.. . wﬁwwmuﬁa
. ...?Q n J. . ...
.. .ww. w»???
.. aw. ... an.

.13

, . . r
. b, .. u ..
(..an ...”..y. ;

.2. ..ﬂw ru- gs}: . p {a «...

.. ....MW “aha. ..me . . .35....)

, a .w..,‘.wx.& aﬁwmmmﬂywam MW.

4. ¢. .ﬁéva.” ,.....‘. ...?m

mamwduwunl m.q,..¢$un...n tank #5... ii.
A «Eran; .3 H 3!. 1......“ *hﬂmwhﬂxh.

I;

. ﬁn}... .n i _ a“? aw...
aw. @vﬁmmnft .... n .
‘gl . i 9,

 

.
. .. , . . 3K». Ivywk I A. .
b , u

i

v p . . O
.. . . 1. . t
. ‘ . 35‘3””..31‘ 5.3%

amamwvfﬂuwmmwr... . .. .. . , . a ...
. .. ‘ .. . .. . . ,3 ..2...h.‘...+m‘.w:.

., 2.:

,1 ‘4.
..1 x t

.-
.1-nn%..... , . q .

,
9
4.. ﬂ
{Biff . I: V 2 . .. .: . n
.t , . , .. .. . . . . ,
n J... .. . . . ... l. . . . I, n . ‘ . .. . .A
. . , , 1. . ‘ .. I . .
v.‘ I [7 . . . . . . . .. ... .. . . , .3. . A. . ‘ . . . . . u
. A . T . .. . . .. . u y f, x .u , . 7 . .. . .. . . .... . .. ., w .n 01"”! 'ylncorﬂ II:
. ... : ...; .h ...I ‘ ... Iv: 01961 . .
. . v ' t . .. . . . W. . 2h .,

.rw ‘

LI
«1"
Mil

. . . ‘ . . .
. ‘ . . _ . ., .. icing. . , z . . . . . . . .. . ‘ . . L if hm...“
A A , a u ‘ .‘ I l r n t I - W ‘ ‘ v . u I A: I

...... . . . , . 431.1. . 3...... . . .. . ‘ . .. . . 1 . 110-2173};
...- . V . . A . . . . .. .....J-uix. . i ..yzul... I . . . ‘ . ‘ . .. . ..(L l . ...».ruéhv .. -.xlf'vXJIIII:~II’03.IL A
. 9. #21. {...-.32... ...of .. . . .. . . . gtuﬂutkuvuuorf tin»... 1:03 ’r
13.5.! 1.: A. 1-4”. I . y ‘ . k . . to. . .-- I . ..hixola)«ol|$.‘|no.lIavn (A, 31”.?

rt» : . . . ...lq. . .. :10 It! a.

. . . r VJ).
‘ ol\

 

‘ 9,.
_ ‘ I . .
1..” , oi b: .. . ,
..‘Mugﬂvtﬂubﬁv... . A. . . ._ n , 3...!» 52.1...
. .l‘l . . s . . v . . n1.....n- titliav: 9.ka .L L
12...... .. . . . , . .. 25.51155... 1.. .3! , .
a. . ._ . ... . . b, . I I c ....5!‘ 245‘; . . .. 1
. , u . ‘ A A ll}. v4: D. s. . ‘ . A '1 I . .
... .. an...» } . . . H. . Z . 1;. 3 § .
, 11 d . F I. n I l I A
>"k. .‘ ' it'l'-‘ ‘ ‘l ' | I 1
Warm ‘ 1 L .. .1!

 

.
VA .-
r . .. . .. ’ -
. iatiiﬂnfl‘WmhMagf . , . ,
. «49".: 3.. .. . . : . , . . . l . . .
1.1.: . . .... v1 . . .. .. ,2
v i . . . . . I“ 11:. {372.}: ... .4 .

. . I '9 191A.) v|-. I, .. . . ‘ . . .../.3 ‘ . ... . ‘ ... 30.1““; .HV

. . .. . V .T. . .. dﬂ‘vfttuvmg.‘ inn

.1... .
.l .
b. in”: 4.
1 E10: 1'.
Hut: by .95.. 1.549“.
Ersvra..-w .
r ....B
. . .0.
oo. ‘ -.....
. in . lb: 75.1....
1 ii 1?:Jﬂ1 ...: .
, . 04.5.... 55.115... v.3. . . . . .
.3312; .4 ....r;..és3u..£ . . w. .e ‘
.. ‘n.. l.‘.I.O . . . 21 vﬁl- ‘4.
. .....Jnl. ; .5!) :33‘ ”5.....iﬂnhlvahﬂ
.1. , . 1.7;. rs, . ...hurulvnnuhnurx 33!»: ...!!z‘ ..
.7118 «1.5.8.13! T... ....ﬂm. KKK!- no
y .. \1. i

 

£2 .

ii. .....t. ii !. 2..
.; .....a-‘¢ b)-o,£..
,5...L.5irLab.!¥: s v

‘9
:3,

«than...

«:6 GI .
. I

 

..
‘l . . , . A

, . ih..ﬁ.,.:..o"h..ui. unﬂiuﬂ . . .
5H,” I. |s"|;.

.l . y 51.2, r
1|> H... 1 WE. ...«dﬁhﬁiu.
...umwumwﬁunmmavﬂamﬂz.
(ﬁstwuu ... . -.Htudormh, a... ....
. 4.1.1? r34 3&1?»
. ... «ﬂavour human.
. . :2? 1.4; I: ...bkﬁbl .
..WH" ”Ogilvhru?
‘. ... Urdlg. l . r

 

‘ Illa .00 1|. n

Y ,ﬂulﬂro. pt . ... v... .7 3' ,I. n ‘ . . ,.. ‘ . .y ‘ , ‘ . , .
....{ﬁxﬂﬂ‘ww .. . . ...... . .. A . ..
.. . .. ., ,. r ‘ t . viliailuwuw

y tilttll 1"!
031:,In\0' d w

 

. . thy”. I3.- ‘ .
Ivo‘ﬂnvl" .‘lv'o. (33.... I z . ‘
l _ ' .am.v1QY .J‘U‘l.1.tu . . . . T 6f :4...-

 

 

. l .. . k
. |\~?' HA.

THESlS

lilllllllllilllilllllllllllllllllllI'llllllllllllllllllllllll

31293 01563 4797 1

This is to certify that the

dissertation entitled

The Applicability of Legal and Attitudinal Models
to the Treatment of Precedent in the Courts of Appeals

presented by

Anna Malia Reddick

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Political Science

 

'54.,

Major professor

Date M

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

LIBRARY
Michigan State
Unlverslty

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINEB return on or before date due.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU to An Afﬁrmative Mien/Equel Opportmlty lm
WM1

_)__—_.__>—————.

 

THE APPLICABILITY OF LEGAL AND ATTITUDINAL MODELS
TO THE TREATMENT OF PRECEDENT IN THE COURTS OF APPEALS

By

Anna Malia Reddick

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Political Science

1997

ABSTRACT

THE APPLICABILITY OF LEGAL AND ATTITUDINAL MODELS
TO THE TREATMENT OF PRECEDENT IN THE COURTS OF APPEALS

By

Anna Malia Reddick

A great deal of research has been conducted regarding whether a legal or attitudinal
model applies to Supreme Court decision making, but it is much less clear which model, if
either, pertains to the US. Courts of Appeals. To address this question, I analyze the
inﬂuence of attitudinal and nonattitudinal factors on the behavior of circuit court judges.
Speciﬁcally, I utilize Shepard ’5 United States Supreme Court Citations to ascertain the
treatment afforded Supreme Court precedents by the Courts of Appeals, and I assess the
extent to which characteristics of Supreme Court decisions and the ideological composition
of appellate panels determine this treatment. I ﬁnd that judicial preferences motivate circuit
court decision making in some types of cases; however, they do not account for behavior
across a wide range of issue areas. Legal considerations must also be incorporated into a

comprehensive explanation of the decisions of the Courts of Appeals.

ACKNOWLEDGMENTS

It would be an oversight on my part not to recognize some of the people who are in
part responsible for the completion of this dissertation. First and foremost, I must thank my
mother and father for their continual love and support. Throughout my twenty-seven years,
they have instilled in me the notion that, with hard work, I could achieve any goal to which
I aspired. Obtaining my Ph.D. would not have been possible without their faith in me, their
guidance, and their encouragement.

There are other members of my family, as well, to whom I owe a debt of gratitude.
My sister, Kalai, has commiserated with me throughout my graduate school education and,
at times, has been more concerned about my progress than even I have been. I appreciate
more than I can express the source of strength that she and her husband, Chris, have been to
me. My sister, Lisa, and her husband, Shawn, have also been very supportive of my
endeavors. My brother, Cam, is my kindred spirit and the most interesting auditor I know.
He has helped me keep my sense of humor during this process. My sister, Hannah, was the
most special surprise a family could receive. I will always remember her attending ICPSR
classes with me and comparing time series methodology to what she does in biology! And,
of course, there is my nephew, Hayden, who has provided a constant temptation not to work

on the dissertation.

iii

The person who has been most instrumental in my development as a public law
scholar is Reggie Sheehan. He took me under his wing from my ﬁrst day of graduate school
and has guided and encouraged me through every stage of my graduate studies. I appreciate
his weekly pep talks, during which we brainstormed about all the research questions that we
could explore if I would only get busy! He has been my ﬁiend as well as my mentor, and I
look forward to maintaining our ﬁ'iendship and our research collaboration.

Harold Spaeth could always be depended upon to provide an alternative point of
view, and to help me keep things in the proper perspective. Stopping off in his ofﬁce on my
way down the hall always proved to be entertaining! I also appreciate the hours he spent
poring over the drafts of my chapters and providing constructive comments.

Finally, I must thank Tom Howard, my professor and friend at Harding University,

whose example inspired me toward a career in education and research in the first place.

iv

TABLE OF CONTENTS

LIST OF TABLES ..................................................... vii
LIST OF FIGURES ..................................................... ix
INTRODUCTION ....................................................... 1
CHAPTER 1
LOWER COURT TREATMENT OF PRECEDENT: A REVIEW ................. 3
Determinants of Lower Court Treatment of Precedent ..................... 5
Approaches to Studying the Treatment of Precedent ...................... 10
Empirical Evidence of Lower Court Treatment of Precedent ............... 13
Decision Making on the Courts of Appeals ............................. 16
CHAPTER 2
RESEARCH DESIGN AND METHODOLOGY .............................. 20
Shepard ’s United States Supreme Court Citations ....................... 22
Courts of Appeals’ Treatment of Four Supreme Court Decisions:
An Exploration ................................................... 27
Courts of Appeals’ Treatment of the Decisions of the 1987 Term:
A Multivariate Analysis ............................................ 28

Courts of Appeals’ Treatment of Civil Rights and Liberties Decisions from 1966-
1975: A Multivariate Analysis and a Comparison of Treatment Among and

Within Circuits ................................................... 36

CHAPTER 3

COURTS OF APPEALS’ TREATMENT OF FOUR SUPREME COURT DECISIONS:

AN EXPLORATION .................................................... 43
US. v. Wade ..................................................... 44
Courts of Appeals’ Treatment of Wade ................................ 47
Griggs v. Duke Power Co. .......................................... 50
Courts of Appeals’ Treatment of Griggs ............................... 52
Lemon v. Kurtzman ............................................... 56
Courts of Appeals’ Treatment of Lemon ............................... 59
Terry v. Ohio .................................................... 64

Courts of Appeals’ Treatment of Terry ................................ 66

Discussion ...................................................... 68
CHAPTER 4
COURTS OF APPEALS’ TREATMENT OF THE DECISIONS OF THE 1987 TERM:
A MULTIVARIATE ANALYSIS .......................................... 72
Hypotheses ...................................................... 75
Findings ........................................................ 81
Discussion ...................................................... 90
CHAPTER 5

COURTS OF APPEALS’ TREATMENT OF CIVIL RIGHTS AND LIBERTIES
DECISIONS FROM 1966-1975: A MULTIVARIATE ANALYSIS AND A COMPARISON

OF TREATMENT AMONG AND WITHIN CIRCUITS ........................ 98
Hypotheses ...................................................... 99
Findings ....................................................... 106
Discussion ..................................................... l 12
Inter- and Intra-Circuit Variation .................................... 115

CONCLUSION ....................................................... 136

APPENDICES
Comparison of Analyses Including and Excluding the Coding of
Dissenting Opinions .............................................. 143
Fit and Inﬂuence Diagnostics for Logistic Regression ................... 144

LIST OF REFERENCES ................................................ 147

LIST OF CASES ...................................................... 154

vi

Table 3.1

Table 4.1

Table 4.2

Table 4.3

Table 4.4

Table 4.5

Table 5.1

Table 5.2

Table 5.3

Table 5.4

Table 5.5

Table 5.6

Table 5.7

LIST OF TABLES

Courts of Appeals’ Treatment of Four Supreme Court Decisions ...... 71

Operationalization of Concepts and Directional Relationships

Between Variables: The 1987 Term ............................ 93
Bivariate Relationships Between Treatment and Legal Factors:

The 1987 Term ............................................. 94
Logistic Regression of Treatment: The 1987 Term ................. 95
Predicted Probabilities of Following Precedent: The 1987 Term ...... 96
Logistic Regression vs. Bounded Inﬂuence Logistic Regression:

The 1987 Term ............................................. 97
Operationalization of Concepts and Directional Relationships

Between Variables: Civil Rights and Liberties ................... 125
Logistic Regression of Treatment: Civil Rights and Liberties

(Model 1) ................................................ 127
Predicted Probabilities of Following Precedent: Civil Rights and

Liberties ................................................. 128
Logistic Regression of Treatment: Civil Rights and Liberties

(Model 2) ................................................ 129
Circuit Composition by Party of Appointing President ............. 130

Rates of Following Precedent by Circuit: Civil Rights and Liberties . . 131

Estimated Coefficients and Standard Errors for Circuit Variables:
Civil Rights and Liberties ................................... 133

vii

Table 5.8

Table 5.9

Table A]

Rates of Following Precedent by Ideological Composition of Panel:
Civil Rights and Liberties ................................... 134

Rates of Following Liberal vs. Conservative Precedent by Circuit and
Ideological Composition of Panel: Civil Rights and Liberties ....... 135

Comparison of Analyses Including and Excluding the Coding of
Dissenting Opinions: The 1987 Term .......................... 143

viii

LIST OF FIGURES

Figure B.1 Model Fit and Inﬂuence of Outlying Cases ...................... 146

ix

INTRODUCTION

In his seminal study of the US. Courts of Appeals, Howard (1981, 8) described these
courts as the “vital center of the federal judicial system.” The Courts of Appeals merit such
a description because they are the courts of last resort in the vast majority of federal cases.
Ideally, it is the Supreme Court that oversees the judiciary and ensures the uniform
application of national law, but in reality the high court exercises very little direct
supervision over the lower courts (Murphy 195 9). Even though the number of requests for
Supreme Court review has gone up dramatically over the past several years, the number of
cases accepted by the Court has declined. According to one study (Songer 1991), less than
one-half of 1 percent of the decisions of the Courts of Appeals are reviewed by the Supreme
Court. Since the 19603, the subject matter of cases litigated in the federal courts has
changed, so that today most of these cases deal with public, rather than private, law issues.
Thus the bulk of the responsibility for interpreting and applying federal law, and for
resolving important political and social questions, falls to the circuit courts.

Recognition of the policy making role of the Courts of Appeals raises two concerns.
First, because of a lack of accountability to a higher court, appeals court judges are free to
exercise a great deal of discretion in decision making. These judges may adhere to Supreme
Court decisions with which they agree and avoid decisions with which they disagree.

Second, the consequence of this independence may be the creation of policy on a regional,

2

rather than a national, basis. Most work on the circuit courts’ application of Supreme Court
precedent indicates that there is a great deal of variation in the extent to which they follow
precedent and that much of this variation is a function of the circuit in which decisions are
made. These products of limited Supreme Court oversight -- judicial independence in policy
making and regional policy making -- foster questions about the legitimacy of the Courts of
Appeals’ capacity to create law. Therefore, the extent to which these concerns represent
empirical reality should be established.

Do appeals court judges behave independently, or are they constrained by the norms
and values of the legal profession; and, does this independence threaten the uniformity of
national law? These questions will be addressed here through the testing of two models of
judicial decision making. A legal model of decision making holds that judges are bound by
precedent set by higher courts, while an attitudinal model posits that judges act according to
their own attitudes and policy goals. In this study, components of both explanations of
judicial behavior are incorporated into models of the Courts of Appeals’ treatment of
Supreme Court precedent, in order to determine which explanation provides a better account

of decision making on the Courts of Appeals.

CHAPTER 1

LOWER COURT TREATMENT OF PRECEDENT: A REVIEW

The judiciary in the United States is organized hierarchically, with the Supreme Court
at the apex and several layers of federal and state courts below. Although the framework of
the court system is undisputed, very different accounts are offered in the judicial politics
literature of the implications such a structure has for the behavior of its components. These
descriptions differ from one another with respect to the degree of control each allows the
Supreme Court over lower courts and the resulting policy making power each attributes to
lower courts.

One explanation of the relationship between the Supreme Court and lower courts is
hierarchical theory, where lower courts within the system are responsible for enforcing the

policy directives of higher courts, and a lower court’s interpretation of a higher court’s

decision is subject to review by the higher court (Baum 1976). According to this view,
Supreme Court policies are routinely and obediently complied with in the lower courts.
Taking this approach, early legal scholars subscribed to what has been termed an “upper-
court myth” (Frank 1963) and focused almost exclusively on Supreme Court decision
making, as that was where authoritative policy making occurred.

More recent studies of the relationship among the courts in the judicial system have
suggested that “the work on the lower courts seems less dependent on the Supreme

3

4
Court...than [hierarchical theory] would indicate” (Richardson and Vines 1970, 144). A

second theory is advanced which compares the decision making freedom that subordinates
in the judicial hierarchy have to that of civil servants in administrative agencies (Murphy
1959). This bureaucratic model‘ likens the inefﬁciency and recalcitrance which the President
encounters in getting bureaucrats to execute his policies to the constraints imposed upon the
Supreme Court by lower courts (Gruhl 1980). Accordingly, lower court judges are depicted
as “independent actors who will not follow the lead of higher courts unless conditions are
favorable for their doing so” (Baum 1976, 91).

Some scholars remain dissatisﬁed with the eminence ascribed to the Supreme Court
by the bureaucratic model and have proposed a third theory of inter-court relationships
(Vines 1963; Richardson and Vines 1970; Howard 1981). Interaction theory2 presumes that
relationships among the different levels of courts are more complicated than either a
hierarchical or bureaucratic model would allow (Vines 1963). This approach recognizes that,
while the Supreme Court formulates policy, it does not dominate decision making on the
lower courts (Richardson and Vines 1970); and, because of the selective nature of Supreme
Court review, the potential exists for important policies to be made throughout the court
system.

In support of this emphasis on the high degree of discretion available to lower courts,

proponents of an interaction model point to the lack of accountability of subordinate judges

 

lBaum (1976) proposed a similar heuristic, which he termed an organizational model.

2This nomenclature originated with Vines (1963), and the work of other scholars has
endorsed his approach.

5

to their superiors. Federal judges are appointed for life and are rarely subject to disciplinary
measures or impeachment; likewise, state judges do not depend upon the appellate courts in
their system to maintain their positions. Lower court judges thus have a good deal of

freedom to make their own decisions and to respond to higher court decisions as they see ﬁt.

DETERMINANTS OF LOWER COURT TREATMENT OF PRECEDENT

In light of the discretion available to lower court judges, scholars have sought to
identify the circumstances under which lower court judges will defer to higher courts and
those under which they will behave independently. A number of factors have been
hypothesized to inﬂuence the extent to which lower courts comply with the mandates of
higher courts. One limitation on the adherence of lower courts to precedent is that there is
no systematic method through which lower courts are informed of higher court decisions
(Wasby 1970, 47; Baum 1976; Johnson and Canon 1984, 54-56). In addition, a higher court
decision may be unclear or ambiguous, due to the complexity of the subject matter, the
majority opinion writer’s accommodation of the views of several judges, or the publication
of a number of separate opinions (Wasby 1970, 44-46; Johnson and Canon 1984, 48-59).
There is also evidence, both systematic and anecdotal, suggesting that judges do not wish to
see their decisions reversed (Murphy 1959; Schick 1970, 145; Baum 1978). Thus the
likelihood of reversal will be a consideration in the treatment of precedent. And, of course,
judges have their own interests and policy goals which they may wish to see advanced.
These, too, may inﬂuence the extent to which they follow a higher court decision (Baum

1978; Johnson and Canon 1984, 68-70).

6

As in any hierarchical organization, the authority that lower courts ascribe to higher
courts provides an important reason to adopt the higher court policy, where “‘authority’
refers to a belief by subordinates that they have an obligation to implement directives handed
down ﬁom above” (Baum 1978, 511). Judges frequently express their acceptance of the
authority of the Supreme Court.3 Such behavior may be attributed to a widespread norm

within the legal community of deference to the decisions of higher courts. Judges are said

to have assimilated this norm of deference to higher courts through their socialization in the
legal profession (Howard 1981, 121-124; Johnson and Canon 1984, 35-37; Baum 1994).
Judges themselves may beneﬁt from the principle of hierarchical authority in the legal
system, as it lends legitimacy to their decisions (Johnson and Canon 1984, 37-3 8).

Judges are thought to be further motivated to adhere to stare decisis by a reluctance
to see their decisions reversed (Carninker 1994). Reversal of a judge’s decision may be
viewed as failure to interpret the law correctly and may negatively affect perceptions of the
judge’s effectiveness within the legal community (Baum 1978). As one scholar stated,
“Judges, no more than other men, enjoy the prospect of public correction and reprimand”
(Murphy 1959, 1030). Some studies have described lower court judges as attempting to
anticipate the reaction of the Supreme Court to their decisions and adjusting their behavior
accordingly in order to avoid reversal (Murphy 1959; Schick 1970, 148-150).

However, reversal has more symbolic than practical effects (Baum 1978). Reversal

is not always a meaningful sanction for deviation from a higher court’s ruling because review

 

3As one example, note Judge Learned Hand’s statement: “I have always felt that it
was the duty of an inferior court to suppress its own opinions and...to try to prophesy what
the appellate courts would do” (Quoted in Schick 1970, 167).

7

by the higher court is not automatic. This is especially true for Supreme Court review of
lower court decisions. The Supreme Court exercises very little direct supervision over lower
courts (Howard 1981, 57).4 Even though the number of requests for Supreme Court review
of lower court decisions has increased dramatically over the past several years, the number
of cases accepted by the Court has declined.5 The Court formally decides only about two
percent of the cases in which petitions for review are ﬁled (Carp and Stidham 1996, 132).
Individuals in any organization will be willing to adopt policies with which they
agree, while strong disagreement with a particular policy provides incentive to resist it
(Baum 1978). Most studies of lower court treatment of higher court decisions have
recognized that, in spite of their positions as subordinates in the judicial hierarchy, lower
court judges are inclined to evaluate higher court policy in terms of their personal policy
preferences (see, e. g., Songer, Segal, and Cameron 1994; Segal, Songer, and Cameron 1995).
Judges who are already committed to their own policies will be reluctant to accept a
dissimilar policy (Baum 1976). However, judges who are favorably predisposed toward the
higher court policy will attempt to expand its application into other policy areas and into

situations where its coverage seems ambiguous (Johnson and Canon 1984, 46-47).

 

4Murphy (1959) suggested that the possibility of Supreme Court review is greater for
federal courts than for state courts. The inferior position of federal district and appellate
courts in the hierarchy of national authority is said to make them subject to more strict
Supreme Court surveillance.

5According to data provided in Carp and Stidham (1996), the number of cases ﬁled
in the Supreme Court increased almost 43% from 1982 to 1992. The Court handed down full
opinions in approximately 150 cases per term during the 19803, but in the 1993 term it issued
only 84 full opinions.

8

Even if lower court judges are socialized to follow precedent and to avoid reversal
by a higher court, and even if they agree ideologically with a higher court ruling, they cannot
follow decisions of which they are unaware. The communication of judicial decisions
represents a potential problem in their implementation (Wasby 1970, 47; Baum 1976;
Johnson and Canon 1984, 55). There are no formal efforts made within the system to inform
other courts of a decision or to ensure that judges have copies of an opinion. Batun (1976)
reported that judges on the US. Courts of Appeals seldom read Supreme Court decisions as
they are issued. Many judges rely on the lawyers who argue cases before them or on their
law clerks to keep them abreast of decisions from higher courts.

Congruence between the policies established by higher courts and the decisions of
lower courts also depends to a great extent upon the effectiveness with which the policy is
communicated.‘5 Much of the discretion exercised by lower court judges may be a product
of the higher court decision which they are interpreting (Romans 1974; Combs 1982;
Johnson and Canon 1984, 49; Marshall 1989). One scholar has said of the Supreme Court:
“[The Court] contributes to this practice when it hands down decisions that are plainly
compromises of competing doctrines advanced by the justices constituting the majority or
which are quite ambiguous” (Schick 1970, 147). Lower court judges may resolve ambiguity
in favor of their own preferences (Peltason 1961 , 13; Carp and Rowland 1983, ch. 2).

An unclear decision may result from a lack of consensus on the higher court

regarding its opinion (Atkins 1972). While the judges may agree on whether a lower court

 

6According to a former Supreme Court justice (Douglas 1949, 739), “it is vital to the
integrity of the judicial process” that the Court’s audience “understand why it did what it
did.”

9

decision should be affirmed or reversed, they may disagree over the legal and constitutional
principles upon which the decision should rest and thus the precedent that should be
established. When a court is divided in its reasoning, one of two things may occur. In an
effort to craft an opinion to which all of the judges can agree, the language of the majority
opinion may be weakened;7 alternatively, several opinions may be written, so that each
opinion expresses the views of an individual judge and no opinion has majority support. So
while multiple opinions may cloud the expectations of the court issuing the decision,
unanimous decisions may be vague and confusing to lower courts (Johnson and Canon 1984,
49-50). Clarity may also be lacking in decisions which deal with complex issues or technical
subject matter (Wasby 1970, 250; Johnson and Canon 1984, 49; Johnson 1987).
According to one comprehensive analysis of the interpretation of judicial decisions
(Johnson and Canon 1984, 56-57), “while the clarity and communication of the policy may
affect the range of interpretations accorded to a decision, the direction of those
interpretations (positive or negative) may be substantially inﬂuenced by the perceived
legitimacy of the policy.” The level of support a decision has on the higher court may serve
as an indication to lower courts of the authority with which the decision was rendered

(Rohde 1972; Marshall 1989). A divided Court or the publication of dissenting opinions

 

7One legal commentator (Hutchinson 1979) has analyzed the role of unanimity in
inﬂuencing lower court responses to Brown v. Board of Education (1954). He reasoned that
the Court strove to present a united front in Brown in order to enhance the acceptability of
the decision, to withstand criticisms of the Court’s expansive civil liberties decisions, and
to mask uncertainty on the Court as to the precise meaning of the decision. He concluded
that “unanimity...operated in time to obscure rather than enhance the Court’s decisions in the
area [of desegregation]. Indeed, the Court’s continuing desire to be united outweighed its
responsibility to be persuasive...” (Hutchinson 1979, 87).

10

may raise questions about the legal correctness of the higher court’s decision, as a sizeable
faction of the Court favored a different outcome (Johnson and Canon 1984, 58). Lower court
judges may be reluctant to comply with policies that lack majority support since such
policies could be altered with a change in Court membership (Marshall 1989; Pacelle and
Baum 1992).

Judicial scholars have sought to operationalize some of the factors which determine
the effectiveness with which a higher court policy is communicated to the lower courts. A
variety of measures have been used in past studies to capture the ambiguity and/or
complexity of a particular Supreme Court decision and the authority of that decision in terms
of Court support. Examples of widely used measures in compliance studies include vote
margin, size of the opinion majority, number of concurring and dissenting opinions written,

and number of issues raised and legal provisions cited.8

APPROACHES TO STUDYING THE TREATMENT OF PRECEDENT
Prior to the behavioral revolution, studies of compliance with Supreme Court
decisions were virtually nonexistent; almost no attention was given to what happened after
Court decisions were rendered. Canon (1991) attributed the surge of interest in the impact
of judicial decisions which occurred during the 19505 to two factors: the behavioralists’
admonition that analysis of actual behavior should be used to explain events, and the lower
courts’ overt resistance to the Court’s decision in Brown v. Board of Education (1954).

Canon (1991) termed the period from 1960-75 the golden age of impact research due to the

 

8F or examples of studies which have incorporated these measures, see Johnson
(1979), Johnson (1987), and Pacelle and Baum (1992).

11

large number of impact studies which were conducted during this time. However, while
Wasby’s (1970) overview of judicial impact research culled over 100 hypotheses from the
literature, few of the studies that he examined were related to general theories of impact.
One of the first theories of the implementation and impact of Supreme Court
decisions was developed by Johnson and Canon (1984). The authors stressed that judicial
decisions are not self-implementing, and that courts must frequently rely on other courts or
on nonjudicial actors in the political system to turn law into action. They developed a
heuristic model of four impact populations which were made up of the actors who respond
to judicial decisions. Lower courts, charged with determining the meaning of policy and
developing rules for matters not addressed in the original decision, were the interpreting
population. This population provided “official” interpretations of Court policies, which were
applicable to the other populations under their jurisdiction (Johnson and Canon 1984, 16).
Judges may employ several strategies to avoid faithful application of higher court
decisions.9 An obvious approach is simply to ignore the decision. Another tactic is to
dispose of the case on technical or procedural grounds. A judge may ﬁnd that one of the
parties does not have standing to bring the case or that the issue is moot. In addition, a judge
may assert that portions of the higher court’s opinion are dicta rather than binding precedent.
Finally, to avoid adherence to precedent, a judge may give the precedent a narrow

application. The judge may argue that the facts of the case before him are distinguishable

 

9Johnson and Canon (1984) viewed a judge’s acceptance of higher court policy as
lying on a continuum from “refusal to accept” to “enthusiastic acceptance,” with many
reactions falling in a “zone of indifference.” Although evidence of noncompliance abounds
in early studies, outright deﬁance of higher court decisions is rare. At the other end of the
continuum, a judge who agrees with the decision of a higher court may attempt to expand
upon its intended scope.

