By Mary A. Connell, Ed.D., ABPP, presented at training seminars for Texas Criminal Defense Lawyers Project, Longview, Texas 2001, El Paso, Texas 2002

The eyewitness is clearly one of the most, if not the most, important players in the criminal justice system. Without eyewitness testimony, the great majority of criminals would evade prosecution. Therefore, the idea that eyewitnesses frequently provide erroneous testimony is jarring to the legal system. Despite the fact that eyewitness research represents psychology’s strongest area of research, and over 2000 studies, most extremely well designed, have been conducted in the past twenty years in this area, the courts have barred eyewitness testimony more frequently than any other category of testimony that psychologists might be asked to provide. To call into question the cornerstone of the prosecution’s strength in most of its cases might be expected to bring the criminal justice system to its knees. However, advances in DNA research have made it possible to obtain definitive information regarding the guilt of the accused and the convicted in a substantial number of cases, so that retrospectively, eyewitness testimony is seen to have been in error. Seventy-four convictions have been proven wrong by DNA testing since the late 1980s, and 81% of those cases had been successfully prosecuted on the basis of mistaken identification. Arizona was the first state to allow testimony about problems with eyewitness identification, but until the mid-1990s, many states failed to follow suit. Presently, only Kansas and Tennessee continue by law to bar such testimony, but many trial courts across the nation continue to exclude it.

When expert testimony regarding eyewitness research is allowed into evidence, it may pertain to any of a number of aspects. Customarily, the forensic psychologist might be asked to review a case and then to provide expert testimony regarding what the research indicates about the methods that were employed to collect eyewitness testimony in the case at hand. There might be testimony regarding the relationship between witness confidence and accuracy, or about the effects of the conditions of observation upon witness accuracy. Issues regarding lineup and photoarray procedures might be offered, and the procedures used in the particular case might be reviewed accordingly. For the case involving a child witness, the forensic psychologist could draw upon regarding methods of interviewing child witnesses in order to preserve accurate testimony, effects of repeated questioning upon child witness reliability, and so forth. Additionally, in cases where there may be a number of conflicting eyewitness reports, the psychologist might be called upon to provide an explanation for why such differences might occur, in order to assist the trier of fact ascertain how much weight to give various eyewitness accounts. It is essential that the expert who is testifying tie the relevant research to the particular case at hand. It is not necessary that the expert witness be a researcher in field of eyewitness reliability, although prosecutors may object to the testimony on the basis that the testifying psychologist is not a researcher and has not worked with lead researchers in the field, and is therefore not an expert. In anticipation of this objection, it may be helpful to consider pairing the practicing psychologist with a research psychologist, asking them to review the case together, and anticipating that the testimony of both may be needed at trial.

In an effort to gain some appreciation for the breadth of research available in this field, we shall look at several categories of eyewitness research. First, the area of child witness accuracy and what can affect it will be examined, with a look at some of the important research by Ceci, Bruck, Loftus, Poole, and others who have made contributions. Second, eyewitness identification procedures and the related research will be examined. The important work of Well, Malpass, Penrod, Cutler, Folero, and others will be briefly reviewed. Finally, general issues about eyewitness reliability will be examined. Again, the important works of Penrod, Cutler, Lindsay, and others will be briefly summarized.

In cases where there are allegations of child sexual abuse, the statements of the child witness may be an extraordinarily important part of the prosecution’s case against the accused. In addition to providing testimony about the research regarding normal sexual behavior and non-abused children, characteristics of sexually abused children, characteristics of familial perpetrators of sexual abuse, characteristics of sexual predators who sexually abuse children outside the family, and research regarding the effects of anatomical dolls on child witness accuracy, for example, psychologists might also be asked to testify regarding the factors that could influence the child’s testimony. It is important to understand that the expert witness should never be asked to comment on whether the child is telling the truth, or on whether the child’s statements are credible or believable. Those are questions for the trier of fact, and psychology has no tools or special expertise to determine whether someone is telling the truth. Rather, what can be said is how various forms of interviewing or questioning might be expected to impact upon the child’s reports, how the child’s cognitive capacity may be expected to influence his or her ability to recall details, discuss times, and make sense of experiences that may be unfamiliar, and why a child’s reports may vary over time, be rich or poor in detail, or be recanted.

