Evidence of the camera perspective bias in authentic videotaped interrogations: Implications for emerging reform in the criminal justice system
Correspondence should be addressed to Dr G. Daniel Lassiter, Department of Psychology, Ohio University, Athens, OH 45701, USA (e-mail: firstname.lastname@example.org).
Objective. Numerous previous experiments have established the existence of a camera perspective bias in evaluations of videotaped interrogations/confessions: videotapes that make the suspect more visually conspicuous than the interrogator(s) by virtue of focusing the camera on the suspect yield assessments of voluntariness and judgments of guilt that are greater than those found when alternative presentation formats are used. One limitation of this body of research is that all the interrogations/confessions used to date were simulations; therefore, no evidence currently demonstrates that the camera perspective bias importantly generalizes to authentic videotapes recorded by police and depicting actual suspects and interrogators. Two experiments addressed this issue.
Methods. Experiment 1 compared judgments of voluntariness based on viewing two authentic videotaped confessions – one recorded with the camera focused on the suspect, the other with the camera focused equally on the suspect and interrogator – with those based on listening only to the audio or reading only a transcript. Experiment 2 compared judgments of voluntariness and guilt of an originally equal-focus videotaped confession that was edited to produce suspect-focus and interrogator-focus versions.
Results. In Experiment 1, participants judged the videotape version of the confession to be more voluntary than either the audio only or transcript versions, but only for the suspect-focus videotape. In Experiment 2, participants viewing the suspect-focus version of the confession (relative to the interrogator-focus version) judged it to be more voluntary and the suspect more likely to be guilty.
Conclusion. The present research further strengthens the policy implications of the literature on camera perspective bias by providing evidence that the bias manifests with authentic interrogations/confessions as well as with simulations.
Since 1989, post-conviction DNA testing in the United States has established that 200 people have spent a combined 2,475 years in prison for crimes that they did not commit, with nearly 70% of these tragic errors having been uncovered just in the last 7 years (Innocence Project, n.d.). As disturbing as these statistics are, they may sadly represent only a fraction of wrongful convictions. That is, for the vast majority of convicted individuals who maintain their innocence, there exist no biological evidence on which to conduct DNA testing to potentially substantiate their claims.
On a more encouraging note, the growing number of DNA exonerations has led legal scholars, social scientists, and government officials to agree that there is an urgent need for reforms in the criminal justice system. In addition, careful examination of the particulars of wrongful conviction cases has yielded insights regarding some of the factors responsible to one degree or another for these miscarriages of justice (e.g. Dwyer, Neufeld, & Scheck, 2000).
One such factor is false confessions, which can occur when innocent suspects succumb to the intense psychological pressure that is a ubiquitous feature of police interrogations in the United States (Kassin & Gudjonnson, 2004; Lassiter, 2004). In fact, false incriminating statements made by suspects during detention were found to be instrumental in more than a quarter of the wrongful-conviction cases on record. Even more alarming, the data reveal that when innocent individuals have been wrongly convicted for the crime of homicide, false confessions are implicated as a contributing factor in fully two-thirds of the cases (Innocence Project, n.d.).
In response to these troubling facts, various constituency groups have called for mandatory videotaping of custodial interrogations. Proponents of a videotaping requirement argue that it will reduce, if not prevent, instances of false confessions influencing trial outcomes in two important ways. First, it is claimed that the presence of the video-camera will discourage police from using highly coercive techniques during interrogations. Such techniques increase the likelihood of a false confession (Kassin & Kiechel, 1996; Leo & Ofshe, 1998; Russano, Meissner, Narchet, & Kassin, 2005), so bringing them to a halt should produce ameliorative effects. Second, the assumption has been made that a complete audio-visual record of what transpired during interrogations will permit later trial fact finders to assess more accurately the voluntariness and veracity of suspects' statements. To the extent that this is true, trial fact finders should demonstrate marked improvement in rejecting incriminating statements or admissions of guilt that are in fact coerced or untrue.
Although there remains some resistance in the law enforcement community to mandated videotaping, officers who have had experience with videotaping interrogations argue that the practice benefits police as well (cf. Sullivan, 2005). For example, supporters believe that videotaping eases society's concerns about how suspects are treated while in custody, and eliminates the need for extensive note taking so that officers can better observe suspects' non-verbal behaviour. In addition, some police point out that videotapes serve as a useful teaching tool for demonstrating appropriate and inappropriate interrogation techniques, and that subsequent viewing of videotapes can reveal incriminating information that was overlooked during the live interrogations.
