Journal of Criminal Law and Criminology

Winter 1998

 

THE CONSEQUENCES OF FALSE CONFESSIONS: DEPRIVATIONS OF LIBERTY AND

MISCARRIAGES OF JUSTICE IN THE AGE OF PSYCHOLOGICAL INTERROGATION

 

Richard A. Leo

Richard J. Ofshe

 

I. Introduction

 

A. DEFINING THE PROBLEM

 

  Because a confession is universally treated as damning and compelling evidence of guilt, [FN1] it is likely to dominate all other case evidence and lead a trier of fact to convict the defendant. [FN2]  A false confession is therefore an exceptionally dangerous piece of evidence to put before anyone adjudicating a case.  In a criminal justice system whose formal rules are designed to minimize the frequency of unwarranted arrest, unjustified prosecution, and wrongful conviction, police-induced false confessions rank amongst the most fateful of all official errors.

 

   As many investigators have recognized, the problems caused by police-induced false confessions are significant, recurrent, and deeply troubling.  [FN3]  Police elicit false confessions so frequently that social science researchers, legal scholars, and journalists have discovered and documented numerous case examples in this decade alone. [FN4]

 

  Yet no one knows precisely how often false confessions occur in the United States, how frequently false confessions lead to wrongful convictions, or how much personal and social harm false confessions cause.  This is because: (1) no organization collects statistics on the annual number of interrogations and confessions or evaluates the reliability of confession statements; (2) most interrogations leading to disputed confessions are not recorded; and (3) the ground truth (what really happened) may remain in genuine dispute even after a defendant has pled guilty or been convicted.  [FN5]  These problems prevent researchers from defining a universe of confession cases, sampling a subset, and confidently determining the truth or falsity of each underlying confession.

 

  Until these methodological obstacles are overcome, no one can authoritatively estimate the rate of police-induced false confessions or the annual number of wrongful convictions caused by false confessions. [FN6]  The lack of such information also prevents researchers from estimating the full magnitude of personal and social harm that police-induced false confessions cause: the days and months innocent persons spend in pre-trial incarceration; the resources, time, and dollars wasted prosecuting and defending them; the months and years defendants languish in prison after wrongful conviction; and the additional crimes carried out by the true perpetrators.

 

  Although it is presently not possible to estimate the magnitude of harm caused by false confessions, this article sheds light on another dark corner of the problem by addressing the following questions: What is the impact of demonstrably unreliable confession evidence on criminal justice officials? What are the consequences of false confessions on defendants as they move through the criminal justice system?  And how much influence does a false confession alone exert on the decision-making of jurors?

 

 B. FALSE CONFESSIONS AND THE ADMINISTRATION OF JUSTICE

 

  Following Edwin Borchard's pioneering study of miscarriages of justice, [FN7] a series of investigators [FN8] have documented numerous cases of wrongful arrest and conviction of the innocent in the United States.  [FN9]  We continue the tradition of research into errors in the criminal justice system by reporting a study of sixty cases of police-induced false confessions in the post-Miranda era, [FN10] and by analyzing the consequences of these errors affecting defendants as they move through the criminal justice system. [FN11]

 

  We suggest that confessions are regarded as the most damning and persuasive evidence of guilt simply because most suspects who confess are guilty, and because most confessions are corroborated by additional evidence.  Under these conditions, however, it is impossible to isolate the effect of the defendant's "I did it" admission [FN12] on the decision-making of criminal justice officials and juries because the confession co-varies with inculpatory witness or physical evidence.  The research reported here isolates the effect of a defendant's "I did it" statement on the decision-making of criminal justice officials and juries by studying only cases in which the defendant's confession is not supported by any physical or reliable inculpatory evidence.  The research design thus allows measurement of the effect of an untrue admission when a detective, prosecutor, judge or jury is required to weigh the admission against evidence that would ordinarily establish the defendant's innocence.

 

  This article explores whether contemporary American psychological interrogation practices continue to induce false confessions like the third degree methods that preceded them.  This article also analyzes how likely police-induced false confessions are to lead to the wrongful arrest, prosecution, conviction, and incarceration of the innocent.  And this article examines with field data [FN13] whether confession evidence substantially biases a trier of fact even when the defendant's statement was elicited by coercive methods. [FN14]  We explore this issue with cases in which the defendant's statement has not only been coerced but is also demonstrably unreliable, and in which other evidence proves or strongly supports the defendant's innocence.