Another productive area of research with regard to which psychologists might be able to provide useful testimony is the area of interviewer bias. When the person questioning the child, whether it is a therapist, a CPS investigator, or a family member, has a strong opinion that abuse has occurred, that opinion may influence what the child ends up saying. A classic study that examined the affects of interviewer bias upon the accuracy of children’s reports was conducted by Ceci, Leichtman, & Bruck (1995). This study, commonly referred to as "Simon Says," involved pre-schoolers playing a game similar to "Simon Says." A month after they played this game, in the second phase of the study, social workers were given a one-page report about the play episode. Some social workers got accurate information about the play episode, and some social workers got erroneous information. The erroneous information would lead those social workers to believe that certain things might have happened in the play episode that did not happen. The task was to conduct an interview to determine what each child could recall about the original play episode. As might be expected, the information provided to the social workers influenced their beliefs about what had transpired, and powerfully influenced the outcome of the interviews. When the social worker had correct information about the play episode, the children interviewed correctly recalled 93% of all events. When the social worker was misinformed, the children provided significantly less accurate information. Fully one-third of the younger group of children (three to four-year-olds) and roughly one-fifth of the older children (five to six-year-olds) gave statements saying that some of the "falsely alleged" events had occurred. Indeed, the children were often reluctant in affirming what the social worker believed, but over time, they capitulated and became more certain, and even provided collaborative details. It was further found that two months after this second portion of the study, the children maintained their stories about what had happened in the play episode. The children who had given false reports continued to do so, even after that period of time, and with a different interviewer asking questions. The children were brought to say that such things happened as marbles being inserted in their ears, or that they had touched a child’s toe, when in fact, they had not. Critics of this study might argue that the children were not being asked about traumatic events that had occurred to them, or were suspected of having occurred. They were not being asked if a loved one had done something terrible to them. Further, they were not being asked these questions by people upon whom they depended and trusted, as a child might depend upon and trust a parent. As is the nature of psychological research, these studies are constructed or designed to measure a specific operation, and the studies that follow may be expected to more highly refine our understanding of the vicissitudes of that operation, controlling more variables. These studies layer, one atop another, until a substantial base of knowledge is amassed.

In other studies, the effects of stereotyping were examined. These studies are particularly important in looking at child sexual abuse allegations in the throes of custody litigation. When a person is systematically identified as a bad person, we might then ask whether it is easier to get a child to make false allegations about that person. Studies that have been done in this area indicate that interview bias has a quite powerful effect upon interview outcome. In one study, known in the literature as "The Incrimination of Dale," researchers examined the effects of asking children what a man had done while they played together in the lab (Lepore & Sesco,1994). What actually had happened was that Dale, the man, played with some toys with each child, and at some point asked the child to help him take off his sweater. Then, the children were questioned by researchers under one of two formats. In one format, the interviewer asked neutral questions about what Dale had done, and about other things that had happened while Dale was in the room. In the other condition, the interviewer asked similar questions, but responded to the child’s answers in a way that insinuated that Dale was a bad person. The interviewer would say, "He wasn’t supposed to do that. That was bad. What else did he do?" Eventually, the interviewer asked even more leading questions, such as, "Didn’t he take off some of your clothes, too?" and "Other kids have told me that he kissed them, didn’t he do that to you?" and finally, "He touched you and he wasn’t supposed to do that, was he?" As might be expected, children who were asked the more leading and suggestive questions gave many more inaccurate responses to an eventual inventory of what had happened. They endorsed allegations that Dale had done bad things, at least a third of them embellishing their incorrect responses to make Dale sound even worse. They identified people that Dale had touched, told of him kissing certain children on the lips, and told of how he took off certain pieces of his clothing. They maintained these inaccurate and embellished stories for a week, and often even made voluntary statements about Dale, further castigating his character (Lepore & Sesco, 1994).

With regard to anatomically detailed dolls, the research that has been conducted on non-abused children’s play suggests that non-abused children do not generally demonstrate explicit sexual activity with the dolls, but do sometimes show great interest in the genitalia of such dolls. When fourteen pre-school children were interviewed with the dolls, and the interviewer asked whether anyone had touched their genitalia, and then followed up with more specific questions, 50% of the children inserted their finger in the opening for the vagina or anus, and most of the children stroked or tugged the penis (Herbert, Grams, & Goranson, 1987). The researchers also reported that most of the children tried to terminate the interview, an activity often observed by children in actual interviews of alleged sexual abuse. Clearly, behaviors that are often cited by "investigative therapists" as indicative that a child has been sexually abused may be behaviors that fall within the purview of normal childhood behavior. Anatomical dolls, while they may be useful to assist children in talking about or identifying parts of the body or activities for which they have few words, also may serve as interesting objects of curiosity to children who have never seen them. It is important not to overinterpret such behaviors as indicative of sexual abuse, and when a therapist does so, expert testimony about this area of research may be useful.

Ornstein, Ceci, & Loftus (2000) noted that when children two and three years of age experience sexual abuse, they may have no words to describe what they have experienced. They may understand that something is wrong if they experience physical pain associated with penetration, but fondling or "accidental" touching may not be perceived by the child as abusive. The child’s perceptions of touching related to hygiene and touching that is fondling may be blurred. The older the child gets, the more likely it is that this distinction can be clearly made. What research has not yet made clear is whether, as the child gets older and learns that certain kinds of touch are inappropriate, the child can then bring into consciousness events that may have occurred during those earlier years. There is no doubt that "post event misleading information" can alter what the child may still experience as "memory." As the authors point out, people have recalled nonexistent "…broken glass and tape recorders, a clean-shaven man as having a mustache, straight hair as curly, stop signs as yield signs, hammers as screwdrivers, and even something as large and conspicuous as a barn in a bucolic scene that contained no buildings at all" (Ornstein, Ceci, & Loftus, 2000, p. 1038). Reporting accuracy following misinformation, the "misinformation effect," is reported to be as large as 30% or 40% (Ornstein, Ceci, & Loftus, 2000).

