .6490854 .000
Governor - - - .3338674 .024
Post-AEDPA - - - .7599104 .000
Robbery - - - .0591871 .719
Rape - - - .2093808 .323
Kidnap - + .1771421 .411
Adult Female - - - .2017898 .229
Elderly - - - .1330317 .668
Minor - - - .0655028 .790
Police Officer - - - .578799 .099
Multi. Murder - + .126527 .485
Constant - .8806758

215
Appendix C

Fully Specified Model - Logistic Regression Analysis

Variables Expected
Effect
Observed
Effect
Coefficient Significance First
Differences
Panel Ideology - - - .5561448 .000 - .376279040
Governor - - - .0883898 .696 - .019540390
Pt-AEDPA - - -1.143462 .000 - .189715160
Robbery - + .0361378 .861 .007815690
Rape - + .1103262 .682 .024478860
Kidnap - - - .1611309 .556 - .034019970
Adult Female - - - .3272484 .120 - .074984890
Elderly - + .0880761 .815 .019470020
Minor - - - .3195112 .311 - .065257210
Police Officer - - - .4369762 .301 - .086925860
Multi. Murder - - - .2228541 .334 - .046463020
1981 - + 2.3382770 .007 .509740986
1982 - + 1.9032720 .011 .439373562
1983 - + .5848028 .497 .137898197
1984 - + .4810928 .537 .112286062
1985 - + 1.4859840 .049 .355292603
1986 - + .7392159 .294 .176403621
1987 - + .5282037 .463 .123884295
1988 - + .6652757 .329 .157932317
1989 - - - .4648310 .573 - .091872125
1990 - + .0762938 .930 .016832008
1991 - + 1.2942070 .061 .311933257
1992 - + .5084251 .427 .119006838
1993 - + .8467416 .216 .203267551
1994 - - - .1075801 .885 - .022954579
1995 - + 1.4137160 .028 .339256983
1997 - + .8699635 .158 .209056716
1998 - - - .1980388 .760 - .041501053
1999 - + .9091759 .181 .218813578
2000 - + 1.0382210 .099 .250672620
2001 - + 1.3043460 .045 .314287581
2002 - + 1.6636830 .009 .392964799
2003 - + 1.4162620 .028 .339828445
2004 - + 1.7425190 .005 .408813990
Circuit 4 - - -2.429795 .000 - .280447157
Circuit 5 - - -1.709752 .000 - .241735941
Circuit 6 - - - .6144994 .054 - .117174166
Note: Pseudo R
2
: .2171, p < .05, N: 918 (1981-2004)
Note: Both the dependent and independent variables were coded using dummy variables. Time
controls were included to capture differences across time by coding 24 of 25 years (1996 was
omitted). Likewise, circuit controls were included to assist in reducing omitted variable bias.
Note: Results from 1980 were dropped by STATA due to the limited number of observations.

216
Appendix D

Fully Specified Model Ordered Regression Analysis

Note: Pseudo R
2
: .1366, p < .05, N: 918 (1981-2004)
Note: Dependent variable was coded (1 to 6) and independent variables were coded
using dummy variables. Time controls were included to capture differences across
time by coding 24 of 25 years (1996 was omitted). Likewise, circuit controls were
included to account for omitted variable bias.
Note: Results from 1980 were dropped by STATA due to the limited number of
observations.
Variables Expected
Effect
Observed
Effect
Coefficient Significance
Panel Ideology + + .4881774 .000
Governor + + .2836854 .157
Pt-AEDPA + + .7323135 .002
Robbery + - - .1775503 .329
Rape + - - .1214315 .609
Kidnap + + .2638582 .277
Adult Female + + .1209913 .509
Elderly + - - .0649951 .845
Minor + + .1610445 .557
Police Officer + + .1703058 .643
Multi. Murder + + .1800290 .374
1981 + - - 2.724122 .001
1982 + - - 2.876288 .000
1983 + - - .7424604 .389
1984 + - - 1.013182 .181
1985 + - - 1.662667 .022
1986 + - - .8890869 .210
1987 + - - .8924508 .220
1988 + - - .9603088 .160
1989 + - - .7960906 .263
1990 + - - .8977492 .269
1991 + - - 1.773984 .008
1992 + - - 1.092040 .086
1993 + - - 1.711961 .009
1994 + - - .4680508 .513
1995 + - - 2.027170 .001
1997 + - - 1.634127 .007
1998 + - - .5895943 .350
1999 + - - 1.094245 .098
2000 + - - 1.260830 .042
2001 + - - 1.430164 .021
2002 + - - 1.949212 .002
2003 + - - 1.522573 .015
2004 + - - 1.841323 .002
Circuit 4 + + 2.312836 .000
Circuit 5 + + 1.626527 .000
Circuit 6 + + .3613947 .189

