THE JOURNAL OF CRIMINAL LAW AND CRIMINOLQQY A non nu: CRIMINAL LAW MAY 2 3 1998 A50 GOLDEN GATE AVENUE OX 35055 (2mm OR A HATE hm: 115mm m1! Judi! A Mi Henry Wm RACE Emcrs Juvam burden-MAKING: We": on A 5?me ANALYSIS Bum-a M. Bishop Charla E. Prater ., Pm 0m AND nu: Ju?maAI. Dm?omss FOR JUVENILE: A NATURAL . 7 Je?'rey anaII J: Martin Guam Tm EN) LINE: Au Sumv? or JUDICIAL WAIVER Harry-Rasmussen Podhpacz?BanyC Felt! II - Con?nm Bomb 31' Tm 0m MARI): ADVERSE IN CIVIL Shannon Noya GUIDING 1m: ?lms Coun?s A harm or PHRASE 0mm? UNDER STATES Gm i ?113 Set]: A. Tim-I: ESSAY THE [ham or SILENCE, 11m PREsuumon or Noel-awn. 11m BURDEN or PROOF, AND A Monm PRorosAI; A Barn 1?0 Bauer! 1.. Inst-aha?! 7 BOOK REVIEW A REVIEW or THE Mm or 11mm MEMORY James E. Beaver Rem-1m BOOKS Julie?: M. Casper REMARKS REMARKS 0F MARVIN E. WOLFGANG AT THE GUNS AND VIOLENCE SYMPOSIUM Marvin E. Wol?m? NORTHWESTERN UNIVERSITY SCHOOL OF LAW VOLUME 86 1 NUMBER 2/ WINTER 1996 THE jomuvu. or Cmnuu. LAW 8: Vol. 86. N0. 2 Copyright 9 1996 by Northwestern University. School of Law in USA. CRIMINAL LAW INSIDE THE RICHARD A. in the odd \ce tirely unheeded in the three docades following the influential Miranda opinioryAlthough law libraries are over?owing with doctrinal analy? I thank the following individuals for offering helpful comments, criticisms, and suggestions on earlier versions of this paper. Mark Cooney, David T. johnson. Richard Lempert. Gary Marx. Fred Pampel, Tom?anlon, Lindsey Simon,}erome Skolnick. Tom Tyler, Eric Wunsch, and Frank Zimring. Assistant Professor of Sociology and Adjoint Professor of law, University of Colo- rado. Boulder. 3A., University of California. Berkeley. 1985: MA, University of Chicago, 1989; University of California. Berkeley. 1994; University of California, Berke- ley. 1994. 1 AMERICAN LEGAL REAUSM (William W. Fisher et al. 1993)- 2 Law AND Socrc'nr: READINGS 0N ?nu: 50cm STUDY or Law (Stewart Macaulay et al. eds, 3d ed. 1995). This edited volume excerpts numerous studies that illustrate the gap between how law is portrayed on the books and how it actually works in practice. 3 Miranda v. Arizona, 384 US. 436. 448 (1966). 4 Id. at 448-58. 266 1996] INSIDE THE INTERROGATION ROOM 267 ses of appellate court cases, there exist no contemporary, descriptive or analytical studies of routine police interrogation practices in America. If, as William Hart has written, ?no law-enforcement func- tion has been more visited by controversy, confusion and court deci- sions than that of the interrogation of criminal then it is not only surprising but also disturbing to note just how little we know about everyday police interrogation practices in America. To be sure, since 1966 there have been a few experimental stud- ies of the social of confessions,6 several early evaluative studies of the judicial impact of the controversial Miranda decision on confession and conviction rates,7 and a few sociolinguistic pr ?conver- sational? analyses of individual police interrogation transcripts.8 But unlike their English counterparts,9 American scholars hav almost al- together ignored or avoided the empirical study of police interroga- tion practices and criminal confessions. In legal schol ip, there have been no empirical studies of police interrogation pra ?ces since the late 19603.lo Instead, law professors, lawyers, and 1a students have created a formidable law review literature that focuses? almost en? tirely on the doctrinal and ethical aspects of interrogation confes? sion case law, rather than on the routine activities of legal tors and? institutions.?1 Since traditional legal scholarship is based 0 an analy- sis of leading cases?which are unrepresentative of the larger universe of court cases and thus may depict atypical police practices as the norm?this literature is by itself both narrow and misltbading. In short, we know scant more about actual police interrogation practices today than we did in 1966 when Justice Earl Warren lamentled the gap 5 William Hart, The Subtle Art qumiort. Poua: MAG., Mar 1981, at 6 See LAWRENCE 8c SAUL Kmn, CONrr-?ssrous IN 1111-: Court-limos: (1993). 7 See 0113 STEPHENS, T111: Surat-1m: COURT AND or GUILT 165-200 (1973) (reviewing this literature). 3 SarWilliam Sanders. Pumps and Pm: Strategic Use of Conwnan?onul in Interm? gan'orrs in, T11: AS AN IN'rrtonucr-rou T0 THODS (Wil- liam Sanders ed.. 1976); DR. Watson, SorneFaatrms qf the Elidtatz'ort ofConf in Murder Inter-regimens, in INTERACTION (George Psathas ed., 1990); T. Wowk. Blame Allocation, Sex and Gender in a Martin Interrogation, in STU 11:13 INTERNA- FORUM (1984). 9 In the last decade in England. there have been numerous empirical $135 of police interrogation practices. For a review of this literature. see Grsu GUDJO N, T111: P57. CHOLOGY or Imuocrmous, AND TasnMONv (1992). '0 However, one law professor has recently gathered data on police interr rgation prac- tices in Salt Lake City. See Paul Cassell 8: Brett Hayman, Polia Interrogation in the 1990:: An Empirical Study of the Eject: qf Miranda, 43 UCLA REV. (forthcoming 1996) (manuscript on ?le with author). '1 The leading two contemporary scholars on police interrogation practices and crimi- nal confessions in the legal literature are Yale Kamisar and Joseph Grano. See YALE Kaursrut, Poucr: Imnoc/mosr AND CONFESSIONS (1980); josrarH GRANO, CONFESSIONS, Tram-1, AND 1111-: Law (1993). ET 268 RICHARD A. LEO 86 problem in Miranda at Arizona. This Article, which is the ?rst in a two-part series, will attempt to ?ll in some of the gaps in our knowledge of routine American police interrogation practices by describing and analyzing the characteristics, context, and outcome Of interrogation and confession in ordinary criminal cases that are not likely to make the published record on appeal. The second Article will analyze the impact of the Court?s ml, ing in Miranda Arizona on the behavior, attitudes, and culture of American police interrogators in the last thirty years.? Both articles are based on nine months (more than 500 hours) of ?eldwork inside the Criminal Investigation Division (CID) of a major, urban police department I shall identify by the pseudonym ?Lacoriia,"3 where I contemporaneously observed 122 interrogations involving forty-?ve different detectives. In addition, I viewed thirty videotaped custodial interrogations performed by a police department I shall identify by the pseudonym ?Southville?? and another thirty video taped interro- gations performed by a police department I shall identify by the pseu- donym ?Northvillef? For each interrogation, I recorded my observations qualitatively in the form of??eldnotes and quantitatively with a forty-seven question coding sheet. Thus, my ?eld research rep? resents a more general, multi?faceted and methodologically diverse study of the history and sociology of police interrogation in America.16 This Article takes the reader inside the interrogation room to un- derstand the characteristics, context, and outcome of contemporary police interrogation practices in America. It is the only study to do so in more than twenty-?ve years, and the ?rst ever to do so in any sus- tained, explicit, or comprehensive manner." I hope to reorient 12 Richard A. Leo, Tull-pact of Miranda Revisited]. Cm. Chmmorocv (forthcom- ing, Spring 1996). 13 In 1990 Walk had a population of 372.242?approximately 43% black, 28% white, 15% Hispanic, and 14% Asian/Paci?c Islander. In 1992 there were 58,668 Part1 offenses in Inconia (10.140 violent crimes and 48.546 property crimes). an of?cial crime rate of 123 per 1,000 members of the population. See Richard A. Leo. Police Interrogation in America: A Study of Violence. Gvility and Social Change 456 (1994) (unpublished dissertation, University of Califomia (Berkeley)). 14 In 1993 Southville had a population of 121.064. Fifty-one percent of Southville's residents were white. 24% Hispanic, 15% Asian. and 10% black. In 1993 there were 8.505 Part I offenses in Southville (1.298 violent crimes and 7,207 property crimes), and of?cial crime rate of 70.3 per 1,000 members of the populaton1993 Northville had a population of 116.148. Forty-six percent of Northville's residents were white, 21% Asian. 20% black. and 11% Hispanic. In 1993 there were 9.360 Part I crimes in Nonhville (1,613 violent crimes and 7.747 property crimes). an of?cial crime rate of 80.78 per 1,000 members of the populationWith the exception of my own empirically-grounded research on the history and 1996] INSIDE TILE INTERROGATION ROOM 269 much of the research and discourse on police interrogation practices in legal scholarship from its near exclusive doctrinal (or ?law-on-the? books?) focus to a more (or ?law?in-action?) perspective, which I believe is necessary to inform the legal. ethical, policy, and theoretical debates in the study of criminal procedure. In Part II of this Article I discuss the potential sources of bias in my data and howl attempted to overcome them. In Part I quan- titatively describe and analyze the patterns in police techniques, sus- pect behavior, and interrogation outcomes in all of the 182 cases I observed.?8 In Part IV, I analyze the e?'ects of police interrogation practices, Miranda warnings, and incriminating statements on the sub- sequent stages of the criminal process, such as the adjudication of guilt, case disposition, and sentencing. Finally, in Part of this Article I offer some concluding thoughts on the ?ndings of this study. II. METHODOIDGICAL CAVEAT: OBSERVER Erracrs AND THE PROBLEM or Burs19 Participant observation may be the ideal method to get as close as possible to the phenomena the researcher intends to analyze and un? derstand. This has been one of the underlying methodological as-? sumptions in my empirical study of American police interrogation practices.?0 However, the problem of studying naturally occurring data confronts the participant observer. Consequently, the partici- pant observer cannot control the parameters of his research nor the e?'ects of his behavior on the research subjects. It is a methodological truism that the field researcher inevitably influences the environment in which he participates during the very process of observation? These socalled ?observer effects? may ?con- sociology of American interrogation practices. see Richard A. Leo, Emu (hen-ion to Decep- tion: MAI-aim, in?rm; (1992) [hereinafter?'om Coercion toDxtp?on}; Richard A. Ito Poliotlnten'ogatian and Social Control, in Sou/u. 8: LEGAL (1994) [hereinafter Social Control}, only two observa- tional studies of police interrogation exist in the American literature. SeeMichael L. Wald ct al.. Interrogation iuNcwHavcn: 111(15an ofMimnda. 1519 (1967); and NEIL A Mruurn, THE AND LOCAL LAW TH: Inner or MIRANDA (1971). Both studies rely on data that was collected more than 25 ago, and neither one is an analysis of a major, urban police department. 13 For a qualitative analysis of the patterns in police techniques. suspect behavior and interrogation outcomes I observed, see Leo, supra note 13, at 170-256; Leo, Social Control, supra note 17, at 99413. 19 For a full account of the methodologies! strategies, challenges. and limitations of my participant observation ?eldwork. see Leo. supra note 13. at 451-95. See also Richard A. Leo, TrialandTribulations: Comm?hnography, Privilege, AM. Spring 1995. at 113-34. 20 Leo. supm note 13. 