Given the many potential positives just described, it is not surprising that one advocate has stated that widespread acceptance of the practice of videotaping interrogations will ‘benefit suspects, law enforcement, prosecutors, juries, trial and reviewing court judges, and the search for truth in our justice system’ (Sullivan, 2005, p. 28). Such enthusiasm notwithstanding, we think it would be rash to ignore possible downsides associated with the videotaping procedure or with the manner in which it might be specifically implemented. To the extent that such drawbacks exist and can be identified, appropriate policies regarding implementation can be developed to minimize, if not eliminate, their influence, thereby increasing the likelihood that the videotape reform will in fact reduce instances of coerced or false confessions, and ultimately result in fewer wrongful convictions.
One concern that has come to light is the possible prejudicial effect of videotapes that depict suspects' final confession statement but none of the interaction between interrogators and suspects preceding it. Two major surveys of actual police practices indicate that such ‘recap videotapes’ are not unusual (Geller, 1992; Sullivan, 2004). Kassin (1997) has noted that recap videotapes are potentially problematic for at least two reasons. First, recaps may convey to trial fact finders that the confession was more voluntary than they would otherwise perceive it to be if the interrogation in its entirety was available for them to observe. Second, recaps often are recorded after suspects have been asked to recount their stories multiple times. By the time the camera is rolling, their statements may be accompanied by little of the emotion and agitation that might have been present when they initially revealed the self-incriminating information. Recap videotapes, then, may make suspects appear far more callous and unremorseful than is in fact the case, which in-turn could bias the jury against them. Awareness of these undesirable possibilities has led most courts and legislatives bodies that have made custodial recordings compulsory to spell out clearly that the entire interrogation must be recorded – from the Miranda warning to the end of the session (cf. Sullivan, 2005).
Whereas most policy makers can readily appreciate the unfairness of not recording earlier parts of an interrogation leading up to a confession, a second issue of concern regarding the videotaping reform is far less obvious. Commonsense suggests that the camera should be focused on the suspect during questioning, as it is this person's statements, demeanour, and overall behaviour that presumably will be most informative to later trial fact finders in rendering correct assessments of the voluntariness and validity of any confession. It is understandable, then, that a suspect-focus camera perspective is in fact currently the norm in the United States with regard to videotaping police interrogations (Geller, 1992; Kassin, 1997). Psychological science, however, indicates that people's perceptions of the cause(s) of an observed other's behaviour are not simply a function of the relevant information available to them. That is, factors that may not be truly causing a person's behaviour have been shown nonetheless to be perceived as causal simply because they are more visually prominent or salient to observers than other factors (Lassiter, Geers, Munhall, Ploutz-Snyder, & Breitenbecher, 2002; McArthur, 1981; Taylor & Fiske, 1978).
This phenomenon, known as illusory causation (McArthur, 1980), suggests the possibility that videotapes of interrogations in which the suspect is more visually conspicuous than the interrogator by virtue of the camera perspective (i.e. the suspect is facing the camera, with only the back of the interrogator visible at best) could bias observers' evaluations. More specifically, the visual prominence of the suspect could lead observers to conclude that incriminating statements made by him or her are largely volitional rather than a consequence of excessive pressure being exerted by the interrogator, irrespective of what the reality of the situation might be. Basic research on illusory causation has examined it in contexts that are very different from custodial police interrogations; for example, many studies of illusory causation focused on judgments regarding which of two college students ‘set the tone’ in an observed get acquainted conversation (see Taylor & Fiske, 1978). Therefore, convincing policy makers to take the possibility of the aforementioned camera perspective bias seriously is likely to require considerable direct evidence of its existence and generalizability to actual legal contexts.
The seminal demonstration that camera point of view could influence evaluations of a videotaped confession was reported by Lassiter and Irvine (1986). Participants viewed one of three versions of a videotaped (mock) confession concerning the crime of shoplifting that differed only in terms of the camera perspective used when the confession was initially recorded. In the suspect-focus version, the front of the ‘suspect’ from the waist up and the back of the ‘detective’ – part of his head and one shoulder – were visible; in the interrogator-focus version, the front of the detective from the waist up and the back of the suspect – part of her head and one shoulder – were visible; and in the equal-focus version, the profiles of both the suspect and detective from the waist up could be seen equally well. After the presentation of the confession, participants were asked to indicate the degree to which they believed it was coerced or involuntary. The confession was judged to be the least coerced in the suspect-focus condition, more coerced in the equal-focus condition, and the most coerced in the interrogator-focus condition. Regardless of condition, participants expressed high confidence in the accuracy of their judgments.