217
Appendix E

Summary of Liberal Votes on Capital Punishment Cases
(Individual, Colleague, and Panel)

Party Panel Colleagues Panel Composition

Summary of Liberal Votes from Various Panel Compositions
RRR RRD RDD DRR DDR DDD

10.3% 17.2% 22.0% 23.9% 35.4% 46.6%

R D D-R RR DD DD-
RR
RRR RRD RDD DDD DDD-
RRR

.163 .323 16% .156 .372 21.6% .109 .181 .338 .529 42%

218
Appendix F
Individual Liberal Votes on Circuit Panels
(2000-2004)

Year Republican Democrat D - R
1980 2/2 = 1.00 2/4 = .500 -.500
1981 7/13 = .538 8/17 = .470 -.068
1982 12/24 = .500 16/27 = .592 +.092
1983 5/18 = .277 2/30 = .066 -.211
1984 4/28 = .142 8/47 = .191 +.049
1985 6/24 = .240 11/32 = .343 +.103
1986 13/57 = .228 10/42 = .238 +.010
1987 5/50 = .100 14/42 = .333 +.233
1988 7/42 = .166 13/51 = .254 +.088
1989 2/41 = .048 3/40 = .075 +.027
1990 4/30 = .133 5/25 =.200 +.067
1991 17/50 = .340 2/25 = .080 -.260
1992 14/92 = .152 18/52 = .346 +.194
1993 9/57=.157 11/35 =.314 +.157
1994 2/60 = .033 10/30 = .322 +.297
1995 20/80 = .250 15/31 = .483 +.233
1996 11/96 = .114 11/39 = .282 +.168
1997 20/114 = .175 25/70 = .357 +.182
1998 10/141 = .070 13/80 = .162 +.092
1999 10/91 = .109 14/65 = .215 +.106
2000 13/107 = .121 31/81 = .382 +.261
2001 10/80 = .125 37/70 = .528 +.403
2002 12/90 = .133 39/86 = .453 +.320
2003 13/118 = .110 30/80 = .375 +.265
2004 40/138 = .289 34/94 = .404 +.115

219
Appendix G
Colleague Liberal Votes on Circuit Appellate Panels
(2000-2004)

Year RR DD DD RR
1980 1/1 = 1.00 1/1 = 1.00 0
1981 2/5 = .400 4/5 = .800 +.400
1982 5/7 = .714 6/10 = .600 -.114
1983 ¼ = .250 1/12 = .083 -.167
1984 1/9 = .111 4/16 = .312 +.201
1985 1/9 = .111 5/10=.500 +.389
1986 5/22 = .045 2/11 = .181 +.136
1987 1/15 = .066 5/15 = .333 +.267
1988 4/11 = .363 2/20 = .100 -.263
1989 0/13 = 0 3/14 = .214 +.214
1990 1/10 = .100 3/8 = .375 +.275
1991 5/20 = .250 3/5 = .600 +.350
1992 6/35 = .171 5/13 = .384 +.213
1993 3/19 = .157 5/12 = .416 +.259
1994 0/23 = .000 4/7 = .570 +.571
1995 8/31 = .258 5/6 = .833 +.575
1996 2/34 = .058 3/12 = .250 +.192
1997 7/42 = .166 7/19 = .368 +.202
1998 3/54 = .055 5/20 = .250 +.195
1999 4/32 = .125 4/20 = .200 +.075
2000 3/37 = .081 12/25 = .480 +.399
2001 2/29 = .068 13/20 = .650 +.582
2002 4/26 = .153 15/33 = .454 +.301
2003 5/41 = .121 10/25 = .400 +.279
2004 17/51 = .333 8/23 = .347 +.014