 

  Part II of this article discusses the selection and classification of the sixty disputed confession cases under study. [FN15]  Part III describes the findings of our research.  Part IV analyzes the deprivations of liberty and miscarriages of justice associated with the sixty cases described in this article.  Finally, Part V discusses the import of this research and offers some concluding remarks.

 

 II. Method

 

A. SELECTION AND CLASSIFICATION

 

  Cases of disputed confessions were identified through multiple sources: electronic media database searches; directly from case files; [FN16] and from secondary sources.  The sixty cases discussed below do not constitute a statistically adequate sample of false confession cases.  Rather they were selected because they share a single characteristic: an individual was arrested primarily because police obtained an inculpatory statement that later turned out to be a proven, or highly likely, false confession.

 

  Based on the information that we obtained and reviewed, all of the cases studied satisfy the following conditions: no physical or other significant and credible evidence indicated the suspect's guilt; [FN17] the state's evidence consisted of little or nothing more than the suspect's statement "I did it;" and the suspect's factual innocence was supported by a variable amount of evidence--often substantial and compelling--including exculpatory evidence from the suspect's post-admission narrative. [FN18] For every case included in this study, there was no credible evidence corroborating the defendant's "I did it" admission or supporting the conclusion that he was guilty. [FN19]

 

  Based on the strength of the evidence indicating a defendant's probable innocence, each case was classified into one of three categories: proven false confession; highly probable false confession; or probable false confession.

 

  For the thirty-four cases classified as proven false confessions, the confessor's innocence was established by at least one dispositive piece of independent evidence. [FN20]  For example, a defendant's confession was classified as proven false if the murder victim turned up alive, the true perpetrator was caught and proven guilty, or scientific evidence exonerated the defendant.  Not only was the confessor definitively excluded by dispositive evidence, but the confession statement itself also lacked internal indicia of reliability.  Any disputed confession case that fell short of this standard--no matter how questionable the confession and no matter how much direct or circumstantial evidence indicated the suspect was innocent--was excluded from this category.

 

  For the eighteen cases classified as highly probable false confessions, the evidence overwhelmingly indicated that the defendant's confession statement was false. [FN21]  In these cases, no credible independent evidence supported the conclusion that the confession was true.  Rather, the physical or other significant independent evidence very strongly supported the conclusion that the confession is false.  In each of these cases, the confession lacked internal reliability.  Thus, the defendant's statement is classified as a highly probable false confession because the evidence led to the conclusion that his innocence was established beyond a reasonable doubt.

 

  For the eight cases classified as probable false confessions, [FN22] no physical or other significant credible evidence supported the conclusion that the defendant was guilty.  There was evidence supporting the conclusion that the confession was false, and the confession lacked internal indicia of reliability.  Although the evidence of innocence in these cases was neither conclusive nor overwhelming, there were strong reasons--based on independent evidence--to believe that the confession was false.  Cases are included in this category if the preponderance of the evidence indicated that the person who confessed was innocent.

 

  We recognize that for any case that could not be classified as a proven false confession, there is a possibility that our classification of the case might be in error.  Despite strong evidence supporting the conclusion that the confession is false, it remains theoretically possible that one or more of the defendants we classify as false confessors may have committed the crime. Nevertheless, we believe that the disputed confessions discussed in this article would be judged false by an overwhelming majority of neutral observers with access to the evidence we reviewed. [FN23]

 

 B. POST-ADMISSION NARRATIVE ANALYSIS

 

  When evaluating the likelihood that a person committed a crime, investigators should first consider witness statements, biological evidence linking the suspect to the crime (fingerprint, DNA, hair, etc.), and alibi evidence.  The identification by an eyewitness, the identification of the person as the donor of one or more type of biological material found at the crime scene, and the lack of an alibi all point to guilt.  By contrast, an opposite pattern of evidence (e.g., no match with eyewitness descriptions, exculpating biological evidence, and the existence of an unimpeachable alibi) all support innocence.

 

  In addition to these traditional sources of evidence, the defendant's post-admission narrative of the crime may provide helpful evidence of guilt or innocence, assuming contamination [FN24] has been eliminated.  If a suspect has made an "I did it" admission and given a post-admission narrative of a crime, the fit--or lack thereof--between the contents of the narrative and the crime scene facts provides evidence of guilt or innocence.  Evaluation of the fit can reveal that a suspect possesses the sort of accurate, personal knowledge of the specifics of the crime that the perpetrator would be expected to have, or it can demonstrate the suspect's ignorance of the crime because his answers about the crime scene evidence are grossly incorrect. [FN25]

 