Since its initial demonstration, the camera perspective bias has also been found to influence judgments of a suspect's likelihood of guilt and sentencing recommendations in the same linear fashion (Lassiter, Beers et al., 2002). Moreover, numerous investigations attest to the considerable robustness and generalizability of the bias. For example, individual differences in the motivation to think effortfully (Lassiter, Slaw, Briggs, & Scanlan, 1992) and in the capacity to reason specifically about complex causal relationships (Lassiter et al., 2005) do not moderate the camera perspective bias, nor is it reduced by situationally increasing observers' sense of accountability for their judgments (Lassiter, Munhall, Geers, Weiland, & Handley, 2001). The bias has also been shown to occur across various types of crime (Lassiter et al., 1992), in the context of realistic trial simulations (Lassiter, Geers, Handley, Weiland, & Munhall, 2002), and with samples of college students and community members from disparate backgrounds (Lassiter, Beers et al., 2002). Finally, relevant expertise does not mitigate the bias as it has been shown recently to affect the evaluations of highly experienced trial judges and police interrogators (Lassiter, Diamond, Schmidt, & Elek, 2007).1
THE PRESENT RESEARCH
Several scholars (e.g. Bornstein, 1999; Bray & Kerr, 1979, 1982; Diamond, 1997) have noted that research programmes in the area of legal psychology must aspire to a high standard with regard to issues of ecological validity. For example, Bornstein (1999, p. 88) has pointed out that ‘courts have not welcomed psycholegal research findings with open arms, especially when derived from methods that are neither very realistic nor representative of actual legal processes’. Bray and Kerr (1982, p. 304) suggested that a reasonable approach to addressing ecological validity ‘is to conduct a series of carefully planned studies that collectively provide data that determine the limits of generalizability’. The present investigations, like those reviewed earlier, adopted this approach, as our aim was to help move the research on the camera perspective bias in videotaped confessions even closer to the high standard of ecological validity needed to ultimately impact the legal system.
The many merits of the studies described above notwithstanding, the likelihood of the legal establishment heeding the existing scientific evidence for a camera perspective bias could be diminished by the fact that none of the experiments reviewed so far exposed participants to authentic confessions obtained during real police interrogations. That is, the prior work used mock interrogations and confessions that were designed to be composites of various elements known to occur in true interrogations, or that were constructed re-enactments developed from transcripts of specific police interrogations. These simulations were required because of the need to produce multiple camera perspectives of the same confession. Because simulated ‘confessors’ faced no serious consequences, it is possible that their behaviour diverged in important ways from that of actual confessors. As such differences could, in-turn, affect judgments, we cannot currently be certain that observers viewing authentic videotaped interrogations and confessions will also manifest the camera perspective bias.
The possibility that observers could render less biased evaluations when viewing authentic videotaped interrogations and confessions is suggested by some recent work in the area of deception detection (Mann & Vrij, 2006; Mann, Vrij, & Bull, 2004). Mann et al. argued that studies (e.g. DePaulo & Pfeifer, 1986; Meissner & Kassin, 2002) demonstrating that laypersons and even professionals (such as police and custom officers) are generally not much better than chance at judging whether a person is lying or telling the truth suffer from a potential drawback – namely, the stakes (negative consequences of being caught and the positive consequences of getting away with the lie) may not have been high enough for liars to exhibit clear cues of their deception (cf. DePaulo, Lanier, & Davis, 1983; DePaulo, Stone, & Lassiter, 1985). Mann et al. (2004), therefore, had participants view clips from videotapes of real-life police interviews in which actual criminal suspects were sometimes telling true, rather than simulated, lies.2 Their results revealed that observers were able to achieve higher rates of accuracy at differentiating deception from truth than is typically found in studies in which the lies are told in a less realistic context.
The implications of the Mann et al. (2004) research are straightforward: in real-world, high-stakes situations, a person may behave in ways that cannot easily be evoked or reproduced in mere simulations, no matter how well the latter are constructed. Moreover, such behaviours may provide vital information capable of improving observers' judgments of the person. With regard to videotaped interrogations, the availability of more potentially diagnostic cues in the behaviour of actual suspects and detectives might help observers avoid being influenced by irrelevant aspects of the videotape, including the camera perspective from which it was initially recorded.