12

from the facts of the higher court case, so that the precedent established in that case is not
controlling.

Songer observed that there are numerous differences in the way Supreme Court
impact on the decisions of lower courts has been conceptualized, and he described four such
approaches (Songer 1988a). A deﬁance approach examines whether a lower court formally
accepted or openly repudiated the authority of the Court. No systematic deﬁance studies
exist; only anecdotal evidence has been presented.10 The most frequently adopted approach
is a noncompliance one, which determines whether lower courts adhere to Supreme Court
precedent in their decisions.‘I Such an approach requires traditional legal analysis and is
therefore somewhat subjective. A legal impact approach takes a broader perspective,
recognizing that lower courts may fail to support the basic policy of the Supreme Court
without being overtly noncompliant.12 This approach requires scholars to read lower court
opinions to determine whether broad or narrow construction is given to Supreme Court
precedent. Finally, an outcomes approach focuses on general decisional trends in lower court
decision making in order to determine whether these courts are responding to changes in

Supreme Court doctrine.l3

 

10Many studies of lower court deﬁance have focused on the school desegregation
decisions. See, e.g., Peltason (1961).

“Studies which have taken a noncompliance approach include Manwaring (1968),
Tarr (1977), and Gruhl (1980).

”The work of Canon (1973) and Romans (1974) examines the application of legal
rules in lower court decision making.

l3Baum (1977) has stressed the importance of focusing on patterns of actual
outcomes. Studies that consider outcomes in the Courts of Appeals include Songer (1987)
and Songer and Sheehan (1990).

13

Songer (1988a) utilized each of these approaches in assessing state court compliance
with Miranda v. Arizona (1966), and he found signiﬁcant differences in the depictions of
Supreme Court impact provided by the different approaches. He encouraged scholars to
employ all of these approaches simultaneously to achieve an accurate assessment of Supreme

Court impact.

EMPIRICAL EVIDENCE OF LOWER COURT TREATMENT OF PRECEDENT
Few comprehensive impact studies have been conducted since Johnson and Canon’s
work, while the undertaking of more narrow analyses is evidenced by the steady production
of journal articles (Canon 1991). These projects have focused on lower court compliance in
speciﬁc issue areas or with particular decisions, and a number of them have reported high
levels of congruence between Supreme Court policy and lower court decisions.
Gruhl (1980) examined libel cases in the Courts of Appeals following New York
Times v. Sullivan (1964) to determine these courts’ application of the actual malice test
developed by the High Court in that decision. He found that these courts consistently
complied with the new precedent and, in some instances, even anticipated the direction of
Supreme Court policy. Songer and Sheehan (1990) replicated Gruhl’s (1980) study of
compliance with New York Times and compared their ﬁndings to those from a similar
analysis of the Miranda decision. For both cases, they discovered high levels of compliance.
However, when an alternative approach to impact was employed, as advocated by Songer

(1988a), their conclusions were modiﬁed. While they found signiﬁcant changes in circuit

14

court outcomes in response to Supreme Court libel policy, there was no such evidence of
shifting trends in response to decisions involving the admissibility of confessions.

Songer (1987) also utilized an outcomes approach in studying appeals court decision
making in labor and antitrust cases. As with libel policy, he reported meaningful shifts in
decisional trends in the Courts of Appeals following alterations of Supreme Court policy,
even after controls were introduced for circuit court composition. Likewise, in analyzing
district court responses to Brown v. Board of Education, Sanders (1995, 744) found that this
decision was “clearly and signiﬁcantly taken into account” in subsequent decisions in all
areas of the country.

In contrast to these examples of lower court compliance, several researchers have
recounted evidence of large-scale noncompliance with Supreme Court decisions. These
studies have typically examined school desegregation or other controversial civil liberties
decisions of the Warren Court (Segal, Songer, and Cameron 1995). In one of the ﬁrst in-
depth studies of the policy making role of lower federal courts, Peltason (1961) explored the
application of the Brown decision. He highlighted the limited acceptance of desegregation
by district and circuit court judges as they encountered the political and social forces that
impinged upon this issue. Other studies have documented widespread opposition in the
lower courts to the Warren Court’s criminal procedure jurisprudence (see, e. g., Manwaring
1968; Canon and Kolson 1971; Romans 1974). In analyzing state supreme court compliance
with the US. Supreme Court’s establishment clause decisions, Tarr (1977) also presented

evidence of refusal to follow Supreme Court precedent.

15

Some of these studies have emphasized the importance of authoritative, unambiguous
Supreme Court precedent in ensuring lower court compliance. Romans (1974) explored the
interaction of the personal preferences of judges with the clarity of Supreme Court decisions.
He found that, when presented with a vague precedent, most state courts followed their own
ideological values; when dealing with an unequivocal precedent, they followed the precedent
but refused to extend its application beyond the boundaries established by the Court. In
examining northern school desegregation cases, Combs (1980) suggested that the Courts of
Appeals had taken advantage of the discretion allowed them by the Court’s decisions in this
area. He reported that various circuits had enacted different legal standards to determine
constitutional violations. As these tests varied in terms of rigor, the choice of which test to
apply had important ramiﬁcations for the litigants.

One work (Songer, Segal, and Cameron 1994) took an entirely different approach

than the studies discussed here, explicitly testing the nature of the interaction between the
Supreme Court and the Courts of Appeals. Borrowing from economic agency theory, the
authors loosely modeled the interaction between higher and lower courts as a principal-agent
relationship. They looked for evidence of such behaviors as congruence, responsiveness, and
shirking, and they applied fact-pattem models developed for the Supreme Court’s search and
seizure decisions to such decisions in the Courts of Appeals in order to test their hypotheses.
The authors reported a high level of responsiveness in the Courts of Appeals to Supreme
Court search and seizure policy. At the same time, the ideologies of the judges motivated

them to ﬁnd opportunities to “Shirk” their responsibilities and pursue their own policy goals.

16

A number of scholars have lamented the dearth of comprehensive models of lower
court responses to Supreme Court decisions (Baum 1978; Songer 1987; Reid 1988; Songer,
Segal, and Cameron 1994). The study presented here is an effort to remedy this shortcoming
with respect to the Courts of Appeals. Before discussing the different approaches that I take
in addressing this question, however, I provide an overview of decision making in these

courts.

DECISION MAKING ON THE COURTS OF APPEALS

The Courts of Appeals were created by the Evarts Act of 1891 as the intermediate
courts of the federal judicial system, located between the federal district courts and the
Supreme Court. In the federal system, the district courts have original jurisdiction in most
cases, the Courts of Appeals have initial appellate review,” and the Supreme Court has ﬁnal,
discretionary review. The Courts of Appeals were created to alleviate the ever-increasing
workload of the Supreme Court and to permit the High Court to concentrate on politically
signiﬁcant or controversial cases. The Judiciary Act of 1925 provided the Supreme Court
with even greater control over its docket, so that the Courts of Appeals became the ﬁnal
arbiter in a substantial proportion of federal cases. According to data presented in the
Supreme Court Compendium (Epstein et. al. 1996), the Court granted review to 20% of the
petitions for review in the period immediately following the passage of the 1925 Act; this

number fell to 11% in the 19608 and to 4% in 1995.

 

1“The Courts of Appeals also have original jurisdiction to review the decisions of
certain administrative agencies.

17

There are currently twelve regional Courts of Appeals, referred to colloquially as
circuit courts. Most appeals court cases are decided by three-judge panels rather than all of
the judges in the circuit. Each circuit determines its own panel assignment procedures, but
the process of selection is random.” The composition of appellate panels changes frequently
so that the same judges do not sit together permanently. To deal with a heavy caseload, the
chief judge of the circuit may assign other federal judges to decision making panels. These
judges may include retired Supreme Court justices, senior (serni-retired) circuit judges,
judges from other circuits, and senior and active district judges. Because of the manner in
which appellate panels are formed, the decision of a given panel will not necessarily reﬂect
the views of a majority of the judges in the circuit.

Occasionally, a case may be heard en banc, where all of the circuit’s active judges
participate in the decision.16 En banc proceedings may be granted for both administrative
and political reasons (Atkins 1972). The Uniform Rules for Appellate Procedure stipulate
that an en banc hearing is “not favored” and should be granted only when consideration by
the full court is necessary to maintain decisional uniformity and when the case involves a
question of “exceptional importance.” Some scholars have suggested that en banc
proceedings send a signal of unity to the Supreme Court and minimize the likelihood of

Supreme Court review (Atkins 1972; Richardson and Vines 1970, 125).

 

”In the past, panel assignments were made by the chief judge of the circuit, a
procedure which, as will be discussed in subsequent chapters, was the subject of some
criticism.

1"’A senior circuit judge who was a member of the original panel, if one was
convened, will also participate in the en banc decision.

18

The Courts of Appeals hear cases which involve both routine and highly controversial
questions. Most decision making consists of review of lower court outcomes under highly
deferential standards of review (George 1997), which is evidenced by the fact that these
courts reverse only about 20% of the cases they decide (Davis and Songer 1988-9). In
addition to resolving these ordinary matters, however, the circuit courts also hear a number
of cases which deal with major issues of public policy and which evoke strong disagreement.
They have an opportunity to sort out and develop the legal issues in such cases, and to shape
those claims which they consider to be worthy of Supreme Court review (Carp and Stidham
1996, 41). In analyzing decision making in the Courts of Appeals, then, it is crucial to
determine the extent to which judges have opportunities to behave independently. As Songer
(1991, 41) reminded students of judicial behavior, “Judges cannot be said to be signiﬁcant
policy makers unless they have substantial discretion to choose among a range of options
when deciding their cases.”

Howard (1981) has conducted the most systematic examination to date of the
motivations of circuit court judges. He interviewed thirty-ﬁve judges from three circuits and
questioned them as to their views of the permissible range of lawmaking discretion (Howard
1981, 164). While almost all of the judges (32 of 35) reported that “clear and relevant”
precedent was “very important” in reaching their decisions, most also agreed that lawmaking
was inherent in the judicial role. F orty-eight percent considered their personal views on
justice to be “very important,” and another 40% said such values were “moderately
important.”

While Howard (1981) estimated that appeals court judges were able to act according

to their personal preferences in only 10% of their cases, Songer (1982) reported that these

19

judges were free to pursue their policy goals in one-ﬁfth to one-third of their cases. He
analyzed unanimous decisions in the Courts of Appeals to determine what proportion of
these decisions were truly consensual in that their outcomes were determined by precedent
or other constraints. He concluded that consensual decisions in unanimous cases were far
less prevalent than most studies have assumed, and that the outcomes of a substantial
proportion of unanimous decisions were inﬂuenced by the ideological perspectives of the
judges. Songer (1982) suggested that circuit court judges feel free to avoid the application
of precedent in a much larger subset of their decisions than they are willing to admit.

It appears that judges on the Courts of Appeals are able to consider their own
attitudes and values in at least a substantial minority of their decisions. A systematic
examination of the Situations in which these judges perceive that they have such latitude, and

the implications that this independence has for judicial policy, is thus warranted.

CHAPTER 2

RESEARCH DESIGN AND METHODOLOGY

Two models of judicial behavior are offered by students of the courts. One model
takes a legal perspective, asserting that judges make decisions based upon legal factors such
as the intent of the Constitution’s framers and precedent. A second model provides that

judges act in accordance with their personal values and policy preferences. There has been

a great deal of research regarding whether a legal or attitudinal model applies to the Supreme
Court, but it is much less clear which model, if either, pertains to the US. Courts of Appeals.
There have also been numerous studies conducted which seek to explain lower court
compliance with Supreme Court precedent. These studies are limited, however, in that they
consider only one, or a few, landmark decisions and they fail to provide systematic tests of
the explanations that are offered. I address these weaknesses here by specifying multivariate
models of the Courts of Appeals treatment of precedent across a broad spectrum of issues.
I also assess the degree of inter- and intra- circuit conﬂict in the application of Supreme
Court precedent.

I utilize Shepard ’s United States Supreme Court Citations to measure the adherence
of the Courts of Appeals to Supreme Court precedent. I code a Court of Appeals case which
Shepard ’s indicates “followed” or “harmonized” a Supreme Court decision as followed,

while I code a case which “distinguished” or “limited” a Supreme Court decision as not

20

21

followed. As for decisions which are classiﬁed as “questioned” or “criticized,” I examine
the actual opinion to determine the treatment the lower court provided the precedent.

I expect that both legal and attitudinal factors will inﬂuence the circuit courts’
treatment of Supreme Court decisions. I include as legal variables measures of the clarity
of the precedent set by a Supreme Court decision and the support on the Court for its
decision. I also identify those cases that are highly signiﬁcant in terms of their policy
implications, as the probability of Supreme Court review should be greater in such cases.
My attitudinal measures are the percentage of the appellate panel that is in line ideologically
with the direction of the Supreme Court decision and the ideological change in the
composition of the Supreme Court Since a decision was handed down. In addition, since
some types of cases are more likely than others to generate intense political beliefs among
judges, I assess differences among issue areas in the factors which inﬂuence the adherence
of circuit court judges to precedent. I also examine whether Supreme Court precedents grow
more persuasive with the passage of time or whether the authority of these decisions is
eroded by subsequent Court rulings.

I take three different approaches to answering the question regarding the determinants
of appeals court behavior. First, I examine four major Supreme Court decisions (Chapter 3).
I study these cases in order to derive measures which capture accurately the strength and
clarity of the precedent established by a particular Supreme Court decision, and thus the
amount of discretion that lower court judges perceive they have. Once I have developed
these measures, I analyze the Courts of Appeals’ treatment of these four decisions in light
of these factors. Second, in order to gain insight into decision making on the Courts of

Appeals in a wide range of contemporary issues, and to test the utility of my measures, I

22

analyze the treatment of all Supreme Court decisions of a particular term (Chapter 4). I
include a variable for the salience of the Supreme Court case in order to assess whether lower
court judges behave differently in cases which have important political or social
consequences. Third, I analyze the circuit courts’ application of civil rights and liberties
decisions of the Supreme Court from a ten-year period (Chapter 5). This approach enables
me to assess variation in the adherence to a particular precedent over time, and to examine

differences in the treatment of precedent among, and within, the circuits.

SHEPARD ’3 UNITED STA TES SUPREME COURT CITA TIONS

This study takes what Songer (1988a) would term a noncompliance approach, in that
it is an examination of whether “lower courts consistently exhibit a ﬁdelity to precedent in
their decisions” (Songer 1988a, 426). Songer asserted that a determination of whether a
given decision is in compliance requires traditional legal analysis, but an alternative to
traditional legal analysis is the use of Shepard ’s Citations. Shepard ’s is a legal research tool
which provides a listing of published opinions in which a particular decision is cited and
categorizes the substantive treatment of each decision. It is utilized by members of the legal
community on a daily basis to ascertain the status of existing precedents. Though designed
for use in the legal profession, Shepard ’s has also been widely employed by judicial behavior
scholars, primarily to identify cases for analysis (see, e. g., Romans 1974; Gruhl 1980; Combs
1982; Johnson 1987; Songer and Sheehan 1990; Pacelle and Baum 1992). Some researchers
have looked to Shepard ’s for purposes similar to the research objective here, i.e., to measure
the deference afforded Supreme Court precedent by the Courts of Appeals (see, e. g., Johnson

1979)

23

Songer (1988b) cautioned researchers in the use of Shepard ’s for this purpose, as well
as for the selection of cases for judicial impact studies. Shepard ’s covers only published
opinions‘ and, according to Songer (1988b), will list only those cases for which full citations
or case names are provided in the opinions. Songer (1988b) illustrated this assertion,
suggesting that a lower court opinion referring to “the Miranda decision” or the rule of law
announced in Miranda would not be included in the Shepard ’s compilation for Miranda v.
Arizona, as an explicit citation is not given. However, in my reading of a number of Courts
of Appeals decisions in conducting this study, I found no such abbreviated references to
decisions which were not cited fully elsewhere in the text of the opinion.

Songer’s (1988b) caveat regarding the use of Shepard ’s which has the most serious
implications for this analysis is that lower court decisions which ignore precedent, whether
intentionally or unintentionally, will not be included. He recognized that noncompliance is
not limited to situations in which a court announces that it is not bound by a Supreme Court
decision, but that it may also occur when a lower court “ignores a ruling of the High Court
which should be a controlling precedent” (Songer 1988b, 570). We evaluated this concern
in coding the treatment of decisions from the 1987 Term (Hurwitz and Reddick 1996) by

examining those appeals court dissenting opinions in which a Supreme Court decision was

 

1Shepard ’s inclusion of only published opinions is problematic. The rules
regulating publication stipulate that unpublished opinions should be characterized by the
“straightforward application of clear, binding precedent” (Songer 1991, 43); however,
some research suggests that descriptions of appeals court decision making which are based
solely upon an analysis of published opinions represent an inaccurate picture of the
behavior of these courts (Davis and Songer 1988-9). Thus any conclusions which may be
drawn from this study should be limited appropriately.

24

cited.2 Our goal was to ascertain whether judges in the minority were calling to task their
colleagues in the majority for ignoring or avoiding an applicable precedent. We found that
such behavior occurred only rarely,3 and that inclusion in the analysis of those cases where
it did occur had only a slight impact on the results. Appendix A provides a comparison of
these results. I concluded on the basis of this examination of decisions from the 1987 Term
that coding circuit court dissenting opinions which cited Supreme Court decisions was
unwarranted for future analyses.

A second reason that Songer’s (1988b) concerns may be unjustiﬁed regarding
Shepard ’s inability to account for the avoidance of precedent resides in our system of legal
advocacy and the nature of legal argument. One legal commentator provides the following
description of the composition of judicial opinions:

[A] well-crafted opinion usually contains numerous citations
to past cases that build and interlock with one another toward
justifying an overall conclusion. Two lawyers representing
opposite sides of a dispute may thus cite two different lines of
precedents. A judge then evaluates the two proposed lines of
[precedents] to arrive at a ﬁnal decision (Heiner 1986, 228).

This quotation indicates that it is a judge’s responsibility to listen to both Sides of an issue

and consider all relevant case law, and to justify his ultimate decision by indicating why

 

2Johnson (1979, 795) dealt with this concern by making the “major assumption that
intentional and unintentional omissions are randomly distributed across all Supreme Court
decisions.” While such a judgment may be appropriate for unintentional omissions (i.e. ,
clerical errors), we were not comfortable with presupposing the randomness of intentional
omissions. We conducted this examination of citations in dissenting opinions to ensure the
validity of our measurement.

3The decisions of the 1987 Term were cited in over 300 dissenting opinions. Of
these 300+ citations, only 19 ﬁt the criteria for inclusion in our analysis.

25

particular precedents are controlling and why others are inapposite.4 If a Supreme Court
precedent is truly relevant to the case at bar, attorneys on one Side or the other will rely on
that precedent in making their argument, and an appeals court judge will be obliged to
address the applicability of that precedent in his opinion. Thus it is unlikely that a judge who
is attentive to his reputation in the legal community will wholly ignore a precedent that is on
point.5

As I have stated, I code a Court of Appeals case which Shepard ’3 indicates
“followed” or “harmonized” a Supreme Court decision as followed. In the context of this
study, a Shepard ’s classiﬁcation of “followed” signiﬁes that an appeals court referred to the
Supreme Court decision as controlling, and a classiﬁcation of “harmonized” denotes that an
apparent inconsistency between an appeals court case and a Supreme Court case is explained
and shown not to exist. I code a case which “distinguished” or “limited” a Supreme Court
decision as not followed. An appeals court which distinguishes a Supreme Court ruling ﬁnds

the case before it to be different in law or fact from the Supreme Court decision, and a

 

4As Sanders (1995, 733) points out, “cases to which clear precedents apply will be
unlikely to move on to the appellate level.”

5Johnson and Canon (1984, 35) describe a judge’s consideration of his perception
among his colleagues:

[J]udges become concerned about their evaluation as
professionals -- about the respect of lawyers and other
judges with whom they interact. Moreover, when the
decisions are published, the judge knows that his or her
interpretation of relevant precedents will be subject to
scrutiny from distant fellow judges as well as from lawyers
and commentators in legal journals around the nation.

26

limiting circuit court decision refuses to extend the holding of the Supreme Court ruling
beyond the precise issues involved.

Decisions which are classiﬁed as “questioned” indicate that the soundness of the
reasoning in the Supreme Court decision is questioned, while those which are categorized
as “criticized” denote that the Supreme Court decision disagrees with the appellate court’s
reasoning. I examine the Court of Appeals opinion to code the treatment of Supreme Court
decisions which are “criticized” or “questioned.” I do this to determine the actual treatment
of the precedent, as a decision may question or criticize a precedent but still follow that
precedent.6

This utilization of Shepard ’s categorizations is justiﬁed in the compliance literature.
Judges who support a precedent on the basis of their personal ideology may apply that
precedent to circumstances where its relevance is questionable, or they may apply it to an
entirely different policy area (Johnson and Canon 1984). At the other end of the compliance
Spectrum, a common technique employed by lower court judges to avoid the application of
a decision with which they disagree is distinguishing that precedent, emphasizing factual
differences between the decision of the higher court and the case at bar (Douglas 1949; Baum
1978; Johnson and Canon 1984; Songer and Sheehan 1990; Caminker 1994). Another 0&-
used method of avoidance is to give a higher court decision a narrow, Situational

interpretation (Murphy 1959; Johnson and Canon 1984).

 

“Wasby (1970, 29) notes that “compliance may exist even when disrespect for the
law or the Court being expressed.”

27

COURTS OF APPEALS’ TREATMENT OF FOUR SUPREME COURT DECISIONS:
AN EXPLORATION

Here I utilize Shepard ’s Citations to analyze the Courts of Appeals’ treatment of four
major Supreme Court decisions: US. v. Wade, Griggs v. Duke Power Co., Lemon v.
Kurtzman, and Terry v. Ohio. The rationale for studying these highly visible cases is
twofold. First, lower court judges should be aware of existing precedent in these cases; and,
second, mistreatment of precedent in these cases should come to the attention of the Supreme
Court. Thus circuit court judges should be particularly conscious of the amount of discretion
they have in applying these decisions.

I explore two cases where the ratio decedendi is deﬁnitive and unambiguous (US.
v. Wade and Griggs v. Duke Power Co.) and two cases where the Court’s application of law
to facts is less straightforward (Terry v. Ohio and Lemon v. Kurtzman). I examine each
decision to determine the aspects that contribute to the effectiveness of its communication,
and I assess the extent to which each decision is followed in the Courts of Appeals. A
secondary goal in analyzing these cases is to derive measures which capture accurately the
strength and clarity of the precedent established by a particular Supreme Court decision.
Descriptive analyses such as this are useful in developing hypotheses to be tested in
subsequent, more comprehensive studies.

I study these particular cases for several reasons. These cases are included in
Congressional Quarterly’s listing of important decisions for each term of the Supreme Court,
which is found in CQ’S Guide to the United States Supreme Court. With the exception of
US. v. Wade, these cases are also identiﬁed in the Supreme Court Compendium (Epstein et.

al. 1996), compiled by noted scholars of the Court, as landmark decisions in their respective

28
policy areas.7 Cases having important policy implications provide interesting subjects for
compliance research. As Baum (1978, 215) noted, the policies that “elicit widespread non-
adoption are likely to be signiﬁcant ones, as minor policies generally lack the capacity to
motivate strong opposition.” An alternative viewpoint is that cases with widespread
ramiﬁcations will receive more favorable treatment in the lower courts, as divergent
decisions in such cases are more likely to be reviewed.

Focusing on these salient cases also allows controlling for other elements which may
inﬂuence the amount of discretion lower court judges exercise, and isolating the factors
which I am interested in exploring here -- the effects of the strength and clarity of Supreme
Court precedent on lower court decisions. The selection of highly visible cases is likely to
overcome the problem of the Courts of Appeals being unaware of the Supreme Court’s
decision. These are also cases in which Supreme Court justices are likely to be concerned
about the interpretations of their policies; thus reversal should be a more signiﬁcant
consideration in a lower court judge’s decision making calculus. Finally, if lower court
judges are motivated by their own policy preferences, these are the types of cases in which

they Should have policy goals which they wish to advance.

COURTS OF APPEALS’ TREATMENT OF THE DECISIONS OF THE 1987 TERM:
A MULTIVARIATE ANALYSIS

In testing the utility of legal and attitudinal models of the Courts of Appeals’

treatment of Supreme Court precedent, I begin by examining the decisions of the Supreme

 

7The authors recognized that any list of the Court’s landmark decisions is a subjective
venture. They emphasized in their listing topics and cases of current interest.

29

Court’s 1987 term. Such an approach permits the development of a general explanation of
the circuit courts’ allegiance, or lack thereof, to the High Court in the wide range of cases
which come before them. The model that is speciﬁed incorporates aspects of both the legal
and attitudinal accounts of judicial behavior, in order to ascertain which is more instructive
in understanding decision making on the Courts of Appeals. Speciﬁcally, variables are
included in the model which capture the strength and clarity of the Supreme Court’s decision,
the policy signiﬁcance of that decision, the ideological composition of the appellate panel
which is asked to apply the decision, and the change in the composition of the Supreme
Court since its decision was handed down.

To test hypotheses regarding the legal and attitudinal motivations for circuit court
judges’ behavior, I utilize a dichotomous dependent variable that identiﬁes whether the
Courts of Appeals followed or did not follow legal precedent established by the Supreme
Court in its decisions of the 1987 Term.8 Using Shepard ’s United States Supreme Court
Citations, I ascertain the treatment accorded these decisions by the circuit courts in cases in
which the decisions have been applied, so that the writ of analysis is the circuit court

decision. The dependent variable is regressed on a number of legal and attitudinal variables.

 

8The 1987 Term was selected for analysis for reasons that do not reﬂect on the
substance of the decisions. First, there is a sufficient time period following this term in
which to trace the Courts of Appeals’ treatment of the decisions. Second, there is variation
in the composition of the circuit courts during this period, as the courts were staffed by both
Carter-Clinton and Reagan-Bush appointees. Finally, the Supreme Court during this term
was ideologically balanced, as it was composed of Associate Justices Marshall, Brennan,
White, Blackmun, Stevens, O’Connor, and Scalia and Chief Justice Rehnquist; Justice
Kennedy joined the Court in early 1988. Although proponents of a legal model would argue
that the composition of a court has no bearing on the decisions that are reached, it would be
imprudent to “stack the deck” in favor of an attitudinal explanation.

30

Because the dependent variable is dichotomous, a logistic regression model is estimated
(Aldrich and Nelson 1984; King 1989).

I include in the analysis only those cases which were orally argued before the Court
(DEC_TYPES l, 6, and 7 in the Spaeth Supreme Court Judicial Database). Memorandum
decisions, decrees, and cases decided by an equally divided vote are excluded, so that I
trace the treatment of 127 Supreme Court decisions from the 1987 Term. The resulting
data set includes 1467 instances in which an appeals court decision treats a Supreme Court

decision.9

AS stated, I predict that both legal and attitudinal factors will inﬂuence the circuit
courts’ treatment of Supreme Court decisions. The following is a description of the

operationalization of these factors. All variables regarding characteristics of Supreme Court

decisions were created from data contained in the Spaeth Supreme Court Judicial Database.
With the assistance of my co-author in a previous version of this analysis (Hurwitz and
Reddick 1996), I collected all data relating to Courts of Appeals decisions.
Legal Factors

Vote Margin. The greater the vote margin in the Supreme Court decision, the higher
the level of agreement among the justices as to the disposition of the case. I hypothesize that
precedent established in cases where the vote margin is large should appear more
authoritative to lower courts, while divisiveness on the Court may indicate a weak precedent

(Wasby 1970).

 

9Similar treatment of the same Supreme Court decision within the same appeals
court decision is counted only once.

31

The inﬂuence of the degree of Supreme Court consensus on the circuit courts’
behavior is likely to be greatest at the extremes. I create two variables ﬁom the vote margin:
Unanimous, where all of the justices agreed with the disposition of the case (e.g., a 9-0
decision); and One- Vote Margin, where only one vote separated the majority and minority
voting blocs (e.g., a 5-4 decision). The circuit courts should be less likely to follow a
Supreme Court decision in which a bare majority of the justices agreed on the outcome,
while they should be more likely to follow a decision supported by all of the justices.

Concurring Opinions. The more concurring opinions the justices write, the more
ambiguous the decision will appear to lower court judges. I expect that unclear decisions

will not be followed consistently in the lower courts. Johnson (1979) relied upon the

number of dissenting opinions while he disregarded concurring opinions. However, as it
is concurring, rather than dissenting, opinions which should weaken the majority opinion,

I examine the number of concurrences in this study, and I count both regular and special
concurrences. While the Spaeth Database differentiates between regular and special

concurrences according to whether the justices indicate that they are “concurring” or
“concurring in the judgment,” I suggest that this is a distinction of semantics rather than

substance. '0

 

10I cite the following examples in support of this assertion:

In the landmark case Terry v. Ohio, 392 US. 1 (1968), Justice Harlan concurred
in the opinion of the Court. He began his concurring opinion by stating, “While I agree
with the Court’s ultimate holding in this case, I am constrained to ﬁll in a few gaps, as I
see them, in its opinion.” Id. at 31. Justice Black also wrote a concurring opinion, though
only one sentence in length, where he “concurred in the judgment and the opinion except
where the opinion quotes from and relies upon [the] Court’s opinion in Katz v. United
States and the concurring opinion in Warden v. Hayden.” Id. at 31. In the Spaeth

32

Complexity. A clear precedent will be difﬁcult to achieve in cases that interpret many
legal provisions and that involve multiple issues. I expect that appeals court judges will be
less likely to follow decisions that deal with complex subject matter. As an indicator of
complexity, I count the number of legal provisions relied upon and the number of additional
issues raised in each Supreme Court case (Johnson 1987). I then calculate the mean number
of legal provisions and additional issues per case. Any case where the number of legal
provisions and additional issues is 1.65 standard deviations above the mean is coded as a
complex case.ll

Legal Provisions. Constitutional provisions by nature are somewhat ambiguous and
open to many different interpretations. Accordingly, I expect that Supreme Court decisions

which apply constitutional provisions will be vague as well and that, as a result, such

 

Database, Justice Black’s concurrence is coded as “Special,” while Justice Harlan’s
concurrence is coded as “regular.”