  The fit between the specifics of a confession and the crime facts determines whether the "I did it" admission should be judged as reliable or unreliable evidence.  There are at least three indicia of reliability that can be evaluated to reach a conclusion about the trustworthiness of a confession. Does the statement: (1) lead to the discovery of evidence unknown to the police? (e.g., location of a missing weapon that can be proven to have been used in the crime, location of missing loot that can be proven to have been taken from the crime scene, etc.); (2) include identification of highly unusual elements of the crime that have not been made public? (e.g., an unlikely method of killing, mutilation of a certain type, use of a particular device to silence the victim, etc.); or (3) include an accurate description of the mundane details of the crime scene which are not easily guessed and have not been reported publicly? (e.g., how the victim was clothed, disarray of certain furniture pieces, presence or absence of particular objects at the crime scene, etc.).

 

  If, for example, a suspect's post-admission narrative either leads the police to missing evidence, or reveals that the suspect knew precisely how the victim was bound and mutilated, or which window was jimmied open with what sort of unlikely tool, then the suspect possesses actual knowledge of the crime that would reasonably be expected of the perpetrator.  Therefore, the suspect's confession should be deemed reliable.  If, on the other hand, the suspect is unable to provide police with accurate information revealing evidence not already known to them (e.g., where to locate the murder weapon or the loot), is demonstrably wrong about the method of killing, or is demonstrably inaccurate about the specifics of the crime scene, then the statement should be judged unreliable and, if anything, treated as evidence of innocence.  Therefore, the statement should be seen as lacking evidence of actual knowledge--something to be expected of a false confessor who has not been contaminated by the police or due to leakage of information into the community.

 

  When the police elicit a post-admission narrative from a suspect, they typically seek only information about major crime elements (e.g., location of the missing weapon, type of mutilation, etc.).  However, a suspect's report about the mundane (but unique or improbable) details of the crime and the crime scene is of great value in establishing a suspect's guilt or innocence.  [FN26]  This is true, in part, because the suspect's knowledge of mundane details is less likely to be the result of contamination by the police.  Mundane details are less likely to have been mentioned during off-tape conversations or during the pre-admission phase of an unrecorded interrogation.

 

  A suspect's post-admission narrative need not demonstrate indicia of reliability in each category for it to reveal personal knowledge of the crime. It is generally accepted that one or more aspects of a crime may be so heinous that a guilty party may refuse to state them even while admitting to other major components of the crime.  For example, Richard Allen Davis, who admitted to kidnapping and killing a child, was not willing to admit that he also raped her. [FN27]  Nevertheless, if a defendant has been properly and thoroughly debriefed, his personal knowledge of the crime should allow him to supply sufficiently detailed information to prove a confession's reliability by demonstrating his specific knowledge of what happened (e.g., the circumstances of the kidnapping, the child's clothing, the location of the killing ground, the description of the killing scene, etc.), even if he resists confessing to certain particularly heinous acts.

 

 C. POLICE-INDUCED FALSE CONFESSION

 

  Police-induced false confessions arise when a suspect's resistance to confession is broken down as a result of poor police practice, overzealousness, criminal misconduct and/or misdirected training. [FN28]  Interrogators sometimes become so committed to closing a case that they improperly use psychological interrogation techniques to coerce or persuade a suspect into giving a statement that allows the interrogator to make an arrest.  Sometimes police become so certain of the suspect's guilt that they refuse to even-handedly evaluate new evidence or to consider the possibility that a suspect may be innocent, even when all the case evidence has been gathered and overwhelmingly demonstrates that the confession is false.  Once a confession is obtained, investigation often ceases, and convicting the defendant becomes the only goal of both investigators and prosecutors.  As the investigative process progresses, some interrogators, who overstepped procedural boundaries to obtain a false confession, engage in criminal conduct to cover up their procedural violations (e.g., coerce false witness statements, suborn perjured testimony from snitches, or perjure themselves at suppression hearings or at trial).  Furthermore, some prosecutors who are determined to convict obstruct justice by withholding exculpatory evidence from the defense.  [FN29]

 

  American police are poorly trained about the dangers of interrogation and false confession.  Rarely are police officers instructed in how to avoid eliciting confessions, how to understand what causes false confessions, or how to recognize the forms false confessions take or their distinguishing characteristics. [FN30]  Instead, some interrogation manual writers and trainers persist in the unfounded belief that contemporary psychological methods will not cause the innocent to confess [FN31]--a fiction so thoroughly contradicted by all of the research on police interrogation [FN32] that it can be labeled a potentially deadly myth.  This fiction perpetuates the commonly held belief that only torture can cause an innocent suspect to confess, and it allows some police to rationalize accepting coerced and demonstrably unreliable confession statements as true. [FN33]

 

 D. FALSE CONFESSION CASES

 

*              *              *              *

 

 III. Findings

 

A. PROVEN FALSE CONFESSIONS

 

  There are four sub-types of proven false confessions: the suspect confessed to a crime that did not happen; the evidence objectively demonstrates that the defendant could not possibly have committed the crime; the true perpetrator was identified and his guilt established; or the defendant was exonerated by scientific evidence.