The most direct way to test whether the camera perspective bias manifests with authentic videotaped interrogations and confessions would be to obtain copies of such videotapes. We managed to secure a few; however, an additional requirement for ideally testing the above hypothesis, unfortunately, could not be met. That is, in the ‘perfect’ experiment, participants would be presented with one of three versions of a given authentic videotaped interrogation: one in which the camera focused on the suspect; one in which the camera focused on the interrogator(s); and one in which the camera focused on both the suspect and the interrogator(s) equally. We were unable to identify any law enforcement agency that videotaped interrogations with two cameras running simultaneously, much less three. Therefore, in the two experiments to be reported, a different approach was taken to answering the question of whether the camera perspective bias occurs when authentic videotaped interrogations are evaluated.
Prior studies (Lassiter, Beers et al., 2002; Lassiter et al., 1992) have shown that audiotapes and transcripts of simulated confessions generally produce evaluations comparable to those obtained with equal-focus videotapes. Based on this pattern of results, we argue that comparing an authentic suspect-focus videotape with audio only and transcript presentations of the same interrogation and confession would constitute a reasonable test of the camera perspective bias under conditions of high ecological validity. That is, if the bias truly occurs with real interrogations and confessions, then an authentic suspect-focus videotape should produce judgments of greater voluntariness than either an audio only or a transcript presentation. Experiment 1 provided a test of this prediction.
Two authentic videotaped police interrogations were used in this experiment. One involved a case of sexual assault and was originally recorded with the camera trained on the suspect (suspect-focus confession). The other involved a murder-by-arson case and was originally recorded with the camera trained on both the suspect and the interrogator (equal-focus confession). Audio only and transcripts versions of each interrogation were derived from the videotapes.
The inclusion of an authentic equal-focus confession allows us to rule out the possibility that the predicted differences between the suspect-focus videotape and its corresponding audio only and transcript presentation formats is due simply to a general tendency to judge the confession as more voluntary when presented on videotape. If, as predicted, the original camera perspective per se is the source of the bias, then, in contrast to the authentic suspect-focus confession, no differences in voluntariness judgments as a function of presentation format (video and audio, audio only, transcript) should occur for the authentic equal-focus confession.
One hundred and three male undergraduates at Ohio University completed the experiment in return for course credit.3
The authentic suspect-focus interrogation/confession was recorded in the fall of 2002. It began at approximately 4 a.m. and lasted 25 minutes. The suspect was a Caucasian male in his late twenties who was taken into custody on suspicion of sexual assault. The camera is stationary and slightly elevated for the entirety of the videotape, so that the front of the suspect from the waist up and the upper back and head of the interrogator are visible. The interrogator, wearing a police uniform, was a Caucasian male estimated to be in his late forties or early fifties (based on his appearance when he first walked into the room and was momentarily facing the camera).
Throughout the interrogation, the suspect appears groggy and dazed, having been awakened in the middle of the night and taken to the police station. The interrogation begins with the officer advising the suspect of his Miranda rights, which the suspect waives by signing a document. The interrogator immediately charges the suspect with having non-consensual sex with a woman at a recent party. It becomes apparent that the alleged victim was someone that the suspect had known for approximately 9 years, and both were intoxicated when the sexual acts in question occurred. The suspect states more than once that he believed the woman wanted to have sex just as he did, however, the interrogator counters that the woman was too incapacitated to give her consent.
Several times the suspect indicates that he has no clear memory of the details of the event. As a means of getting the suspect to keep talking, the interrogator at one point says, ‘you're not digging yourself in a hole, you're digging yourself out of a hole.’ Later, the interrogator, still refusing to accept that the suspect cannot remember what happened states, ‘but if I get the feeling you're lying to me, I'll put it in this report – “He was lying to me. He showed no remorse for what he did. He tried to snake his way out of it” – so just tell me the truth.’ Although the interrogator never accepts the suspect's contention that he cannot remember, the interrogator does make remarks that indicate he is quite willing to accept the fact that the woman can ‘barely remember what happened’. By the end of the interrogation the suspect acquiesces and acknowledges that the officer's characterization of the events could be accurate.