220
Appendix H

Panel Liberal Votes on Circuit Panels
(2000-2004)

Year RRR RRD RDD DDD DDD-RRR
1980 0/0 = 0 1/1 = 1.00 0/0 = 0 1/1 = 1 +1.00
1981 0/0 = 0 2/5 = .400 2/3 = .666 2/2 = 1.00 +1.00
1982 0/1 = 0 4/6 = .666 5/8 = .625 ½ = .50 +.500
1983 0/0 = 0 ¼ = .250 1/10 = .100 0/2 = 0 0
1984 0/2 = 0 1/7 = .142 2/9 = .222 2/7 = .285 +.285
1985 0/0 = 0 1/9 = .111 5/8 = .625 0/2 = 0 0
1986 1/3 = .333 4/18 = .222 3/13 = .230 0/0 = 0 0
1987 0/5 = 0 1/7 = .142 4/16 = .250 0/0 = 0 0
1988 0/2 = 0 4/9 = .444 2/20 = .100 0/0 = 0 0
1989 0/2 = 0 0/11 = 0 3/13 = .230 0/1 = 0 0
1990 0/1 = 0 1/9 = .111 3/8 = .375 0/0 = 0 0
1991 3/6 = .500 3/15 = .200 2/3 = .666 0/1 = 0 -.500
1992 0/10 = 0 6/25 = .240 4/12 = .333 1/1 = 1.00 +1.00
1993 1/8 = .125 2/10 = .200 4/12 = .333 1/1 = 1.00 +.875
1994 0/7 = 0 0/14 = 0 4/8 = .500 0/0 = 0 0
1995 2/11 = .181 6/20 = .300 5/6 = .830 0/0 = 0 -.181
1996 1/18 = .050 1/15 = .06 3/11 = .270 0/1= 0 -.050
1997 2/14 = .140 5/28 = .178 6/17 = .352 1/2 = .500 +.360
1998 0/16 = 0 3/38 = .078 5/18 = .277 0/1 = 0 0
1999 0/9 = .111 3/24 = .125 1/17 = .058 3/3 = 1.00 +1.00
2000 0/8 = 0 4/30 = .133 9/22 = .409 2/2 = 1.00 +1.00
2001 0/5 = 0 2/24 = .083 9/17 = .529 4/4 = 1.00 +1.00
2002 1/7 = .140 2/13 =.153 11/26 = .423 4/6 = .666 +.526
2003 1/12 = 083 3/22 = .136 6/18 = .333 4/7 = .57 +.570
2004 4/15 = .260 12/33 = .363 7/18 = .055 1/5 = .200 -.100

221
Appendix I

Liberal Voting of Various Panel Combinations
(2000-2004)