2? Melvin Pollner 81: Robert Emerson, The Dynamics and Distant in Fieldwork 270 RICHARD A. LEO [Vol. 86 taminate" the data that the participant observer seeks to collect. In the context of my research, my presence may have altered the behav- ior of the detectives during the custodial interrogations I observed. Indeed. whether the participant observer alters the behavior of law enforcement o?icers by his mere presence is a classic methodological problem that has bedeviled sociologists of policing since Westley?s groundbreaking ?eld research more than forty years ago.? It stands to reason that the presence of a third party in?uences police behavior, yet the precise outcome of this e?'ect is often dif?cult, if not impossi- ble, to accurately assess since participant observers usually lack any independent or hidden controls. I do not believe that my presence in the interrogation room sig- ni?cantly altered the behavior of the detectives I observed. Although I will never know the true e??ect of my presence, I offer the following observations. First, I sometimes put my ear to the door and listened to those interrogations from which I was purposely excluded, and each time the Mimnda warnings were given properly. Nor did I overhear any threats or promises. Conversely, I occasionally observed behavior inside the interrogation room?such as yelling, table pounding, or highly aggressive questioning?that straddled the margins of legality. After one such interrogation, one of the two interrogating detectives informed me that he could be ?red if I reported his behavior to the Captain. As we will see in Part of this Article, I viewed a few interro- gations that were clearly ?coercive? by the standards of contemporary appellate courts. In one of these interrogations, the primary detective ignored the suspect?s repeated invocations of his Miranda rights to silence and counsel, though ultimately the detective failed to convince the suspect to talk. After the interrogation session, the detective asked me what I thought he could have done differently to elicit admissions. When I responded that it did not matter since any subsequent confes- sion would have been suppressed by the court, the detective casually replied that neither one of us would have remembered the Miranda violations in court. That one of the detectives could so naturally as- sume that I would perjure myself to advance the cause of crime con- trol is, I think, good evidence that my presence, at least in some instances, had little effect on the interrogation practices I was observing.25 Marianas, in Frau) Rnsmacu (Robert Emerson ed., 1983). 22 William A. Westley. Wand WM 59 Soc. 34 (1953). 23 Other police ?eld researchers have reported similar feelings of invisibility. For ex- ample, Richard Uviller writes: ?For the most part, my presence was simply ignored while police activities were in progress.? UVILLEK. Tun-mu Em; A COLUMBIA Law Pxorrssort's YEAR on rm: Smarts rm: New Your Cm Poua: (1988). And David 1996] INSIDE THE MERROGATION ROOM 271 Second, the more time I spent inside the CID, the more the detectives became accustomed to my presence. As I became part of the ?furniture? inside the Laconia Police Department (LPD), the detectives frequently treated me as one of their own. They would, for example, describe their cases to me by penal code sections or their actions by police codes or jargon, apparently forgetting that I did not know what they meant. In addition, many of the detectives shared with me explicitly con?dential information about their co?workers or superiors, information whose exposure could have damaged their rep- utations, if not their careers, within CID. At the same time, detectives told me of their own indiscretions and sometimes questionable behav- iors, for which they could have been administratively sanctioned, and in some instances held civilly liable, had I publicly revealed their con- ?dences. In a sense, then, I became the archetypical Sirnmelian ob- server to whom social secrets were entrusted.? Strangely and unexpectedly, I realized midway through my research that the police code of silence (perhaps more accurately described as a code of soli- darity) applied to me as well. Third, I interviewed prosecutors and public defenders who were knowledgeable of LPD custodial interrogation practices. They agreed with my general descriptions of police interrogation practices at LPD, con?rming that the methods and techniques I observed were repre- sentative of what they knew of interrogation methods at LPD from their daily cases as well. Signi?cantly. the public defenders?the most strident ideological critics of police in an adversarial system of crimi- nal justice?generally spoke respectfully of custodial interrogation practices at LPD, agreeing that detectives rarely engage in any illegali- ties during custodial questioning. Nevertheless, despite these com- ments, and although 1 established high levels of trust with the LPD detectives and sometimes even felt literally invisible during the inter- rogations, inevitably my presence must have exerted some e?'ect on their behavior. Thatl tended to be (but was not always) excluded from the more serious cases raises the general methodological issue of bias, and more speci?cally the problem of the representativeness of my data within LPD and the generalizability of my ?ndings beyond LPD. In ?eld studies, the researcher can overcome these problems only in degrees. I was unable to select randomly those cases whose interrogations I ob- served for three reasons. First, I had no control over those interroga- Simon writes: became a piece of furniture in the unit, a benign part of the detectives' daily scenery." DAVID SIMON, Humane: ON rm: Krumc 5mm 596 (1991). 2* KURT Wou-?r, THE SOCIOLOGY or Grout: SIMMFJ. (1950). 272 RICHARD A. LEO 86 tions from which some detectives chose to exclude me. Second, even in those cases in which I was allowed to attend the interrogation, the suspect had sometimes posted bail and left the jail prior to any custo~ dial questioning. A class bias that is naturally present in the practice of interrogation at LPD was therefore inevitably present in my data. Third, occasionally circumstances conspired to provide me with a choice over which interrogations I would attend. Although sometimes hours would pass during which no interrogations would occur, at other times multiple interrogations would be occurring simultane- ously. To correct for the inherent biases in my data, I always re- quested to sit in on the interrogation of the most serious cases available to me.? Nevertheless. the 122 interrogations I observed con- temporaneously were still not entirely representative of the general category of interrogations occurring within LPD. In addition. I attempted to correct for the bias against serious cases in the data I collected in my non?participant observations of sixty videotaped interrogations at the Southville and Northville Police De? partments. I speci?cally requested videotapes of interrogations in the more serious felony crimes, especially homicide, rape, and assault, from these two departments. Observing the videotaped interroga- tions of two other police departments served as a check on'any idio- interrogation practices at Certainly the interrogation practices at LPD, which has been generally regarded as one of the most professional police departments in America during the last three decades. are not represenmtive of the interrogation practices of all police departments in America. Nevertheless, observing the interro- gation practices of two other police departments was one way I at: tempted to redress the intrinsic limitations of the case study method. 2? I considered the seriousness of cases in the following order: 1) homicide. 2) sexual amult. 3) felony assault; 4) robbery, 5) property crimes. 2" Although both the Southville and Northville Police Departments routinely videotape felony interrogations, neither department has a policy of storing their videotapes. and neither department permitted me to view videotaped interrogations in cases still pending. Therefore. neither department had many videotaped interrogations on hand from which I could choose to view cases. In Northville. Iwas provided with a master list of all the felony cases during the last several years in which interrogations had been videotaped. I was then penniued to meet with the evidence technician and retrieve as many tapes as we could locate. Unfortunately. most of the tapes had been destroyed once the case had been con- cluded. In Southville. the Captain ofG?lD circulated a memo to the detectives. asking them to identify the homicide. sexual assault. and felony assault cases in which they had interro- gated suspects. From this list. the Captain and I retrieved existing videotapes. While my ?eldwork in Northville and Southville made up for some of the biases of my participant observations in Iaconia, the method through which I obtained videotaped interrogations inevitably contained its own set of biases. Ultimately. I had little control over which videos both departments provided me. 1996] INSIDE THE MERROGAUON ROOM 273 THE CONTEXT AND OUTCOME or Poucr. INTERROGATION: Exrcomnc THE DATA In all the interrogations I observed, both contemporaneously at LPD as well by videotape at the Southville and Northville Police De- partments, I coded for a number of independent variables class, race, gender, and social distance between the suspect and victim, strength of evidence against the suspect, and prior conviction record of the suspect) and a number of dependent variables whether suspect waived/ invoked Miranda, length of interrogation, outcome of interrogation, and ultimate casedisposition). I now turn to a system- atic analysis of these and other variables. A brief statistical description of the demographic, legal, and case variables in my sample reveals the variation in my data and the typical characteristics of the suspect, victim, and interrogation procedures in the cases I observed. more than one-third of the inter- rogations were conduc?ed in an interrogation room located inside the jail; the remaining were conducted in the interrogation rooms located inside the CID. In the large majority of interrogations one detective questioned the suspect; in the remaining interrogations two detectives conducted questioning. The primary detective was typically white though not infrequently African-American or Hispanic when present, the secondary detective was also white most of the time and African-American some of the time There was less variation in the gender of the detectives: virtually all (over 90% of the primary and 86% of the secondary detec- tives) were men. Most of the primary detectives had been police of~ ?cers for between ten and twenty years and detectives for one to ?ve years most secondary detectives had also been police of?cers for ten to twenty years but detectives for only zero to three years The typical suspect in my sample was a young, lower or working class, African-American male. Although the age range spanned from the middle teens to the late sixties, approximately two-thirds of the suspects were less than thirty years old. More than 87% of the suspects were from the lower or working class; almost 12% were mid- dle class; and only 1% were upper middle class.? Sixty-nine percent of the suspects were 14% were white, 13% were Hispanic, and the remaining 4% were either Asian or Native Ameri- 27 I coded for the class status of the suspects in my sample by asking the primary detec- tive to rate the suspect as either below middle class. middle class, or above middle class based on 1) the suspect?s occupation and (2) the location of the victim?s residence. The same procedure was followed to code for the class status of the victims in my sample. 