In another oft-cited case, US. v. Wade, 388 US. 218 (1967), Chief Justice Warren
and Justice Douglas joined the opinion of Justice Fortas, in which Justice Fortas concurred
in part and dissented in part. The opinion expressed disagreement with Part I of the
Court’s opinion, which held that compelling a criminal defendant to utter certain words at
a lineup did not violate his privilege against self-incrimination. In the Spaeth Database,
this concurrence is classiﬁed as “regular.”

These examples serve to illustrate my point. The “regular” concurrences that I
have highlighted here raise just those issues that might foster ambiguity or confusion in the
decision making calculus of lower courts. They might also serve as bases for avoiding
application of these decisions. For these reasons, making a distinction in this study
between regular and special concurrences is not warranted.

111.65 is the critical value for a one-tailed test of statistical Signiﬁcance at a=0.05.
The mean number of legal provisions and additional issues per case was 1.28, with a
standard deviation of 0.81. Thus any case which dealt with three or more legal provisions
and additional issues was coded as complex. Nine of the 127 Supreme Court cases in this
study were coded as complex cases.

33

decisions will receive disparate treatment in the lower courts. I include a dummy variable
for whether the Court relied on a constitutional provision in reaching it decision.

Salience. In cases that have important policy implications, avoidance of precedent
on the part of lower courts is more likely to come to the attention of the Supreme Court. If
lower court judges are reluctant to be reversed on appeal, they Should be more likely to
follow precedent in highly salient cases. Since salience is deﬁned here in terms of policy
implications, an appropriate measure of salience may be derived from the number of amicus
briefs that are ﬁled in the case by parties who have a stake in the outcome.'2 I count the
number of amicus briefs that were ﬁled in each case. Any case where the number of amicus
briefs ﬁled is 1.65 standard deviations above the mean is coded as a salient case.l3
Attitudinal Factors

Ideological Consistency. I expect that when the personal policy preferences of the
members of an appellate panel are in accord with a Supreme Court decision, the panel is

more likely to follow that decision. Ideological consistency is measured as the percentage

 

12Maltzman and Wahlbeck (1996) developed this measure of case importance. I
originally considered as an indicator of salience the measure recommended by Cook (1993)
and Brenner and Spaeth (1995). These authors utilized Congressional Quarterly’s listing
of important decisions for each term of the Court, which is found in CQ’S Guide to the
United States Supreme Court. Of the 127 Supreme Court cases traced in this data set, only
four were dubbed important by CQ. These cases were treated less than 20 times by the
Courts of Appeals. As there are almost 1,500 Courts of Appeals cases in the data set, the
number of salient cases was too minuscule to capture any statistical effects. In addition,
the indicator used here better measures the importance of a decision in terms of political
and social policy implications.

13The mean number of amicus briefs ﬁled per case was 3.55, with a Standard
deviation of 3 .78. Thus any case in which ten or more arrricus briefs were ﬁled was coded
as salient. Twelve of the 127 Supreme Court cases in this study were coded as salient
cases.

34

of the appellate panel that is in line attitudinally with the direction of the Supreme Court
decision. In order to obtain an independent measure of the ideology of circuit court judges,
I utilize the party of the appointing president, which has been Shown to be a reliable measure
of judicial ideology (Tate 1981; Carp and Rowland 1983; Tate and Handberg 1991).

To illustrate, where two of three judges on a circuit court panel were appointed by
President Carter, while the third was appointed by President Bush, and the Supreme Court
issued a liberal decision, ideological consistency of the panel for that particular decision is
67%. Conversely, if that same panel treated a conservative Supreme Court decision, then
ideological consistency is 33%. By coding the makeup of every Court of Appeals panel in
this manner, I am able to incorporate en banc decisions. Such a measure also captures more
information than would a dummy variable for majority consistency or inconsistency.

Issue Areas. In their comprehensive analysis of Supreme Court decision making,
Segal and Spaeth (1993) ﬁnd that justices’ attitudes, as measured by past voting behavior,
are powerful predictors of subsequent voting behavior in civil rights and liberties and
criminal procedure cases. They found, additionally, that attitudes are less successful in
predicting the justices’ votes in economic, judicial power, and federalism cases.

Thus, I expect that the inﬂuence of ideology will be moderated by the types of issues
involved in the cases. According to the issue areas identiﬁed in the Supreme Court Judicial

Database, I group the cases into three broad categories: Civil Rights and Liberties,”

 

1"This category includes civil rights, First Amendment, due process, and privacy
cases.

35

Criminal Procedure, and Economic Activity.15 I include three multiplicative terms in the
model, where each issue category is interacted with Ideological Consistency, in order to
assess the impact of ideology in each of these areas.

Change in Supreme Court Composition. Appeals court judges may also be
inﬂuenced in their decision making by the ideological composition of the Supreme Court.
If lower court judges do, in fact, consider the likelihood of reversal, they should be attentive
to changes that have occurred in the makeup of the Supreme Court since a particular
precedent was established. My argument is that lower court judges will feel less constrained
to follow a given decision when the ideology of the Court has moved away from that
decision.

I use the Segal and Cover (1989) scores as measures of the ideological values of
Supreme Court justices. In order to derive attitudinal measures from sources independent
of votes, Segal and Cover analyzed the content of newspaper editorials that were published
between the nomination and conﬁrmation of each justice. These scores have been updated
and backdated to include the nominees of Franklin Roosevelt through Clinton (Segal et. al.
1995)

To determine the ideological makeup of a given Supreme Court, I calculate the mean
of the ideology scores of the justices on that Court. I subtract the mean of the Court that
issued a decision from the mean of the Court at the time the appeals court treats the decision,

to obtain a measure of the change in the ideological composition of the Court. I then

 

15Thirty-three of the Supreme Court cases I examine do not fall into any of these
categories.

36

combine this measure with the direction of the Supreme Court decision so that the ﬁndings
may be interpreted consistently. To illustrate, if the composition of the Supreme Court has
become more conservative and the Courts of Appeals are applying a conservative decision,
the direction of this measure is positive. If the Court has become more liberal and the Courts
of Appeals are applying a conservative decision, the direction of this measure is negative.
The reverse is true when an appeals court is applying a liberal decision.

The estimation of a logistic regression model which includes these legal and
attitudinal variables will shed light on the extent to which each variable inﬂuences the

behavior of circuit court judges in the treatment of precedent.

COURTS OF APPEALS’ TREATMENT OF CIVIL RIGHTS AND LIBERTIES DECISIONS
FROM 1966-1975: A MULTIVARIATE ANALYSIS
AND A COMPARISON OF TREATMENT AMONG AND WITHIN CIRCUITS
The analysis performed in this chapter is similar to that of Chapter 4, in that I
examine the role of legal and attitudinal factors in inﬂuencing circuit court treatment of
precedent. However, rather than focusing on the decisions of a single term, a model is
speciﬁed which explains the treatment of the Supreme Court’s civil rights and liberties
decisions from 1966-75."3 Civil rights cases are among the most controversial cases that
courts hear and are decisions that may potentially impact a substantial number of citizens.

Thus it iS important to study the interpretations afforded these decisions by the actors

responsible for giving them effect.

 

1"This time period was selected for two reasons. First, it encompasses decisions
of both the Warren and Burger Courts; and second, it allows an adequate span following
the decisions to assess variance in the adherence to a particular precedent over time.

37

Research on the Supreme Court suggests that different considerations guide judicial
decisions in cases involving civil rights and liberties issues. For example, Segal and Spaeth
(1993, 225-6) found that measures of the justices’ ideological attitudes and values were
highly successﬁrl predictors of their votes in civil rights and liberties and criminal procedure
cases, while personal preferences were not as explanatory in economic, judicial power, and
federalism cases. A comparison of the civil rights ﬁndings with those of the 1987 Term will
indicate whether attitudinal motivations dominate appeals court decision making in civil
rights and liberties cases to the exclusion of legal considerations.

I conﬁne this analysis to the Court’s formally decided, full opinion cases
(DEC_TYPES 1 and 7 in the Supreme Court Database), and I include civil rights, First
Amendment, due process, and privacy cases (VALUES 2, 3, 4, and 5) in the category of civil
rights and liberties. I also restrict this study to cases which were appealed from the circuit
courts. Thus I trace the treatment of 163 civil rights and liberties decisions of the Supreme
Court. The resulting data set includes 5664 instances in which an appeals court decision
treats a Supreme Court decision.

The variables that are included in the civil rights model are operationalized in the
same manner as in the model of the 1987 Term, with a few exceptions. As the issues raised
and the legal provisions relied upon in these cases are substantially similar, a measure of
Complexity which incorporates these factors is inappropriate. In addition, an alternative
measure of Salience is more suitable for this analysis. In civil rights and liberties cases, a
large number of amicus briefs are likely to be ﬁled, as these cases may have widespread
policy implications; thus the ﬁling of amicus briefs does not provide an adequate basis for

differentiating these cases. Instead, Congressional Quarterly’s listing of the important

38

decisions of each term, provided in CQ’S Guide to the US. Supreme Court, is utilized (Cook
1993; Brenner and Spaeth 1995).17

Two variables relating to the strength of Supreme Court precedent are measured more
precisely in the civil rights model. Rather than simply indicating whether there were dissents
to a particular decision, as in the model of the 1987 Term, the Unanimous variable utilized
here captures those cases in which no dissenting votes were cast and no concurring opinions
were written. Precedents for which both outcome and reasoning had the approbation of all
of the justices are thus coded as unanimous. The One- Vote Margin variable in the model of
the 1987 Term is intended to distinguish those cases decided by the narrowest possible
margin. According to the rules of Supreme Court procedure, cases in which a lower court
decision is reversed by a vote of ﬁve to three are also decided by a minimum winning
coalition. The Minimum Winning Coalition variable in the civil rights and liberties model
measures 5-4 and 4-3 decisions, as well as 5-3 reversals.

Additional legal factors are incorporated into the civil rights model, including
whether the Supreme Court decision emanated from the Burger Court, whether the precedent
has been modiﬁed by subsequent decisions, the amount of time that has passed Since the
precedent was announced, and whether the appeals court decision was rendered en banc.

Burger Court. A dummy variable for a Burger Court decision is employed as an
additional indicator of the clarity of a particular precedent. Carp and Rowland (1983, 11)
characterize Burger Court precedents as ideologically imprecise and inconsistent in contrast

to the “Warren Court’s narrow, but more consistent, liberalism.” Another scholar states that

 

17According to CQ’s determination, 14 of the 163 cases in this analysis are coded as
salient.

39

“one of the most Signiﬁcant characteristics of the Burger Court is ambiguity in the legal rules
it [applies]” (Walker 1978, 360). AS I trace decisions of both the Warren and Burger Courts
in this analysis, lower court treatment of these Courts’ decisions may be compared. The
objective is to assess whether there is empirical support for the hypothesis that Burger Court
precedents are more confusing than those of the Warren Court and thus are followed less
frequently in the circuit courts.

Doctrinal Modiﬁcation. A number of studies have suggested that a consistent series
of Supreme Court decisions may be more pivotal for compliance than the clarity of a single
decision (Gruhl 1980; Johnson and Canon 1984, 51; Johnson 1987). The analysis of the four
decisions in Chapter 3 illustrates how a precedent may be modiﬁed or obscured by ensuing
Court decisions. A dummy variable is included in the model to indicate those Courts of
Appeals’ decisions that took place after a signiﬁcant change in Supreme Court doctrine, to
assess how these courts are responding to such modiﬁcations and to control for alterations
in precedent.

The Court’s treatment of Rosenbloom v. Metromedia,18 one of the cases traced here,
illustrates the importance of such a variable. In this libel decision, the Court ruled that
private persons who had become involved in public issues would have to Show actual malice
to recover compensatory damages. However, in its subsequent decision in Gertz v. Welch,l9
the Court implicitly overruled Rosenbloom, holding that such persons would not have to

Show actual malice to recover damages. It is critical to distinguish those appeals court cases

 

18403 US. 29 (1971).

‘9418 US 323 (1974).

40

which treated Rosenbloom after Gertz, as the lower courts should no longer be following the
Rosenbloom precedent.20

Age of Precedent. In order to determine whether a new or old precedent is more
persuasive, I include a measure of the length of time between the Supreme Court and circuit
court decisions. A precedent may grow stronger with the passage of time as it is applied and
clariﬁed in subsequent decisions and even extended to other areas (Johnson 1987); or, the
authority of a Supreme Court decision may erode over time due to shifts in Court personnel
and doctrine (Marshall 1989; Pacelle and Baum 1992). Time is measured as the number of
years between High Court and lower court decisions.

En Banc. I include a dummy variable for those appeals court decisions that are
rendered by all of the judges in a circuit, in order to test competing hypotheses regarding
decision making in these cases.21 An explanation of en banc decisions has been advanced

which parallels that of salient Supreme Court decisions; i.e., en banc cases involve

 

20In addition to Rosenbloom, other Supreme Court decisions which were modiﬁed
signiﬁcantly by later decisions include Pierson v. Ray, 386 US. 547 (1967); Lou v.
Nichols, 414 US. 563 (1974); Cleveland Board of Education v. LaFleur, 414 US. 632
(1974); and Wood v. Strickland, 420 US. 308 (1975). Pierson and Wood outlined
subjective and objective components to be established in qualiﬁed immunity analysis, but
the Supreme Court eliminated the subjective component in Harlow v. Fitzgerald, 457 US.
800 (1982). The Lau Court said that, while an equal protection violation under the
fourteenth amendment requires proof of discriminatory intent, all that is required to
establish a Title VI violation is proof of disparate impact. However, in Board of Regents
v. Bakke, 438 US. 265 (1978), the Court held that Title VI prohibits only uses of racial
criteria that would violate the fourteenth amendment. In deciding LaFleur, the Court
employed “irrebuttable presumption” analysis, but its subsequent decision in Weinberger
v. Salﬁ, 422 US. 749 (1975), severely limited the use of this doctrine as a generally
acceptable mode of analysis in cases involving constitutional challenges to statutes or
regulations.

21223 of the 5664 appeals court cases in this analysis were decided en banc.

41

signiﬁcant and controversial issues, and the Supreme Court will be attentive to the resolution
of these cases (George 1997). If lower courts perceive that review is likely, they may feel
more constrained by precedent in reaching their decisions. Alternatively, the Supreme Court
may be less likely to review decisions reached by the circuit as a whole, as such a procedure
allows a fuller consideration of the case (Atkins 1972; Richardson and Vines 1970). Circuit
court judges may discern that they have more freedom in decision making in such cases.

The civil rights and liberties data is also used to assess the extent of inter- and intra-
circuit conﬂict in the treatment of these decisions. A number of studies have noted variation
among the business and behavior of these courts in general (Baum, Goldman, and Sarat
1981; Howard 1981; Carp and Rowland 1983; Davis and Songer 1988-9; Wenner and Dutter
1989) and the differential application of Supreme Court decisions in particular (Vines 1963;
Combs 1982). Panel assignment procedures foster inconsistent decision making within
circuits as well as among them. A number of scholars have recognized the potential for
panel decisions to represent the preferences of a minority of circuit judges (Richardson and
Vines 1970; Atkins 1972; Goldman 1975; Howard 1981; Songer 1982).

In order to determine the degree of variation among the circuits in the treatment of
precedent, I use diﬁ‘erence-of-proportions tests to compare the rate of following precedent
in the Courts of Appeals as a whole to the following rate in individual circuits. I also run
twelve iterations of the civil rights model, including a dummy variable for a different circuit
each time, in order to determine whether the treatment of Supreme Court decisions in a
particular circuit differs Signiﬁcantly from such treatment in the other circuits. In order to
evaluate diversity in the adherence to precedent within circuits, I compare the decisions of

liberal-dominated panels with those of conservative-dominated panels. These tests will

42

provide evidence of the extent of inter- and intra-circuit inconsistency in the treatment of

precedent.

CHAPTER 3

COURTS OF APPEALS’ TREATMENT OF FOUR SUPREME COURT DECISIONS:
AN EXPLORATION

The legal model of judicial decision making suggests that there is a widespread norm
within the legal community of deference to the decisions of higher courts. There is also
evidence, both anecdotal and systematic, suggesting that judges do not wish to see their
decisions reversed. It is in highly salient cases where lower court judges should be most
aware of existing precedent, and where avoidance of precedent is most likely to come to the
attention of the higher court. In a preliminary examination of the relationship between the
Supreme Court and the Courts of Appeals, and the legal and attitudinal forces which may
deﬁne this relationship, I study the Courts of Appeals’ treatment of four Supreme Court
decisions. I focus speciﬁcally here on the clarity of Supreme Court precedent and the support
on the Court for that precedent, and I explore the inﬂuence these elements have on the Courts
of Appeals’ application of these four decisions.

I look ﬁrst at two decisions where the Court’s rulings are unambiguous and where
its reasoning is well-explicated in the opinions of the Court. In US. v. Wade, the Supreme
Court held that a criminal defendant’s sixth amendment right to counsel applies to all critical
stages of the prosecution, including a post-indictment lineup. In Griggs v. Duke Power Co. ,
the Court held that, under Title VII of the Civil Rights Act of 1964, any employment practice
which excludes blacks and is not related to job performance is prohibited.

43

44

I also examine two cases that are equivocal in terms of the justiﬁcations for the
Court’s decisions and the guidelines for the decisions’ application in the lower courts. In
Lemon v. Kurtzman, the Court held that, in order to avoid a violation of the establishment
clause of the ﬁrst amendment, a statute must have a secular legislative purpose, its primary
effect must be one that neither advances nor inhibits religion, and it must not foster an
excessive entanglement with religion. In Terry v. Ohio, the Court held that certain situations
present exceptions to the warrant requirement of the fourth amendment; in such situations
the appropriate analysis is a balancing of the justiﬁcation for the search or seizure against the
invasion it entails.

In the following sections, I discuss each of these four cases, paying particular
attention to the reasoning employed by the Supreme Comt in reaching its decisions. I trace
the interpretation of these decisions in subsequent Court cases to determine whether these
precedents have been modiﬁed, obscured, or even abandoned. I also provide a discussion
of the perspectives of legal scholars who have analyzed these decisions and their application
in the lower courts. Finally, I examine the treatment these decisions have been afforded in
the Courts of Appeals and consider characteristics of the Supreme Court’s decisions which

may have determined this treatment.

U.S. V. WADE, 388 US. 218 (1967)
On April 2, 1965, Billy Joe Wade and an accomplice were arrested and charged with
robbery. Several weeks after his arrest, and without notice to his appointed counsel, Wade
was exhibited in a lineup to witnesses of the robbery and was compelled to repeat the words

allegedly spoken by the robber. The witnesses identiﬁed Wade as the robber. They

45

conﬁrmed their identiﬁcation on direct examination at trial. Wade’s counsel argued that the
conduct of the lineup violated Wade’s ﬁfth amendment privilege against self-incrimination
and his sixth amendment right to the assistance of counsel, but Wade was convicted by the
district court. The circuit court, holding that the lineup was a violation of Wade’s sixth
amendment rights, reversed the conviction and ordered a new trial at which the in-court
identiﬁcation was to be excluded. The Supreme Court granted certiorari. The Court
reversed the judgment of the circuit court and remanded the case to the district court.

Writing for the Court, Justice Brennan ﬁrst addressed the ﬁfth amendment issue. He
found the controlling case to be Schmerber v. California, 384 US. 757 (1966), where the
Court held that the privilege against self-incrimination protects the accused only from being
compelled to provide evidence of a testimonial or communicative nature. Wade was
compelled merely to exhibit his person for observation by witnesses prior to trial, which
involved no compulsion to give evidence having testimonial signiﬁcance (3 88 US. 218 at
222). Neither was the statement which Wade was compelled to utter testimonial in nature,
as his voice was being used solely as an identifying physical characteristic (Id. at 222). The
Court concluded that the lineup did not violate Wade’s ﬁfth amendment privilege against
self-incrimination.

Regarding Wade’s right to counsel at the post-indictment lineup, the Court
recognized that the criminal court process requires confrontations of the accused by the
prosecution at pretrial proceedings which could determine the accused’s fate, and that prior
decisions had construed the sixth amendment guarantee to apply to “critica ” stages of the
proceedings (Id. at 224). The language of the Sixth amendment encompassed counsel’s

assistance whenever necessary to assure a meaningful “defence” (Id. at 225). Thus any

46
pretrial confrontation of the accused should be scrutinized to determine whether the presence
of his counsel is necessary to preserve his right to a fair trial (Id. at 227). The Court found
that there was grave potential for prejudice in a pretrial lineup, and they held that Wade’s
post-indictment lineup was a critical stage of the prosecution at which he was entitled to
counsel (Id. at 236-237).

Having resolved these issues, the Court considered whether the courtroom
identiﬁcation of Wade should have been excluded due to the absence of counsel at the
lineup. The Court held that the proper test to be applied was whether the courtroom
identiﬁcation was the product of the illegal lineup (Id at 244). The Supreme Court reversed
the judgment of the circuit court and remanded the case to the district court to determine
whether the in-court identiﬁcations had an independent source, and if not, whether the use
of the identiﬁcations as evidence was harmless error (Id. at 242).

Justices Black, Harlan, Stewart, and White cast dissenting votes. Justice Black
argued that the lineup violated Wade’s privilege against self-incrinrination. He reasoned that
if Wade had been compelled to speak in open court, he would have been acting as a witness
against himself; being forced to supply evidence against himself by talking outside the
courtroom was equally violative of his ﬁfth amendment right (Id. at 245). He also objected
to the rule of evidence which the Court established regarding use of the courtroom
identiﬁcation. He viewed the rule as unsound, as well as an encroachment on the
constitutional prerogative of the states and the federal government to determine their own
rules of evidence (Id. at 248, 250).

Justice White, writing for Justices Stewart and Harlan, objected to the broadness of

the Court’s decision, which he claimed would bar the use of a wide range of relevant

47

evidence solely because a step in its acquisition occurred outside the presence of defense
counsel (Id at 250). He also took issue with the basis for the Court’s decision, which was
the widespread utilization of improper police procedures (Id at 251). Justice White believed
that the decision would not contribute measurably to more reliable pretrial identiﬁcations and
that it could lead to fewer convictions (Id at 259).

Justice Douglas wrote a brief concurrence in which he joined the opinion of the
Court, except for the Court’s holding that compulsory lineups do not violate the ﬁfth
amendment. Justice Fortas also wrote a concurring opinion which Chief Justice Warren and
Justice Douglas joined. Justice Fortas agreed with Justice Black that the accused should not
be compelled in a lineup to speak the words of the person who committed the crime. He
argued that it was the “kind of volitional act...which falls within the historical perimeter of
the privilege against compelled self-incrimination” (Id. at 260). Justice Clark also wrote a
concurring opinion, in which he expressed his agreement with the Court’s holding that a

lineup is a critical stage of the prosecution.

COURTS OF APPEALS’ TREATMENT OF WADE
There has been a state of confusion among lower court judges concerning the
Supreme Court’s intentions in Wade (Sobel 1971; Comment, Iowa Law Review 1972-3).
This uncertainty likely stems from the Court’s fractionalized opinion. The majority opinion
in Wade has been “charitably referred to as a splintered opinion” (Read 1969-70, 351). Only
Justices Brennan and Clark Signed on to the two major portions of the Court’s opinion,
holding that there was no ﬁfth amendment violation but that there was a sixth amendment

violation. The Court’s opinion obviously was not satisfactory to either of the largest

48

coalitions on the Court. Four justices saw both ﬁfth and Sixth amendment violations, and
three saw neither ﬁfth nor sixth amendment violations. The Court’s exclusionary rule on all
in-court identiﬁcations that are tainted by illegal out-of-court lineups received a bare ﬁve-to-
four majority.

Another factor which could contribute to divergent treatment of the Wade decision
in the lower courts is the Court’s decision ﬁve years later in Kirby v. Illinois, 406 US. 682
(1972). In Kirby, the Court was asked to determine whether an accused had a constitutional
right to counsel at a police station lineup that took place after an arrest but before an
indictment or formal charges had been brought. A plurality of the Court refused to extend
the right to counsel to preindictrnent identiﬁcations. The Kirby Court departed from Wade’s
approach to the right to counsel by focusing on what constituted a “criminal prosecution”
rather than what constituted a “critical stage in the proceedings” (Comment, Iowa Law
Review 1972-3, 413). The Court’s judgment stressed that the accused is entitled to counsel
at any stage of the prosecution and declined to extend that right to a routine police
investigation (406 US. 682 at 690). While the Wade Court had sought to ensure a
meaningful defense for the accused, the plurality in Kirby emphasized society’s interest in

the “prompt and purposeful investigation of an unsolved crime” (Id at 691).I

 

1Three years after Kirby, the Supreme Court ruled that a probable cause determination
is also not a “critical stage” in a criminal prosecution requiring appointed counsel. See
Gerstein v. Pugh, 420 US. 103 (1975). The Court reasoned that, while pretrial custody may
affect to some extent the defendant’s ability to assist in the preparation of his defense, the
absence of counsel at this stage does not present the high probability of substantial harm
identiﬁed as controlling in Wade. The Court characterized probable cause hearings as
informal procedures, where the consequences of determination were less severe, the ﬁne
resolution of conﬂicting evidence was not required, and credibility determinations were
seldom crucial (Id. at 121).

49

The Kirby decision considerably restricted the role of counsel in protecting the
pretrial rights of the accused and was unexpected in the legal community (Comment, Iowa
Law Review 1972-3). In view of the possible prejudicial factors associated with lineup
proceedings which the Wade Court had emphasized, legal commentators expected the lower
courts to reject a pre-indictment--post-indictrnent dichotomy when applying Wade (Kelly
1973). Extending the right to counsel to preindictrnent confrontations seemed logical based
on the Court’s reasoning in Wade, and, as the dissenters in Kirby noted, the majority of lower
courts since the Wade decision had taken this interpretation.2 As Kelly (1973) argued, such
an extension was the logical next step in the line of decisions that began with Powell v.
Alabama, 287 US. 45 (1932), and extended through Escobedo v. Illinois, 378 US. 478
(1964), and Miranda v. Arizona, 377 US. 201 (1966).

Data on the Courts of Appeals’ treatment of the four cases in this analysis is
presented in Table 3.1. Support for the hypotheses that weak and unclear precedents are not
well-received in the lower courts is found when we examine the treatment accorded Wade
by the Courts of Appeals. The decision in Wade was followed in only 42% of the cases in
which it was applied by the Courts of Appeals. This ﬁgure is especially striking given that

the four cases examined here were followed 75% of the time.

 

2AS one example, note the Tenth Circuit’s statements in Wilson v. Gaﬁirey, 454 F.2d
142, 144 (10th Cir. 1972):

[S]urely the assistance of counsel, now established as a post-
indictrnent right, does not arise or attach because of the return
of an indictment...The conﬁontation of a lineup...cannot have
a constitutional distinction based upon the lodging of a formal
charge. Every reason set forth by the Court in Wade...for the
assistance of counsel post-indictment has equal or more
impact when projected against a pre-indictrnent atmosphere.

50

When the higher court is Sharply divided as to both the holding and the reasoning of
a particular decision, as the Supreme Court was in Wade, lower court judges may be unsure
of the direction the Court wants them to take in applying that decision. The ﬁnal vote in the
case was ﬁve to four. The justices wrote four concurrences and two dissents, so that a total
of six separate opinions were written. Clearly, there was a low level of support among the
justices for the Court’s decision. Lack of support on the higher court may limit the authority
which is attached to its decision (Rohde 1972), and it may also raise questions about the legal
correctness of the decision (Johnson and Canon 1984, 58).

Failure on the part of the higher court to provide “consistent, continuing cues” to the
lower courts also leaves these courts free to make their own inferences regarding the scope
and applicability of higher court decisions (Johnson and Canon 1984, 51). The Kirby
decision may have created uncertainty in the lower courts as to the direction the Supreme
Court was taking in this area. Whether due to lack of agreement in the Court over the
reasoning for the decision or the seemingly contradictory ruling in Kirby, the Wade decision
seems to have left the lower courts unsure of the Supreme Court’s policy intentions. The

result is a low level of compliance in the Courts of Appeals.

GRIGGS V. DUKE POWER Ca, 401 US. 424 (1971)

In 1968 a group of black employees brought a class action suit against Duke Power
Company, the power generating facility at which they worked. The company required a high
school diploma or the passing of intelligence tests for employment in or transfer to jobs at
the plant. These requirements were not shown to be Signiﬁcantly related to successful job

performance, and they operated to disqualify blacks at a substantially higher rate than whites.

5 l

The employees challenged the company’s policy as a violation of Title VII of the Civil
Rights Act of 1964. Section 703 (a) of the Act makes it unlawful for an employer to limit,
segregate, or classify employees in order to deprive them of employment opportunities or
adversely to affect their status because of race, color, religion, sex, or national origin.
Section 703 (h) authorizes the use of any professionally developed ability test, provided that
it is not designed, intended, or used to discriminate.

The district court found that prior to July 2, 1965, the effective date of the Civil
Rights Act of 1964, the company had openly discriminated on the basis of race in the hiring
and assigning of employees, but that such conduct had ceased since the Act went into effect.
That court held that since Title VII was prospective only, it did not apply to prior actions.
The circuit court reversed in part. It held that residual discrimination arising from prior
practices was not insulated from remedial action, but it agreed with the lower court that there
was no Showing of a discriminatory purpose in the adoption of the diploma and test
requirements. Therefore there was no violation of the Act. The employees appealed to the
Supreme Court, arguing that the requirements were unlawful under Title VII unless they were
Shown to be job related. The Supreme Court granted certiorari and reversed the judgment
of the circuit court.

