Ill/1M!!!III/HMl/Illlllll/IIIIIIIIIIHllIlI/lI/l/H/U/IUIII , + 300546 8537 f ; LIBRARY Michigan State University This is to certify that the dissertation entitled A Cognitive-Cybernetic Theory of Judicial Decision Making: A Theory and Empirical Analysis of United States Supreme Court Decision Making presented by Timothy M. Hagle has been accepted towards fulfillment of the requirements for Ph.D. degreein Political Science / . LW \ Majoryésor , . / , Datej fl/V KO/ 0-12771 MS U is an Affirmative Action/Equal Opportunity Institution MSU RETURNING MATERIALS: Place in book drop to remove this checkout from LIBRARIES . your record. FINES will be charged if book is returned after the date MAGICZ stamped below. SEP 21 1998 gm 1 3 W ,’ 1.." - film-W we}; (989 02W A COGNITIVE-CYBERNETIC THEORY OF JUDICIAL DECISION MAKING: A THEORY AND EMPIRICAL ANALYSIS OF UNITED STATES SUPREME COURT DECISION MAKING BY Timothy Mark Hagle A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirement for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1988 12 pr ml] “8' Wei! liqu x. S \ w Q s. Q ABSTRACT A COGNITIVE-CYBERNETIC THEORY OF JUDICIAL DECISION MAKING: A THEORY AND EMPIRICAL ANALYSIS OF UNITED STATES SUPREME COURT DECISION MAKING BY Timothy M. Hagle This study examines United States Supreme Court decision making from a cognitive-cybernetic perspective. Neither the comprehensive rational model of decision making or the tradi- tional theory of legal decision making can account for the ability of judicial decision makers to operate effectively in a highly complex decisional environment requiring the pro— cessing of voluminous amounts of information. The cognitive- cybernetic theory of judicial decision making is character- ized by four main principles: simplification of the complex problem, a stable decisional environment, a focus on a small number of major factors, and a limited choice set. The search and seizure decisions of the United States Supreme Court from 1961 through 1987 are used to test the theory. The dichotomous choice set, concentration on major factors, and simplification of the decision making process result in a form of judicial balancing. The major factors Weigh for or against a specific choice; the specific combina— tion determining the outcome. The maximum-likelihood tech~ nique probit is used to test the model derived from the theorj ion wt ular c tive-c not ac Tw- jUStiC¢ justice The dec also ex data is vidual - allows , sum of : The fill at c justices ationali eXplanat‘ examinatj increaSeS internall theory. Probit emphasizes the weighing of factors in a fash- ion which increases or decreases the probability of a partic— ular choice. This technique is more analogous to the cogni- tive-cybernetic theory than regression techniques which do not account for the limited choice set. Two judicial decision makers are examined: the individual justices and the Supreme Court. The estimated votes of the justices are aggregated to estimate the Court's decisions. The decisions of the Court as the judicial decision maker are also examined. Estimating the Court's decisions from primary data is more appealing than aggregating estimations of indi- vidual votes. Examination of both judicial decision makers ‘ allows consideration of whether the whole is greater than the sum of its parts. The model derived from the theory proves highly success— ful at correctly classifying the votes of the individual justices and the decisions of the Court. The variables oper- ationalized from the hypotheses provide high degrees of i explanatory power for both judicial decision makers. The examination of the Court's decisions demonstrates how increases in the stability of the decisional environment, internally and externally, increase the ability of the model to correctly classify the decisions. Copyright by TIMOTHY MARK EAGLE 1988 To my father Gordon E. Hagle who was patient Profes: througl its gre ous suk proved in one thanks . ACKNOWLEDGMENTS Professor Harold J. Spaeth provided advice and guidance throughout the course of this project. His guiding hand had its greatest impact when it was applied most lightly. Numer- ous substantive and strategic discussions with Joseph Ignagni proved to be most helpful. Many more people provided support in one form or another. To them I give a collective sigh of thanks. vi List 0; l. 1111 Juc‘ TABLE OF CONTENTS List of Tables 1. 2. Introduction The Cognitive—Cybernetic Theory of Judicial Decision Making Rationality in Decision Making Human Decision Making General Limitations Cognitive Limitations Human Limitations and Judicial Decision Making Cognitive—Cybernetic Decision Making Data Human Learning Adaptive Strategies Cybernetic Decision Making Cognitive Theory Cognitive-Cybernetic Judicial Decision Making Hypotheses To Be Tested Operationalization, and Methodology Operationalization A. B. l. 2. 3. C. l. 2. 3. 4. 5. 6. Data, A. B. C. Methodology ix l6 16 25 31 33 34 36 45 60 63 85 99 99 102 116 5. SI Notes Append Append List 0 4. Results A. The B. The 5. Summary Notes Appendix A. Appendix B. and Analysis Individual Justices Court and Conclusions Cases Contained in the Data Set Cases Cited in Text and Notes List of References viii 130 130 145 177 194 213 219 220 10. ll. 12 13. LIST OF TABLES Probit estimation of individual justices' votes. Aggregation of estimated individual justices' votes. Probit estimation of Court's decisions. Probability fit for estimation of Court's deci— sions. Probit estimation of reduced location choice set model. Probit estimation of reduced location choice set model with LEDCITE. Value systems of U.S. Supreme Court Justices since 1958. Comparison of DECISION by group controlling Court. Probit estimation of reduced location choice set model for the period 1975 through 1987. Probit estimation of reduced location choice set model with variable PERIOD for the period 1975 through 1987. Probit estimation of reduced location choice set model for the period 1975 through 1981. Probit estimation of reduced location choice set model for the period 1981 through 1987. Comparison of classification rates for selected time periods based on probit estimation. ix 131 143 146 151 155 158 161 164 166 169 170 173 175 Th States scient: branche all so: a persp ernment tionall, lUdicia. litical laIldnar} judicial Justice the Mean Pointing were int; and Only arbiters shall We: km“! as classic 6: Follox Clary has Each Case PreCeded i 18 differs. Chapter 1: Introduction The judicial branch is a coequal part of the United States government, and yet it has escaped the degree of scientific scrutiny given to the executive and legislative branches. This is not to say the judicial branch has lacked all scrutiny, only that it has traditionally been viewed from a perspective different from the other two branches of gov— ernment. The executive and legislative branches have tradi- tionally been viewed as political entities. Judges and the judicial branch have fostered the idea that they are nonpo— litical arbiters of the law. In Marbur v. Madison, the landmark United States Supreme Court case which established judicial review under the United States Constitution, Chief Justice John Marshall rhetorically asked who should determine the meaning of the Constitution. He answered himself by pointing to the fact that members of the other two branches were intimately involved in that nasty business of politics, and only judges were qualified to be truly nonpolitical arbiters of the law. These statements by Chief Justice Mar- shall were certainly not the beginning of what is generally known as the "cult of the robe,“ but they are certainly a classic example in American jurisprudence. Following Marshall's reasoning, the study of the judi- ciary has traditionally concentrated on individual cases. Each case must be decided on the basis of cases which have Preceded it. Although it may be acknowledged that each case is different from any other in many ways, past cases must 1 still then case I be ab: rate 1 to ste the de abilit Sis of Of the Kilmer , tree, t asa pa Alt to the 1 there d< raphies' taining CaSQ Stu JudiCial jUrist, c the rem] isnecQsS more Comp This ‘ 2 still be examined to find the general principles which are then to be applied to the present dispute. This reliance on precedent, known in legal terms as stare decisis, and its accompanying detailed examination of each case has caused legal scholars, to paraphrase Wieland, to not be able to see the forest for the trees. To get a more accu— rate picture of the workings of the judiciary it is necessary to step back from the cases. One must remain cognizant of the details, but not to such a degree that they inhibit the ability to see the greater whole. This is not to say analy- sis of individual cases has no place in the scientific study of the judiciary. Indeed, as was pointed out by Joyce Kilmer, there is always a place to appreciate the beauty of a tree, but there are also times when we must consider the tree as a part of the greater forest. Although judicial scholars by and large do not subscribe to the myth that judges are nonpolitical arbiters of the law, there does seem to be substantial interest in judicial biog— raphies, case studies, and narrowly drawn issue areas con— taining only a few observations. Judicial biographies and case studies are certainly useful in interpreting particular judicial decisions, examining the opinions of a particular jurist, or discussing specific points of law, but to optimize the results of such efforts, in terms of scientific study, it is necessary to View such research within the framework of a more comprehensive theory of judicial decision making. This work will construct such a comprehensive theory of judicial decision making by building on a theory of human decisi ration be exa Next, ‘ light 1 tations necesse rationa include and rec A C} Cogniti‘ IIIIIIIIIIIIIIIIIIII3___________________________—_____1IIIlllllllllllllllfiififififiifN 3 decision making. The assumptions and requirements of rationality and the comprehensive rational actor model will be examined in the context of the Cuban missile crisis. Next, these assumptions and requirements will be examined in light of real world and human limitations. Real world limi- tations include various barriers to obtaining the information necessary to fulfill the requirements of the comprehensive rational actor model of decision making. Human limitations include the cognitive limits on the processing, retention, and recall of information. A cybernetic model of decision making supplemented by cognitive theory will be offered as an alternative to the comprehensive rational actor model. Evidence will be pre- sented to show that decision makers, especially those who must deal with inordinate amounts of information, find short— cuts and ways of streamlining the processing of information. In essence, this involves searching for a relatively small number of important cues or factors on which the decision can be based. In this respect, the decision making process is like a servomechanism. The mechanism is built (or trained) to monitor a small number of factors. When the factors change the mechanism changes accordingly. Humans are clearly not automatons, but cognitive limitations and the need to process large amounts of information force decision makers to adopt strategies designed to cope with their limitations. These strategies focus on a limited number of factors in making decisions. Judges are professional decision makers, but they are still in par Court, arrivi limita‘ system form or adopt 5 Process sion. In ' Itakers l are two that mod llOdel in indi. indi‘ dict inte: consu say n faCe What (1975 The c; are those (Beer 1964 described human d805,; every facu IIIIIIIIIIIIIIIIIIIIl33:__________________________________________________________________7__7_W 4 still subject to human cognitive limitations. Judges, and in particular the justices of the United States Supreme Court, must process voluminous amounts of material before arriving at a decision in a case. In addition to cognitive limitations, judges are also constrained by the judicial system itself which places demands on the judges as to the form and content of their decisions. Thus, judges must also adopt strategies and shortcuts designed to allow them to process the information necessary for them to reach a deci- sion. In the behavior of rational actors and human decision makers we do not try to predict individual decisions. There are two reasons for this. First, Lave and March point out that models of individual decision making may not be used to model individuals. They make this point as follows: Although our model is formally a model of individual behavior, we rarely use it to predict individual behavior. Instead, we use it to pre- . dict the behavior of groups of people. If we are . interested in applying our model of choice to consumer behavior, for example, we do not try to say what each individual consumer will do in the face of a price change. Rather, we try to say what the aggregate movement of consumers will be. (1975:134—135) The cybernetic decision makers which are most interesting are those which are exceedingly complex and probabilistic (Beer 1964). This means that their inner workings cannot be described in a complete and precise fashion. With respect to human decision makers, it means we cannot know with certainty every factor which has an impact on a human decision maker. — . ,l’ 5 Even the decision maker may not know, since many factors may operate on a subconscious level. Many of these factors will be specific to the particular individual (§.g., a fear of water due to some childhood mishap). Other factors will be consistently important across individuals, especially if there are similarities in the group or the decision—making situation forces a certain amount of homogeneity into the decisions, as is the case with the justices of the Court. Nevertheless, if we attempt to model the behavior of individ- uals we may be confronted with fluctuations and random errors which cause the model to do poorly in correctly predicting individual behavior. Using the model of individual behavior over a large group of decisions, these fluctuations and ran— dom errors will even out and the model will do much better. Second, as noted by Steinbruner, cognitive theory acknowledges that in terms of substantive content human beings differ so greatly that "empirical generalizations are simply overwhelmed" (1974:95). However, in terms of the structure of beliefs and the decision making process itself there is a greater similarity across individuals. In examining judicial decision making, and in particular the decisions of the United States Supreme Court, one must decide whether to examine the decisions of the individual justices or those of the Court. Can a court composed of several individual justices be considered a decision maker in the theoretical sense? One of the proponents of the cybernetic model does not even assume the decision maker has a consciousness of its own IIIIIIIIIIIIlIIIIlIIIIIlIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIl-IP"I . y n y 6 in order to be considered a cybernetic decision maker (Ashby 1960). Another author notes that it is not necessary to posit a utility function for a goal-seeking organism (Simon 1957:261-273). This allows for a group of individual deci— sion makers to be considered as an individual "actor" or "decision maker." We often speak of large groups of people as if they were an individual decision maker. In interna— tional politics, whole nations may be considered the actor; in elections, whole segments of the population are often analyzed on the basis of their demographic characteristics (g.g., blacks, Catholics, union members); in sports, we talk about what a team must do to beat another team; and in judi— cial politics, we are concerned with the decisions of courts, not individual justices. In addition, certain factors specific to multi—judge courts suggest a preference for examination of the Court's decisions rather than those of the individual justices. When the Court makes a decision, that decision has the same force of law whether the decision was unanimous (§.g., 9—0) or minimum winning (§.g., 5-4). It has been argued that minimum winning decisions may be less stable and more prone to resis— tance (Brenner, Hagle, and Spaeth 1988:2). However, resis- tance can be just as strong against unanimous decisions (§.g. Brown v. Board of Education), and although it is mathemati- cally obvious that only one justice need defect to change the outcome in a minimum winning decision, five of nine justices are still required to form a majority.1 One may wonder if the preferences of the individual justices may simply be summed to arrive at a model for the Court's behavior. A group of individual decision makers may each have utility functions and goal directed behavior and yet the actions of the group may still be suboptimal. Often these suboptimal results are due to a voting paradox or col- lective action problem (Frohlich and Oppenheimer 1978:15-48). More simply, it may be explained by the maxim "the whole is greater than the sum of its parts." Despite the comment that the Court functions like nine separate law firms, interac- tions occur among the justices before, during, and after the individual votes are cast, but before the Court announces its decision. The cognitive-cybernetic theory developed in Chapter 2 is not dependent on the type of decision maker being examined (i.