Young children are known to be disproportionately vulnerable to suggestive influences. Pre-schoolers are more susceptible to influences of misleading post-event information than are older children and adults, and children can certainly be caused to make false reports based on this information. When children were told to "…think real hard about each" suggestion, or to "…try to remember if it really happened," pre-schoolers were inaccurate between 25% and 44% of the time in their judgments of false events, but were always accurate regarding true events, in a study conducted by Ceci, Huffman, Smith, & Loftus (1994).

Finally, children have tremendous difficulty distinguishing between perception and imagination, particularly during the pre-school years (Orenstein, Ceci, & Loftus, 2000). They could not distinguish whether they had said a word or imagined saying it, in certain studies, but could certainly distinguish whether they or someone else had said it. Pre-schoolers were more error prone in distinguishing real or imagined acts or words when both concerned themselves, but they could distinguish as accurately as adults whether the act was performed or the words were spoken by themselves or others.

Another interesting body of research within the arena of child witness accuracy is that surrounding the question of how children respond to repeated questioning. It is hypothesized that children come to believe, when they give the same answer to a question four or five times, and the question keeps being asked, that the answer they are giving is not the right answer. Then, we assume, they try a different answer. Thus, if a child is asked whether somebody has touched him or her inappropriately, and the child says that the person did not, and the child is asked that question again and again, he or she may eventually provide a different answer. We see this again and again in reviewing therapy notes when children are being treated in order to "obtain an outcry." In addition to being repeatedly questioned, the child may also be read books at these sessions, about good touch, bad touch, and secret touch, about being able to say "no," and so forth. The child may further be asked many questions about the suspect, wherein the therapist/investigator attempts to determine whether the child has negative feelings about that person. Thus, the idea of bad touch and the aura or incrimination of a specific person may be transmitted to the child, so that there are multiple forces working upon the child.

In one such study of the effects of repeated questioning, children were shown an ambiguous event and then questioned immediately following the event and one week later, in one trial, or only one week later, in the second trial (Poole & White, 1991). What was found was that four-year-olds were most likely, of four, six, and eight-year-olds, to change their responses, both within interviews when asked repeated questions, and across interview sessions. When they were asked a question about which they could have no information, many children offered speculative responses, rather than saying they did not know. Uncertainty disappeared across time as well, with answers of children of all ages becoming more and more certain (Poole & White, 1991). In actual experience, of course, children are not just asked these questions repeatedly, but there is often an admixture of repeated questions, leading questions, incrimination, and interviewer bias, so that the result may be highly contaminated testimony.

Our interest is not exclusively given to defining areas in which expert testimony can challenge the child witness’s accuracy. Rather, we take a great interest in determining how to best gather data from children so as to preserve their accurate testimony. Unfortunately, it often happens that the well-intentioned attorney leads his or her client somewhat astray in this very process. When, for example, a child custody litigant calls her attorney to say she believes her child has been sexually abused during a visitation with the other parent, the attorney may urge her to take the child to a therapist as quickly as possible, to call Child Protective Services, and to perhaps take the child to a doctor or emergency room. The call to CPS may occur immediately, but it may be that the child will not be interviewed by CPS for several days. In the interim, the child may be taken to a therapist and to the emergency room, and the child may be interviewed several times by different people. In selecting a therapist, the client may choose her own therapist, or someone recommended by the attorney, or may ask friends and acquaintances for the names of therapists who have experience treating child sexual abuse. Generally, such therapists do not have training in how to preserve child witness testimony, and their aim may be to help the child work through the trauma of the sexual abuse and help the parent find ways of protecting the child. Thus, before the child is ever seen by CPS, much less by the custody evaluator who is later asked to further evaluate the child regarding alleged sexual abuse, there may be a number of interviews of a highly leading nature that render eventual testimony useless. Ideally, the child should be protected from interviews until the CPS worker can conduct an interview on videotape. While it may be the standard practice of CPS investigators to interview the child first, without the video camera running, the ideal situation would be to capture that initial interview on camera, in order to insure, as nearly as possible, preservation of the child’s initial statements. Further, the parent should be advised to ask the child nothing further about the issue. This may seem counterintuitive to the parent, but it is of extreme importance. The single most troublesome issue in these cases is often how the parent may have repeatedly questioned the child, or led the child to refine the allegation. Finally, it seems to me that very frequently the attorney advises the mom to withhold the child from an upcoming visit, in order to protect the child. Thus, the issue becomes further clouded by what may be communicated to the child regarding the potential danger of seeing the other parent, the aura of urgency that may surround the matter as protective orders are sought, and so forth. Taking the matter slowly and carefully, and preserving an aura of neutrality in which the child can give his or her statement to the proper authorities in due time, seems to be a much more effective way of insuring that the child who needs protection gets protection.