Chief Justice Burger delivered the opinion for a unanimous Court.3 He described
Congressional intent in enacting Title VII:

What is required by Congress is the removal of artiﬁcial,

arbitrary, and unnecessary barriers to employment when the
barriers operate invidiously to discriminate on the basis of

 

3Justice Brennan took no part in the consideration or decision of this case.

52

racial or other impermissible classiﬁcation (401 US. 424 at
431)

The Court read the Act to proscribe “not only overt discrimination, but also practices that are
fair in form, but discriminatory in operation” (Id at 431). The Act was said to be directed
to the “consequences of employment practices, not simply the motivation” (Id at 432,
emphasis in original). In the words of Chief Justice Burger:

[G]ood intent or absence of discriminatory intent does not

redeem employment procedures or testing mechanisms that

operate as ‘built-in headwinds’ for minority groups and are

unrelated to measuring job capability (Id at 432).

The Court held that an employment practice which operates to exclude blacks and
which is not related to job performance is prohibited. The standard to be applied is business
necessity, or “manifest relationship to the employment in question” (Id at 432). The Court
relied on the legislative history of Title VII, along with the fact that the Equal Employment
Opportunity Commission had adopted guidelines consistent with the Court’s decision, to

support its holding (Id at 436). The Court concluded that the requirement that employment

tests be job related comported with congressional intent.

COURTS OF APPEALS’ TREATMENT OF GRIGGS
One expert in the ﬁeld of employment discrimination has praised the Griggs decision
for “pouring decisive content into a previously vacuous conception of human rights” by
deﬁning discrimination in terms of consequence and effect rather than motive and purpose
(Blurrrrosen 1972, 62). Under Griggs, discrimination became conduct rather than a state of
mind (Blunrrosen 1972). Such a deﬁnition was new to the ﬁeld of employment

discrimination, where a subjective test had previously been used. According to one

53

commentator, the lower federal courts “quickly grasped and applied” this concept of
discrimination (Blumrosen 1972, 74). Lower courts have not limited Griggs ’ application to
employment credentials such as high school diplomas and intelligence tests, but have
extended it to such employment policies as the use of conviction records in hiring4 and wage
garnishment rules.5

While the lower courts expanded upon the Griggs decision by extending it to other
areas, most of these courts have construed the business necessity defense more narrowly,
“consistently demonstrat[ing] an extremely restrictive application of the business necessity
defense...[so that] only the most essential business practices will survive attack” (Comment,
University of Chicago Law Review 1979, 920). The lower courts have taken this initiative
in spite of the Supreme Court’s liberal application of the business necessity doctrine in its
subsequent decisions.6 Some commentators have attributed this behavior to the Court’s
failure to discuss adequately the scope of the business necessity defense, especially in terms
of the types of business purposes that justify an employer’s policy and how “necessary” the
policy must be (Comment, University of Chicago Law Review 1979).

Lower court compliance with the Griggs decision may have been ﬁ'ustrated by
ensuing Supreme Court decisions, particularly that of McDonnell Douglas Corp. v. Green,
411 US. 792 (1973), where the Court recognized a second type of Title VII violation. The

Griggs Court held that a neutral employment policy which has a “disparate impact” on

 

4Carter v. Gallagher, 452 F.2d 315 (8th Cir. 1971); Green v. Missouri Pac. R. R. Co.,
523 F.2d 1290 (8th Cir. 1975).

5Johnson v. Pike Corp. of America, 332 F.Supp. 490 (CD. Cal. 1971).

6See, e. g., New York Transit Authority v. Beazer, 440 US. 568 (1979).

54

persons protected under Title VII is prohibited. In McDonnell Douglas, the Court ruled that
“disparate treatment” (i.e., treating some people less favorably because of race, color,
religion, sex, or national origin) also violates Title VII. One legal scholar has observed that,
since the recognition of these two types of Title VII violations, the Court has begun to merge
the two types of cases, developing “identical orders of proof, as well as similar substantive
defenses...for each type of case” (Furnish 1982, 445). This commentator has characterized
the Court’s disparate impact and disparate treatment cases as a “maze” through which a
suitable path must be found (Furnish 1982).

According to Griggs and McDonnell Douglas, intent is irrelevant in a disparate
impact case, but in a disparate treatment case it is crucial. Unlike a disparate impact plaintiff
under Griggs, a McDonnell Douglas disparate treatment plaintiff must establish
discriminatory intent, or that a “legitimate, nondiscriminatory reason” for an employee’s
rejection was a “pretext or [was] discriminatory in its application” (411 US. 792 at 780).
In subsequent cases,7 however, the Supreme Court developed a similar burden of proof for
plaintiffs in disparate impact cases. By showing that an alternative practice would
accomplish an employer’s job related purpose with a less discriminatory impact, a plaintiff
could demonstrate that an employer enacted a given policy as a pretext for intentional
discrimination (Albemarle, 422 US. 405 at 428). The Court’s decision four years later in
Beazer “clearly injected the issue of intentional discrimination into disparate impact cases”
(Furnish 1982, 424). The Beazer Court failed to mention the relevance of lesser-impacting

alternatives under Albemarle and Dothard, and instead relied solely upon the absence of

 

7Albemarle Paper Co. v. Moody, 422 US. 405 (1975); Dothard v. Rawlinson, 433
US. 321 (1977); New York City Transit Authority v. Beazer, 440 US. 568 (1979).

55
discriminatory intent in holding that the plaintiff had failed to rebut the business necessity
defense.

The defense available to an employer in a disparate impact case is “business
necessity” or job-relatedness (Griggs, 401 US. at 431), while in a disparate treatment case
an employer must articulate a “legitimate non-discriminatory reason” for a discriminatory
policy (McDonnell, 411 US. at 802). Subsequent decisions, however, have blurred the
distinction between the two defenses, reducing the burden to be met by defendants in
disparate impact cases (F urnish 1982). Note the Court’s decision in Beazer which, according
to one commentator, “allowed the establishment of a defense in a disparate impact case when
the defendant did nothing more than articulate a legitimate non-discriminatory reason”
(Furnish 1982, 439).8

In addition to the confusion which may have been created by McDonnell Douglas and
the disparate treatment line of cases, a second factor may hinder the application of the Griggs
decision in the lower courts. This factor is the standard by which employment tests are to
be judged under Title VII (Wilson 1972; Comment, University of Chicago Law Review
1979). The Court suggested that the legality of such tests turns on the degree to which they
predict job success. Test validation is both technical and difﬁcult, and the establishment of

this legal standard opened a “complex area of inquiry to judicial scrutiny” (Wilson 1972,

 

8This assertion received further support in the Court’s most recent disparate impact
decision. In Ward’s Cove Packing Co. v. Atonio, 490 US. 642 (1989), the Court held that
in disparate impact cases, the employer’s burden of proof regarding business necessity
“should have been understood to mean an employer’s production--but not persuasion--
burden” (490 US. 642 at 660). Justice Stevens dissented, charging the majority with
blurring the distinction between the required defenses in disparate impact and disparate
treatment cases and diminishing the employer’s burden (Id at 668-669).

56
844). According to one legal commentator, the Griggs decision left numerous questions
regarding test validation unresolved, “thus providing the lower courts with considerable
leeway for interpreting its mandate” (Wilson 1972, 847).
Griggs was followed in 69% of the decisions in which it was applied. This is six

points less than the overall following rate for the four decisions. We might have expected
more consistent application, given that the Court’s ruling was straightforward and that the
reasoning was agreed to by all of the justices. Baum (1976) suggested that clear decisions
are relatively easy to achieve in cases with a narrow scope, and this seems to be an accurate

description of the Griggs decision. The fact that the decision was unanimous, with a Single
majority opinion, also contributes to its clarity. The hypotheses advanced in previous studies
seem to suggest that Griggs would have been followed even more faithﬁrlly than it was by
the lower courts. One aspect of Griggs that could create problems with its interpretation is
the Court’s test validation standard, which introduced a degree of complexity into the
decision. Even more crucial may be the confusion created by the Court’s decisions in

McDonnell Douglas and its progeny.

LEMON V. KURTZMAN, 403 US. 602 (1971)

In 1968 the Pennsylvania legislature passed the Nonpublic Elementary and Secondary
Education Act, which authorized ﬁnancial support for nonpublic schools through
reimbursement for the cost of teachers’ salaries, textbooks, and instructional materials in
speciﬁed secular subjects. The act was passed in response to a crisis that the legislature

believed existed in the state’s nonpublic schools due to rapidly rising costs. Several citizens

57
and taxpayers of Pennsylvania challenged the constitutionality of the act as a violation of the
establishment and free exercise clauses of the ﬁrst amendment. The three-judge district court
dismissed the claim, and the Supreme Court noted probable jurisdiction. The Court reversed
the judgment of the district court and remanded the case.

Chief Justice Burger wrote the opinion of the Court.9 He began by articulating three
tests that the Court had developed over the years in analyzing establishment clause cases
(403 US. 602 at 612, 613). First, the statute must have a secular legislative purpose.
Second, its principal or primary effect must be one that neither advances nor inhibits religion,
Board of Education v. Allen, 392 US. 236 (1968). Finally, the statute must not foster “an
excessive entanglement with religion,” Walz v. Tax Commission, 397 US. 664 (1970).

The Court accepted the Pennsylvania legislature’s stated purpose in passing the
statute, which was to enhance the quality of the secular education offered in schools covered
by compulsory attendance laws (Id at 613). In the Court’s view, the legislature recognized
that the program “approached, even if it did not intrude upon, the forbidden areas under the
Religion Clauses” (Id at 613). Such a realization was evidenced in the precautions taken by
the state to guarantee that the aid supported only secular educational functions (Id at 613).
The Court held that it did not need to consider whether these precautions adequately
restricted the principal or primary effect of the program, Since the “cumulative impact of the
entire relationship arising under the statutes...involves excessive entanglement between

government and religion” (Id at 613-614).

 

9Justice Marshall took no part in the consideration or decision of the case.

58

Chief Justice Burger acknowledged that some relationship between government and
religious organizations is unavoidable (Id at 614). He described the separation that must be
maintained: “Judicial caveats against entanglement must recognize that the line of
separation, far from being a ‘wall,’ is a blurred, indistinct, and variable barrier depending on
all the circumstances of a particular relationship” (Id at 614). The circumstances which
should be considered are the character and purposes of the institutions that are beneﬁted, the
nature of the aid that the state provides, and the resulting relationship between the
government and the religious authority (Id at 615).

The Court found that the restrictions and oversight necessary to ensure that teachers
play a “strictly nonideological role” promoted entanglements between church and state (Id
at 620). The schools were required to maintain accounting procedures that identiﬁed the
“separate” cost of the “secular educational service,” and these accounts were subject to state
audit (Id at 610). Aid was also limited to courses which were part of the curriculum in
public schools. The fact that the program entailed a continuing cash subsidy was also an
indication that control and surveillance would be necessary. The Court concluded that the
statute created an “intimate and continuing relationship between church and state” and was
therefore unconstitutional (Id at 622).

Chief Justice Burger also discussed a broader base of entanglement which was
presented by the “divisive political potential” of the program (Id at 622). The issues
presented by state aid to parochial schools would divide states and communities and would
tend to “confuse and obscure other issues of great urgency” (Id at 622-623). One of the
principal goals of the ﬁrst amendment was to protect against such political divisions along

religious lines (Id at 622).

59

Justice Douglas wrote a concurring opinion, which Justice Black joined, in which he
argued that the use of taxpayers’ money to support parochial schools violates the ﬁrst
amendment, regardless of whether the money is used for solely secular educational functions
(Id. at 640-641). Justice Brennan also concurred, setting forth his views that the statute
impermissibly involved the state with the “essentially religious activities” of sectarian
educational institutions, and that the state government was using “essentially religious means
to serve governmental ends, where secular means would sufﬁce” (Id at 658). Justice White
published a concurring opinion in which he stated that an agreement between the school and
the state that the state ﬁrnds would be used only to teach secular subjects did not violate the
ﬁrst amendment (Id at 670). However, he agreed with the Court that the district court’s
judgment should be reversed, as the taxpayers’ claim should not have been dismissed for

failure to state a cause of action.

COURTS OF APPEALS’ TREATMENT OF LEMON
While the Lemon test appears straightforward, it has not led to predictable
adjudication (Paulsen 1986). The Supreme Court itself has continued to struggle with the
Lemon test and has voiced repeated dissatisfaction with both the test and the “vagaries of its
application” (Paulsen 1986, 330). One commentator describes the Court’s ambivalence: “At
times the Court has described the test as a helpful signpost,” at other times the Court has

suggested that it can be discarded in certain circumstances,11 and at still other times the Court

 

”See Lynch v. Donnelly, 465 US. 668 (1984); Mueller v. Allen, 463 US. 388 (1983);
Hunt v. McNair, 413 US. 734 (1973).

llSee Marsh v. Chambers, 463 US. 783 (1983).

60
has held that it must be rigorously applied”'2 (Marshall 1986, 497). The result is a
“patchwork of ad hoc decisions inside a legal framework that [has lost] its intellectual
integrity” (Marshall 1986, 498).

The Lemon test appeared headed for obsolescence following several establishment
clause decisions of the early 19805. In Marsh v. Chambers, the Court ignored the test
entirely in rejecting a challenge to a state legislature’s practice of opening its sessions with
a prayer led by a state-paid chaplain. In Lynch v. Donnelly, the Court applied the test in fornr
only to allow a city to include a nativity scene in its annual Christmas display. In Mueller
v. Allen, upholding tax deductions for tuition and other expenses of parochial school
education, the Court indicated that it no longer viewed the Lemon test as the proper measure
of establishment clause violations.

However, the establishment clause decisions of the Court’s 1984 Term represented
a shift in direction, with the revitalization of the Lemon test (Simson 1987). The Court
applied the test in invalidating a state moment-of-silence law,l3 publicly ﬁnanced programs
offering remedial and enrichment classes to parochial school children in parochial school

classrooms,” and a state law barring employers from requiring employees to work on their

 

12See Stone v. Graham, 449 US. 39 (1980); Committee for Public Education and
Religious Liberty v. Nyquist, 413 US. 756 (1973).

13Wallace v. Jaﬁi'ee, 472 US. 38 (1985).

l"Aguilar v. Felton, 473 US. 402 (1985); Grand Rapids School District v. Ball, 473
US. 373 (1985).

61
Sabbath.l5 The Court afﬁrmed its commitment to the Lemon test in its only establishment
clause decision of the 1985 Term.”

In spite of the Court’s professed ﬁdelity to the analysis set forth in Lemon, the Lemon
test has remained the subject of continued criticism, and even derision, among members of
the Court. In his colorful concurrence in Lamb’s Chapel v. Center Moriches Union Free
School District, 508 US. 384 (1993), Justice Scalia provided the following analogy to the
Court’s invocation of the Lemon test:

Like some ghoul in a late-night horror movie that repeatedly

sits up in its grave and shufﬂes abroad, after being repeatedly

killed and buried, Lemon stalks our Establishment Clause

jurisprudence once again, frightening the little children and

school attorneys of Center Moriches Union Free School

District (1d at 398).
Scalia went on to note that, over the years, ﬁve of the justices Sitting on the Court when
Lamb ’s Chapel was decided had “driven pencils through the creature’s heart” and a Sixth had
joined an opinion which did so (Id. at 398).

The Lemon test has been criticized by the legal community on a number of grounds.
With regard to the second part of the test, the requirement that the principal or primary effect
be one that neither advances nor inhibits religion, some critics have objected to the Court’s
broad reference to advancements and inhibitions of religion (Simson 1987). They have

pointed out the difﬁculties involved in deﬁning what is and is not “religious” and have

argued that it is not a task for which courts are particularly well-suited (Paulsen 1986).

 

15Estate of Thornton v. Caldor, Inc., 472 US. 703 (1985).

1"’Witters v. Washington Department of Services for the Blind, 474 US. 481 (1986).

62

The third part of the test prohibits legislation that will foster excessive government
entanglements with religion. The Court interpreted this to require that the “character and
purpose of the institutions” and the “resulting relationship between the government and the
religious authority” be assessed to determine the likelihood of such an effect taking place in
the future. This standard has been criticized as an invitation to judges to decide the fate of
a law based on a set of circumstances that does not exist, or, in effect, to render an advisory
opinion (Simson 1987). According to one legal scholar, “While the role of subjectivity has
always been especially pronounced in religion clause litigation, the excessive entanglement
test invites a whole new degree of judicial subjectivity” (Ripple 1980, 1216-7).”

The ambiguity in the Lemon test may stem from the language of the ﬁrst amendment
(Simson 1987). In the Lemon opinion, Chief Justice Burger characterized the religion
clauses as “at best opaque” (Id at 612). He acknowledged that, in deciding what is permitted
and what is forbidden by the religion clauses, the Court could “only dimly perceive the lines

of demarcation” (Id at 612). The Court’s use of the religion/nonreligion dichotomy is seen

 

17The author also addressed the Court’s discussion of political entanglement:

When the Court turns to an assessment of ‘political
entanglement,’ an even greater opportunity for reliance on the
subjective is present. An assessment of how politics and
religion will mix in any given political environment
necessarily involves a personal judgment based in large part
on the Justice’s own political experience, observation, and,
possibly, tolerance for the methods of a particular religious
sect. It is a judgment easily swayed by the contemporary
political climate and by the inﬂuences of regionalism from
which no member of the Court entirely escapes (Ripple 1980,
1217)

63
as further evidence of the Court’s unclear thinking about the establishment clause (Paulsen
1986)

The lack of clarity in the Lemon decision may have an additional source. The ﬁnal
vote was unanimous; while a unanimous decision indicates support on the Court for the
judgment, it does not necessarily lead to a clear and coherent mandate. Rohde (1972)
showed that in cases with important policy implications, the justices may suppress their
disagreements in order to present a united front.” Such a tactic should minimize the
probability and degree of noncompliance (Rohde 1972); however, the result may be that the
language of the opinion is weakened, as it is written to accommodate a number of different
positions. Opinions which are compromises of a variety of views are likely to be ambiguous.
In fact, three concurring opinions were written in Lemon, in which the justices expressed
their own views.

Johnson and Canon (1984, 54) suggested that interpretations of ambiguous decisions
are less likely to be noncompliant, since noncompliance is difficult to deﬁne for such
decisions. In other words, unclear decisions may provide lower courts with leeway to

deviate ﬁ'om the higher court’s ruling while technically remaining in compliance. Consistent

with this hypothesis, the Lemon decision was followed in an overwhelming 90% of the cases

in which it was treated in the Courts of Appeals. The vagueness of the standards set forth

 

l"Rohde (1972) described such cases as “threat” situations. One type of threat
situation exists in issue areas in which there is likely to be noncompliance with the Court’s
mandate. There has been widespread resistance to the Court’s line of establishment clause
cases, particularly the school prayer decisions.

64
in the Lemon test may have created a precedent with which almost any lower court decision
could be harmonized.

Another possibility is that Shepard ’s codes as followed any lower court decision
which nominally applies the Lemon test, regardless of the interpretation of the individual
standards of the test. It is unlikely that a lower court would decide an establishment clause
case without citing Lemon v. Kurtzman as controlling. The high rate of compliance with
Lemon, then, may be a function of Shepard ’s coding conventions, rather than evidence of

faithful application in the lower courts.

TERRY V. OHIO, 392 US. 1 (1968)

While on patrol, a Cleveland police detective observed three men engaging in
behavior which he interpreted as “casing a job.” The ofﬁcer approached the men, identiﬁed
himself, and asked for their names. Following their responses, he grabbed the petitioner
Terry and patted down the outside of his clothing. He felt a revolver in Terry’s overcoat and
directed the men into a store, where he removed Terry’s overcoat and found the revolver.
The trial court convicted Terry of carrying a concealed weapon, denying Terry’s motion to
suppress the weapon. The intermediate appellate court afﬁrmed, and the state supreme court
dismissed the appeal on the ground that no substantial constitutional question was raised.
The Supreme Court granted certiorari and subsequently affirmed the judgment of the Ohio
appellate court.

Chief Justice Warren delivered the opinion of the Court. The Court recognized that
whenever a police ofﬁcer has restrained an individual’s freedom to walk away, he has

“seized” that person; and when a police ofﬁcer has explored the outer surfaces of a person’s

65

clothing, a “search” has taken place (392 US. 1 at 16). Thus the fourth amendment comes
into play as a limitation upon police conduct in “stop and frisk” encounters (Id at 19).

The situation in this case was one which required swift action on the part of the police
and therefore, as a practical matter, could not be subjected to the warrant procedure (Id at
20). However, the “notions which underlie both the warrant procedure and the requirement
of probable cause remain ﬁrlly relevant in this context” (Id at 20). The police ofﬁcer must
be able to point to “Speciﬁc and articulable facts which, taken together with rational
inferences from those facts, reasonably warrant the intrusion” (Id at 21). The appropriate
test for determining the reasonableness of fourth amendment activity is a balancing of the
need to search or seize against the invasion necessitated by that search or seizure (Id at 21).

A police ofﬁcer has an immediate interest in ensuring that the person he confronts
is not armed with a weapon that could be used against him (Id at 23). The test to be applied
in establishing an ofﬁcer’s authority to conduct a search is whether a reasonably prudent
man, based upon the circumstances and in light of his experience, is warranted in his belief
that his safety or the safety of others is in danger (Id at 27). A search for weapons must be
limited to that which is necessary for their discovery and may be characterized as something
less than a “full” search (Id at 26). In the present case, the Court found that the search was
permissible and that the pistol seized from Terry was properly admitted into evidence.

Justice Black concurred in the Court’s judgment and opinion (Id at 31), except where
the opinion relied upon the Court’s opinion in Katz v. United States, 389 US. 347 (1967),
and the concurring opinion in Warden v. Hayden, 387 US. 294 (1967). Justice Harlan wrote
a concurring opinion to make explicit his view that “the ofﬁcer’s right to interrupt Terry’s

freedom of movement and invade his privacy arose only because circumstances warranted

66
forcing an encounter...in an effort to prevent or investigate a crime” (Id at 34). Justice White
issued a concurrence to express his belief that, if an investigative stop is justiﬁable, there is
no constitutional violation if relevant questions are asked and the person is brieﬂy restrained
(Id at 35).

Justice Douglas dissented from the Court’s decision. He objected to the Court’s
disclaiming probable cause. He argued that the crime committed was carrying a concealed
weapon and that there was no basis for concluding that the officer had probable cause to
believe this crime was being committed (1d at 36). If a warrant had been sought by the
ofﬁcer, a judge would therefore have been unauthorized to issue one (Id at 36). The effect
of the Court’s decision was to give greater power to the police to make a seizure and conduct
a search than a judge has to authorize one (Id at 36). A seizure of a person is reasonable
under the fourth amendment only if the police are required to possess probable cause before

they seize him (Id at 38).

COURTS OF APPEALS’ TREATMENT OF TERRY

The Terry decision has been subjected to widespread criticism due to the lack of
Speciﬁc guidelines it provided to lower courts (LaF ave 1968; Williamson 1982; Dix 1985).
The Terry Court held that a seizure in order to conduct a weapons search is not an
investigatory seizure and is therefore not subject to the same fourth amendment requirements
as investigatory seizures; but it “completely failed to address the criteria for determining
when an investigatory seizure occurs, the factors distinguishing investigatory seizures from
other detentions, and the validity of nonarrest detentions other than those incident to weapons

searches” (Dix 1985, 857). One critic has accused the Terry Court of making a “conscious

67
effort to leave sufﬁcient room for later movement in almost any direction” (LaF ave 1968,
46). Another legal scholar has attributed the vagueness of the Terry decision to a more noble
purpose, “the development of a doctrine ﬂexible enough to accommodate both the impact
of different kinds of detentions upon citizens’ fourth amendment interests and the variety of
law enforcement interests involved in different detentions” (Dix 1985, 855).

In subsequent cases, the Supreme Court has addressed many of the issues left
unresolved in Terry. According to one commentator, however, the Court “has yet to provide
standards that will produce consistent and logical treatment of the issues in the lower courts
and workable guidelines for law enforcement ofﬁcials” (Williamson 1982, 817). Another
legal expert (Dix 1985) cited several examples which lend support to this assertion. In
Adams v. Williams, 407 US. 143 (1972), the Court assumed, virtually without discussion,
the validity under the fourth amendment of nonarrest detentions for investigatory purposes.
In other cases, as well, the Court has been “unable or unwilling to address the questions
clearly and deﬁnitively” (Dix 1985, 858). In Dunaway v. New York, 442 US. 200 (1979),
and Florida v. Royer, 460 US. 491 (1983), the Court ruled that nonarrest investigatory
detentions which were valid at their inceptions had become unreasonable, but failed to
discuss why this was so, or even when the detentions became invalid. In Hayes v. Florida,
470 US. 811 (1985), the Court held that movement to the stationhouse of a suspect detained
pursuant to a nonarrest detention was impermissible, but did not address the constitutional
limits on the transportation of suspects.

Problems in lower court application of Terry are likely to arise, not from lack of
Court support, but from uncertainty about the intended scope of the rule. The Terry decision

seemed to have the support of the Court, in that only one justice dissented and two of the

68
three justices who concurred did so only to emphasize particular portions of the Court’s
reasoning; but the insufﬁcient guidance regarding the substance of the fourth amendment’s
requirements for nonarrest detentions may allow lower courts a great deal of freedom in their
applications of Terry. The Terry decision was followed in 86% of the lower court decisions
which relied on it. This ﬁnding is consistent with Johnson and Canon’s (1984) hypothesis
that the Court’s failure to provide precise standards regarding the scope of a particular
decision may enable lower courts to provide their own interpretations, which are expressed

so as to be in accord with the decision.

DISCUSSION

The legal scholars who have analyzed the Supreme Court’s decisions in each of these
cases have agreed that the decisions left numerous questions unanswered and thus provided
the lower courts with considerable freedom in interpreting them (Sobel 1971; Wilson 1972;
LaF ave 1968; Dix 1985). A single decision cannot answer all questions about the scope of
its applicability; judging is to some degree subjective. This study has sought to identify
characteristics of higher court rulings which tend to enhance, and to limit, this subjectivity.
A few hypotheses may be advanced.

A decision which was unambiguous but which lacked support on the Court (Wade)
was not faithfully implemented in the Courts of Appeals, while another clear decision which
had the support of all the justices (Griggs) was followed more consistently but not to the
degree that we might expect. Two decisions established standards which were vague and

subjective and which had the support of the justices (Lemon and Terry), and these were

69

uniformly followed in the lower courts. This suggests that a precedent which provides
straightforward and coherent guidelines for its application may be more restrictive on lower
court decision making, while a precedent which sets forth vague and equivocal standards
may give lower court judges more room to maneuver. Examination of all four cases
highlighted the importance of the perceived authority of the Court’s decision, in terms of
Court support, in inﬂuencing lower court treatment.

The analyses of Lemon and Terry indicate that cases which involve constitutional
provisions may permit a wider range of interpretations than cases which involve statutory
provisions or other rules. The examination of Lemon also introduced an interesting
hypothesis regarding support on the higher court for its decision. This hypothesis is a
corollary of the clarity vs. ambiguity hypothesis discussed above. Court support may interact
with the type of case the Court is deciding as a predictor of lower court treatment. Some
unanimous decisions will provide clear directions to the lower courts and will not leave much
room for judges to reconcile divergent rulings; but some unanimous decisions, especially
those that establish policies which may encounter resistance in the lower courts, will be the
product of compromise. In the latter type of unanimous decisions, the Court’s intentions
may be ambiguous and thus easier to comply with.

This exploration also suggests that the Court’s rulings in subsequent cases may limit
the precedential impact of its decisions. The Court may later reach decisions which seem to
contradict prior rulings, develop alternative modes of analysis for highly similar cases, or

exhibit only sporadic allegiance to its previous decisions; and lower courts may be left in

70

doubt as to how and when past precedents should be applied. When precedent becomes
obscured, other forces are likely to act upon decision making in the lower courts.

It Should be clear from this analysis that legal variables relating to Supreme Court
decisions do not provide an adequate explanation of the treatment of these decisions in the
Courts of Appeals. This examination illustrates the need for multivariate tests of the
inﬂuence of ideological, as well as legal, factors on lower court behavior in light of Supreme

Court precedent. I conduct such analyses in the chapters which follow.

TABLE 3.1

COURTS OF APPEALS’ TREATMENT OF FOUR SUPREME COURT DECISIONS

 

 

 

 

 

 

 

NOT FOLLOWED FOLLOWED
58.0% 42.0%
US. V. WADE (76) (55)
31.3% 68.7%
GRIGGS V. DUKE POWER CO. (36) (79)
10.4% 89.6%
LEMON V. KURTZMAN (11) (95)
14.2% 85.8%
TERRY V. OHIO (47) (283)
24.9% 75.1%
TOTAL (170) (512)

 

 

 

 

CHAPTER 4

COURTS OF APPEALS’ TREATMENT OF THE DECISIONS OF THE 1987 TERM:
A MULTIVARIATE ANALYSIS‘

According to the most recent, and most comprehensive, study of Supreme Court
decision making (Segal and Spaeth 1993, 255), “attitudes...are crucial to explaining the votes
of the Supreme Court...Virtually no support is found for non-attitudinal factors.” Supreme
Court justices, unlike their lower court colleagues, are said to be free to act in accordance
with their personal ideologies and values for three reasons (Segal and Spaeth 1993, 73).
First, the legal rules governing decision making (e.g., precedent, plain meaning, intent of the
framers, balancing) do not limit the discretion of the justices; second, as the Supreme Court
is a court of last resort, its decisions cannot be overturned by other courts; and third, Supreme
Court justices lack electoral and political accountability and ambition for higher ofﬁce, so
that they need not respond to other political actors.

It is important to note that Segal and Spaeth (1993) emphasize the applicability of the

attitudinal model to the Supreme Court due to these unique institutional characteristics?

 

‘An earlier version of this chapter was presented as a co-authored paper,
“Ambivalence at the Courts of Appeals: Does the Legal or Attitudinal Model Apply?,” at the
1996 Meeting of the Midwest Political Science Association (Hurwitz and Reddick 1996).