 

 1. The Suspect Confessed to a Crime That Did Not Happen

 

  Police interrogators may extract a confession to a crime that did not, in fact, occur.  In Austin, Texas in 1990, after twice failing a polygraph test, Billy Gene Davis confessed that he killed his ex-girlfriend; she subsequently turned up alive in Tucson, Arizona. [FN136]  Even if the underlying event did in fact occur, police may induce a confession to a non-existent crime.  In 1993, Phoenix, Arizona police elicited a confession from Christina Mason to killing her three-month-old infant by letting another woman inject the child with heroin and cocaine to prevent it from crying. [FN137] The autopsy, however, revealed no drugs other than Tylenol in the baby's body, and the medical examiner concluded that the likely cause of death was pneumonia or a viral infection. [FN138]

 

 2. The Defendant Could Not Have Committed The Crime

 

  Police may extract a confession from an individual who could not have committed the crime.  In 1987, Los Angeles, California police interrogators elicited false confessions from two suspects--Ruben Trujillo and Pedro Delvillar--to the same double murder and robbery. [FN139]  Yet both men were in police custody (one in a county jail and the other at a California Youth Authority facility) for other crimes when the murders occurred. [FN140]  In another example of flawed interrogation, police in Laguna Beach, California obtained a confession to arson from Jose Soto Martinez in 1993, but prosecutors dismissed charges when they discovered that Martinez had been in a Mexican prison at the time of the arson. [FN141]  Similarly, in 1986 Montana police elicited a false confession to a sexually motivated killing from Ivan Reliford, but later discovered that Reliford was in custody when the crime was committed. [FN142]

 

  The cases in this study reveal many reasons why someone could not have committed the crime to which he confessed.  In 1973, Connecticut State Police elicited a confession from Peter Reilly to killing and mutilating his mother.  [FN143]  After a jury trial, conviction, and then reversal by an appellate court, the prosecutor handling the second trial discovered that the former prosecutor's files contained documents showing that Reilly arrived at the scene of the murder only minutes before the police and thus could not have committed the crime. [FN144]

 

  In 1982, James Harry Reyos confessed in New Mexico that he had killed a Catholic priest a year earlier. [FN145]  The victim died between 7 p.m. and midnight in Odessa, Texas, [FN146] but gas receipts and an eyewitness established that Reyos was in Roswell, New Mexico (200 miles away) at 8 p.m. that evening, [FN147] and a speeding ticket proved that he was also in Roswell shortly after midnight. [FN148]  To have committed the murder, Reyos would have had to drive 200 miles to the murder site, kill the priest in no more than one minute and speed 215 miles back to where he received the speeding ticket--in four hours (averaging well over 100 miles an hour on narrow, country roads).  Eventually the state's attorney handling Reyos' appeal conceded that Reyos could not have committed the crime. [FN149]

 

  In 1995, in Portland, Oregon, police extracted false confessions from Rick Nieskins and Christopher Cole to the 1991 murder of John Sewell.  [FN150]  Both men were charged with homicide, and both spent thirteen months in jail awaiting trial--even though two other men had been convicted of Sewell's murder in 1991 and had always maintained that they acted alone.  [FN151]  Prosecutors eventually dropped charges against Nieskins after records showed that he could not have committed the crime because he was at a homeless shelter in Seattle at the time of the killing. [FN152]  Once they acknowledged Nieskins' false confession, prosecutors admitted that Cole also could not have been involved in the crime and dropped charges against him.  [FN153]

 

 3. The True Perpetrator Was Identified and His Guilt Established

 