§., individual versus group). Since the theory applies to both types of judicial decision makers, the model derived from the theory will be used to examine the votes of the individual justices and the decisions of the Court. The dual application of the model provides an opportunity to observe the differences between a reductionist and a holistic view of judicial decision making. The cognitive—cybernetic theory of judicial decision making requires a stable decisional environment. Thus, a single issue area is chosen for examination: search and sei-, zure. Chapter 3 provides the details regarding the data set, Operationalization of variables, and the methodology used. Chapter 4 presents the results of the examination. Chapter 5 Presents a summary and conclusion. IIIIIIIIIIIIIIIIIIlIIII:—________________—_77_7_777777_———_—__—__—___——_____'777fl777747 Chapter 2: The Cognitive-Cybernetic Theory of Judicial Decision Making This chapter begins by presenting the comprehensive rational actor model. The next section examines a number of the limitations which make the comprehensive rational actor model unworkable as a model of human decision making. The third section presents a cognitive-cybernetic theory of human decision making which incorporates the methods of human learning and adaptation of cognitive theory in a cybernetic structure applicable to judicial decision making. In the final section, several hypotheses are drawn from the cogniti- ve—cybernetic theory and the specific issue area to be examined. A. Rationality in Decision Making When we speak of a "rational" actor (be it an individual, a group, or an institution), it is generally assumed the actor is engaged in goal-directed behavior (Laver 1981:21). Rational actors choose those actions which maximize the prob- ability of obtaining the desired goals. The assumption of rationality, with respect to the study of social choice theory, allows observers to examine actions with the assump- tion that the actor chose a course of action because it was the most preferred, in terms of obtaining the desired goal (Frohlich and Oppenheimer 1978:1—13). If rationality cannot be assumed, the study of social choice would be pointless, or at least reduced to the statistics of a distribution of 8 random actions. When behavior is to be considered rational it meets cer- tain conditions. First, the preferences of an actor must be connected and transitive (Frohlich and Oppenheimer 1978:7-9). Connectedness requires that any two outcomes can be ordered (A>B, Av to 78. At 159 last decision was in favor of reasonableness by as much as .46 (if the other variables multiplied by their estimated coefficients sum to zero). The standard error for LEDCITE is also very small (.005) which produces the second largest (in absolute value) t-statistic of this version of the model (3.64).14 With two exceptions, the estimated coefficients of the other variables remain essentially the same. The first exception is the increase (in absolute value) of the esti— mated coefficient of PROPERTY to -1.21, making it the largest for the location variables, though its standard error is still larger than that of HOME. The second exception is the sharp increase (in absolute value) in the estimated coeffi- cient of SEVERE to -1.51, a 63.3% increase. The standard error of SEVERE remains the same which pushes the size of the t-statistic to a minimally acceptable level of significance. Unfortunately, the estimated coeffi- cient for TIME drops just enough to lower its t-statistic below a minimally acceptable level Of significance (p < .07). LEDCITE also has an impact on the classification rate of the model. Seven additional decisions are correctly classi- fied, and all seven are decisions where the Court found the search or seizure to be unreasonable. The percent correctly classified rises to 58.3% for this category. Though still much lower than the other category, it is a substantial improvement. The overall percent correctly classified rises to 78.43% with a corresponding increase in ROE to 38.88%. At first glance there does not seem to be any theoretical justi: the c. corre: occurI decis: 1979:: attitI in the opinic cognit makes Opinic 0f decisi ronmen requir of opi Proced sions . enviro LE1 time p. not be: S°nne1 Table : these I Tal Spaeth. 160 justification for the impact of LEDCITE since it is merely the citation of each decision. However, these citations also correspond to a time element, with the earliest decision occurring in 1961 and the latest in 1987.15 Most theories of decision making (e.g., Rohde and Spaeth 1976:75—76, Spaeth 1979:119—120), including this one, make the assumption that attitudes are consistent over time. This is especially true in the area of judicial decision making where rulings and Opinions must be consistent with previous decisions. The cognitive-cybernetic theory of judicial decision making also makes this assumption as to the consistency of attitudes and opinions, so another explanation must exist. Of primary importance in modeling cognitive-cybernetic decision making is the existence of a stable decisional envi- ronment. Many aspects of judicial decision making meet this requirement of stability, both from an internal (consistency of opinions and attitudes) and an external (judicial rules of procedure) standpoint. In one respect, however, the deci- sions being estimated are not made in a stable decisional environment. LEDCITE is measuring a change which occurs during the time period covered by the data set (1961-1987), but which is not being measured by the independent variables. Eight per- sonnel changes occurred on the Court during this period. Table 7 presents evidence as to how LEDCITE may be measuring these personnel changes. Table 7 is an adaptation of Tables 2 and 3 in Harold Spaeth's supreme Court Policy Making (1979:133-137) . Spaeth 1958-2 Dougl: WarreI Brenna Black StewaI Clark Whitta FrankI Harlar M Dougl Warren Fortas Brenna Marsha Black Stewar White Harlan Value systems 161 Table 7 of U.S. Supreme Court Justices since 1958. 0 1 2 1958-1961 F E ND 1962-1965 F E ND 1965-1967 F E ND Douglas + + + Douglas + + + Douglas + + + Warren + + + Warren + + + Warren + + + Brennan + + + Goldberg + + + Fortas + + + Black + — + Brennan + + + Brennan + + + Stewart o o O Black + - + Black + - + Clark — — + Stewart 0 O 0 Stewart 0 o O Whittaker - - - White 0 o 0 White 0 o o Frankfurter - — - Clark - - + Clark - - + Harlan - — — Harlan — — Harlan — — — 3 4 5 1967—1969 F E ND 1970—1971 F E ND 1972-1975 F E ND Douglas + + + Douglas + + + Douglas + + + Warren + + + Brennan + + + Brennan + + + Fortas + + + Marshall + + + Marshall + + + Brennan + + + Black(l970) + - + Stewart o o 0 Marshall + + + Stewart 0 o 0 White 0 o 0 Black + - + White o o o Blackmun - - — Stewart 0 o o Blackmun - ~ - Powell - - - White 0 o o Burger - - - Burger - - - Harlan — - - Harlan(1970)- - - Rehnquist - - - 6 7 8 1975-1981 F E ND 1981-1986 F E ND 1986-1987 F E ND Brennan + + + Brennan + + + Brennan + + + Marshall + + + Marshall + + + Marshall + + + Stevens 0 o 0 Stevens 0 o 0 Stevens 0 o o Stewart 0 o 0 White o o 0 White o o o White 0 O O Blackmun - - — Blackmun - - - Blackmun - - - Powell - - — Powell - - — Powell - — — O'Connor - - - O'Connor - - - Burger - — - Burger - - — Scalia — — - Rehnquist - - - Rehnquist - - - Rehnquist — - — Values: F = freedom, E = equality, ND = new dealism + = support of value, - = nonsupport of value, 0 = neutrality to value Adapted from Harold Spaeth, Supreme Court Policy Making; W.H. Freeman and Company 1979, pages 133—135, Tables 2 and 3. const: issue: that ' Deali: fied I to (0 nonsuI justiI from 1 nine 1 t0 thI perio< of pa] based Freedc this I II SPlit the ne Change FrankI White iSSUes Chang. and it Suppox Ir Porter 162 constructed scale scores for the justices on narrowly defined issues. The scales associated with each other sufficiently that three main values emerged: Freedom, Equality, and New Dealism. Based on his scale scores, each justice was identi- fied as either supporting (+), nonsupporting (—), or neutral to (0) each value.16 The particular combination of support, nonsupport, or neutrality identifies the value system of a justice. Table 7 lists the value systems of the justices from 1958 through 1987.17 This information is divided into nine periods, identified by the period number, corresponding to the personnel on the Court at the beginning of the time period covered by the data set and at each subsequent change of personnel. The ordering of the justices in each list is based on the size of their average scale scores for the value Freedom, because the issue area search and seizure is part of this value. In the initial time period (period = 0) there is an even split between supporters and nonsupporters of Freedom with the neutral Stewart the swing vote (4-1—4). With the first changes in the Court's personnel, nonsupporters Whittaker and Frankfurter are replaced by supporter Goldberg and neutral White giving the supporters an outright majority on Freedom issues (5-2-2). This majority continues after the second change where supporter Fortas replaces supporter Goldberg, and increases to 6-2-1 when nonsupporter Clark is replaced by supporter Marshall. In the next period (4) this majority is lost when sup- porters Warren and Fortas are replaced by nonsupporters Burgel porteI nonsuI Rehan replac Tl depart O'Conr period Burger E) data 5 SuppOI trol" (Peric sonne] Freedc Freedc 0f the Nonsu‘ Strike efficj for LE Change C°I1rt. Ta Change 163 Burger and Blackmun (4—2—3). Further erosion of the sup- porter's strength occurs in period 5 when supporter Black and nonsupporter Harlan are replaced by nonsupporters Powell and Rehnquist (3—2-4), and in period 6 when supporter Douglas is replaced by neutral Stevens (2-3—4). The nonsupporters gain a majority in period 7 with the departure of neutral Stewart and the arrival of nonsupporter O'Connor (2-2-5). This majority is maintained in the final period (8) when nonsupporter Scalia replaces nonsupporter Burger. Except for the initial period (only two decisions in the data set come from this period), the general trend is from support of the value Freedom (periods 1, 2, and 3) to "con- trol" by the neutrals (periods 4, 5, and 6) to nonsupport (periods 7 and 8). As the changes occur in the Court's per— sonnel, the Court becomes less likely to support the value Freedom. In terms of this research, support of the value Freedom indicates a tendency to strike the balance in favor of the rights of the individual ceteris paribus. Nonsupport of the value Freedom indicates a tendency to strike the balance in favor of protection of the public and efficient law enforcement. The small estimated coefficient for LEDCITE and its large cumulative effect measure this change in support due to the changes in personnel on the Court. Table 8 provides additional evidence these personnel changes on the Court affect the stability of the decisional environment. In the periods where the supporters have a Un: DECISI Rea majori rights outnun Ods (n Percer to 33. the no rights in the Change that i Th Person. h°°d t] able. 164 Table 8 Comparison of DECISION by group controlling Court. CONTROLLING GROUP Supporter Neutral Nonsupporter Majority "Control" Majority Unreasonable 26 35 11 (56.5) (33.3) (20.8) DECISION Reasonable 20 70 42 (43 5) (66.7) (79.2) N 46 105 53 = 204 majority (1, 2, and 3) decisions in favor of individual rights (i.§., the search or seizure was deemed unreasonable) outnumber decisions against by 26 to 20 (56.5%). In the peri- ods where no outright majority exists (0, 4, 5, and 6) the percentage of decisions in favor of individual rights drops to 33.3% (35 of 105). In the final periods (7 and 8) where the nonsupporters have a majority, support for individual rights drops to only 20.8% (11 of 53). The fact that changes in the Court's personnel resulted in a somewhat orderly change in the support of the value Freedom is fortuitous in that it allowed LEDCITE to measure the differences. The important point, however, is that changes in Court personnel change the decisional environment and the likeli- hood the Court will decide a search or seizure to be reason- able. This finding is inherent in the theory. The cogni‘ maker: seizu: easilj est tI justiI When I change addit: unknOI 0r er] decisi tend I ever, any or time n Tl Obserw 0n the mint mation ferent cognitive-cybernetic theory does not require all decision makers to make the same decision. In the area of search and seizure, the major factors considered by the justices are easily identified; however, numerous minor factors, of inter— est to only one or two justices, may influence an individual justice's decision, and in turn, the decision of the Court. When personnel changes occur these individual factors also change causing instability in the decisional environment. In addition, decision makers can simply make mistakes due to unknown consequences, limited information, false assumptions, or erroneous perceptions. In examining a large group of decision makers these minor fluctuations and aberrations will tend to even out (Lave and March 1975:135). The Court, how— ever, is a small group with a maximum of only nine members at any one time and the composition of the Court at any point in time may influence its decisions. The more stable the Court is during the period under Observation, the less influence personnel changes will have on the Court's decisions. Tables 9, 10, 11, and 12 make this point clear. These tables present the results of probit esti- mation using the reduced location choice set model for dif- ferent time periods. The period from 1975 to 1985 is one of the most stable in the Court's history, with only one change of personnel occur- ring: the resignation of Justice Stewart and the appointment of Justice O'Connor in 1981. From Table 7 we can see that this was an important change since it gave the nonsupporters of the value Freedom an outright majority on the Court. DECIS] ...... 166 Table 9 Probit estimation of reduced location choice set model for the period 1975 through 1987. Independent Estimated Standard t- Variable Coefficient Error Statistic CONSTANT 1.33 0.50 2.65 HOME -1.25 0.41 -3.08*** BUSINESS -0.94 0.51 -1.84* PROPERTY -1.22 0.44 -2.76*** CONTROL -0.57 0.33 -1.74* CUSTODY -0.92 0.41 -2.21* TIME 0.45 0.35 1.30 DRUG/GUN 0.63 0.35 1.82* SEVERE -1.27 0.70 -1.80* US 0.86 0.35 2.45** Auxiliary Statistics Log Likelihood At Convergence Initial Adjusted -49.62 "79.02 -66.71 N 114 CROSSTABLULATION OF DECISION BY CLASSIFICATION CLASSIFICATION I I I I COUNT I I I ROW I ROW PCT I 0 I 1 I TOTAL DECISION I I I I | 0 | 16 I 15 I 31 I I 51.6 I 48.4 I 27.2 I I I I I 1 I 6 I 77 I 83 I I 7.2 I 92.8 I 72.8 I I I I COLUMN 22 92 114 TOTAL 19.3 80.7 100.0 Percent Correctly Classified 81.58 Percent in Modal Category 72.81 ROE (%) 32.23 —2*LLR (chi-square w/10 d.f.) 34.18**** (using adjusted value) *Significant at .05 **Significant at .01 ***Significant at .005 ****Significant at .001 occur varia but ti (see ' ally . also 1 the s] 31 (5: This 1 data I the re are cc data 5 perCer be not 167 Chief Justice Burger's resignation in 1986, and the subsequent appointment of Justice Scalia did not have the same impact since both are considered nonsupporters of this value. Assuming that the change from Burger to Scalia was minimal in terms of the variables we are measuring (although the minor factors which influence these justices will undoubtedly vary), the period from 1975 to 1987 has only one significant personnel change. Table 9 presents the results of a probit estimation for this time period using the reduced location choice set model. Roughly 56% (114 of 204) of the decisions in the data set occur during this period. All but one of the independent variable reach minimally acceptable levels of significance, but the levels are somewhat lower than for the full data set (see Table 5). The magnitude of the coefficients is gener- ally larger for this estimation, but the standard errors are also larger. The larger standard errors are due in part to the smaller number of observations in this data set. For this period, the model correctly classifies 16 of the 31 (51.6%) searches or seizures considered unreasonable. This represents a slight increase over the 48.6% of the full data set. An increase is also made in the classification of the reasonable searches or seizures where 77 of 83 (92.8%) are correctly classified, as opposed to 89.4% for the full data set. This represents a combined increase of 6.6% in the percentage of decisions correctly classified. It should also be noted that the percentage in the modal category increases by 8. cate peric incre one m this decis indep on th 7), a Spond this other coeff no re CONST, level 168 by 8.1%. This increase in the modal category does not indi- cate that the model does not do as well for the 1975-1987 period. In this stable periOd, ROE is 32.23%, a 3.09% increase over the reduction of error for the full data set. The gains made are small, but encouraging, and there is one major personnel change during this period. To see if this change (from Stewart to O'Connor) has an impact on the decisions of the Court the variable PERIOD is added to the independent variables. PERIOD is coded as "6" if Stewart is on the Court (1975-1981) (corresponding to period 6 on Table 7), and "7" during O'Connor's tenure (1981-1987) (corre- sponding to periods 7 and 8 on Table 7). The results of this probit estimation are contained in Table 10. The large values for PERIOD, especially relative to the other independent variables, causes the size of the estimated coefficient for CONSTANT to become quite large. This is of no real importance. If PERIOD is recoded as a 0/1 variable, CONSTANT's estimated coefficient decreases to its former level. The estimated coefficient for PERIOD reaches a high level of significance (p < .001) which supports the position that the change from Stewart to O'Connor was one of substance as well as form. The large positive value of PERIOD's estimated coefficient indicates the increased likelihood that a search or seizure will be considered reasonable during periods 7 and 8. This is supported by Table 7, where we see the nonsup- porters (including O'Connor) have a majority in these peri- ods, and from the fact that during these two periods (7 and DECIS ..... 169 Table 10 Probit estimation of reduced location choice set model with variable PERIOD for the period 1975 through 1987. Independent Estimated Standard t— Variable Coefficient Error Statistic CONSTANT -8.35 2.66 -3.13 PERIOD 1.57 0.43 3.67**** HOME ~1.99 0.51 -3.93**** BUSINESS -1.18 0.58 -2.04** PROPERTY —1.80 0.55 -3.30**** CONTROL -0.68 0.36 -1.91* CUSTODY -1.50 0.52 -2.92*** TIME 0.86 0.42 2.05** DRUG/GUN 0.79 0.38 2.07** SEVERE -2.08 0.74 -2.81*** US 1.32 0.43 3.07**** Auxiliary Statistics Log Likelihood At Convergence Initial Adjusted -41.09 -79.02 -66.71 N 114 CROSSTABLULATION OF DECISION BY CLASSIFICATION CLASSIFICATION | I I I I COUNT I I I ROW I Row PCT I o I 1 I TOTAL DECISION I I l I I 0 I 21 I 10 | 31 I I 67.7 I 32.3 I 27.2 I I l I I 1 I 8 I 75 I 83 I I 9.6 I 90.4 | 72.8 I I I I COLUMN 29 85 114 TOTAL 25.4 74.6 100.0 Percent Correctly Classified 84.21 Percent in Modal Category 72.81 ROE (%) 41.93 -2*LLR (chi—square w/ll d.f.) 51.24**** (using adjusted value) *Significant at .05 **Significant at .025 ***Significant at .005 ****Significant at .001 DECIS ..... 