274 RICHARD A. LEO 86 Can.28 Thus, more than 85% of the suspects in my sample were minor- ities. As with the detectives who interrogated them, virtually all the suspects were male (more than Unlike the suspects, the victims in my sample did not ?t so clear a demographic pro?le, because many of the victims (more than 25%) were organizations. Excluding organizations, most of the victims came from the lower or working class a higher number of vic- tims came from the middle class than did suspects but, as with suspects, only a negligible number were from the upper middle class (1 . just as African-Americans comprised the largest racial group of suspects in my sample, so too were they more likely to be the victims of crime than any other racial group. Excluding organizations, a full 42% of the victims in my sample were African-American; 28% were white; 22% were Hispanic; and the remaining 8% were Asian. A far higher percentage of victims were likely to be female than were the . suspects: excluding organizations, 39% of the victims were female while only 61% were male. All the interrogations I observed were for felony offenses. A fre? quency distribution of these o?enses, divided into ?ve categories, is listed below in Table l.29 Table FREQUENCY DISTRIBUTION or TYPE or CRIME Type of Crime Freq. Percent Theft 9 4.95% Burglary 21 11.54 . Robbery 78 42.86 Assault 44 24.18 Homicide 22 12.09 1 Other 8 4.40 Total 182 100.00 . I r23 The non-minority white) suspects in my sample were drawn disproportionately llrom the cases I observed at the Southville and Northville Police Departments. Although more than two-thirds of the interrogations I observed occurred at the Laconia Po? lice Department, only one-third 27) of the interrogations involving white suspects in my tmple occurred at this department; 26% 27 occurred at the Southville Police Depart- . em; and 41% (11/27) occurred at the Northville Police Department. 29 My division of criminal offenses into ?ve primary categories parallels the categoriza- iion of o?'enses inside the Laconia CID Unit. The reader should note that allegations of iape or child molestation were classi?ed as assaults. The few crimes in the ?other? category were mostly drug offenses. Although these categories appear relatively straightforward. if multiple offenses were involved, the suspect was classi?ed into the category representing the most serious offense. For example, if a suspect allegedly committed bath a theft and a homicide, his crime was classi?ed as a homicide. I I 1996] INSIDE THE EVTERROGATION ROOM 275 While the type of crime I was likely to observe varied by depart- ment, 81% of the offenses were crimes against persons robbery, assault, homicide), and the remaining 19% were crimes against prop- erty burglary, theft). Two potentially important case factors that may in?uence a sus- pect?s treatment by legal authorities are the strength of the evidence and the suspect?s prior criminal record. In 33% of the cases in my sample, the strength of the evidence against a suspect prior to an in- terrogation was weak (highly unlikely to lead to charging); in 32% of the cases the strength of the evidence was moderate (probably likely to lead to charging); and in 35% of the cases it was strong (highly likely to lead to charging). In 13% of the cases in my sample, the suspect did not have a prior criminal record; in 29% of the cases, the "suspect had a misdemeanor record; and in 58% of the cases?, the sus pect had a felony record. Not surprisingly, almost 90% of the suspects interrogated were repeat players with prior criminal records. This va- riable distinguished the suspects from their victims. Almost 69% of the victims had no prior criminal record; approximately 15% had mis- demeanor records; and approximately 16% had felony records. The formal interrogation process must, of course, be preceded by\ the well-known Miranda warnings. Table 2 lists the frequency distribu? tion for suspect?s responses to Miranda. Table 2 FREQUENCY DISTRIBUTION or RESPONSE TO MMWA WARNINGS Suspect?s Response To Miranda Warnings Freq. Percent Waived 136 74.73% Changed to Waive 1 0.55 Invoked 36 19.78 Changed to Invoke 2 1.10 Not Applicable 7 3.85 Total 182 100.00 In seven (almost of the cases I observed, the detective did not provide any Miranda warnings because the suspect technically was not ?in custody? for the purpose of questioning.30 Therefore, in these seven cases the detectives were not legally required to issue Miranda 30 In other words. the suspect was neither under arrest nor was his freedom restrained ?in any signi?cant way" (in each case, the detective?) informed the suspect that he did not have to answer their questions and that he was free to leave at any time). 276 RICHARD A. LEO 86 warnings.51 With the exception of these cases, the detective(s) read each of the fourfold Miranda warnings verbatim from a standard form prior to virtually every interrogation I observed.? A suspect might respond in one of four ways: waiving his rights, invoking them, or changing his initial response either to a waiver or an invocation. As Table 3 below indicates, 78% of my sample ultimately waived their Miranda rights. while 22% invoked one or more of their Miranda rights, thus indicating their refusal to cooperate with police questioning. Table 3 . FREQUENCY DISTRIBUTION or ULTIMATE RESPONSE 'no MWA Whether Suspect Waived or Invoked - Freq. Percent Waived 137 78.29% Invoked 38 21.71 Total 175 100.00 If a suspect chooses to waive his Miranda rights, the custodial in- terrogation formally begins. If a suspect chooses to invoke one or more of his Miranda rights, typically the detective terminates the inter- rogation and retums the suspect to jail (if he was under arrest). How- ever. in seven of the cases I observed, the detectives questioned suspects even after receiving an invocation. In each of these cases, the detective(s) informed the suspect that any information the suspect provided to the detective could not and therefore would not be used - against him in a court of law. The detective told the suspect that the sole purpose of questioning was to learn ?what really happened.? Of course, what the detectives knew and did not tell the suspect was that although the prosecution could not use such evidence as part of its case-in-chief, any information the suspect provided to the detective nevertheless could be used in a court of law to impeach the suspect's credibility, and indirectly incriminate the suspect if he chose to testify at trial.33 In the remaining thirtyone cases in which the suspect in- voked his Miranda rights at some point during questioning (82% of all 3? Miranda warnings are legally required only ?after a person has been taken into cus tody or otherwise deprived of his freedom of action in any signi?cant way.? Miranda v. Arizona, 384 US. 436. 444 (1966). 32 In two of the burglary interrogations I observed. one investigator recited the Miranda warnings verbatim from memory. One robbery interrogator had the habit of reading the M?undawamings from a standard form but crossing out (and thus not reading) the words ?and will? in the second of the four warnings. See Leo. supra note 13, at 174-81. 33 Harris v. New York. 401 US. 222 (1971). 1996] INSIDE TIE MERROGATION ROOM 277 cases in which a suspect invoked a Miranda tight), the detective terminated the interrogation. In any session in which the detective questioned a suspect beyond the Miranda warnings (whether or not a suspect invoked), I coded for twenty-?ve potential interrogation techniques.? Table 4 lists the fre- quency distribution for the total number of tactics employed by detec- tives during each interrogation. The number of tactics a detective employed per interrogation ranged from zero the suspect spon- taneously confessed or the detective did not genuinely try to elicit a confession) to ?fteen. The cumulative percentage ?gure represents the percentage of interrogations in which detectives used at least that many interrogation tactics. Table 4 FREQUENCY DISTRIBUTION or TACTICS EMPLOYED Pea Number of Interrogation Tactics Used by Detectives Cumulative per Interrogation Freq. Percent Percentage 0 2 1.31% - 1 8 5.23 99% 2 19 12.42 93 3 17 11.11 81 4 16 10.46 70 5 16 7 10.46 59 5 16 10.46 49 7 19 12.42 39 8 10 6.54 26 9 11 7.19 20 10 9 5.88 12 11 5 3.27 7 12 3 1.96 3 15 2 1.31 Total 153 100.00 The detectives employed a median of 5 and a mean of 5.62 tactics per interrogation. Clearly, however, the detectives used some interro- gation tactics more frequently than others. Table 5 below lists each of the twenty-?ve tactics, and the frequency of. their use during the inter- rogations I observed. 3? 1 derived this list from the tactics (1) advocated in contemporary police interroga- tion training annuals: (2) taught in police interrogation training courses; and (3) used by police detectives in popular culture. 278 RICHARD A. LEO 86 Table 5 TYPES or INTERROGATION TAcncs AND THEIR FREQUENCY No. of Cases of Cases In Which In Which Type of Interrogation Tactic Tactic Used Tactic Used TACTICS USED MOST OFTEN Appeal to the suspect's self-interest 134 88% Confront suspect with existing evidence of 130 85 guilt TACTICS USED OFTEN Undermine suspect?s con?dence in denial of 66 43 guilt Identify contradictions in suspect's story 65 42 Any Behavioral Analysis Interview questions 61 40 Appeal to the importance of cooperation 56 37 Offer moral justi?cations/ 52 34 excuses 7 Confront suspect with false evidence of guilt 46 30 Use praise or ?attery 46 30 Appeal to detective?s expertise/ authority 45 29 Appeal to the suspect?s conscience 35 23 Minimize the moral seriousness of the offense 33 22 TACTICS USED LEAST OFTEN Touch suspect in a friendly manner 17 11 Invoke metaphors of guilt 15- 10 Minimize the facts/ nature of the offense 9 6 Refer to physical of guilt 7 5 Exaggerate the facts/ nature of the o??ense 6 4 Yell at suspect 5 3 Exaggerate the nature/ purpose of questioning 3 2 Exaggerate the moral seriousness of the 3 2 offense Accuse suspect of other crimes 2 1 Attempt to confuse the suspect I Minimize the nature/ purpose of questioning 1 1 Good cop/Bad cop routine 1 1 Touch suspect in an unfriendly manner 0 0 As Table 5 indicates, there is great variation in the distribution of the interrogation tactics I observed. A couple of the tactics were used in virtually all of the cases, several others were used in approximately one?third to one-half of the cases, a couple were used in approxi- mately one-?fth of the cases, a few others were used only sparingly, and others virtually not at all. Ifa portrait of the typical interrogation emerges from the data, it involves a two-prong approach: the use of 1996] INSIDE TIE MERROGATION ROOM 279 negative incentives (tactics that suggest the suspect should confess be- cause of no other plausible course of action) and positive incentives (tactics that suggest the suspect will in some way feel better or bene?t if he confesses). In my sample, detectives typically began the interro- gation session by confronting the suspect with some form of evidence, whether true or false suggesting his guilt and then at- tempting to undermine the suspect?s denial of involvement while identifying contradictions in the suspect?s alibi or story But detectives relied on positive incentives as well, most often by ap- pealing to the suspect?s self-interest but also by frequently of- fering the suspect moral justi?cations or excuses using praise or ?attery minimizing the moral seriousness of the offense appealing to the importance of cooperation with legal authorities or appealing to the detective?s expertise or appealing to the suspect?s conscience In approxi? mately 90% of the interrogations I observed, the detective confronted the suspect with evidence (whether true or false) of his guilt and then suggested that the suspect?s self-interest would be advanced if he confessed}5 Of course, the interrogations in my sample also varied by length, ranging from literally seconds (when the suspect invoked before the detective even introduced himself) to four and one-half hours. Table 6 shows the frequency distribution of the length of the interrogations for those cases in which the detective chose to question a suspect excluding the twenty-nine cases in my data in which the suspect in- voked his Miranda rights, and the detective terminated all question- ing). As Table 6 indicates, more than 70% of the interrogations in my sample lasted less than an hour, and only 8% lasted more than two hours. Table 6 LENGTH or INTERROGATION ONLY WHERE AN INTERROGATION OCCURRED Length of Interrogation Freq. Percent Cum. Less Than 30 Minutes 53 34.64% 34.64% 31?60 Minutes 56 36.60 71.24 1-2 Hours 32 20.92 92.16 More Than 2 Hours 12 7.84 100.00 . Total . 