The authentic equal-focus interrogation/confession was a 16-minute segment of a longer videotape that documents questioning occurring at different junctures over a 2-day period in late summer of 2002. We used only a portion of this interrogation to keep it comparable in duration to the suspect-focus interrogation.4 It should be noted that this segment was contiguous and represented rather well the tone and content of the full videotape.
The suspect in this interrogation was a middle-aged Caucasian woman and the plain-clothes detective was a Hispanic male who appeared to be thirty something. The camera is stationary throughout the segment and positioned to the side of both the suspect and interrogator, so that each is seen in profile from the waist up. The suspect is accused of setting a fire that resulted in the death of her daughter. For most of the interrogation, the suspect repeatedly denies the accusation; however, the detective never acknowledges these denials.
On several occasions the interrogator emphasizes that he is trying to help the suspect. The interrogator asserts that he knows the suspect is guilty and at multiple points provides false information that would seem to back up or give some weight to his assertion. The interrogator makes frequent use of the minimization strategy (cf. Kassin & McNall, 1991) – for example, saying, ‘I think you didn't try to do it on purpose but it happened, it got out of hand.’ The interrogator also generates two possibilities about what happened and directs the suspect to choose which is the truth (a technique recommended in the leading interrogation manuals, cf. Leo, 2004):
We've got two persons here. We've got a cold-blooded killer that did this because she wanted for some ungodly reason to kill her daughter. Or we've got a person who loves her daughter that was going into prostitution, that was going to leave the house, and was losing control … and somehow by accident this happened [the fire] and feels real bad about it now.
As with the previous interrogation, the suspect eventually acquiesces and makes comments suggesting the death of the daughter may have been her fault.
Small groups of participants (no more than six at a time) were present at an experimental session. The experimenter informed participants that their task was to assume the role of trial jurors for the purpose of helping the researchers ‘discover how people in real courtrooms make decisions about the validity of confession evidence’. Participants learned that their specific charge would be to evaluate the voluntariness of an actual confession obtained during a police interrogation. A brief presentation on the concept of voluntariness and its importance to accurately evaluating a confession was provided by the experimenter. Depending on which of the two interrogations participants were randomly assigned to evaluate, they next received a short description of either the sexual assault or murder case. Participants were then randomly presented with either the video and audio, audio only, or transcript of one of the two authentic interrogations/confessions described above.
Following the confession presentation, participants, individually and without any prior group discussion, responded (on 9-point scales) to two items pertaining to perceptions of voluntariness: one assessed their impression of the interrogation in its entirety by asking whether the suspect or the detective was most in control of, or responsible for, what transpired over the course of the interrogation (1 = Detective, 9 = Suspect). The other focused on their impression of the extent to which the suspect's self-incriminating statements specifically were the result of coercion (1 = To a large degree, 9 = Not at all). These items were combined into a single voluntariness index, with higher numbers indicating a judgment that the interrogation in general was not overly coercive and the confession in particular was given freely and intentionally.5 Following a thorough debriefing, participants received credit and were dismissed.
Results and discussion
A 2 (authentic confession: suspect focus/sexual assault vs. equal focus/murder)×3 (presentation format: visual and audio vs. audio only vs. transcript) analysis of variance (ANOVA) was performed on the voluntariness index. The ANOVA revealed a significant main effect for confession, with the sexual-assault confession being judged to be more voluntary (M=6.15) than the murder confession (M=4.81), F (1, 97)=11.80, p<.01. There was also a significant main effect of presentation format, F (2, 97)=3.87, p<.05. The video-and-audio format led participants to rate the confessions as more voluntary (M=6.25) than did the audio-only or transcript formats (Ms=5.10 and 5.09, respectively). Importantly, this effect was qualified by the predicted two-way interaction, F (1, 97)=9.11, p<.01.
Planned comparisons examining the effect of presentation format on evaluations in the suspect-focus/sexual-assault condition indicated, as anticipated, that the video-and-audio format produced significantly higher voluntariness judgments than did the audio-only format, t (97)=2.72, p<.01, and the transcript format, t (97)=2.50, p<.05 (see Table 1 for means and standard deviations). Identical comparisons examining the effect of presentation format on evaluations in the equal-focus/murder condition revealed no significant differences, ts<1. The fact that the effect of presentation format diverges for the two confessions makes it likely that camera perspective, rather than the video-and-audio format, was responsible for the relatively high assessments of voluntariness made by participants viewing the suspect-focus/sexual-assault videotape.