170 Table 11 Probit estimation of reduced location choice set model for the period 1975 through 1981. Independent Estimated Standard t- Variable Coefficient Error Statistic CONSTANT 1.23 0.91 1.35 HOME -2.06 0.71 -2.90*** BUSINESS ~1.73 0.81 -2.14* PROPERTY -2.39 0.94 -2.55** CONTROL -0.18 0.52 -0.35 CUSTODY -2.07 0.81 -2.57*** TIME 1.49 0.70 2.14* DRUG/GUN 0.20 0.54 0.37 SEVERE 1.68 20.14 0.08 US 1.24 0.64 1.94* Auxiliary Statistics Log Likelihood At Convergence Initial Adjusted —22.51 -42.28 -38.59 N 61 CROSSTABLULATION OF DECISION BY CLASSIFICATION CLASSIFICATION I I I I COUNT I I I ROW I ROW PCT I 0 I 1 I TOTAL DECISION I I I I I 0 I 13 | 7 I 20 I I 65.0 I 35.0 I 32.8 I I I I I 1 I 3 I 38 I 41 I I 7.3 I 92.7 I 67.2 I I | I COLUMN 16 45 61 TOTAL 26.2 73.8 100.0 Percent Correctly Classified 83.61 Percent in Modal Category 67.21 ROE (%) 50.02 -2*LLR (chi—square w/10 d.f.) 32.16**** (using adjusted value) *Significant at .05 **Significant at .001 ***Significant at .005 ****significant at .001 8)7 able ity majo icanI hood incrI for I were made were decre error more only Signi Varia reach lEVQl This dhrin a gma the e due t 171 8) 79.2% of the searches and seizures were considered reason- able compared to 67.2% during period 6. The addition of the variable PERIOD increases the stabil- ity of the period being estimated by controlling for the major personnel change from Stewart to O'Connor. The signif- icance of the model is also increased raising the log likeli- hood ratio to 54.24 (p < .001). In addition, ROE is increased by almost 10 percentage points to 41.93%. Table 11 presents the results of the probit estimation for the period 1975 through 1981. During this period there were no changes in the personnel on the Court. The Court made 61 search and seizure decisions during this period; 41 were considered reasonable, 20 were considered unreasonable. The estimated coefficients for CONTROL and DRUG/GUN decrease (in absolute value) sharply, while the standard error for the estimated coefficient of SEVERE increases even more sharply. These increases cause these variables to be the only ones which do not reach minimally acceptable levels of significance. Although the estimated coefficients for the variables HOME, BUSINESS, PROPERTY CUSTODY, TIME, and US do reach minimally acceptable levels of significance, these levels are generally lower than in previous estimations. This may be due in part to the preferences of the justices during this period, but it is also a problem associated with a small data set. In small data sets the standard errors of the estimated coefficients have a tendency to become large due to the less frequent presence of the variables which in turn able SEVE the 0f t1 (65.I love: the I job I able The c sligl The ] high] categ Allow deSpi for t Court Small for t] faile< in Ta! this I I It 172 turn increases the collinearity among the independent vari— ables. A prime example of this is the standard error for SEVERE. SEVERE only occurs once in this period which forces the standard error to be exceptionally large. The classification rates for this period are quite good. Of the 20 searches or seizures considered unreasonable 13 (65.0%) are correctly classified. This is just slightly lower than for the 1975-1987 period, but much better than for the entire data set. The model continues to do a very good job classifying the searches and seizures considered reason- able, correctly classifying 38 of the 41 decisions (92.7%). The overall success rate is 83.6% (51 of 61) which is slightly lower than the previous estimation in Table 10. The log likelihood ratio is also lower, (32.16) though still highly significant (p < .001). The percent in the modal category falls to 67.2% for this period. This decrease allows ROE to increase by over eight percent to 50.02%. despite the slight decrease in the overall success rate. Table 12 presents the results of the probit estimation for the final period: 1981 through 1987. In this period the Court made 53 search and seizure decisions. Once again, the small number of observations has inflated the standard errors for the estimated coefficients. The three variables which failed to reach minimally acceptable levels of significance in Table 11 (CONTROL, DRUG/GUN, and SEVERE) all do so for this period. Two other variables (BUSINESS and TIME), how- ever, do not. It is interesting that not only does the estimated CON! HOM] BUS: PRO] CON? CUSI TIMI DRUC SEVI US Auxil Log I DECIS ~~~~~ 173 Table 12 Probit estimation of reduced location choice set model for the period 1981 through 1987. Independent Estimated Standard t- Variable Coefficient Error Statistic CONSTANT 4.89 2.28 2.15 HOME -4.36 2.29 -1.91* BUSINESS 0.06 15.05 0.00 PROPERTY -3.41 1.75 -1.96** CONTROL -2.97 1.73 -1.72* CUSTODY -1.73 1.03 -1.68* TIME 0.39 0.69 0.53 DRUG/GUN 2.66 1.35 1.96** SEVERE -4.51 2.08 -2.18** US 1.99 1.10 1.81* Auxiliary Statistics Log Likelihood At Convergence Initial Adjusted -10.86 -36.74 ~27.07 N 53 CROSSTABLULATION OF DECISION BY CLASSIFICATION CLASSIFICATION I l | I COUNT | I I ROW I ROW PCT I 0 I 1 I TOTAL DECISION I I I I I 0 | 8 I 3 I 11 I I 72.7 I 27.3 I 20.8 I I I I | 1 I 2 I 40 I 42 I I 4.8 I 95.2 I 79.2 I I I I COLUMN 10 43 53 TOTAL 18.9 81.1 100.0 Percent Correctly Classified 90.57 Percent in Modal Category 79.25 ROE (%) 54.55 32.42*** -2*LLR (chi-square w/lO d.f.) (using adjusted value) *Significant at .05 **Significant at .025 ***Significant at .001 coef valu smal cons busi duri Cour busi two' trat over coef: to a will seveI that issue legaI in u resu] the p wiret vehic Fortu affec 174 coefficient for BUSINESS drastically decrease (in absolute value), it is now positive, indicating an increase (albeit a small one) in the likelihood that a search or seizure will be considered reasonable when the location is a business. A business is the location of six of the searches or seizures during this period; all six were deemed reasonable by the Court. Five of the six were not typical of searches of a business. Two involved illegal aliens working in a business, two were searches pursuant to a statute authorizing adminis- trative searches, and the fifth involved a plane which flew over a factory and took pictures from the air. These unusual cases explain the change in the estimated coefficient for BUSINESS for this period, but they also point to a potential problem. Quite often a great deal of interest will occur in a particular issue area, and the Court may make several decisions on that issue. Eventually, the cases on that issue appear less often and are replaced by some other issue which has sparked the interest of the Court and the legal community (see Schattschneider 1975:65). These changes in the cases decided by the Court could very well affect the results of multivariate modeling of judicial decision making. One example of this occurs in the full data set. During the Warren Court there are numerous decisions involving wiretaps and listening devices, but very few involving vehicles. During the Burger Court this trend is reversed. Fortunately, this difference between the two Courts did not affect the results of the models used here, but it certainly coulI corrI able and . corrI of 5: reac] modal ~~~~~ 175 could be a problem in other circumstances. Returning to Table 12 we can see an improvement in the correct classification rate for both reasonable and unreason- able searches and seizures. Eight Of 11 (72.7%) unreasonable and 40 of 42 (95.2%) reasonable searches or seizures are correctly classified. The overall success rate of 90.6% (48 of 53) is the best of any of the time periods, and ROE also reaches its highest value (54.55) despite an increase in the modal category to 79.25%. Table 13 Comparison of classification rates for selected time periods based on probit estimation. ‘ Correct Overall % in Modal I Period N Classification Rates Success Category Unreasonable Reasonable Rate [ROE] 1961 - 1987 204 35/72 118/132 153/204 64.71 (Table 5) (48.6) (89.4) (75.0) [29.16] 1975 - 1987 114 16/31 77/83 93/114 72.81 (Table 9) (51.6) (92.8) (81.6) [32.23] 1975 - 1987 114 21/31 75/83 96/114 72.81 W/ PERIOD (67.7) (90.4) (84.2) [41.93] (Table 10) 1975 - 1980 61 13/20 38/41 51/61 67.21 (Table 11) (65.0) (92.7) (83.6) [50.02] 1981 - 1987 53 8/11 40/42 48/53 79.25 (Table 12) (72.7) (95.2) (90.6) [54.55] Table 13 summarizes the correct classification rates, modal category percentages, and ROE values of the models and periods estimated in Tables 5, and 9 through 12. In general, ther exam Cour more majo stab the esti repr this tant that resu the ing I set) exam: the 1 the I do me for c eXter Peric Hal c fact OCCuI 176 there is a trend toward better results as the periods examined become more stable, in terms of changes in the Court's personnel, and as the justices on the Court become more uniform in their outlook (e.g., the nonsupporter majority during the 1981-1987 period). With regard to a stable Court, the cumulative correct classification rates of the two halves of the 1975-1987 period exceed those of the estimation of the whole period even with the variable PERIOD representing the change from Stewart to O'Connor. Although this finding supports the position that stability is impor- tant, the number of observations for these periods requires that one be careful in drawing conclusions from these results. One conclusion which can be drawn from these results is the importance of a stable decisional environment. Increas- ing stability in the internal (reducing the location choice set) and external (choosing more stable time periods for examination) decisional environments increased the ability of the model to estimate and correctly classify the decisions of the Court. Second, the decisions of the individual justices do make a difference. The Court is simply too small a group for differences between the justices to even out to the same extent as a larger group. When the Court experiences a period of stability in its personnel the effects of individ- ual differences can be minimized. Unfortunately, it is a fact of life that changes in the personnel on the Court occur quite frequent1Y~ deep tion in t part cati dent of c beli aPPr know actu. sit . unreI the ( inst] Supre is tl Withc bl‘dnc respe Carri devic which Chapter 5. Summary and Conclusions United States Supreme Court decision making has its roots deep in American and English legal traditions. These tradi- tions emphasize the importance of stability and consistency in the decisions handed down by courts. Stability is ensured partially through the use of previous decisions as a justifi- cation for present decisions. This reliance on past prece— dent, called stare decisis, lends continuity to the decisions of courts. This reliance on past decisions leads some to believe that the role of courts is simply to identify the appropriate rule of law and apply it accordingly. This is known as the declaratory theory of judicial decision making. It does not require a very detailed examination of the actual decisions to realize that courts, and the judges who sit on them, do much more than merely discover heretofore unrevealed legal maxims. Certainly lower courts must follow the dictates of higher courts, but when no higher court instructions exist or when speaking of the United States Supreme Court, tradition is not binding. The judicial branch is the least subject to the vagaries of public opinion, but without the enforcement machinery held by the other two branches of government, courts must rely a great deal on respect for the judicial institution for their will to be carried out. Nevertheless, several traditionally acceptable devices exist for courts to distinguish or ignore precedents which are contrary to a court's desired position. If the cult of the robe and the declaratory theory of 177 jud. cou: can is I but the make alit unde Ther and with tion: on t1 maxim decis ing t than of pr makin. goals may or inc to thi 178 judicial decision making are mere straw men, on what basis do courts make decisions? We would hope that courts and judges can be considered rational decision makers, but rationality is one of those slippery concepts which everyone talks about but few define precisely. Rationality involves normative considerations regarding the appropriateness of the actions chosen by a decision maker. Rational choices imply successful actions, irration- ality implies bad choices, but choices regarding what? The underlying assumption is that behavior is goal directed. Therefore, choices which help to achieve goals are rational, and choices which do not are irrational. Various scholars have attempted to define rationality with a great degree of precision. Generally, these defini- tions require some form of comprehensiveness or maximization on the part of the decision maker. Comprehensiveness or maximization is due in part to the method of representing decisions in terms of costs and benefits. Certainly, accord- ing to the theory, if decision X produces more net benefits than decision Y, X should be chosen over Y. Two major types of problems exist with this formulation of rational decision making. First, decision makers may not always know what their goals are. Just as candidates for political office may be fuzzy on their policy positions, decision makers may be fuzzy or indeterminate about the specifics of their goals. Related to this is the problem of quantifying costs and benefits. To determine when benefits outweigh costs, both benefits and cost bene huma the requ bene ingl to t rati CODS are rati that unde natu: alter requ fits Inake: for ' make: Vari. eXpe. b0un< Conc« 179 costs must be quantifiable. Unfortunately, some costs and benefits are very difficult to quantify (e.g., how much is a human life worth?). The second major type of problem involves the scope of the search for the "best" choice or solution. If rationality requires the decision maker select the choice which maximizes benefits, even a relatively simple choice can become exceed— ingly costly in terms of time and expense. Thus, conforming to the comprehensiveness or maximization requirement of rationality often produces results which most people would consider irrational. Bounded rationality and decision making under uncertainty are attempts to deal with these two major problems. Bounded rationality recognizes that there are limits on the search that can be undertaken by a decision maker. Decision making under uncertainty attempts to deal with the probabalistic nature of future states of the world. Neither of these two alternatives to comprehensive rationality rejects the requirement that the decision maker attempt to maximize bene— fits. Within the bounds of bounded rationality the decision maker is still expected to carry out a comprehensive search for the optimum solution. Despite uncertainty, the decision maker is expected to calculate the expected utilities of various courses of action and select the one with the highest expected utility. A third major problem, which is not solved or avoided by bounded rationality or decision making under uncertainty, concerns human cognitive limitations. According to human lea: proc The chur rati Give maki tive smal cour does only 128 limi thre occu comb adOp leVe Cybe nor arms make make enti: 180 learning and memory studies, short term memory is capable of processing only a limited amount of information at one time. The amount seems to lie somewhere in the range of seven chunks of information, plus or minus two. Herbert Simon has developed his own version of bounded rationality which is commonly referred to as satisficing. Given the cognitive limitations inherent in human decision making extensive maximization is not possible. As an adap- tive strategy, the decision maker will focus attention on a small number of factors important to the decision and adopt a course of action based on these factors. The decision maker does not maximize with respect to these factors. Focusing on only seven factors with only two values each still produces 128 possible combinations; a number well beyond the cognitive limits. Instead, the decision maker has certain limits or thresholds which must be passed before changes in behavior occur. The relevant factors are monitored and when their combined values reach a utility threshold the decision maker adopts a new course of action consistent with the different level of utility. This type of decision making fits the cybernetic theory of decision making. Servomechanisms such as the Watt governor are often used to illustrate this type of behavior. Because the Watt gover— nor only has to monitor one factor (the angle of the rotating arms) and because it does so continuously, the adjustments it makes in its "behavior" are also continuous. Human decision makers, whether individuals or groups, do not focus their entire attention continuously on the problem they are gene var: tail priz tral is I vehi olds tib] same fact behe tior impc thei cate miss huma tati make high nor rule Whic Sist Thes 181 generally concerned with solving. This is why the utility of various combinations of the relevant factors must cross cer- tain thresholds to elicit changes in behavior. A more appro— priate model for this type of behavior is the automatic transmission in a car. The shift from lower to higher gears is made automatically when the relevant factor, speed of the vehicle, crosses specific thresholds. Between these thresh— olds exists a range of values of the factor which are compa- tible with that gear. The human decision maker maintains the same behavior for a range of utility values for the relevant factors. Perhaps more utility could be gained by adapting behavior to reflect smaller ranges, but cognitive limita- tions, as well as the time and effort involved, make it impossible to do so. Human decision makers are not machines. They must make their choices in decisional environments much more compli— cated that those of the Watt governor or an automatic trans- mission. Cybernetic theorists often fail to account for human cognitive limitations or the fact that cognitive limi— tations are internally imposed. It is the human decision maker who must find adaptive strategies for operating in highly complex decisional environments. For the Watt gover— nor, the automatic transmission, or even a supercomputer, the rules are externally determined. Steinbruner identifies five elements of cognitive theory which supplement cybernetic theory: inferential memory, con- sistency, reality, simplicity, and stability (1974:95-103). These