2Note some of Segal and Spaeth’s (1993, 358) concluding comments:

[A]lthough we do not think the Supreme Court is necessarily

72

73

Decision making on lower courts, however, is not similarly unfettered. Judges are
socialized, through their law school training, their membership in legal organizations, and
their interactions with colleagues, to defer to the decisions of higher courts (Segal, Songer,
and Cameron 1995). Thus precedent should play a signiﬁcant role in their behavior. In
addition, the Supreme Court has discretionary jurisdiction, so that it may choose to review
those lower court decisions which conﬂict with the preferences of the justices. Finally,
unlike Supreme Court justices, subordinate judges may aspire to positions on higher courts.
Thus their decisions may be tempered by political forces.3

Studies of lower court decision making have revealed that a variety of factors are
brought to bear upon outcomes in these courts (see, e. g., Emmert 1992; Songer and Haire
1992; Brace and Hall 1993; Songer, Segal, and Cameron 1995). These factors include, but
are not limited to, case facts, legal issues raised by the litigants, nature of the litigants,
background characteristics of the judges, environmental and contextual features, and decision
making rules and procedures. Thus the most successful models of lower court behavior are

those which integrate legal, attitudinal, and institutional approaches.

 

unique in being driven attitudinally, we suspect that
intervening variables of a non-attitudinal sort appreciably
affect lower-court decision making...lf neo-institutionalism
has taught political science anything over the past decade, it
is that institutional structures shape incentives.

3Political ambition as a constraint on lower court judges is particularly germane to
circuit court judges. Segal, Songer, and Cameron (1995) reported that fourteen of the last
twenty justices appointed to the Supreme Court were members of the Courts of Appeals at
the time of their nomination.

74

Likewise, at the level of the Supreme Court, political scientists have grown wary of
a description of judicial decision making which relies primarily upon attitudinal motivations.
In Spite of the force of Segal and Spaeth’s (1993) argument, and the empirical support
provided, judicial behavioralists have continued to assert that legal considerations are not
absent from Supreme Court decision making. An analysis of the Court’s death penalty
decisions (George and Epstein 1992) revealed that the most successful prediction of
outcomes resulted from an incorporation of both legal and extralegal factors; and in an
examination of search and seizure cases before the Court, Segal (1984) found legally relevant
facts to be much more inﬂuential than extralegal facts in accounting for the Court’s
decisions.4 AS one study of the Supreme Court’s abortion and death penalty jurisprudence
concluded (Epstein and Kobylka 1992, 302), “the law and the legal arguments grounded in
the law matter, and they matter dearly.”

In light of this continued reliance on the signiﬁcance of legal doctrine and the
institutional boundaries of the attitudinal model, I examine here the utility of legal and
attitudinal explanations in describing the treatment of Supreme Court precedent in the Courts
of Appeals. AS discussed in Chapter 1, there is a paucity of such studies in the literature.
Thus this analysis should provide a meaningful contribution to our understanding of judicial

decision making in these intermediate federal courts.

 

4The studies of George and Epstein (1992) and Segal (1984) specify so-called “fact-
pattem” models of Supreme Court decision making. Segal and Spaeth (1993) argue that the
inﬂuence of fact patterns is consistent with both legal and attitudinal models.

75

HYPOTHESES

Most public law scholars have contended that Courts of Appeals judges have policy
goals which they would like to pursue if possible (see, e. g., Goldman 1966, 1975; Howard
1981; Johnson 1987; Songer and Haire 1992; Songer, Segal, and Cameron 1994). However,
these judges may be constrained from carrying out their personal preferences to the full
extent, as their legal socialization may motivate them to adhere to Supreme Court precedent
and the accompanying principle of stare decisis. Consequently, these judges will be
reluctant to have their decisions reversed on appeal (Baum 1978; Pacelle and Baum 1992),
and they should act accordingly to ensure that the Supreme Court does not reverse their
decisions.

Scholars have reported that Supreme Court behavior affects the manner in which
lower courts react. For instance, Wasby (1970) set forth a number of hypotheses which
suggested that lower courts are more likely to abide by a clear Supreme Court decision that
has the support of the members of the Court. On the other hand, the greater the ambiguity
of the policy enacted by the Court, the less likely lower courts are to apply faithfully such
precedent. Other scholars have advanced similar hypotheses, arguing that divisiveness on
the High Court leads to less compliance in the courts below (Rohde 1972; Johnson and
Canon 1984; Marshall 1989; Pacelle and Baum 1992).

Although many researchers have offered these hypotheses, they have not been
adequately tested. Johnson (1979, 792) sought to ﬁll this void by examining what he
referred to as the “widely accepted, but generally untested, idea that attributes of the original

Supreme Court decisions inﬂuence lower court reactions to and acceptance of the decisions.”

76

Johnson’s (1979) analysis focused on levels of Supreme Court support for precedent and
employed such indicators of support as the size of the voting and opinion majority, the
number of dissenting votes and opinions, and the author of the majority opinion. He
calculated various correlation coefficients for these indicators and lower court treatment of
precedent. Contrary to his hypotheses, however, none of these relationships was statistically
Si gniﬁcant.

The study conducted here is similar to that of Johnson (1979), as Shepard ’s is utilized
to code lower court treatment of precedent and the inﬂuence of Court support is considered
as a motivating force in this treatment; but it differs in that a multivariate model is speciﬁed
and estimated, incorporating both legal and attitudinal factors which are hypothesized to
affect the behavior of circuit court judges. My ﬁndings also differ from those reported by
Johnson.

The contrast in ﬁndings between this study and Johnson’s (1979) may be attributable
to the different ways in which Court support is operationalized. Johnson utilized continuous
variables in his measurements; however, my data indicates that the inﬂuence of Court
support is strongest at the extremes, i.e., in unanimous decisions and decisions reached by
a one-vote margin, and that there is little variation in the treatment of precedents established
by other decisional coalitions (see Table 4.2).

Additionally, Johnson’s measures may not accurately capture the factors which limit
the authority of precedent in the lower courts. It is hypothesized here that it is concurring,
rather than dissenting, opinions which detract from the majority opinion. A justice who

concurs in the Court’s decision desires the same outcome as the other justices in the majority

77
but would reach that outcome for different reasons. Concurring opinions are likely to
provide circuit court judges with the bases upon which to distinguish precedents in their
decisions.

In order to evaluate the applicability of legal and attitudinal models to Courts of
Appeals treatment of Supreme Court precedent, I test a number of hypotheses. Hypotheses
1 and 2 relate to the strength or authoritativeness of Supreme Court precedent. According
to the research of Wasby (1970), Rohde (1972), and others, lower court judges will be
constrained from deciding cases on the basis of their beliefs when the Court is unequivocal
in issuing a decision. In other words, the stronger the precedent, the more likely it is that
circuit court judges will abide by stare decisis, regardless of whether they agree with the
precedent. I quantify the strength of precedent in terms of the support on the Court for a
decision.

Hypothesis 1. There is a positive relationship between a
unanimous Supreme Court precedent and the following of
that precedent in the Courts of Appeals.

A judge who does not follow a unanimous Court decision is more likely to provoke
the Court into hearing and reversing the judge’s decision (Songer, Segal, and Cameron
1994); wishing to avoid this circumstance, the judge is more likely to follow the precedent
(Johnson 1979; Pacelle and Baum 1992).

Hypothesis 2. There is a negative relationship between a
Supreme Court precedent which was established by a one-
vote margin and the following of that precedent in the Courts
of Appeals.

A 5-4 or 4-3 decision is weak in that the members of the Court barely agree on the

outcome. A lower court judge who disagrees with this decision will not be as likely to

78

follow it, in part because he realizes that the chances of reversal are slim (Songer, Segal, and
Cameron 1994). Moreover, the authority of the decision could be in doubt, since the justices
are closely divided (Johnson and Canon 1984).

Similar to Hypotheses I and 2, a lower court judge is more likely to follow a Supreme
Court decision which is clear and unambiguous (Baum 1978). Hypotheses 3, 4, and 5
employ alternative measures of decisional clarity.

Hypothesis 3. There is a negative relationship between the
number of concurring opinions that the Supreme Court issues
in establishing a precedent and the following of that precedent
in the Courts of Appeals.

When the Court issues a decision in which a number of justices write separate,
concurring opinions, there apparently is some dissension on the Court with respect to the
legal reasoning underlying the decision. As a result, circuit court judges will perceive that
this is not a clear precedent; therefore, they will be less likely to follow such a decision than
when the Court is united and issues few, if any, concurring opinions.5

Hypothesis 4. There is a negative relationship between a
complex Supreme Court precedent and the following of that
precedent in the Courts of Appeals.

The literature is somewhat contradictory on this point, as some scholars argue that

the greater the complexity of a Court decision, the less likely the decision is to be followed

 

5The following is offered as anecdotal support for this hypothesis:

The Chief did not like concurrences. He felt that they were
often nitpicking, that they added little to the law, and that they
split majorities... They confused as much as they enlightened
lower court judges, Burger felt. Also, concurrences detracted
from the main opinion... (Woodward and Armstrong 1979,
58).

79

below (Wasby 1970); others (Johnson 1987, 333) depict complex precedents as providing
“many ‘handles’ that lower court judges may grasp to justify their decisions.” Hypothesis
4 espouses the former view, asserting that complicated Supreme Court decisions will foster
confusion in the lower courts and will not be applied consistently.

Hypothesis 5 concerns constitutional issues in a case.

Hypothesis 5. There is a negative relationship between a
Supreme Court precedent which interprets a constitutional
provision and the following of that precedent in the Courts of
Appeals.

This hypothesis is justiﬁed for two reasons. First, the language of most constitutional
provisions is somewhat ambiguous and has been subject to a wide range of interpretations
(Solan 1993). Second, as constitutional provisions generally are controversial, they “may
be precisely those in which the impact of the Supreme Court on lower courts is at its
minimum” (Songer 1987, 830).

The next hypothesis concerns the Signiﬁcance of the issues within a case.

Hypothesis 6. There is a positive relationship between a
salient Supreme Court precedent and the following of that
precedent in the Courts of Appeals.

Circuit court judges are more likely to follow decisions with important policy
implications (Johnson 1979; George 1997). When a lower court does not abide by a salient

precedent, it is more likely to gain the attention of the Court, thus increasing the likelihood

of review.6

 

61 recognize Baum’s (1978, 215) claim that “policies that do elicit widespread non-
adoption are likely to be signiﬁcant ones, because minor policies generally lack the capacity
to motivate strong opposition.” Despite the apparent logic of this assertion, due to the
increased likelihood of review, I maintain that highly salient cases are more likely to be

80

In addition to the above hypotheses which concern aspects of the legal model,

attitudinal factors should be pivotal as well.
Hypothesis 7. There is a positive relationship between an
appellate panel’s ideological consistency with the direction of
a Supreme Court precedent and the following of that
precedent in the Comts of Appeals.

Very simply, Courts of Appeals judges will vote attitudinally if at all possible, as they
will be more likely to follow decisions with which they agree (Segal and Spaeth 1993).

Furthermore, attitudes should be more inﬂuential when certain issues are at bar. As
the following hypothesis demonstrates, consistent with the ﬁndings of Segal and Spaeth
(1993), attitudes should matter most in civil rights and liberties and criminal procedure cases.

Hypothesis 8. In civil rights and liberties and criminal
procedure cases, there is a positive relationship between an
appellate panel’s ideological consistency with the direction of
a Supreme Court precedent and the following of that
precedent in the Courts of Appeals; in economic activity
cases, this relationship is not as strong.

Civil rights and liberties and criminal procedure cases generate potentially intense
beliefs among judges. In these cases, therefore, appeals court judges should be more likely
to follow their own preferences as opposed to Supreme Court precedent. At the other
extreme, ideology should be less important in cases involving economic activity and other
issues. Since these cases raise fewer passions, appeals court judges are more likely to abide
by the norm of deference (Songer and Sheehan 1990; Segal and Spaeth 1993).

Finally, the changing makeup of the Supreme Court should make a difference in

whether Courts of Appeals judges apply precedent.

 

followed by lower courts.

81

Hypothesis 9. There is a positive relationship between a
Supreme Court precedent which has gained greater
ideological support on the Court, due to changes in Court
composition, and the following of that precedent in the Courts
of Appeals.

Generally Speaking, circuit court judges will feel less constrained by a Speciﬁc
precedent when it appears that the current Court may be less likely to support that decision,
as the likelihood of review Should decrease in that instance (Wasby 1970).

These hypotheses incorporate both legal and attitudinal components, as both types
of factors should inﬂuence the decision making of circuit court judges. The
operationalization of these factors is described in detail in Chapter 2 (see pp. 30-36) and is
summarized in Table 4.1. Table 4.1 also Speciﬁes the direction of the relationships which
are hypothesized to exist between these legal and attitudinal variables and the treatment of
precedent. In the next section, I estimate a multivariate logistic regression model in order

to determine the extent to which these attitudinal and nonattitudinal variables motivate the

treatment of precedent in the Courts of Appeals.

FINDINGS
In light of the norm of deference to hierarchical authority, it is not surprising that the
Courts of Appeals followed Supreme Court precedent in 70.8% of the cases in our sample.
Table 4.2 provides preliminary evidence of the importance of the strength and clarity of
precedent, as well as considerations of Supreme Court review, in moderating this obligation
to adhere to precedent. According to the bivariate relationships between treatment and

characteristics of Supreme Court decisions, an ambiguous precedent is not afforded such

82

favorable treatment in the Courts of Appeals. The rate of following precedent drops to
57.7% when two concurring opinions are written, to 60.5% when the case involves multiple
issues and legal provisions, and to 65.5% when the application of a constitutional provision
is involved. Precedents which issue from a sharply divided Court (i.e., 5-4 decisions) are
followed only 64.2% of the time. In reaching decisions which are likely to gain the attention
of the High Court, the circuit courts are more likely to follow precedent, i.e., in 73.1% of
such cases. These ﬁndings illustrate the importance of legal elements in Courts of Appeals
decision making and suggest the need for more sophisticated analysis.

Incorporating these legal factors, along with the ideological measures already
discussed, into a model of appeals court treatment of precedent permits a determination of
the effects of each of these factors when other potentially inﬂuential factors are taken into
account. The following model is estimated:

P(Treatrnent=1) = b0 + b,(Unanimous) + b2(One- Vote Margin) +
b3(Concurring Opinions) +b4(Complexity) + b5(Constitutional Provision) +
b6(Salience) + b7(Ideological Consistency) + b3(Ideological Consistency x Civil Rights) +
b9(1deological Consistency X Criminal Procedure) +
b,0(Ideological Consistency >< Economic Activity) +
bl 1(Change in Supreme Court Composition)

Each bi indicates the change in the log of the odds of following precedent due to changes in
the value of variable i, controlling for the values of the other variables. It is hypothesized

that b1, b6, b7, b8, b9, bio, and bll > 0 and that b2, b3, b4, and b5 < 0.

Table 4.3 presents the estimated coefﬁcients and signiﬁcance levels for this logistic

regression model which predicts whether the Courts of Appeals will follow a particular

Supreme Court decision. A likelihood ratio test, which tests the null hypothesis that all

coefﬁcients in the model except the constant are zero, indicates that the block of coefﬁcients

83

is signiﬁcantly different from zero at p<0.001.7 Each coefficient achieves statistical
signiﬁcance at p<0.05 and is in the expected direction. The model correctly classiﬁes 71.3%
of appeals court decisions.8

Appeals courts are less likely to adhere to precedent in cases decided by a one-vote
margin. Additionally, as the nmnber of concurring opinions the justices write in a given case
increases, the likelihood that the decision will be followed decreases. Circuit courts are also
less likely to follow precedent in cases involving constitutional provisions and cases dealing
with complex subject matter.

Consistent with the ﬁndings of Segal and Spaeth (1993) regarding Supreme Court
decision making, in both civil rights and liberties cases and criminal procedure cases, Courts
of Appeals judges are motivated by their personal policy goals. Speciﬁcally, appellate
panels become more likely to follow precedent in civil rights and criminal procedure cases

as ideological agreement with Supreme Court decisions increases. They are also more likely

 

7The likelihood ratio test statistic is calculated as -210g(Lo/L,), where L1 is the
likelihood ﬁrnction for the full model as ﬁtted and L0 is the maximum value of the likelihood
function if all coefﬁcients except the constant are zero. The test statistic is xz-distributed,
with degrees of freedom equal to the number of independent variables.

8The model achieves only a 1.71% reduction in error ﬁ'om predicting treatment based
upon the proportion of decisions followed. Percent reduction in error is calculated according
to the following formula:

100 X W
100 - % in modal category

It is important to point out that the goal in developing this model is not to improve
predictive accuracy with respect to circuit court treatment of precedent. Rather, the purpose
is to identify circumstances under which circuit court decisions will follow precedent and
when they will avoid precedent. The model performs extremely well in this regard.

84

to adhere to precedent when the Supreme Court has become more predisposed ideologically
toward that precedent. Finally, appeals courts are more likely to defer to the Supreme Court
in cases that have important policy implications.

In estimating the model, likelihood ratio tests were calculated in order to determine
whether some of the variables had zero coefficients and thus should be deleted from the
model.9 Interestingly, these tests indicated that Ideological Consistency and Unanimous did
not contribute to the explanatory power of the model. The expectation that ideology was less
important in economic cases was confrrmed, as these tests indicated that Ideological
Consistency >< Economic Activity was also a redundant variable.

Interesting results obtain when the probability of following precedent under certain
conditions is calculated. To Show the substantive impact of the independent variables, I ran
several simulations to predict the likelihood that a circuit court would follow a Supreme
Court decision. By holding the other independent variables constant at their mean or modal

values and varying the values of the variables of interest, one can demonstrate how these

 

’Ihe likelihood ratio test which was calculated above as a measure of goodness-of-ﬁt
can also be used to determine whether individual coefﬁcients are zero. In such a test, L1 is
the value of the likelihood function for the complete model, and L0 is the value of the
likelihood function for the reduced, or nested, model. The number of degrees of freedom is
equal to the number of coefﬁcients hypothesized to be zero.

When Ideological Consistency was omitted from the model, the likelihood ratio was
0.483 with p=0.487. When Ideological Consistency X Economic Activity was omitted, the
likelihood ratio was 0.358 with p=0.549. When Unanimous was omitted, the likelihood ratio
was 0.016 with p=0.901.

85

variables of interest affect the probability that a circuit court will defer to Supreme Court
authority.” Table 4.4 presents these predicted probabilities.

I highlight some of the more intriguing aspects of this table here. The baseline
probability of following precedent indicates that in the most common circumstances, i.e.,
when the independent variables take on their mean or modal values, precedent will be
followed 74% of the time. Complexity has the most dramatic effect on the likelihood of
following precedent. Speciﬁcally, when a Supreme Court decision deals with multiple issues
and/or legal provisions, the probability that the decision will receive favorable treatment in
the Courts of Appeals drops to 0.56. On the other hand, Salience provides the most
substantial increase in the probability of following a Supreme Court decision, in that judges
will follow precedent established in highly Signiﬁcant cases 87% of the time. When the
justices agree with the legal reasoning set forth in the opinion of the Court and no concurring
opinions are written, precedent is also more likely to be followed than under normal
conditions, at a rate of 76%; this ﬁgure drops to 65% when three concurring opinions are
written.

In order to determine the degree of conﬁdence that may be placed in these ﬁndings,
some diagnostic tests should be performed. One should assess the leverage that particular
patterns of independent variables have on the coefﬁcient estimates, to ensure that certain
combinations of values are not exerting disproportionate inﬂuence on the results. Appendix

B details the diagnostic checks that were performed. Speciﬁcally, these diagnostics indicate

 

”The appropriate equation is 1/(1 + exp(-Xﬂ)).

86

that certain cases in the data set are poorly ﬁt by the model and are exerting undue leverage
on the results.

Several strategies for treating such inﬂuential cases have been suggested. One is to
consider what makes these cases unusual, which may lead to better model speciﬁcation. I
examined the outlying cases in the data but found nothing to suggest that an important
variable had been omitted from the model. Another approach to dealing with outliers is to
delete problematic cases. While this would allow a better ﬁt of the remaining data, it
obviously is not an appropriate solution (Granato, Inglehart, and Leblang 1996).

Some methodologists, (e. g.,Hamilton 1992; Granato, Inglehart, and Leblang 1996)
have suggested an alternative estimation procedure -- bounded inﬂuence estimation.
Bounded inﬂuence estimation is one type of robust estimation. When applied to outlier-
ﬁlled data, robust techniques are more efﬁcient than traditional techniques and give more
accurate conﬁdence intervals and tests (Western 1995). With bounded inﬂuence estimation,
the inﬂuence of outlying cases on coefﬁcient estimates is reduced, or bounded, based upon
the magnitude of their inﬂuence. In practical terms, poorly ﬁtting observations with large
deviances are given less weight in the estimation. This weighted estimation is repeated,
updating the weights with each iteration, until there is little difference from one iteration to
the next. Such a technique is appropriate for this model.

Bounded inﬂuence methods are designed to resist the leverage of unusual values of
the independent variables. Hamilton (1992, 207) suggests a “quick-and-dirty” bounded
inﬂuence technique which assigns weights to cases on the basis of leverage, and I executed

this technique here. First, I identiﬁed the 10% of the cases in the data set that are most

87

inﬂuential based upon the values of hj, the leverage of pattern j. For the remaining 90% of
the data set, h]. was less than 0.166. I then assigned a weight of 1 to any case wherejh <
0.166, and a weight of(0.166/hj)2 to any case where hi 2 0.166.” Finally, I reestirnated the
logistic regression, weighting the cases by these values.

There was little difference in results between the logistic regression and bounded-
inﬂuence logistic regression models. Table 4.5 allows a direct comparison of these results.
Generally Speaking, in the bounded inﬂuence model, the values of the coefficients are
slightly smaller, but the standard errors are smaller as well. Notably, all of the coefﬁcients
are in the hypothesized direction and remain statistically signiﬁcant at p<0.05. Because the
bounded inﬂuence results represent the “bulk of the data, not just a few outliers” (Hamilton
1992, 211), and because these robust ﬁndings are similar substantively and statistically to
the logit results, I am conﬁdent that the model ﬁts the data well, in spite of some outlying
cases revealed by the diagnostics.

It is interesting, as well as instructive, to consider some empirical evidence in support
of these ﬁndings or, in other words, to consider how these considerations play out in the
decision making of appeals court judges. In Bowen v. Kendrick, 487 US. 589 (1988), the
Supreme Court upheld against an establishment clause challenge the Adolescent Family Life

Act, which prohibits federal funding of organizations involved with abortions, while

 

“1 17 cases were assigned a weight of 0.70, 11 cases were assigned a weight of 0.98,
and 102 were assigned a weight of slightly less than 1. Thus, a total of 230 cases received
less than full weight. Hamilton (1992) suggests that the magnitudes of the weights may be
instructive as diagnostic tools. The sizes of the weights assigned here indicate that, for all
but the 117 cases in our data set which were assigned a weight of 0.70, leverage is not a
serious problem.

88

allong funding of religious groups advocating self-discipline as a form of birth control.
This decision was rendered by a vote of 5-4, two concurring opinions were authored, and a
constitutional provision was at issue. These factors have been identiﬁed as decreasing the
likelihood that such precedents will be followed. However, the decision also involved a civil
liberties issue, and such cases tend to activate the personal preferences of judges. Consistent
with this expectation, in the six appeals court cases which applied Bowen v. Kendrick, this
precedent was followed when ideological consistency was 67% or 100% (three cases), and
it was not followed when ideological consistency was 33% (three cases). These ﬁndings
indicate the importance of judges’ policy goals in the resolution of certain types of cases.
While four of these Six decisions involved the issue of taxpayer standing to bring suit,
the remaining two involved Kendrick ’s interpretation of the “principal or primary effect”
prong of the Lemon test. The Kendrick Court stated that a statute is void if “motivated
wholly by an impermissible purpose” (487 US. 589 at 602). One circuit court opinion,
issued by a liberal-dominated panel, labeled this statement as dictum. In holding a city
ordinance unconstitutional, the court relied instead on the less restrictive standards of other
establishment clause decisions.” In contrast, a second decision turned to Kendrick for
“guidance for the appropriate application of the purpose prong of the Lemon test” and found
that the statute in question satisﬁed this prong.l3 One judge, a Carter appointee, dissented

from the panel’s opinion in this case. She questioned the majority’s selective reliance on

 

”Church of Scientology Flag Service Org., Inc. v. City OfClearwater, 2 F.3d 1514
(11th Cir. 1993).

l3Cammack v. Waihee, 932 F .2d 765 (9th Cir. 1991); quotation at 775.

89

Kendrick and advocated a “primary” or “actual” purpose test, concluding under this approach
that the statute was unconstitutional. These are instances in which appeals court decision
making appears to be ideologically driven, as the model speciﬁed here suggests.

Another example of the utility of this model is found in the treatment of the Court’s
decision in Shapero v. Kentucky Bar Association, 486 US. 466 (1988). Here the Court held
that a state may not categorically prohibit lawyers from soliciting business by sending
truthful letters to individuals known to face particular legal problems, but it may require
lawyers to ﬁle a copy of solicitation letters with a state agency. The Court’s opinion
received only limited support, in that Six justices Signed on to Parts I and II, but only four
justices signed on to Part III. In addition, the application of the ﬁrst and fourteenth
amendments was in question. In light of the results presented here, it is not surprising that
the Shapero precedent was followed in only three of the Six circuit court decisions in which
it was treated.

The reasoning of those decisions which did not follow the Shapero precedent is
illustrative of the tactics employed by lower courts to limit the application of decisions which
are equivocal or of doubtful authority. In one decision, the appellate panel ruled that
solicitation laws requiring charitable organizations to provide a state agency with copies of
any oral solicitation violated the ﬁrst amendment.” The panel refused to extend the Shapero
decision to charitable solicitations, as solicitation by lawyers was commercial Speech and not
subject to the exacting scrutiny required for charitable solicitations. Another case dealt with

the constitutionality of a state law disallowing access to criminal justice records to be used

 

l"Telco Communications, Inc. v. Carbaugh, 885 F.2d 1225 (4th Cir. 1989).

90

for direct mail advertising by attorneys, but allowing such access for noncommercial
purposes.'5 The appeals court upheld the statute on the basis of the privacy issues involved
and distinguished the Shapero precedent, as the attorneys in Shapero presumably already
possessed the information necessary to solicit. In a third decision, the circuit court held that
the case was not yet ripe for review, an issue that the appellate panel raised sua sponte.'6 In
all three of these decisions, the Courts of Appeals relied upon questionable grounds to avoid

compliance with a relevant, but legally ambivalent, precedent.

DISCUSSION

One of the more interesting ﬁndings of this study is that attitudinal factors, as
measured by ideological consistency, are not generally explanatory with respect to behavior
in the Courts of Appeals. In fact, as statistical tests demonstrated (see fit. 9), ideology has
no independent effect on appeals court decision making across the broad range of issues
considered. Attitudes have a profound inﬂuence when civil rights and liberties and criminal
procedure issues are at play; but when other issues are applicable, ideology has no effect, as
legal factors seem to be most inﬂuential here. Thus, Hypothesis 7 with respect to ideological
consistency is not supported by these ﬁndings. On the other hand, Hypothesis 8, which deals
with the importance of ideology in particular issue areas, is fully conﬁrmed by these results.

Unanimity carried little weight with the circuit courts, as there was almost no

distinction between unanimous decisions and those decided by more narrow margins. In

 

l5Lanphere & Urbaniak v. Colorado, 21 F .3d 1508 (10th Cir. 1994).

”Felmeister v. Oﬂice of Attorney Ethics, 856 F.2d 529 (3rd Cir. 1988).

91

fact, contrary to Hypothesis 1, the unanimity variable carried so little statistical weight that
it was dropped from the ﬁnal model (see ﬁt. 9). Competing forces may be at work with
respect to unanimous decisions. Some of these decisions may be straightforward and
unequivocal, so that there is no basis for disagreement. Others may be the product of
compromise, where the justices weaken the language of the majority opinion in order to
achieve consensus and present a united front. In any event, this is yet another study which
has demonstrated that, contrary to popular belief, unanimous decisions are not provided any
greater deference in the lower courts.

Signiﬁcantly, however, appellate panels are much leSS likely to follow close Supreme
Court decisions. Indeed, there was a clear distinction between lower court treatment of cases
decided by one-vote margins and virtually all other coalitions. Apparently, when they
confront a close case (i.e., 5-4 or 4-3), judges perceive either that there is dissension on the
Court or that the Comt is less inclined to reverse a contrary decision below, Since a change
in Court membership or a one-vote swing could alter the precedent. In any event, it seems
that appeals courts feel less constrained and work within their “room to maneuver” (Songer,
Segal, and Cameron 1994, 693) when the Court is closely divided. Hypothesis 2, therefore,
is conﬁrmed by this analysis.

Hypotheses 3, 4, and 5 set out various measures of the clarity of a Supreme Court
decision, and each of these three hypotheses is borne out by the results. Speciﬁcally, circuit
court panels are less likely to follow a decision when an increasing number of concurring

opinions are written, when complex subject matter is involved, and when constitutional

92

provisions are at issue. These ﬁndings, therefore, strongly suggest that lower courts are
much less likely to follow ambiguous as opposed to clear Court precedent.

Finally, as for Hypothesis 6, these ﬁndings indicate that decisions with important
policy implications are more likely to be followed by circuit courts. Accordingly, and
contrary to Baurn’s (1978) contention (see ﬁr. 6), this hypothesis is conﬁrmed. Hypothesis
9 is also supported by the results; when lower courts perceive that a prior Court decision may
not be supported by the current Court, they do not feel as obliged to follow that decision.

These results suggest that attitudinal factors are not universally instructive in
explaining decision making in the Courts of Appeals, as ideology has an impact only in civil
rights and liberties and criminal procedure cases. In other issues areas, legal factors appear
to control behavior, which is partially demonstrated by judges following precedent at better
than a 70% clip. Lower court judges abide by the norm of deference, at least in some types
of cases, as evidently that is a part of what judges do. In any event, these results indicate that

both legal and attitudinal variables are important in explaining outcomes in the circuit courts.