153 100.00 35 For a fuller description of contemporary American police interrogation techniques. see Leo. Social Contml, supra, note 17, at 99-113. 280 RICHARD A. LEO 86 The outcome of an interrogation is, of course, the most impor- tant aspect of questioning from the perspective of the police, and po- tentially the most important aspect of a case from the perspective of the suspect. In each interrogation, I coded for one of four possible outcomes: the suspect provided no information to the police that they considered incriminating (whether or not the suspect invoked); the suspect provided some information that police considered incrim- inating (whether or not intentionally) but did not directly admit to any of the elements of the crime;96 the suspect admitted to some, but not all, of the elements of the crime; and the suspect provided a full confession to the detectives. Table 7 displays the frequency distribu- tion for the outcome of the interrogations in my sample. Table 7 OUTCOME or Isrmnocmrons Suspect's Response to Interrogation Freq. Percent No Incriminating Statement 65 35.71% Incriminating Statement 41 22.53 Partial Admission 32 17.58 Full Confession 44 24.18 Total 182 100.00 If an interrogation is successful when the suspect provides the detec- tive(s) with at least some incriminating information, then almost two- thirds of the interrogations I observed produced a successful result. The rate of successful interrogations in this sample is notably higher than the success rate reported by Wald et al. Younger Neubauer Seeburger and Wettick or Leiken but is lower than the success rate reported by Witt42 If we exclude from my sample those cases in which 35 Typically this consisted of implausible or contradictory denials that the detectives believed corroborated other evidence pointing to the suspect?s guilt or that locked the suspect into a false alibi. and/or that could be used successfully to impeach a suspect's credibility. and thus incriminate him. in subsequent judicial proceedings. 3'7 Wald et al.. supra note 17. at 1566. 38 Evelle Younger. Midst of 6 Sunny WW in theDisn-ia Attorney?s O??ia of Las Angeles Upon ?unmiss- ?JFelony Cases, 5AM. Cam. L. Q. 32. 35 (1966). 39 David W. Neubauer. (?infusion in City: San: Cam and was, 65]. Cum. L. 8: Cmumorocv 103. 104-06 (1974). 40 Richard Seeburger 8c Stanton Wettick, Miranda in WA Statistical Study, 29 U. L. Rev. 1, 11 (1967). ?1 Lawrence S. Leiken. Pals}: Interrogation in Gatorade: TRIM ofMimnda. 47 DENV. I, 13 (1970). 41? james W. Witt, Nan-Comiw Interrogation and the MW of Criminal justice: The 1996] INSEE THE MERROGATION ROOM 281 the police terminated questioning upon the invocation of a Miranda right (and thus the detective or detectives made no effort to incrimi- nate the suspect), more than tlrreevfour'ths of the interroga- tions I observed produced a successful result. To the extent that these studies are representative of general trends in American policing, detectives have become increasingly successful at eliciting incrirninat. ing information from criminal suspects. Following a suspect?s waiver of his Miranda rights, any informa- tion that he provides to a detective during custodial questioning? whether an incriminating denial, a partial admission. or a full confes- sion?must be rendered ?voluntarily? if it is to be used against that suspect in subsequent judicial proceedings.? In other words. the prosecutor will not be able to use any incriminating information po- lice elicit from a custodial suspect if the interrogation methods they employed are deemed ?coercive? by the courts. The issue of coercive questioning has been the fundamental concern of the appellate courts that have traditionally regulated police interrogation proce- dures in America.? Yet the meaning that courts have attributed to the concept of ?coercion? has always been relative and historically contingent. How courts de?ne the concept of ?coercion? and where they draw the line between coercive and non?coercive interrogation tactics has varied dramatically in American history and continues to vary (from one jurisdiction to another as well as between courts within similar jurisdictions) in contemporary America.? For ?coercion? in the context of interrogation is not an external thing one can indepen- Impact q'Mimnda on mam. 64]. Crust. L. 8: Cammorocv 320. 325 (1973). ?3 One must interpret these comparisons with some caution. for several of the resesrch? en de?ned a successful interrogation di?'erently or coded the outcomes of interrogations di?'erently or relied on di?'erent methodologies when gathering their data. Witt uses a de?nition of success identical to my own any interrogation yielding a confession, admission. or incriminating statement). SssWitt. supra note 42, at 325. I computed the Wald et a1. success rate by imposing on their data a de?nition of success identical to mine. SuWald et al.. supra note 17. at 1566. Younger's de?nition of a successful interrogation?a confession. admission. or other (presumably incriminating) statement?appears similar to mine own. SssYounger. supra note 38. at 35; sass also Evellej. Younger. Interrogation ?Jaimi- naqu'sndants?Sasrs View an Miranda Arizona. 35 FortortAM L. REV. 255. 255-62 (1966). Neubauer?s de?nition of a successful interrogation. which includes any confession. admis? sion, or statement to police. appears more inclusive than mine because ?statement to po- lice? need not be incriminating. See Neubauer. supra note 39. at 104-106. By contrast, Seeburger 8c Wettick and Leiken employ a more exclusive de?nition of successful interro- gations. which includes only confessions and admissions. not incriminating statements. See Seeburger 8c Wettick. supra note 40. at 10; Leilten. supra note 41. at 13. Despite the lack of identical de?nitions. however, these studies offer valuable thta for rough comparisons of the ef?cacy of modern interrogation practices. ?4 Brown v. Mississippi. 297 US. 278 (1936). ?5 WAYNE LAFAVE &jr.aou) Emu CRIMINAL PROCEDURE (2d ed. 1992). ?5 SssLeo. mpmnote 13. at 12-66. 282 RICHARD A. LEO 85 dently observe or something whose existence one can objectively ver- ify, but rather a concept that courts attach to a variable and sometimes quite amorphous set of police behaviors." In hard cases how courts draw the line between coercive and non-coercive interrogation meth- ods depends on the. judge?s predisposition toward crime control or due process values and perhaps ultimately on the judge's philosophi- cal conceptions of moral responsibility and the limits of human freedom.?8 To Operationalize the concept of ?coercion,? I attempted to cap- ture those set of police behaviors and interrogation practices that con- temporary appellate courts generally tend to label as ?coercive.? Thus, I coded any interrogation in my sample as ?coercive? if at least one of the following ten conditions were present during the interrogation: (1) The detective failed to read the Miranda warnings; (2) The suspect was not permitted to invoke his Miranda rights; (3) The detective touched the suspect in an unfriendly manner, (4) The suspect was in obvious physical or pain (whether or not related to the detective?s actions); (5) The detective threatened the suspect with physical or harm; (6) The detective promised the suspect leniency in exchange for an admission of guilt; (7) The detective deprived the suspect of an essential necessity (such as water, food, or access to a bathroom); (8) The detective?s questioning manner was unrelenting, badgering or hostile; The intenogation lasted an unreasonable amount of time (more 1 than six hours); or (10) The suspect's will appeared to be overbome by some other factor or combination of factors. Although some may disagree with where or how I chose to draw the lin between coercive and non-coercive interrogations, I believe that I 1 on the side of ruling as ?coercive? questioning methods that many contemporary trial and appellate courts would otherwise deem to non-coercive49 and thus, my criteria for coercive tactics generally 1? lve any doubts in favor of the suspect, not the police. Neverthe- in my sample of 182 custodial interrogations, police questioning Heathods in only four (or of the cases rose to the level of ?coer- Id. See Hunt-1n Panama, THE or THE CRIMINAL (1968). To my knowl- no one has empirically studied how judges reason. think about and distinguish ?coer- ci from ?voluntary' confusions. See Ysu: Kamsut ET AL. Moor-2m Gamma. PROCEDURE CASES, COMMENTS, Qua; no Us 452-61 (8th ed. 1994). 1996] INSDE TIE ROOM 283 cion? according to these criteria. Since four is too small a number to warrant any statistical analysis, I can only qualitatively describe the patterns, if any, I observed in these cases. All four cases involved the use of coercive methods; none involved the use of physically coercive methods. In one interrogation, detectives questioned a heroin addict who was quite obviously experiencing extremely painful withdrawal While the detectives did nothing to contribute to the suspect?s agony, they intentionally questioned him during the second day of his incar- ceration when they knew his withdrawal would be most acute. Although the police arrested the suspect on probable cause for felony gun possession, the detectives considered him a potential, not an actual, suspect in their robbery case. The detectives questioned him as if he were an informant, promising to release him just as soon as he provided them with information about a couple of robberies. Shortly thereafter, the police released the suspect from custody with- out charging him. In another case I coded as ?coerciVe,? two detec- tives employed the ?good cop-bad cop? routine on a young gang member who witnessed a violent gang beating. As one detective kindly promised to release him from custody if he named the perpe-5 trators of the assault, the other detective angrily threatened to provide the prosecutor with incriminating information that would send the suspect to prison. The suspect provided the detectives with the infor- mation they desired and was without charge. In another case, an alleged violent armed robbery by an individ- ual with a long criminal record who had been recently released from prison, the detectives failed to acknowledge the suspect?s repeated in? vocation of silence in response to the initial Miranda admonition. Af- ter repeatedly trying to talk the suspect out of waiving his Miranda rights, the detectives terminated their questioning after approximately ?ve minutes. The suspect, against whom strong eyewitness evidence