five elements recognize certain features of human memc the cont tive cybe thec plii of t limi rule desi mode be e wit} the desj (int the attz nece set! tie: the Cip] 182 memory and information processing which are concerned with the structure and processing of information rather than its content. The result of incorporating these five elements of cogni- tive theory into cybernetic theory yields a cognitive— cybernetic theory of decision making. The essence of this theory is found in four basic principles. The first is sim— plification. Human decision makers, because the complexity of their decisional environments exceeds their cognitive limitations, attempt to simplify the decision making process. Second, the decision maker relies on stable decision rules. The stability is internal due to the decision makers desire for consistency in his actions. For purposes of modeling the decision making process, the stability must also be external. A decision maker who is continually presented with vastly different decisional environments will not have the opportunity to develop stable decision rules. The third principle is derived from the first two. A desire for simplification and a stable decisional environment (internal and external) allow the decision maker to focus on the "important" attributes of the situation. These important attributes or factors are the minimal pieces of information necessary for the decision maker to make a choice. Fourth, the decision maker must have a limited choice set. A large choice set requires large amounts of informa- tion processing; amounts beyond the cognitive limitations of the decision maker. A limited choice set fits with the prin— ciples of simplification and a focus on the important factors of thel tra cen‘ One: of 1 impe casc The] 0th: the whi< tiox the: to 1 as a impc traj must they Cons Vidu dete deci of a situation. Judicial decision making fits the cognitive-cybernetic theory of decision making very well. Lawyers and judges are trained in fact pattern analysis. They are trained to con— centrate on the important factors in a given issue area. Once trained the stability of their learning and the desire of decision makers to find and use shortcuts reinforces the importance of these learned factors. When lawyers argue cases they do so based on the important facts of the case. They attempt to emphasize the facts which are consistent with other cases which were decided in their favor. Lawyers for the other side also try to emphasize the facts of the case which are consistent with other cases which favor their posi- tion. Both sides deemphasize the facts which detract from their preferred outcome. In an appellate case, it is then up to the court to determine which facts are indeed important, as a matter of law, in the resolution of the conflict. Judges are also trained in the use of the same sets of important factors, usually because they were originally trained as lawyers. But even nonlawyers who become judges must quickly adapt to the decisional environment in which they find themselves. This means picking out the factors considered important by the legal community. How any indi- vidual decision maker makes use of these important factors is determined by that decision maker's personal preferences. How is this determination made? Many argue that judicial decisions are based on the ideological orientation and per- sonal policy preferences of the judges (e.g., Schubert 1965, 197 att ior Gut' jUdt but Blac Arm: pen: ever cast ment sati pres side tors Firs that rule infll mati< recos them to Vc quick 184 1974; Rohde and Spaeth 1976). This position is derived from attitude theory as developed and applied by judicial behav— ioralists such as Schubert (1958, 1959), Eysenck (1954), Guttman (1950, 1954), and Rokeach (1968a, 1968b). This approach helps to explain the general attitude of judges, and the justices of the Supreme Court in particular, but more is needed. With very few exceptions (e.g., Justices Black and Douglas in First Amendment cases (Woodward and Armstrong 1979:193, gt §§g-) and Justice Brennan in death penalty cases (PBS's "This Honorable Court," part II, 1987)) even the most ideologically oriented justice occasionally casts votes in opposition to his or her preferred position. Von Neumann and Morgenstern point out that all measure— ment must ultimately based on sensation (1980:16). The sen- sations measured by judicial decision makers are the factors presented to them during the course of the litigation. Con- sidering the Supreme Court in particular, the types of fac- tors presented to the justices are limited in two main ways. First, the rules of judicial behavior and etiquette require that many potential factors not be considered. These include rules of standing and evidence and a general avoidance of influences outside the legal process (such as lobbyists). Even with these limitations, inordinate amounts of infor- mation reach the justices. A number of the justices have recognized that to deal with the amount of material before them, they had to develop ways of quickly determining how to vote on a case or petition. They developed a system to quickly determine whether certain relevant factors are pre sea and for fac ind use cas tio cia rat goa dif use‘ bri1 the 185 present and make their decision accordingly. In the area of search and seizure, the requirements of the Fourth Amendment and the general similarity of the fact patterns of the cases, force the Court to focus its attention on a small number of factors relevant to the conflict between the rights of the individual and protection of the public. The monitoring and use of specific factors for the purpose of determining how to cast a vote consistent with one's preferred ideological posi— tion is the basis of the cognitive-cybernetic theory of judi- cial decision making. Cognitive-cybernetic theory assumes the justices are rational decision makers in that they are engaged in goal-directed behavior. Where cognitive-cybernetic theory differs from other theories of rationality is in the method used by the decision maker in determining which decision will bring the justice the greatest utility (i.e., is closest to the justice's preferred ideological position). Before discussing the specific factors relevant to the area of search and seizure, we must address the question of who decides; the Court or the individual justices. Clearly, the Court is composed of the justices who sit on it, but is it more than that? Section 1 of Article III of the United States Constitution begins by stating, "The judicial Power of the United States, shall be vested in one supreme Court . . ..“ The Constitution does not give this power to the indi- vidual justices, it does not even specify the number of jus— tices to be on the Court other than a Chief Justice. Deci— sions are made by the Court, and the size of the winning coa] mini majc app] ual whic a pc app] ing way. his may of t res; stra tors vari infl nUmb mode resu erro resu deci rand. --——-4] .v 186 coalition is immaterial from a legal standpoint. Even in a minimum winning coalition, the winning side is composed of a majority of the justices who voted. Quite often, theories of individual behavior are not applied to predict or classify individual behavior. Individ- ual decision makers are just that; individuals. Factors which prove to have a certain level of explanatory power over a population, may take on greater or lesser importance when applied to individuals. No theory of rational decision mak— ing requires all decision makers to treat factors in the same way. It is generally assumed that each decision maker has his or her own personal utility scale on which the factors may score differently depending on the personal preferences of the individual decision maker. This is also true with respect to judicial decision making. Because of the con- straints placed on the Court and the justices the major fac— tors considered are often readily identifiable. However, various minor factors peculiar to specific justices will also influence their decision. If these minor factors are consistent across a large number of the justices their continued exclusion from the model will result in systematic error which invalidates the results. When these minor factors are more localized, random error is the result. Random error also invalidates the results with respect to the individual, but when a group of decision makers is examined the assumption is that these random errors will even out. Nevertheless, the cognitive-cybernetic theory does not dif whi tic aPP the sys mak que. or : gr01 faC' by ' and war: fiw becz made tor: 681 War] 0the denc natt othe 187 differentiate between the two judicial decision makers with which we are concerned: the Court and the individual jus— tices. This allows for a comparison of a reductionist approach to judicial decision making which concentrates on the individual justices and a method of aggregation and a systemic approach which emphasizes the Court as the decision making entity. Returning to the area of search and seizure, the basic question facing the Court is the reasonableness of a search or seizure. The major factors considered fall into two groups: legal factors and situational and case characteristic factors. The consideration of the legal factors is mandated by the Fourth Amendment. The requirement of probable cause and a warrant, the existence of probable cause and a valid warrant, and the violation of the exclusionary rule are the five legal factors. These five factors are considered legal because their presence or absence is a legal determination made by the Court. It quickly becomes clear that these fac- tors are merely an alternate measure for the Court‘s decision as to reasonableness. The Fourth Amendment makes it clear that a search cannot be considered reasonable unless probable cause and a valid warrant are present when they are also required. This and other relationships among these legal factors provide evi- dence of severe collinearity, as well as their tautological nature. The legal nature of these five factors precludes any other way of measuring them. Thus, none of these variables - .. - .-- W _~_,___.___.__~_— can pm the par var af f Seg cal ing van var of man is maj str of are to rea tec tiv of men how reg 188 can be used in a model to estimate judicial decisions. In a previous examination of the search and seizure decisions of the Court, Segal (1984) used a number of legal variables as part of his model (warrant, probable cause, and three arrest variables). Certainly, the inclusion of these variables affected the results obtained. However, the purpose of Segal's study was to test the applicability of the statisti- cal technique probit to the study of judicial decision mak- ing, not the formation of a theory and set of factors rele- vant to the area of search and seizure. The situational and case characteristic factors and the variables drawn from them examine what underlies the question of reasonableness. In the area of search and seizure, as in many other areas of constitutional interpretation, the Court is faced with striking a balance between minority rights and majority rule. In search and seizure the balance must be struck between the rights of the suspect and the protection of the public. Certainly the rights of any criminal suspect are impinged to some extent, but it is the job of the Court to make sure they are not violated beyond constitutional reasonableness. On the other hand, the Court will not pro— tect the individual's rights to an extent which makes effec- tive law enforcement impossible. The relevant factors include the permanence and privacy of the location searched, the amount of control the govern- ment Officials exercised over the location or item seized, how much time was available for fulfilling the technical requirements of obtaining a valid warrant, the nature of the ite: whe‘ fro: dir of rea who var eXp var of Thi whi coe cog tic OPP the to Sio nal inf tio and ide 189 item seized, the severity of the search or seizure, and whether the case came from a state or federal court. The estimated coefficients of the eleven variables drawn from these factors all prove to be in the hypothesized direction for both the decisions of the Court and the votes of the individual justices. The levels of significance reached for the estimated coefficients and the model as a whole support the rejection of the null hypothesis. Of the variables which were expected to have a high degree of explanatory power, only the estimated coefficient for the variable SEVERE failed to reach a minimally acceptable level of significance in the estimation of the Court's decisions. This was primarily due to the low number of instances in which SEVERE is coded as being present. The levels of significance attained by the estimated coefficient and the model as a whole support the use of cognitive—cybernetic theory to explain the votes of the jus- tices as well as the decisions of the Court. In the examination of the decisions of the Court an opportunity exists to see how an increase in the stability of the decisional environment affects the ability of the model to estimate the those decisions. The stability of the deci- sional environment is increased both internally and exter- nally. Internal stability is increased by reducing the information considered by the justices. Although the loca- tion choice set is drawn from a combination of the privacy and permanence factors, six possible search locations are identified. Reducing the location choice set reduces the amc dec inc pe: fie of im; jus of set cor of fax pe1 to axe dii moc‘ 0f inc PEI Per Vot Of Me 190 amount of information which must be processed to reach a decision. External stability is increased by choosing increasingly stable time periods, in terms of the Court's personnel, for estimation. The reduced location choice set model correctly classi- fies 75.0% of the Court's decisions for a reduction of error of 29.2%. These results are good, but there is room for improvement. An examination of the value systems of the justices reveals a steady shift in the ideological position of the justices on the Court. In the early portions of the period covered by the data set, a majority of the sitting justices are more likely to consider searches or seizures unreasonable. After a number of personnel changes, no clear majority of the justices favored either position. During the latter stages of the period, justices who preferred to find searches or seizures to be reasonable held a majority on the Court. Although examination of the Court's decisions eliminates many of the differences of the individual justices, it appears that the model is not accounting for all the differences. Examination of selected portions of the entire time period reveals an increased accuracy of the model's classifications as the period examined becomes more stable in terms of the Court's personnel and ideological orientation. The results of the aggregation of the estimation of the votes of the individual justices and those of the estimation of the Court's decisions were quite similar. At first, one might think this indicates a lack of any group effect being imp cas dec tic jue dow not vot the dec boo and Bre doe enc int of : mat. Sup; ali: POW< Sig] idel for app] 191 imposed on the decisions of the Court, but this is not the case. A group effect may very well exist, but comparing the decisions of the Court with the votes of the individual jus— tices cannot identify it. Group effects, if any exist, occur prior to the time the justices officially cast their votes in the opinions handed down. Once the votes are cast, the Court's decisions are nothing more than a simple aggregation of the justices' votes. To test for possible group effects, we must examine the behavior of the justices during earlier stage of the decision making process. We may want to examine the docket books of the justices to determine how they initially voted and if their votes changed (e.g., Brenner and Spaeth 1988; Brenner, Hagle, and Spaeth 1988). But even this approach does not identify all group effects. Arguably, even confer— ence votes and votes on certiorari are influenced by previous interactions of the justices. To conclude, support for the cognitive-cybernetic theory of judicial decision making is found in the results of esti- mating the search and seizure decisions of the United States Supreme Court. Several factors are identified and operation- alized which prove to have a high degree of explanatory power. The model as a whole also reaches a high level of significance. Room for improvement still exists, either by identifying additional factors or improving the measures used for the factors already identified. The value of this study lies in the development and application of a theoretical framework to the decision making 192 of the Court. The theory takes into consideration many of the problems and pitfalls of other theories of decision mak- ing and provides an alternative which fits the general and cognitive limitations which face human decision makers. From this theory, a number of hypotheses which identify relevant factors considered by the Court are derived and tested with respect to a specific area of the law. The support found for these hypotheses will be of value in the legal community because of the increased understanding of the Court's deci— sions and decision making process, but also as a means of identifying the factors present in a case which may make it an eventual winner or loser in the eyes of the Court. Future research should concentrate on the identification of additional factors or alternative measures of the identi- fied factors which do a better job of explaining the deci- sions in the Court's nonmodal category. A closer examination of the incorrectly classified decisions and further analysis of the decisions of the individual justices may help in this regard. This theory and model should also be tested on courts other than the United States Supreme Court. The additional limitations imposed on other courts by the presence of a higher court is at least one additional consideration in estimating their decisions. Several of the factors identified in this study are rele- vant to the constitutional conflicts which occur in other areas of the law (e.g., First Amendment free speech and Fifth Amendment involuntary confessions). Identification of proper 193 measures for these factors as applied to other areas of the law may allow us to peek into the black box of the judicial decision maker and gain a greater understanding of how and why courts and judges make their decisions. _fi—i "' .. ,w1'm'“"‘-’mt‘=—U' ""“‘“‘~""'w' ’w“""-"" "— . _ - T. . . . . NOTES Chapter 1 1. The importance of a majority decision by the Court versus the decisions of individual justices is evident in how tie votes are treated. When there is a tie vote (4-4 or 3-3) no opinion is written and no issues are resolved. The decision of the lower court stands until the ques- tions presented in the case once again come before the Court. Chapter 2 Chapterg; 1. This basic assumption is evidenced in a number of works such as Simon's Models of Thought and Harris' America's Democracy, and is illustrated in Allison's The Essence of Decision. It is unlikely that any potential alternative set could be exhaustive. Alternatives which appear random and unrelated to the decision could be chosen. For example, Kennedy could have decided to change the color of navy uniforms due to the mistaken belief that blue attracts missiles. Under Frohlich and Oppenheimer's definition this would have been a rational, though erroneous deci- sion. They might attempt to rule out such an action as a violation of the nonperversity rule, but a preference for having the missiles removed does not necessarily rule out a change in wardrobe as a potential course of action. A "reasonable" person would probably not consider such an alternative, but it must be remembered that rational actors can make mistakes and preferences are not directly observable. (Consider the Hertz example in Chambers 1982:32-34.) Browning and Browning refer to this type of analysis as "benefit-cost" but the order chosen to refer it is of no particular consequence. A more complete description of Zero Base Budgeting can be found in Lyden and Lindenberg's Public Budgeting, Wildav- sky's Politics of the Budgetary Process, and Hammond and Knott's A Zero-Base Look at Zero Base Budgeting. 