93

 

000.0000 0.500
0:00:03 05 .00 0000000 05 w:_0000&00:_ .8500

5:600:80

 

+ 030 0003000 80000 30.002 :02: 5 00000005 0.500 08055 :_ 0w0000
00200
0.500 080.05 05 .00 00:00.00 05 605 30030—002
+ 0000w0 005 3:0: 80:0000 05 mo 0m0000000m 5:036:00 _0o_m0_000_ 000500.005 30:03
.00E 003 £010 0:280 0:08 :0 2 00003
+ 00000 0.500 0:00:05 $00205 0E0t0> ugh—5 00:0:0m 00:00¢_:w_m 5:8
000009005 003 5630:: 3:003:88 0 0055
I 00000 0.500 0:00:95 w:_000_0:_ 03010,» 03:55 :20305 000030200
00302:
0:03 003$ 000:0 0:00.032: _0w0_ 000:: :0 m 000:3
I 0000 EEO 080003 w:u00_0:_ 05019 090805 baa—9:00
.0000 :30 0:00:05
I 0 :_ 00:02 0:03 005 002500 w:_t:0:00 .«0 00:52 0:258 w:_t:0:0u 8000005 00 5:05
.mé :0 Tm .00 08> 0 3 000600 0:03
I 006. 00000 0.500 0:00:96 $50205 0300.0,» 035:5 £302 000>-0:O
.00 ..0 .04. .0-” 6-0 00 08> 0 .3 002000 0.03
+ 005 00000 00:00 050095 $000205 05000:, 9:55 30830:: 3000005 ‘00 0%:00m
8000005 m530=0m 8080000002 5000200000000 0:00:00

00 038000—3—

 

EMm—H. hwoﬁ mIH ”mm4m<_m<> meahmm mEImZOF<4mM J<ZOEOMM=Q QZ< Wham—0200 m0 ZO_._.<N5<ZOF<MmmO

:0 mam/Er

94

TABLE 4.2
BIVARIATE RELATIONSHIPS BETWEEN TREATMENT AND LEGAL FACTORS: THE 1987 TERM

 

 

 

 

 

 

 

NOT FOLLOWED FOLLOWED
UNANIMOUS 27.3% (172) 72.7% (457)
7 31.0% (39) 69.0% (87)
6 22.4% (13) 77.6% (45)
Von: 5 45.8% (11) 54.2% (13)
MARGIN
4 28.9% (56) 71.1% (138)
3 27.0% (20) 73.0% (54)
2 27.8% (40) 72.2% (104)
I 35.8% (78) 64.2% (140)
0 28.9% (241) 71.1% (592)
0 0
CONCU G 1 26.54 (132) 73.54 (367)
OPINIONS 2 42.3% (55) 57.7% (75)
3 20.0% (1) 80.0% (4)
YES 34.5% (123) 65.5% (234)
CONSTITUTIONAL
PROVISION NO 27.6% (306) 72.4% (804)
0 0
Comm-mm YES 39.54 (30) 605/0 (46)
NO 28.7% (399) 71.3% (992)
0 0
SAUENCE YES 26.9/o (32) 73.1 /o (87)
NO 29.5% (397) 70.5% (951)

 

 

 

 

 

95

TABLE 4.3

LOGISTIC REGRESSION OF TREATMENT: THE 1987 TERM

 

 

 

Explanatory
Variable Coefﬁcient Standard Error t Statistic
_Constant 0.961 0.092
i ' F t
Ideological Consistency ><
Civil Rights and Liberties 0.631 0.262 2.411
Ideological Consistency ><
Criminal Procedure 0.844 0.254 3.326
Change in Supreme Court
Composition 0.712 0.299 2.382
W15
One-Vote Margin -0.550 0.180 -3.057
Concurring Opinions -0.177 0.090 -1.977
Constitutional Provision -0.3 83 0. 155 -2.471
Complexity -0.821 0.287 -2.860
Salience 0.791 0.254 3.107
N = 1467
12 (8) = 45.72

Log-likelihood = -863.672

p > x2 = 0.000

96

TABLE 4.4
PREDICTED PROBABILITIES OF FOLLOWING PRECEDENT: THE 1987 TERM

 

Baseline 0.744

Ideological Consistency in Civil Rights Cases

100% 0.819

Ideological Consistency in Criminal Procedure Cases

100% 0.848

Change in Supreme Court Composition

-0.35 0.692

0.35 0.787
One-Vote Margin 0.627
Concurring Opinions

0 0.762

3 0.653
Constitutional Provision 0.665
Complexity 0.562

Salience 0.865

 

97

 

 

52...” Sad dwnd hdﬁm emmd and 852%
gm.” vwmd w ﬂ wd- dowd- Smd Sad- bio—980
Ndmd- mm _ d mwmd- Kim- mEd mwmd- 53>on 3533580
33. ago Ed- E. _- 890 Ed- 8258 3588
add- de demd- hmdd- 22d dmmd- EmEE Sex/-25
and»; o

ddvd mdmd ﬂ Ed mwmd ddmd NKd 8368800
2:00 08055 E omega

Ema End Quad ommd emmd Edd 9:63on 35:90
x 5:038:00 323303

weed mmmd vmdd SYN momd Red 8:535 98 magma :30
x zocoammmaoo 323303
gag
god nmdd mood Edd ESESUI

owmuﬁm N “ohm dadgm BBQ—boom 2335 u Sum dadgm 820580 28353 bounce—axm

commmocwom owmmwoq cocoa—m5 dovgom commmouwom onmmwou—

 

same 32 m5 ”205355“ 0550A 82%qu omQZDOm .m> zoammxoma 2.553
We mqmst

CHAPTER 5
COURTS OF APPEALS’ TREATMENT OF CIVIL RIGHTS AND LIBERTIES

DECISIONS FROM 1966-1975: A MULTIVARIATE ANALYSIS AND
A COMPARISON OF TREATMENT AMONG AND WITHIN CIRCUITS

The analysis of Chapter 4 indicated that Courts of Appeals judges behaved differently
when civil rights and liberties issues were under consideration. Speciﬁcally, circuit court
judges were more likely to be motivated by their personal policy goals in such cases. In light
of these ﬁndings, it is important to undertake a closer examination of decision making in
these cases, in order to determine whether ideological values predominate over legal forces
in inﬂuencing appeals court treatment of precedent.

The civil rights and liberties decisions investigated here include decisions of both the
Warren and Burger Courts. Cases involving civil rights and liberties issues dominated
overwhelmingly the agenda of the Burger Court, and on the Warren Court agenda they were
second only to economic activity cases (Segal and Spaeth 1993, ch. 6). While Warren Court
decisions in these cases are identiﬁed with “freewheeling libertarianism” (Schubert 1974,
198), the decisions of the Burger Court are characterized by “cautious conservatism” (Segal
and Spaeth 1993, 117). The Warren Court favored individual over community rights and
national over state power; alternatively, the Burger Court exhibited preferences for
community rights and state power. The Warren Court made ten of the speciﬁc guarantees

of the Bill of Rights applicable to the states, but the Burger Court limited or denied federal

98

9.3- "-

99

protection to several of these provisions (Walker 1978). One political scientist (Steamer
1976, 5) provides this somewhat facetious description of the distinction between the two
Courts:

The Warren Court liked criminals, minorities, and people who

wear sweatshirts with four letter words on them. It did not

like police, prosecuting attorneys, or draft card burners. The

Burger Court like[d] the police (although not enough to

permit them to wear long hair), welfare people, Indians,

women, and drugstores...The Burger Court dislike[d]

prisoners, shopliﬁers, pushcart dealers, homosexuals, and

human fetuses during the ﬁrst three months of gestation.

The views of Court commentators have been conﬁrmed by empirical analysis.
According to data provided in Segal and Spaeth (1993, ch. 6), the percentages of liberal
decisions rendered by the Warren Court in civil rights, First Amendment, due process, and
privacy cases were 77.1, 71.7, 55.8, and 50.0, respectively. The percentages of liberal
decisions announced by the Burger Court in the same issue areas were 51.7, 49.2, 35.6, and
28.6. The stark contrast in the policy making of the Supreme Court under these two Chief

Justices provides fertile ground for analyzing lower court treatment of the precedents

established by these Courts.

HYPOTHESES
According to the ﬁndings in Chapter 4, the attitudes and values of Courts of Appeals
judges are important determinants of their decisions in civil rights and liberties cases.
However, as their decisions are subject to Supreme Court review and as they may have
ambitions of membership on the Supreme Court, circuit court judges are constrained in their

capacity to follow their own preferences. The hypotheses tested here incorporate legal forces

100
which may come into play in appeals court treatment of precedent. These legal
considerations include the strength and clarity of precedent and the political and social
signiﬁcance of precedent. Hypotheses which control for the inﬂuence of ideological factors
are also speciﬁed.

The ﬁrst four hypotheses deal with the strength of a particular precedent, in terms of
the authority which is attributed to a precedent. As Baum (1976, 101) notes, “acceptance of
authority represents one of several motivations which help to determine the treatment of
appellate decisions.” A precedent which is viewed as carrying the full authority of the High
Court should be more readily espoused by the lower courts.

One indicator of the strength or authority of a decision is the level of support a
decision garnered from the Court’s members (Johnson 1979; Johnson and Canon 1984, 58).
Level of support is operationalized in Hypotheses 1 and 2.

Hypothesis 1. There is a positive relationship between a
unanimous Supreme Court civil rights and liberties precedent
and the following of that precedent in the Courts of Appeals.

A number of scholars have suggested that the Supreme Court may strive to achieve
unanimity in decisions which have important policy implications and which may encounter
resistance in their implementation (see, e. g., Rohde 1972; Hutchinson 1979). Consensus in
decision making may enhance the perceived legitimacy of the decision and may lend the
decision greater weight in the lower courts (Johnson and Canon 1984, 58). Hypothesis 1
asserts that unanimous decisions will be treated more favorably in the circuit courts.

On the other hand, a decision to which the justices cannot agree may raise questions
in the lower comts regarding the legal correctness of the decision. Hypothesis 2 concerns

precedents which were established by a minimum winning coalition on the Court.

101

Hypothesis 2. There is a negative relationship between a
Supreme Court civil rights and liberties precedent which was
established by a minimum winning coalition and the
following of that precedent in the Courts of Appeals.

In these decisions, a substantial bloc on the Court favors an alternative outcome. As
a result, lower court judges may be reluctant to comply with such rulings, since these rulings
could change in conjunction with membership turnover on the Court (Marshall 1989; Pacelle
and Baum 1992).
Hypothesis 3 deals with the Supreme Court’s treatment of precedent in later
decisions.
Hypothesis 3. There is a negative relationship between a
Supreme Court civil rights and liberties precedent which has
been modiﬁed substantially by subsequent decisions and the
following of that precedent in the Courts of Appeals.
There are two justiﬁcations for including this variable in the model. First, it is

important to control for ensuing alterations in precedent, as stare decisis only binds judges

to current doctrine.l Second, a number of studies have highlighted the importance of a clear

 

'There is not a perfectly collinear relationship between doctrinal modiﬁcation and the
treatment of precedent in the circuit courts. Many of these modiﬁed decisions are not
explicitly overruled, and more signiﬁcantly, appeals court opinions continue to refer to them
for guidance even when they have been substantially altered.

One example of this behavior is found in the Courts of Appeals’ treatment of
Rosenbloom v. Metromedia (1971). As discussed in Chapter 2 (see pp. 39-40), the
Rosenbloom Court held that the actual malice standard in libel cases applied to private
persons who had become involved in issues of “public or general concern” (403 US. 265 at
44). However, in Gertz v. Welch (1974), the Court implicitly overruled Rosenbloom, noting
the difﬁculty of requiring judges to determine which issues were and were not of “public or
general interest” (Id at 41). The Court went on to delineate three classes of “public ﬁgures”
to whom the actual malice standard would apply, one of which was individuals who “have
thrust themselves to the forefront of particular public controversies” (418 US. 323 at 345).

Even after the Gertz decision, circuit court opinions continued to cite Rosenbloom

102
and consistent line of precedent in achieving lower court compliance (Gruhl 1980; Johnson
and Canon 1984, 51; Johnson 1987; Segal, Songer, and Cameron 1995). When a Supreme
Court decision has been altered or obscured by later decisions, such revisions should be
acknowledged by the circuit courts.

Some scholars have reasoned that, over time, the authority of a particular precedent
is eroded by changes in Court personnel and doctrine (Marshall 1989; Pacelle and Baum
1992). On the other hand, some researchers have argued that precedent becomes more
forceful as it is clariﬁed in subsequent decisions and applied in other areas (Johnson 1987).
Hypothesis 4 takes this view, stipulating that a precedent grows stronger, and thus is treated
more favorably in the lower courts, with the passage of time.

Hypothesis 4. There is a positive relationship between the
length of time that has passed since a Supreme Court civil
rights and liberties precedent was established and the
following of that precedent in the Courts of Appeals.

In addition to the strength or authority of precedent, a second factor which may
contribute to divergent lower court responses to Supreme Court directives is the clarity with
which the Court makes these directives known. Baum (1976, 92) argues that ambiguity in

decision making may preclude faithful implementation of decisions in two ways: “by leaving

loyal judges uncertain as to their superior’s intent and by providing leeway which recalcitrant

 

as authority. Note a portion of the First Circuit’s opinion in Bruno & Stillman, Inc. v. Globe
Newspaper Co., 633 F.2d 583 (lst Cir. 1980):

Distinguishing what would be “an issue of public or general
concern” under Rosenbloom from a “public controversy”
under Gertz is not a clearcut task...[W]e ﬁnd ourselves
returning to the job from which the Court in Gertz felt it had
liberated us (Id. at 590).

103

judges may use to evade obedience to a directive.” Hypotheses 5, 6, and 7 relate to the
importance of an unequivocal Supreme Court decision. In these hypotheses, clarity is
operationalized in three different measures: the number of concurring opinions that are
written, whether a constitutional provision is at issue, and whether the decision is rendered
by the Burger Court.

Hypothesis 5. There is a negative relationship between the

number of concurring opinions that the Supreme Court issues

in establishing a civil rights and liberties precedent and the
following of that precedent in the Courts of Appeals.

Concurring opinions present differing viewpoints and provide alternative
interpretations of legal provisions. These individual expressions may serve to weaken the
majority opinion.

Hypothesis 6. There is a negative relationship between a
Supreme Court civil rights and liberties precedent which

interprets a constitutional provision and the following of that
precedent in the Courts of Appeals.

As Solan (1993, 170) asserts in his linguistic analysis of judicial decisions, “the
words in the Constitution...obviously underdeterrnine their applicability to particular
situations.” When a constitutional provision is the basis of adjudication, there is greater
potential for numerous, and conﬂicting, interpretations. Thus we would expect to see more
variation in lower court treatment of constitutionally based precedents.

Hypothesis 7. There is a negative relationship between a
Supreme Court civil rights and liberties precedent which was
established by the Burger Court and the following of that
precedent in the Courts of Appeals.

One study of policy making in the federal courts has hypothesized that the

inconsistency and ambiguity of Burger Court opinions contributes to the discretion that

104

lower court judges are able to exercise (Carp and Rowland 1983, 11). Another states that
“the Burger Court’s decisions in...civil liberties are unlikely to generate uniform compliance”
due to their lack of precision (Walker 1978, 384). Accordingly, Hypothesis 7 speciﬁes that
civil rights decisions emanating from the Burger Court should be followed less frequently
in the Courts of Appeals.

Competing hypotheses are offered with regard to lower court treatment of Supreme
Court decisions which have important political or social ramiﬁcations. One view is that
lower court judges are more apt to follow such decisions, as a deviant decision is more likely
to come to the attention of the High Court and to be selected for review (Johnson 1979;
George 1997). An alternative belief is that highly signiﬁcant decisions will encounter more
resistance in the lower courts, precisely because of their widespread implications (Baum
1978). It is hypothesized here that, because of the legal socialization that judges undergo,
the possibility of review will be the prevailing inﬂuence. Hypothesis 8 suggests that this
consideration will lead to more faithful application of salient decisions.

Hypothesis 8. There is a positive relationship between a
salient Supreme Court civil rights and liberties precedent and
the following of that precedent in the Courts of Appeals.

There is general agreement that circuit courts will decide cases en banc when
“difﬁcult, complex, highly political, or simply signiﬁcant questions” are involved (George
1997, 9). However, many scholars disagree about the deference the Supreme Court pays to
en banc decisions. Some assert that because of the importance of such decisions, they will

be monitored closely by the Court (George 1997). Alternatively, one study reports that the

“higher court rarely conﬁrms or reverses decisions made by the entire circuit membership”

105

(Richardson and Vines 1970, 123). The view adopted in Hypothesis 9 is that the Supreme
Court should give more credence to en banc judgments, and thus that circuit court judges

should have more discretion in their decision making.

Hypothesis 9. There is a negative relationship between en
banc decisions in the Courts of Appeals and the following of
Supreme Court civil rights and liberties precedents in those
decisions.
As students of all levels of courts have demonstrated, personal attitudes and values
and values are crucial determinants of judicial behavior (see, e. g., Epstein and George 1992;
Songer and Haire 1992; Brace and Hall 1993; Segal and Spaeth 1993; Segal, Songer, and
Cameron 1995). Hypothesis 10 relates to the importance of ideology in Courts of Appeals’

treatment of Supreme Court precedent.

Hypothesis 10. There is a positive relationship between an
appellate panel’s ideological consistency with the direction of
a Supreme Court civil rights and liberties precedent and the
following of that precedent in the Courts of Appeals.

Because of the established importance of individual policy goals, review-minded
judges should be aware of the preferences of the High Court’s members. When subordinate
judges perceive that the ideological conﬁguration of the Court has changed since a precedent
was handed down, they should respond accordingly (Wasby 1970). Hypothesis 1] describes
the nature of this response.

Hypothesis 1 I. There is a positive relationship between a
Supreme Court precedent which has gained greater
ideological support on the Court, due to changes in Court

composition, and the following of that precedent in the Courts
of Appeals.

106

The operationalization of the legal and attitudinal elements contained in these
hypotheses is described in detail in Chapter 2 (see pp. 37-41) and is summarized in Table
5.1. Table 5.1 also speciﬁes the direction of the relationships which are hypothesized to exist

between these legal and attitudinal elements and the treatment of precedent. In the next

section, these factors are included in a logistic regression model of the Courts of Appeals’
treatment of Supreme Court civil rights and liberties decisions, in order to ascertain their role

in determining this treatment.

FINDINGS
As with the decisions of the 1987 Term, the Supreme Court’s civil rights and liberties
decisions from 1966-75 received deferential treatment in the Courts of Appeals. The
intermediate federal courts followed the precedents established by these decisions in 74.2%
of their own decisions. In approximately one fourth of their decisions, however, the circuit
courts did not follow a relevant Supreme Court precedent. Thus it is critical to identify the
circumstances under which these courts fail to apply precedent. To accomplish this, the
following model of appeals court treatment of Supreme Court civil rights decisions is
estimated:
P(Treatment=l) = b0 + b1(Unanimous) + b2(Minimum Winning Coalition) +
b3(Doctrinal Modiﬁcation) + b4(Age of Precedent) + b5(Concurring Opinions) +
b6(Constitutional Provision) + b7(Burger Court) + b8(SaIience) + b9(En Banc) +
b,o(1deological Consistency) + bn(Change in Supreme Court Composition)
Each bi indicates the change in the log of the odds of following precedent due to changes in

the value of variable i, controlling for the values of the other variables. It is hypothesized

that b1, b,, b,, bm, and bll > 0 and that b2, b3, b5, b6, b7, and b, < 0.

107

Table 5.2 presents the estimated coefﬁcients and signiﬁcance levels for this logistic
regression model which predicts whether the circuit courts will follow a particular Supreme
Court decision. A likelihood ratio test, which tests the null hypothesis that all coefficients
in the model except the constant are zero, indicates that the block of coefﬁcients is
signiﬁcantly different from zero at p<0.001.2 Each coefﬁcient achieves statistical
signiﬁcance at p<0.05 and is in the expected direction. The model correctly classiﬁes 74.7%
of appeals court decisions.3

According to the results in Table 5.2, the Courts of Appeals become less likely to
follow a Supreme Court decision as more concurring opinions are written by the justices.
The interpretation of a constitutional provision also decreases the likelihood that a precedent
will be followed. These courts are attentive to Supreme Court modiﬁcations of prior
precedents, as they are less likely to adhere to a precedent that has been substantially altered

by a subsequent decision.

 

2The likelihood ratio test statistic is calculated as -210g(L0/L l), where L 1 is the
likelihood function for the full model as ﬁtted and L0 is the maximum value of the likelihood
function if all coefﬁcients except the constant are zero. The test statistic is xz-distributed,
with degrees of freedom equal to the number of independent variables.

3The model achieves a 1.94% reduction in error from predicting treatment based upon
the proportion of decisions followed. Percent reduction in error is calculated according to
the following formula:

100 x W
100 - % in modal category

As noted in Chapter 4, the goal in developing these models is to determine when
circuit courts fail to follow Supreme Court precedent. This model is very useful in
identifying such situations.

108

Appellate panels are more likely to follow a decision which was agreed to by all of
the justices, and they become more likely to follow precedent as the length of time increases
since the precedent was established. Additionally, they are more likely to follow a precedent
which comports with their personal preferences. Courts of Appeals judges are also aware
of changes in the ideological composition of the High Court, as they are more likely to follow
a decision which has the approval of the current Court. Contrary to Hypothesis 7, a Burger
Court decision is more likely to be applied favorably in the circuit courts.

Likelihood ratio tests were calculated in order to determine whether some of the
variables had zero coefﬁcients and should therefore be dropped from the model.‘ I
hypothesized that appeals court judges would be less likely to follow precedent in their en
banc decisions and that they would be less likely to follow decisions reached by a minimum
winning Supreme Court coalition. I also theorized that cases with important policy
implications would be followed more often in the lower courts. However, tests for redundant
variables indicated that Minimum Winning Coalition, En Banc, and Salience had little

explanatory power in a model of circuit court treatment of precedent.

 

“The likelihood ratio test which was calculated above as a measure of goodness-of-ﬁt
can also be used to determine whether individual coefﬁcients are zero. In such a test, Ll is
the value of the likelihood ﬁrnction for the complete model, and L0 is the value of the
likelihood function for the reduced, or nested, model. The number of degrees of freedom is
equal to the number of coefﬁcients hypothesized to be zero.

When En Banc was omitted from the model, the likelihood ratio was 0.575 with
p=0.448. When Minimum Winning Coalition was omitted, the likelihood ratio was 0.528
with p=0.467. When Salience was omitted, the likelihood ratio was 2.096 with p=0.148.

109

Predicted probabilities of following precedent under certain conditions are more
readily interpretable than are logit coefﬁcients.5 Some of these predictions are displayed in
Table 5.3. The baseline probability indicates the likelihood of following precedent when
each independent variable takes on its mean or modal value. Under the most common
circumstances, the probability of a circuit court following a Supreme Court civil rights
precedent is 0.72. This ﬁgure increases notably, taking a value of 0.79, when the Supreme
Court’s membership becomes more attitudinally supportive of a past decision. A unanimous
decision is followed at a similar rate of 0.78. A Warren Court decision is substantially less
likely to receive favorable treatment in the circuit courts, as such decisions are followed only
59% of the time. A precedent which has been signiﬁcantly revised by subsequent Supreme
Court decisions is followed in only 47% of appeals court decisions. A ﬁve-year-old
precedent is followed at a rate of 0.71 , while a twenty-year-old precedent is followed at a rate
of 0.7 8.

Those who advocate a strictly attitudinal description of judicial decision making
could reasonably assert that appeals court judges do not fail to follow a particular Supreme
Court precedent because the precedent lacks authority or because they are unsure of the
Court’s intentions for its application; rather, these judges avoid precedent because they see
an opportunity to exercise their own preferences when the chances of Supreme Court review
are slim. In order to investigate this possibility, I estimated the model again with interactive
variables. The purpose of including these interaction terms is to ascertain whether attitudinal

or legal factors determine treatment when both are present.

 

5The equation for calculating these predicted probabilities is 1/(1 + exp(-X,B)).

110

As a measure of Ideological Consistency, I used a variable which indicated whether
the majority of the panel or circuit agreed with the direction of the Supreme Court decision.
I interacted this variable with two of the measures of decisional clarity, Concurring Opinions
and Constitutional Provision. The results of this analysis, as reported in Table 5.4, are
inconclusive. The model does not allow one to predict treatment conﬁdently when a panel
or circuit approves of the outcome of a Supreme Court decision, but when the decision’s
interpretation is ambivalent. Perhaps other considerations come into play in such instances,
which are not accounted for in the model. In any event, the argument that a lower court
judge is relying on doctrinal justiﬁcations to mask his ideological inclinations is not
supported by these ﬁndings.

Some diagnostics tests were employed to determine the robustness of the ﬁndings
reported in Model 1. These diagnostics are the same as those performed in Chapter 4 (see
pp. 84-85) and described in detail in Appendix B. The results of these tests indicated that
only ninety-seven of the 5,656 cases are poorly ﬁt by the model, according to the deviance
residuals which were calculated. These cases truly ﬁt the description of outliers; the ninety-
seven cases which are problematic are comprised of forty-three different patterns of
independent variables, so that each pattern on average consists of only between two and three
cases. This indicates that the outlying patterns are unusual assortrnents of legal and
attitudinal factors.

The diagnostic tests further demonstrated that no combinations of independent
variables are exerting undue inﬂuence on the results, as deleting particular patterns did not
produce signiﬁcant changes in the estimated values. In fact, the leverage statistic hj for each

pattern never exceeds 0.03. According to these diagnostics, a great deal of conﬁdence may

111

be placed in these ﬁndings, as the model ﬁts 98% of the data extremely well and no case
pattern is having a disproportionate impact on the results.

Some trends were discernible in the 2% of cases which were poorly ﬁt by the model.
Of these ninety-seven cases, forty-two applied Supreme Court precedents which were
established by a unanimous vote margin. It is noteworthy that nearly half of the forty-two
unanimous decisions (18) dealt with school desegregation. Desegregation was a politically
divisive issue, in which the Supreme Court strove to present a united front in order to
discourage public resistance (see, e. g., Rohde 1972; Hutchinson 1979). However, in eight
of the eighteen desegregation cases where a unanimous precedent was cited, the Court’s
solidarity did not promote compliance in the lower courts, as the precedent was not followed.

In fact, contrary to what the model predicts, a unanimous precedent was not followed
in thirty-two of the forty-two circuit court decisions where it was found to be relevant. Ten
of these treatments of unanimous decisions dealt with issues of due process, and in none of
these instances was precedent adopted. It seems that, generally speaking, support on the
Court for a precedent fosters obedience in the Courts of Appeals; but in certain issue areas,
consensus is not sufﬁcient to ensure favorable treatment of precedent.

This examination of outlying cases provides evidence that some issues may be more
complex than others. Forty-six cases of the ninety-seven cases which were not well-
explained by the model dealt with various due process issues, and twenty involved school
desegregation.6 Perhaps certain aspects of these issues raise more difﬁcult questions, and

further exploration is needed to account for their treatment.

 

6Ten such cases dealt with one decision, Swann v. Charlotte-Mecklenburg County
Board of Education, 401 US. 1 (1971).

112

DISCUSSION

Consistent with the ﬁndings of Segal and Spaeth (1993) and those reported in
Chapter 4 of this study, judicial attitudes are an important determinant of decision making.
The estimates reported in Table 5.2 illustrate that when appeals court judges agreed
ideologically with the policy established by a Supreme Court decision, they were
signiﬁcantly more likely to apply that precedent favorably in their own decisions. These
judges were also attentive to the ideological composition of the Supreme Court, for if a later
Court was not as supportive of a decision as was the Court that rendered the decision, circuit
court judges were less inclined to afford a positive treatment to that decision. Hypotheses
10 and 1 I , the two hypotheses relating to attitudinal factors, are therefore conﬁrmed by these
ﬁndings.

However, personal attitudes and values were not the only forces which inﬂuenced
behavior in the Courts of Appeals, nor were they the strongest motivators. The actions of
circuit courts judges were affected substantially by their legal training, which socialized them
to defer to the decisions of higher courts and to subordinate their own preferences. As the
simulations displayed in Table 5.3 demonstrate, many of the legal elements which were
operationalized in the model had greater consequence for the treatment of precedent than did
the attitudinal elements.

Partial support is obtained for the importance of the authoritativeness or strength of
precedent in affecting circuit court behavior. Hypothesis 1 is corroborated, as unanimous
civil rights and liberties decisions had greater precedential value in the circuit courts than did
nonconsensual decisions. I hypothesized that unanimous decisions would be followed more

frequently in the Courts of Appeals because they carried the full authority of the Court. I

113

reasoned that the Court’s agreement could act to mitigate any resistance the decisions might
encounter in the lower courts. Pacelle and Baum (1992) offered an alternative explanation
of the favorable treatment that might be afforded unanimous decisions. They argued that
nonunanimous decisions are more likely to be controversial to the Court’s audience; thus
lower courts may respond more positively to unanimous decisions simply because they are
less likely to disagree with them. It may be that these two considerations were working in
concert to motivate circuit court deference to these decisions.

On the other hand, contrary to Hypothesis 2, appeals court judges were not inﬂuenced
by closely divided Supreme Court rulings. I theorized that decisions which were reached by
a minimum winning coalition on the Court were more likely to encounter opposition in the
lower courts, as the justices were in sharp disagreement over the outcome. However,
statistical tests (see ﬁt. 4) showed that narrow decisional margins did not signiﬁcantly
diminish the propensity of circuit court judges to follow Supreme Court precedent.

Of particular import to circuit court judges was whether a Supreme Court precedent
had been altered by subsequent decisions. As Hypothesis 3 stipulates, lower court judges
were acutely aware of modiﬁcations in precedent, and they reacted accordingly to these
changes in doctrine. This ﬁnding is entirely consistent with the principle of stare decisis.