What should be clear from a cursory review of this literature, then, is that it is rich, far ranging, and illuminating to our understanding of child witness testimony. The psychologist practicing in this area may be able to examine the events surrounding the cultivation of a particular child’s testimony and offer some useful observations to the trier of fact. Those observations, grounded in the research in this area and applied to the facts of the case, may assist the trier of fact in determining how much or how little weight to give each aspect of the witness testimony. When the end product is a child making a statement that her father put his pee-pee in her privates and it got big and then it had white stuff, the average lay person may find it extremely difficult to imagine that the child could make such a statement if it were not the truth. The research has demonstrated clearly, however, that with biased interview techniques, leading questions, misinformation, and negative stereotyping, it is quite possible to lead a child to make such statements, and for the child to then have tremendous difficulty sorting out whether that event, now pictured in the child’s "mind’s eye," occurred or was memory produced by talking about it. No doubt, such inappropriate questioning is often done with the best of intentions. Neither the therapist/investigator, nor the child’s mother, likely consciously construed the potential benefits of gaining a false allegation from the child. Rather, in a climate of distrust and suspiciousness, a child’s normal behavior or behavior reflective of trauma (perhaps surrounding the divorce) may have been taken as indicative of sexual abuse. Four and five-year-old children engage in a wide variety of sexual exploration during normal childhood development. When a child is seen to play with himself or herself, or to show interest in other children’s genitals, this is sometimes perceived as evidence that sexual abuse may have occurred, and the questioner is "off to the races."

As a result of the extraordinary body of research in the area of eyewitness identification procedures, we now have a very clear understanding of the kinds of things that can interfere with accurate eyewitness identification. We have come to understand much more about techniques of interviewing eyewitnesses, techniques for obtaining identifications in lineups and photo arrays, individual differences in eyewitness abilities, the effects of stress and of the presence of weapons upon witness recall, the effects of racial differences between witness and suspect upon recall, and a broad range of other important issues. Gary Wells, a lead researcher in the area of eyewitness identification, has provided good information about the research on his web site, http://www.psychology.iastate.edu/faculty/gwells/homepage.htm. The reader is urged to explore this site for information about this of research, information on individuals who were serving life sentences or were on death row as a result of eyewitness evidence but were exonerated by DNA, and other useful information.

Wells, Small, Penrod, Malpass, Fulero, & Brimacombe (1998) identified some important themes that had emerged from this literature relative to lineup methods. These included "relative-judgment processes," "lineups as experiment-analogy," and "confidence malleability." Their important paper on these three trends, the white paper on eyewitness research, serves as an excellent review of the literature. The authors concluded that review with some recommendations, including the use of double-blind lineup arrangements, eyewitnesses being forewarned that the culprit may not be present in the lineup, choosing distracters that are based on the eyewitness’s verbal description of the perpetrator, and assessing and recording confidence of the eyewitness immediately after an identification. Further information will be provided about each of these areas below; it should be noted, however, that as a result of this important research, the Department of Justice adopted these four recommendations and incorporated them in their own guide, Eyewitness evidence: A Guide for Law Enforcement, and made it available to every law enforcement agency in the country. A copy of that Guide can be downloaded from the official web site of the Department of Justice, http://www.ojp.usdoj.gov. This publication represents a remarkable application of psychological research to a social problem, the first direct such application of its kind, and as such, serves as extraordinary positive testimony regarding the quality of this research.

Wells et al. (1998), in their white paper that was adopted as a position paper by the American Psychology-Law Society (American Psychological Association Division 41), educated ces to this research by providing a comprehensive review of the literature. First, the question of relative judgment theory was explored. Eyewitnesses tend to identify the person in the lineup who looks most like the culprit, relative to the other members of the lineup. This becomes a problem when the subject is not actually in the lineup, and under such circumstances, research indicates that wrong identifications tend to occur. That is, witnesses identify someone, even if the suspect is not in the lineup, and they identify the person who most resembles the suspect. An example of a study of this effect might be that eyewitnesses to a staged crime are shown a lineup where the actual culprit is present. The rate at which he identified is recorded. Another set of eyewitnesses see the same lineup, except that the culprit’s photo is not there. We would expect that the number of people who correctly identified the culprit in the culprit-present lineup would now indicate "none of the above" in the culprit-absent lineup. In one such experiment (Wells, 1993), 200 eyewitnesses to a staged crime were shown either culprit-present or culprit-absent lineups. All eyewitnesses were told that the actual culprit might or might not be present. In the culprit-present group, 54% of the eyewitnesses accurately identified the culprit. In the culprit-absent group, the majority of those 54% of people simply chose someone else. The "none of the above" category only increased from 21% to 32%. A specific suspect, who apparently looked most like the culprit, captured most of the vote (Wells, 1993).