194 195 There may not have been one definite preferred position for the group. It is possible that one preferred posi— tion could simply have been a return to the status quo. Another might have been action meant to clearly demon- strate United States supremacy. Given the particular situation, Kennedy's preferences were the preferred posi- tion since the final decision was his, but we must remem— ber that his advisors might not have shared those prefer- ences. Von Neumann and Morgenstern do develop an axiomatic structure for the manipulation of utilities in an attempt to move beyond the most basic measurements and proceed to more sophisticated levels of analysis (1980:17-31). They do note, however, that the conditions on which these numerical utilities are achieved are an extension of the conditions under which indifference curves are based (1980:17). It will be shown below that in many situa- tions, including judicial decision making, these condi- tions are unrealistic for practical applications. This is often known as the law of diminishing returns. To illustrate, suppose the costs and benefits 2-1 are for a project to stop pollution. Initial expenditures yield minimal benefits (though the cost-benefit ratio may be large). Efficiency is reached at the point where the marginal costs equal the marginal benefits (the first derivatives of the cost and benefit curves respectively). Suppose 80 percent of the pollution would be eliminated with a project of this scale, but it may be politically advantageous to try for 90 percent and increase the scale of the project. The project is no longer efficient as defined by Browning and Browning but may still be accept- able. To try to clean up the remaining 10 percent would cost much more that it is worth. At this point the pro— ject may be running into technological inadequacies and budgetary limitations as well as greater opportunity costs associated with transferring funds to this project from other sources. Thus, efficiency requires the toler- ation of some pollution. As indicated in note 7, by taking the first derivative of the cost and benefit curves with respect to the scale of the project, setting the resulting equations equal to each other and solving, will yield the point at which marginal costs equal marginal benefits. It is at this point that the scale of the project is considered effi- cient according to Browning and Browning (1983218). This point is not where the two curves intersect. Von Neumann and Morgenstern, Browning and Browning, and Frohlich and Oppenheimer all discuss the rational actor in terms of economic behavior. Economic behavior allows for a certain amount of quantification which makes motives easier to understand. This also allows for a 10. 11. 12. 13. 14. 15. 196 f more rigid specification of the model. As will be shown below, certain characteristics of economic behavior may not be present in noneconomic decisional situations which ‘ in turn may invalidate certain assumptions and require- ' ments of the rational actor model. Consider the game of chess. There are twenty possible f opening moves and twenty possible countermoves which allows for 400 possible combinations after only one round “ of play. Obviously, not all moves are equally likely, especially for players with some knowledge of the strategy of chess, and certain moves will prevent other possible moves. Nevertheless, even in a controlled situ- ation with a specified number of actors (pieces) the possible variations are enormous. Such advice for beginners can be found in books such as Chess by William R. Hartston, An Invitation to Chess by Irving Chernev & Kenneth Harkness, and other beginning chess books. If we write xPy for the preference X is preferred to Y (Frohlich and Oppenheimer 1978:7), then xPy and sz implies xPz by the rule of transitivity. If, however, we are not sure that X is preferred to Y and write the rela- tion as x?y, we may still know that sz, but we can only conclude that x?z. If in fact we can determine the spe- cific preference relation between X and Y, from the rule of transitivity we can say something about their prefer- ence relation with Z. If not, we can only determine that one of three possibilities exists: xPz, or sz, or sz (where I stands for indifference). Since these three alternatives comprise the complete alternative set we have gained nothing without an initial nonprobabilistic preference. "It is interesting to consider that psychologists have been using seven-point rating scales for a long time, on the intuitive basis that trying to rate into finer cate- gories does not really add much to the usefulness of the ratings." (Miller 1956:84) The same may be said for political scientists. In addition, the magical number seven has found its way into popular reading (Newsweek 9/29/86z48-54). Decision making is essentially problem solving. In deci- sion making the actor has a goal to be reached and the problem is to decide on the course of action which brings the actor closest to that goal. The Cult of the Robe is a phrase used to denote a number of symbols used by members of the judiciary to lend more dignity and authority to their position. These symbols include the wearing of black robes, being seated above the rest of the courtroom, having everyone rise when they 16. 17. 18. 19. 20. 197 enter, addressing them as "Your Honor", and so on. Other manifestations of the Cult will be discussed later in the context of traditional judicial dogma. Amicus curiae literally means "friend of the court" (Ehrlich 1985:42). When the Court hears cases of consti— tutional importance it is likely that a number of amicus briefs will be filed. The record seems to be held by Regents of the University of California v. Bakke, 438 U.S. 265 (1978) in which 58 amicus briefs were filed (Murphy and Pritchett 1986:193). For example, in The Memory Book by Harry Lorayne and Jerry Lucas, the reader is shown methods of memorizing lists of words. To try the method the reader is asked to memorize the states in alphabetical order. The key to their "Peg" system is to form associations from one state to the next. Thus one may associate Alabama with "album." On the album is a picture of baked "Alaska" which is being eaten by someone in an "arid zone" (Ari— zona) and so on. For the beginner the method requires a great deal of practice. Continued use of the system improves the speed and ability to use the method, not the span of immediate memory. What this knowledge might guarantee is an inability to find an opponent willing to put up with the excessive amounts of time required to trace the moves of the game through the diagram. Most computerized chess games can be set for a number of levels of difficulty with the computer examining more possible moves at the more diffi- cult levels. In the higher levels of difficulty the computer may take an average of one (Radio Shack 1650) to four (Sargon II) hours between moves. I imagine very few players would have the determination and resolve to play such a lengthy game. In fact, in tournament chess strict time limits are placed on the players. If a player exceeds the time allotted he or she forfeits the match. A basic problem with estimating the goals, costs, and benefits of another decision maker is the nature of the utilities generally used to provide some measure of com- parison. Whether the units are utiles or dollars, the basic fact remains that utilities are personal and cannot be compared across individuals. With "adequate" informa- tion such utilities may be estimated with a fair degree of success but they will never be known with absolute certainty. There are two types of uncertainty. Uncertainty may be present when there is simply a lack of any information. This may occur when a consumer is shopping for a new appliance and does not know about the reliability of a particular manufacturer's products. A driver in a new city may have a lack of information about local side 21. 22. 198 streets if he accidentally gets off at the wrong exit. In both situations the decision maker can seek out infor- mation. The consumer may ask a sales clerk for informa- tion and the driver may stop and ask someone for direc- tions. It is also possible for both decision makers to fill in the gaps in their information with certain assumptions. If the consumer frequents the store on a regular basis he may assume that the store would not carry a product which was unreliable. He might also assume that if the product carries a strong warranty, it is more reliable. The driver may assume that if he back- tracks he can get on at the same place again. He might also assume that the roads are laid out in a basically rectangular pattern and if he continues parallel to the expressway he will find the next onramp without undue difficulty. In both cases the decision maker is utiliz- ing the pattern of past information to fill in the gaps with respect to current conditions. The second type of uncertainty occurs when the decision maker has a great deal of information but does not know what information represents the current state of the world. Kennedy had a great deal of information on Soviet motives and goals, but he did not know specifically what inspired them to place missiles in Cuba. Again, the strategy is to rely on previous experience and patterns to make a decision. Kennedy relied on trusted and knowledgeable advisors to help sort through the information which was being received by the group. Best estimates were made and the American response chosen. Alexander's diagram seems to fail to account for the possibility of interrelations between the concepts, but this is not the case. The interrelationships manifest themselves at the top of the subpyramids. Effects of production on economics would not be considered until the three functional concepts are satisfied and then function and economics are considered jointly. If there is a conflict, it may be necessary work back down the paths, find a different solution, and work back up to function and economics again. This is not necessarily the most efficient method for problem solving, but it does allow all the requirements to be considered in turn. Cyert and March note that when a problem is thus divided and approached in a sequential fashion local rationality may result (1963:117). The goals and problems of the subunit handling production will be quite limited and may only be expanded when the solution is returned for revision. What Simon refers to as his principle of bounded ration- ality is different from what economists and other adher- ents of copmrehensive rationality call bounded rational— ity. As used in Section A of this Chapter, bounded rationality merely means that certain limitations on the choice set or information available are recognized, but the decision maker must still maximize within those ”94‘ —-,: - <-—. r a 191..., _ . 23. 24. 25. 26. 199 limitations. As Simon uses the term, the decision maker not only constructs a simplified version of the problem, but is also not required to find the solution which pro— vides maximum benefits, only one which is satisfactory. Hence, the term satisficing has come to be associated with Simon's conception of bounded rationality. The behavior patterns of the Watt governor and the thermostat are well known and understood. Aside from breakdowns or extreme operating conditions they will operate precisely as planned. This certainty is a major distinguishing feature these machines have with human decision making. At present, we can never be sure how a decision will be made. One might adopt a reductionistic determinism point of view to eliminate this distinction, i.§., by knowing and being able to measure all the influences on the human brain, decisions would be as understandable as those of the servomechanism. This argument begins to enter the conflict between determinism and freewill; a topic beyond the scope of this work. But since such measurements are not presently available, we must rely on the notion of the Black Box which will be discussed below. Steinbruner, paraphrasing Simon, tries to make this dis- tinction by comparing the rational choice model with a "blueprint" and the cybernetics model with a "recipe“ (1974:55). Although this analogy does point out the more flexible aspects of the cybernetic model it seems to require the conscious consideration of the problem which Ashby has deemed unnecessary. At this point Beer states that the third characteristic of cybernetic systems has been revealed: exceedingly complex, probabilistic, and homeostatic. If this is true, why does Beer use less complex deterministic machines such as the Watt governor and the thermostat to illustrate the basic notions of cybernetics? I suspect the reason has to do with the ease of illustrating these concepts in a simpler environment before moving on to more complex examples where the notions may not be as easily isolated for discussion. Steinbruner accepts servomechanisms as being cybernetic machines (1974:51), but also moves on to more complex areas (1974:61). Beer does not flatly state that servomechanisms are not cyber- netic machines, only that their operation has been ade— quately described by other fields and it is with the more complex probabilistic systems that cybernetics is of greatest value (1964:18). Steinbruner states that the feedback loop is between the engine speed and the throttle opening (1974:52). Both Steinbruner and Beer agree that information inputs are feedback, thus, inputs to any element of a system can be considered feedback to that element. The key is in 27. 28. 200 choosing which element to be considered the starting and ending point of the information loop. Since the discus— sion is intended to illustrate the concepts of cybernet- ics via the workings of a servomechanism (in this case a Watt governor), this control device should logically be the focus of our attention. The feedback loop indicated by Steinbruner seems to ignore what should be the focal point of the system. The reason the number of inputs is small has to do with the cognitive limitations of the human decision maker, although this is not explicitly stated by either Stein- bruner or Beer. Steinbruner rejects the idea that a cybernetic machine can act with a purpose (1974:62-66). He accuses Ashby of letting the "critical values" and their monitoring become a purposeful action on the part of the decision maker even though his (Ashby's) model purports to reject out— come calculations and payoff scales. This attitude on the part of Steinbruner seems in conflict with the example he uses to illustrate the cybernetic model: the tennis player (1974:60). Clearly the tennis player has a purpose in mind when he advances to hit the ball, namely to return it to his opponent's side of the court. He rejects both optimization and satisficing for survival (1974:64-65). How can it be said that the tennis player is engaging in survival techniques? This inconsistency may be explained, I think, by two points. First, Stein— bruner does not seem to be using the term "purposeful" in the same manner as Ashby and Beer. His reference to the calculation of outcomes (1974:63) is more consistent with the rational choice model which he had rejected. Ashby and Beer seem to allow purpose to be some goal of the machine. As with other concepts, it is difficult to believe that the Watt governor has a purpose, but it is not difficult to accept the statement with regard to human decision makers. This leads us to the second point. Both Ashby and Beer make use of the Black Box in their treatments of the cybernetic model. As will be noted later in the text, it is within this box that the extreme complexity of the human mind is contained, including such concepts as goals, values, and purposes. Ashby and Beer do not ask why the tennis player wants to hit the ball back into his opponent's court, but rather, given the desire, how does the machine accomplish the task. Steinbruner covers this material in the form of the cognitive paradigm, which attempts to discover the contents of the Black Box. The problem with Stein- bruner's approach to the cybernetic paradigm is that by eliminating the notion of a Black Box, he has forced any consideration of human decision making out of the realm of cybernetics. 29. 30. 31. 32. 33. 34. 35. 36. 37. 