) on the voluntariness index as a function of confession presentation format for the two authentic confessions (Experiment 1)
The findings of Experiment 1 are the first to indicate that the camera perspective bias found previously with simulated confessions is also likely to occur when authentic police interrogations and confessions are evaluated. Although we believe our approach to ascertaining whether the camera perspective bias generalizes to authentic interrogations and confessions was a reasonable one, it is not beyond reproach. One criticism is that the crucial interaction reported in Experiment 1 between the authentic confessions (suspect focus/sexual assault vs. equal focus/murder) and presentation format (video and audio vs. audio only vs. transcript) may have resulted from content differences across the two confessions, rather than being due specifically to the varying camera perspectives.
To eliminate this potential confound, in a second experiment we transformed the authentic equal-focus confession used in Experiment 1 into two new versions. By editing the original videotape, we were able to create a version in which only the suspect was visible (suspect focus) and a version in which only the interrogator was visible (interrogator focus). The essential content in these two versions is thus identical, with only the visual conspicuousness of the suspect and interrogator differing. If, as we have argued, the interaction pattern obtained in Experiment 1 was caused by changes in camera perspective and not changes in content, then observers viewing the new suspect-focus version of the murder videotape should judge the confession to be more voluntary than those viewing the new interrogator-focus version.
Two additional limitations of the first study were addressed in Experiment 2 as well: both males and females were recruited as participants and a measure of the suspect's likelihood of guilt was included along with items assessing the voluntariness of her confession.
Twenty-six male and female introductory psychology students at Ohio University were recruited to participate in the study in exchange for partial course credit.
The equal-focus/murder interrogation that was used in the first experiment was modified to produce two new videotapes. Video-editing software was employed to split the original videotape essentially down the middle, with each half saved as new separate videotapes. In the first of these, only the suspect could be seen from the waist up and in profile (suspect-focus videotape), and in the other only the interrogator could be seen from the waist up and in profile (interrogator-focus videotape). It is important to emphasize that for both of these altered versions of the murder interrogation/confession, the audio portion was exactly the same and unchanged from the original equal-focus videotape. Thus, the actual verbal exchange between the suspect and interrogator (including intonation, volume, etc.) was identical across the two new versions of the murder videotape. Finally, we note that during the debriefing of Experiment 2, participants made no remarks that would suggest they found the framing of either of these new videotapes unusual in any way.
The procedure of Experiment 1 was employed once again with the following modifications. Participants were randomly assigned to view either the suspect-focus or interrogator-focus version of the murder videotaped interrogation/confession. Participants also responded (on 9-point scales) to two new voluntariness items, both focused on the confession rather than the interrogation more generally (‘The “suspect's confession was …” forced out of the suspect by the detective  or given freely by the suspect  and “To what degree do you believe the detective tricked the suspect into confessing”, with 1 = To a large degree, 9 = Not at all’). As in the first study, these items were combined into a single voluntariness index for purposes of analysis.6 Finally, participants were asked to indicate how likely it was that the suspect was guilty on a 9-point scale, with high numbers denoting a greater likelihood of guilt.
Results and discussion
The data were analysed in a 2 (camera perspective: suspect focus vs. interrogator focus)×2 (measure: voluntariness vs. guilt) ANOVA, with the first variable a between-subjects factor and the second a within-subjects factor.7 The analysis revealed a significant main effect of camera perspective, F (1, 24)=4.70, p<.05, with participants in the suspect-focus condition rendering assessments of voluntariness and judgments of likely guilt that were greater than those rendered by participants in the interrogator-focus condition (see Table 2 for means and standard deviations). The two-way interaction was not significant (F<1), which indicates that the effect of camera perspective on the voluntariness and guilt measures was comparable.
) on the voluntariness index and likelihood-of-guilt measure as a function of camera perspective (Experiment 2)
These results extend those of Experiment 1 by demonstrating that even when the same authentic interrogation is being considered, which person (suspect or interrogator) is more visually salient affects observers' evaluations in a manner consistent with the many prior studies that have documented the camera perspective bias in simulated videotaped confessions.
The present investigations are the first to demonstrate that the camera perspective bias found previously with simulated videotaped interrogations/confessions also emerges when authentic videotapes – recorded by police and depicting actual suspects and interrogators – are evaluated. The current research, then, contributes to the literature on camera perspective bias in videotaped confessions by moving it still closer to the high standard of ecological validity that many argue is required for psychological science to sway a typically sceptical legal community (cf. Bornstein, 1999).