existed, was eventually charged by the prosecutor.? The last instance of a ?coercive? interrogation in my sample involved a suspect who po- lice arrested for selling drugs and offered an explicit promise of leni? ency if he became an informant and provided names of more highly placed drug dealers. The suspect refused to turn state?s witness, and received a four year prison sentence instead. Of course, four cases is too small a sample from which to draw any meaningful generalizations. If a common thread exists among these cases, however, it is that the detectives perceived that they had 50 Due to a regretful error in note-taking, I was unable to track down the ultimate disposition of this case. 284 RICHARD A. LEO 86 nothing to lose by exerting ?coercive? means or intentionally eliciting an involuntary statement. For as all detectives know, witnesses do not enjoy the same constitutional protections as suspects, and the judicial suppression of a suspect's statement may not be very signi?cant to the state?s case when there exists other compelling physical evidence of a suspect's guilt. As in earlier studies,? the detectives did not, in this limited sample, appear to be any more likely to resort to coercive methods as a result of any personal or social characteristics of the sus- pect under questioning. Police interrogation involves only one stage of the larger criminal process through which an individual may be convicted and ultimately incarcerated. The detective?s primary goal during interrogation is to gather enough incriminating evidence to convince the prosecutor to ?le criminal charges against the suspect. Eventually, prosecutors charged 69% of the suspects in my sample and released the remaining Of those individuals who were charged by prosecutors, 88.5% were convicted. Approximately 60% of the suspects in my entire sam- ple were eventually convicted, thoughathis ?gure includes only those cases whose dispositions were resolved not pending trial or un- known) when I left the ?eld altogether.? Eighty-?ve percent of those su5pects received felonies while the remaining 15% received misdemeanors. . Whether or not a suspect was ultimately convicted of the offense for which he was under questioning, his case might be resolved by the criminal justice system in one of several ways: by dismissal; by a viola- tion of parole or probation (in which the suspect is returned to prison for up to one year due to a prior conviction); by plea bargaining; or by- trial. No charge or dismissal after charging occurred in approxi- mately 38% of the cases in my sample; in approximately 10% of the cases the suspect violated his probation or parole from a prior convic- tion and was returned to prison; plea bargaining resolved approxi- mately 43% of the cases; and in approximmly 10% of the cases, the suspect?s case was resolved by a trial. 5' SaWald et al., supra note 17. 1 52 I coded parole/ probation violations as being charged, even though technically a sus- pect whose parole or probation is violated is not charged. Prosecutors may prefer to cite a aparole/probation violation rather than charge a suspect for his current o?'ense because the standard of proof for a violation is only ?preponderance of evidence? more likely than not) rather than ?beyond a reasonable doubt,? and thus violating a suspect's parole or probation typically results in the suspect's automatic reinmrceration for a previous convic- vtion. Prosecutors ?violated? the parole/probation of 18 (or approximately 10%) of the suspects in my sample. 53 Although I left the ?eld in late September 1993 after nine months of ?eldwork. I briefly returned to the Laconia. Southville, and Northville Police Departments in the spring of 1994 to collecr information on outcomes for the outstanding cases in my sample. 1996] INSIDE TI-E ROOM 285 The ?nal stage of the criminal justice process for defendants is sentencing. In my sample, approximately 42% of the suspects were either not charged or not convicted and thus received no sentence. The other suspects in my sample received sentences ranging from pro? bation to life imprisonment. Table 8 lists the length of- sentence re ceived by those suspects for whom dispositions were available at the time of this writing.? While there was considerable variation in the range of sentences in my sample, I have receded them into four categories. Table 8 or SENTENCE Susrecr RECEIVED Length of Suspect's Sentence Freq. Percent Cum. None 68 41.98% 41.98% Low (Less Than One Year) 56 34.57 76.54 Medium (1-5 Years) 17 10.49 87.04 High (More than 5 Years) 21 12.96 100.00 Total 162 100.00 IV. THE CONTEXT AND OUTCOME or: POLICE ANALYZING THE DATA What explains how suspects respond to Miranda warnings? How do their responses vary by social, legal, and case factors? Are the dif- ferences statistically signi?cant? How does a suspect?s decision either to waive or invoke his Miranda rights a?'ect the processing of his case, the likelihood of conviction. and the ?nal case resolution? What ex- plains how suspects respond to police questioning? Why do some sus- pects confess while others do not? Does the likelihood that a suspect will provide detectives with incriminating information vary by social, legal, and case factors? Are these di??erences statistically signi?cant? Which interrogation tactics are likely to be most effective and ineffec- tive in eliciting incriminating information from suspects? What ex- plains the varying length of custodial interrogations and efforts detectives expend in trying to elicit incriminating information from criminal suspects? What effect does providing incriminating informa- tion to a detective have on the suspect?s fate in the criminal justice system? These are among the questions that I will seek to answer in this section. 5? Sumnete 53. 286 RICHARD A. LEO 86 A. A SUSPECT RESPONSE TO MRANDA AND ITS EFFECTS Despite the passing of almost thirty years since their judicial crea? tion, the Miranda warnings remain one of the most controversial is- sues in American criminal justice, even as Miranda has become settled doctrine in the appellate courts, standard policy in police depart- ments, and a household word in American popular culture.55 The conventional wisdom in legal and political scholarship is that virtually all suspects waive their rights prior to interrogation and speak to the police.m However, as we saw above, almost one-fourth of my sample exercised their right to terminate police questioning, while 78% of the suspects chose to waive their Miranda rights. Nevertheless, one might expect that certain individuals are more likely to waive their rights than others. Indeed, the Warren Court in Miranda specu- lated that underprivileged suspects were less likely to comprehend or exercise their constitutional rights to silence and counsel than their more advantaged counterparts.? Though I tested for twelve social, legal and casespeci?c variables, the only variable that exercised a sta- tistically signi?cant effect on the suspect?s likelihood to waive or in? voke his Miranda rights was whether ?a suspect had a prior criminal record As Table 9 below indicates, while 89% of the sus- pects with a misdemeanor record and 92% of the suspects without any record waived their Miranda rights, only 70% of the suspects with a felony record waived their Miranda rights. Put another way, a suspect with a felony record in my sample was almost four times as likely to invoke his Miranda rights as a suspect with no prior record and almost three times as likely to invoke as a suspect with a misdemeanor record. This result con?rms the ?ndings of earlier studies,58 as well as the conventional wisdom among the detectives I studied, who complained that ex?felons frequently refuse to talk to them as a matter of course. The more experience a suspect has with the criminal justice system, the more likely he is to take advantage of his Miranda rights to terrni? nate questioning and seek counsel. 55 See ?rst) P. Gum?, THE WOUND (1970); LIVA BAKER, MIRANDA: Came, Law AND Pouncs (1983); Leo, supra note 12. 55 Leo, mom note 12. 5'7 Miranda v. Arizona. 384 Us. 436. 471-73 (1966). 58 Wald et al., supra note 17. at 1562-77; Neubauer. supra note 39, at 106-07. 1996] INSEE THE INTERROGATION ROOM 287 Table 9 RESPONSE TO BY PRIOR CRIMINAL RECORD Whether Suspect Waived or lnvoked Suspect?s Prior Record Waived Invoked Total None 22 2 24 91 .67% 8.33% 100.00% Misdemeanor 42 A 5 47 89.36% 10.64% 100.00% Felony 72 31 103 69.90% 30.10% 100.00% Total 136 38 174 78.16% 21.84% 100.00% Punch chi2(2) - 10.1340 Pr - 0.006 At least as important as a suspect's response to the Miranda wam- ings is the effect that either a waiver or an invocation will exert on the processing of his case, the likelihood of conviction, and the ?nal case resolution. While the police may consider a suspect?s interrogation less likely to be successful if a suspect invokes his Miranda rights, this is neither necessarily nor obviously true. In my sample, the detectives - acquired incriminating information against a suspect in six (approxi- mately 16%) of the thirty-eight interrogations in which the suspect at some point invoked his Miranda rights.?3 Despite its potential effect on the outcome of an interrogation, a suspect?s case was 4% less likely to be charged if be waived his Miranda rights than if he invoked his Mimnda rights prior to or during interrogation (approximately 73% vs. While counterintuitive, this difference, as Table 10 below indicates, is not statistically signi?cant and thus not signi?cantly re- lated to the prosecutor?s decision to charge the suspect with a crimi? nal offense. 59 In my sample, detectives questioned seven suspects after they had invoked their Mi- randa rights and two suspects who subsequently invoked their Miranda rights. Of these nine cases, six suspects provided incriminating information to detectives. 288 RICHARD A. LEO 86 Table 10 Emcr OF RESPONSE TO Manama ON DECISION T0 CHARGE CASE Suspect's Response . Whether Suspect was Charged by Prosecutor To Miranda Warnings Not charged Charged Total Waived 42 95 137 30.66% 69.34% 100.00% Invoked 10 27 37 27.03% 72.97% 100.00% Total 52 122 174 29.89% 70.11% 100.00% Pearson chi2(l) 0.1832 Pr 0.669 While the suspects in my sample who waived their Miranda rights were only 4% less likely to be charged by the prosecution, they were approximately 10% more likely to be convicted of an olfense than those who invoked their Miranda rights (63% vs This di?'er- ence may seem large, but it is not statistically signi?cant, as Table 11 below indicates. Table LIKELIHOOD 0F CONVICTION BY RESPONSE TO MEAIVDA Snspect?s Response Whether Suspect Was Convicted To Miranda Warnings Not Convicted Convicted Total Waived 48 81 129 37.21% 62.79% 100.00% Invoked 15 17 32 46.88% 53.13% 100.00% Total 63 98 161 39.13% 60.87% 100.00% Pearson chi2(l) 1.0057 Pr 0.316 Although a suspect?s response to Miranda is not signi?cantly re- lated to either the prosecutor?s charging decision or the likelihood of conviction, it is signi?cantly related to the process by which the sus- pect?s case will be resolved As Table 12 below indicates, a suspect who waives Mimnda is twice as likely to have his case resolved through plea bargaining, and this difference is highly significant And in my sample more than 98% of the plea bargains re- sulted in convictions. That a suspect?s decision to waive his Miranda rights signi?cantly increases the likelihood that his case will be re- 1996] INSIDE THE INTERROGATION ROOM 289 solved by plea bargaining con?rms Neubauer?s earlier ?nding??D and may be the most notable effect of a suspect?s response to the pre?inter- rogation Miranda warnings. Presumably, the greater evidence accu- mulated against suspects who Speak to their interrogators (and likely provide them with incriminating information) accountsfor this statis- tically signi?cant relationship. However, it is also possible that this relationship results from the selection bias created by Miranda; those suspects who waive their constitutional rights and allow police interro- gation may be more cooperative individuals and thus more predis? posed toward less adversarial means of case resolution such as plea bargaining, while those suspects who invoke their Miranda rights may be more inclined to press their claims aggressively through the court system. Table 12 THE RELATIONSHIP BENEEN Manama AND Pm BARGAINING Whether Suspect Case Was Suspect's Response Resolved by Plea Bargaining To Miranda Warnings No Yes Total Waived 69 65 134 51.49% 48.51% 100.00% Invoked 28 9 37 75.68% 24.32% 100.00% Total 97 74 171 56.73% 43.27% 100.00% Pearson chi2(l) 6.9076 Pr 0.009 The ?nal stage of the criminal process in which a suspect?s re- sponse to the Miranda waiver may exert an effect is sentencing. In particular, one might reasonably expect that suspects who waived their Miranda rights during interrogation would be likely to receive more severe sentences than those suspects who had invoked their rights. Although suspects who waive their Miranda warnings are more likely to receive punishment than their counterparts who invoke, the di?'erences in the severity of punishment they receive are not statisti? cally signi?cant, as Table 13 indicates. 