201 The use of a Black Box to indicate processes that cannot as yet be clearly defined may seem to be the easy way out of a sticky situation. Beer notes, however, that, "the black Box treatment of the system itself avoids the pit- falls of oversimplifying its nature through some wild and misleading analogy" (1964:70). In addition, the cyber- netic model is not a perfectionist model, it allows for mistakes, breakdowns, and random occurrences. At some point these failures of the system may be explained by the model, but for present purposes it is better to put them into the Black Box than to fashion exceptions to an already strained analogy. One can find many collections of optical illusions which illustrate this point. One such collection is Gyles Brandreth's The Great Book of Optical Illusions. The illusions presented on pages 6 and 7 are particularly appropriate to the discussion. Much of the following material on the trappings of the cult of the robe will not be directly citeable to spe- cific sources. Generally it is based on information con— tained in Murphy and Pritchett (1986), Rodell (1955), Rohde and Spaeth (1976), Spaeth (1979), and Woodward and Armstrong (1979), as well as personal experience at two law schools. This does not count the thankfully brief period when President Richard Nixon had the White House staff wear outlandish uniforms. For a more complete treatment of the origins of the English legal system see generally, A Constitutional and Legal History of Medieval Encland by Bryce Lyon. Woden: "The West Germanic form of Odin“ (Evans 1970:296). Odin is the All-Father of Northern mythology. It is true that the jury, which is to be composed of one's peers, decides what the facts are in a case, but remember that the judge is still the controlling factor and decides what evidence the jury is allowed to hear and consider in making its decision. This statement applies generally to all courts, espe- cially the federal courts, and particularly the Supreme Court of the United States. State courts also must con- sider state constitutions as well as the federal consti- tution but the latter remains the ultimate authority. The factual situation surrounding Marbury led to what was probably the most fortuitous political maneuvering in United States history. This is ironic and paradoxical given the arguments made by Marshall in claiming judicial review as the province of the judiciary. Nevertheless, 202 the facts are quite enlightening and are summarized in Murphy and Pritchett (1986) as follows: Before going out of office after their crushing defeat in 1800, the Federalists enacted the Judiciary Act of 1801. While this law made several needed reforms, its creation of many new judicial posts——to which Adams nominated and the lame-duck Senate confirmed deserving Federal— ists-~made it vulnerable to a charge of court packing. In signing and delivering the commis- sions the night before Jefferson's inauguration, Adam's Secretary of State, John Marshall, neglected to deliver a number of appointments to justice-of-the—peace courts in the District of Columbia. The new Secretary of State, James Madison, refused to deliver some of these com— missions. William Marbury, one of the disap- pointed appointees, brought suit in the Supreme Court under a provision of the Judiciary Act of 1789 that Marbury claimed gave the Court origi— nal jurisdiction in such cases. (1986:16) Marshall's opinion did not give Marbury the commission, but the reasoning behind the decision left Jefferson with a difficult choice. Jefferson did not want Marbury to get the commission but in order to accept the Court's decision, he was forced to accept the Court's reasoning, which included the premise that the Supreme Court had the authority to determine if federal action was constitu- tional. To have challenged the decision would have meant more Federalists in office; not a desirable outcome to Jefferson. By essentially giving Jefferson what he wanted, Marshall managed to greatly expand the powers of the court. (For a more detailed account of the setting and the opinion see Nowak, Rotunda, and Young 1983:2-15.) Marshall's ability to seize the moment was certainly brilliant. On the other hand, I would think Marshall no less brilliant--and I know of no evidence to support this--if he had purposefully withheld the commissions to force the situation into existence. 38. This was an incredible argument on Marshall's part given the fact many members of Congress had taken part in the drafting of the Constitution and Marshall had not. 39. This material quoted by Statsky and Wernet is footnoted and attributed to Swift, J., Gulliver's Travels and other Writin s, p. 203 (R. Quintana, editor), (Random House, The Modern Library, 1958). 40. This unwillingness on the part of the Court to make unnecessary decisions may be an overlap from the consti- tutional requirement of deciding only actual cases or 41. 42. 43. 44. 203 controversies (U.S. Constitution, Article III, Section 2). If some particular point was not necessary then it could be argued that it was not in controversy or part of the case. The importance placed on factual analysis begins with the first year of law school where law students are taught to "brief" cases for later discussion and study. Briefing involves determining the facts of the case, the law applied, and the conclusions reached by the court. Ostensibly, the purpose of briefing is to improve the analytical skills of the student. It does improve these skills, but it also deeply reinforces the importance of the facts and the need to find precedent which duplicates the facts of the present case as closely as possible. From this point on the discussion will focus primarily on the United States Supreme Court, referred to simply as the Court (capital "C“). One reason is that the data necessary to illustrate various points is more readily available for the Supreme Court than for other federal or state courts. For the most part any statements and argu— ments on caseloads, decision making, and the like, will apply to both the Supreme Court and other courts as well. Differences will be noted when this is not the case. A second reason for focusing on the Supreme Court at this point is the fact that the data set will be drawn from its decisions.. Woodward and Armstrong offer contradictory bits of evi- dence with respect to Burger's attitude on this point. Early in his tenure as Chief Justice, Burger issued a memo to his law clerks essentially telling them that although they occupied the same building as the clerks of the other justices, they were not to consider themselves part of one big law firm, but rather nine separate law firms (1979:34-35). Later, in an effort to lessen the workload of the justices, Burger proposed and implemented a cert pool. Rather than have a duplication of effort_by having the clerks of each justice write summaries of certiorari petitions, the petitions would be divided among the clerks and the summaries copied for each jus- tice; the proposal met with mixed reactions (Woodward and Armstrong 1979:272-273). Spaeth notes that if a great deal of vote switching did take place it would disrupt the Court's decisional pro- cess since the majority opinion is originally assigned by the Chief Justice or the most senior Associate Justice (when the Chief is not in the majority) of the majority formed by the initial vote (1979:22). On the other hand, Woodward and Armstrong found evidence of a tendency on the part of Chief Justice Burger to switch his vote fairly regularly, much to the dismay of some of his bre- thren (1979:417-418). 45. 46. 47. 48. 204 The importance to the Court of the fact pattern of a case is made clear in O'Connor v. Ortega. In this search and seizure case eight of the nine justices who participated indicated that cases in this area must be decided on a case—by-case basis. Only Justice Scalia refused to accept this position, stating, ". . . I would object to the formulation of a standard so devoid of content that it produces rather than eliminates uncertainty in this field" (480 U.S. _, _; 94 L.Ed.2d 714, 731). (It should be noted that this case is not in the data set because the Court did not reach the merits of the case and remanded it for further review on procedural grounds.) Segal notes the following: "Although some have modeled Supreme Court Justices as utility maximizers (e.g., Rohde, 1972; Teger, 1977), Niemi and Weisberg (1974) quite clearly point out that unidimensionality and single peakedness, in the sense used by formal theorists, can not be inferred from Guttman scaleability" (1983:56). Technically speaking, every case that the Court chooses to hear is different from any other case they have heard. If this were not true the Court would be duplicating their efforts and wasting their time. Since, as has been argued, the Court is a policy making body, it is reason— able to assume that they want to make their policies as clear as possible, and general enough to give enforcement officials sufficient guidance in applying the policy. Thus, when the Court hears cases relating to a past precedent, these new cases generally explore some new aspect of the general problem which the Court was unwill- ing or unable to address in previous cases. These are the "additional problems" which must be handled by the Court. In the codebook to his United States Supreme Court Judi- cial Data Base, Professor Spaeth lists eleven possibili— ties for the disposition of a case before the Supreme Court. These possibilities are as follows: 0 = stay, petition, or motion granted 1 = affirmed 2 = reversed 3 = reversed and remanded 4 = vacated and remanded 5 = affirmed in part and reversed in part 6 = affirmed in part and reversed in part and remanded 7 = vacated 8 = petition denied or appeal dismissed 9 = certification to a lower court (blank) = a case arising under the Supreme Court's original jurisdiction, in which situa- tion there is no lower court decision to review 49. 50. 205 The original jurisdiction of the Court is rarely invoked, and the remaining ten possibilities can be placed into three basic categories: affirm, reverse, and no deci- sion. Since we are interested in decisions of the Court, cases where no formal decision was reached on the merits of the case are not considered, and we are left with the two basic possibilities of affirm or reverse. The words "cues" and "cue theory" have taken on a special meaning in the discipline, namely a reference to the cue theory of certiorari. I prefer to use these terms in a more general sense. The preceding discussion has indi— cated that justices make their decision based on certain facts in a case. The presence of these facts signals the justice that a particular decision is in order, just as the fact that the rotating arms are swinging too far outward signals the Watt governor that the engine speed is too fast. Such signals are cues to particular behav- ior necessary to maintain the stability of the system. Thus the mention of "cue theory" refers to the assump- tions that behavior is dependent on the presence of cues or signals in the decision maker's inputs. The "cue theory of certiorari" refers to the theory developed by Tanenhaus, et al. The Fourth Amendment of the United States Constitution states: / i The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated, and no Warrants shall issue but upon probable cause, supported by Oath / or affirmation, and particularly describing the 1 place to be searched, and the persons or things , to be seized. Chapter 3 1. The Exclusionary Rule was created to operate as a deter- rent to unconstitutional police conduct. If police actions are considered constitutionally unreasonable, any evidence obtained cannot be used against the subject of the illegal search. The exclusion of such illegally obtained evidence is intended to give the police an incentive to operate within the law and punish them if they do not. It should be pointed out that Mapp v. Ohio is not part of the data set. In Ma , the Court did not reach the question of the reasonableness of the search. They made the procedural decision regarding the applicability of the Exclusionary Rule to state law enforcement agencies ._._-_._’._;f 206 and sent the case back to the Ohio courts for further review. A concern exists when there seems to be some combination of the location areas, such as when the suspect uses part of a residence as a business office (but the Court did not confront this question). A second questionable area concerns motor homes which are used as homes on wheels. The Court did face this issue in California v. Carney and took the position that the mobility of the vehicle required that it be treated as a car rather than a home (Brennan, Marshall, and Stevens disagreed). Although a motor home is neither strictly a home nor a vehicle, a majority of the Court decided to fit it into one of the established categories rather than fashion another, as the dissenters seemed willing to do. An additional assumption is that the underlying probabil— ity function is logistic as opposed to normal in the case of logit. The difference between the two functions is minimal and occurs at the tails (Hanushek and Jackson 1977:204). An additional problem associated with the use of a popu- lation rather than a sample involves the type of statis- tics used in analysis of the results. For the most part, both regression and probit analysis utilize the t-statis— tic as an indicator of the significance of the variables used in the model. The t—statistic, however, is intended for use with samples from a population whose underlying variance is unknown (Hanushek and Jackson 1977:122—124), and the normal distribution should properly be used. For purposes of statistical analysis, the population chosen for study may be considered a sample. This may seem to be playing fast and loose with the definitions and dif- ferences between samples and populations, but it is jus— tified for two main reasons. First, by narrowly defining a particular data set as the population, g.g. Supreme Court search and seizure cases decided from 1961 through 1987, the ability to explain and draw conclusions beyond the scope of the population is severely limited. Never— theless, this does not obviate the fact that this "popu- lation" is still part of a larger population of all search and seizure cases decided by the Burger Court, which is in turn part of the larger population of all search and seizure cases decided by both the Warren and Burger Courts, and so on. Defining a narrow population primarily affects the conclusions that can be drawn from the statistical analysis and not the analysis itself. Second, as noted by Kmenta (1971:143) as the sample size becomes large the t-distribution approaches the normal distribution. "Large" is sometimes understood to be over 30 observations, and at other times to be over 100 obser- 207 vations. The sample set for the research at hand con— tains over two hundred observations. Chapter 4 1. Nine justices voting in each of the 204 decisions would yield 1836 individual votes. There were times, however, when the Court was forced to operate with less than a full compliment of justices due to vacancies, illness, or recusal of one of the justices. 2. Eleven independent variables would seem to be larger than the magical number seven (plus or minus two), however, five of these independent variables relate to the loca- tion of the search or seizure. Unlike the variables drawn from Hypotheses 2 through 6, the location of the search or seizure cannot be dichotomized as being present or not. Five categories of locations are identified (six if one includes the declaration by the Court that no search took place). The impact a particular location will have on the judicial decision maker can be logically derived to the extent that the locations can be ranked in order of their estimated importance in the model. How— ever, the derived rankings are not specific enough to place them in one variable with a meaningful numerical value assigned to each location. Thus, each of the five locations is assigned a dichotomous variable. Viewing the location variables as one factor, the model estimates seven factors considered by the judicial decision maker. 3. In the interest of parsimony, the model was estimated a second time without the variable CAR with no appreciable change in the results. The inclusion in a model of irrelevant variables can produce suboptimal results. In this case, CAR is not irrelevant (it being part of the location choice set) and its deletion from the model only minimally affects the estimation results. 4. ROE measures the reduction in error over the modal cate- gory and is calculated as follows: 0 % correctly classified - 6 in modal category ROE(%)=100 x 100% - % in modal category where % correctly classified is the percentage of obser— vations which the model correctly classifies, and % in modal category is the percentage in the modal category of the dichotomous dependent variable. ROE is based on the idea of diminishing returns and cost-benefit curves (see Browning and Browning 1983:114—125) and offers a way of comparing the success of one or more models over data sets with differing distributions of the dependent vari- able. 208 ROE is too simple of a statistic not to believe that it has not been used before. In fact, I have found two previous uses. Freeman (1965) discusses Guttman's coef- ficient of predictability as a method of determining how the values of one variable can be used to predict the values of another (1965:71-78). Weisberg (1978) dis— cusses Goodman and Kruskal's lambda which uses a similar formulation for a determination of a baseline in predict- ing roll-call vote in the House of Representatives. Both the Guttman and Goodman and Kruskal formulations are mathematically equivalent to the formulation of ROE. One difference lies in ROE's use of the modal category as the baseline. Use of the modal category provides a quick and easily ascertainable baseline against which the success of multivariate estimations may be compared. Weisberg indicates Goodman and Kruskal's lambda uses party identi— fication as the baseline, or naive model. This is appro— priate for roll—call voting, but would require a differ— ent standard when examining other areas. Thus, ROE is a more generic formulation. The correctly and incorrectly classified votes were com- pared to each of the independent variables included in the model, as well as a number of alternative measures of the factors found in the hypotheses. None of these com- parisons revealed systematic differences between cor- rectly and incorrectly classified votes. In addition, an examination of the estimated probabilities and error terms for autocorrelation did not reveal any systematic pattern to the size or direction of the errors. It did appear that a larger portion of the misclassified votes for unreasonableness are from the earlier part of the time period covered by the data set. This is simply because a greater portion of all the votes for the unrea- sonableness of a search or seizure occurred during the earlier part of the time period. This point will be discussed again later in the chapter. In the course of estimation probit generates an estimated probability that each observation takes on the highest value of the dependent variable. This value is often referred to as PHAT. The classification rule utilizes PHAT to divide the observations into those classified as taking on the high value of the dependent variable and those which have the low value. The common decisional cutoff point is PHAT = .5, where values of PHAT greater than or equal to .5 are classified as taking on the high value of the dependent variable and those less than .5 are classified as taking on the low value. For any observation, VOTE is classified as 1 if the probability of the observation taking on the value 1 is greater than or equal to .5, and VOTE is classified as -1 if the prob- ability of the observation taking on the value 1 is less " — f—7¢'_-7._,’ "" — '—'V' ' 111,111,17 ._-: ._'-_'~a:._‘r9.:z;e.;xd ., .1 1) 209 than .5. 