Ideally, we would have randomly selected our authentic interrogation/confession stimuli from a large sample of such items, and used several different interrogations/confessions to assure maximum generalizability. As acknowledged in the Introduction and Footnote 4, however, we were able, after considerable effort, to obtain very few authentic interrogations/confessions, with only one of those exhibiting an equal-focus camera perspective. This, of course, means our findings are open to criticism on the grounds of insufficient stimulus sampling (cf. Wells & Winschitl, 1999). That is, we cannot assert that our results would hold for different authentic stimulus materials. That being said, it is difficult for us to imagine that the two authentic interrogations/confessions that we used would be the only ones that would produce the results we obtained in Experiments 1 and 2. Nevertheless, future research should strive to determine whether the camera perspective bias generalizes across interrogations that may differ in a number of different ways (e.g. whether the interrogation is inquisitorial or accusatorial in nature) or across suspects/interrogators who may offer different impressions (stereotypes) to perceivers. As authentic videotaped interrogations/confessions become more readily available to researchers, such investigations will be feasible and will more definitively establish the extent to which the camera perspective bias manifests where it really matters – in the real world.
In the early 1990s, the Police Executive Committee of New Zealand approved the videotaping of police interviews/interrogations on a national basis. In implementing this policy, various procedural guidelines were established. One critical issue that had to be decided was the vantage point of the camera. One of the authors of ‘The New Zealand Video Interview Project’ (L. W. Takitimu, personal communication, November 3 1993) noted that the seminal research on camera perspective and videotaped confessions (Lassiter & Irvine, 1986) led them to opt for showing side profiles of both the police officer and the suspect, although they knew at the time that this deviated from the primarily suspect-focus approach used in Australia, Canada, and the United Kingdom.
Follow-up surveys have indicated that judges in New Zealand feel that the policy mandating that police interrogations be videotaped from an equal-focus perspective has led to significant improvements and protections for all criminal-justice participants (L. W. Takitimu, personal communication, November 3 1993). Based on New Zealand's success and the current and previous findings on the camera perspective bias, we believe it would be prudent for the United States and the other aforementioned countries to seriously consider adopting a similar policy: one that requires ‘full disclosure’ in terms of both time – the entire interrogation is recorded – and space – the interrogator(s) and suspect are equally visible.
Funding for this research was provided by National Science Foundation grant SES-0453302. We thank Katrina Brickner for her assistance in conducting the research and Kim Lassiter and Matt Lindberg for their constructive comments on various aspects of the research and manuscript.
1 For more complete coverage and discussion of the literature on the camera perspective bias in videotaped confessions, the reader is directed to Lassiter, Geers, Munhall, Handley, and Beers (2001), Lassiter (2002), Lassiter and Geers (2004), and Lassiter, Ratcliff, Ware, and Irvin (2006).
2 Statements from reliable, independent witnesses and forensic evidence were used to establish whether any given statement of the suspects' was true or false.
3 Early in our data collection, we discovered that some females indicated that they were experiencing considerable discomfort with some of the content of the confession dealing with sexual assault. We therefore made the decision to recruit only males for Experiment 1. Experiment 2, however, included both male and female participants.
4 We attempted to locate an authentic equal-focus videotape that was of similar length to the authentic suspect-focus videotape but were unsuccessful, as the former camera perspective currently is not that common. Indeed, the interrogation we used was the only one recorded from an equal-focus camera perspective that we could obtain. This fact supports our assumption that policy makers, on their own, would not likely come to the conclusion that a suspect-focus camera perspective is problematic and should be avoided.
5 Cronbach's α for this index was only .23, which is not entirely surprising given that one item was directed at perceptions of the interrogation more globally and the other was directed at perceptions of the confession per se. We view this as a strength of the index in that it captures participants' broader impressions of the suspect's liberty to behave as he or she wished in the situation. An alternate analysis was also conducted in which the two voluntariness items were treated as separate, within-subjects factors. The results of this analysis were entirely consistent with that presented in the text.
6 Cronbach's α for this index was .76, no doubt higher than in Experiment 1 because both constituent items were concerned specifically with the confession.
7 To make them equatable, values on the voluntariness and guilt measures were transformed to standard (z) scores prior to the analysis.