In addition to the level of Supreme Court agreement on a precedent, the age of a
precedent may also serve as an indication of its authority. These ﬁndings indicate that, as
Johnson (1987) speculated, lower courts may ﬁnd older precedents more persuasive, or
authoritative, than newer ones. Consistent with Hypothesis 4, as time passed since a

precedent was established, the Courts of Appeals became more likely to follow that

114

precedent. This result may also be attributable to the fact that precedent is deﬁned more
clearly as its appropriate application is articulated in other decisions.

The clarity of Supreme Court policy was critical to its implementation in the Courts
of Appeals. As more, and divergent, judicial viewpoints were expressed, appellate judges
became uncertain of the direction the High Court wanted them to take. When the High Court
was divided in its reasoning, the lower courts were divided in their applications of precedent.
Hypothesis 5 is thus supported. The interpretation of provisions of the Constitution may
have also fostered confusion in the circuit courts, as speciﬁed by Hypothesis 6. This is
evidenced by the fact that precedents which did not involve constitutional issues were more
likely to be afforded favorable treatment.

Hypothesis 7 asserted that Burger Court decisions should be followed less often in
the Courts of Appeals, as such decisions have been characterized as ambiguous and
inconsistent (see Walker 1978; Carp and Rowland 1983). The ﬁndings in Table 5.2
contradict this hypothesis, as it was the decisions of the Warren Court which engendered
negative responses from the circuit courts. There may be an attitudinal, rather than a legal,
explanation for this phenomenon. The Burger regime has been termed the “era of
Contemporary Conservatism” (Schubert 1974, 189-213). It is reasonable to suggest that the
decisions of the Burger Court met with greater agreement in the Courts of Appeals because
appointees of conservative presidents dominated these courts. In fact, 59% of active appeals
court judges during the period under investigation were appointed by Republican presidents
(see Table 5.5).

Hypothesis 8 relating to the salience of precedent is negated by the results of tests for

redundant variables (see ﬁr. 4). I expected that the Courts of Appeals would be particularly

115

attentive to Supreme Court decisions which had signiﬁcant political or social ramiﬁcations,
as contrary lower court decisions might be more apt to foster review. However, there was
no evidence that the circuit courts treat these decisions differentially. Appeals court judges
may have viewed all of the civil rights and liberties decisions traced here as having important
policy consequences and, as a following rate of nearly 75% suggests, afforded these
decisions a high degree of deference.

Hypothesis 9, which asserted that circuit courts will exercise more discretion in their
en banc decisions, was also disproved on the basis of tests for redundant variables (see fn.
4). Clearly, en banc cases are distinguishable from those heard by three-judge panels, since
the circuit is willing to expend the requisite resources for adjudication by all of its members.
As Howard (1981, 9) explains, “circuit judges tend to avoid en bancs as cumbersome.” In
fact, the Administrative Ofﬁce Of United States Courts reported that in a recent year, less
than one percent of the appeals terminated on the merits were handled en banc. It is possible
that there were not enough en banc decisions in the data set to evaluate differential behavior,

as just over 200 of the nearly 6,000 decisions were reached en banc.

INTER- AND INTRA-CIRCUIT VARIATION
Because of the infrequency of Supreme Court review, the Courts of Appeals have a
great deal of freedom in their decision making. A number of scholars have recognized that
the product of this discretion may be the creation of policy on a circuit-by-circuit, rather than
a national, basis (Howard 1981; Carp and Rowland 1983; Davis and Songer 1988-9; Wenner
and Dutter 1989). Goldman (1966, 382) found the institutional diversity of the circuit courts

to be so pronounced that it “imposed limitations on data collection and analysis.” Combs’

116

(1982) study provides empirical support for this allegation in the area of school
desegregation, where the various circuits contemporaneously employed four different tests
in determining whether proof of racial discrimination had been established.

Several explanations of inter-circuit differences in decision making are offered in the
literature. One such explanation attributes decisional dissirnilarities to regional forces which
may act upon the circuits courts. For example, Howard (1981, 33) asserts that the
organization and structure of these courts “may well spawn regionalization and regionalized
national law.” Carp and Rowland (1983, 88) also suggest that “some of the circuit-to-circuit
variance may be caused by extralegal political and environmental inﬂuences.” Regional
values and interests may be relevant for some circuits such as the fourth, ﬁfth, and eleventh,
which are composed solely of southern states, and the seventh, which is made up of only
midwestem states; but while some circuits clearly have a regional character, other circuits
ignore such boundaries. For example, the ninth circuit consists of states from Alaska to
Arizona and from Montana to Hawaii, and even includes the territory of Guam. In spite of
the fact that these states share the western region of the United States, many of the policies
pursued by Northwestemers are in direct conﬂict with those espoused by Californians,
particularly in the area of environmental regulation. This ideological discord has prompted
many legislators to lobby for the division of the ninth circuit.

Other scholars ascribe inter-circuit diversity to the disparate nature of the cases the
various circuits hear. Howard (1981, 55) reported that each circuit he studied “became a
magnet for certain subjects of federal law by virtue of location, special jurisdiction, and
predilections.” Even in those cases where the circuits dealt with similar issues, he detected

differences in the types of parties which each circuit favored. Howard’s (1981) ﬁndings for

117

the 1960s were conﬁrmed by Davis and Songer’s (1988-9) analysis of circuit court decision
making in the 19805. Subject matter specialization and preferences for certain classes of
litigants enhance the potential for the creation of circuit-speciﬁc policy.

A ﬁnal explanation of circuit-based variation is an attitudinal one, asserting that
decisional inconsistencies among the circuits may be due to differences in circuit
composition. A number of studies have documented distinct judicial behavior according to
the party afﬁliation of the judge or of the president who appointed the judge (see, e. g.,
Goldman 1966, 1975; Tate 1981; Carp and Rowland 1983; Tate and Handberg 1991). Table
5.5 displays the percentages of active judges in each circuit who were Democratic and
Republican appointees.7 During this time, the Courts of Appeals as a whole were dominated
by appointees of Republican presidents, although the makeup of each circuit varied. Nearly
three-fourths of seventh circuit judges were appointed by Republican presidents. While the
third, fourth, tenth, and DC circuits were reasonably well-balanced, only in the ﬁfth circuit
were there more Democratic appointees than Republican ones.

In order to determine whether civil rights and liberties precedents are treated
differentially among the circuits, I ﬁrst employ standard difference-of-proportions tests. I
compare the rate of following precedent in each circuit to that in the Courts of Appeals as a
whole. The null hypothesis is HO: RC = itc, where rte is the proportion of decisions followed
in all of the circuits and 1tc is the proportion of decisions followed in individual circuits. The

alternative hypothesis is HA: 1% e 1;, . Rejection of the null hypothesis indicates that

 

7In relating the makeup of each circuit to its decisional tendencies, it is important to
point out that, as noted in Chapter 1, it is not only the active circuit judges who may sit on
appellate panels. Retired circuit judges, judges from other circuits, district court judges, and
even retired Supreme Court justices may be assigned to these panels.

118

precedent is followed more often, or less often, in a particular circuit. A two-tailed test is
used, where H0 is rejected at a = 0.05 if the t-statistic is 1.96 or greater.8

Table 5.6 provides the rates at which precedent is followed in each circuit, as well
as a breakdown of the treatment of liberal and conservative precedents. The difference-of-
proportions tests demonstrate that there is substantial variation in the treatment of precedent
among the circuit courts. While the eighth and eleventh circuits were more likely than the
other circuits to follow precedent, the ﬁrst circuit was less likely to follow precedent. The
ninth circuit may also have been less likely to follow precedent.

An analysis of treatment based upon the ideological direction of precedent suggests
that some of this behavior may be attitudinally motivated. The ninth circuit was less likely
than other circuits to follow liberal precedents, while the ﬁrst and second circuits were less
likely than other circuits to follow conservative precedents. The eighth and eleventh circuits
were more likely than other circuits to follow conservative precedents. Additionally, the
eighth and ninth circuits were more likely to follow conservative precedents than they were
to follow liberal ones.

I also reestimate the civil rights model twelve times, including a dummy variable for
a different circuit in each run. The results of these estimations will indicate whether the

decision making of individual circuits is signiﬁcantly different from that of the Courts of

 

8A difference-of-proportions test is calculated as

t = (PA ' PB)
{PC (1 ’ PC) (nA + “B / nA x “3)

where PC = (n APA + nBPB) / n A + n B. H 0 is rejected at at = 0.10 if the t-statistic is 1.64 or
greater and at a = 0.01 if the t-statistic is 2.57 or greater.

119

Appeals as a whole when other factors are controlled for. The coefﬁcients and standard
errors are presented in Table 5.7. The results of the difference—of-proportions tests are
conﬁrmed to some extent, as the eighth circuit was more likely than the other circuits to
follow precedent and the ﬁrst circuit was less likely to follow precedent.

To summarize the ﬁndings in Tables 5.6 and 5.7, in the third, fourth, ﬁﬁh, sixth,
seventh, tenth, and DC circuits, no patterns of variation in the treatment of precedent were
discernible. Decision making in these circuits closely resembled that of the Courts of
Appeals as a whole. This ﬁnding is of particular interest with respect to the fourth, ﬁfth, and
seventh circuits. As Table 5.5 shows, the seventh circuit was dominated overwhelmingly by
conservative appointees; yet this circuit was not more likely than others to follow
conservative precedents, nor did it treat conservative precedents differently from liberal ones.
Regarding the fourth and ﬁfth circuits, these circuits have historically been said to rule
conservatively in their race relations and Fourteenth Amendment cases (see, e. g., Carp and
Rowland 1983), but no such trends were apparent here.

However, in the ﬁrst, second, eighth, ninth, and eleventh circuits, deﬁnite
precedential biases were found. The ﬁrst and second circuits were less likely than the other
circuits to apply conservative precedents. The ﬁrst and second circuits consist of the New
England states and New York, and the literature reports that East Coast Republicans are more
liberal than their fellow Republicans in the rest of the nation (see, e. g., Carp and Rowland
1983). It is not surprising, then, that even though two-thirds of the judges of both the ﬁrst
and second circuits were Republican appointees, these circuits made fewer conservative

rulings than did their GOP colleagues in other parts of the country.

120

The eighth and ninth circuits exhibited similar behavior in some respects. Both
circuits favorably applied conservative precedents more often than they did liberal ones; and
while the eighth circuit espoused conservative precedents more often than the other circuits,
the ninth circuit complied less frequently with liberal precedents than its counterparts.
However, although the eighth circuit was more deferential to precedent overall than were the
other circuits, the ninth circuit may have been less obedient.

The eighth circuit is composed of six midwestem states, plus Missouri and Arkansas.
The conservatively oriented behavior of this circuit is explainable in light of the ﬁndings of
one study of the federal courts (Carp and Rowland 1983), where the authors discovered only
minimal partisan differences among the judges of the eighth circuit. In fact, the liberalism
scores of Republican judges were higher than those of Democratic judges, so that even the
“liberal” judges were likely to support conservative policies.

As with the eighth circuit, Carp and Rowland (1983) reported only slight differences
in liberalism indexes among ninth circuit judges. Their data also indicated that western
judges reached more conservative decisions than did northern and eastern judges, particularly
in civil rights and liberties cases. Thus the ninth circuit’s conservative approach to the
treatment of Supreme Court precedent is understandable.

The eleventh circuit was notably obedient to precedent. It was signiﬁcantly more
likely to adopt all precedents and conservative precedents than were the other circuits. It was
also more likely than the other circuits to adhere to liberal precedents, although the t-value
is not statistically signiﬁcant. These ﬁndings are especially intriguing when one considers

a possible institutional motivation for this behavior. The eleventh circuit was not created

121

until 1981, when it was carved out of the ﬁfth circuit.9 Research on the US. Supreme Court
indicates that in the years following its creation, consensus was the norm, with the justices
casting dissenting votes only infrequently (see Epstein et. al. 1996). In fact, substantial
dissent did not appear until the 19405. The eleventh circuit may have behaved similarly
following its inception, showing especial deference to Supreme Court precedent, regardless
of its ideological tenor, in order to establish its legitimacy as a decision making institution.

Two of the federal circuits tended to avoid precedent in general. The ﬁrst circuit was
signiﬁcantly less likely than the other circuits to apply Supreme Court decisions in its
rulings, and there is some evidence that this is also true of the ninth circuit. These circuits
seemed to perceive that they had a great deal of independence in their decision making, and
they took advantage of this opportunity to shape their own public policies.

The analysis presented here does not permit an evaluation of Howard’s (1981)
explanation for inter-circuit variation in decision making. He reported that differences in
circuit outcomes were due, at least in part, to differences in the types of cases each circuit
adjudicated; but, as this study is limited to circuit decision making in civil rights and liberties
cases, this claim cannot be assessed. There is limited evidence of regional patterns in
decision making, as the northeastern circuits exhibited more liberal tendencies than did the
rest of the nation, while the western circuit was more conservative. However, no such trends
were observable in the southern circuits; and while the eighth circuit ruled more
conservatively than other circuits, this behavior was not apparent in the other midwestem

circuits, i.e., the sixth and seventh circuits. Finally, some support is found in the eighth and

 

9See Barrow and Walker (1988) for a thorough account of the division of the ﬁfth
circuit.

122

ninth circuits for an attitudinal explanation of circuit diversity. It is important to note,
however, that in seven of the twelve circuits, decision making is consistent with that of the
Courts of Appeals as a whole.

In his extensive study of the Courts of Appeals, Howard (1981, 8) contended that the
decentralization of the federal courts could potentially lead to the “Balkanization of national
law.” The evidence presented here indicates that his concerns are justiﬁed with respect to
decision making in at least some circuits, as there is meaningful variation in the extent to
which individual circuits adhere to Supreme Court doctrine in their decisions. In some
circuits, the application of precedent appears to be attitudinally determined. But in the
majority of circuits, judges appear to be guided in large part by past Supreme Court rulings.

In addition to variation among circuits in the treatment of precedent, many scholars
have acknowledged the potential for variation within circuits as well (Richardson and Vines
1970; Atkins 1972; Goldman 1975; Howard 1981; Songer 1982). This variation is said to
be a product of decision making procedures, as the majority of Courts of Appeals decisions
are made by three-judge panels rather than all of the circuit judges. Richardson and Vines
(1970, 123) assert that “the panel system...enhances the opportunity for intracourt conﬂicts
on policy positions,” and according to Howard (1981, 9), panel rotation is “a potential source
of disharmony in federal law.”

Few circuits have been able to avoid accusations of “panel packing” (Howard 1981).
One study (Atkins and Zavoina 1974) has illustrated the capacity of a chief judge to Obtain
outcomes favorable to his policy position through panel assignments. This analysis reported

that, from 1961 to 1963, a minority of ﬁfth circuit judges decided a disproportionate number

123

of race relations cases, usually in favor of black plaintiffs. The authors attributed this ﬁnding
to the fact that Chief Judge Elbert Tuttle, a member of the liberal minority voting bloc,
assigned judges to these cases who shared his pro-civil rights ideology. One of his fellow
circuit judges had publicly accused Tuttle of such a strategy in a dissenting opinion.10

If panel decisions are truly a function of which judges comprise the panel, we would
expect to see a great deal of diversity in the treatment of precedent according to the
ideological makeup of the panel. Table 5.8 presents the rates of following precedent for
appellate panels according to their ideological composition. Panels consisting of three liberal
judges were more likely than other panels to follow liberal precedents. Panels made up of
two conservative judges and one liberal judge were more likely to follow conservative
precedents than they were to follow liberal ones, and this may also be true of panels
composed of three conservative judges.

Additional evidence of ideologically motivated decision making by Courts of Appeals
panels is provided in Table 5.9, which breaks down panel behavior by circuit. Liberal-
dominated panels in the sixth circuit were more likely to follow liberal precedents than
conservative ones, while conservative-dominated panels in the eighth and ninth, and perhaps
seventh, circuits were more likely to follow conservative precedents than liberal ones.
Interestingly, in the third circuit, liberal-dominated panels were more likely to follow
conservative precedents than they were to follow liberal precedents.

This analysis of panel decision making indicates that the ideological composition of

appellate panels can have a signiﬁcant inﬂuence on the decisions that are reached. In the

 

10See Judge Ben Cameron’s dissent in Armstrong v. Board of Education in the City
of Birmingham, 323 F.2d 337 (5th Cir. 1963).

IL Z“:

124

sixth, seventh, eighth, and ninth circuits, Supreme Court precedent was followed more often
when it comported with the personal preferences and policy goals of panel members. Thus
the anecdotal ﬁndings of Atkins and Zavoina’s (1974) study are conﬁrmed here. However,
there is also evidence that panel members are not driven solely by their own attitudes, but
that they take the authority Of Supreme Court precedent into account in their decisions. It
is noteworthy that panel decisions on the third circuit appear to be motivated by stare decisis
rather than the policy preferences of the judges, while judges in the remaining seven circuits

seem to be inﬂuenced by a combination of legal and extralegal factors.

 

-‘n .
.A

 

125

.2000 00w0=m 20 >0 000300 203 35
t 00000 2000 08296 w5000m05 2.33? bus—:0 2:00 ..0mSm

022922 003 003320 3000300000 0 2003
t 000.00 2000 05295 203205 0:32? >EE=Q 003320 300030200

.0000 2000 08296 a

t E 00:02 203 35 0000000 20200000 .00 000802 22500 20200000 00000005 00 >220
000300
+ $3 0000 2000 08296 0 00:3 200% .00 000052 0000000.:— 00 0w<

02.5000 00: 00000020
.00 02320008 0:02.:an 0 .000 03003.,
.«0 2.500 05 E 00:00“ 0.3 0033 20000020
t 2000 2:295 mat—W205 0733.3 >EE=D 02305002 35.0000

.m-m .00
20> 0 >0 0020>2 00>» 003000 2:00 .030. a. 20:3
0020 0:0 mtv ..0 VA .00 20> 0 >0 000300 203 35
t 0.00.00 2000 08295 203205 07333 >EE=Q 002300 20:53 82552

d0 00 dtn d-» dd .00 20> 0 >0 002000 203 35
+ 3000 2000 08205 203202 032.03 bus—:0 30.5025 00000005 .00 52.25

 

E00002,” w=_>>0=0m 0008200002 023530202000 “000000
2 020023—0M

 

8:300: 02< 2.203 020 ”$000,305 meBEm 250203502 320200005 02< 0200200 00 zoﬁ<~3<zo§§20
0m mammary

126

 

022000 2000
0.0295 05 .00 0200020 05 w2000000002 .02000
+ 025 0002500 02000 32002 0002 2 00020.29

022000
2000 00.295 05 .00 0200020 05 2% 3.0022002
+ 00020 0000 _0000 20:0000 000 .00 0w20020m

2000000020 20020 02000 20 >0 002000 203 0000

0020000000
2000 2020.5 2 0w0000

080220000 _00_w0_000_

000020.020 .2203.

 

I 00000 200090. .00 02000 w200202 0.0200> E0230 050m :0 00020009 20202000—
.0022000
00020002 .00 w200= 0.00 2 000202 20 0000
+ 00000 2000 02295 w200202 0_00_._0> E00030 0000005 000005025 >000;
00000020 w2>>0=0m 00000200022 02000000202000 0000000
00 200020230

 

8.0280 0.0 0003.

127
TABLE 5.2

LOGISTIC REGRESSION OF TREATMENT: CIVIL RIGHTS AND LIBERTIES (MODEL 1)

 

 

 

Explanatory
Variable Coefficient Standard Error t Statistic

_Constant 0.321 0.136

Attitudinal E actors

Ideological Consistency 0.326 0.106 3.082

Change in Supreme Court

Composition 0.321 0.106 3 .027

W

Unanimous 0.271 0.094 2.898

Doctrinal Modiﬁcation -1.071 0.253 -4.228

Age of Precedent 0.024 0.005 4.761

Concurring Opinions -0.154 0.044 -3.498

Constitutional Provision -0.257 0.072 -3.572

Burger Court 0.613 0.102 6.028
N = 5656
32 (8) = 193.72

Log-likelihood = -3130.084

p > x2 = 0.000

128

TABLE 5.3
PREDICTED PROBABILITIES OF FOLLOWING PRECEDENT: CIVIL RIGHTS AND LIBERTIES

 

Baseline 0.725
Ideological Consistency
0% 0.691
100% 0.756

Change in Supreme Court Composition

-1 0.659
1 0.786
Unanimous 0.776
Doctrinal Modiﬁcation 0.475
Age of Precedent
5 Years 0.707
20 Years 0.776
Concurring Opinions
0 0.739
4 0.604
No Constitutional Provision 0.773

Warren Court 0.588

129

TABLE 5.4
LOGISTIC REGRESSION or TREATMENT: CIVIL RIGHTS AND LIBERTIES (MODEL 2)

 

 

 

Explanatory
Variable Coefﬁcient Standard Error t Statistic

_Constant 0.365 0.146
WE:
Ideological Consistency 0.288 0.144 1 .994
Change in Supreme Court

Composition 0.326 0.106 3.062
L l t r
Unanimous 0.272 0.094 2.894
Doctrinal Modiﬁcation -1.073 0.253 -4.235
Age of Precedent 0.024 0.005 4.668
Concurring Opinions -0.175 0.059 -2.975
Constitutional Provision -O.262 0.090 -2.91 7
Burger Court 0.589 0.102 5.760
Attitudinal and Legal
11000101205
Ideological Consistency X

Concurring Opinions 0.054 0.079 0.680
Ideological Consistency ><

Constitutional Provision -0.003 0.112 -0.029

N = 5632

12 (10) = 189.77

Log-likelihood = -3116.439 p > x2 = 0.000

130

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

the Auburn judicial background data set.

TABLE 5.5
CIRCUIT COMPOSITION BY PARTY OF APPOINTING PRESIDENT
DEMOCRATIC REPUBLICAN
APPOINTEES APPOINTEES
33.3% 66.7%
lsT (4) (8)
31.3% 68.8%
2ND (10) (22)
45.5% 54.5%
3RD (15) (18)
42.3% 57.7%
4TH (11) (15)
55.6% 44.4%
5TH (30) (24)
40.6% 59.4%
6TH (13) (19)
26.9% 73.1%
7m (7) (19)
32.0% 68.0%
8TH (8) (17)
42.0% 58.0%
9TH (21) (29)
45.5% 54.5%
1001 (10) (12)
14.3% 85.7%
1 1TH (1) (6)
44.8% 55.2%
DC (13) (16)
41.1% 58.9%
TOTAL (143) (205)
NOTE: Figures include only active circuit judges. Data were Obtained from

 

131

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1.000.0- 00.0- 000.0 000.0- 025500200
.5> 3mm...—
00000 0.00.0.0- 0.00 000.0- 0000 000.0 0000 000.0 00000 25500200
000.0 000.0 E0 0000 E0
5000 00.0.0- 300 000.0- @000 000.0- 8000 000.0 :03 0.0000:
000.0 000.0 000.0 000.0 0:0
0.0000 ...-20.0- 00000 0.00- 00000 000.0- 0000 000.0 0000 .5.
000.0 000.0 000.0 30.0 000.0
44000 En EN. Eb Em
:00- E;- 000 000.0 25500200
5> Emmi
00000 000.0- 30: 000.0- 85 2.03.0 0000 2000.0 0.00 055500200
000.0 000.0 000.0 000.0 000.0
3.000 000.0- 00000 000.0 A000 0.20- 803 0:0 A030 .2000:
000.0 000.0 000.0 000.0 30.0
0.0000 0000- 0.000 000.0 000 0:; A003 ......0000 0.000 d<
000.0 000.0 000.0 0:0 000.0
0.20 E0 95 QZN #50 .FZQOUmM—m
mhé mDA<>I-_- ”...—km m5.._<>--0 Bé mDA<>Ih mhé $5.050- m-Hé

 

 

 

 

 

 

 

 

 

 

05 mama?

500-50094 92¢. 5.5.005“ 0500 ”0.50050 >m Ewan-“Oman oak/00.00% LO mmhé

 

132

mcdvm:
26V}.

200800 005 00.0 2 .88 05 88:05 0002000088 E 0038: Z ”mi-OZ

 

 

 

 

 

 

 

 

 

 

 

00000. 0- 0 0 0.00 1000.0- 002-0005000200
.m> 3mm:
000000 000.0- 00000 0:00.0- 00000 000.0 00000 000.0 00000 020020028
000.0 000.0 000.0 000.0 000.0
000000 000.0 00000 000.0- 00000 000.0- 00000 .....0000 00000 00000:
000.0 000.0 000.0 000.0 0000
000000 000.0 00000 $000.0- 00000 000.0- 00000 000.0 00000 :0.
000.0 000.0 000.0 000.0 000.0
3.00 on :00 0 :00: :00 020000060200
02-; 05.10.0230 wk; 05-35-0- mFé 05-05230. 0.0-$00M mbq<>nr FEM

 

 

 

 

 

 

 

 

 

 

 

000.0280 0.0 0:020

133

TABLE 5.7
ESTIMATED COEFFICIENTS AND STANDARD ERRORS FOR CIRCUIT VARIABLES:
CIVIL RIGHTS AND LIBERTIES

 

 

Circuit Coefﬁcient Standard Error t Statistic

lst -0.314 0.132 -2.373
2nd -0.139 0.110 -1.262
3rd -0.053 0.118 -0.447
4th 0.192 0.130 1.470
5th 0048 0.088 0545
6th 0.007 0.1 18 0.060
7th 0.125 0.098 1.280
8th 0.280 0.113 2.465
9th 0144 0.092 -1.567
10th 0.039 0.129 0.300
11th 0.169 0.161 1.048
DC -0.049 0.126 -0.392

 

8.0V0h00...
00.0v0:
000 v0...

609900 005 80 2 083 2.00 80065 8005:2000 5 0030052 “ma-OZ

 

 

 

 

134

 

 

 

 

 

 

 

000000 000000 000000 .0080
000.0 000.0 000.0
... 0 00. 0- 000 .0 00000 000.0 00000 000.0 00000 0095200200 0
000.0 000.0 000.0
33.000.0- 000.0- 000000 000.0 000000 000.0 00 0000 00>E<>0mmzoo 0
000.0 000.0 000.0 00000: 0
000.0 000.0 00000 000.0- 00000 000.0- 000000 0200500200 0
000.0 000.0 000.0 0.2000: 0
000.0 000.0 00000 3.0000- 0000 000.0- 0:00 000000: 0
000.0 000.0 000.0
05003-0 05.750 000,090 05005-0 00.050 05005-0 00.050
0.000QO055 HZmDmumMm Ewan-0055 .02QO0th .3<
m>F<>meZOO m>0h<>mmmZOU 3mm:
.0> 05000005

 

 

 

 

 

 

0000000005 oz< 0:550 .0020 :0sz “5 20000000200 0005000000: 0000 020580200 02553000 ”5 00:50
0.0 mam/0:-

 

135

00.30:...

“HUN-«0. -

00.0 v A:

00.01..

00003080002 .Bcovoooa o>000>homaoo 00:0 :23: 00.0 02
:38 80:20 000200002000 E 0038: Z 0300085 gag-60:8 505 :33 22: “00326.0 0.03 000208200 00005: 005 0808:0000 855.00% 2500000 .0. “mi-OZ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0000000 0000000 00000000 00.000000 8002028

000. 0- 000.0- 000.0- 000.0- 000.0 000.0 1.000.0- 0000. 0.00000200on

00000000 0000000 0000000 000000000 000002028

000.0- 000.0- 000.0- 000.0- 000.0 000.0 000.0- 000.0- -300:
on E0 0 E00 :00

000 :00: 00000000 0000000 00 0:00: 00002028

...-0.000.0- 0000- 0000.0. 000.0- 000.0- 000.0- 000.0- 000.0- 00000000000600

00000000 00000000 000000000 000000000 0000202000

000.0- 000.0- 000.0 000.0 :000 .0 00 0.0 000.0- 000.0- -3005
:00 E0 :00 :00

0000000 000000000 00000000 0000000 00.002028

000. 0- 000.0- 000.0- 000.0- 000.0 000.0 :00- 000.0- 0500000028

00000 0000000 0000000 0000000 00002028

000.0- 000.0- 2,000.0- 000.0- 000.0 000.0 00 0. _ 0 :0 00000:
$.50 QM»,- QZN hm—

mDA<>-.0- muzmmmmhmﬂ mag-«>3.- mozmmmwbﬂ mD-0<>.-0. mUme-EEQ 05-0059 momen-EED

 

000000.05 Qz< 0:550 0.20
50020.0 00 20000000200 05000-5050 0.20. 058:0 0000 0205060200 0020090000200 .0> 0500005 02500-5000 ..5 000.050
m6 mam/00.0-

 

CONCLUSION

The three sets of Supreme Court decisions which were analde here were favorably
applied by the Courts of Appeals in about three-fourths of their decisions. This is strong
evidence of these judges’ internalization of the norm of stare decisis, or deference to the
decisions of higher courts. The following excerpt from a circuit court opinion summarizes
the position which seems to be espoused by many circuit jurists:1

For us, of course, there is no question whether to adhere to
Jones and McCraty; they are part of our marching orders,
mandates which we can either obey or seek other work.
Bhandari v. First National Bank of Commerce, 829 F .2d
1343, 1349 (5th Cir. 1987).

However, due to the infrequency of Supreme Court review of lower court decisions,
those who study the Courts of Appeals continue to point out the independence which judges
are free to exercise. These scholars’ concern is realized in the statements which immediately
followed the above acknowledgment of the authority of Supreme Court precedent:

But that is not the issue. The issue is whether, because the
Supreme Court reasoned in a particular manner on one
subject, we are obliged to extend that reasoning, when it
seems to us -- and not to us only -- severely ﬂawed, to a new

series of situations to which the Court has not yet spoken. Id.
at 1349.

 

1The opinion author refers here to two past Supreme Court decisions: Jones v. Alﬁ'ed
Mayer Co., 392 US. 409 (1968), and Runyon v. McCrary, 427 US. 160 (1976).