Malpass & Devine (1981) broke ground in demonstrating the importance of telling the eyewitness that the suspect might or might not be present. They found that when they failed to warn the eyewitness of this, 78% of them attempted to identify someone from the culprit-absent lineup. That rate dropped to 33% when they were explicitly warned that the culprit might not be present in the lineup. This is extraordinarily important data, and indicates that the instruction discourages the eyewitnesses from relying on the relative judgment process.

Other studies supported this notion that the relative judgment process results in misidentification of culprits, and eventually, lead researchers to the conclusion that the most accurate identification process is that obtained from sequential presentation of suspects, rather than simultaneous presentation of suspects. This may not be intuitively obvious to triers of fact, a potentially important argument for allowing testimony, if there is an effort to bar it. That is, asked if they believe they could make a more accurate identification by seeing a photo array of six people that included the suspect, versus seeing one photo after another, and having to make an absolute judgment about each of them (either it is the suspect or is not the suspect) without the opportunity for repeated viewing, or comparison, many people might suggest they could do a better job of identification if they had an opportunity to see all of the photos at once. Research consistently indicates that this is not the case, and the most accurate identification can be made through sequential presentation. Nevertheless, it remains the common practice of law enforcement agencies to present photo arrays with simultaneous presentation of six suspects. Cutler & Penrod (1988) found that sequential presentation of suspects substantially improved eyewitness capacity to identify the suspect.

One such experiment involved subjects viewing a video- taped re-enactment of an armed robbery in Session 1, and in Session 2, attempting to identify the culprit from a robber-present or a robber-absent lineup (Cutler & Penrod, 1988). It should be noted that the researchers were testing some other issues as well as the simultaneous versus sequential presentation issue. For the present purposes, we will focus upon the presentation question. Of the 175 college undergraduate subjects, 63 correctly identified the robber from a robber-present lineup. Five identified a foil from a robber-present lineup, 13 incorrectly rejected a robber-present lineup, and 66 correctly rejected the robber-absent lineup. Finally, 28 identified a foil from the robber-absent lineup. The overall proportion of correct identifications was .74. Varying the presentation mode, the researchers found marginally significant improvement in sequential presentation conditions. The false condition rate was .39 in the simultaneous presentation condition, but .19 in the sequential presentation condition, supporting the hypothesis that sequential presentation improves the probability of avoiding a false identification (Cutler & Penrod, 1988).

In their lineups as experiments analogy, Wells et al (1998) offered the following analogy (p. 13):

A lineup is like an experiment: the police have a hypothesis (that the suspect is the culprit); they collect materials that could be used to test the hypothesis (e.g., picture of the suspect and filler pictures), they create a design (e.g., placing suspect’s picture in a particular position in an array); instruct the subject(s) (eyewitness or eyewitnesses); run the procedure (show the lineup to the eyewitness); record the data (identification of the suspect or not); and interpret the hypothesis in light of the data (decide whether the identification decision changes their assessment of whether the suspect is the culprit).

Based on this analogy, they hypothesized that as many things can go wrong in this identification procedure as can go wrong on the typical scientific experiment. For example, there may be some demand characteristics of the situation, such as: pressure for the eyewitness to make a choice; the influence of confirmation bias, such as asking the eyewitness specifically about the suspect, but not asking questions about the other people in the lineup; response biases, such as encouraging the subject who seems to loosely recognize a culprit, until he or she becomes more certain; selectively recording and interpreting the data; and so forth. Other problems, such as making it obvious to the eyewitness which person in the lineup is the suspect, may also confound the results of the process. Double-blind experiments are common in science, because it is understood that interviewer or researcher bias may have a strong and profound impact upon the subject’s performance. It is found, indeed, that more accurate identifications will be made when the person conducting the lineup is not aware of which person in the lineup is the suspect. Nevertheless, it is common practice in police departments for the lineup to be conducted by the main investigator in the case, so that the potential for confounding the identification procedure is great.

Another area in which there is data from research that may not be intuitively obvious is that of eyewitness confidence in identifications. The lay person would assume that the more certain the eyewitness of his or her identification, the more likely the identification is to be accurate. This is one of the most researched questions in all of the scientific eyewitness literature. It turns out that the relationship between accuracy of identification and confidence expressed by the eyewitness is a positive one, as expected, but it is not a substantially positive correlation. Wells, et al. (1998) reported the result of surveys affirming that prosecutors have much more of a tendency to associate eyewitness confidence with accuracy than do defense attorneys, and surveys of the lay public reveal that there is a substantial cross cultural belief in the positive correlation between confidence and accuracy. Further, post verdict interviews of jurors revealed that confidence expressed by an eyewitness regarding identification was the most powerful single determinate of whether jurors would believe the testimony. (Wells, et al. (1998).