  Police may elicit a confession that is proven false when the true perpetrator is identified.  Sometimes this occurs fortuitously when police encounter the perpetrator in connection with another crime and obtain a demonstrably reliable confession.  In 1979, after twenty-one hours of interrogation by West Virginia State Police, Paul Reggetz confessed to murdering his wife and two children.  [FN154]  Reggetz spent eleven months in pre-trial incarceration before one of his neighbors confessed. [FN155]  In 1990, Suffolk County, New York police interrogated Anthony Atkinson for three-and-a-half hours before he confessed to murder and sodomy. [FN156]  Later, two other men confessed to the crime, and charges against Atkinson were dismissed. [FN157]  In 1994, Guy Lewis confessed to Memphis, Tennessee police to shooting and killing his girlfriend. [FN158]  The prosecutor was preparing to bring charges against him when Tony Hedges and Michael Maclin were arrested and each confessed to the murder. [FN159]  In 1996, Robert Moore confessed to the capital murder and robbery of a taxi driver after Nassau County, New York detectives interrogated him for twenty-five hours. [FN160]  Moore was released only because police happened to arrest one of the actual killers on unrelated charges, and he confessed and identified his two co-perpetrators. [FN161]  In 1996, in Daytona Beach, Florida, police extracted a confession to capital murder and robbery from Donald Shoup, a mentally handicapped teenager. [FN162]  While Shoup was awaiting trial, the true killer confessed. [FN163]

 

  In one of the century's most dramatic and disturbing false confession cases, prosecutors dismissed charges against three false confessors after routine detective work identified the true killers. [FN164]  In 1991, during interrogations that lasted up to twenty-one hours, Maricopa County Sheriffs in Phoenix, Arizona coerced false confessions from Leo Bruce, Mark Nunez, and Dante Parker to the mass murder of nine persons at a Buddhist temple.  [FN165]  While prosecutors were preparing capital cases against the defendants, a ballistics test was carried out on a rifle that was picked up for testing the same day that Bruce, Nunez, and Parker were interrogated.  [FN166]  It proved to be the murder weapon.  The rifle had been in the possession of Jonathan Doody and Alex Garcia the night of the murder.  Searches led to the discovery of loot in the possession of both Doody and Garcia.  Both adolescents confessed to the murders, and Garcia supplied the police with a detailed account of how he and Doody planned and carried out the killings.  [FN167]

 

  Garcia not only confessed to the nine Temple murders, but also to murdering Alice Marie Cameron shortly before being arrested for the Temple murders. [FN168]  The police delayed doing the ballistics test on the rifle that led to Garcia's arrest because they were occupied first with coercing false confessions from Bruce, Nunez, and Parker and then with the media storm and public protests against the police that followed the disputed confessions.  [FN169]

 

  To make matters even worse, Maricopa County Sheriffs had also extracted a confession to Cameron's murder from George Peterson, a mentally ill adult, during a sixteen hour interrogation. [FN170]  When Garcia admitted to the Cameron murder fourteen months later, Peterson was awaiting trial for capital murder for the same crime. [FN171]

 

 4. The Defendant Was Exonerated By Scientific Evidence

 

  Police may elicit a confession that is conclusively proven false by scientific evidence.  In 1996, police in West Palm Beach, Florida elicited a confession to capital murder from Martin Salazar, but prosecutors dropped charges when the defense discovered that fingerprint evidence clearing Salazar had been withheld by the police and the prosecutor. [FN172]  During an interrogation in 1980, Chicago police reshaped a dream by Steven Linscott into a murder confession, but DNA testing established his innocence many years later. [FN173]  In 1983, Virginia police elicited several confessions from Earl Washington--including one to the rape and murder of Rebecca Williams.  [FN174]  In 1993, DNA evidence established that Washington could not have been responsible for any of these crimes. [FN175]  In 1996, in Sitka, Alaska, Richard Bingham confessed to being the lone rapist and killer of seventeen-year-old Jessica Baggen. [FN176]  DNA testing excluded Bingham as the source of the semen found in the victim. [FN177]  The foreign hair found on the victim's body was not Bingham's nor was the fingerprint found on a cigarette pack at the crime scene. [FN178]  Bingham was also unable to describe the unusual properties of the physical scene where the body was found nor the unusual way in which the victim had been silenced. [FN179]  Bingham was acquitted at trial. [FN180]

 

  Notwithstanding the numerous examples of proven false confessions reported in this article, it is difficult to establish conclusively that a defendant's confession is false even when the evidence of innocence is compelling.  Once a suspect has confessed, it is rare for the crime to evaporate, for the true perpetrator to be apprehended, for police or prosecutors to discover that the defendant could not have committed the crime, or for scientific evidence to exonerate him.  The standard for inclusion into the proven false category--established innocence--is a formidable barrier.

 

 B. HIGHLY PROBABLE FALSE CONFESSIONS

 

  While our research has unearthed numerous examples of highly probable false confessions, only a small number of these cases are reported here.

 

 1. Bradley Page

 

  In 1984,