5? Neubauer, note 39, at 109-10. 290 RICHARD A. LEO 86 Table 13 RELATIONSHIP BETWEEN MFAMA AND SENTENCE Suspect's Response Severity of Suspect's Sentence To Miranda Warnings None Low Medium High Total Waived 48 46 15 15 124 38.71% 37.10% 12.10% 12.10% 100.00% Invoked 15 9 2 6 32 46.88% 28.13% 6.25% 18.75% 100.00% Total 63 55 17 21 156 40.38% 35.26% 10.90% 13.46% 100.00% Pearson chi2(3) 2.6350 Pr - 0.451 Even if we control for conviction exclude from our analysis those suspects who were not convicted), the relationship between a sus- pect?s response to the Miranda warnings and the severity of his sen- tence remains statistically insigni?cant B. THE SUCCESS 0F INTERROGATIONS Why do some suspects confess while others manage to resist po? lice pressures to incriminate themselves? What social and legal cir- cumstances make the probability of a successful interrogation more or less likely? Several earlier studies have attempted to answer these questions. Leiken found that younger suspects were much more likely to confess than older ones, and that suspects without a prior criminal record were more likely to confess than suspects with a prior record.? The differences captured in his data, however, were not statistically signi?cant at the .05 level.62 Leiken also found that socially disadvan- taged suspects (as measured by years of education) were no more likely to confess than more socially privileged suspects.65 Neubauer found that suspects without criminal records were substantially more likely to confess than suspects with criminal records, and that suspects accused of property crimes were more likely to confess than suspects accused of crimes against persons.? But younger suspects were no more likely to confess than older ones nor were suspects from disad? vantaged social groups any more likely to confess than suspects from privileged social groups.? Neubauer theorized that what really ex- Lcikcn. supra note 41. at 19-20. 62 Id. at 19-21. 63 Id. at 20. 6* Neubauer, supra note 39, at 104. 65 Id. at 105. 1996] INSEE TI-HE ROOM 291 plained the differential confession rate in his data was the evidence against a suspect prior to interrogation, which, he posited, was typi- cally much higher in property crime cases than crimes against per- sons.66 Like Leiken, however, Neubauer failed to provide tests of signi?cance to support his assertions, and thus we _do not know whether the strength of the associations in his data were arbitrary. Employing more sophisticated methods, Wald et al. found that suspects without a prior record were signi?cantly more likely to pro- vide incriminating information during interrogation than suspects with prior criminal records, and that suspects were signi?cantly more likely to provide incriminating information during interrogation the stronger the evidence against them prior to custodial questioning.? Additionally, in the more serious o??enses suspects were signi?cantly more likely to provide incriminating information during interroga~ tion.?58 However, Wald et al. found no statistically signi?cant relation- ship between either the suspects? race or age and their likelihood to confess.?59 Unlike Wald et al., Leiken, and Neubauer, I examined the effects of a wide range of sociological and legal variables on the likelihood of successful interrogation outcomes. None of the sociological vari: ables-"the class, race, or gender of the suspects, victims, or of?cers? were signi?cantly related to the likelihood of obtaining incriminating information from the suspect.70 Neither were many of the legal and case speci?c variables.7 Although my data con?rm Neubauer?s (un- substantiated) assertion that the strength of the evidence against a sus- pect prior to interrogation is signi?cantly higher in property crimes 55 Id. at 106. 5'7 Wald et al., supra note 17. at 1643-48. 53 Id. 69 Id. at 1644-46. 70 Multivariate regression con?rmed the CHI2 ?nding that suspects' demo- graphic data is not signi?cantly related to succeuful interrogation outcomes. 71 It is possible that unsuccessful or successful interrogation outcomes were accounted for by the suspect's innocence or guilt rather than by the length of interrogation or the number of tactics. Since I could not tell whether a suspect was innocent or guilty prior to his interrogation, this possibility cannot be altogether falsi?ed. However, it seems highly unlikely for at least two reasons. First, many unsuccessful interrogations were the result of a suspect invoking his Mimnda rights, and. as we have seen, suspects with prior felony records were four times more likely to invoke their Miranda rights than suspects without any prior criminal record, a ?nding that is statistically signi?an it stands to reason that suspects with prior felony records, who were more likely to be guilty than suspects without a prior criminal record. thus accounted for a disproportionate number of the unsuccessful interrogation outcomes. Second, the strength of the evidence prior to the interrogation was the best indicator of the suspect's innocence or guilt prior to interrogation. and it was not signi?cantly related to successful interrogation outcomes. When controlling for the strength of the evidence prior to the interrogation, however, the length of interrogation and number of tactics remain statistically signi?cant at the .01 level. 292 RICHARD A. LEO 86 than in crimes against persons they do not support his argu- ment that there is a signi?cant relationship between the type of crime and the likelihood of confession. Nor do my data con?rm Leiken?s assertion that younger suspects are much more likely to provide in- criminating information during interrogation than older suspects. Nor do my data corroborate Wald et al.?s ?ndings that the absence of a prior record, the strength of the evidence prior to questioning, and the seriousness of the offense exert a statistically signi?cant effect on the likelihood that the suspect will provide incriminating information during interrogation. The only variables in my sample that were signi?cantly related to the likelihood of a successful interrogation'were the number of tactics employed by detectives and the length of the interrogation That the number of metics employed by detectives and the length of interrogation are signi?cantly related to the likelihood of confession suggests that the effort and energy expended by detectives is one of the most important factors in explaining successful interro- gation outcomes. The more interrogation tactics detectives use, the more likely they are to ?nd something that works. The longer detec- tives interrogate, the more likely they are to wear the suspect down and elicit incriminating statements. C. THE EFFECT OF INTERROGATION TACTICS Although the earlier empirical studies mentioned above all ana- lyzed the e?ect(s) of selected sociological and legal variables on the likelihood of confession, no study has ever examined (either qualitaf tively or quantitatively) the effect of interrogation tactics on the likeli- hood that suspects will provide incriminating information to detectives during interrogation. Yet according to the rhetoric of po- lice interrogation training manuals and courses, as well as the conven- tional wisdom in police culture, the interrogation tactics that a detective uses should be the decisive in?uence in a suspect?s decision to provide police with incriminating information.72 The Miranda Court echoed a similar sentiment when it excoriated police interroga- tion training texts for compelling confessions through subtle and sophisticated questioning methods.73 My statistical analysis in the previous section revealed that the number of interrogation tac- tics detectives employ during custodial questioning is signi?cantly re- lated to the likelihood of obtaining incriminating information from suspects Are certain interrogation methods and strategies 72 See Leo, supra note 18, at 67-127. 73 Miranda v. Arizona, 384 U.S. 436, 448-55 (1966). 1996] INSEE THE INIERROGATION ROOM 293 also signi?cantly likely to be effective at eliciting incriminating admis- sions from custodial suspects? And is the use of these and other inter- rogation tactics by detectives socially and/ or legally patterned? in other words, under what circumsrances or conditions do some interro- gation tactics yield a statistically signi?cant likelihood of eliciting in- criminating information from a suspect? As Table 14 indicates, when detectives employed certain interro- gation tactics they were signi?cantly likely to elicit incriminating infor- mation from suspects.74 In particular, the tactic of identifying contradictions in the suspect?s denial of involvement was successful at eliciting incriminating information in 91% of the interrogations in MN which it was used the tactic of offering the suspect a moral justi?cation or excuse for his behavior was successful in 90% of the interrogations the tactic of using praise or ?at- tery was successful in 91% of the interrogations and the tac- tic of appealing to the suspect?s conscience was successful in 97% of the interrogations in which it was used Although these four tactics were the on] ones that exercised a statistically signi?cant effect on successful the tactic of appeal- ir?ig to the importance of cooperating with legal authorities (p<.098) and using Behavioral Analysis Interview questions? (p<.090) were also highly likely to be effective. 7* Since 71 di??erent detectives participated in the 182 interrogations in my sample, we can safely rule out the possibility that this table measures the techniques that only good interrogators use. Most of the interrogations in my sample did not involve the same of- ?cers. Moreover, as Table 5 indicates, several tactics were commonly used in a great many of the interrogations. 