Further, research dramatically illustrates the malleability of witness confidence. Wells, et al. (1998) summarized a number of studies that demonstrated this. One such study, conducted by Luus & Wells (1994), involved the use of a staged crime to secure false identifications from 136 eyewitnesses. The eyewitnesses viewed a theft in pairs, and were separated shortly after the theft. After being separated, false identifications were obtained from the eyewitnesses using a photo spread, and the witnesses were not made aware that they had made a false identification. After the identification process, they were randomly assigned to one of several experimental conditions. In the control condition, they were told nothing about the identification decision of their co-witness. In experimental conditions, they were given information that their co-witnesses had identified the same person, or had identified someone else, or had said the culprit was not present in the lineup. They were then interviewed by someone posing as a campus police officer, and were asked about their confidence, on a 10-point scale, in the accuracy of their identifications. There was an extraordinary increase in confidence among those who were told the co-witness had identified the same person (8.8 confidence on a 10-point scale, versus 6.9 confidence in the control condition where they were given no information about their co-witness). Witnesses who were told that their co-witness had identified someone else had the lowest confidence level (4.7), and there were predictable differences for the various control groups along the continuum. Similar results have been found when eyewitnesses were told that the identification they have made is of the suspect (the person the police believe to be the culprit). Thus, the very natural tendency of the lead investigator to express appreciation to the eyewitness who has identified the suspect may have a profound impact upon the outcome of the trial. The witness may come across as far more confident than he or she would be without such feedback, and jurors may give far greater weight to the confident witness identification than they would to a witness who expressed uncertainty.

As a result of this comprehensive review of the literature, Wells, et al. (1998) made four recommendations regarding rules for lineup and photo array procedures. They recommended the following:

Rule 1: Who conducts the lineup. The person who conducts the lineup or photo spread should not be aware of which member of the lineup or photo spread is the suspect.

Rule 2: Instructions on viewing. Eyewitnesses should be told explicitly that the person in question might not be in the lineup or photo spread, and therefore should not feel they must make an identification. They should also be told that the person administering the lineup does not know which person is the suspect in the case.

Rule 3: Structure of lineup or photo spread. The suspect should not stand out in the lineup or photo spread as being different from the distracters based upon the eyewitness’s previous description of the culprit, or based on other factors that would draw extra attention to the suspect.

Rule 4: Obtaining confidence statements. A clear statement should be taken from the eyewitness at the time of the identification and prior to any feedback as to his or her confidence that the identified person is the actual culprit.

Several of these rules have been adopted by the Department of Justice and promulgated in their Guide for collecting eyewitness evidence. Regarding simultaneous versus sequential lineups, the Guide remains neutral. It includes an admonition that the identification procedure should be conducted in a manner that promotes reliability, fairness, and objectivity of the witness’s identification (Guide, Section V.C). Directions are provided for conducting simultaneous photo and live lineups and sequential photo and live lineups. While it is noted that the Guide is not a legal mandate, it is considered by the Department of Justice to be reflective of sound professional practices and of available research in the area.

There has been a great deal of research on the question of how retention intervals affect eyewitness memory for events and for attributes of actors in crime scenes. For review of some of these studies, see Ebbesen & Rienick (1998). In their own study on this issue, the authors found that there were some variations on the generally accepted tenet that memory drops off rapidly and then the rate of decline levels off. This occurred in their study, in the following way. The total number of correct events remembered did decline rapidly and then level off. The memory for personal attributes, however, remained more stable, with small decay in steady increments. Further, they observed that memory loss appeared to be "protected" by repeated recall attempts, such that there was very little overall loss of memory function when there were repeated interviews after the event. Thus, if the witness is interviewed immediately after a crime, and then is interviewed periodically between that time and time of the trial, there may be relatively little memory loss. If, however, the witness is interviewed some time after the crime, memory loss may already have occurred, such that witness recollection may be significantly compromised.

Much of the research regarding eyewitness identification procedures can be generalized to police investigation techniques wherein witness statements have been gathered by investigating officers, utilizing a broad variety of investigation techniques. Psychological research suggests that there are ways to enhance the probability that witnesses give accurate statements, and that nevertheless, police interview techniques often incorporate techniques known to elicit contaminated data. In a study in which eleven interviews conducted by experienced members of a Florida police department were analyzed, Fisher, Geiselman, & Raymond (1987), found that striking features of the interviews were that (a) there was little uniformity in the structure of the interviews; (b) almost all questions about facts were asked in a direct fashion (for example "Was he wearing jeans?"); and (c) no assistance was offered to enhance the witness’s recollection. Analysis revealed that investigators often interrupted the witness, asked many short answer questions, and interfered with the witness narration. Rarely did the interviewers seem to resonate with the witness’s mental imagery of the scene. The interviews would begin with a routine set of questions about the perpetrator’s physical characteristics, in order that they could complete the crime report that they needed to write, while the witness was clearly wanting to talk about the crime that had been observed, for example (Fisher, Geiselman, & Raymond, 1987).

