Judicial gatekeeping of police-generated witness testimony.

• Author: Thompson, Sandra Guerra
• Position: III. Pretrial Reliability Hearings in Criminal Cases through V. Conclusion, with footnotes, p. 364-395

3. PRETRIAL RELIABILITY HEARINGS IN CRIMINAL CASES

   In addition to best practices for the police, researchers have recommended that courts conduct pretrial reliability hearings for police-generated testimonial evidence. (181) The recommendations for pretrial reliability screening call into question the traditional practices of trial courts in admitting possibly unreliable prosecution evidence and would require a new approach to admitting these three common forms of evidence. For confessions and eyewitness identifications, courts have routinely exercised a limited gatekeeping role confined to reviewing the prosecution's evidence for possible constitutional violations. (182) Remarkably, informant testimony is generally admitted without any type of reliability or constitutional screening by courts. (183)

   The following sections reveal the inadequacies of current due process screening as a means of ensuring the reliability of police-generated witness testimony. They also show that courts have engaged in reliability gatekeeping under the rules of evidence for various types of lay witness testimony, most often to exclude defense witnesses.

   1. DUE PROCESS FAILS TO ENSURE RELIABILITY

      Due process under federal law has only limited effectiveness in regulating police-generated witness testimony. With regard to informant testimony, due process fundamental fairness may call for exclusion, but only in cases where the defense can show "outrageous government conduct" in using the informant to manufacture the evidence and set up a person who was not otherwise involved in criminal activity. (184) The more typical jailhouse informant provides information about a person who has already been arrested, so this application of due process would not apply. For jailhouse informant testimony, the central constitutional protections require limited disclosure and discovery of "impeachment material" regarding the informant. (185) Due process has not been applied to require a reliability assessment by the courts.

      For confessions and identifications, due process protects against inappropriate police procedures in gathering the evidence, which may affect reliability. However, unreliability without more does not mandate exclusion under federal due process law. The Supreme Court has made it clear that no matter how unreliable a confession may be, unless it is a product of overreaching state action, no due process violation occurs. (186) Moreover, the defense must show that the overreaching by the police caused the defendant to confess. (187) In Colorado v. Connelly, the defendant, who suffered from schizophrenia, confessed and provided critical details about a previously unsolved murder. (188) There was no suggestion that the

      to conduct an effective cross-examination. See NATAPOFF, supra note 2, at 58-60 (noting limited nature of discovery requirements); see also Raeder, supra note 132, at 1439-47 (arguing that current constitutional jurisprudence does not promote sufficiently high standards of ethical behavior relating to the disclosure of informant information). police had in any way coerced Connelly to confess. The Court thus found no due process violation.

      Viewed purely as a matter of evidentiary reliability, Connelly's confession may well have been reliable. If he provided the police with important information about a murder that would only have been known to the killer--and which was not even known to the police at the time--then it is likely his confession was true. The point here is that under the due process analysis, courts are not required to make a reliability determination as a condition of admissibility. Indeed, the Supreme Court has noted that reliability per se should be treated as an evidentiary concern, not a constitutional one. (189)

      What makes matters worse is that even when there is evidence of police lies, threats, and other coercive conduct, the due process voluntariness test fails to ensure the reliability of the suspect's statements. After reviewing thousands of cases from the 1990s and 2000s challenging the voluntariness of confessions, Paul Marcus concludes:

      One necessarily comes away with a feeling of being unclean and tainted by government activities that are not honorable even given the environment needed for interrogations. Many judges allow confessions into evidence in cases in which police interrogators lied and threatened defendants or played on the mental, emotional, or physical weaknesses of suspects. While judges write that they do not condone such conduct and find such practices repugnant, reprehensible, or deplorable, some of those same judges have upheld the admission of such confessions that result from those practices after applying the totality of circumstances test. (190) It bears noting as well that the due process voluntariness test does not take into account whether critical best practices have been followed, such as videotaping, lack of contamination by suggesting facts of the crime, and independent corroborating evidence.