136

137

Because the absence of direct Supreme Court oversight creates opportunities for
policy-minded judges to deviate from precedent, and because the Courts of Appeals are
charged almost exclusively with the task of ensuring the consistent interpretation and
application of federal law, it is critical to determine what motivates these judges. In other
words, when supervision and sanctions from the High Court are rare, do circuit court judges
feel constrained in their decision making by Supreme Court doctrine, or do they feel free to
pursue their own policy goals? Identifying the factors that inﬂuence the treatment of
precedent in the Courts of Appeals has been the purpose of this study.

The ﬁrst step in addressing this question was to explore circuit court treatment of four
major Supreme Court decisions (Chapter 3). This preliminary examination focused on the
clarity and authoritativeness of precedent. Speciﬁcally, I analyzed the Supreme Court’s
decisions and reasoning in these cases in order to determine whether characteristics of the
Court’s rulings were related to their application in the Courts of Appeals. I found that the
extent of Supreme Court agreement on the outcomes seemed to affect responses to the
decisions in the lower courts, as appellate panels treated more favorably those decisions
which had garnered the support of the justices.

Such a deﬁnitive assertion may not be made regarding decisions which were
characterized as ambiguous, based upon the divergent views expressed in separate opinions
and the legal provisions which were involved. One decision in which a constitutional
amendment was at issue and in which concurring and dissenting opinions were written
received disparate treatment in the circuit courts, while a similarly equivocal decision was
followed consistently. I concluded that these legal considerations provided an incomplete

account of appeals court behavior in these cases.

138

The analysis of Chapter 4 buttressed this inference. Here I incorporated attitudinal
measures for appeals court judges with legal factors in a multivariate model of the treatment
of decisions from the Supreme Court’s 1987 Term. I found that the policy preferences of
circuit judges were indeed signiﬁcant inﬂuences in their application of precedent, as these
judges tended to be more supportive of a precedent which comported with their personal
views; however, the importance of judicial ideology was limited to cases involving civil
rights and liberties and criminal procedure. These are precisely the issues raised in the cases
which were examined in Chapter 3, so it is understandable that legal elements did not
provide a comprehensive explanation of appeals court treatment of these decisions.

In contrast to the limited relevance of personal attitudes and values, legal factors were
crucial determinants of the actions of circuit judges with respect to the broad range of 1987
Term decisions. When a Supreme Court decision was equivocal, as evidenced by the
application of a constitutional provision, the issuance of separate opinions, or complex
subject matter, appellate panels were much less likely to afford a positive treatment to the
decision. Appeals court judges also seemed to be affected by the likelihood of Supreme
Court review, as they were substantially more deferential to precedents which had important
policy implications.

Many of these ﬁndings were replicated in Chapter 5. Because of the additional,
attitudinal forces which motivated appeals court judges in civil rights and liberties and
criminal procedure cases, as established in Chapter 4, I restricted this analysis to treatment
of the Supreme Court’s civil rights and liberties decisions. In conﬁrmation of the indications
in Chapter 4, circuit court judges applied civil rights and liberties precedents in accordance

with their own attitudes and values. Clearly, as has been suggested in other studies of

139

judicial behavior, these are highly salient issues which appeal to the personal views of judges
and encourage them to pursue their favored policy outcomes.

In conjunction with personal preferences, legal considerations were signiﬁcantly
related to the treatment of civil rights and liberties precedents in the Courts of Appeals. The
circuit courts exhibited attentiveness to the Supreme Court’s treatment of its own decisions
by selectively adopting precedents which the Court had subsequently modiﬁed, while
precedents which were not altered in ensuing Court decisions seemed to grow more
authoritative over time. As with the decisions of the 1987 Term, ambiguous civil rights
decisions did not fare well in the intermediate federal courts. The publication of several
concurring opinions and the application of constitutional provisions seemed to foster
confusion in the lower courts, as such decisions were followed only sporadically.

Each of the three analyses in this study suggested that the level of Supreme Court
consensus may serve as an indicator to the lower courts of a decision’s legitimacy and may
thus inﬂuence judicial responses to Supreme Court decisions. In Chapter 3, anecdotal
evidence of the importance of Court support for a decision was presented. In the models of
Chapters 4 and 5, this element manifested itself differently. While the circuit courts
responded negatively to precedents of the 1987 Term which were established by a minimum
winning Court coalition, they were signiﬁcantly more likely to obey civil rights and liberties
precedents when the Court was unanimous in both its ruling and its reasoning. Regardless
of the way in which it is operationalized, the extent of Supreme Court agreement is obviously
crucial in accounting for lower court compliance.

Much of the behavior of Courts of Appeals judges with respect to the treatment of

precedent may be motivated by the perceived likelihood of Supreme Court review. These

140

judges may be more apt to adhere to unanimous precedents and to disregard narrowly
established ones in part because the probability of garnering the votes of four justices for
review is substantial in the former case and only slight in the latter. Additionally, circuit
judges may apply more favorably those precedents which relate to important political or
social issues because they believe that the High Court is more concerned about the resolution
of such cases. The importance of the age of precedent also suggests that appellate judges
may be review-minded; when a precedent has been ﬁrmly established in Supreme Court
jurisprudence, the justices should be more attentive to its application in the lower courts.
As a further test of the proposition that appeals court judges wish to avoid Supreme
Court reversal, I included a variable in the models of the 1987 Term decisions and of the
civil rights and liberties decisions which measured the change in the ideological composition
of the Court since a precedent was handed down. This variable may serve as an indirect
indicator of the likelihood of review. If circuit court judges do, in fact, take this into account
their decision making, they should be more deferential when the ideology of the current
Court has moved in the direction of a past precedent; alternatively, when there appears to be
less attitudinal support for a precedent, these judges should not feel similarly bound by the
precedent. This hypothesis was conﬁrmed in both analyses, suggesting that the perceived
probability of review is a critical component in any explanation of circuit court behavior.
It is apparent from the case studies and the model estimations presented here that
there are some situations in which appeals court judges take advantage of the opportunity to
pursue their own policy goals. When the Supreme Court’s intentions are unclear, when there
is substantial disagreement among the justices, and when they care deeply about the outcome

of a case, appellate judges tend to deviate from strict adherence to precedent. Alternatively,

141

circuit judges are more faithful to Supreme Court mandates when the High Court is united
in rendering its decision, when a precedent has become entrenched in Court doctrine, and
when they perceive that a deviant ruling is likely to be reviewed.

Upon concluding that circuit court judges do indeed behave independently in some
situations, I analyzed the effect this discretionary decision making had on the uniformity of
national law. I found that in a few circuits the motivation of stare decisis was strong, as
these circuits exhibited high rates of following precedent. In several circuits, adherence to
precedent was ideologically determined; some circuits were more deferential to conservative
precedents while others favored liberal ones. However, treatment of precedent in a majority
of circuits was substantially similar to that of the Courts of Appeals as a whole; that is, they
adopted precedent, regardless of its ideological tenor, in about three-fourths of their
decisions. This examination suggests that much of the apprehension regarding the diverse
application of federal law is unwarranted.

In spite of the convincing evidence which has been presented in this study, the
ﬁndings remain incomplete. Analyses of cases which were poorly ﬁt by the model indicated
that the majority of these outliers were instances in which the Courts of Appeals did not
follow a relevant Supreme Court decision. Thus, while I have determined a number of
scenarios in which appellate panels will be less deferential to precedent, some of these
circumstances have not been identiﬁed.

A number of scholars have pointed to institutional differences among the circuit
courts, ranging from whether district court judges should prepare written justiﬁcations for
their decisions to the deﬁnition of what constitutes auto theﬁ (see Carp and Rowland 1983,

ch. 4). In analyses of other levels of courts, researchers have found such institutional

142

contexts and internal rules to be pivotal to explanations of judicial decision making (see, e. g.,
Brace and Hall 1993). This may be a fruitful avenue of exploration for Courts of Appeals’

behavior as well.

APPENDICES

143

 

 

000.0 000.0 000.0 000.0 000.0 000.0 85000
000.0- 000.0 000.0- 000.0- 000.0 00.0- 0002088
00 0.0- 000.0 000.0- 000.0- 00 0 .0 00 0.0- 0000320 0080030008
00 0.0- 000.0 000 .0- 000. 0 - 000.0 00 0 .0- 0.000.000 000.0808
000.0- 00 0 .0 000.0- 000.0- 000.0 000.0- .0032 35-800
009005003

000.0 000.0 000.0 000.0 000.0 000.0 00008058
0.500 08295 5 $525

000.0 000.0 000.0 000.0 000.0 000.0 200880 .8055
x 00208000050 28030000000

0 0 0.0 000.0 000.0 :00 000.0 000.0 02:00: 000 00002 :5
x 00208000080 0303—33
00000002000000.0000
000.0 000.0 000.0 00.0 00000000:

omumﬁSm 0 00.0-0m pats—dam «00305000 omwmtaum u hobm ﬁaﬁgm “ComoEooU oswmua> 000350—me

00000 n 20 00000000 00005005 002003 00000 n 20 0020000 00005005 003

 

200:- 0000 00:0 ”02002000 00200-200005 00 02500 000.0 02005-6080 920. 002005.620 000050.200 00 2000500200
70. 0590-0-

40 0292-0300400

APPENDIX B

FIT AND INFLUENCE DIAGNOSTICS FOR LOGISTIC REGRESSION

The deviance residual is one kind of standardized residual that may be calculated.‘
AXZDG) measures the change in deviance that results from deleting all cases with a particular
pattern (1') of independent variables.2 Thus it provides a measure of how poorly the model
ﬁts a particular pattern. Large values of szm) indicate that the model would ﬁt the data
much better if the jth pattern were deleted. A rule of thumb is that values of AXZD greater
than 4 indicate a signiﬁcant difference in model ﬁt between a model which includes a
particular pattern of independent variables and a model which excludes these cases.
Nineteen patterns, including 138 cases, produced a AxZD higher than 4, with the largest szn
of 9.836 from a pattern comprised of four cases.

AB]. measures the standardized change in estimated parameters that results from
deleting all cases with the jth pattern of independent variables.3 A large value of AB,-

indicates that the jth pattern exerts substantial inﬂuence, where a value greater than 1 is

 

'ag. = i: {2 [I;1n(lj/ ntAPj) + (n; - I;)ln(n} - If) / n}(1 - All?)]}” , wherej represents a
particular pattern of independent variables, and mjis the number of cases with the jth pattern.
The sum of squared deviance residuals, 230‘2 or x20, can be used to test whether a given logit
model is signiﬁcantly worse than a perfect-ﬁt model.

ZAXZDO) = -2[lnL - lnLj], where L is the likelihood of a model estimated from the ﬁrll
data set, and Lj is the likelihood for the same model estimated after deleting all cases with
the jth pattern.

3A3]. = rjzhj / (l - h)’, where it. is the leverage of any one jth-pattern case times the
number of times the pattern occurs. rj is the Pearson residual, calculated as jY - I" ’}P /
x/mjAPJU -"P).

144

145

considered large. Only one pattern, consisting of eleven cases, produced a AB greater than
1. This pattern produced a AB of 1.15 and a AxZD of 9.628.

Figure 3.1 is a graph of szn against predicted probability, with the size of the
plotting symbols proportional to AB. Thus it combines information about ﬁt and inﬂuence.
Each circle represents a particular combination of independent variables. The larger the
circle, the more inﬂuential a particular combination is. A case where precedent is followed
(Y=1) despite a low predicted probability of following will fall toward upper left. A case
where precedent is not followed (Y=0) despite a high predicted probability of following will
fall toward upper right.

Pattern 267 represents the four cases with a szn of 9.836. The model predicts that,
for this combination of independent variables, the probability of following precedent is 0.68.
However, precedent is not followed in any of these four cases. Pattern 204 represents the
eleven cases with large values of both szn and AB. The model predicts that, for this
combination of independent variables, the probability of following precedent is only 0.69,

but precedent is followed in all eleven of these cases.

h-l dd

146

10 J @

 

 

 

 

 

 

® a
4-4
T '“ I " ‘ T ”" '1
4 .6 a 1
MM)
FIGUREB.1

MODEL FIT AND INFLUENCE OF OUTLYING CASES

LIST OF REFERENCES

LIST OF REFERENCES

Aldrich, John H., and Forrest C. Nelson. 1984. Linear Probability, Logit, and Probit Models.
Beverly Hills: Sage.

Atkins, Burton M. 1972. “Decision-Making Rules and Judicial Strategy on the United States
Courts of Appeals.” Western Political Quarterly 25:626-642.

Atkins, Burton M., and William Zavoina. 1974. “Judicial Leadership on the Court of
Appeals: A Probability Analysis of Panel Assignment in Race Relations Cases in the
F iﬁh Circuit.” American Journal of Political Science 18:701-71 1.

Barrow, Deborah J ., and Thomas G. Walker. 1988. A Court Divided: The F iﬁh Circuit Court
of Appeals and the Politics of Judicial Reform. New Haven: Yale University Press.

Baum, Lawrence. 1976. “Implementation of Judicial Decisions: An Organizational
Analysis.” American Politics Quarterly 4286-114.

Baum, Lawrence. 1977. “Judicial Impact as a Form of Policy Implementation.” In Public
Law and Public Policy, edited by John A. Gardiner. New York: Praeger.

Baum, Lawrence. 1978. “Lower Court Response to Supreme Court Decisions: Reconsidering
a Negative Picture.” Justice System Journal 3 :208-21 9.

Baum, Lawrence. 1994. “Specialization and Authority Acceptance: The Supreme Court and
Lower Federal Courts.” Political Research Quarterly 47:693-704.

Baum, Lawrence, Sheldon Goldman, and Austin Sarat. 1981. “The Evolution of Litigation
in the Federal Courts of Appeals.” Law and Society Review 16:291-3 09.

Blumrosen, Alfred W. 1972. “Strangers in Paradise: Griggs v. Duke Power Co. and the
Concept of Employment Discrimination.” Michigan Law Review 71 :59-107.

Brace, Paul, and Melinda Gann Hall. 1993. “Integrated Models of Judicial Dissent.” Journal
of Politics 55:914-931.

Brenner, Saul, and Harold J. Spaeth. 1995. Stare Indecisis. New York: Cambridge
University Press.

147

148

Caminker, Evan H. 1994. “Why Must Inferior Courts Obey Superior Court Precedents?”
Stanford Law Review 46:817-873.

Canon, Bradley C. 1973. “Reactions of State Supreme Courts to a US. Supreme Court Civil
Liberties Decision.” Law and Society Review 8: 109-134.

Canon, Bradley C. 1991. “Courts and Policy: Compliance, Implementation, and Impact.”
In The American Courts: A Critical Assessment, edited by John Gates and Charles
A. Johnson. Washington, DC: Congressional Quarterly Press.

Canon, Bradley C., and Kenneth Kolson. 1971. “Compliance with Gault in Rural America:
The Case of Kentucky.” Journal of Family Law 10:300-326.

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

Carp, Ronald A., and Ronald Stidham. 1996. Judicial Process in America, 3rd Edition.
Washington, DC: Congressional Quarterly Press.

Combs, Michael W. 1982. “The Policy-Making Role of the Courts of Appeals in Northern
School Desegregation: Ambiguity and Judicial Policy-Making.” Western Political
Quarterly 35:359-375.

Comment. 1972-3. “Kirby v. Illinois: A New Approach to the Right to Counsel.” Iowa Law
Review 58:404-420.

Comment. 197 9. “The Business Necessity Defense to Disparate-Impact Liability Under Title
VII.” University of Chicago Law Review 46:911-934.

Cook, Beverly B. 1993. “Measuring the Signiﬁcance of US. Supreme Court Decisions.”
Journal of Politics 55:1 127-1 140.

Davis, Sue, and Donald R. Songer. 1988-9. “The Changing Role of the United States Courts
of Appeals: The Flow of Litigation Revisited.” Justice System Journal 13:323-340.

Dix, George E. 1985. “Non-Arrest Investigatory Detentions in Search and Seizure Law.”
Duke Law Journal 1985:849-959.

Douglas, William O. 1949. “Stare Decisis.” Columbia Law Review 49:735-758.
Emmert, Craig F. 1992. “An Integrated Case-Related Model of Judicial Decision Making:

Explaining State Supreme Court Decisions in Judicial Review Cases.” Journal of
Politics 54:543-552.

...—
0 .___..—-I

1.. ‘_; ‘7 _

 

149

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

Epstein, Lee, Jeffrey A. Segal, Harold J. Spaeth, and Thomas G. Walker. 1996. The Supreme
Court Compendium: Data, Decisions, and Developments, 2nd Edition. Washington,
DC: Congressional Quarterly Press.

Frank, Jerome. 1963. Courts on Trial: Myth and Reality in American Justice. New York:
Atheneurn.

Furnish, Hannah Arterian. 1982. “A Path Through the Maze: Disparate Impact and Disparate
Treatment Under Title VII of the Civil Rights Act of 1964 After Beazer and
Burdine.” Boston College Law Review 23 :419-445.

George, Tracey E. 1997. “The Politics of Deciding Appeals: A Study of the En Banc
Decisions of the US. Court of Appeals for the Fourth Circuit.” Paper presented at the
Annual Meeting of the Midwest Political Science Association.

George, Tracey E., and Lee Epstein. 1992. “On the Nature of Supreme Court Decision
Making.” American Political Science Review 86:323-337.

Goldman, Sheldon. 1966. “Voting Behavior on the United States Courts of Appeals, 1961-
1964.” American Political Science Review 60:370-385.

Goldman, Sheldon. 1975. “Voting Behavior on the United States Courts of Appeals
Revisited.” American Political Science Review 69:491-506.

Granato, Jim, Ronald Inglehart, and David Leblang. 1996. “The Effect of Cultural Values
on Economic Development: Theory, Hypotheses, and Some Empirical Tests.”
American Journal of Political Science 40:607-631.

Gruhl, John. 1980. “The Supreme Court’s Impact on the Law of Libel: Compliance by
Lower Federal Courts.” Western Political Quarterly 33:502-519.

Hamilton, Lawrence C. 1992. Regression with Graphics: A Second Course in Applied
Statistics. Belmont: Duxbury Press.

Heiner, Ronald A. 1986. “Imperfect Decisions and the Law: On the Evolution of Legal
Precedent and Rules.” Journal of Legal Studies 15:227-261.

Howard, J. Woodford, Jr. 1981. Courts of Appeals in the Federal Judicial System: A Study
of the Second, F iﬂh, and District of Columbia Circuits. Princeton: Princeton
University Press.

150

Hurwitz, Mark S., and Malia Reddick. 1996. “Ambivalence at the Courts of Appeals: Does
the Legal or Attitudinal Model Apply?” Paper presented at the Annual Meeting of
the Midwest Political Science Association.

Hutchinson, Dennis J. 1979. “Unanimity and Desegregation: Decisionmaking in the
Supreme Court, 1948-1958.” Georgetown Law Journal 68:1-87.

Johnson, Charles A. 1979. “Lower Court Reactions to Supreme Court Decisions: A
Quantitative Examination.” American Journal of Political Science 23 :792-804.

Johnson, Charles A. 1987. “Law, Politics, and Judicial Decision Making: Lower Federal
Court Uses of Supreme Court Decisions.” Law and Society Review 21 :325-340.

Johnson, Charles A., and Bradley C. Canon. 1984. Judicial Policies: Implementation and
Impact. Washington, DC: Congressional Quarterly Press.

Kelly, Michael E. 1973. “Criminal Procedure: Restricting Right to Counsel at Lineups.”
North Carolina Law Review 51:630-639.

King, Gary. 1989. Unifying Political Methodology: The Likelihood Theory of Statistical
Inference. Cambridge: Cambridge University Press.

LaF ave, Wayne R. 1968. “‘Street Encounters’ and the Constitution: Terry, Sibron, Peters,
and Beyond.” Michigan Law Review 67:39-126.

Maltzrnan, Forrest, and Paul J. Wahlbeck. 1996. “May It Please the Chief? Opinion
Assignments in the Rehnquist Court.” American Journal of Political Science 40:421-
443.

Manwaring, David R. 1968. “The Impact of Mapp v. Ohio.” In The Supreme Court as
Policy-Maker: Three Studies on the Impact of Judicial Decisions, edited by DH.
Everson. Carbondale: Southern Illinois Press.

Marshall, Thomas R. 1989. “Policy Making and the Modern Court: When Do Supreme
Court Rulings Prevail?” Western Political Quarterly 42:493-507.

Marshall, William P. 1986. “‘We Know It When We See It.’ The Supreme Court and
Establishment.” Southern California Law Review 59:495-550.

Murphy, Walter F. 1959. “Lower Court Checks on Supreme Court Power.” American
Political Science Review 53: 1017-103 1 .

Pacelle, Richard L., Jr., and Lawrence Baum. 1992. “Supreme Court Authority in the
Judiciary: A Study of Remands.” American Politics Quarterly 20:169-191.

151

Paulsen, Michael A. 1986. “Religion, Equality, and the Constitution: An Equal Protection
Approach to Establishment Clause Adjudication.” Notre Dame Law Review 61:311-
371.

Peltason, J .W. 1961. F iﬁy-Eight Lonely Men: Southern Federal Judges and School
Desegregation. Urbana: University of Illinois Press.

Read, Frank T. 1969-70. “Lawyers at Lineups: Constitutional Necessity or Avoidable
Extravagance?” UCLA Law Review 17:339-407.

Reid, Traciel V. 1988. “Judicial Policy-Making and Implementation: An Empirical
Examination.” Western Political Quarterly 41:509-527.

Richardson, Richard J ., and Kenneth N. Vines. 1970. The Politics of Federal Courts: Lower
Courts in the United States. Boston: Little, Brown and Company.

Ripple, Kenneth F. 1980. “The Entanglement Test of the Religion Clauses: A Ten Year
Assessment.” UCLA Law Review 27:1195-1239.

Rohde, David W. 1972. “Policy Goals and Opinion Coalitions in the Supreme Court.”
Midwest Journal of Political Science 16: 208-224.

Romans, Neil T. 1974. “The Role of State Supreme Courts in Judicial Policy Making:
Escobedo, Miranda, and the Use of Judicial Impact Analysis.” Western Political
Quarterly 27:38-59.

Sanders, Francine. 1995. “Brown v. Board of Education: An Empirical Reexamination of Its
Effects on Federal District Courts.” Law and Society Review 29:731-756.

Schick, Marvin. 1970. Learned Hand ’s Court. Baltimore: The Johns Hopkins Press.

Schubert, Glendon. 1974. Judicial Policy Making, 2nd Edition. Glenview, 111.: Scott,
Foresman.

Segal, Jeffrey A. 1984. “Predicting Supreme Court Cases Probabilistically: The Search and
Seizure Cases, 1962-1981.” American Political Science Review 78:891-900.

Segal, Jeffrey A., and Albert D. Cover. 1989. “Ideological Values and the Votes of US.
Supreme Court Justices.” American Political Science Review 83: 557-565.

Segal, Jeffrey A., Lee Epstein, Charles M. Cameron, and Harold J. Spaeth. 1995.
“Ideological Values and the Votes of US. Supreme Court Justices Revisited.”
Journal of Politics 57:812-823.

152

Segal, Jeffrey A., Donald R. Songer, and Charles M. Cameron. 1995. “Decision Making on
the US. Courts of Appeals.” In Contemplating Courts, edited by Lee Epstein.
Washington, DC: Congressional Quarterly Press.

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

Simson, Gary J. 1987. “The Establishment Clause in the Supreme Court: Rethinking the
Court’s Approach.” Cornell Law Review 72:905-935.

Sobel, Nathan R 1971. “Assailing the Impermissible Suggestion: Evolving Limitations on
the Abuse of Pre-Trial Criminal Identiﬁcation Methods.” Brooklyn Law Review
38:261-325.

Solan, Lawrence M. 1993. The Language of Judges. Chicago: University of Chicago Press.

Songer, Donald R. 1982. “Consensual and Nonconsensual Decisions in Unanimous Opinions
of the United States Courts of Appeals.” American Journal of Political Science
26:225-239.

Songer, Donald R. 1987. “The Impact of the Supreme Court on Trends in Economic Policy
Making in the United States Courts of Appeals.” Journal of Politics 49:830-841.

Songer, Donald R. 1988a. “Alternative Approaches to the Study of Judicial Impact.”
American Politics Quarterly 16:425-444.

Songer, Donald R. 1988b. “Case Selection in Judicial Impact Research.” Western Political
Quarterly 41:569-582.

Songer, Donald R. 1991. “The Circuit Courts of Appeals.” In The American Courts: A
Critical Assessment, edited by John Gates and Charles A. Johnson. Washington,
DC: Congressional Quarterly Press.

Songer, Donald R., and Susan Haire. 1992. “Integrating Alternative Approaches to the Study
of Judicial Voting: Obscenity Cases in the US. Courts of Appeals.” American
Journal of Political Science 36:963-982.

Songer, Donald R., Jeffrey A. Segal, and Charles M. Cameron. 1994. “The Hierarchy of
Justice: Testing a Principle-Agent Model of Supreme Court-Circuit Court
Interactions.” American Journal of Political Science 38:673-696.

Songer, Donald R., and Reginald S. Sheehan. 1990. “Supreme Court Impact on Compliance
and Outcomes: Miranda and New York Times in the United States Courts of
Appeals.” Western Political Quarterly 43:297-316.

153

Steamer, Robert J. 1976. “Contemporary Supreme Court Decisions.” Paper delivered at the
Annual Meeting of the Southern Political Science Association.

Tarr, G. Alan. 1977. Judicial Impact and State Supreme Courts. Lexington, Mass:
Lexington Books.

Tate, C. Neal. 1981. “Personal Attribute Models of Voting Behavior of US. Supreme Court
Justices: Liberalism in Civil Liberties and Economics Decisions, 1946-1978.”
American Political Science Review 75:355-367.

Tate, C. Neal, and Roger Handberg. 1991. “Time Binding and Theory Building in Personal
Attribute Models of Supreme Court Voting Behavior, 1916-1988.” American Journal
of Political Science 35:460-480.

Vines, Kenneth N. 1963. “The Role of Circuit Courts of Appeal in the Federal Judicial
Process: A Case Study.” Midwest Journal of Political Science 7:305-319.

Walker, Darlene. 1978. “The Burger Court and Civil Liberties: The Problem of Moving the
Law Backward.” In Nationalizing Government, edited by Theodore J. Lowi and Alan
Stone. Beverly Hills: Sage Publications.

Wasby, Stephen L. 1970. The Impact of the United States Supreme Court: Some
Perspectives. Homewood: The Dorsey Press.

Wenner, Lettie McSpadden, and Lee E. Dutter. 1989. “Contextual Inﬂuences on Court
Outcomes.” Western Political Quarterly 41:113-134.

Western, Bruce. 1995. “Concepts and Suggestions for Robust Regression Analysis.”
American Journal of Political Science 39:786-817.

Williamson, Richard A. 1982. “The Dimensions of Seizure: The Concepts of ‘Stop’ and
‘Arrest’.” Ohio State Law Journal 43:771-818.

Wilson, Hugh Steven. 1972. “A Second Look at Griggs v. Duke Power Co.: Ruminations
on Job Testing, Discrimination, and the Role of the Federal Courts.” Virginia Law
Review 58:844-874.

Woodward, Bob, and Scott Armstrong. 1979. The Brethren. New York: Simon and Schuster.

LIST OF CASES

LIST OF CASES

Adams v. Williams, 407 US. 143 (1972).

Aguilar v. Felton, 473 US. 402 (1985).

Albemarle Paper Co. v. Moody, 422 US. 405 (1975).
Armstrong v. Board of Education in the City of Birmingham, 323 F.2d 337 (5th Cir. 1963).
Bhandari v. First National Bank of Commerce, 829 F.2d 1343 (5th Cir. 1987).
Board of Education v. Allen, 392 US. 236 (1968).

Board of Regents v. Bakke, 438 US. 265 (1978).

Bowen v. Kendrick, 487 US. 589 (1988).

Brown v. Board of Education, 347 US. 483 (1954).

Bruno & Stillman, Inc. v. Globe Newspaper Co., 633 F.2d 583 (lst Cir. 1980).
Cammack v. Waihee, 932 F.2d 765 (9th Cir. 1991).

Carter v. Gallagher, 452 F.2d 315 (8th Cir. 1971).

Church of Scientology Flag Service Org, Inc. v. City of Clearwater, 2 F.3d 1514 (11th Cir.
1993).

Cleveland Board of Education v. LaFleur, 414 US. 632 (1974).

Committee for Public Education and Religious Liberty v. Nyquist, 413 US. 756 (1973).
Dothard v. Rawlinson, 433 US. 321 (1977).

Dunaway v. New York, 442 US. 200 (1979).

Escobedo v. Illinois, 378 US. 478 (1964).

154

155

Estate of Thornton v. Caldor, Inc., 472 US. 703 (1985).

F elmeister v. Oﬂice of Attorney Ethics, 856 F.2d 529 (3rd Cir. 1988).
Florida v. Royer, 460 US. 491 (1983).

Gerstein v. Pugh, 420 US. 103 (1975).

Gertz v. Welch, 418 US. 323 (1974).

Grand Rapids School District v. Ball, 473 US. 373 (1985).

Green v. Missouri Pac. R.R. Co., 523 F.2d 1290 (8th Cir. 1975).
Griggs v. Duke Power Co., 401 US. 424 (1971).

Harlow v. Fitzgerald, 457 US. 800 (1982).

Hayes v. Florida, 470 US. 811 (1985).

Hunt v. McNair, 413 US. 734 (1973).

Johnson v. Pike Corp. of America, 332 F.Supp. 490 (CD. Cal. 1971).
Jones v. Alﬁ'ed Mayer Co., 392 US. 409 (1968).

Katz v. United States, 389 US. 347 (1967).

Kirby v. Illinois, 406 US. 682 (1972).

Lamb’s Chapel v. Center Moriches Union Free School District, 508 US. 384 (1993).
Lanphere & Urbaniak v. Colorado, 21 F.3d 1508 (10th Cir. 1994).
Lau v. Nichols, 414 US. 563 (1974).

Lemon v. Kurtzman, 403 US. 602 (1971).

Lynch v. Donnelly, 465 US. 668 (1984).

McDonnell Douglas Corp. v. Green, 411 US. 792 (1973).

Marsh v. Chambers, 463 US. 783 (1983).