Other studies have demonstrated that witnesses sometimes cannot attribute memory to its appropriate sources, or that they make source attribution errors. When witnesses get information from other witnesses and from the police, then their own recollection is likely to be contaminated (Fisher, 1995). As has been demonstrated, post-event questioning can contaminate children’s testimony, and certainly the same can be said for adult witness recollections. Leading and misleading questions and suggestions are known contaminants, and yet police regularly make suggestive comments during their investigations. Further, research regarding juror assessment of witness credibility applies not just to eyewitness identification testimony, but to all eyewitness testimony. Fisher suggested that one way of improving on this body of research would be to bring the law enforcement community into the process of designing research studies. Experienced prosecuting and defense attorneys could instruct researchers, for example, about aspects of witness’s statements that are susceptible to attack in court. Experienced police officers might offer some suggestions about the kinds of information that are most helpful in apprehending a suspect. Through understanding what kind of information is needed from witnesses, researchers may be able to make increasingly helpful suggestions to law enforcement about how to extract and preserve that kind of information to meet the goals of law enforcement; that is, to close the case or gain a successful prosecution.

The eyewitness evidence Guide includes recommendations to police officers that when they are initially responding to the report of a crime, they should ask open ended questions, avoid asking suggestive or leading questions, and provide the reporter with an opportunity to give any other information that might be known about the incident, beyond that asked by the officer (Guide, Section I.A.2.3.4). Further, it is recommended that witnesses be separated and instructed to avoid discussing details of the incident with other witnesses (Section I.B.5). In the section that instructs investigating officers about how to obtain information from witnesses (Section I.C), the Guide states that the manner in which information is obtained from a witness has a direct impact on the amount and accuracy of that information. It is a stated policy that the officer should obtain and accurately document and preserve information from witnesses, and should engage in several procedures in order to accomplish that. Open-ended questions are recommenced, and it is further recommended that the information be augmented with closed-ended questions. For example, the officer might ask, "What can you tell me about the car?" and then might follow with, "What color was the car?" The interviewer is cautioned to avoid leading questions, such as, "Was the car red?" It is suggested that the witness be encouraged to avoid contact with the media or exposure to media accounts concerning the incident (Section I.C.7), and that the witness be instructed to avoid discussing details of the incident with other potential witnesses (Section I.C.8). These, then, would be some areas that might be worth exploring with the witness offering damaging testimony, with the police investigator who collected the eyewitness evidence, and finally, with the expert witness who can cite the relevant research.

To be sure, the research on factors that may influence eyewitness memory is broadly applicable in the criminal justice system and in civil court matters having to do, for example, with alleged child sexual abuse. Additionally, in civil tort matters, witness reliability or accuracy may be quite important, and the relevant research may be applied there as well. In all cases, it is critically important to focus on events or interview techniques that may have contaminated the witness’s capacity to provide accurate information, rather than on the credibility of the witness. It is important that the focus not appear to be one of trying the witness, or of calling the witness’s honesty into question. No doubt, witnesses attempt to give accurate information to law enforcement authorities. It is the case, however, that there is a range of factors that affect their ability to do so. Research indicates that some factors influencing witness accuracy are witness characteristics, perpetrator characteristics, and event variables or characteristics. Among witness characteristics that might influence accuracy are the witness’s age, state of mind, relationship with the alleged perpetrator, alertness, general cognitive capacity, and comfort providing information or testimony. Perpetrator characteristics that may affect witness reliability are distinguishable or unique features, racial similarity or difference from witness, manner of behavior, and other such variables. Event characteristics that may affect accuracy include the presence of a weapon, the sense of crisis or emergency, the length of time the witness has to observe the perpetrator, lighting and other factors that might affect the witness’s capacity to make an accurate mental record, the recency of the event, the frequency of the event (for repeated episodes of sexual assault, for example), the uniqueness of the situation, and factors that might cause the witness to fear consequences if accurate testimony is provided.

Finally, interviewer techniques obviously interact with witness accuracy in a number of ways that have been carefully researched. Interviewer bias has a profound effect. Leading questions, misinformation, and confirmatory feedback are known contaminants. Interrupting the witness’s narrative, asking closed-ended questions, and failing to formulate or time questions to resonate with the witness’s mental state at the time evidence is being gathered are all potential contaminants. The familiarity of the interviewer may have a profound effect, so that a parent, for example, might get quite different information from a child than would a stranger in a law enforcement uniform, or a doctor in a white coat. Repeating questions, in the face of consistent answers, may communicate to the witness that the answers being provided are not considered correct.

Unwitting communication of the interviewer’s theory of the case is almost inevitable, so that premature formulation of a hypothesis or lack of objectivity is likely to distort witness reports. Witnesses are known to have difficulty attributing source to memory of events or persons, so that child witnesses, for example, may have difficulty distinguishing between a dream, a real event, or an event about which they have been questioned numerous times. Witnesses who hear discussion about a crime in the media, or from other witnesses, may have difficulty recalling where they got pieces of information and may believe they saw something or heard something at the crime scene that, in fact, they did not. Witness certainty is a poor predictor of witness accuracy, a fact that is somewhat counterintuitive. Also counterintuitive is the fact that sequential presentation of photos or suspects in a live lineup will produce more accurate identifications that will simultaneous presentation of the suspect and foils.

The vast body of research about eyewitness accuracy is difficult to summarize in a brief presentation, and it is also somewhat difficult to communicate it in a brief time on the witness stand. It is important that testimony offered about this research be tied to the facts of the case, be kept to a manageable amount of information, and be presented in a way that is interesting to the trier of fact. The ideal situation might be to have the case material reviewed by a practicing psychologist and by a research psychologist with a special interest in this area, and offer opinions from both experts at trial. The researcher would be able to lay a foundation for the relevant research that can be applied to the case at hand, and the practitioner could then tie that research to the case at hand, going through the material point by point to identify those aspects of the case material that might be considered problematic, based on the relevant research.

While most trial judges have probably had some experience hearing the presentation of eyewitness research, certainly most members of the jury will not have had such experience. It is important to take sufficient time to build a foundation for the ideas that are to be presented, and yet it is equally important not to overwhelm the jury with dry data. The testifying expert will benefit from some idea about how much time can be spent in developing the testimony on the stand. If the attorney anticipates getting to the nexus of the expert witness’s testimony within thirty minutes, while the expert witness anticipates spending an hour or more developing the points, then the effectiveness of the testimony is likely to be significantly compromised. If the expert witness has prepared a report that provides the review of relevant studies and summaries of research findings, and then ties those findings to the case at hand, the attorney may be able to work from that report in developing the testimony.

Ceci, S. J. & Bruck, M. (1999). Jeopardy in the courtroom: A scientific analysis of children’s testimony. Washington, DC: American Psychological Association.

Ceci, S. J., Huffmann, M. L. C., Smith, E. & Loftus, E. F. (1994). Repeatedly thinking about a non-event: Source misattributions among pre-schoolers. Consciousness and Cognition, 3. pp. 388-407.

Ceci, S.J., Leichtman, M. D., & Bruck, M. (1995). The suggestibility of children’s eyewitness reports: Methodological issues. In F. Weinert & W.P. Schneider (Eds.), Memory development: State of the art and future directions. pp. 323-347. Englewood Cliffs, N.J.: Erlbaum.

Cutler, B. L. & Penrod, S. D. (1988). Improving the reliability of eyewitness identification: Lineup construction and presentation. Journal of Applied Psychology, 73(2), pp. 281-290.

Ebbesen, E. B. & Rienick, C. B. (1998). Retention interval and eyewitness memory for events and personal identifying attributes. Journal of Applied Psychology, 83(5), pp. 745-762.

Fisher, R. P., Geiselman, R. E., & Raymond, D. S. (1987). Critical analysis of police interview techniques. Journal of Police Science and Administration 15, pp. 177-185.

Fisher, R. P., (1995). Interviewing victims and witnesses of crime. Psychology, Public Policy, and Law, 1(4), pp. 732-764.

Herbert, C. P., Grams, G. D., & Goranson, S. E. (1987). The use of anatomically detailed dolls in an investigative interview: Preliminary study of "non-abused" children. Unpublished manuscript cited in S. J. Ceci and M. Bruck, Jeopardy in the courtroom: A scientific analysis of children’s testimony, (1995). p. 128.

Lepore, S. J. & Sesco, B. (1994). Distorting children’s reports and interpretations of events through suggestion. Journal of Applied Psychology, 79, pp. 108-120.

Luus, C. A. & Wells, G. L. (1994). The malleability of eyewitness confidence: Co-witness and perseverance effect. Journal of Applied Psychology, 79, pp. 714-724.

Malpass, R. S. & Devine, P. G. (1981). Eyewitness identification: Lineup instructions in the absence of the offender. Journal of Applied Psychology, 66, pp. 482-489.

Ornstein, P. A., Ceci, S. J., & Loftus, E. F., (2000). Adult recollections of childhood abuse: Cognitive and developmental perspectives. Psychology, Public Policy, and Law, 4 (4), pp. 1025-1051.

Poole, D. & White, L. (1991). Effects of question repetition on the eyewitness testimony of children and adults. Developmental Psychology, 27, pp. 975-986.

Wells, G. L. (1993). What do we know about eyewitness identification? American Psychologist; 48, 553-571.

Wells, G. L., Small, M., Penrod, S., Malpass, R. S., Fulero, S. M., and Brimacomb, C. A. E., (1998). Eyewitness identification procedures: Recommendations for lineups and photo-spreads. Law and Human Behavior, (22) 6, pp. 603-647.

U. S. Department of Justice Technical Working Group of Eyewitness Evidence (1999). Eyewitness evidence: A guide for law enforcement. N.C. J. 178240. http://www.ojp.usdoj.gov.

Back to homepage