Year DDD DDR DRR RDD RRD RRR
1980 2/3 = .666 0/0 = 0 0/1= 0 0/0 = 0 2/2 = 1.00 0/0 = 0
1981 4/6 = .666 3/6 = .500 2/5 = .400 2/3 = .660 4/10 = .400 0/0 = 0
1982 3/6 = .500 10/16 = .620 4/6 = .660 3/8 = .370 8/12 = .666 0/3 = 0
1983 0/6 = 0 3/20 = .150 1/4 = .250 1/10 = .100 2/8 = .250 0/0 = 0
1984 6/21 = .285 5/18 = .277 1/7 = .140 2/9 = .222 2/14 = .140 0/6 = 0
1985 0/6 = 0 12/16 = .750 1/8 = .120 5/8 = .625 2/18 = .111 0/0 = 0
1986 0/0 = 0 6/26 = .230 4/18 = .222 3/13 = .230 8/36 = .222 3/9 = .333
1987 0/3 = 0 9/32 = .281 2/7 = .285 1/16 = .062 2/18 = 111 0/15 = 0
1988 0/0 = 0 7/40 = .175 5/9 = .555 1/20 = .050 8/18 = .444 0/6 = 0
1989 0/3 = 0 6/26 = .230 1/11 = .090 3/13 = .230 0/22 = 0 0/6 = 0
1990 0/0 = 0 6/16 = .370 2/9 = .220 1/8 = .125 2/18 = .110 0/3 = 0
1991 0/3 = 0 4/6 = .666 2/15 = .130 1/3 =.330 6/30 = .200 9/18 = .500
1992 3/3 = 1.00 8/24 = .333 6/25 = .240 3/12 = .250 12/50 = .240 0/30 = 0
1993 3/3 = 1.00 9/24 = .375 2/10 = .200 2/12 = .160 3/20 = .150 3/24 = .120
1994 0/0 = 0 9/16 = .500 0/14 = 0 1/8 = .125 0/28 = 0 0/24 = 0
1995 0/0 = 0 10/12 = .833 7/20 = .350 4/6 = .666 11/40 = .270 5/33 = .150
1996 0/3 = 0 8/22 = .363 2/15 = .130 3/11 = .272 2/30 = .066 3/54 = .055
1997 3/6 = .500 13/34 = .382 7/28 = .250 5/17 = .294 8/56 = .142 6/42 = .140
1998 1/6 = .166 12/36 = .333 4/38 = .105 2/18 = .111 7/76 = .092 0/48 = 0
1999 8/9 = .888 3/34 = .088 4/24 = .166 0/17 = 0 6/48 = .125 3/24 = .125
2000 6/6 = 1.00 20/44 = .454 8/30 = .266 7/22 = .318 7/60 = 116 0/24 = 0
2001 12/12 = 1.00 19/34 = .558 7/24= .291 7/17 = .411 4/48 = .083 0/15 = 0
2002 9/24 = .375 24/52 = .461 4/13 = .307 4/26 = .153 4/26 = .153 3/24 = .125
2003 14/21= .666 13/36 = .361 5/22 = .227 3/18 = .166 6/44 = .136 6/57 = .105
2004 3/15 = .200 16/36 = .444 14/33 = .420 5/18 = .277 22/66 = .333 12/45 =.222

222
Appendix J

Percentage of Liberal Decisions in the Supreme Court
and the Courts of Appeals
1980 2004

Year USSC N 4
th
Cir. N 5
th
Cir. N 6th
Cir.
N 9
th

Cir.
N All
Cir.
N

1980 0% 3 0% 0 100% 4 0% 0 0% 0 100% 4
1981 0% 2 0% 0 60% 10 0% 0 0% 0 60% 10
1982 40% 5 0% 0 64.7% 17 0% 0 50% 2 63.1% 19
1983 0% 6 0% 2 20% 15 0% 0 0% 0 17.6% 17
1984 25% 4 11.1% 9 26.6% 15 0% 1 0% 0 20% 25
1985 40% 5 20% 5 30.7% 13 0% 0 100% 1 31.5% 19
1986 37.5% 8 0% 3 21.4% 28 0% 0 66.6% 3 23.5% 34
1987 50% 10 100% 1 15.3% 26 0% 0 20% 5 18.7% 32
1988 66.6% 9 50% 4 10.7% 28 0% 0 50% 6 21% 38
1989 28.5% 7 0% 4 5% 20 0% 1 75% 4 13.7% 29
1990 27.2% 11 0% 8 50% 4 100% 1 40% 5 26.3% 19
1991 55.5% 9 0% 6 30% 10 0% 0 54.5% 11 33.3% 27
1992 80% 10 12.5% 8 11.5% 26 100% 1 46.6% 15 24% 50
1993 12.5% 8 14.2% 7 21.4% 14 0% 2 45.4% 11 26.4% 34
1994 28.5% 7 0% 6 0% 15 0% 0 38.4% 13 14.7% 34
1995 66.6% 3 16.6% 12 18.1% 11 66.6% 3 72.7% 11 37.8% 37
1996 50% 6 0% 13 5.2% 19 0% 2 28.5% 14 10.4% 48
1997 33.3% 3 14.2% 14 8.6% 23 66.6% 6 48% 25 29.4% 68
1998 12.5% 8 0% 30 0% 26 0% 2 43.4% 23 12.6% 79
1999 0% 3 10% 20 5.2% 19 33.3% 3 27.2% 11 13.2% 53
2000 50% 4 0% 16 9% 22 41.6% 12 53.3% 15 23% 65
2001 100% 2 0% 5 11.1% 18 35.7% 14 68.7% 16 33.3% 54
2002 42.8% 7 22.2% 9 14.8% 27 33.3% 15 83.3% 12 33.3% 63
2003 40% 5 5.8% 17 12.5% 32 45.4% 11 60% 10 24.2% 70
2004 57.1% 7 12.5% 8 31.1% 45 46.1% 13 46.1% 13 34.1% 79