8. The purpose of using a runs test is to see if there is systematic variation based on the ordering of the obser— vations. The variable selected to order the observations is, therefore, of great importance. Quite often, the observations are collected in a time sequence, as is the case with the search and seizure data. However, the time sequence in which the observations were gathered is not directly related to the dependent variable, and only indirectly due to the sequence of personnel changes on the Court (which will be discussed in detail later in the text). In addition, such an ordering would still leave us with the problem of numerous tied ranks and a depen- dency on additional inappropriate variables to break them. 9. One method of differentiating among the justices in a given case is to create dummy variables which indicate the justice who is casting the vote in a particular observation. Nineteen justices cast votes in the period covered by the data set. Two of these justices (Frank— furter and Scalia) cast too few votes to make creation of a dummy variable for them feasible. Inclusion of the 17 variables representing the justices results in a dramatic increase in the percentage of votes correctly classified. The additional variables increase the correctly classi- fied votes to 1368 of 1775 (77.1%). Examination of these results, however, reveals that the additional variables are forcing classification of the votes of a justice into the modal category of that justice's actual votes. The A high percentage of votes in the modal category for many 1 of the justices (Justices Black, Brennan, Burger, Dou- i glas, Goldberg, Marshall, O'Connor, and Rehnquist are all ; over 80%) causes the increase in the correct classifica- ‘ tion rate. Aggregation of the estimated votes only increases the correct classification of the Court's deci— i sions by two decisions to 148 of 204 (72.5%). 10 Despite the limitations placed on the estimation of the I votes of the individual justices, the results on a justi— : ce—by-justice basis should be noted. No consistent pat- terns exist with respect to the number of votes correctly I classified, the percentage of votes in a justice's modal I category, and the percentage of agreement a justice has with the majority. The lowest correct classification 1 rates are for Justices Marshall (50.0%), Douglas (51.2%), I and Brennan (52.0%). These low rates can be explained by l a combination of two factors. First, the modal category percentage for each of these justices is over 80%. Sec— f ond, for all three of these justices the majority of their votes were cast against the reasonableness of the search or seizure. This results in very low percentages of agreement with the majority (less than 60%). On the other end of the ideological spectrum, Justices Burger, 11. 210 O'Connor, and Rehnquist each have modal category percent- ages in excess of 80%, but have higher correct classifi- cation rates because more of their votes were cast with the majority. Justice Powell has the highest rate of agreement with the Court at 93.0%, but both Justices Rehnquist and Stewart have higher correct classification rates (77.3% and 70.7% respectively as compared with Powell's 69.7%). As indicated in the text, the number of votes a justice casts in dissent directly affects the number of votes which can be correctly classified. To see how PF values are determined, let us examine the decision where PFf = .027. In this case, Almeida—Sanchez v. U. S. the independent variables and their estimated coefficients are as follows: CONSTANT 1.22 * 1 = 1.22 HOME —1.55 * 0 = 0 BUSINESS -1.20 * 0 = 0 PROPERTY -1.15 * 0 = 0 CAR —0.30 * 1 = -0.30 PERSON -0.64 * 0 = 0 CONTROL -0.47 * 0 = 0 CUSTODY -0.69 * 1 = —0.69 TIME 0.45 * 1 = 0.45 DRUG/GUN 0.64 * 1 = 0.64 SEVERE -0.94 * O = 0 US 0.60 * 1 = 0.60 Z-score (phat) 1.92 From the standard normal distribution table, a Z-score of 1.92 translates into a probability of .9726 that the dependent variable takes on its high value. By the clas- sification rule, this decision would be classified as a reasonable search or seizure. Unfortunately the Court did not agree, and the dependent variable, DECISION, took on its low value. Thus, PFf = 1 - phat = .0274, which makes this the most badly missed classification. One may wonder why the model missed so badly on this decision. An examination of the facts do provide some clues. The case involved a roving border patrol on the lookout for illegal aliens, but the circumstances of the search led a majority of the Court to decide the patrol exceeded its authority to search without a warrant. It is quite pos- sible that if illegal aliens had been found instead of drugs, the lesser penalties involved would not have pushed the balance so far toward the protection of the individual and the search might have been allowed. This suggests a reverse logic. The roving border patrol's actions were somewhat questionable. If they had found illegal aliens, the penalties for the suspects would not be particularly severe, at least in relation to the penalties for drug smuggling. Thus, the higher 12. 13. 14. 15. 16. 211 penalties involved in drug charges caused the Court to View the actions of the patrol more carefully. The result is the suspect goes free, despite the model's estimate that the presence of drugs increases the likeli- hood the search or seizure will be considered reasonable. In a general sense, we do want PFf values to be close to the cutoff point. This indicates the possibility that these observations may be correctly classified if the specification of the model is slightly altered or other minor adjustments are made. However, when the PFf values are close to the cutoff point it may not be obvious what alterations are needed. If the PFf values are very low, closer examination of the observations may reveal simi— larities which account for their misclassification. If no such similarities exist or can be found, the observa— tions may be considered outliers and treated accordingly. The decisional cutoff point for classification, and thus , for success, is .5. Therefore, by definition, when 1 examining the PF values of correctly classified observa- tions (PFS) the minimum value must be greater than or equal to .5. For this model, the minimum PF value is .501. Similarly, the maximum PF value for misclassified observations (PFf) must be less than .5. For this model the maximum PFf value is .499. Clearly, an estimated coefficient of .02 and a standard error of .005 should yield a t—statistic of 4.00. This and all other discrepancies in the Tables are due to roundoff errors. Two other possible measures could have been used to cap- ture this time element. Segal (1984) used a variable called CHANGE to indicate personnnel changes on the Court. A second alternative is a variable, TERM, which indicates the year of the term in which the search or seizure was decided. LEDCITE is used simply because it is readily available; LEDCITE is coded for every observa- tion for identification purposes. CHANGE, TIME, or any other similar variable could have been used to make the point. "The statement that a . . . Justice does not support the values of freedom, equality, or New Deal economics is not meant to be pejorative. Support or nonsupport is a matter of degree——a measure of comparative support or nonsupport. . . . Moreover, a Justice may well believe that other considerations outweigh support for such val— ues as freedom and equality. He may strongly believe that individuals must bear full responsibility for their won actions and that the government must not engage in social engineering regardless of “humanitarian" consider— ations. Such a Justice would likely appear hostile to equality as it has been described . . . . Similarly, a 17. 212 Justice who was strongly supportive of states' rights and who thus believed that primary responsibility for enforcement of the criminal laws rested with the states. Such a Justice would likely be tolerant of state efforts to cope with crime, with the result that he would appear to be antifreedom" (Spaeth 1979:134). The data Spaeth used to construct the value systems and underlying scale scores covers the period from 1958 through 1977. Since that time minor fluctuations in the scores and/or the ordering of the justices may have occurred. Since the terms of Justices O'Connor and Scalia postdate the original data period, their placement is based on informal educated guesses of Professor Spaeth. Justices Black and Harlan served through the 1970 term. For a more complete explanation of the construction of scales values, and value systems, see Spaeth (1979:109-139). APPENDICES APPENDIX A. Cases Contained in the Data Set Listed below are the 204 cases contained in the data set. They are listed in decending order based on their citation in the United States Supreme Court Reports. Lawvers' Edition. 2d. Case Name Date U.S. L.Ed 2d Griffin v. Wisconsin (1987) 483/ * 097/0709 New York v. Burger (1987) 482/ 096/0601 Illinois v. Krull (1987) 480/ 094/0364 Arizona v. Wicks (1987) 480/ 094/0347 U. S. v. Dunn (1987) 480/ 094/0326 Maryland v. Garrison (1987) 480/0079 094/0072 Colorado v. Bertine (1987) 479/0367 093/0739 DOW Chemical Co. V. U. S. (1986) 476/0227 090/0226 California v. Ciraolo (1986) 476/0207 090/0210 New York v. Class (1986) 475/0106 089/0081 U. S. v. Montoya De Hernandez (1985) 473/0531 087/0381 Maryland v. Macon (1985) 472/0463 086/0370 California v. Carney (1985) 471/0386 085/0406 Tennessee v. Garner (1985) 471/0001 085/0001 Tennessee v. Garner (1985)l 471/0001 085/0001 Hayes v. Florida (1985) 470/0811 084/0705 Winston v. Lee (1985) 470/0753 084/0662 U. S. V. Sharpe (1985) 470/0675 084/0605 U. S. v. Johns (1985) 469/0478 083/0890 New Jersey v. T. L. o. (1985) 469/0325 083/0720 U. s. v. Hensley (1985) 469/0221 083/0604 Thompson v. Louisiana (1984) 469/0017 083/0246 Florida v. Rodriquez (1984) 469/0001 083/0165 I. N. S. v. Lopez—Mendoza (1984) 468/1032 082/0778 Massachusetts v. Sheppard (1984) 468/0981 082/0737 U. S. V. Leon (1984) 468/0897 082/0677 Segura v. U. s. (1984)2 468/0796 082/0599 Segura v. U. s. (1984) 468/0796 082/0599 U. S. v. Karo (1984) 468/0705 082/0530 Hudson v. Palmer (1984) 468/0517 082/0393 Welsh v. Wisconsin (1984) 455/0740 080/0732 Massachusetts v. Upton (1984) 465/0727 080/0721 Florida v. Meyers (1984) 466/0380 080/0381 I. N. s. v. Delgado (1984) 466/0210 080/0247 Oliver v. U. S. 1984)3 462/3178 gig/8:1: ' 46 Oliver v. U. S. 1984) 466/0109 080/0085 U. S. v. Jacobsen (1984) 213 214 (1984) (1984)4 (1984) (1983) (1983) U. S. v. Place 1983) Illinois v. Lafavette (1983) U. S. v. Villamonte-Margpez (1983) Illinois v. Gates (1983) Texas v. Brown (1983) Florida v. Royer (1983)5 Florida v. Royer (1983) U. S. v. Knotts (1983) Michigan v. Thomas (1982) U. S. v. Ross (1982) Washington v. Chrisman (1982) New York v. Belton (1981) Robbins v. California (1981) Michigan v. Summers (1981)6 Michigan v. Summers (1981) Donovan v. Dewey (1981) Steagald v. U. S. (1981) U. S. v. Cortez (1981) Colorado v. Bannister (1980) Reid v. Georgia (1980) Rawlings v. Kentucky (1980)7 Rawlings v. Kentucky (1980) U. S. v. Payner (1980) Walter v. U. S. (1980) Walter v. U. s. (1980)8 U. s. v. Mendenhall (1980)9 U. S. v. Mendenhall (1980) Pavton v. New York (1980) Edyton V. New York (1980)10 Ybarra v. Illinois (1979) Brown v. Texas (1979) Michigan v. DeFillippo (1979) Arkansas v. Sanders (1979) Smith v. Maryland (1979) Torres v. Puerto Rico (1979) Lo—Ji Sales. Inc. v. New York (1979) Dunawav v. New York (1979) Bell v. Wolfish (1979)11 Bell v. Wolfish (1979) Dalia v. U. S. (1979) U. S. v. Caceres (1979) Delaware b. Prouse (1979) Michigan v. Doran (1978) Rakas b. Illinois (1978) Franks v. Delaware (1978) Mincey v. Arizona (1978) Zurcher v. Stanford Daily (1978) Zurcher v. Stanford Dail (1978)12 Michigan v. Tyler (1978)¥3 Michigan v. Tyler (1978) 464/0408 464/0287 464/0287 463/1032 463/0765 462/0696 462/0640 462/0579 462/0213 460/0730 460/0491 460/0491 460/0276 458/0259 456/0798 455/0001 453/0454 453/0420 452/0692 452/0692 452/0594 451/0204 449/0411 449/0001 448/0438 448/0098 448/0098 447/0727 447/0649 447/0649 446/0544 446/0544 445/0573 445/0573 444/0085 443/0047 443/0031 442/0753 442/0735 442/0465 442/0319 442/0200 441/0520 441/0520 441/0238 440/0741 440/0648 439/0282 439/0128 438/0154 437/0385 436/0547 436/0547 436/0499 436/0499 078/0567 078/0477 078/0477 077/1201 077/1003 077/0110 077/0065 077/0022 076/0527 075/0502 075/0229 075/0229 075/0055 073/0750 072/0572 070/0778 069/0768 069/0744 069/0340 069/0340 069/0262 068/0038 066/0621 066/0001 065/0890 065/0633 065/0633 065/0468 065/0410 065/0410 064/0497 064/0497 063/0639 063/0639 062/0238 061/0357 061/0343 O6l/0235 061/0220 061/0001 060/0920 060/0824 060/0447 060/0447 060/0177 059/0733 059/0660 058/0521 058/0387 057/0667 057/0290 056/0525 056/0525 056/0486 056/0486 215 Marshall v. Barlow's Inc. (1978) Scott v. U. S. (1978) U. S. v. Ceccolini (1978) U. S. v. New York Telephone Co. (1977) Pennsylvania v. Mimms (1977) Nixon v. Administrator of General Services (1977) U. S. v. Chadwick (1977) U. S. v. Ramsey (1977) U. S. v. Donovan (1977) G. M. Leasing Corp. v. U. s. (1977)14 G. M. Leasing Corp. v. U. S. (1977) U. S. v. Martinez-Fuerte (1976) U. S. v. Martinez-Fuerte (1976)15 U. S. v. Janis (1976) South Dakota v. Opperman (1976) Andersen v. Maryland (1976) U. S. v. Santana (1976)16 U. S. v. Santana (1976) U. S. v. Miller (1976) U. s. v. Watson (1976)17 U. S. v. Watson (1976) State of Texas v. White (1975) U. S. v. Ortiz (1975) U. S. v. Brignoni-Ponce (1975) Gerstein v. Pugh (1975) Cardwell v. Lewis (1974) Air Pollution Variance Board v. Western Alfalfa Corp. (1974) U. S. v. Chavez (1974) U. S. v. Chavez (1974) U. S. v. Giordano (1974) Gooding v. U. S. (1974) California Bankers Assn. v. Shultz (1974) California Bankers Assn. v. Shultz (1974) U. S. V. Edwards (1974) U. S. v. Calandra (1974) U. S. v. Matlock (1974) Gustafson v. Florida (1973) U. S. v. Robinson (1973) Roaden v. Kentucky (1973) Cady v. Dombrowski (1973) Almeida-Sanchez v. U. S. (1973) Cupp v. Murphy (1973) Schneckloth v. Bustamonte (1973) Brown v. U. S. (1973) U. S. v. Mara (1973) U. S. v. Dionisio (1973) Gelbard v. U. S. (1972) Gelbard v. U. s. (1972)20 Shadwick v. City of Tampg (1972) U. S. v. U. S. District Court (1972) Adams v. Williams (1972) 436/0307 436/0128 435/0268 434/0159 434/0106 433/0425 433/0001 431/0606 429/0413 429/0338 429/0338 428/0543 428/0543 428/0433 428/0364 427/0463 427/0038 427/0038 425/0435 423/0411 423/0411 423/0067 422/0891 422/0873 420/0103 417/0583 416/0861 416/0562 416/0562 416/0505 416/0430 416/0021 416/0021 415/0800 414/0338 415/0164 414/0260 414/0218 413/0498 413/0433 413/0266 412/0291 412/0218 411/0223 410/0019 410/0001 408/0041 408/0041 407/0345 407/0297 407/0143 056/0305 056/0168 055/0268 054/0376 054/0331 053/0867 053/0538 052/0617 050/0652 050/0530 050/0530 049/1116 049/1116 049/1046 049/1000 049/0627 049/0300 049/0300 048/0071 046/0598 046/0598 046/0209 045/0623 045/0607 043/0054 041/0325 040/0607 040/0380 040/0380 040/0341 040/0250 039/0812 039/0812 039/0771 039/0561 039/0242 038/0456 038/0427 037/0757 037/0706 037/0596 036/0900 036/0854 036/0208 035/0099 035/0067 033/0179 033/0179 032/0783 032/0752 032/0612 Williams (1972) Adams v. U. S. v. U. S. v. Biswell (1972) Harris (1971) Coolidge v. New Hampshire (1971)22 Coolidge v. New Hampshire (1971) Hill v. California (1971) U. s. v. White (1971) Whiteley v. Warden of Wyoming State Penitentiary (1971) Wyman v. James (1971) Chambers v. Maroney (1970) Vale v. Louisana (1970) U. S. v. Leeuwen (1970) Colonnade Catering Corp v. U. S. (1970) Shiplly v. California (1969) Von Cleef Chimel v. California (1969) 216 v. New Jersey (1969) Frazier v. Cupp (1969) Davis v. Mississippi (1969) New York (1969) U. 5. Kaiser v. Desist v. (1969) Spinelli v. U. s. (1969) Recznik v. City of Lorain (1968) Lee v. State of Florida (1968) Mancusi v. DeForte (1968) Sibron v. New York (1968) Sibron v. New York (1968)23 Terry v. Ohio (1968) Sabbath V. U. S. (1968) Bumper v. North Carolina (1968) Dyke v. Taylor Implement Mfg Co. Harris v. U. S. (1968) Katz v. U. S. (1967) Berger v. New York (1967) City of Seattle (1967) Camara v. Municipal Court (1967) Warden, Maryland Penitentiary v. Hayden See V. McCray v. Osborn v. (1967) Illinois (1967) Cooper v. California (1967) U. S. Hoffa v. U. S. Hoffa V. U. S. Hoffa v. U. S. Hoffa v. U. S. Lewis v. U. S. (1966) (1966) (1966)24 (1966)25 (1966)26 (1966) Schmerber v. California (1966) Riggan v. Virginia (1966) Louisana (1965) U. S. v. Ventresca (1965) Texas (1965) James v. Stanford v. Beck v. Ohio (1964) Aguilar v. Texas (1964) Clinton v. Virginia (1964) Rugendorf v. U. s. (1964) (1968) 407/0143 406/0311 403/0573 403/0443 403/0443 401/0797 401/0745 401/0560 400/0309 399/0042 399/0030 397/0249 397/0072 395/0818 395/0814 395/0752 394/0731 394/0721 394/0280 394/0244 393/0410 393/0166 392/0378 392/0364 392/0040 392/0040 392/0001 391/0585 391/0543 391/0216 390/0234 389/0347 388/0041 387/0541 387/0523 387/0294 386/0300 386/0058 385/0323 385/0293 385/0293 385/0293 385/0293 385/0206 384/0757 384/0152 382/0036 380/0102 379/0476 379/0089 378/0108 377/0158 376/0528 032/0612 032/0087 029/0723 029/0564 029/0564 028/0484 028/0453 028/0306 027/0408 026/0419 026/0409 025/0282 025/0060 023/0732 023/0728 023/0685 022/0684 022/0676 022/0274 022/0248 021/0637 021/0317 020/1166 020/1154 020/0917 020/0917 020/0889 020/0828 020/0797 020/0538 019/1067 019/0576 018/1040 018/0943 018/0930 018/0782 018/0062 017/0730 017/0394 017/0374 017/0374 017/0374 017/0374 017/0312 016/0908 016/0431 015/0030 013/0684 013/0431 013/0142 012/0723 012/0213 011/0887 217 WWkE—ifi‘ builhgicrx: am can: arm-H — Stoner v. California (1964) 376/0483 011/0856 Preston v. U. S. (1964) 376/0364 011/0777 Ker v. California (1963) 374/0023 010/0726 Lopez v. U. s. (1963; 373/0427 010/0462 Sun v. U. s. (1963)2 371/0471 009/0441 Sun V. U. S. (1963) 371/0471 009/0441 Lanza v. New York (1962) 370/0139 008/0384 Marcus v. Search Warrant (1961) 367/0717 006/1127 * 10. 11. 12. 13. 14. 15. 16. Blanks indicate volume and/or page numbers have not yet been assigned. A second case Memphis Police Department v. Garner, docket number 83-1070, combined with the lead case. There are two separate searches in this case. The first deals with the search prior to securing a warrant, and the second after the warrant was obtained. A second case Maine v. Thornton, docket number 82-1273, combined with the lead case. There are two separate searches in this case. The first deals with the search at the time of the fire, and the second concerns a search which took place several hours later. There are two separate incidents in this case. The first deals with the search, and the second with the seizure. There are two separate searches in this case. The first deals with the search of defendant's person, and the sec- ond concerns the search of the home. There are two separate searches in this case. The first deals with the third party search, and the second with the search of defendant's person. A second case Sanders v. U. S., docket number 79-148, ' combined with the lead case. There are two separate searches in this case. The first deals with the search of suspect's person, and the second concerns the initial stop. A second case, Riddick v. New York, docket number 78-5421, combined with the lead case. There are two separate searches in this case. The first deals with the searching of prison cells, and the second with the searching of body cavities. A second case, Bergna v. Stanford Daily, docket number 76-1600, combined with the lead case. There are two separate searches in this case. The first deals with the searches conducted days after the fire, the second concerns the searches which took place at the time of the fire and immediately thereafter. There are two separate searches in this case. The first deals with the search of the business premises, and the second of a car located on a public street. A second case, Sifuentes v. U. 8., docket number 75-5387, combined with the lead case. There are two separate incidents in this case. The first deals with the search, and the second concerns the seiz- ure. l7. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. r__w_ 218 There are two separate incidents in this case. The first deals with the search, and the second concerns the seiz- ure. There are two separate wiretaps in this case. The first observation deals with the first wiretap and the second observation with the second wiretap. A second case, Stark v. Shultz, docket number 72-1196, combined with the lead case. A second case, U. S. v. Egan, docket number 71-263, com— bined with the lead case. There are two separate searchs in this case. The first is the search of the car incident to the arrest and the second is the search of the person. There are two separate incidents in this case. The first concerns the evidence provided by the wife and the second the search of the car. A second case, Peters v. New York, docket number 74, com- bined with the lead case. A second case, Parks v. U. S., docket number 33, combined with the lead case. A third case, U. S. v. Cam bell, docket number 34, com- bined with the lead case. A fourth case, King v. U. S., docket number 35, combined with the lead case. There are two separate defendants in this case. The first observation concerns defendant Sun and the second observation concerns defendant Toy. APPENDIX B. Cases Cited in Text and Notes The following table is an alphabetic list of the cases cited in the text and notes. Cases preceded by an asterisk (*) belong to the data set. Case name date Citation *Almeida-Sanchez v. U.S. (1973) 413 U.S. 0266 Brown v. Board of Education (1954) 347 U.S. 0483 *California v. Carney (1985) 471 U.S. 0386 Flast v. Cohen (1968) 392 U.S. 0083 *Florida v. Royer (1983) 460 U.S. 0491 Frothingham v. Mellon (1923) 262 U.S. 0447 Mapp v. Ohio (1961) 367 U.S. 0643 Marbury v. Madison (1803) 1 Cranch 0137 O'Connor v. Ortega (1987) 480 U.S. Plessy v. Ferguson (1896) 163 U.S. 0537 Regents of the University of California v. Bakke (1978) 438 U.S. 0265 Rochin v. California (1952) 342 U.S. 0165 Roe V. Wade (1973) 410 U.S. 0113 *Segura v. U.S. (1984) 468 U.S. 0796 219 LIST OF REFRENCES LIST OF REFRENCES Agassi, ,"How are Facts Discovered?" Impulse, 1959, 3, No. 10 pp.2-4 Aldrich, J., & C. Cnudde, "Probing the Bounds of Conventional Wisdom: A Comparison of Regression, Probit and Discrimi- nant Analysis." American Journal of Political Science, 1975, 19:571-608. Aldrich, J. & F. Nelson, Linear Probability, Logit. and Pro- bit Models. Beverly Hills: Sage Publications, 1984. Alexander, Christopher, Notes on the Synthesis of Form. Cam— bridge: Harvard University Press, 1964. Allison, Graham T., The Essence of Decision. Boston: Little Brown and Company, 1971 Ashby, W. Ross, An Introduction to Cybernetics. New York: John Wiley & Sons, 1956. Ashby, W. Ross, Design for a Brain, 2d ed. rev. New York: John Wiley & Sons, 1960. Beer, Stafford, Cybernetics and Management. New York: John Wiley & Sons, 1964. Brandreth, Gyles, The Great Book of Optical Illusions. New York: Sterling Publishing Co., 1986. Brenner, Saul, Timothy M. Hagle, and Harold J. Spaeth, "Fac~ tors Affecting the Size of the Final Coalition on the Warren Court." Paper presented at the Annual Meeting of the Midwest Political Science Association. Chicago: April 1988. Brenner, Saul, and Harold J. Spaeth, "Majority Opinion Assignments and the Maintenance of the Original Coali- tion on the Warren Court." American Journal of Political Science, 1988, 32:72-81. Browning, Edgar K., and Jacquelene M. Browning, Public Finance and the Price System, 2d ed. New York: Macmil— lan, 1983. 220 221 Bruner, Jerome S., Jacqueling J. Goodnow, and George A. Aus- tin, A Study of Thinking. New York: John Wiley & Sons, 1956. Chalmers, A. F., What Is This Thing Called Science?, 2d ed. St. Lucia: University of Queensland Press, 1982. Chapanis, N. P. and J. A. Chapanis, "Cognitive Dissonance: Five Years Later." Psychological Bulletin, 1964, 60:1023. Chernev, Irving, & Kenneth Harkness, An Invitation to Chess. New York: Simon & Schuster, 1985. Cook, Thomas D., and Donald T. Campbell, Quasi- Experimentation: Design & Analysis Issues for Field Set- tings. Boston: Houghton Mifflin, 1979. Cyert, Richard, and James March, A Behavioral Theory of the Firm. Englewood Cliffs: Prentice-Hall, 1963. Ehrlich, Eugene, Amo, Amas, Amat and More. New York: Harper & Row, 1985. Evans, Bergen, Dictionary of Mythology. New York: Dell Pub- lishing Company, 1970. Eysenck, Hans J., The Psychology of Politics. London: Rout- ledge and Keqan Paul. 1954. Flew, Antony, editorial consultant, A Dictionary of Philos- ophy, rev. 2d ed. New York: St. Martin's Press, 1984. Freeman, Linton C., Elementary Applied Statistics: For Stu- dents in Behavioral Science. New York: John Wiley & Sons, 1965. Freidman, Jerome H. and Lawrence C. Rafsky, "Multivariate Generalizations of the Wald-Wolfowitz and Smirnov Two- Sample Tests." The Annals of Statistics, 1979, 7:697-717. Freund, John E., Mathematical Statistics, 2d edition. Englewood Cliffs, NJ: Prentice-Hall, 1971. Frohlich, Norman, and Joe A. Oppenheimer, Modern Political Economy. Englewood Cliffs, NJ: Prentice-Hall, 1978. Gleick, James, Chaos: Making a New Science. New York: Viking, 1987. Greenhouse, Linda, "Burger Proposes New Panel to Ease High Court's Load." The New York Times, February 7, 1983:A1.6. 222 Guralnik, David B., editor in chief, Webster's New World Dic— tionary, 2d college ed. New York: Simon and Schuster, 1982. Guttman, Louis, "The Quantification of a Class of Attributes: A Theory and Method of Scale Construction," in P. Horst pp g1., The Prediction of Personal Adjustment. New York: Bulletin No. 48, Social Science Research Council, 1941, pp. 319-348. Guttman, Louis, "The Basis for Scalogram Analysis." In S. A. Stouffer, et. al., (eds.), Measurement and Prediction. Princeton: Princeton University Press, 1950. Guttman, Louis, "The Principle Components of Scalable Atti- tudes." In P. F. Lazarsfeld (ed.), Mathematical Think- ing in the Social Sciences. New York: Free Press, 1954. Hammond, Thomas, and Jack Knott, A Zero-Base Look at Zero— Base Budgeting. City Transaction Books, 1980. Hanushek, Eric A., and John E. Jackson, Statistical Methods for Social Scientists. New York: Academic Press, 1977. Harris, Fred R., America's Democrac , 2d ed. Glenview: Scott Foresman and Company, 1983. Hartston, William R., Chess. New York: David McKay Company, 1985. Heckman, James J., "The Common Structure of Statistical Mod- els of Truncation, Sample Selection and Limited Depen- dent Variables and a Simple Estimator For Such Models." Annals of Economic and Social Measurement, 1976, 5:475-492. Hogg, Robert V., and Elliot A. Tanis, Probability and Statis— tical Inference, 2d ed. New York: MacMillan, 1983. Howard, J. W., "Query: Are Heavy Caseloads Changing the Nature of Appellate Justice." Judicature, 1982, 66:57-59,102. Kilmer, Joyce, "Trees." In Roy J. Cook, ed., One Hundred and One Famous Poems, rev. ed. Chicago: Contemporary Books, 1958. King, Gary, "How Not to Lie with Statistics: Avoiding Common Mistakes in Quantitative Political Science." American Journal of Political Science, 1986, 30:666-687. Kmenta, Jan, Elements of Econometrics. New York: Macmillan Publishing Company, 1971). -_- ..a a... _.__v.l.. ..1 . -- l. . . ‘:"-'v'~' ", .Q—V- , y 223 LaFave, Wayne R., and Jerold H. Israel, Criminal Procedure, vols. 1-3. St. Paul: West Publishing Company, 1985. LaFave, Wayne R., and Jerold H. Israel, Criminal Procedure, student edition. St. Paul: West Publishing Company, 1985. Lakatos, Imre, "Falsification and the Methodology of Scien- tific Research Programmes" in Criticism and the Growth of Knowledge, edited by Imre Lakatos and Alan Musgrave. Cambridge: Cambridge University Press, 1970. Lave, Charles A. and James G. March, An Introduction to Mod- els in the Social Sciences. New York: Harper & Row, 1975. Layer, Michael, The Politics of Private Desires. New York: Penguin Books, 1981. Lorayne, Harry, & Jerry Lucas, The Memory Book. New York: Ballantine, 1974. Lyden, Fremont, and Marc Lindenberg, Public Budgeting in Theory and Practice. New York: Longman, 1983. Lyon, Bryce, A Constitutional and Legal History of Medieval England. New York: Harper & Row, 1960. Maddala, G. S., Limited-Dependent and Qualitative Variables in Econometrics. Econometric Society Monographs No. 3. Cambridge and New York: Cambridge University Press, 1986. Marschak, J., "Elements for a Theory of Teams." Management Science, 1955, 1:127—137. McDonald, Forrest, Novus Ordo Seclorun. Lawrence: University Press of Kansas, 1985. McKelvey, Richard D., and William Zavoina, "A Statistical Model For the Analysis of Ordinal Level Dependent Vari- ables." Journal of Mathematical Sociology, 1975:103-120. "Memory." Newsweek, 9/26/86:48-54. Miller, George A., "The Magic Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Informa- tion." The Psychological Review, 1956, 63:81-97. Murphy, Walter F., and C. Herman Pritchett, Courts Judgesy_g Politics, 4th ed. New York: Randon House, 1986. Neimi, Richard, and Herbert Weisberg, "Single-Peakedness and Guttman Scales: Concept and Measurement." Public Choice, 1974, 20:33. 224 Nowak, John E., Ronald D. Rotunda, and J. Nelson Young, Con- stitutional Law, 2d ed. St. Paul: West Publishing Com— pany: 1983. Ordeshook, Peter C., Game Theory and Political Theory: An Introduction. Cambridge: Cambridge University Press, 1986. Public Broadcasting System, "This Honorable Court," 1987. Pindyck, R. S., & D. L. Rubinfeld, Econometric Models and Economic Forecasts, 2d ed. New York: McGraw-Hill, 1981. Reber, Arthur S., The Penggin Dictionary of Psychology. New York: Penguin, 1985. Rehnquist, W. H., The Supreme Court: How It Was. How It Is. New York: Morrow, 1987 Ripple, K. F., "The Supreme Court's Workload: Some Thoughts for the Practitioner." American Bar Association Journal, 1980, 66:174-176. Robinson, P. M., "On the asymptomic properties of estimators with limited dependent variables." Econometrica, 1982, 50:27-42. Rodell, Fred, Nine Men: A Political History of the Supreme Court of the United states from 1790 to 1955. New York: Random House (Vintage Books), 1955. Rodell, Fred, 55 Men: The Story of the Constitution. Harris— burg: Stackpole Books, 1986. Rohde, David W., "Policy Goals and Opinion Coaltitions in the Supreme Court." Midwest Journal of Political Science, 1972, 1972:208. Rohde, David W., and Harold J. Spaeth, Supreme Court Decision Making. San Francisco: W. H. Freeman and Company, 1976. Rokeach, Milton, Beliefs, Attitudes. and Values. San Fran- cisco: Jossey-Bass, 1968a. Rokeach, Milton, "The Nature of Attitudes." International Encyclopedia of the Social Sciences, 1968b, 1:449-457. Salmon, Wesley, The Foundations of Scientific Inference. Pittsburgh: University of Pittsburgh Press, 1967. Schattschneider, E. E., The Semisovereign People. Hinsdale: The Dryden Press, 1975. 225 Schubert, Glendon A., "The Study of Judicial Decision Making As an Aspect of Political Behavior." American Political Science Review, 1958, 52:1007—1025. Schubert, Glendon A., Quantitative Analysis of Judicial Behavior. Glencoe: Free Press, 1959. Schubert, Glendon A., The Judicial Mind. New York: Oxford University Press, 1965. Schubert, Glendon A., The Judicial Mind Revisited. New York: Oxford University Press, 1974. Schwartz, Bernard, Super Chief: Earl Warren and His Supreme Court--A Judicial Biography, unabridged edition. New York: New York University Press, 1983. Segal, J. A., Predicting Supreme Court Decisions Probabilis- tically. Unpublished Ph.D. dissertation, Michigan State University, 1983. Segal, J. A., "Predicting Supreme Court Decisions Probabilis- tically: The Search and Seizure Cases, 1962-1981." Amer- ican Political Science Review, 1984, 78:891-900. Segal, J. A., "Measuring Change on the Supreme Court: Examin— ing Alternative Models." American Journal of Political Science, 1985, 29:461-479. Segal, J. A., "Supreme Court Justices as Human Decision Mak- ers: An Individual Level Analysis of the Search and Seizure Cases." Journal of Politics, 1986, 48:938—955. Simon, H. A., Administrative Behavior. New York: Macmillan Company, 1955 . Simon, H. A., Models of Man. New York: John Wiley & Sons, 1957. Simon, H. A., "Theories of Decision-Making in Economics and Behavioral Science." American Economic Review, 1959, 49:253-283. Simon, H. A., "Theories of Bounded Rationality," in Decision and Or anization, edited by C. B. Radner and R. Radner. Amsterdam: North-Holland Publishing Company, 1972:161-176. Also in Models of Bounded Rationality: Behavioral Economics and Business Organization, vol. 2, H. A. Simon. Cambridge, MA: The MIT Press, 408-423. Simon, H. A., Administrative Behavior, 3d edition. New York: Free Press, 1976. Simon, H. A., Models of Thought. New Haven: Yale University Press, 1979. 226 Simon, H. A., The Sciences of the Artificial, 2d edition. Cambridge: Massachusetts Institute of Technology Press, 1981. Spaeth, Harold J., Supreme Court Policy Making. San Fran- cisco: W. H. Freeman, 1979. Statsky, William P., and R. John Wernet, Jr., Case Analysis and Fundamentals of Legal Writing. St. Paul:West Pub— lishing Company, 1977. Steinbruner, J. D.,The Cybernetic Theory of Decision. Prince- ton: Princeton University Press, 1974. Supreme Court of the United States, The Supreme Court of the United States. Washington, D.C.: Supreme Court of the United States and the Supreme Court Historical Society, 1982+. (An informational phamplet available to visitors of the Supreme Court Building; no publishing date given, but references are made to the end of the 1981-1982 term.) Tanenhaus, Joseph, Marvin Schick, Matthew Muraskin, and Daniel Rosen, "The Supreme Court's Certiorari Jurisdic- tion: Cue Theory" in Schubert, Glendon, ed., Judicial Decision Making. New York: Macmillan, 1963. Teger, Stuart, "Presidential Strategy for the Appointment of Supreme Court Justices." Public Choice, 1977, 31:1. Teger, Stuart H., and Douglas Kosinski, "The Cue Theory of Supreme Court Certiorari Jurisdiction: A Reconsider— ation." Journal of Politics. 1980, 42:833—846. Tribe, Laurence H., God Save This Honorable Court. New York: Random House, 1985. Tversky, A. and D. Kahneman, "Judgment under uncertainty: Heuristics and biases," in Judgment Under Undercer- tainty: Heuristics and Biases, edited by D. Kahneman, P. Slovic, and A. Tversky. Cambridge and New York: Cam— bridge University Press, 1982. Ulmer, s. Sidney, William Hintze, and Louise Kirklosky, "The Decision to Grant or Deny Certiorari: Further Consider- ation of Cue Theory." Law and Societ , 1972, 6:637—643. Von Neumann, John and Oskar Morgenstern, Theory of Games and Economic Behavior, 3d ed. Princeton: Princeton Univer- sity Press, 1980. Wald, A. and J. Wolfowitz, "On a Test Whether Two Samples Are From the Same Population." Annals of Mathematical Stat— istics, 1940, 11:147-162. 227 Weisberg, Herbert F., "Evaluating Theories of Congressional Roll-Call Voting." American Journal of Political Science, 1978, 22:554-577. Wieland, Christoph Martin, Musarion, canto II, 1768. Wildavsky, Aaron, The Politics of the Budgetary Process, 4th ed. Boston: Little Brown and Company, 1984. Woodward, B., and S. Armstrong, The Brethren. New York: Simon and Schuster, 1979. Yntema, Douwe B., and Gayle E. Mueser, "Remembering the Pre- sent States of a Number of Variables." Journal of Exper— imental Psychology, 1960, 60:18-22. 1).-.11.. » it I] I \ lllllil‘II ||||l ‘1 1 WIllllllHllllllllWWIIHIIJWIIHUlilNWlHH/IW