      In judging eyewitness identifications, the Supreme Court considers reliability the "linchpin" of its due process analysis. (191) Yet scholars have consistently bemoaned the many failures of the test as a reliability-screening tool. (192) First, as with confessions, due process only protects against inappropriate evidence-gathering procedures. But unreliability due solely to other causes is not a sufficient ground for exclusion. (193) The fundamental unfairness of being convicted largely or even solely on patently unreliable identification evidence has no federal due process traction, unless the police also acted in an unduly suggestive way. (194)

      Second, even if the police do use unduly suggestive procedures, due process does not require exclusion if the identification is nonetheless reliable. Reliability, as the Court defines it, does not take police suggestiveness into account. (195) The Court's reliability checklist ignores the several ways in which police suggestiveness can produce and reinforce a false identification. Instead, it provides what has come to be an exclusive list of five factors that bear on reliability. Four of the factors take into account characteristics of the witness and the circumstances surrounding the viewing, which are appropriate considerations in a reliability assessment. (196) However, one of the factors listed by the Court (which federal courts are still required to consider) has been scientifically shown not to have a bearing on reliability--"the level of certainty demonstrated by the witness at the confrontation." (197) One might take the view that this reliability test is mostly appropriate and good enough, until one considers all the other relevant factors that are missing. The most crucial omissions are whether the identification was cross-racial and whether the witness was intrinsically less reliable due to age, mental ability, or intoxication. (198) It also bears reiterating that whether the identification was "reliable" (applying the Court's definition of the term) ultimately does determine admissibility, but the Court's definition of reliability omits the critical factors of police suggestiveness or coercion, witness vulnerability, and cross-racial identifications. (199)

      Some state courts have tweaked the Manson test by discarding factors that have been shown not to correlate with reliability and incorporating important factors that do bear on reliability. (200) These judicial fixes have not gone so far as to mandate that police follow best practices.

      In its landmark decision of State v. Henderson, the New Jersey Supreme Court completely sidestepped the Manson due process test for identifications on the grounds that it fails to meet any of the reliability goals set for it. Instead the court replaced the test with a mandate for pretrial reliability hearings which take into account whether the police have followed best practices. (201)

      In short, neither the "voluntariness" test for confessions nor the "reliability" test for identifications ultimately protects against unreliable evidence. The many wrongful convictions produced by confessions and identifications that passed muster under these anemic constitutional tests lay bare this truth. (202)

   2. CURRENT PRACTICES IN RELIABILITY SCREENING FOR LAY WITNESS TESTIMONY

      While police-generated witness testimony has not traditionally been subject to effective reliability screening, there is certainly precedent for such screening in relation to other types of evidence. The most instructive example involves the testimony of the lay witness whose memory has been hypnotically refreshed. The Supreme Court found that the preferred approach is for courts to conduct pretrial reliability screening to ensure that the evidence has been generated by professionals following accepted protocols.

      Hypnotically refreshed witness testimony raises reliability concerns strikingly similar to those raised by police-generated witness testimony: (1) hypnosis makes a witness more vulnerable to the memory-distorting effects of suggestion; (2) biased interrogators may use suggestive questioning in order to elicit certain desired statements; (3) the process of suggestive questioning of a hypnotized subject may cause the subject to "confabulate"; (203) and thus (4) a witness may experience a heightened and unwarranted degree of confidence in the memory. (204)

      In Rock v. Arkansas, when the defense offered to introduce the defendant's own hypnotically refreshed testimony, the prosecution filed a motion to exclude the testimony, and the trial court held a reliability hearing to determine its admissibility. (205) The trial court, applying a per se rule excluding such testimony as unreliable, limited the defendant's testimony to memories that had been shared with the examiner before hypnosis. (206) In overturning the defendant's conviction, the Supreme Court found that such a per se exclusion violated the defendant's...