223
Appendix K

Cumulative Percentage of Liberal Decisions in the Supreme Court
and the United States Courts of Appeal from 1980 to 2004

USSC 4
th
Circuit 5
th
Circuit 6
th
Circuit 9
th
Circuit

60/152 = 39.4% 17/207 = 8.2% 89/487 = 18.2% 35/87 = 40.2% 113/226 = 50.4%

Abstract (if available)

Abstract Political scientists have long suggested that the Supreme Court is affected by both legal and extra-legal variables. This quandary has spread beyond the Supreme Court as the high court is only able to address a small fraction of the cases that are presented to it each year. For this reason, political scientists have begun to scrutinize the decision making of the United States Courts of Appeals as the true court of last resort. Some in the political science community suggest the law itself is the primary motivator for judicial decisions while others argue that judges base their decisions on their own personal policy preferences. Institutionalists believe the institutional structure of the courts itself drives decision making whereas some see the hierarchical structure of the courts as impacting intermediate appellate decisions. A clear consensus among political scientists has yet to be achieved.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Supreme Court decision-making in slavery cases

PDF

A government of words: the social practice of judging in a rule-of-law system

PDF

The separation of powers and the Supreme Court: a new institutional analysis of inter-branch disputes, 1946-2005

PDF

The new dynamics of bureaucratic autonomy: courts, litigation, and agencies in the modern American state

PDF

The Supreme Court and the governing regime in political time

PDF

The life cycle of jurisprudential regimes: the supreme court and the constitutional law of civil rights

PDF

Egalitarianism begins at home and in cohorts: egalitarianism in United States public opinion, 1952-2000

PDF

A two-level analysis of foreign policy decision making: an empirical investigation of the case of China-Taiwan

PDF

Shaking the foundations of violent civil conflict: institutions, disasters, and the political economies of state-rebel interaction

PDF

Pathways of local economic development: tales of cities in the United States and South Korea

PDF

Transnational (after)life: migrant transnationalism and engagement in U.S. and Mexican politics

PDF

State restructuring and pathways to local democracy: idea and politics of decentralization in Japan and Korea

PDF

Courts and collective action: Legal mobilization in the United States environmental movement

PDF

Neither purse nor sword: strategic minority opinions and judicial policy-making

PDF

Institutional variance of the democratic peace, 1816-2002: electoral, executive, and federal institutions in time and space

PDF

The impact of 9/11 on the judicial treatment of Middle Eastern asylum applicants in the U.S. courts

PDF

Private interests in American government institutions

PDF

Fictions of representation: narrative and the politics of self-making in the interwar American novel

PDF

Assessing the implementation of the RECLAIM cap-and-trade market for pollution: measurement issues in counterfactuals, goal attainment, and command-and-control alternatives

PDF

A discussion of inter-rater reliability and content validity of Westphal's decision making inventory (1967), describing developmental processes of adults with developmental delays

Asset Metadata

Creator Auerbach, Arthur H. (author)

Core Title Death and politics in the United States Courts of Appeals

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Political Science

Publication Date 04/20/2007

Defense Date 02/14/2007

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag judicial decision making,judicial politics,OAI-PMH Harvest,United States Courts of Appeals

Language English

Advisor Gillman, Howard (committee chair), Barnes, John E. (committee member), Yackee, Susan Webb (committee member)

Creator Email aauerbac@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m429

Unique identifier UC1116763

Identifier etd-Auerbach-20070420 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-480840 (legacy record id),usctheses-m429 (legacy record id)

Legacy Identifier etd-Auerbach-20070420.pdf

Dmrecord 480840

Document Type Dissertation

Rights Auerbach, Arthur H.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu