'bv1531—J RETURNING MATERIALS: Place in book drop to LJBRAfiJES remove this checkout from “ your record. FINES win be charged if book is returned after the date stamped below. . A“? If? 615-?tv .4 : ".r’ "iii aft)“- - "JV” PREDICTING SUPREME COURT CASES PROBABILISTICALLY: THE SEARCH AND SEIZURE CASES. 1962-1981 BY Jeffrey Allan Segal A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science Copyright by JEFFREY ALLAN SEGAL 1983 ABSTRACT PREDICTING SUPREME COURT CASES PROBABILISTICALLY: THE SEARCH AND SEIZURE CASES. 1962-1981 BY Jeffrey Allan Segal The Fourth Amendment, which protects against unreason- able searches and seizures, is "one of the laws most vital to our liberty“ (Amsterdam. 1974:377). Yet we are still not sure what makes a particular action by the government an un- reasonable search or seizure. According to Dworkin (1973). the search and seizure cases are “a mess" (p. 329). It was felt that our incomprehension of the Fourth Amendment cases was not a result of the decisions them- selves. but the manner in which the decisions were studied. Previous quantitative attempts to explain court decisions lacked any theoretical base and used methods that were rather inappropriate. Qualitative attempts to explain court decisions place an overreliance on precedent, and either have too many exceptions or unwieldy classifications. In order to attack this problem a theoretical model of Supreme Court decision making was developed that assumes members of the Court are satisficers who monitor certain facts from the lower court's Opinion. These facts strongly predispose the members towards finding a search reasonable or unreasonable. The model was tested using the facts of the case. such as whether there was a warrant or probable cause, as the independent variables. The decisions of the Court and the individual justices as to the reasonableness of the search or seizure were the dependent variables. The data were the Supreme Court's search and seizure decisions from the Fall of 1962 until the Spring of 1981. The parame- ters were estimated by the multivariate technique probit. The model predicted 76% of the Court's decisons cor- rectly and explained 57% of the variance in those decisions. The point estimates were almost universally in the right direction. of the expected relative magnitude and statisti- cally significant. The place of the search, the extent of the intrusion, the existence of a warrant, a few mitigating circumstances, the presence of the United States as a party and the particular makeup of the Court were all found to be important variables. The results from the justices were fairly consistent with those of the Court itself, but many of the estimates failed to attain statistical significance. In memory of my Father ACKNOWLEDGEMENTS I would like to thank John Aldrich. Daniel McGoldrick, Gary Miller and Charles Ostrom for their help on this pro- ject. Special thanks go to my dissertation chairperson, Harold Spaeth. Harriet Dhanak of the Politimetrics Laboratory at Michigan State University and Roger Pijacki of the Social Science Data Lab at SUNY at Stony Brook were more than generous in helping me with the computer work. Emotional support was supplied in abundance by Susan Herbel, Daniel McGoldrick and Genny Shields. TABLE OF CONTENTS 1. Introduction The Literature 2. Search and Seizure 3. Supreme Court Decision Making The Nature of the Choice The Process Specification 4. Methodology The Data 5. Results Operationalization Findings Discussion The Justices 6. Summary and Conclusions Policy Implications Limitations Suggestions for Future Research Bibliography 16 41 50 52 57 66 84 86 86 90 105 108 125 130 131 136 139 LIST OF TABLES Values Along the Cumulative Normal Probability of Voting for McGovern. 1972 Probabilities of Finding Searches Reasonable With and Without Warrants Probit Estimates for Search and Seizure Decisions Probit Estimates with United States as Party Probit Estimates for Individual Justices O 78 79 82 91 103 108 LI ST OF FI GURES 3.1 A Justice's Utility for Freedom and Order 4.1 The Linear Probability Model 4.2 A Comparison of Linear and "S" Shaped Models 4.3 The Cumulative Probability Function 54 69 71 74 CHAPTER 1 INTRODUCTION The right of the pe0p1e to be secure in their persons, houses, papers, and effects, against un- reasonable searches and seizures, shall not be violated. and no warrants shall issue. but on probable cause, supported by oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized. The above declaration was deemed important enough by the nation‘s founding fathers to be included as the Fourth Amendment to our Constitution. The desire for its protec- tion lies deep in both English and American democratic his- tory. Few are not aware of William Pitt's assertion that: ”The poorest man in his cottage may bid defiance to all the force of the Crown. It may be frail-- its roof may shake--the wind may blow through it- -the storm may enter--but the King of England can- not enter: all his force dares not cross the thre- shold of the ruined tenement” (Landynski. 1966:23). Less well remembered is John Adams' recollection of the stirring speech James Otis. Jr. had given against the writ of assistance, the colonial version of England's general warrant. The speech. Adams wrote, was the first in "the Chain of Causes and Effects" leading to independence (Landynski, p. 37). Today. the Fourth Amendment is ”one of the laws most vital to our liberty“ (Amsterdam, 1974:377). Nevertheless, we are still not sure what makes a parti- cular action by the government an unreasonable ”search or seizure" within the meaning of the Fourth Amendment. According to Justice Frankfurter. ”the course of true law ‘I pertaining to searches and seizures . . . has not run smooth“ (Amsterdam, p. 349). Justice White. joined by Justice Burger, claims that the Court has ceased even “to strive for clarity and consistency of analysis.” Coolidge v.New Hampshire, 403 0.8. 443 (1971). Scholarly opinion agrees. Professor LaFave has observed that ”no area of law has more bedeviled the judiciary" (in Amsterdam. p. 349). Dworkin (1973) simply states that the Fourth Amendment cases are “a mess" (p. 329). Amsterdam (1974) considers all of these comments to be understatements. The question of whether a search or seizure violates the Fourth Amendment becomes even more important when one considers what happens to unconstitutionally obtained evidence. According to the Supreme Court, it may not be presented as evidence at a trial. This exclusionary rule. as it is called, means, in the immortal words of Judge Cardozo, that "the criminal is to go free because the const- able has blundered." People 3; Defore.242 N.Y. 13 (1926). If the Fourth Amendment seems incomprehensible to those who study the Supreme Court, perhaps the fault lies not in the court decisions but in the method chosen to study those decisions. The traditional method has been the case method where decisions are examined one at a time. Idiosyncrasies are emphasized: common ground is played down. However, with the coming of the behavioral revolution in public law, many have attempted to use quantitive analysis to explain deci- sions of the Supreme Court. As will be discussed later. most of those attempts leave tremendous room for theoreti- cal. methodological. and empirical improvements. This dissertation has several goals. but the 'ultimate is to contest the conventional wisdom in the Supreme Court's search and seizure decisions. It is my belief that these decisions can be shown to make sense if only studied in the proper framework. In order to do this a model of Supreme Court decision making will be developed. The model will then be applied to the Court's search and seizure decisions using the multivariate estimation procedure probit. The data will be derived from the Supreme Court's search and seizure decisions from 1962 until 1981. The decisions of the Court and the individual justices as to whether or not a search violates the Fourth Amendment will be the dependent variables. The facts of the case, such as the existence of a warrant, probable cause or a lawful arrest will be the independent variables. It is felt that when a strong theoretical understanding as to what the court has said on search and seizure and how judges decide cases is combined with a methodology that is more appropriate than what has heretofore been used in this realm, order will be found where until now ladk of theory and inappropriate methodolo- gies have found only chaos. Besides the substantive improvements it is hOped that important theoretical and methodological benefits will fol- low from this work. Theoretically, I hOpe to develop a mod- el that is useful for understanding Supreme Court decision 4 making. This will be done not only because such a model is necessary before the search and seizure decisions can be ex- plained, but because such a model should prove useful in other realms of judicial decision making as well. Methodologically, I hOpe to show probit to be an extremely valuable tool in explaining and predicting Supreme Court de- cisions. As with the theoretical section, what is useful in search and seizure should be generalizeable to other subs- tantive areas of judicial behavior. This is of great signi- ficance since, to my belief (and to be discussed in greater detail later), there currently exist no satisfactory objec- tive, quantitative techniques for predicting judicial deci- sions. Previous attempts that employ scaling (Schubert 1963; Rohde and Spaeth 1976; Spaeth. 1979) have been suc- cessful but in the final analysis require a subjective judgement as to where the Court will "split." The more ”objective" attempts (Kort, 1957; Kort, 1963; Schubert, 1959: and Ulmer, 1964) as will be discussed below, are all to one degree or another seriously flawed. This concern with prediction is not only of practical concern to the attorney and the citizen but is perhaps the subject of highest concern to the scholar of jurisprudence. According to Holmes, “The prOphecies of what the courts will do in fact and nothing more pretentious are what I mean by law (in Murphy and Pritchett, 1979:21). The law is not the black letter written by our legislative branch. These stat- utes must be interpreted by judges. The determination as to whether what we do today is legal will be made by judges tomorrow. “The object of our study, then, is prediction” (Holmes, in Murphy and Pritchett, 1979:19-20). The following, then, represents how this dissertation will proceed. The remainder of this chapter will review and characterize the literature on the quantitative attempts to predict Supreme Court decisions. Chapter II will discuss our current qualitative understanding of search and seizure by examining its history and analyzing the more important judicial decisions. Chapter III will provide a theoretical model of Supreme Court decision making. Specification of the variables to be used in the model will be included. Probit will be discussed in the fourth chapter, and the em- pirical results will be presented in Chapter V. Chapter VI. will contain conclusions and suggested areas for future re- search. The Literature Logically, there are three distinct types of explana- tion concerning Supreme Court decision making. First, we can try to explain differences (variance) in the actual de- cisions of the Supreme Court. That is, we can attempt to determine why the Court voted one way in one case and anoth- er way in another. The type of case and the particular ideological makeup of the Court are among the more obvious independent variables that would be used. Second, we could try to explain the variance in any particular justice's votes. Again, the type of case would be important. Third, we could lock at variance between justices: why Justice A votes liberal and Justice B votes conservative. Independent variables in this type of study would fall under the rubric “background characteristics.” One of the first attempts to mathematically explain de- cisions of the Court as a whole was Kort's (1957) investiga- tion of the “right to counsel" cases. From the 1932 deci- sion in Powell v. Alabama, 287 U.S. 45, until the historic Gideon v. Wainright, 337 U.S. 335 (1963), the Supreme Court occupied itself with a case by case examination of whether the Sixth Amendment's "right to counsel“ clause, made appli- cable to the states through the Fourteenth Amendement, was violated by state refusal to provide legal assistance to ac- cused felons. Kort devised a mathematical formula vaguely resembling discriminant analysis for predicting Supreme Court decisions. The independent variables were those "factual elements which have been emphasized by the justices in their Opinions and on their votes to affirm or set aside convictions" (p. l). Twenty-six variables were applied to fourteen cases. The process involved dividing the popula- tion of cases into two sets, determining weights for vari- ables in the first set (the source cases) according to Kort's self-invented algorithm, and then applying those weights to the second set (the test cases). "Pro" (requiring an attorney) and ”con“ (not requiring an attor- ney) cases were successfully discriminated in twelve of the fourteen cases. The seeming success of Kort's methodology led to a number of related works. Schubert (1959) attempted a rea- sonably precise replication of Kort using the Supreme Court's search and seizure cases from 1937 to 1957. Twenty variables were used on thirteen source cases, ten federal test cases, and six state test cases. The variables were classified as to the manner of the search (physical or wire- tap), the scope of the search (where the search took place), the justification for the search (warrant and probable cause requirements), the object of the seizure (speech, property or contraband) and the use made of the objects seized (direct evidence or as leads). The model predicted the ten federal test cases perfectly but failed in five of the six state test cases. Kort (1963), responding to harsh criticism from Fisher (1958), made two significant changes in a new model. First, he employed factor analysis as a means of keeping the number of variables less than the number of cases. Second, he used ordinary least squares (OLS) to determine factor weights. High prediction rates were obtained for involuntary confes- sion cases. Ulmer (1964) devised an admittedly experimental model, again resembling in function discriminant analysis, to further examine the search and seizure cases. Twenty vari- ables applied to nineteen cases yiefded perfect prediction. Four of these were chosen for additional analysis. (The method of selection was not discussed.) These four variables (search of a living quarters, search in the absence of an arrest warrant, search involving state officers and search after forced entry) were used to correctly discriminate eighteen of the nineteen cases. Kort (1968) tested a nonlinear model of judicial deci- sion making. After applying OLS to 128 Connecticut workmen‘s compensation cases, he added, one at a time, every possible pairwise multiplicative variable. Each new model was compared to the purely linear model to judge whether a significantly greater amount of variance was explained. Since there were twenty-two original variables, 253 nonlin- ear models were tested. Two combinations were found to add significantly to the model. Despite the fact that by chance alone we would have found more than two significant combina- tions, Kort concludes by supporting the notion of multipli- cative relationships between variables. Finally, Kort (1973) observed and demonstrated that techniques such as discriminant analysis, which take into account the dichotomous nature of the Supreme Court's deci- sions, are more appropriate than linear regression tech- niques. Unfortunately, the above mentioned studies that attempt to explain and predict Supreme Court decisions are to one extent or another seriously flawed. First, not one of the studies presents any sort of theory of Supreme Court deci- sion making, nor do any of them posit theoretical justifica- tion for expecting the Court to behave as it has. Indeed, given the assumptions that are made, this would be quite difficult to do. We find, for example, that each scholar assumes the Court to be a static institution; changes in membership are deemed to have no effect on the Court's deci- sions. If the importance of new members needs defense, let it be remembered that the Warren Court's decisions in 5222. v. Ohio, 367 U.S. 643 (1961), Gideon v. Wainright, 372 U.S. 335 (1963) and Miranda v. Arizona, 384 U.S. 436 (1966) (1966) would render the static models of Schubert (1959), Kort (1957) and Kort (1963) useless. Second, the Kort (1957) and Schubert (1959) studies in- volve more variables than cases. This makes it nearly im- possible not to predict perfectly, and renders variable weights meaningless in that there are literally an infinite number of values that can predict perfectly. It is thus im- possible to gauge the relative importance of variables. Third, in all but the latter Kort studies, the methodo-l logy imposed is random. That is, ad hoc formulae are deve- loped to discriminate between “pro" and "con“ cases. The algorithms do not minimize error in terms of sums of squares, do not maximize percent predicted correctly or meet any other established criterion. Finally, variables are used whose values can only be determined by reading the opinion of the Court. For exam- ple, Schubert (1959) used as an independent variable whether or not the Supreme Court finds probable cause. Even if we ignore the practical difficulty that we can never predict 10 based on his model (the values of the independent variables are not known until the same day that the value of the de- pendent variable is known), we are left nonetheless with the theoretical problem that there is no way of knowing whether a justice's decision on probable cause precedes, as a cause, or succeeds, as a justification, the decision on the consti- tutionality of a search. The cause must come before the ef- fect, and this we do not know. Kort (1963), it should be noted, recognized this difficulty and later developed a mod- el using discriminant analysis as a means of predicting the Court's evaluation of subjective variables (1973). Although the above techniques could have been easily applied to the study of individual justices, this has not been done. Much of the work that has attempted to explain and predict the decisions of individual justices falls into two distinct categories, scaling analysis and bivariate ta- ble analysis. Let us first consider scaling analysis. Quantitative prediction of the decisions of individual justices has been left mainly to those who employ the meth- ods of scaling analysis. Originally developed during World War II to study the attitudes of American soldiers (Stouffer et al., 1950), scaling became one of the major tools in the investigation of judicial attitudes (Schubert, 1963b). The initial purpose of scaling analysis is to determine whether the votes cast by judges can be arranged in an ordinal rela- tionship (Tanenhaus, 1966). Consider a hypothetical series of cases on the First Amendment. How can we know if one 11 judge is more supportive of free speech rights than another? If n votes are taken, and Judge A supports free speech in x of these cases while Judge B does so in (x + i),where i is greater than zero, we can not at this point be sure that Judge B is more supportive since the cases in which Judge A supported free speech may have been more important than the cases in which Judge B supported free speech. However, if Judge 8 supported free speech in all x of the cases in which Judge A did, plus i more, we.are confident that Judge 8 is more supportive of free speech than Judge A. If a series of cases meets this type of criterion for a series of judges, the probability that the cases are unidimensional is greatly enhanced (Spaeth and Peterson, 1971). The knowledge provided by these scales has been used by Schubert (1959), Rohde and Spaeth (1976) and Spaeth (1979) to predict the votes of individual Supreme Court justices. Rohde and Spaeth report accurate predictions in 86% of the votes analyzed, while Spaeth (1979) reports 85% accuracy for the justices and 88% for the Court as a whole. As impres- sive as these results are, they are not ”objective." Scaling provides support for assumptions of unidimensionality and ordinality; it would lead us to believe, for example, that if Rehnquist voted liberally on a search and seizure case so would Stevens, Douglas and Brennan. It can not tell us how Stevens will vote if Rehnquist votes conservatively, nor can it tell us how Rehnquist will vote. A subjective judgement, however informed it might be, must be made as to how 12 "severe" a case is and where it will split. Herein lie the limits Of scaling: two people with the same data can predict different results. Although many judicial behavioralists concerned them- selves with different applications Of scaling (Schubert, 1963a), many still followed the more traditional behavioral- ist modes of inquiry, such as bivariate analysis. Two subs- tantive areas that might be of concern to this study are the responsiveness of the Court to the United States as a liti- gant and the relationship between votes on certiorari and the votes on the merits. Much has been written claiming that the Supreme Court's responsiveness to the United States as a party is a vital factor in the Court's decision to grant certiorari (i.e., Ulmer et al., 1972). Does this favorable predisposition carry over to final decisions? Analysis by Spaeth and Teger (1982) shows the Court supporting the government 69.7% of the time in the 1966-1967 terms and 67.8% of the time in the 1969-1973 terms. Throughout these terms, Douglas was the only justice to support the federal government less than half the time. The general support trend holds for criminal procedure cases. These and other studies showing support for the federal government as Opposed to its adversaries are enlightening and might lead us to believe that in our search and seizure cases the Court may, on the whole, be predi- sposed to favor the government. But, before we can use this information to help us explain specific Court decisions, we 13 must have a point Of comparison. In our pOpulation of search and seizure cases, those cases that do not involve the federal government do involve the state government. Our question is not whether the Supreme Court is more favorable to the U.S. than its adversaries: rather, we are concerned with whether the Supreme Court is more favorable to the U.S. than it is to the states. That is, overall, the United States is more likely than not to win. What is of interest is whether the United States is more likely to win than are the states. The traditional response, based on the notion Of feder- alism, is that the Court should show more leeway toward the states. Harlan, for example, argued that the “fundamental fairness" doctrine supported more leniency for the states as to searches and seizures. His eight brethren, on the other hand, supported equal treatment. Nevertheless, Spaeth and Teger (1982) found that across all issues, each and every justice from the Burger Court (1969-1977) was more suppor- tive of the federal government than the states. In search and seizure cases, seven justice and the Court as a whole were more supportive of the U.S., three were more supportive of the states, and one was equally supportive of both. What else might have a zero-order association with the decisions of judges? Baum (1977) states that the decision to grant or deny hearing does. His article suggests an "error correcting“ strategy, limited by salience and the de- sire tO limit workload. That is, a judge will grant hearing 14 when the lower court's decision is sufficiently far from‘his own preference. If correct, Baum's model would lead us to predict, first, that a judge who votes for hearing feels that the lower court's decision was not his preference. Such a judge will vote to reverse. Second, we would predict that a judge who votes against hearing either agrees with the lower court decision or disagrees with it, but prefers not to take the time to hear the case. Such a judge may or may not vote to reverse the decision if and when it is for- mally decided. These predictions are given some support by Ulmer (1972), who found association among justices between their votes to grant certiorari and their votes on the mer- its. As far as decisions of the Court itself are concerned, if the full Court is voting, a minimum of four votes are needed in order to hear a case. Thus, most cases heard have at least four justices that arguably are predisposed to re- verse the lower court. We might therefore want to extend the above hypothesis with the notion that the Court as an institution hears those cases it wishes to reverse. A zero-order association would be expected. Both the findings on the United States as a litigant and the relationship between votes on certiorari and the merits are interesting whether or not they stand after other factors are controlled. Nonetheless, we must control for those other factors before we can truly understand the na- ture of these relationships. Finally, as was said earlier, variance between judges 15 can be explained. Nagel (1961) showed a bivariate relation- ship between political party affiliation and judicial decision-making. Ulmer (1973) attempted a regression analy- sis using twelve background variables On fourteen justices to determine differences between the brethren. Tate (1981) introduced thirty-seven variables into a stepwise regression with twenty-five justices and claims to have found “impressive" results. These studies are presented here merely as a major example of quantitative explanation Of ju— dicial behavior. They can not, however, be used to explain or predict variance in the decisions of individual justices. As such, they are beyond the scOpe of this inquiry. What this literature review has meant to show is that there exists great room for imprOVement in terms of explana- tion and prediction of Supreme Court decisions. Scaling analysis has worked but is too reliant on subjective judge- ment. Bivariate analysis has obvious problems of control. The multivariate studies are, to be blunt, failures. This failure, I believe, is not inherently flawed in its concept, only in its practice. How this can be corrected will be discussed in Chapter IV. Currently, we must examine the substantive area of this study, search and seizure. CHAPTER 2 SEARCH AND SEIZURE The purpose of this chapter is to examine what can be considered the conventional wisdom on search and seizure de- cisions. This will be done in order to bring to the reader the substantive knowledge necessary to fully appreciate the empirical results of Chapter V, and to serve as a point of comparison when examining the utility of quantitative analy- sis Of judicial behavior. This discussion will be prefaced by an introduction as to what are the traditional means Of studying court decisions. In his seminal exposition, The Nature of the Judicial Process (1921), Benjamin Cardozo asks Of himself and of oth- er jurists, “What is it that I do when I decide a case? To what sources of information do I appeal for guidance?”(p.10). The everyday working rule, he replies, is stare decisis, the following of previous decisions. "Adherence to precedence," he exclaims, "should be the rule and not the exception"(p.149). However, the following of precedent is: A “a process of search, comparison and little more. Some judges seldom get beyond that process in any case. Their notion of their duty is to match the colors of the case at hand against the colors of many sample cases spread out upon their desk. The sample nearest in shade supplies the applicable rule. But, Of course, no system of living law can be evolved by such a process, and no judge of a high court, worthy of his Office, views the func- tion of his place so narrowly. If that were all 16 17 there was to our calling, there would be little of intellectual interest about it. The man who had the best index of the cases would also be the wis- est judge. It is when the colors do not match, when the references in the index fail, when there is no decisive precedent, that the serious busi- ness Of the judge begins"(PP.l9-20). When precedent does not provide a just answer the judge, according to Cardozo, will turn to history as well as tradition and social welfare. So in our explanation of the conventional understanding Of search and seizure we must therefore examine not only the previous judicial decisions in this realm but also its historical background. Behavioralists may argue that the methods of precedent and history are merely justifications for decisions, not causes Of them. This will be addressed later. For now, we must see how much can be learned from this approach. English history of search and seizure dates back at. least as far as 1335, when the crown allowed innkeepers to search guests for imported false money. Henry VI (1422- 1461) followed by granting the Dyers Company of London the privilege of searching and seizing cloth dyed with logwood. This was in turn followed by permissible searches for Oil (Lasson, 1937). Broad search and seizure powers were first granted dur- ing the Tudor period as a means of regulating the press (Siebert, 1952). The Stationers Company, a private monopoly guild, was empowered in 1557 "to search whenever it shall please them in any place, shop, house, chamber, or building Of any printer, binder or bookseller whatever within our 18 kingdom of England..." (Siebert, p. 82). The Star Chamber confirmed this power in 1586 and added to the places to be searched anyplace where “they shall have reasonable cause of suspicion" (Siebert, p. 84). The Printing Act of 1662, Parliament's first endeavor in this field, granted "the authority of the principal secretary to issue general and special warrants for searches and seizures” (Siebert, p. 253). This was, though, the height of England's authority to intrude upon the privacy of its citizens. In 1665, Chief Judge Scroggs was impeached by Parliament for issuing gener- al warrants. Judicial sentiment was likewise growing against broad search powers. The jurist Sir Matthew Hale declared general warrants void despite their precedent under common law (Landynski, 1966). Specific warrants required "judicial approval, and were not,to be issued unless they followed an examination of the complainant under oath and a finding of probable cause" (Landynski, p. 27). In 1762 M. P. John Wilkes was awarded three hundred pounds damage after his seizure under a general warrant. For the first time a sanction had been judicially imposed against an illegal search or seizure. A similar sanction was imposed in Entwick !;’ Carrington. These decisions, according to Landynski, spurred Parliament to greater action against the general warrants. In 1694 the House of Commons vetoed continuation of the Printing Act. Warrants were still issued but greater parti- 19 cularization was required. In the eighteenth century, searches and seizures were used more for tax collection and less for press control. Not surprisingly, this led to even greater popular resentment than before. William of Orange abolished a tax because its search requirements constituted “a badge of slavery upon the whole people“ (Landynski, p. 25). In 1733, a tobacco and wine tax failed because it al- lowed general warrants for the search of warehouses. Thirty years later, William Pitt gave his previously quoted speech against the enforcement provisions of the cider tax. In 1766 Parliament declared general warrants illegal. The colonies were likewise plagued by general warrants. The original purpose was the enforcement of the Molasses Act of 1733. The warrant used was called a ”writ of assis- tance," but it was actually even more dangerous to liberty than the general warrant for it was not returnable after execution. Rather, it was good through the lifetime of the reigning sovereign. The discretion delegated was absolute and unlimited (Lasson, p. 264). In February 1761, six months after the death of George 11, all writs expired. Arguments ensued as to the legal authority for the issuance of new writs. James Otis Jr., arguing in front of the Massachussetts Superior Court, as- serted that the writs were repugnant to Magna Carta and thus void. The court ruled against Otis but the people responded by electing him to the colonial legislature. The legisla- ture in turn lowered the salaries of the Superior Court. 20 Specific attempts by the Crown at enforcing the writs led to armed resistance. The struggle continued through the Stamp Act, the Townsend Act and through the revolution. The revolutionary legislatures wasted no time in redressing this particular set of grievances. The Virginia Constitution had denounced general warrants as "grievous and oppressive" even prior to July 1776. Before 1776 ended Pennsylvania, Maryland and North Carolina adopted constitu- tional safeguards regulating searches. Maryland instituted an oath or affirmation requirement while Pennsylvania and Maryland prevented all unreasonable searches, warranted or otherwise. Massachussets in 1780 and New Hampshire in 1784 first used the actual language "unreasonable searches." The Constitutional Convention of 1787, however, did not include a Bill of Rights. Hostility from the people, the states and leaders like Patrick Henry led to the Bill Of Rights adOp- tion. Included was a provision drafted by Madison and re- written in committee that read: The right Of the people to be secure in their persons, houses, papers and effects against unrea- sonable searches and seizures shall not be vio- 1ated by warrants issuing without probable cause, supported by oath or affirmation, and not particu- larly describing the place to be searched and the persons or things to be seized" (Landynski, 1966: 41). Benson of New York proposed strengthening the amendment by replacing "by warrants issuing" with "and no warrant shall issue.“ This motion, though, was soundly defeated! (Landynski, 1966:42). According to Lasson (1937: 101-103), Benson as committee chairman simply reported his own ver- sion, and no one noticed. The amendment as passed, then, contains two parts. The first is a protection against un- reasonable searches and seizures, while the second specifies the condition under which a warrant may be specified. Three possible interpretations result (Landynski, 1966: 42-43). First, reasonable searches are those that meet the warrant requirements of the second clause. Second, the first clause places added restrictions on the second: certain searches that meet the warrant requirements can still be unreason- able. Third, the first clause is an exception to the se- cond: searches that do no meet the warrant requirements may nonetheless be constitutional if they are reasonable. All three positions have been arqued by different justices. Reguardless, a constitutional protection against unrea- sonable searches and seizures does exist. Unfortunately, the enumeration of a constitutional right does not guarantee the realization of that right. An enforcement mechanism is needed. Consider the First Amendment's declaration that Congress shall pass no law abridging freedom of speech. How are we protected if Congress does pass a law abridging freedom of speech? The answer is judicial review, the power of the Supreme Court to declare laws repugnant~ to the Constitution unconstitutional and thus void. Thus, if I am convicted of violating a law that prohibits free speech the Supreme Court will, in theory, overturn the conviction and declare the law unconstitutional. This same sort of protec- 21 22 tion can and has been used to protect citizens from inva- sions of Fourth Amendment rights. One can consider, for ex- ample, Marshall v. Barlow‘s, 56 L.Ed. 2d 305 (1978), in which laws that sanctioned warrantless searches by OSHA were declared unconstitutional. On the other hand, most searches that violate the Fourth Amendment occur without any law permitting them. They usually occur by police Officers in the field as part of their day-tO-day investigative routines. Again, no law permitting searches is necessary before a search may take place. As such, the power Of the Supreme Court to declare laws void is in and of itself ineffective in fighting uncon- stitutional intrusions into our privacy; if the passage of a law is not necessary in order to violate the constitution, prohibition of such laws Obviously cannot prevent its viola- tion. For many years this left the guarantees Of the Fourth Amendment an empty promise. It was not until 2219 v. United States, 116 U.S. 616 (1886), that the Supreme Court breathed life into the Fourth Amendment. The case revolved around the seizure of thirty- five cases Of plate glass, alleged to have been fradulently brought into this country in civil violation of the Revenue Acts of 1863 and 1874. In an attempt to determine the value of the thirty-five cases, Edward and George Boyd were or- dered to produce the invoices Of twenty-nine cases previous- ly imported. The act stated that in any noncriminal pro- ceedings under the act, "the attorney representing the go- 23 vernment, whenever in his belief, any business book, invoice or paper belonging to...the defendant will tend to prove any allegation... the court...may, at its discretion, issue a notice to the defendant...to produce such book, invoice or paper” (116 U.S. 616, 620). NO more than the belief Of the attorney was required. Probable cause is never mentioned. Further, if the defendant refuses to produce such papers “the allegations stated in the said motion shall be taken as confessed" (116 U.S. 616, 620). The Boyds produced the pap- ers but protested the order as violating their Fourth and Fifth Amendment rights. As there was little in the way of precedent, Justice Bradley, for the majority, turned heavily to history to ex- plain his decision. First, he declared that the Act of 1863 was the first in either the U.S. or England which authorized the search and seizure of a man's private papers. ”Even the Act under which the Obnoxious writs of assistance were issued did not go as far as this,“ stated Bradley (116 U.S. 616, 623). He went on to explain that there has always been a legal difference between stolen goods, forfeitable goods or contraband, on the one hand, and papers or mere evidence on the other. Seizures of the former had long been allowed under common law and statute law. Seizures of the latter had never been permissible. In determining whether the search in question might nonetheless be considered reasonable within the founder's meaning of the Fourth Amendment, Bradley turned to ”the con- JIYHM“—‘fl -..__._ 24 temporary or then recent history of the controversies of the subject, both in this country and in England." (116 U.S. 616, 624-5). After reviewing the. speeches of Otis and Adams, Bradley deemed it obvious that the Fourth Amendment was meant to solve the problems in the United States that Entwick v. Carrington had done in England. Finding little difference between searching for papers and compelling their production, Bradley ruled that the papers must be returned to the Boyds. Any future trial would have to proceed with- out this evidence. We thus see in this case the historical respect given papers and books, historical evidence on the conditions that led to the adoption of the Fourth Amendment, and even the use of English precedent. The Boyd case left unclear whether all unreasonably Ob- tained evidence must be excluded from use at a criminal trial. In the 1914 decision, Weeks v. United States, 232 U.S. 383, the Supreme COurt unanimously ruled that exclusion was constitutionally required. The case involved the use of the U.S. mails for the purpose of conducting a lottery. Twice, the defendant's house was searched, resulting in the confiscation of certain papers and articles. In neither in- stance was a warrant procured. Weeks urged that the papers be returned before trial but his timely request was denied. Some of the papers were used against him. Justice Day, for the Court, spOke at great length about how the Fourth Amendment was a reaction against the writs of assistance, and that it had been placed in the Constitution 25 to uphold the maxim that "every man's house is his castle" (34 S. Ct. 341, 343). Again, the distinction was also drawn between stolen goods and mere evidence. SO, the search, considering both precedent and history, was unreasonable. Day then discussed what was to be done. ”The tendency of those who execute the criminal laws“ he stated "to obtain convictions by means of unlawful searches and seizures...should find no sanction in the judgements of the courts" (34 S. Ct. 341, 344). "If letters and private documents can thus be seized and held and used in evidence against a ci- tizen accused of an offense, the protection of the Fourth Amendment...is of no value, and, so far as those thus placed are concerned, might as well be stricken from the Constitution“ (34 S. Ct. 341. 344). Day was not exlicit in answering whether the exclusion- ary rule was “judicially created" as a means of enforcing the Fourth. or was actually required by the commands of the Fourth. He implied the latter though there was little his- torical evidence to support this position; in fact, common law explicitly rejected such a doctrine. Refusal to impose the exclusionary rule, according to Day, would be manifest neglect of the constitution, yet he would not call it Open defiance. This issue was to figure in two future cases, 321$ v. Colorado, 338 U.S. 25 (1949) and HEEE.V' ghig, 367 U.S. 643 (1961). The Weeks decision institutionalized the exclusionary rule as a sanction in federal trials where the Fourth Amendment has been violated. The Fourth Amendment, though, 26 like the other Bill Of Rights provisions, was only meant as a protection against the federal government. With the pas- sage of the Fourteenth Amendment and its prohibition against any state depriving “any person of life, liberty or prOperty without due process of law,“ the Supreme Court began the process of selective incorporation: the doctrine that those provisions in the Bill of Rights ”implicit in the concept of ordered liberty" were incorporated as binding against the states. In ngf v. Colorado (1949) the Supreme Court had its first Opportunity to decide whether the exclusionary rule was required in state courts. The Court, speaking through Justice Frankfurter, accepted as given the unreasonableness of the search; the facts of the case were not even pre- sented. Instead, Frankfurter commenced with the question of whether the privacy protections of the Fourth were in fact “implicit in the concept of ordered liberty and thus en- forceable through the due process clause.” Without any dis- cussion, Frankfurter merely stated that they were. As to whether this required the states to follow the exclusionary rule, Frankfurter first noted his belief that the Esskg_decision was not constitutionally based; the deci- sion was a matter of judicial implication. Second, he noted that within the United States, only seventeen states had ac- cepted the ngkg doctrine, and further, no other English speaking nation held illegally obtained evidence to be inad- missible. As to the relationship between the right to pri- 27 vacy and the exclusionary rule, Frankfurter concluded: "When we find that in fact most of the English-speaking world does not regard as vital to such protection the exclusion of evidence thus Ob- tained, we must hesitate to treat this remedy as an essential ingredient of the right" (338 U.S. 25, 29). Black, concurring, labeled the exclusionary rule a "judicially created rule of evidence which Congress might negate" (338 U.S. 25. 40). Douglas, in dissent, likewise 1a- belled it a rule of evidence, but contrary to Black felt that the rule must be enforced. Arguments in behalf of this were presented by Murphy, who attempted to show that all other remedies would fail. Only Rutledge, though, specifi- cally stated that the Fourth Amendment itself requires the exclusion of illegally obtained evidence. This view became the majority opinion in Mapp v. Ohio, 367 U.S. 643 (1961). On May 23, 1957. seven Cleveland po- lice officers, in search of a fugitive, searched the home of Dollree Mapp without a warrant. The only peoPle they found were in pictures, Obscene pictures. Mapp was arrested for possession of lewd and lascivious books and pictures in vio- lation of 2905.34 of Ohio's Revised Code. Clark, speaking for five members of the Court, reversed the decision in ngf. First, he claimed that the rule of evidence in Week: is of constitutional origin. Second, he stated that since ngf, over half the states that have passed on the exclu- sionary rule have adopted it in whole or in part. Third, he presented evidence that remedies other than the exclusionary 28 rule have been "worthless and futile.” Citing Weeks, Clark reiterated that the Fourth Amendment would be a ”form of words" without the exclusionary rule. What remained Of the federalism issue was for the Court to decide whether the same standard of reasonableness ap- plied to the states as applied to the U.S., and likewise whether the requirements of probable cause were the same. In both Ker v. California, 374 U.S. 23 (1963) and Aguilar v. Texas, 378 U.S. 108 (1964) the Supreme Court answered af- firmatively. The major changes that have occured in Fourth Amendment decisions Of the Supreme Court since Mapp, Ker and Aguilar fall into two main categories. The first change, involving what constitutes a protected area, was dramatically revised in a single case, Katz v. United States, 389 U.S. 347 (1967). The other major category of change involves the growing list of exceptions to the probable cause and warrant requirements, such as the right to frisk. IEEEX v. Obie, 392 U.S. l (1968). Let us consider the fundamental change that occured in 5235 v. United States. The Constitution provides for the right to be secure in one's person, houses, papers and effects against unreason- able searches or seizures. So, for an action of the govern- ment to be unreasonable within the meaning of the Fourth Amendment it must first be considered to be a search or se- zure. It is, of course, the Supreme Court that determines what is and is not a search. This is Of more than academic 29 concern. Consider the case of Olmstead v. United States, 227 U.S. 438 (1928), in which Chief Justice Taft, writing for the Supreme Court, ruled that a tap on telephone wires did not amount to a search within the meaning of the Fourth ' Amendment because the “wires are not part of his house or office, any more than are the highways along which they are stretched.“ The only constitutionally protected areas are those enumerated in the Constitution: persons, houses, pap- ers and effects. Despite a brilliant and forewarning dis- sent by Brandeis, who admonished that "the progress of science in furnishing the government with means of espionage is not likely to stop with wiretapping," so stood American Constitutional doctrine until 1967. Katz, involved in an interstate gambling Operation, was convicted as a result of evidence Obtained from a bug at- tached to the- outside of a telephone booth that he fre- quented. Potter Stewart for the majority, rearranged the Court's interpretation of the Constitution in a single sen- tence. "(T)he Fourth Amendment protects peOple, not places.“ He expanded on this by declaring that “(w)hat a person knowlingly exposes to the public, even in his home or office, is not a subject of the Fourth Amendment's protec- tion. But what he seeks to preserve as private, even in an area accesible to the public, may be constitutionally pro- tected.” This is known and followed today as the “expectation of privacy test.“ The other major changes since Mapp have more or less 30 involved limitations on the Fourth Amendment and or the ex- clusionary rule. One such case was Terry v. Ohio, 392 U.S. l (1968). Terry was arrested after a frisk by the police yielded a weapon. Rejecting traditional requirements, the Court ruled that “there must be a narrowly drawn authority to permit a reasonable search for weapons for the protection of the police Officer, where he has a reason to believe that he is dealing with an armed and dangerous individual, re- guardless of whether he has probable cause." The Court cre- ated ‘a balancing test between the extent of the intrusion and the government's interest in investigating crime. Other major examples Of limitations on the bounds of the Fourth Amendment include Camara v. Municipal Court, 387 U.S. 523, in which the Court ruled that probable cause was not absolutely required for inspections of residential prem- ises, Collonage Catering v. United States, 25 L. Ed. 2d 60 (1970), which upheld certain types of warrantless searches of businesses if based on Congressional authorization and Schneckloth v. Bustamonte, 412 U.S. 218 (1973), which upheld consent searches. One exception, the right to search after a lawful arrest, was restricted in Chimel v. California, 395 U.S. 722 (1969) to include only searches Of the arrestee's person and areas within his immediate con- trol. These decisions are much affected by the fact that a finding Of unreasonableness Often, though not always, re- quires that the evidence obtained be excluded. Thus, for 31 these decisions to be understood in their prOper context, a further examination of the purposes of the exclusionary rule is necessary. "The Court's perception of these purposes will determine the sCOpe...of the exclusionary rule.“ (LaFave, 1978, vol.1: 18) Since the overwhelming majority of Fourth Amendment cases involve the exclusionary rule, this will affect many of the Court's ultimate decisions. The Supreme Court has in one form or another noted three purposes to the exclusionary rule (LaFave,1978). The first and most important is deterrence. The Constitution protects us against unreasonable searches and seizures. Yet, as was stated earlier, judicial review, the major means of effecting the commands Of the constitution, is insuffi- cient protection against such searches and seizures. Laws permitting them are not necessary for them to take place. Thus, the Supreme Court turned to the exclusionary rule as the most effective available tool in deterring unconstitu- tional police conduct. In W213 the Court declared that "the exclusion Of evidence may be an effective way of deterring unreasonable searches,” 338 U.S. 25, 31 (1949). The Elkins' Court stated that "(t)he rule is calculated to prevent, not to repair," 364 U.S. 206, (1960). EEEEX.V' Ohio, 392 U.S. l (1968), explicitly gave priority to this purpose, proclaim- ing that the rule's "major thrust is a deterrent one.“ If this were the sole purpose of the exclusionary rule, we would expect that it would only be applied in cases where the actual effect would be to deter, and that evidence that 32 the rule has not deterred would be sufficient cause to per- mit tainted evidence. This is in fact what happened in United States v. Calandra, 414 U.S. 338 (1974), United States v. 32233, 428 U.S. 433 (1976), and S2223 v. Powell, 428 U.S. 465 (1976). Yet, other justices have occasionally noted justification in addition to deterrence. A second purpose as declared in Elkins, 364 U.S. 206, (1960), is "the imperative of judicial integrity.“ In 2255!! 392 U.S. l (1968), the Court expanded on this doctrine. "Courts which sit under our Constitution cannot and will not be made party tO lawless invasions of the constitutional rights of citizens by permitting unhindered governmental use Of the fruits of such invasions." Finally, a trust in go- vernment argument has been presented. The exclusionary rule is useful in "assuring the people...that the government would not profit from its lawless behavior, thus minimizing the risk of seriously undermining popular trust in govern- ment,“ United States v. Calandra, 414 U.S. 338 (1974) (dissent). Against these benefits, the Court must weigh the costs Of the exclusionary rule, both on a case by case basis and as a general rule of evidence. The major arguments and their rebuttals (see LaFave, 1978 , Dworkin, 1973, and Oaks, 1970) are as follows: (1) “The criminal goes free because the constable has blundered.“ People X; Defore, 242 NY 13,21 (1926). The most clear result of the exclusionary rule is that alleged cri- 33 minals are released, not because the evidence is of ques- tionable validity (as in coerced confessions), but because of the manner in which it was obtained. (2) The exclusionary rule fails to aid innocent vic- tims.. If I, as an innocent citizen, have my home ransacked by the police, the exclusionary rule provides no basis for reparations. Supporters of the rule, however, can claim that, as a deterrent, it does aid the innocent, and that there is nothing in the rule that prohibits aid to innocent victims. (3) The exclusionary rule fails to deter unconstitu- tional police behavior. Chief Justice Burger took this po- sition in Bivens v. Six Unknown Federal Agents, 403 U.S. 388 (1971). He supported' his evidence by relying on Oaks (1970), who himself felt there was no strong evidence to support or refute the deterrence theory. This argument, though, is a burden of proof argument. Whichever side faces the burden is bound to lose. In a related point, it is claimed that the exclusionary rule cannot work when the po- lice have goals other than successful prosecution. Confiscation of contraband and satisfying public outcries for visible enforcement are two such examples. What this argument suggests though, is not that the exclusionary rule is too strong, but that there is a need for additional sanc- tions. (4) The exclusionary rule is drastic and inflexible. The same sanction is imposed in willfull and extreme viola- 34 tions as in unknowing and technical violations. It gives the same relief to the murderer and rapist that it gives to the possessor Of marijuana.- Its supporters will claim, though, that this inflexibility is what makes it such a strong deterrent. (5) The exclusionary rule does not punish the police officer; it punishes society. This is closely related to the deterrent argument. If the officer is interested in arrest, but not in conviction, there is little to deter him from illegal conduct. Since he is the one who broke the constitution, it is he who should be punished, not society. (6) The exclusionary rule punishes prosecutors. Conviction rates will Obviously fall. Dworkin (1973) rep- lies that this criticism is valid only if the goal of the prosecutor is to gain convictions and is not to serve the law. (7) The exclusionary rule has handcuffed the police and thus led to a surge in the crime rate. Kamisar (1962), though, ‘ has shown these statistics to be faulty. Regardless, the courts have held that this cost argument was rejected when the Fourth Amendment was adopted. People X; 92232, 44 Ca. 2d 434 (1955). The arguments that the costs of the exclusionary rule outweigh the benefits have been held with increasing consis- tency by members of the Supreme Court. As a result, more and more exceptions have been found, and it could be claimed that the groundwork for overruling Mapp has been laid. 35 Chief Justice Burger stated in Bivens, 403 U.S. 388 (1971), that the exclusionary rule as a doctrine is "conceptually sterile and practically ineffective“ and urged that alterna- tives be investigated. In S3223 v. Powell, 428 U.S. 465 (1976), he suggested that the rule be dropped even without waiting for alternatives. In the same case, White called for substantial limitations. Blackmun, in Coolidge v. New Hampshire, 403 U.S. 443 (1971) claimed that "the Fourth Amendment supports no exclu- sionary rule." Rehnquist's strategy, on the other hand, ap- pears to be one of chipping away at the rule, bit by bit. The Court's majority Opinion in Calandra, 414 U.S. 338 (1974), to a limited extent did limit Mgpp by recharacteriz- ing the exclusionary rule as "judicially created.“ If the Court retains this position, the rule could then be over- ruled by a Congressional statute, since it is not a “constitutional" requirement. Pushing this chapter's “traditional" investigation of search and seizure to its logical limit, we can summarize the “black-letter“ Of Fourth Amendment law. The most basic requirements for a search to be reasonable are a warrant and probable cause. Probable cause is generally required wheth- er there is, United States v. Harris, 403 U.S. 573 (1971), or is not a warrant. Chambers v. Maroney, 399, U.S. 42 (1970). Warrants require a finding of probable cause, Aguilar v. Texas, 378 U.S. 108 (1964), by a neutral and de- tatched magistrate, Coolidge v. New Hampshire, 403 U.S. 443 36 (1971), supported _by oath or affirmation, Whitele v. Warden, 401 U.S. S60 (1971), and particularly describing the place to be searched, United States v. Ventrusca, 380 U.S. 102 (1965), and the things to be seized. Marron v. United States, 275 U.S. 192 (1927). For a search to be unreasonable, it must occur at a place where the accused has an ”expectation of privacy.” 5325 v. United States, 389 U.S. 347 (1967). This includes one's home, Silverman v. United States, 365 U.S. 505 (1961), one's place of business, Mancusi v. DeForte, 392 U.S. 364 (1976), and since 1967, one's person. EEEE V. United States, 389 U.S. 347 (1967). One's car gets minimal protection, Carroll v. United States, 414 U.S. 132 (1925), but this is nonetheless great compared to the protection one receives when the search takes place in an area in which he has no property interest ,United States v. Calandra, 414 U.S. 338 (1974), such as the house of a third person. Further, the less intrusive an actual search is, the less justification that is needed. A frisk, for example, is, other things be- ing equal, more reasonable than a full search, Tg££y_ v. 9232, 392 U.S. l (1968), as is detentive questioning. United States v. Mendenhall, 64 L. Ed. 497 (1980). Finally, we must consider exceptions to the warrant and probable cause requirements. The first of these is the right to search incident to a lawful arrest. From 1950 through 1969, this included the right to search the person of the arrestee and the entire premises where the arrest oc- 37 curred. This was limited by the 1969 Court to only include the person and the area within his immediate control. Chimel v. California, 395 U.S. 752 (1969). When a search occurs after, but not incident to a lawful arrest, the Court has been split on the need for a warrant. Chambers v. Maroney, 399 U.S. 752 (1970). Since some agree and some do not, the end result has probably been that such searches are more reasonable than searches without a lawful arrest, but less reasonable than searches incident to a lawful arrest. A second mitigating factor concerns permission by those who have a prOperty interest in the place to be searched. This includes the suspect, Schneckloth v. Bustamonte, 412 U.S. 218 (1973), a cohabitant, Frazier v. Cupp, 394 U.S. 731 (1969), and any participant to a wiretapped conversation. 92' Lee v. United States, 343 U.S. 747 (1952). Other miti- gating circumstances include searches after hot pursuit, Warden v. Hayden, 387 U.S. 294 (1967), searches at fixed or functional borders, United States v. Ramsey, 431 U.S. 606 (1977), searches statutorily allowed pursuant to regulation Of business, Collonade Catering v. United States, 397 U.S. 72 (1970), and searches for evidence to be used solely in administrative, man v. ggpgg, 100 U.S. 309 (1970), or grand jury hearings. United States v. Calandra, 414 U.S. 338 (1974). As well, seizures of Objects in plain view are gen- erally permitted. Coolidge v. New Hampshire, 493 U.S. 443 (1971). The above inforamation can be divided into three cate- 38 gories that the Court considers when deciding the reason- ableness of a search. The first involves the prior justifi- cation for the search. This inclludes the probable cause and warrant requirements. Next is the nature Of the search itself. This has twO dimensions: where the search took place and the extent of the intrusion (i.e., a full search vs. a stop and frisk or other limited intrusion). Finally the Court considers the exceptions to the probable cause and warrant requirements. This is, in my opinion, a fair repre- sentation of our traditional knowledge of search and sei- zure. Does it, though, provide a true "explanation" of the Supreme Court's decisions? My belief is that it does not. There are both theoretical and empirical problems with this approach. First, it assumes as did Blackstone, that the law is never changing. Far preferable is Holmes' statement that the law is the prophecies of what judges will do. Second, this traditional position assumes that judges are bound by history and precedent; therefore we can explain judicial de- cisions by examining these factors. The following chapter will expand on the following points, but suffice it to say that first, neither history nor precedent are as clear as they are made to seem by judicial Opinions, and second, there is no way to require Supreme Court justices to follow these traditions. Third, this view of judicial decision making ignores what is the most fundamental cause of the de- cisions given by the Supreme Court: politics. A series of liberal search and seizure decisions in the 1960's were fol- 39 lowed by more conservative decision making in the 1970's. Neither precedent nor history had changed. What had changed was the makeup of the Supreme Court. Warren, Fortas, Black, Harlan and Douglas were replaced by Burger, Blackmun, Powell, Rehnquist and Stevens. The Burger Court decisions cannot be explained by precedent, nor can those of earlier courts. Let us use 1948 case as an example of the failure of precedent. During the search in question, Officers stood on a chair and locked through the transom above the door. They observed a gambling Operation in progress and therefore en- tered the premises, arrested the tenant and seized the para- phenalia. The government argued that: the officers were lawfully in the hallway and that the hallway was a common use area unprotected by the Fourth Amendment; looking into the apartment was not a search since the eye cannot commit a search; the observations by the officers gave them probable cause for an arrest: the warrantless entry was therefore permissible; and the gambling paraphenalia was legally se- ized as under the incident to a lawful arrest exception to the warrant requirement. Each statement was backed by precedent, yet the court simply rejected the government's claim without exclaiming why. McDonald v. United States, 335 U.S. 451 (1948). Explanations based on precedent and history are useless. The Court rejected the search, I would hold, because of the personal policy preferences of a ma- jority of its members. 40 Finally, we note that this traditional approach to law leads either to numerous exceptions or unwieldy classifica- tions. Amsterdam (1974) recites over one-hundred Fourth Amendment rules and talks of his "compression" of the law. LaFave (1978) divides his treatise into eleven chapters and eighty-five sections. Each section contains from four to eight subsections. Each case is then ”explained“ by pigeon-holing it into one of the approximately 500 subsec- tions. After considering the admitted failure Of the tradi- tional scholars in search and seizure, a revised theoretical outlook on Supreme Court decision making might prove fruit- ful. 41 CHAPTER 3 SUPREME COURT DECISION MAKING The questions to be answered here are how do the Supreme Court justices make decisions and what factors will affect the decisions they make. Let us start by making an assumption: that the goals of the justices are primarily po- licy goals (Rohde and Spaeth, 1976). Between two choices, a justice will choose the alternative that comes closest to approximating his personal policy preferences. This assumes that the justices are not motivated by the desire to follow precedent, follow the Constitution, consider public wishes or consider the wishes of significant political others. Although this assumption can never be proven, I would like to give it support by showing that there are no external forces that require that anything but personal policy pref- erences be followed. Let us start with precedent. Precedent, quite simply, is the notion that similar cases should be decided similar- ly. Precedent, it is generally argued provides stability and justice to the law (see, for example, Cordozo, 1921). It does not, however, decide cases that reach the Supreme Court. The most basic reason for this that when the law is clear, when the precedents are all on one side of a con- flict; the conflict generally will not be taken to court (Carter,l979). Since cases that reach the Supreme Court 42 will typically have precedents on both sides, it would be difficult to claim that the decisions are determined by precedent. Now, let us say that the precedents are all or mostly on one side of a case. Will the justices be bound by prece- dent? The answer again is “no.” There are several means by which precedent can be avoided. The Court's power of fact freedom-~the ability to determine what facts are relevant to a case-- allows them to "distinguish a precedent." The facts of two cases are never exactly the same. Any case can be distinguished from any other case, and it is the judge who must determine whether two cases are similar enough for the prevoius Opinion to apply. Let us consider state aid to parochial schools. The constitutional test as to whether a program is consistent with the establishment clause of the First Amendment.is that “there must be a secular legislative purpose and a primary effect that neither advances nor in- hibits religion" Abington Township School District v. Schemp, 374 U.S. 203 (1963). Relying on that ruling, the Court decided in 1968 that furnishing books to all secondary school children, public, private and parochial, did not vio- late the Constitution. Board gf Education v. Allen, 392 U.S. 236. Three years later, the Court distinguished this case from a Rhode Island plan that paid 15% of the salary Of teachers of secular subjects in parochial schools. In over- ruling the plan, the Court ruled that ”a textboOk's content is ascertainable, but a teacher's handling of a subject is 43 not." Lemon v. Kurtzman, 403 U.S. 602 (1971). This in fact seems to be a reasonable distinction. Yet, in 1973, the Court overruled on separation of church and state grounds a New York plan that provided financial assistance for main- tenance and repair of school facilities and equipment. Committee for Public Education v. Nyguist, 413 U.S. 756 (1973). The Court was able to distinguish maintenance from textbooks because there was no guarantee that the money would be spent on maintenance. Distinguishing a precedent basically leaves the pre- vious decision intact. It merely refuses to extend the precedent to a "dissimilar" case. When the Court wishes to avoid a previous precedent, not only in a current case, but for the future as well, stronger measures may be taken. One such measure is to declare much of the previous case to be dicta. A dictum is "any expression in an Opinion that is unnecessary to the decision reached in the case." (Murphy and Pritchett, 1979: 489). A classic example of avoiding precedent through dicta occurs in Humphrey's Executor v. United States, 295 U.S. 602 (1935). Humphrey had been re- moved from his Federal Trade Commission post by President Roosevelt. His estate sued, claiming that the removal was illegal. Unfortunately, Chief Justice Taft's opinion in M1352 v. United States, 272 U.S. 52 (1926), stated that the President had the power to remove all presidential appoin- tees, including heads of federal regulatory commisions. The Humphrey Court, noting that Myers was a postmaster and not a 44 member of an dependent regulatory commission, ruled that the part of Taft‘s Opinion that dealt with such commissions was not to be applied. If even more damage is sought to a precedent, the precedent may be limited in principle. Mild but relevant examples pertain to the exclusionary rule. The M3pp_ deci- sion incorporated the exclusionary rule as part of the Fourth Amendment. The Court limited the scope of this deci- sion in Calandra, stating that the rule is a "judicially created doctrine." United States v. Calandra, 414 U.S. 338 (1974). Thus the Court was able to limit the scope Of Mapp so that it would not apply to grand jury hearings (Calandra), civil proceedings, United States v. Janis, 428 U.S. 43 (1976), or habeas corpus relief, S5223 v. Powell, 428 U.S. 465 (1976). It also happens that undesirable precedents can be ig- nored. Murphy and Pritchett (1979) note that in 1917, the Court upheld a ten-hour workday law for factory workers. In its Opinion the Court failed to mention its 1905 decision of Lochner v. §£!_Yggk, 198 U.S. 45, in which it declared un- constitutional a ten-hour law for bakers. More honest than letting a precedent fade into Obscuri- ty is the act of overruling a precedent. Although the Supreme Court has specifically overruled itself on average less than once per year (Abraham, 1975), the ramifications are often tremendous. Brown v. Board pf Education, 347 U.S. 483 (1954), overruled Plessy v. Ferguson, 163 U.S. 537 45 (1896), and the "separate but equal" doctrine. Mapp v. Ohio, 367 U.S. 643 (1961), overruled Wolf v. Colorado, 338 U.S. 25 (1949). Not only can precedents be overruled, but there is legitimate legal justification for doing so. Stated William 0. Douglas: "The place of stare decisis in constitutional law is ... tenuous. A judge loOking at a constitu- tional decision may have compulsions to revere past history and accept what was once written. But he remembers above all else that it is the Constitution which he swore to support and defend, not the gloss which his predecessors may have put on it. SO he comes to formulate his own views re- jecting some earlier ones as false and embracing others. He cannot do otherwise unless he lets men long dead and unaware of the problems of the age in which he lives do his thinking for him" (in Carter, 1979: 183). Or, as Oliver Wendell Holmes said more succinetly, "(i)t is revolting to have no.better reason for a rule of law than that so it was laid down in the time of Henry IV.“ (in Murphy and Pritchett, 1979:23). Clearly, if Justices desire not to follow precedent, there are numerous paths which they might follow. Now that we have seen that precedent need not be a 1i- mit on Supreme Court policy making, we will be able to see that the Constitution need not be either. Just as there are precedents on both sides of most cases, there are constitu- tional arguments on both sides of cases. The reason for this is that the meaning of the Constitution is usually am- biguous; no one has a monopoly on how the Constitution is to be interpreted. To illustrate, let us examine the different modes of constructing the meaning of the Constitution. 46 The simplest method Of giving meaning to Constitutional commands is the plain meaning rule. This viewpoint says that the Constitution literally means what it says. This literalist approach is found most frequently in the writings of Justice Hugo Black, who claimed among other things that the First Amendment's command that Congress shall pass no law abridging the freedom Of speech means 22 law. This mode of reasoning though, is not often followed or even possible to follow. Consider the Fourth Amendment. What does it mean that the people shall be free from "unreasonable searches and seizures?“ A "plain meaning" approach can not even tell us what a search or seizure is, no less an unrea- sonable one. The English language is generally too ambi- guous for any sort of plain meaning rule. Nevertheless, even when the words seem to be clear, as in the First Amendment, they are not a limit upon the Court's policy making abilities. Consider the meaning given to the Amendment by the 1919 Schenck v. United States, 249 U.S. 47, Court. According to that Court, the Amendment meant that you can say whatever you want so long as your speech does not create a "clear and present danger." That same year, the Court upheld the conviction Of Jacob Abrams for protesting U.S. support of the White armies in Russia and calling for a general strike. Abrams v. United States, 250 U.S. 616 (1919). By 1925, clear and present danger had given way to "bad tendency" as the new judicial test of whether speech could be restricted. Gitlow v. United States, 47 250 U.S. 652. This, in turn, was replaced in 1951 by a "gravity of the evil" test. Dennis v. United States, 341 U.S. 494. At no time has the Supreme Court ruled that "no law“ means no law. Let us consider a less controversial clause. Article 1, Section 10 of the Constitution says, "no state shall . . . pass any . . . law impairing the obligation of con- tracts." Yet when the State of Minnesota passed a mortgage moratorium law, the Court, in upholding the law, ruled that the founding fathers could not have predicted the exigencies of the Great Depression and that the statute must be seen “in the light of our whole experience” Home Building and Loan v. Blaisdell, 290 U.S. 398 (1934). The words of the Constitution bind the Court only when they want the words to bind them. As the Court Of last resort, there is no higher court to prevent them from ruling Otherwise. If the Court does not or need not rely on the words of the Constitution, perhaps it relies on the “intent of the framers." As is pointed out by Spaeth (1979), Carter (1979), and Murphy and Pritchett (1979). this is not so simple a so- lution. First, our knowledge as to what the framers meant is rather incomplete. Madison, the only delegate to take consistent notes at the Constitutional Convention, admitted- ly left an incomplete record. The Federalist Papers are quite valuable but represent the efforts of Hamilton, Madison and Jay to affect the beliefs of the New Yorkers; they do not represent the intent of the framers. Second, it 48 is not at all clear that there exists anything that could be labelled the intent Of the framers. The same clause could be supported by different peOple for totally disparate rea- sons. Nor is it clear that the framers always knew on what they were voting. The Fourth Amendment is a perfect exam- ple. The First Congress proposed as the Fourth Amendment: The right of the people to be secure in their per- sons, houses, papers, and effects, shall not be violated by warrants issuing without probable cause, supported by oath or affirmation, and not particularly describing the place to be searched, and the person or things to be seized. (Lasson, 1937: 101) Congressman Benson sought to strengthen the amendment by replacing "by warrants issuing" with "and no warrant shall issue. This motion lost by a considerable majority. (Lasson: 1937). Nevertheless, the rejected amendment is what was passed into law. Acording to Lasson (1937), Benson, as committee chairman, reinserted his own language and nobody noticed. Finally, even if there had been a "legislative intent" and that intent is known and well documented, there is no necessity of following the intent. According to Justice Brennan: "A too literal quest for the advice of the Founding Fathers upon the issues of these cases seems to me futile and misdirected . . . (P)ractices which may have been Objectionable to no one in the time of Jefferson and Madison may today be highly offensive to many person.“ Abington Township v. Schempp, 374 U.S. 203 (1963). Like precedent, the Constitution is quite malleable in the hands of the justices of the Supreme Court. Neither a lit- 49 eral reading of the words nor historical research can con- trol the decisions they reach. It should be fairly Obvious that once on the Court jus- tices have no need to follow public wishes since they are not electorally accountable. It can and has been claimed though, that the Court's actual power is based~ to a large extent on the public's perception of their legitimacy, and that more can be accomplished by moderation than by arro- gance (McCloskey, 1960: Kessel, 1966; Murphy and Tannenhaus, 1968). However, the Court's decisions on segregation, school prayer, criminal procedure and abortion suggest that the public's will can be stretched rather far. Nor is it likely that a justice will seek favor with the peOple to gain elected office. Charles Evan Hughes is the only jus- tice to resign from the Court to seek the Presidency. None have ever resigned to run for any other elected Office. As for Congress, there have been only two succussful attempts at limiting the Court's appellate jurisdiction. Impeachment has only affected one justice, Samuel Chase, but the vote for removal failed. The President, as well, has little power over sitting Supreme Court justices. The President might Offer a better job in return for favorable votes, but there are few better jObs to Offer. Since Hughes, only Byrnes and Goldberg have left the Court for other offices. And with the exception of Justice Jackson, there is little evidence of sitting justices actively seek- ing the position of Chief Justice. Thus, the justices need 50 not please anyone but themselves. What this set of arguments has been meant to show is not that justices g2 vote by their policy preferences, but that they ggp_. There are neither political nor legal res- trictions that prevent them from doing so. 322 Nature pf_£§g Choice Accepting for now the assumption that Supreme Court de- cisions are motivated by policy preferences and are free to decide consistently with them, the nature of the decision facing the justices in search and seizure decisions can be examined. We start by noting that Spaeth (1979) has found there to be three underlying values that explain Supreme Court decisions: freedom, equality, and economic regulation. Important to this study is the freedom dimension, which in- cludes issues such as freedom of speech, freedom of religion and criminal procedure. Criminal procedure itself includes several substantive areas of decision-making, most notably involuntary confessions and search and seizure. Ideally, we would expect justices to want all the freedom they could get for society. Unfortunately, freedom Often comes at the cost of order: setting felons free because of police impropriety is a relatively clear example of the tradeoffs involved. It will therefore be assumed that each justice receives some level of utility from society's balance between freedom and order. That utility will be maximized at the point which he believes society should have. 51 How does this translate into actual decisions as to the reasonableness of a search and seizure? We start by assum- ing that each search can be measured in terms of its severi- ty. The more severe a search is, the less reasonable it is. If severity can be considered unidimensional, and scaling results suggest that it can (see Rohde and Spaeth, 1976), then we can set up a continuum that measures the severity of a search. The severity is determined by the presence or ab- sence of a warrant, probable cause, and other related facts. Let us also say that the current Court has allowed all searches measured less than X, where X is a point on the se- verity of search scale. These are the “less severe" searches. All searches measured greater than X are not al- lowed. These are the "more severe" searches. X then, is the status quo of the Court's policy on search and seizure. As a result of these decisions, a particular balance between freedom and order has been reached. We now consider how the Court's decision can affect the status quo. Consider a search at S>X, where S represents the severity of a new search whose constitutionality is be- ing considered by the Court. Since S>X, the search is more severe than searches already declared unconstitutional. Thus, holding this search unconstitutional is consistent with the status quo, and it therefore should not be claimed that this ruling changes the status quo. Such a ruling would leave the freedom order balance unchanged. Likewise, deciding that a search is reasonable when SX, is constitutional. They are declaring as reasonable a search that is more severe than searches previously held un- reasonable. This is inconsistent with the status quo and thus represents a change in policy. The new search will represent the new status quo. The result will be a balance favoring relatively more order. Likewise, if the Court de- clares unreasonable a search that is more reasonable than the status quo, it is creating a stricter standard for searches and thus a new policy. The new search becomes the new status quo. The result will be a balance favoring rela- tively more freedom. To summarize, if the search is more severe than the status quo, the Court can move the status quo to the right by holding the search reasonable. Otherwise, the status quo remains constant. If the search is less severe than the status quo, the Court can move the status quo to the left by holding a new search unreasonable. Otherwise, the status quo remains. ng'Process How the above relates to specific decisions depends upon how the justices make decisions. Much of the debate in recent years on decision making in general has focused on the contrast between raional choice and cybernetic theories (e.g. Allison, 1971; Steinbruner, 1974). Briefly, the ra- tional choice model "holds that decisions will be taken 53 which maximize value (utility) within the constraints of the situation" (Steinbruner, 1974:28). If the theory is realis- tic and we could assume unidimensionality of the search and seizure cases and single peaked utility curves of the jus- tices the results would be as follows. As stated before the status quo has resulted in a cer- tain balance between freedom and order. Changing the status quo will result in a new balance. Figure 3.1 (below) shows the utility that a justice gets from different balances be- tween freedom and order. His optimal point is represented by “0(1)" and the sta- tus quo by "sq.” The justice in each case must choose be- tween keeping the status quo and the current balance or changing the status quo and obtaining a new balance. He will vote for the status quo if and only if the status quo balance is at least as close to his ideal point as the new balance will be. Remember that: 1. A balance toward more freedom can only result from invalidating a search that is less severe than the search corresponding to the status quo; 2. A balance toward more order can only result from upholding a search that is more severe than the search corresponding to the status quo; and 3. Upholding a search that is less severe than the search. coresponding to the status quO or invalidating a search that is more severe than the search corresponding to the status quo preserves the status quo. 54 Given the above, we can determine how a judge will react to any given search and its resulting balance. If in- validating a search that is more reasonable than the search corresponding to the status quo would lead to a society at point A, the search will be upheld and the status quo re- mains at B. If upholding a search that is more severe that the search corresponding to the status quo would lead to a society at C, the search will be upheld and the new status quo will be C. Similar arguments hold for searches leading to societies at D and E. The justice is indifferent to the societies at B and F. Given such a choice, we will assume precedent to have a decisional effect. The justice will in- validate a search at F, leaving society at B. To the right of point F, The utility decreases as compared to the status quo, so all searches in this area will likewise be inva- lidated. . Unfortunately there are several problems with this approach, both generally and with reguards to Supreme Court Justices in particular. Generally, Simon (1976) has pointed out that humans are not capable of maximizing. Rather, the decision maker satisfices. The cybernetic theory of deci- sion making takes this into account an assumes a repertory of responses (Steinbruner, 1974). “In brief, cybernetics provides an analysis of extremely simple decision making mechanisms which are nonetheless highly successful in the proper environments" (Steinbruner, 1974:13). The decision process focusses “on a few incoming variables while elim- 55 / Freedom sq 0(1) Order A B C D E F G Figure 3.1. A Justice's Utility for Freedom and Order 56 inating entirely any serious calculation of probable out- comes" (Steinbruner, 1974:66). 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 unidimen-' sionality and single peakedness, in the sense used by formal theorists, can not be inferred from Guttman scaleability. On the other hand suggestions of a more simplified, cyber- netic approach come from various sources. First, note that the Supreme Court's caseload has come under tremendous attention recently. The number of cases filed has more than doubled in the past twenty years (Howard, 1982). The number of full opinions has multiplied from sixty-five in 1953 to 141 in 1981 (Greenhouse, 1983). This seems to have greatly affected the nature of the Court's decision making. Claims Chief Justice Burger, “(O)nly fundamental changes in structure and jurisdiction will insure prOper time for reflection" (Greenhouse, 1983: 86.5), insinuating of course that prOper time for reflection is not now being taken. States Blackmun: “(t)he heavier the burden, the less is the poss- ibility of adequate performance and the greater is the probability of less-than-well considered adju- dication. . . .One, therefore, to a large degree, relies on innate and hOpefully already developed prOper judicial reaction" (Ripple, 1980:175). Moreover, rational choice decision making becomes in- creasingly unlikely under conditions of uncertainty (March and Simon, 1958). The goal of the exclusionary rule may be 57 to prevent police misbehavior, yet neither the Court nor, judging by the disparate conclusions reached by various scholars (Oaks, 1970: Spiotto, 1973: Canon, 1973-74), the legal discipline knows what effects the Court's decisions have had. Further evidence of a simplified decision structure comes from the success that qualitative and semi-quantitive predictors have had. Whitaker (1971) reports 97% accuracy in Supreme Court simulations involving graduate students at Temple University, with one Justice assigned to each stu- dent. Spaeth (1979) has attempted to predict the decisions of each justice on a given Court and reports an average suc- cess rate of 84.7% from 1970 through 1977. As difficult as rational choice decisibn making is, it would be incompre- hensible if one could replicate the decision making calculus Of nine others. This again suggests that the justices use a more simplified decision structure. Previous studies present some clues as to how this might be accomplished. Tanenhaus et a1. (1963) have shown that a small number of "cues" present in briefs influence the Supreme Court's certiorari decisions. Ulmer (1969) sug- gests that various "signs" derived from the lower court re- cord affected Justice Frankfurter's civil liberty decisions. It is suggested herein that the justices monitor certain "facts" from the case and that the presence or absence of these facts strongly predispose the Court in its decisions on the reasonableness Of a search and seizure. 58 Specification Using the framework Of prior justification, nature of the intrusion and mitigating circumstances developed in Chapter Two, the variables to be included in this model can be specified. First, looking at prior justification, probable cause is included. It is perhaps the most basic requirement Of a valid search. Although a precise definition is impossible, Landynski (1966) states that in a search, "there must exist reasonable grounds for believing that the dwelling contains the things to be searched for. In the case of arrest there must exist reasonable grounds for believing that the suspect has committed the offense" (p.46). According to the Court, probable cause is "facts and circumstances . . . such as to (warrant a man Of prudence and caution in believing that the Offense has been (or is being) committed.” Stacey v. Emery, 97 U.S. 642, 645 (1878). What is crucial is that the police have more than "mere suspicion." W2 g S22 v. United States, 371 U.S. 471. (1963). Constitutionally, the basis for the probable cause re- quirement is in the Fourth Amendment itself, commanding that "no warrants shall issue, but on probable cause.“ This was in large part a response to the general warrant, in which the name of the person to be arrested or a description of the things to be seized were left blank. The constitution though, as interpreted by the Supreme Court, does not re- quire that all searches be pursuant tO a warrant. The first 59 clause, securing to the peOple the right to be free from un- reasonable searches and seizures, grants additional police power. Warrantless searches may be allowed under certain circumstances provided they are reasonable. The most basic guide to reasonableness is probable cause. Thus, probable cause is generally required whether there is a warrant or not. Its presence makes a search more reasonable: its ab- sence makes it less so. Although a warrant is often not necessary for a search to be reasonable provided probable cause exists, the Supreme Court has time and time again expressed a strong preference for its use. Obtaining a warrant, acccording to the Supreme Court (as noted by Lafave), ”interposes an orderly procedure involving 'judicial impartiality' whereby 'a neutral and de- tatched magistrate' can make 'informed and delib- erate determinations' on the issue of probable cause. To leave such decisions to the police is to allow 'hurried actions'" (1978, vol. 1:445) The presence of a warrant will make a search more reason- able, independent Of probable cause. "(I)n a doubtful or marginal case, a search under a warrant may be sustainable when without one it would fail." United States v. Ventrusca, 380 U.S. 102 (1965). This is for good reason; without such a rule there would be little incentive to ever Obtain a war- rant. For the warrant to make the search more reasonable, it must, of course, be valid. This requires a finding of prob- able cause, Aguilar v. Texas, 378 U.S. 108 (1964), by a neu- 60 tral and detatched magistrate, Coolidge v. New Hampshire, 403 U.S. 443 (1971), supported by oath or affirmation, Whiteley v. Warden, 401 U.S. 560 (1971), and particularly describing the place to be searched, United States v. Ventrusca, 380 U.S. 102 (1965), and the things to be seized. Marron v. United States, 275 U.S. 192 (1927). Prior justification in the form of a warrant and or probable cause is perhaps the most crucial aspect in the de- termination of whether or not a search is reasonable. It is through these protections that arbitrary police power can be retarded. However, unreasonableness is not necessarily equal to arbitrariness. Searches with probable cause and warrants have been disallowed; searches without them have been approved. Before a determination as to the reasonable- ness of a search can be made, the search itself must be ex- amined. What must be determined is the nature of the intru- sion. The greater the intrusion, the less reasonable the search, all other things being equal. The two factors that determine the extent Of the intrusion are where the search took place and the extent of the intrusion. Searches may take place in homes or businesses, in cars or in public. The Constitution itself gives protection to: one's person, including bodies, Schmerber v. California, 384 U.S. 757 (1966), and clothing, Beck v. Ohio, 379 U.S. 89 (1966): houses, including apartments, Clinton v. Virginia, 377 U.S. 158 (1964), hotel rooms, Stoner v. California, 376 U.S. 483 (1964), and business offices, United States v. 61 Lefkowitz, 285 U.S. 452 (1932): papers including letters, £5 pgpgg Jackson, 96 U.S.727 (1878); and effects, such as automobiles, Preston v. United States , 376 U.S. 364 (1964). Until 1967, the Supreme Court's definition of what con- stituted a search extended only to physical intrusions. This was overruled in 1967 in 5355 v. United States, 389 U.S. 347. The Court ruled therein that the Constitution protects peOple, not places. The doctrine of a "constitutionally protected area" was replaced with the different, but somewhat overlapping “expectation of privacy test." As a result, the Court has declared that different areas demand different expectations Of privacy. It is in one's house that one has the greatest expecta- tion of privacy. “At the very core (of the Fourth Amendment) stands the right of a man to retreat into his own home and there be free from unreasonable governmental intru- sion." Silverman v. United States, 365 U.S. 505 (1961). Still protected, but less so, is one's place of business. Commercial premises were protected even before the expecta- tion of privacy test. "The businessman, like the occupant of a residence, has a constitutional right to go about his business free from unreasonable official entries upon his private commercial property.“ .§ES v. 313! g£_Seattle, 387 U.S. 541 (1967). Because businesses are Open to the public, and because the government has the right tO closely regulate many types of business, the protection of commercial prOper- ty is not as great as the protection of one's home. Much 62 less protection is given to one's car. Such rulings date back to Caroll v. United States, 267 U.S. 132 (1925), where the Court ruled that in the case of an automobile, where the car "can be quickly out of the locality or the jurisdication in which the warrrant must be sought," procurement of said warrant would not necessarily be required. Nor is it always necessary that a valid warrant could have been obtained und- er the probable cause requirement when searching a car (Lafave, vol. II). "There is a constitutional difference between houses and cars." Chambers v. Maroney, 399 U.S. 42 (1970). Likewise, searches in public do not infringe priva- cy as much as searches at home. Nevertheless, the protection given to those in cars is much greater than the protection granted one who has no property interest in the place that had been searched. One may not claim protection against the unreasonable search of the property of another. Generally, this rule is based on the doctrine of standing which requires that the person bringing suit has had his own rights violated. Related to the Fourth Amendment, the reasonable expectation of privacy requirement becomes the defendant's reasonable expectation of privacy. Mancusi v. DeForte, 392 U.S. 364 (1968). There is a severely lessened expectatin of privacy in the house, ngg_§gg v. United States, 371 U.S. 471 (1963), business, United States v. Miller, 425 U.S. 435 (1976), or car,PeOple !;_Raskas, 360 N.E. 2d 1252 (1977), of another. As was stated above, how the search took place is the 63 second prong of the extent of the intrusion. A full search is the most extreme intrusion and thus needs the greatest amount of prior justification. Such a search may be characterized as entrance into one's house, car, clothes, or person. This may be contrasted to the stOp and frisk, where the pOlice stop someone and "pats them down, usually in search Of weapons. It was not until 1968 that the Supreme Court took the opportunity to rule on such practices. They ruled at that time that probable cause under such circums- tances generally will not be required. Even less prior ju- stification will be required when one is stopped by the po- lice for questioning, Terr v. ghig, 392 U.S. 1 (1968). The Court, it is assumed, will weigh the prior justifi- cation not only against the extent of the intrusion, but against any one of several mitigating circumstances. The most well known mitigating circumstance is the right to search incident to lawful arrest. The authority to search subsequent to an arrest dates back to ngkg’ v. United States, 232 U.S. 383 (1914), where the Court ruled that the case at hand was ”not an assertion of the right on the part of the Government, always recognized under English and American law, to search the person of the accused when le- gally arrested to discover and seize the fruits or evidence of crime.“ In United States v. Rabinowitz, 339 U.S. 56 (1950), the Court specifically ruled that police had the right to search the person of the arrestee and the entire premises where the arrest occurred. In 1969, this was li- 64 mited to include only the person and the area within his im- mediate control. Chimel v. California, 395 U.S. 752 (1969). The right to search flows automatically from the lawful arr- est; probable cause or even reasonable suspicion are not re- quired. United States v. Robinson, 414 U.S.218 (1973). When a search occurs after, but not incident to a lawful arrest, the Court has been split on the need for a warrant. Chambers v. Maroney, 399 U.S. 42 (1970). The end result has been that such a search is more reasonable than one without an arrest and less reasonable than a search that immediately follows an arrest. SO, a lawful arrest gives added justifi- cation to a search. But if the arrest is unlawful the search is no more reasonable. Nevertheless, a search that is otherwise reasonable should not become unreasonable be- cause it was preceeded by an unlawful arrest. A second class of mitigating factors, as discussed in the previous chapter, concerns permission by those who have a property interest in the place to be searched. This in- cludes the suspect, Schneckloth v. Bustamonte, 412 U.S. 218 (1973), a cohabitant, Frazier v. gppp, 394 U.S. 731 (1969), and any participant in a wiretapped conversation, prgsg v. United States, 343 U.S. 747 (1952). Other mitigating cir- cumstances include searches after hot pursuit, Warden v. Hayden, 387 U.S. 294 (1967), searches at fixed or functional borders United States v. Ramsey, 431 U.S. 606 (1977), searches statutorily allowed pursuant to Congress' authority to regulate business, Colonnade Catering Cor . v. United 65 States, 397 U.S. 72 (1970), and searches for evidence to be used solely in administrative, Wygan v. James, 400 U.S. 309 (1977), or grand jury hearings, United States v. Calandra, 414 U.S. 388 (1974). As well, seizures of objects in plain view are generally permitted, Coolidgg v. New Hampshire, 493 U.S. 443 (1971). The presence or absence of these factors, it is herein argued, will add to or subtract from the reasonableness of a search. In addition to these a variable to measure the change in the Court's membership.) As the transition was made from the Warren Court to the Burger Court, it would not be surprising if the Court's decisions became more conserva- tive, even after the facts of the cases have been con- trolled. The Operationalization of the variables will be dis- cussed in Chapter V. What we need now is a means by which the effects Of these veriables can be gauged. CHAPTER 4 METHODOLOGY Most political scientists are fairly familiar with the general linear model Y1 = a+BXi+ei where Yi is the dependent variable,“ is the constant, B is the change from Y that results in a one unit increase in X and 6i is the error term. Under the assumptions that 1. the model is correctly specified, 2. the expectation of the error term is zero, 3. the error term has constant variance, 4. there is no autocorrelation between the error terms, and 5. the X5 are nonstochastic variables whose values are fixed it can be shown that the ordinary least squares (OLS) esti- mator 6 is the best (smallest variance) in the class of lin- ear unbiased estimators (BLUE). If we added the assumption that the error term is normally distributed, we are in a po- sition to conduct classical hypothesis tests of the model. There are many reasons why OLS regression has become one Of the most widely used techniques in political science: interpretation is extremely straightforward; program pack- ages such as SPSS solve most programming problems; and com- putation costs are low. But, the only true justification is 66 67 the desirability of the best linear unbiased criterion. When the above stated assumptions are not met, the justifi- cation for using OLS must be questioned. Let us now consider Supreme Court decisions. What we observe is a dichotomous decision: the Court must hold that a search is reasonable or unreasonaable. What herein is be- lieved to exist though, is a real but unobservable predispo- sition to act in a given matter. For example, let us imag- ine two searches: one involves a search, with probable cause and a warrant, of a car; the other is also a search of a car, again with probable cause, but this time, no warrant. Several, if not most, Of the justices would find both searches reasonable. In coding such decisions, both deci- sions get the same value. However, each justice was more likely to find the first search reasonable. What exists is a continuum; what is measured is a dichotomy. The most elementary manner in which a dichotomous choice can be modeled is the linear probability model Yi=a+BXi+ei where Yi equals 1 if the first Option is chosen (finding a search reasonable) and 0 if the second Option is chosen (finding a search unreasonable). These are the only values Yi can have. Thus, we can describe the probability distri- bution of Y by letting Pi equal the probability that Yi=1 and 1-Pi equal the probability that Yi=0. It therefore fol- lows that 68 E(Yi) = 1(Pi) + 0(1-Pi) =Pi The regression equation may thus be interpreted as de- scribing the probability that a justice will find a search , reasonable (Pindyck and Rubinfeld, 1976). Since Yi=a+BXi+el 8i = l -»0- 8X1 when Yi = l and Si = -CI- BXi when Yi = 0 the probability that Yi = 1 remains Pi, and the probability that Yi = 0 remains l-Pi. From here it can be shown that the variance Of the error term is heteroskedastic (not con- stant for all observations), thus violating the third as- sumption of the OLS model. This is done as follows. When Xi is fixed (assumption 5), the probability dis- tribution of ei must be equivalent to the probability dis- tribution of Yi (Pindyck and Rubinfeld, 1976). Since the expectation of the error term is zero (assumption 2), it should be clear that E(ei) a (1-a-BXi)Pi + (-a-BXi)(1-Pi) = 0 Solving for Pi gives us the result that Pi =a+BXi l-Pi = l -.5. If the observed value of the dependent variable for case i is 0, the model is considered to have predicted correctly if the Prob(Yi=l) <.5. Besides testing the fit of the model, we are, as in re- gression, interested in significance tests on the parameter estimates. We can test whether an estimate is signifantly different from any other value, such as zero, by taking ad- vantage of the fact that the maximum likelihood estimate di- vided by the standard error of the estimate is approximately 1a standardized normal random variable or Z score (Aldrich and Cnudde, 1977). Thus, in a large sample one tailed test, an MLE/SE greater than the familiar benchmark 1.64 (1.94 in the two tailed test) is statistically significant at p<.05. Unfortunately, interpreting the coefficients is not as straightforward as it is in simple regression, where represents the change in Y that occurs with a one unit change in X, or its variation for dichotomous dependent variables, the linear probability model, where represents the change in the probabilitly that Y=1 that occurs with a one unit change in X. As discussed earlier, represents the change in the cumulative normal function that occurs with a one unit change in X. Again, as stated before, a constant change in the cumulative normal function will have a vari- able effect on the probability that Y=1. The effect will be greatest when Prob(Y=l) is closest to .5. Let us consider as a hypothetical example the variable Warrant, which takes on the value of 0 when there was no 83 TABLE 4.3 Probability of Finding Search Reasonable With and Without Warrants Case W/O Warrant (Z Score) With Warrant (Z Score) Change 1 .01 (-2.33) .09 (-1.33) .08 2 .10 (-1.28) .39 (-0.28) .29 3 .25 (-o.67) .63 ( 0.33) .38 4 .50 ( 0.00) .84 ( 1.00) .34 5 .75 ( 0.67) .95 ( 1.67) .20 ' 6 .90 ( 1.64) .996 ( 2.64) .096 7 .99 ' ( 2.33) .999 ( 3.33) . .009 84 warrant, and 1 when there was Let us also say, for simplicity's sake that the MLE estimate for this variable was 1. Thus, times X equals 1 (1x1) when there was a war- rant and 0 (1x0) when no warrant was present. Now let us consider seven hypothetical cases of warrantless searches where the probabilities of the Supreme Court finding the searches in question reasonable are, for various reasons, .01, .10, .25, .50, .75, .90, .99 respectively. What would the probabilities have been in these cases had a warrant been Obtained? Table 4.3 (above) summarizes the re- sults. In the first column, the probability of finding the gi- ven warrantless search is presented. The second column pre- sents the corresponding 2 score. The third column presents the probability of finding a similar search reasonable if a warrant had been obtained. The probabilities are computed from the Z scores in the fourth column. The final column consists of the change in the probability of finding each search reasonable by adding in the condition of having a warrant . As can be seen, the change in probability increases and re- aches its maximum in case 3, where the probability moves from under .5 to over .5. Moving away from .5, the changes in the probabilities start to decrease again. The Data The data to be used in this study have been gathered 85 from all formally decided cases between Mapp v. Ohio (1961) and the end of the end of the 1981 term in which the Supreme Court decided whether or not a search by state or federal Officials was unreasonable within the meaning of the Fourth Amendment. The dependent variables are the decisions of the Supreme Court and of the individual justices. The independ- ent variables are the facts of the cases, as discussed in Chapter 3. CHAPTER FIVE RESULTS Heretofore I have discussed the current state of be- havioral reseacrch in public law as applied to the predic- tion of Supreme Court decisions (Chapter One), the substan- tive issues of search and seizure (Chapter Two), 8 model Of Supreme Court decision making (Chapter Three), and how pro- bit can be used to measure that model. This has been done. The results are the subject matter of this chapter. OPERATIONALIZATION The model will be specified using the information pre- sented in Chapter Three regarding the reasonableness of the search. There will, however, be one limit and one addition. The limit is this: the value of all variables used must be knowable prior to the release of the opinion of the Court. For example, the Supreme Court's decision on whether there was probable cause or whether an arrest was lawful can not be used. As has been discussed earlier, this is for two reasons. Theoretically, we do not know that the Supreme Court's decision on probable cause precedes as a cause its_ decision on the reasonableness of a search or succeeds it as a justification. Even if this were not a problem the Supreme Court's decision on probable cause obviously can not be used for predictive purposes since its value is not known until the day the Court renders its decision. For these 86 87 facts lower Court decisions will be used as indicators. Additionally, rather than assume as Kort (1957) and others have, that changes in membership do not affect the Court's decisions, the model herein explicitly takes this into account. As the Court's membership changed with the Nixon and Ford appointees replacing those appointed by Roosevelt through Johnson, one would quite reasonably expect that the search and seizure decisions Of the Court have got- ten more conservative, even after controlling for the facts of the case (see for example, Wasby, 1976). The variables to be used and their operationalizations are as follows. REASONABLENESS pg THE SEARCH Decision: The decision Of the Court in a search or seizure case. Takes on the value 1 if the Court either found a search to be reasonable or allowed obtained evidence to be used. Otherwise takes on the value 0. NATURE 9E THE INTRUSION: PLACE House: Takes on the value 1 when the search or seizure took place at the home of the defendant, zero otherwise. The es- tiamate is predicted to be negative. Business: Takes on the value 1 when the search or seizure took place at the place of business of the defendant, zero otherwise. The estimate is predicted to be negative, but less negative than House. Car: Takes on the value 1 when the search took place in the 88 car of the defendant, zero otherwise. The estimate is predicted to be negative, but less negative than Business. Person: Takes on the value 1 when the search or seizure took place on the person of the defendant, zero otherwise. The estimate is predicted to be negative, but less negative than Business. When all of the PLACE variables take on the value zero, the search took place in an area over which the defendant does not have a prOperty interest. NATURE 9E THE INTRUSION: EXTENT Search: Takes on the value 1 when there was a full search or seizure. Takes on the value 0 during limited intrusions such as a "stop and frisk" or detentive questioning. The estimate is predicted to be negative. PRIOR JUSTIFICATION Warrant: Takes on the value 1 when a search took place after a search warrant had been Obtained or when an arrest took place after an arrest warrant had been Obtained, zero other- wise. The estimate is predicted to be positive. Probable Cause: The proportion Of lower court judges finding probable cause. The estimate is predicted to be positive. ARREST Incident Lawful: The proportion of lower court judges find- ing that the search took place incident to a lawful arrest. The estimate is predicted to be positive. 89 After Lawful: The prOportion of lower court judges finding that the search took place after but not incident to a law- ful arrest. The estimate is expected to be positive, but less than Incident Lawful. Unlawful: The proportion of lower court judges finding that the search took place incident to or after an unlawful arr- est. The estimate is expected to be substantively and sta- tistically insignificant (see Chapter Three). When all of the ARREST variables take on the value 0, the search took place without a prior arrest. EXCEPTIONS Exceptions: Since the probability of any of the following exceptions appearing is quite small, and since there is no a priori reason for believing that any one exception is signi- ficantly more important than another, an Exceptions variable was created that takes on the value of the number of the following questions that can be answered "yes." It can range from zero to five but in fact ranged from zero to three. 1. Was the search or seizure following hot pursuit? 2. Was the search or seizure at a fixed or functional border? 3. Was the search statutorily allowed persuant to Congress' authority to regulate business? 4. Was the evidence seized used only for administra- tive or grand jury hearings? 5. Was the evidence seized in plain view? 90 The estimate is predicted to be positive. CHANGE Change: Takes on the value 0 during the Warren Court, 1 after Burger was appointed, 2 after Blackmun was appointed, 4 after Rehnquist and Powell were jointly appointed and 5 after Stevens was appointed. The estimate is expected to be positive. Findings The estimates are presented below in Table 5.1. As can be seen, the model does a praise worthy job of explaining the Court's decisions. Looking at the summary measures, we first note that the overall significance of the model, -2xLLR, is significant at p< .01. The results clearly can- not be explained by chance alone. The estimated R2, .57, is explaining over half of the variance from the mean. The mean is .54, meaning that the Supreme Court ruled that the searches in question were reasonable 54% of the time. Having no further information would therefore lead us to predict correctly 54% of the decisions of the Court. By adding our independent variables, predictive accuracy has increased to 76%. At first blush, it seems as if the model has taken one of the Courts most difficult areas, one characterized as inconsistent, bedeviling and even a mess (Amsterdam, 1974), and found, for the first time, a clear and logical form. Does this consistency hold when we look at the particu- 91 lar parameter estimates? With one or two possible excep- tions, the answer seems to be yes. This part of the analy- sis will start with the prior justification Variables, move to the place of the intrusion and the extent of the intru- sion. Exceptions to the probable cause and warrant require- ments will then be considered. The first prior justifica- tion variable to be considered is whether or not there was a warrant. The existence of a warrant should, of course, add to the reasonableness of a search. To add to the reason- ableness of a search, though, the warrant must be valid. An invalid warrant cannot add to the search's reasonableness, but, it should not subtract from an otherwise reasonable search. Just as we cannot Operationalize probable cause to include the Supreme Court's decision on probable cause (see Chapter One), we cannot Operationalize the valid warrant re- quirement to include the Supreme Court's decision on whether the warrant was valid or not. We do not know whether the. Court's decision on whether a warrant was valid precedes or succeeds its decision on the reasonableness of a search. Instead, the existence of a warrant will be used as a meas- ure of whether there was a valid warrant. Thus, the strength of the parameter estimate will be weakened by the fact that some of the warrants were declared by the Court to be invalid. These could not have helped make a search more reasonable. Nevertheless, -our estimate is a moderately large and-statistically significant .95. This is the inde- pendent effect (controlling for other variables) that the 92 TABLE 5 .1 Probit Estimates for Search and Seizure Decisions Variable MLE S.E. MLE/S.E. House -2.31 .61 -3.78** Business -2.19 .64 -3.4l** Car -2.02 .63 -3.21** Person -1028 054 -2037** Search - .77 .44 -l.76* Warrant .95 .45 2.14* Incident Lawful Arrest 2.30 .81 2.84** After Lawful Arrest 1.22 .50 2.47** After Unlawful Arrest - .57 .63 - .91 Exceptions 1.36 .35 3.91** Change .16 .07 2.26* Constant 1.34 mean .55 Estimated R2 .57 % correct 76 -2xLLR 56.68** n=123 * sig. at {65 ** sig. at .01 93 the addition of this variable has on the cumulative normal. Since finding a search reasonable was coded as l and finding a search unreasonable was coded as 0, the larger the esti- mate, the more that factor adds to the reasonableness of the search. A search that had a prior probability of .5 of be- ing found reasonable would, with the addition of a warrant, have a probability of being found reasonable equal to .83. In a questionable search, it would clearly make it worth the while of a law enforcement officer to obtain a warrant. If, as the Court ruled in Elkins v. United States, 364 U.S. 206 (1960), the purpose of the exclusionary rule is to prevent or deter unconstitutional police behavior, the Court has done its job in signifying the importance of prior judicial approval before searches take place. The "probable cause" variable did not fare as well. As discussed earlier, probable cause can not be operationalized as the Supreme Court's decision on probable cause since its value is not known prior to the day the Court's Opinion is released and more importantly, since the Court's decision on probable cause is not clearly independent of its decision on the reasonableness of the search. As such, probable cause was operationalized as the lower court's decision on prob- able cause. This proved to be not a very reliable indica- tor. The estimate of -.32, although substantively small and statistically insignificant, suggests a negative relation- ship between a lower court finding of probable cause and the reasonableness of the search. The Supreme Court did, howev- 94 er, uphold 53% of the searches in which it did not find probable cause, as Opposed to 65% of the searches in which it did. Probable cause is undoubtedly important; the trouble remains in trying to Operationalize it. We next turn to where the search took place. The search could be at one's house, one's place of business, of one's person or in one's car. All of the variables take on the value zero when the search takes place in an area over which one does not have a property interest. In comparisons between places in which one has a property interest we see, as expected,that one's house gets the most protection. A search of one's house, as compared with a search in an area in which one does not have a property interest, results in 2.31 units being subtracted from the cumulative normal prob- ability function when determining whether the search is rea- sonable. A search that had a probability of being upheld of .85 if it took place at the house of another (a fair guess for such a search) would have a probability of being upheld if it had taken place at the defendant's house of .10. Less protection, although not decidedly so, is given one's place of business. The estimate here is -2.19. This lesser protection is consistent with the Court's determina- tion that at the very core of the Fourth Amendment "stands the right of a man to retreat into his own home and there be free from unreasonable governmental intrusion." Silverman v. United States, 365 U.S. 505 (1961). The Court has consist- ently ruled that businesses are protected by the Fourth 95 Amendment, Sgs_v. City of Seattle, 387 U.S. 541 (1967), yet, as LaFave (1978, vol.1:338) notes, “businesses and com- mercial premises are not as private as residential prem- ises." Thus, while the Supreme Court has upheld 33% of the searches in houses, they have upheld 40% of the searches of businesses. Less protection is given to one's car and even less to one's person. Sixty-three percent of searches of the former and 60% of searches of the latter have been upheld. This evidence of lesser protection holds after other factors are controlled. The probit estimates for these two variables are -2.02 and -l.28 respectively. The right to search a car, for the period of this study, did not seem to include the right to search suitcases within a car without first ob- taining a warrant. Although relatively few cases were de- cided on this issue, the Court was rather adamant on this point. United States v. Chadwick, 433 U.S. 1 (1977). This was, however, reversed at the end of the 1982 term in United States v. Eggs (1982). Despite this aberration, the model seems to suggest that the Court is being rather consistent with its avowed doctrines. The second of the nature of the intrusion is the extent of the intrusion. As noted in Chapter Two and Chapter Three, the Supreme Court considers stOps and frisks and si- milar lesser intrusions as being inherently more reasonable than full searches or seizures. In fact, they have upheld 71% of the cases involving lesser intrusions brought before 96 them, while upholding only 53% of the full searches or sei- zures. The parameter estimate for the search variable, -.77, shows that this is a trend that holds after con- trolling for other variables that might be related to it. If a stop and frisk has a probability of .70 of being found reasonable, a full search, under the same circumstances, has a probability of .40 of being found reasonable. Outside of the general framework of prior justifica- tion, the place of the intrusion and the extent of the in- trusion, the Court has created several exceptions to the warrant and probable cause requirements that must be weighed before it decides the constitutionality of a search. [Let us look at the most important of these, the ARREST variables: -Incident to Lawful, After Lawful and After Unlawful. At least one of these will be nonzero whenever the search at hand was preceeded by an arrest. As discussed in Chapters Two and Three, the Supreme Court has long upheld the right to search incident to a lawful arrest. United States v. Robinson, 414 U.S. 218 (1973). Less justification exists for a search after, but not incident to, a lawful arrest. Chimel v. California, 395 U.S. 752 (1969). When the arrest is unlawful, the fact that the arrest was made cannot add to the justifiability of an otherwise unreasonable search, al- though it is not clear that an unlawful arrest would sub- tract from an otherwise reasonable search. The Supreme Court appears to be ruling consistently with these established principles. They have upheld 56% of 97 the searches incident to a lawful arrest, 50% of the searches after a lawful arrest and none of the searches oc- ' curing after an unlawful arrest. The probit estimates are 2.30, 1.22 and -.57 respectfully. The existence of a search incident to an arrest adds to the cumulative normal the val- ue of 2.30. Thus, if the probability of chance of finding a search reasonable was .5, the addition of a search incident to arrest would make the new probability of finding the search reasonable .99. A search after a lawful arrest would make the resulting probability .89. The value for an unlawful arrest was expected to be zero. An unlawful arrest certainly can not justify an otherwise unreasonable search, but it also seems plausible that the fact that a person was unlawfully arrested should not make an otherwise reasonable search unreasonable. The estimate of -.57 is statistically insignificant. Yet if this is the true value, it would have a sizeable affect on the reasonableness of a search. A search that had a prob- ability of .50 of being reasonable without an arrest has only a probability of .28 of being reasonable if preceeded by an unlawful arrest. The other exceptions mentioned are all relatively rare, but substantively significant events. There were thirteen cases that involved permission, ten that involved statutory regulation, six each for administrative and grand jury or civil hearings, five involving plain view, two involving hot pursuit and two involving searches at fixed or functional 98 (e.g., airports) borders. Since (1) the Supreme Court has never expressed a hierarchy of exceptions, (2) the substan- tive significance of each of these particular variables se- parately is not extremely crucial to the study and (3) the low n's would lead to extremely high standard errors, it was decided that a general exception category, measured by the number of exceptions present in each case, would be prefer- able to estimating each parameter separately. This was done in full knowledge that some of subtle or not so subtle differences between the variables would be masked. 0n bal- ance though, it was felt a relatively precise measure of "exceptions" would better serve our understanding of search and seizure than would extremely imprecise estimates of each exception individually. The evidence from the cases supports the notion that. the Supreme Court treats these exceptions rather seriously. The Court found the searches in question to be reasonable 45% of the time that no exceptions were present, 67% of the time that one exception was present and 100% of the time that two or more exceptions were present. The probit estimate, 1.32, is quite consistent with our expectations. Controlling for other facators, adding one exception could increase the probability of the Court finding a search rea- sonable from .5 to .91. Two exceptions could raise the probability from .5 to .997. Similarly, we could start with a more unreasonable search, one in which the probability of finding the search reasonable was .10. Adding one exception 99 would raise the probability of finding the search reasonable to .53, while adding a second would raise the probability to .93. Once again, at the risk of sounding repetitive, the Court is enunciating Constitutional doctrine, and then fol- lowing it. The confusion that exists seems to emanate from the method in which we have studied the decisions of the Court, not from the decisions themselves. Finally, we come to our last variable, change. The early attempts at quantitative prediction of Supreme Court decisions (Kort, 1957, 1963, 1968, 1973; Schubert, 1959; and Ulmer, 1964) universally included no measurement that ac- counted for change in the personnel of the Court. Kort (1957), for example, in his right to counsel cases, exclaims that “(i)t should not be surprising that a gradually chang- ing group of justices . . .should behave according to sta- tistical regularities" (pp.11-12). Actually, it should be surprising if a gradually changing group of justices act the same way. The Supreme Court explicitly ended its case by adjudication on the right to counsel with the historic Gideon v. Wainright, 372 U.S. 335 (1963). With one or two more Reagan appointments, the exclusionary rule may become history. Even without such a dramatic change, we would ex- pect, in fact, that the change from the liberal Warren Court to the conservative Burger Court would leave the Supreme Court less likely to find searches unreasonable. What we find is that in 62% of the Warren Court deci- sions, the Court found the search at hand to be unreason- 100 able, while only 39% of the Burger Court decisions found the same. As there was not a sudden shift from the Warren Court to the Burger Court, but a gradual replacement, we would ex- pect slightly more conservative decisions with each addition of a Nixon or Ford appointee. Basically, this is what the aggregate data tends to show. Until the retirement of Warren, the Supreme Court voted liberally in 62% of the search and seizure decisions. Burger, the first Nixon ap- pointee, sat on eight search and seizure decisions before Blackmun joined the Court. The Court voted liberally in five of these, or 62.5%. A lone Nixon appointee could not make a budge in the solidly liberal Court. When Blackmun replaced the liberal Fortas, two of the four cases heard be- fore Rehnquist and Powell joined were against the defendant. This is clearly to small a sample upon which to make any inferences. When Powell and Rehnquist replaced Harlan and Black, the picture begins to clear. Three-quarters of the decisions made were in favor of the government, and against the claim that evidence was obtained illegally. The final change, Stevens, a moderate, for Douglas, the liberal, did not push the percentage of cases found unrea- sonable any higher. In fact, only sixty percent of the de- cisions were in favor of the government. At least two ex- planations are possible. The easiest to reject is that the Court became more liberal with the addition of Stevens. Given that Stevens is solidly a "moderate“ (Spaeth, 1979), and Douglas was perhaps the most liberal of the 101 Warren/Burger Court justices (Ibid.), this is quite unlike- ly. More likely, the Court was hearing different cases be- fore Douglas retired than afterwards. Lower courts, usually unwilling to shift direction until a clear change from the Supreme Court is signalled, may have been slow in responding to the new, more conservative decisions of of the Burger Court. Add to that the time lag between lower court deci- sions and those of the Supreme Court, and it may be that many cases in which the lower courts found searches to be unreasonable were extremely reasonable within the eyes of the Burger Court. Once this pattern became clear, it could be that less cases involving "extremely reasonable" searches (within the eyes of the Supreme Court) were being declared unreasonable below. At this point, the more difficult deci- sions, involving closer judgements, could be handled. The above, though, is hopefully well reasoned, but at this point, conjecture. We need to be able to control for the facts of the case, and then see if the decisions are getting more conservative. Probit, as a multivariate tech- nique, is able to accomplish this. A variable measuring change was included in the model. It was operationalized as zero during the Warren Court, with one added to its value for each Nixon or Ford appointee. The estimate for Change was .16. Thus, The full difference between the Warren Court and the Burger Court after the ap- pointment of Stevens is .80 (.16x5). A search that for the Warren Court had a probability of being upheld of .38 (they 102 in fact upheld 38% of the searches before them) would have a probability of .69 of being upheld by the Burger Court. A search that for the Warren Court had a probability of being upheld of .50 would have a probability of being upheld by the pre-O'Conner Burger Court of .79. It is interesting to note that a search with a warrant during the Warren Court would be almost as likely to be upheld as a similar search without a warrant by the Burger Court. The United States as a Party to the Suit. Thus far the only variables considered in the model are those that relate directly to the reasonableness of the search and the one variable that measures changes in the Court's personnel. The reasons for the former should be obvious enough. The latter is included not because it helps to gauge the reason— ableness of a search, but because it helps to gauge the Court's subjective belief as to the reasonableness of a search. It could be, however, that the Court's decisions in search and seizure cases are concerned with more than the reasonableness of the search; there might be factors unre- lated to the reasonableness of the search that nevertheless affect the Court's decisions. One such possibility is the presence of the United States as a party to the suit. Much has been written claim— ing that the Supreme Court's responsiveness to the United States as a party is a vital factor in the Court's decision to grant certiorari (e.g., Ulmer et al., 1972). Does this favorable predisposition carry over to final decisions? 103 Analysis by Spaeth and Teger (1982) shows the Court sup- porting the government 69.7% of the time in the 1966-1967 terms and 67.8% of the time in the 1969-1973 terms. Throughout these terms, Douglas was the only justice to sup- port the federal government less than half the time. The general support trend holds for criminal procedure cases. These and other studies showing support for the federal go- vernment as opposed to its adversaries are enlightening and might lead us to believe that in our search and seizure cases the Court may, on the whole, be predisposed to favor the government. But, before we can use this information to help us explain specific Court decisions, we must have a point of comparison. In our population of search and sei- zure cases, those cases that do not involve the federal go— vernment do involve the state government. Our question is not whether the Supreme Court is more favorable to the U. S. than its adversaries; rather, we are concerned with whether the Supreme Court is more favorable to the U. S. than it is to the states. That is, overall, the United States is more likely than not to win. What is of interest is whether the United States is more likely to win than are the states. The model was rerun including a dummy that measured when the U.S. was a party to the suit. The results for the second model are presented in Table 5.2. Although the esti- mated R2 barely improved and the percent categorized cor- rectly did not improve, the overall fit of the model, as 104 TABLE 5.2 Probit Estimates with United States as Party Variable MLE S.E. MLE/S.E. House -2.12 .62 -3.42** Business -2.07 .64 -3.22** Car -1.87 .64 -2.91** Person -1015 055 -2011* Search - .91 .46 -1.98* Warrant .94 .46 2.03* Probable Cause - .30 .39 - .75 Incident Lawful Arrest 2.30 .77 2.98** After Lawful Arrest 1.34 .52 2.59** After Unlawful Arrest - .33 .62 - .54 Exceptions 1.23 .35 3.56** Change .16 .07 2.28* United States .59 .30 '1.97* Constant 2 1.06 mean .55 Estimated R .58 % correct 76 -2xLLR 57.51** n=123 * sig. at .0547 ** sig. at .01 105 measured by -2xLLR, is a statistically significant improve- ment over the fit of the first model (see Aldrich, 1977). Furthermore, the point estimate of .59 is of moderate size and is statistically significant. A state search that had a probability of .50 of being upheld would have had a prob- ability of .72 of being upheld if it had been a federal case. What is so important about this finding is that it is the result after the facts of the case have been taken into account. The preference given to the United States does not appear to be the result of federal searches being less se- vere than state searches. ' Predicting from the Model: The 1981 Term. It is one thing to fit a line to previous behavior; it is quite another if that line fits future behavior as well. The original model was applied the search and seizure decisions of the October 1981 term. This set involves seven different decisions over five different cases. (Two of the cases had two distinct decisions.) Five of the decisions were conservative, two were liberal. The model predicted all five of the conserva- tive decisions correctly and one of the two liberal deci- sions. The incorrect prediction involved the admissability of a voluntary confession made after an unlawful arrest. To the 5-4 majority, the confession could not be admitted since it was the result of an unlawful seizure. Although we are dealing with a limited population, the fact that six of sev- en cases were predicted correctly does give hope the the model has predictive as well as explanatory value. 106 Discussion The results are very pleasing.. It seems as if the mod- el has taken one of the Court's most difficult areas, one characterized as "inconsistent," "bedeviling" and even "a mess" and found for the first time a clear and logical form. Perhaps the reason is that those who have been bedeviled have been too reliant on the case method as the means for analyzing these decisions. Such reliance seems reasonable enough. The opinion delivered in any given decision more often than not relies upon a single aspect of the case, e.g., was there probable cause? or is a search several hours after a lawful arrest reasonable? Yet the decision itself is dependent upon many factors-—presumably the ones I have mentioned above-—that go undiscussed in the main body of the opinion. It is well and good to know in a probable cause determination that probable cause was sufficiently esta- blished, but when it comes to understanding the decision it- self we must know where the search took place, the extent of the intrusion, if any exceptions were present as well as the makeup of the Court. The parameter estimates are certainly interesting, but great care must be taken in interpreting them. Since dummy variables were used it is important not to compare the mag- nitude of the coefficients without considering the baseline from which they were measured. For the prior justification variables, the arrest variables and the exceptions variable, the baseline was zero. When "probable cause“ had the value 107 zero, there was no probable cause; when the "arrest" vari- ables all had the value zero, there was no arrest. Compare this to the "search“ variable. When the search variable was zero, there was no search, but there still was some form of intrusion. If we were measuring this from a baseline of no intrusion the value would be very large, for if there were no intrusion, it would be extremely doubtful that the search could be unreasonable. Alternatively, consider the vari- ables that measure where the search took place. They all take on the value zero when the search took place in an area over which one does not have a prOperty interest. The value for "house" is not so large because one's house receives ex- traordinary amounts of protection; it is so large relatively speaking it receives tremendously more protection than areas over which one does not have a property interest. Notwithstanding the above, the parameter estimates tell us quite a bit about the Supreme Court's search and seizure decisions. Although one's house seems to get more protec- tion than one's business, and one's business gets slightly more protection than one's car, the differences are not as substantial as one might have guessed. A search of a house that has a probability of .50 of being upheld would have a probability of .61 of being upheld if it instead were a search of a car. This certainly affects the reasonableness of the search, but compared to other factors that can inter- vene, it is far from determinative. Other points might seem to belabor the obvious-- it is better to have a warrant than 108 not to, full searches are inherently less reasonable than a st0p and frisk--yet these points have not heretofore been demonstrated through multivariate analysis. More important though is the evidence that Supreme Court decisions, though no two are exactly alike, are succeptible to this type of study. This is true despite the Court's assertions that the reasonableness of a search and seizure is not determinable by mathematical formulations. The Justices The Supreme Court is a collegial body. Whichever side, that of the appellant or that of the appellee, gets at least five votes wins. It is the votes of the justices that de- termine the decisions of the Court. One of the advantages of the theoretical and methodoligical framework used herein is that it is consistent not only with the decisions of the Court, but with the individual justices as well. The pur- pose of this section, then, is to examine how the justices voted. The results of running the model on each of the jus— tices who have heard over fifty search and seizure cases be- tween 1963 and 1981 are presented in Table 5.3, below. The most immediate thing one notices when looking at the results for the individual justices are the occasional extremely high point estimates and extraordinarily high standard errors. For example, the estimate for Marshall for After Unlawful is -35.34, stating that in searches involving M'EF’“-W . - 109 TABLE 5.3 Probit Estimates for Individual Justices: Douglas Variable MLE S.E. MLE/S.E. House -1065 075 -2021* Business -l.29 .80 -1.61 car -1029 087 -1048 Person - .71 .67 -l.06 Search .56 .69 .81 Warrant .32 .64 .49 Probable Cause .28 .65 .44 Incident Lawful Arrest -6.46 7235.08 - .00 After Lawful Arrest .24 .74 .32 After Unlawful Arrest -34.22 15741.93 - .00 Exceptions .88 .42 2.10* United States - .07 .41 - .18 Constant -’i76 mean .17— Estimated R2 .997 % correct 88 -2xLLR 21.23* n=68 * 519. at .05 ** sig. at .01 110 TABLE 5.3 contd. Probit Estimates for Individual Justices: Bre nnan Variable MLE S.E. MLE/S.E. House -1007 054 -1099* Business -1.03 .60 -l.73* Car "’ 098 060 ’1064 Person - .66 .49 -1.33 Search - .43 .43 -1.01 Warrant .13 .43 .30 Probable Cause .41 .40 1.02 Incident Lawful Arrest .29 .66 .44 After Lawful Arrest - .01 .57 - .01 After Unlawful Arrest -33.79 9777.63 - .00 Exceptions .63 .25 2.52** United States .37 .30 1.24 Constant - .18 mean .22 Estimated R2 .99 % correct 78 -2xLLR 21.08* n=ll9 * 819. at 00377 ** Sig. at .01 111 TABLE 5.3 contd. Probit Estimates for Individual Justices: Marshall Variable MLE S.E. MLE/S.E. House - .92 .65 -l.4l Business - .70 .65 -1.08 Car - .58 .68 - .86 Person - .29 .56 - .52 Warrant - .03 .48 - .07 Probable Cause .50 .41 1.24 Incident Lawful Arrest .12 .85 .14 After Lawful Arrest - .41 .65 - .64 After Unlawful Arrest -35.34 11878.53 - .00 Exceptions .50 .29 1.71* United States .02 .34 .04 Constant' - .57 mean .20 Estimated R2 .99 % correct 80 -2xLLR 11.44 n=94 * sig. at .057 ** sig. at .01 112 TABLE 5.3 contd. Probit Estimates for Individual Justices: Stewart Variable MLE S.E. MLE/S.E. House - .73 .55 -l.33 Business -l.32 .55 -2.40** Car -l.25 .59 -2.12* Person —l.16 .52 -2.24* SearCh - 094 045 “2011* Warrant .37 .40 .94 Probable Cause - .73 .55 -1.32 Incident Lawful Arrest 1.38 .67 2.05* After Lawful Arrest .00 .45 .00 After Unlawful Arrest .20 .53 .38 Exceptions .51 .25 2.03* United States .53 .27 1.97* Constant 2 1.56 mean .58 Estimated R .34 % correct 73 -2xLLR 30.89** n=121 * sig. at .05 ** sig. at .01 113 TABLE 5.3 contd. Probit Estimates for Individual Justices: Wh ite Variable MLE S.E. MLE/S.E. House - 099 056 ‘1078* Business -l.54 .57 -2.70** Car -1.44 .61 -2.34* Person -1.09 .53 -2.04* Warrant .39 .41 .96 Probable Cause .13 .35 .38 Incident Lawful Arrest .92 .63 1.45 After Lawful Arrest 1.04 .48 2.18* After Unlawful Arrest .54 .56 .96 Exceptions .88 .31 2.87** United States .58 .27 2.11* Constant 1.27 mean .63 Estimated R .36 % correct 70 -2xLLR 28.26** n=ll9 * sig. at .05 ** sig. at .01 114 TABLE 5.3 contd. Probit Estimates for Individual Justices: Burger Variable MLE S.E. MLE/S.E. House - 026 094 " 027 Business -l.48 .92 -1.61 Car -1.34 .84 -l.59 Person - .74 .73 -1.01 Warrant 6.65 4769.31 .00 Probable Cause .19 .52 .37 Incident Lawful Arrest 29.69 13344.04 .00 After Lawful Arrest 1.30 .76 1.70* After Unlawful Arrest - .25 .70 - .36 Exceptions .97 .49 1.99* United States .27 .42 .65 Constant 2 1.27 mean .79 Estimated R .98 % correct -2xLLR 21.79* n=81 * sig. at .05 ** sig. at .01 115 TABLE 5.3 contd. Probit Estimates for Individual Justices: Blackmun Variable MLE S.E. MLE/S.E. House -1.92 .92 -2.10* Business -l.56 1.02 -1.52 Car - 047 092 OS]. Person - .52 .75 .70 Search - .07 .65 .11 Warrant 1.15 .78 1.47 Probable Cause .66 .68 .97 Incident Lawful Arrest 24.32 11993.48 .00 After Lawful Arrest 12.10 7923.65 .00 After Unlawful Arrest - .37 .75 .49 Exceptions 1.25 .61 2.05* United States .86 .50 1.71* Constant .62 mean .80 Estimated R2 .98 % correct 80 -2xLLR 25.62* n=75 * sig. at .05 ** sig. at .01 116 TABLE 5.3 contd. Probit Estimates for Individual Justices: Powell Variable MLE S.E. MLE/S.E. House -1.77 .91 -1.94* Business -l.59 .96 -l.66 Car -1.53 .88 -l.75* Person - .88 .80 -1.10 Search - .62 .60 -l.03 Warrant 1.20 .75 1.59 Probable Cause .57 .54 1.06 Incident Lawful Arrest 25.17 13023.81 .00 After Lawful Arrest 1.33 .80 1.67* After Unlawful Arrest .82 .85 .96 Exceptions 1.20 .51 2.35* United States .31 .44 .71 Constant 1.03 mean .69 Estimated R .97 . % correct 81 -2xLLR 24.50* n=67 * sig. at .05 ** sig. at .01 117 TABLE 5.3 contd. Probit Estimates for Individual Justices: Rehnquist Variable MLE S.E. MLE/S.E. House -1.10 7204.83 .00 Business -7.23 5394.61 .00 Car " 052 1021 043 Person - .81 .89 .91 Search .03 .84 .04 Warrant 6.13 4599.60 .00 Probable Cause 35.09 4589.60 .01 Incident Lawful Arrest 12.46 7934.13 .00 After Lawful Arrest 5.70 5062.16 .00 After Unlawful Arrest .04 .97 .05 Exceptions 6.54 5394.61 .00 United States - .28 .99 .29 Constant .95 mean .90 Estimated R2 .997 % correct 90 -2xLLR 21.23* n=68 * sig. at .05 ** sig. at .01 118 unlawful arrests 35.34 is subtracted from the cumulative normal in determining the probability of finding a search reasonable. To keep this somewhat in perspective, a Z of -3.53 would make the probability of finding such a search reasonable .0002. At -35.34, the probability of finding such a search reasonable is too small to be called infinite- simal. 0n the other hand, the standard error is a gargan- tuan 11,878.53, suggesting that the true coefficient could be anywhere; -35.34 is, given the data, just more likely than any other value. Before any further analysis can be done results like these for Marshall and his brethren must be explained. The reason results like this occur is that for the po- pulation of cases before us there are examples of extremely “rigid" decision making. There were only nine cases in which evidence had been seized after what the lower court considered to be an unlawful arrest. In each of those cases Marshall voted liberally. Analagously, in all five cases during Burger's tenure in which evidence had been obtained incident to what the lower court considered to be a lawful arrest Burger voted conservatively. In attaing the best fit to the data possible, the probit formula will give such highly negative or highly positive values because in the data there are neither examples of Marshall voting conserva- tively in After Unlawful searches nor examples of Burger voting liberally in Incident Lawful searches. Thus, for a search in the data set, a .9999 probability of Burger find- 119 ing an Incident Lawful search reasonable is more accurate than a .99 probability. Therefore, the statistical fit is improved by the extraordinary estimates. An additional consequence of this is that the estimated R will be uncom- monly high since, in many of the cases, virtually all of the variance from the mean has been explained. Statistically then, the results can be explained. Do the large estimates make sense theoretically? Briefly, I believe that they do. In Chapter 3 it was proposed that Supreme Court decision making resembles cybernetic decision making in that the justices take short cuts from rational decision making; they monitor a few facts from the lower court decisions and give somewhat standard responses based on these facts. The facts do not cause the justices to vote one way or another, but rather, strongly predispose them in a given direction. Since some of these facts seem as if they might be sufficient conditions for a vote in a given direction for some of the justices, the theory is supported. It could in fact be claimed that some of the facts do more than predispose: they cause. I would be unwilling to go that far. Rehnquist has voted conservatively in all 24 cases in which the lower court found probable cause. The point estimate is 14.67. Yet, I do not feel it can be con- cluded that he would always do so. Given that he votes con- servatively 90% of the time there is, according to binomial probability tables, a .07 probability that given a random sample of 24 cases, none would be decided liberally. Though 120 it would make little sense to conclude therefore that the lower court's decision on probable cause does not affect his decision on the reasonableness of a search, it is by no means certain that the lower court's probable cause determ- ination necessarily causes his own decisions on the merits. Substantively, I am not convinced that such a lower court determination places Rehnquist 14.67 standard deviations above zero on the cumulative normal. A single case could dramatically lower that figure. The existence of such a case is, in my mind, entirely conceivable. At this point the basic findings can now be discussed. In terms of the summary statistics, the model did reasonably well. It explained a statistically significant amount of variance for eight of the nine justices. It was only for Marshall that the null hypothesis could not be rejected. The estimated R2 ranged from .997 for Rehnquist to .34 for Stewart and .36 for White. As noted above, however, the es- timated R2 is so high because of the rigid decision struc- ture that most of the justices show for at least one vari- able. A single case of Rehnquist voting liberally when the lower court found probable cause would drop the estimated R2 tremendously. It is only for Stewart and White that con- fidence can be placed in the stability of all the estimates and in the amount of variance explained by the model. Though their estimated R is much lower, values in the .30- .40 range are certainly acceptable. It should be kept in mind that the model was specified so as best to fit the 121 Court's decisions, not any individual justice's. Running different models tailored for each justice almost certainly improve the fit. The percent correctly categorized likewise seem rather high. White was the most unpredictable justice; only 70% of his decisions were accurately gauged. The most predictable judges was Rehnquist, upon whom the model was accurate 90% of the time. Prediction rates, though, are not necessarily an accurate measure of how well the model predicted. It is crucial to examine prediction rates relative to the mean of each justice. The mean of Douglas, for example, is .17. This states that Douglas voted conservatively in 17% of the cases. If we had no information as to what the facts of the case were, we could still predict Douglas' decisions with 83% accuracy just by guessing each time that he would vote liberally. Probit, a multivariate technique, can only im- prove on this rate. Thus, the 86% prediction rate for Douglas is high, and for practical purposes is extremely useful. The model, though, did not contribute tremendously to that result. One method by which one can account for the mean is to determine the percent reduction in error. By only knowing the mean in the case of Douglas one would be in error 17% of the time. With the model, one's rate of error is reduced to 14%. The percent reduction in error is 3 divided by 17, or 18%. Of all the justices, the largest reduction in error was attained for Powell. With a mean of .69, we would ex- 122 pect no more than 31% errors, given no other information. With the model, the prediction rate jumps to 81%, leaving only 19% errors. The reduction in errors is 39%. The low- est reduction of errors were obtained for Brennan, Marshall, Blackmun and Powell. For none of them could the model im- prove upon the mean. The scores for the remaining justices were 34% for Stewart and 7% for Burger. The parameter estimates for the individual justices did not come out as well as they had for the Court as a whole, but they did not come out too poorly either. The biggest disappointment was in terms of statistical significance. The only variables that were significant for over half the justices were House (five of nine) and Exceptions (eight of nine). Neither Probable Cause nor Warrant were significant for any of the justices. Looking at this from the point of View of the justices, Stewart and White each had seven of their twelve estimates significant. But, for the remaining justices, not one had more than three significant estimates. Burger and Rehnquist had none. This is not, however, as en- tirely bad as it seems. The extremely large estimates that result from the rigid decision making leave tremendously larger standard errors. The variables are nevertheless clearly important. If even a single exception occured the estimates would rapidly approach significance. Under these rather odd circumstances, if the estimates were less subs- tantively significant they would be more substantively sig- nificant. Moreover, the fact that it is Stewart and White, 123 and then Brennan, i.e., those justices who have served the longest, that have the largest number of statistically sig— nificant estimates gives rise to the belief that the insig- nificance for the others may often be the result of too few cases having been decided. Finally, it should again be noted that the model was specified using those variables the Court found to be important. There is no reason to expect that this means that each justice finds each variable to be consequential. The signs of the parameter estimates were almost en- tirely in the expected direction; eighty-two of the ninety were correct. (Note: this does not include the Unlawful and ELEL variables, for which, respectfully, the expectation was zero and no prediction had been made.) Of the eight that were not in the expected direction four were practically zero. Two of the remainders belonged to Douglas, one to Marshall and one to Stewart. Of particular note among the Arrest variables was that for each of the four Nixon appointees every lower court finding of a search incident to a lawful arrest was followed by a conservative vote. For the three liberals on the Court, Brennan, Douglas and Marshall, every lower court finding of a search after an unlawful arrest was followed by a liberal vote. Not surprisingly, the balance was in the hands of Stewart and White. Probable Cause was in the expected direction for eight of the nine justices in this part of the analysis, as was 124 Warrant. Rehnquist voted conservatively in every case in which the lower court found probable cause, and Burger did the same in every case in which there was a warrant. All of the place variables were in the expected direc- tion, showing the universal acceptance of the need for a prOperty interest in the place to be searched. House was more negative than Business for five of the nine justices and more negative than Person and Car for six of the nine. Business unanimously higher than Person and Car. The Search variable was negative for all except Douglas, who dissented to and never accepted the logic of Terry v. Ohio, 392 U.S. 1 (1968). The ELBL variable was po- sitive for Brennan, Stewart, White, Powell, Blackmun and Burger, and statistically significant for Stewart and White. For Douglas and Marshall it was about zero, and for Rehnquist it was slightly negative. The Exceptions variable proved to be the most hardy of them all. It was positive for each justice and statistically significant for all ex- cept Burger and Rehnquist. If the votes of any given justice were the votes of the Court itself, there might be some cause for alarm. With the possible exception of Douglas none seem inexplicable, yet none seem entirely consistent either. Nevertheless, out of this morass we find for the Court a unified, consistent and logically sound set of decisions. The whole truly seems to be greater than the sum of its parts. CHAPTER 6 SUMMARY AND CONCLUSIONS Let me review what heretofore has been said. The Fourth Amendment is "one of the laws most vital to our lib- erty," (Amsterdam, 1974:377) yet we are still not sure what makes a particular action by the government an "unreasonable search or seizure." The Fourth Amendment cases have been considered "inconsistent, Coolidge V. New Hampshire, 403 U.S. 443 (1971), "bedeviling" (LaFave, in Amsterdam, 1974: 349) and even a "mess" (Dworkin, 1973: 329). The reason for this, I felt, was not in the decisions themselves, but in the ways the decisions had been studied. Quantitative studies lacked any theoretical perspective and were mired with methodologies that were and still are unjus- tifiable. The qualitative/case study techniques suffered from an overreliance on precedent, too many exceptions and/or unwieldy classifications. Since "the object of our study... is prediction,‘ (Holmes, in Murphy and Pritchett, 1979:19-20) it was felt that a renewed effort at deVising a means by which Supreme Court decisions could be explained and predicted would be well worthwhile. While chapter two reviewed the history of search and seizure, Chapter Three develOped a model of Supreme Court decision making. First it was explained that Supreme Court justices are free to vote according to their policy prefer- 125 126 ences. Neither the Constitution nor precedent are effective legal limits on their power; neither Congress nor the President are effective political limits on their power. Assuming that Supreme Court justices receive some meas- ure of utility for various outcomes, it was at first sug- gested that the justices might be viewable as utility maxi— mizers. Unfortunately, there were problems with that view in regards to decision makers in general and with regards to Supreme Court justices in particular. Evidence about the Court's workload, the predictability of Supreme Court deci- sions and one or two empirical studies suggest that the cy- bernetic theory of decision making more closely fits the justices' decision making calculi. The justices, it was as- sumed herein, monitor certain facts from the lower court opinions that strongly predispose them toward finding a search reasonable or unreasonable. A model using the deci- sions of the Court and the individual justices as to the reasonableness of a search and seizure as the dependent variables, and the facts of the case (e.g., whether there 'was a warrant, where the search took place) as the independ- ent variables was run. The parameters _were estimated through the multivariate technique probit. The data were the Supreme Court search and seizure decisions from the Fall of 1962 through the Spring of 1981. The results for the decisions of the Court itself were quite pleasing. The model fit quite well. The estimated.R3 was .57, and 76% of the decisions were categorized correct- 127 ly. Moreover, the parameter estimates were almost univer- sally as expected. In terms of the of the place of the search, it was shown that the most protection was granted one's house, followed by one's place of business, one's car and one's person. Areas over which one did not have a prop- erty interest received virtually no protection. Further, it was demonstrated that lesser intrusions, such as a stop and frisk, are inherently more reasonable than full searches. The existence of a warrant was shown to be quite important, but unfortunately, the lower court's decision on probable cause was insignificant. As expected, the right to search incident to a lawful arrest was a strong exception to the probable cause and war- rant requirements. Decidedly less leeway, but still a sig- nificant amount, was granted searches after but not incident to a lawful arrest. Searches after an unlawful arrest seem to make searches slightly less reasonable. The estimate, however, was statistically insignificant. Other exceptions, such as border searches and searches of objects in plain view, were grouped together into a single Exceptions vari- able. This proved to be moderately strong and statistically significant. A variable was included that measured change in the Court's membership. As the Warren Court gradually became the present day Burger Court, it was expected that the Court's decisions would have gotten more conservative even after controlling for the facts of the case. This was in 128 fact what was shown. Finally, the model was rerun including one additional variable: whether the United States was a party to the suit. Despite claims by the Court that it treats searches by the states the same as searches by the federal government, it was found that the Court was significantly more likely to find federal searches reasonable, even after the facts of the case were controlled. In order to generate an early gauge of the model's predictive ability, the results were applied to the Fall, 1981 term. Six of seven cases were predicted accurately. The results for the individual justices were not quite as pleasing. Although the estimated Rz's ran from .34 to .997, the extremely high scores were to a large extent sta- tistical aberrations. The percent categorized correctly ranged from 70% to 90%, but for no justice was the percent reduction in error higher than 39%. For four of the jus- tices the percent reduction in error was zero. The parame- ter estimates were almost universally in the expected direc- tion, and the relative magnitude was more often than not as hypothesized. Nevertheless, very few of the variables at- tained statistical significance. Much has been learned from this study. By combining substantive knowledge of the search and seizure decisions with a theoretical framework for understanding those deci- sions and a proper methodolOgical approach, order has ap- peared where there once was chaos. Substantively, it is 129 fairly clear what tends to make a search reasonable or un- reasonable. The framework of nature of the intrusion (place and extent), prior justification (probable cause and war- rant) and exceptions (arrest and others), plus change in the Court's membership appears to fit the cases quite well. If it was not known prior to this study what makes a search reasonable or unreasonable, I believe it is known now. Theoretically, it does seem that the justices follow somewhat simple decision rules in determining the reason- ableness of a search or seizure. Burger, Blackmun, Rehnquist and Powell all voted conservatively every time the lower court found a search occuring incident to a lawful arrest. Brennan, Douglas and Marshall each voted liberally every time the lower court found a search was after an un- lawful arrest. Burger voted conservatively every time there was a warrant. Rehnquist did the same every time the lower court found probable cause. Methodologically, it was shown that the Court's deci- sions can be understood through multivariate analysis, de- spite various assertions by the court and legal scholars that every case is different. Probit, the estimation proce- dure used herein, is theoretically more appropriate than any of the objective quantitative techniques used to predict ju- dicial decisions. The results were consistent enough with prior expectations to make it seem substantively useful as well. 130 Policy Implications Although this dissertation was written with the purpose of enhancing our theoretical understanding of Supreme Court decision making, knowledge does not and should not exist in a vacuum. When evidence is illegally obtained it has been the policy of the United States Supreme Court to exclude that evidence in federal trials, Weeks v. United States, 232 U.S. 383 (1914), and at state trials, Mapp v. 9332, 367 U.S. 643 (1961). This exclusionary rule has never been very pOpular. To repeat what Judge Cardozo said, "(t)he criminal is to go free because the constable has blundered" Peo le X; Defore, 242 N.Y. 13 (1926). The rule has come under increasing attack since the Mapp_ decision. Among the arguments against it is the view that it can not deter because Fourth Amendment law can not be understood. According to Chief Justice Burger: "Whatever educational effect the rule conceivably might have in theory is diminished in fact by the realities of law enforcement work. Policemen do not have the time, inclination or training to grasp the nuances of the appellate Opinions that ultimately define the standards of conduct they are to follow. The issues that these decisions resolve often admit of neither easy nor obvious answers, as sharply divided courts on what is or is not 'reasonable' amply demonstrate. Nor can judges, in all candor, forget that opinions some- times 1ack helpful clarity." Bivens v. Six Unknown Federal Agents, 403 U.S. 388 (1971)(dissentlng). This point is particularly relevant to the Court's cur- rent consideration of the so called "good faith" exception to the exclusionary rule. According to Schlesinger (1977): 131 "the rule does not distinguish between a willful and flagrant violation by an officer, and an exer- cise, under strained circumstances, of his best judgement, a judgement which, only after several appeals, may be held by a perhaps divide appellate court to have been improPer. The issues which ap- pellate courts resolve are both controversial and difficult, as evidenced by courts sharply divided on ue of what is a reasonable search and seizure. After all, if appellate judges are not sure about the legality of an officer's action long after consideration of a written record, it is asking a great deal of the officer that he judge the matter correctly in a few moments. Nor can judges be oblivious to the fact that their opinions some- times lack the kind of helpful guidance which will aid a law enforcement officer in the actual per- formance of his duty. Police officers should not be deterred from exercising 'reasonable judge- ment,‘ though that judgement may not represent the final judgement of the appellate courts as to the proper behavior in that situation" (p.62). Yet, contrary to popular belief, the Supreme Court's search and seizure decisions do make sense. The point of this is obviously not that police officers should carry cu- mulative normal probability tables with them at all times. However, they can and ought to be able to sensibly consider whether the factors present in a given situation would tend to make a given search reasonable or unreasonable. What those factors are should by now be fairly clear. This study should somewhat undermine the type of arguments against the exclusionary rule presented above. Limitations The type of study conducted herein has been very useful in helping to explain (or retrodict) and predict Supreme Court decisions. There were, however, both potentially so- luble and necessarily insoluble problems with the explana- 132 tion of previous decisions, and certain inherent limits in applying this sort of model to future decisions. The major potentially soluble problem in terms of ex- planation was the Operationalization of subjective Vari- ables. As has been stated several times, the Supreme Court's decisions on whether an arrest was lawful, whether a warrant was valid and whether there was probable cause can not be included in the model as independent variables. As such, indicators had to be used, and when indicators are used there will be error in the measurement of what is ac— tually sought. So, rather than measuring probable cause, the lower court's decision on probable cause was measured. Rather than measuring whether there was a valid warrant, the existence of a warrant was measured. Rather than measuring whether an arrest was lawful, the lower court's decision on whether an arrest was lawful was used. The greatest problem was in measuring probable cause. The estimate for the lower court decision was not only sta- tistically insignificant, but it was in the wrong direction. The Supreme Court voted conservatively 48% of the time the lower Court found probable cause and 59% of the time it did not. Yet the Court voted conservatively 65% of the time the Court itself found probable cause and 53% of the time it did not . The indicator for a valid warrant, the existence of a warrant, was not a very reliable indicator either. Of the eighteen cases involving warrants, in only five did the 133 Court find the warrant to be valid. The value of the Warrant variable was severely attenuated by the thirteen in- valid warrants, which should not be able to add to the rea- sonableness Of a search. The Court voted to allow the evidence to be used in all five of the cases involving valid warrants and in five of the thirteen cases involving invalid warrants. Possible solutions for these measurement problems will be discussed below in the suggestions for future re- search. The second major limit to the explanation of the deci- sions involves the results for the individual justices. Although much was learned (see Chapter V), any inferences that can be made must be done with the delicacy due statis- tically insignificant estimates. Further, although most of the estimates are in the expected direction and many are of the expected relative magnitude, several are not. A truly coherent View of the search and seizure decisions does not apppear from an analysis of the individual justices. It would certainly be useful in the future to try to find di- sparate models to fit their own idiosynchratic decision functions. These are problems that hopefully can be solved. A li- mit that this model can not take into account is that it re- lies solely on the decision of the Court as its unit of analysis, ignoring entirely the Opinion of the Court. Often the opinion tells us far more than the decision. A recently decided case points this out dramatically. In Illinois v. 134 Gates (1983) the Supreme Court upheld the search of Gates' car and house. The police had first obtained a warrant, based on information from an anonymous tip. The lower court overturned the search, citing the Supreme Court decision in Aguilar v.Texas, 378 U.S. 108 (1964), which ruled that prob- able cause could not be established for the validity of a warrant when the source of information was anonymous and there was no concrete evidence to support the claim. In admitting the evidence in Gates the Court had two options open to it. First, it could have overturned Aguilar and ruled that probable cause had been established. Second, it could have, as many Court watchers expected, upheld Aguilar but nevertheless created a good faith exception to the exclusionary rule. The former, the route the Court ac- tually took, was in fact a minor decision. The latter route, if it had been accepted, would have amounted to a ma- jor revision of Fourth Amendment law. The original model correctly predicted that the search of the car would be upheld (p=.58) but incorrectly predicted that the search of the house would be invalidated (p=.46). The model with the U.S. as a party yielded similar results. The important point is that even if both predictions had been right there is still no way the model can tell us what was truly crucial in the case: whether the Court would adopt the good faith exception. The model, quite obviously, can only predict decisions, not doctrines. 135 "The object of our study" said Holmes, "is prediction" (in Murphy and Pritchett, 1979:21). It is also one of the goals of this study. Yet most of the work herein is ex- planation, or retrodiction. The limited attempt at predic- tion worked reasonably well, but it is important to point out that studies like this, like most studies of human be- havior, will not be able to predict the future as well as it can retrodict the past. One reason for this is that the es- timates generated are the estimates of the true parameters that maximize the likelihood that the given sample of data would have been observed. So even if the present data were a random sample of all past and future search and seizure decisions, prediction would not work as well as retrodic- tion. The data, of course, are not a random sample. It is the population of cases from 1962 until 1981. Decisions in 1982 and 1983 are far more like decisions in 1980 and 1981 than they are decisions in 1963 and. 1964. Some of the differences can be controlled by the dynamic element of the model, the Change variable, but others can not. New doc- trines which can lead to different decisions can not be predicted and thus can not be taken into account. Furthermore, new members sometimes have predictable effects (Burger and Rehnquist for example) but sometimes do not (Warren and Brennan). Predicting the effects of future ap- pointments is far more difficult than modeling the effects of past appointments. Overall, that the past can be modeled successfully gives hope that the future can be, but it does 136 not guarantee it. Suggestions for Future Research Quite a bit has been learned about the Supreme Court's search and seizure decisions. What makes a search reason- able or unreonable is specifiable, however this can not be done as for the individual justices. The most obvious sug- gestion for future research is to try to determine the deci- sion functions of each of the justices. A second suggestion within the framework of what was actually done in this work is to improve the indicators for probable cause. Kort (1973) has suggested one means by which this might be done. Kort was interested in determin- ing whether the Supreme Court accepted as fact that a de— fendant was not advised of his right to counsel in an invo- luntary 'confession case. To do so he noted if the defendant's advisement of his right to counsel was either acknowledged, not mentioned or denied in (1) the allegations of the petitioner in the transcript of the lower court re- cord, (2) the allegations of the respondent in the tran- script of the lower court record, (3) the Opinion of the lower appellate court (4) petitioner's briefs and statements to the Supreme Court and (6) the respondent's briefs and statements to the Supreme Court. An acknowledgement in each of these was coded as l, a nonmention was coded as 0 and a denial was coded as —l. A system similar to this could be used in trying to determine probable cause. 137 Other suggestions take us well beyond the sc0pe of this dissertation. First, remember that the measure of the fit of the probit model, -2xLLR, is distributed as a chi square with degrees of freedom equal to the number of independent variables. The difference in the degree of fit between two models is also distributed as a chi square, with degrees of freedom equal to the difference in degrees of freedom be— tween the two models. Aldrich (1977) has used this to test for the equality of coefficients in two models, similar to the Chow test for multiple regression (Chow, 1960). Thus, two pOpulations of cases can be compared directly. Numerous hypotheses can be tested. For example, more precise esti- mates of the change from the Warren Court- to the Burger Court could be developed. We could see whether the Burger Court is just generally more predisposed to vote conserva- tively, or if only certain aspects of the cases are treated differently. Lower court decisions can be examined to see how closely they follow Supreme Court decisions. As well, the Supreme Court's certiorari jurisdiction could be exam- ined with greater SOphistication than before. Ulmer (1972) has demonstrated a relationship between the decision to grant certiorari and the vote on the merits. Justices who vote to grant certiorari are more likely to reverse the low— er court's decision than those who do not. With some preci- sion we can estimate whether the Court would have upheld most of the cases in which it denied certiorari. Finally, what has been done in search and seizure can 138 and should be done in other substantive areas of judicial decision making. Sex discrimination and anti-trust law are two areas that I plan to investigate. Once more, Holmes, "The prOphecies of what the courts will and nothing more pretentious are what I mean by Murphy and Pritchett, 1979:21). "The object of to quote do in fact I. law (in our study, then, is prediction" (in Murphy and Pritchett, 1979:19-20). B I BLIOG RAPHY BI BLIOGRAP HY Abraham, Henry (1975). The Judicial Process. New York: Oxford University Press. Aldrich, John (1977). "Electoral Choice in 1972: A Test of Some Theorems. of the Spatial Model of Electoral Competition." Journal of -Mathematical--Sociology 5:215- 237. ——' Aldrich, John and Charles F. Cnudde (1975). "Probing the Grounds of Conventional Wisdom: A Comparison of Regression, Probit and Discriminant Analysis." American Journal of Political Science 19:571-608. Allison, Graham (1971). 'Essencew of Decision. Boston: Little, Brown and Company. Amsterdam, Anthony G. (1974). “Perspectives on the Fourth Amendment." Minnesota Law Review_58:349. Baum, Lawrence (1977). "Policy Goals in Judicial Gatekeeping: A Proximity Model of Discretionary Jurisdiction." American Journal of Political Science 21:13. Black, Duncan (1958). The Theory of Committees and Elections. Cambridge: Cambridge University Press. Canon, Bradley C. (1973-74). "Is the Exclusionary Rule in Failing Health?" KentuckyALaw Journal 62:681. Cardozo, Benjamin (1921). The Nature of the Judicial Process. New Haven: Yale UniVersity Press. Carter, Lief (1979). Reason in Law. Boston: Little Brown and Company. Chow, G.C. (1960). ”Tests of Equality Between Sets of Coefficients in Two Linear Regressions." Econometrica 28:591-605. Dworkin, Roger B. (1973). ”Fact Style Adjudication and the Fourth Amendment: The Limits of Lawyering." Indiana Law Journal 48:329. Fisher, Franklin M, (1958). "The Methodological Analysis of Supreme Court Decisions: The Use and Abuse of Quantitative Methods." American POlitical Science Review 53:321. 140 Goldberger, Arthur S. (1964). Econometric Theory. New York: John Wiley and Sons. Greenhouse, Linda (1983). "Burger Proposes New Panel to Ease High Court's Load." The New York Times February 7, 1983: Al.6+. Hanushek, Eric and John E. Jackson (1977). Statistical Methodsgfor SOcial Scientists. New York: Academic Press. Holmes, Oliver Wendell (1979). "The Path of the Law,“ in Walter Murphy and C. Herman Pritchett (ed.) Courts, Judges'and Politics (New York: Random House) pp. 19-23. Howard, J. Woodford (1982). "Query: Are Heavy Caseloads Changing the Nature of Appellate Justice." Judicature 66: Kessell, John H. (1966). "Public Perceptions of the Supreme Court." idwe§t Qggrgal of Eglitical Sgience 10:167-191. Kort, Fred (1957). "Predicting Supreme Court Cases Mathematically: Quantitative Analysis of the 'Right to Counsel' Cases.” American Political Science Review 57:1. Kort, Fred (1963). "Content Analysis of Judicial Opinions and Rules of Law," in Glendon Schubert (ed.) Judicial_ Decision- Making . Glencoe, 111.: The Free Press. Kort, Fred (1968). "A Nonlinear Model for the Analysis of Judicial Decisions." American Political Science Review 62:546. Kort, Fred (1973). "Regression Analysis and Discriminant Analysis: An Application of R. A. Fisher's Theorem to Data in Political Science." American Political Science Review 67:555-59. LaFave, Wayne (1978). Search and Seizure. Vols. 1-3. St. Paul Minn.: West PuinshingICO. Landynski, Jacob (1966). Search and Seizure and the Supreme Court. Baltimore, Md.: The’jfihns Hopkins Press. Lasson, Nelson B. (1937). The History and Development of the Fourth Ameggment to the United States Constitution. Baltimore, Md.: The Johns HOpkins Press. March, James and Herbert Simon (1958). Organization . New York: John Wiley and Sons. McCloskey, Robert (1960). _The American Sgpreme Court. Chicago: University of Chicago Press. 141 McKelvey, Richard D. and William Zavoina (1975). "A Statistical Model for the Analysis of Ordinal Level Dependent Variables." Journal of Mathematical Spciology 4:103-120. ' Murphy, Walter and C. Herman Pritchett, eds. (1979). Courts, Judges, and Politics. New York: Random House. Murphy, Walter and Joseph Tanenhaus (1972). The Study of Public Law. New York: Random House. Nagel, Stuart S. (1961). "Party Affiliation and Judge's Decisions." American Political Science Revi§w_55:843-50. Neimi, Richard and Herbert Weisberg (1974). "Single- Peakedness and Guttman Scales: Concept and Measurement." Public Choice 20: 33- Oaks, Dallin (1970). "Studying the Exclusionary Rule in Search and Seizure." University_ of Chicago Law Review 37:665. Pindyck, R.S. and D.L. Rubinfeld (1976). Econometric Models and Economic Forecasts. New York: McGraw Hill and Company. Ripple, Kenneth F. (1980). "The Supreme Court's Workload: Some Thoughts for the Practitioner." American Bar Association Journal 66: 174. Rohde, David (1972). "Policy Goals and Opinion Coalitions in the Supreme Court." Midwest Journal of Political Science 1972: 208. Rohde, David and Harold Spaeth (1976). Supreme Court Decision Making. San Francisco: W. H. Freeman and Company. Rosenstone, Steven J. and Raymond E. Wolfinger (1978). “The Effect of Voter Registration Laws on Voter Turnout." American Political Science Review 72:22-45. Schlesinger, Steven R. (1977). Exclusionary Injustice. New York: Marcel Decker, Inc. Schubert, Glendon (1959). Quantitative Analysis of Judicial Behavior. Glencoe, Ill.: The Free Press. Schubert, Glendon (1963a). “Bibliographical Essay: Behavioral Research in Public Law." American Political Science Review 57:433. Schubert, Glendon (1963b). "Prediction from a Psychometric Model." Jurimetrics 28:100-142. 142 Siebert, Frederick Seaton (1965). Freedom of the Press in England. Urbana, Ill.: University of Illinois Press. Simon, Herbert (1976). Administrative Behavior 3rd Ed., New York: Free Press. Spaeth, Harold (1979). Supreme Court Policy Making. San Francisco: W. H. Freeman and Company. Spaeth, Harold and David Peterson (1971). "The Analysis and Interpretation of Dimensionality." Midwest Journal of Political Science 15:415. Spaeth, Harold and Stuart Teger (1982). “Activism and Restraint: A Cloak for the Justices' Policy Preferences," in Steven Halpern and Charles Lamb Supreme Court Activism and Restraint. Lexington, Mass.: D. C. Heath‘and Company, pp. 277-302. Spiotto, James (1973). "Search and Seizure, An Empirical Study of the Exclusionary Rule and Its Alternatives." Journal of Legal Studies 2:243. Steinbruner, John D. (1974). Theory of Decision. Princeton, N.J.: Princeton U Press. Stouffer, Samuel et al. (1950). Measurement and Prediction. Princeton, NJ: Princeton University Press. . Tanenhaus, Joseph (1966). "The Cumulative Scaling of Judicial Decisions." Harvard Law Review 79:1583. Tanenhaus, Joseph, Marvin Schick, Matthew Muraskin and Daniel Rosen (1963). "The Supreme Court's Certiorari Jurisdiction: Cue Theory," in Glendon Schubert (ed.) Judicial Decision Making. Glencoe, 111.: The Free Press. Tate, C. Neal (1981) "Personal Attribute Models of Voting Behavior of the U.S. Supreme Court Justices.“ American Political Science Review 75:355-367. Teger, Stuart (1977). “Presidential Strategy for the Appointment of Supreme Court Justices.“ Public Choice 31:1. Ulmer, Sidney (1964). "Discriminant Analysis and an Error Criterion," in Glendon Schubert (ed.) Judicial Behavior. Chicago: Rand McNally and Company. Ulmer, S. Sidney (1969). "The Discriminant Function and a Theoretical Context for its Use in Estimating the Votes of Judges," in Joel Grossman and Joseph Tanenhaus (ed.) Froniers 'of Judicial Research. New Yorszohn Wiley and Sons. 143 Ulmer, Sidney (1972). "The Decision to Grant Certiorari as an Indicator to Decision 'On the Merits.'“ Polity 4:429. Ulmer, Sidney (1973). "Social Background as an Indicator to the Votes of Supreme Court Justices in Criminal Cases." American Political Science Review 67:622-30. AAAAA ~'7.111111111191711111H