75 The Behavioral Analysis Interview consists of a structured set of non-investigative hy- pothetical questions that are thought to evoke particular behavioral responses from which interrogators are taught to ascertain the truthfulness of suspects? responses and infer de- ception prior to commencing formal interrogation. See Firm E. [New Er AL, CRIMINAL Imuocsnow AND CONFFSIONS 63-68 (3d ed. 1986). lnbau et al. recommend approxi- mately 15 questions to pox to the suspect. ranging from general questions. such as why does the suspect think someone would have committed the crime, to speci?c ones, such as would the suspect be willing to take a polygraph. Inbau et argue that guilty suspects react defensively and with discomfort to these questions; they equivocate. stall. and provide evasive or noncommittal answers. By contrast, innocent suspects are thought to produce cooperative. direct, and spontaneous responses to these questions. In their introductory and advanced training seminars, the Chicago-based ?rm of Reid 8: Associates advise inter- rogators to treat as guilty any suspect whose answers to four or more of the ?fteen ques- tions appear deceptive to the interrogator. 294 RICHARD A. LEO 86 Table 14 THE Err-?ram OF INDIVIDUAL INTERROGATION Tacncs . Success"6 Interrogation Tactic Rate MOST SUCCESSFUL INTERROGATION TACTICS Appeal to the suspect?s conscience 97% .001? Identify contradictions in suspect?s story 91 .000* Use praise or ?attery 91 .005* Offer moral justi?cations/ excuses 90 LEAST SUCCESSFUL Touch suspect in a friendly manner 88% .225 Invoke metaphors of guilt 87 .327 Any Behavioral Analysis Interview (BAI) questions 84 .090 Appeal to the importance of cooperation 84 .098 Appeal to detective's expertise/ authority 84 .133 Confront suspect with false evidence of guilt 83 .241 Minimize the moral seriousness of the offense 81 .414 Undermine suspect?s con?dence in denial of guilt 80 .182 Confront suspect with existing evidence of guilt 78 .168 Appeal to the suspect's self-interest 77 .760 ?p<.01 The effectiveness of speci?c interrogation tactics also varies by the social characteristics of the suspects under questioning as well as the legal characteristics of their cases. Table 15 displays the relation- ships between the techniques that are signi?cantly likely to yield in- criminating information by the social and legal variables in my data for which the number of observations available was large enough to warrant statistical analysis. These ?ndings tell us not merely which police techniques are most likely to be successful, but also the tactics to which different suspects are most likely to be vulnerable. For exam- ple, younger suspects seemed far more vulnerable to appeals of con- science and justi?cation, perhaps because they are more naive, inexperienced, or idealistic than older suspects, who, by contrast, seemed far more vulnerable to pragmatic appeals based on self-inter- est and the strength of evidence suggesting their guilt. 75 As we saw in Table 7, detectives were successful at eliciting incriminating information in 64% of the cases in my sample. This ?gures rises to 76% if we exclude the cases in which suspects invoked one or more of their Miranda rights and questioning subsequently ceased. This ?gure. then. is the base rate of success against which the percentages in Table 14 should be compared. 77 I excluded from my analysis any interrogation tactic that was not used in at least 15 (or approximately 10%) of the interrogations because otherwise the number was too small to permit adequate statistical analysis. 1996] INSIDE TIE MERROGATION ROOM 295 Interestingly, some interrogation techniques that were not gener~ ally signi?cant become signi?cantly likely to yield incriminating infor~ mation under certain conditions. For example, suspects who are below middle class and suspects with prior felony records were signi?~ cantly likely to be vulnerable to physical evidence ploys. Due to the lack of variation in many of the independent variables in my data, however, explicit comparisons in the ef?cacy of techniques and the vulnerability of suspects could not always be made. For example, we can see- the speci?c tactics to which men, minorities, and below mid? dle class suspects in my sample were most vulnerable, but I can pro- vide no such data for their female, white, or middle class counterparts because so few of my subjects fell into these categories. Moreover, as we will see below, detectives were signi?cantly likely to use more inter? rogation tactics against certain types of suspects and in certain types of cases, thus increasing the possible number of, efficacious tactics in those cases and against those suspects. For example, detectives were signi?cantly likely to use more interrogation tactics the more serious the crime (p<.042) as well as in those cases in which the strength of the evidence prior to the interrogation was not already high Table 15 EFFECTIVENESS or Immocanou TACTICS BY SOCIAL AND LEGAL CHARACTERISTICS OF THE CASE Variable/Success?il Tactics Success Rate Younger Suspects (Less than 30 Years Old) Identify Contradictions 90% .018 Use Praise or Flattery 93 .015 O??er Moral Rationalizations 97 .002 Appeal to Suspect's Conscience 100 .003 Older Suspects (Older than 30 years) BAI Questions 90 .024 Identify Contradictions 92 .004 Confronting Suspect with Existing Evidence 80 .004 Appeal to Suspect's Self-Interest 78 .048 Male Suspects Identify Contradictions 90 .001 Offer Suspect Moral Rationalizations 90 .008 Use Praise or Flattery 93 .004 Appeal to Suspect?s Conscience 97 .001 Minority Suspects Identify Contradictions 93 .000 O?'er Suspect Moral Rationalizations 93 .007 Use Praise or Flattery 91 .030 Appeal to SuspeCt's Conscience 100 .002 Appeal to Importance of Cooperation 88 .055 296 RICHARD A. LEO 86 Suspects Below Middle Class Identify Contradictions 93% .000 Confront Suspects with False Evidence 88 .050 O?'er Suspect Moral Rationalizations 93 .001 Use Praise or ?attery 93 .004 Appeal to Suspect?s Conscience 97 .002 Touch Suspect in a Friendly Manner 100 .028 Suspects With Prior Felony Records Use BAI Questions 94 .007 Identify Contradictions 95 .004 Confront Suspect with False Evidence 96 .027 Offer Suspect Moral Rationalizations 96 .013 Use Praise or Flattery - 95 .033 Appeal to Suspect's Conscience 95 .030 Suspects Without Prior Felony Records Identify Contradictions 86 .031 Use Praise or Flattery 87 .044 Appeal to Suspect's Conscience 100 .009 Crimes Against Persons Identify Contradictions 90 .001 O?'er Suspect Moral Rationalizations 90 .014 Use Praise or ?attery 91 .022 Appeal to Suspect's Conscience 97 .003 Appeal to Importance of Cooperation 86 .048 Crimes of Low or Medium Seriousness Identify Contradictions 97 .002 Offer Suspect Moral Rationalizations 93 .011 Appeal to Suspect?s Conscience 100 .006 Crimes of High Seriousness Identify Contradictions 86 .054 Use Praise or Flattery 100 .016; Strength of Evidence is Low or Medium Prior to Interrogation Identify Contradictions 91 .001 Offer Suspect Moral justi?cations 88 .036 Use Praise or Flattery 89 .031 Appeal to Suspect's Conscience 95 .013 Appeal to Importance of Cooperation 89 .008 Strength of Evidence is High Prior to Interrogation Offer Suspect Moral Rationalizations 95 .040 Appeal to Detective?s Expertise 100 .042 Appeal to Suspect?s Conscience 100 .033 ?p<.05 D. THE LENGTH AND EFFORT or cusronuu. mrmocxnons What factors determine the amount of time detectives put into interrogating suspects and attempting to elicit incriminating informa- tion from them? When are detectives more likely to interrogate sus- 1996] INSIDE THE MERROGATION ROOM 297 pects aggressively? How does the time and effort detectives expend during custodial questioning vary by the legal, case-speci?c, and socio- logical factors in my data? Which relationships are statistically signi?cant? My data revealed a statistically signi?cant relationship between the amount of time detectives spend interrogating suspects and three other variables: the seriousness of the offense, the success of the inter- rogation, and the gender of the victim. The more serious the crime, the longer detectives spent attempting to elicit incriminating informa- tion from the suspect The high seriousness crimes were more than twice as likely to result in long interrogations (more than one hour) than low seriousness crimes (42% vs. conversely, a crime of low seriousness was approximately three times as likely to result in a short interrogation (less than thirty minutes) than a crime of high seriousness (53% vs. Not surprisingly, the length of the interrogation is also signi?cantly related to its success suc- cessful interrogations were six times more likely to last more than one hour than unsuccessful ones (36% vs. conversely, unsuccessful interrogations were more than twice as likely to be under thirty min? utes than successful ones (58% vs. Finally, the gender of the victim was also signi?cantly related to the length of the interrogation in my sample If the gender of the victim was female, the interrogation was more than twice as likely to be long (46% vs. This ?nding is likely due to the interrogations of suspects accused of rape in my sample, interrogations which almost always lasted more than one hour and in which the victim was always female. However, this ?nding is dif?cult to assess because 89% of the suspects in my sample were men, thus rendering an extremely small comparison group. Like the amount of time detectives spend interrogating suspects, the number of tactics detectives employ during custodial questioning was also signi?cantly related to three independent variables: the seri- ousness of the offense, the race of the suspect, and the strength of evidence against a suspect. The more serious the offense, generally the more interrogation tactics detectives employ in their attempts to gather incriminating information from custodial suspects Interrogations for crimes of low seriousness were more than twice as likely to last under thirty minutes than either crimes of medium seri- ousness or high seriousness (53% vs. 26% vs. Not surprisingly, the number of interrogation tactics detectives employed during their interrogations of suspects accused of crimes against persons was signif- icantly higher than for those accused of property crimes Detectives were also signi?cantly likely to employ more tactics during i U) 298 RICHARD A. LEO 86 their interrogation of minority suspects However, the small number of non-minority suspects makes this ?nding dif?cult to assess because 85% of the suspects in my sample were nonwhite, once again rendering an extremely small comparison group. Finally, the strength of the evidence against a suspect prior to questioning is signi?cantly related to the number of tactics detectives are likely to employ during interrogation Generally, detectives tend to use fewer tactics when the evidence against a suspect is already strong and there ap- pears to be little need to obtain more incriminating evidence. Detec- tives in my sample were more than twice as likely to use a high number of interrogation tactics when the evidence against the suspect prior to questioning was either weak or intermediate than when it was strong (47% vs. E. THE EFFECT OF CONFESSIONS ON CASE PROCESSING What happens to suspects who incriminate themselves during in? terrogation? What eifect does providing incriminating statements, ad- missions, and confessions in the interrogation room have on the likelihood that a suspect will subsequently be charged and convicted? The process through which a suspect?s case is resolved? The severity of sentencing? Is it true, as critics have argued, that once a suspect confesses to police his case is largely over; in effect, the rest of the judicial process is mostly form rather than substance?m This study suggests that confessions may well be the most damn- ing and persuasive evidence of criminal guilt. a ?nding that con?rms the beliefs of many detectives and prosecutors, as well as the outcome of mock jury experiments.79 Incriminating statements provided to poi lice during interrogation cast a long shadow over the defendant?s fate within the criminai justice system. Suspects who provide incriminating in- provide incriminating information during interrogation. As Table 16 below indicates, suspects in my sample who incriminated themselves during interrogation were 20% more likely to be charged by prosecutors 24% less likely to have their cases dismissed 25% 73 SeeCharles Ogletree, Are ConfusionsRmUy Miranda. 100 Hanv. L. REV. 1826 (1987); Arthur Sutherland, Crimeand Confession. 79 HARV. L. RBI. 21 (1965). 79 SxSmou. supra note 23:1eo. Soda! Control. supra note 17, at 99; Gerald R. Miller 8: F.joseph Boner, 77:121qu Trial: inPw- csrouocv IN THE LEGAL Pam 19 (Bruce Dennis Sales ed.. 1977). 30 The statistically signi?cant differences at this stage of the criminal process remained even when I recoded parole or?probation violations at a decision by the prosecutor not to charge the suspect Still, a suspect who provided police with incriminating information . 1996] INSIDE THE MERROGATYON ROOM 299 Table 16 THE Emcr or PROVIDING INCRIMINATING STATEMENTS 0N CASE PROCESSING Stage of Criminal Process Percentages Whether Suspect Charged .006? Successful Interrogation 76% Unsuccessful Interrogation 56 Whether Suspect's Case Was Dismimed Successful Interrogation 29 Unsuccessful Interrogation 53 Whether Suspect?s Case Resolved .001* By Plea Bargaining Successful Interrogation 52 Unsuccessful Interrogation 27 Whether Suspect Was Convicted .001* Success?il Interrogation 69 Unsuccessful Interrogation 43 Severity of Sentence Received .012? NONE Successful Interrogation 33 Unsuccessful Interrogation 57 LOW SENTENCE (Less than One Year) Successful Interrogation 43 Unsuccessful Interrogation 21 MEDIUM SENTENCE (One to Five Years) Successful Interrogation 19 Unsuccessful Interrogation 13 LONG SENTENCE (More than Five Years) Successful Interrogation 13 Unsuccessful Interrogation 13 more likely to have their cases resolved by plea bargaining and 26% more likely to be found guilty and thus convicted Suspects who incriminated themselves during interrogation were also signi?cantly likely to receive more punishment following their convic- tion These ?ndings con?rm the View of many criminal jus- tice professionals that what happens during police interrogation will be fateful for the subsequent processing of the suspect?s case in the criminal justice system.82 during interrogation was 24% more likely than his tight-lipped counterpart to be charged by prosecutors 3? Moreover, if we examine only those cases in which the suspect spoke to detectives after the Miranda admonition those cases in which an interrogation actually oc- curred). those suspects?whether or not they invoked their Miranda rights and whether or pot argue; incriminated themselves?were 35% more likely to be eventually convicted . 32 Multiple regression anahmes that controlled for whether an interrogation was suc- 300 RICHARD A. LEO 86 F. SUMMARY In sum, there are three sets of important ?ndings. First, police use many of the standard interrogation tactics taught by training ?rms and advertised in training manuals; in my sample of cases the detec- tives employed an average of 5.62 tactics per interrogation. If a por- trait of the typical interrogation emerges from the data, it involves a two-prong approach: the use of negative incentives (tactics that sug- gest the suspect should confess because no other course of action is plausible) and positive incentives (tactics that suggest the suspect will in some way feel better or bene?t if he confesses). In approximately 90% of the interrogations I observed, the detectives confronted the suspect with some evidence (whether truthful or false) of his guilt and then suggested that the suspect?s self-interest would be advanced if he confessed. Of the many interrogation tactics that detectives em- ployed, only four were signi?cantly likely to result in a successful inter- rogation: the appeal to the suspect?s conscience identifying contradictions in the suspect?s alibi the use of praise or flat? tery and offering moral justi?cations or ex- cuses to account for his behavior The second important ?nding concerns the e?ects of Miranda on the criminal process. Almost one-fourth of the suspects in my sample chose to invoke their Miranda rights and thus either prevent or termi- nate police questioning. This is a far higher percentage than we. might expect, since the conventional wisdom among police profes- sionals and in the academic research literature is that virtually all sus- pects choose to waive their rights and speak to police.33 Nevertheless, detectives successfully elicited incriminating information from sus- pects in 64% of the total number of cases in my sample, and in 76% of the cases in which any questioning occurred. The effects of Miranda on the subsequent processing of a suspect's case were limited, how- ever. Suspects who waived their Miranda rights were 4% less likely to have their cases charged by prosecutors and 10% more likely to be convicted than suspects who invoked, but neither of these differences were statistically signi?cant, nor was the relationship between a sus- pect's response to Miranda warnings and the severity of punishment statistically signi?cant. The only statistically signi?cant e?'ect of Mi- randa in the criminal process was that suspects who waived their Mi? randa rights were twice as likely to have their case resolved by plea cessful by the strength of the evidence prior to the interrogation con?rmed the (II-112 ?ndings in Table 16. In other words, as mentioned above, whether an interrogation is successful is signi?cantly related at the .01 level to the likelihood that a suspect will be treated di??erently at every subsequent stage of the criminal process. 83 Leo. supra note i=3- 1996] INSIDE THE MERROGATION ROOM 301 bargaining than suspects who invoked their Miranda rights And in my sample more than 98% of the plea bargains ultimately re- sulted in guilty verdicts. The third, and perhaps most important, ?nding is that what hap- pens inside an interrogation room exercises a statistically signi?cant effect on the subsequent processing of a suspect?s case'at every stage in the criminal justice system. While a suspect?s response to Miranda was not signi?cantly likely to affect the subsequent processing of his case, a suspect?s decision to provide detectives with incriminating in- formation was fateful. Suspects who incriminated themselves during interrogation were signi?cantly more likely to be charged by prosecu- tors signi?cantly less likely to have their case dismissed signi?cantly more likely to have their cases resolved by plea bargaining signi?cantly more likely to be convicted and signi?cantly more likely to receive more punishment than their counterparts who did not provide interrogators with any incriminating information These ?ndings o?er support for the widely held view among criminal justice of?cials that admissions and confessions are highly persuasive and damning evidence of guilt against a criminal suspect.? . V. CONCLUSION: CIDSING THE GAP I began this Article by pointing to the familiar contrast between how law is written in the books and how it is actually practiced by legal actors in the social world, arguing that the gap in our knowledge be- tween legal ideals and empirical realities remains as wide as ever in the study of American police interrogation. In this Article I have tried to ?ll in this gap?a gap that has widened considerably in the last two decades due to the complete absence of any empirical research on police interrogation practices?by providing quantitative data from the almost 200 interrogations I observed in nine months of partici- pant observation ?eldwork at three police departments. I have sys- tematically described the patterns in police techniques, suspect behavior, and interrogation outcomes in all of the 182 cases I observed, and I have analyzed the effects of police interrogation practices, Miranda warnings, and incriminating statements on the sub- sequent stages of the criminal process in my entire sample. I have sought to bring the reader inside the interrogation room in order to understand the characteristics, context, and outcomes of interroga? tion and confession in ordinary criminal cases that are not likely to make the published record on appeal. Although this Article breaks 3" leo.Soa'alConlromemnote I7.at99. 302 RICHARD A. IEO 86 new ground as the ?rst empirical study of its type in more than two decades, our understanding of contemporary American police inter- rogation practices and outcomes remains highly incomplete. lfwe are to close the gap in our knowledge between the ideal and reality that the Miranda Court decried, we need more primary data and empirical studies of everyday police investigative practices, especially in other regions of the country. This study offers a number of ?ndings with important academic, legal, and policy implications. For example, the Miranda Court la? mented the tactics advocated in police interrogation training manuals and texts for exploiting the weaknesses of criminal suspects and threatening to overbear their rational decision-making capacity.??Yet Unis study indicates that these techniques?undEmining a susp?ct?s ml con?dence in his denial of guilt, o?'ering moral justi?cations for his behavior, and confronting suspects with fabricated evidence of their guilt, to name but a few?appear to be exceedingly common in con- temporary Arnerican police interrogations. In addition, this study sug- gests that detectives have become increasingly successful at eliciting incriminating information from custodial suspects in the last thirty years; that one in ?ve custodial suspects invokes his or her constitu- tional right to avoid cooperating with custodial police questioning; that most of the suspects who invoke their Miranda rights to silence or counsel have prior criminal records; that very few everyday police in- terrogations are ?coercive? by contemporary judicial standards; that the overwhelming majority of everyday police interrogations last less than one hour, and that suspects who provide incriminating informa- tion to detectives are signi?cantly more likely to be treated differently at every subsequent stage of the criminal process (from charging to sentencing) than their counterparts who do not. These ?ndings con- ?rm the view of many criminal justice professionals that what happens inside the interrogation room exerts a fateful effect on the processing of a defendant?s case at every subsequent stage in the criminal justice system.86 One might thus ask: what has been the effect of Miranda? It is to this question that I will turn in the second article of this two part se- ries. In that article, I will examine the broader legal, social, and polit- ical context in which the empirical ?ndings of this study must be interpreted. In particular, I will analyze the general impact of the Ma'- randa warnings?which have remained an ongoing source of legal and political controversy since their judicial creation almost thirty years 35 Leo. supu note 13, at 67-127. 35 Se: Leo, Social Control, supra note 17, at 99: SIMON, supra note 23, at 193-207. 1996] INSHJE THE INTERROGATION ROOM 303 again?on the behavior, ideology, and culture of contemporary Amer- ican police interrogators. 3" Set :upm text at note 55.