The first fully randomized field experiment of police interrogations, investigates the impact of video recording

Saul M. Kassin, John Jay College of Criminal Justice of the City University of New York
Melissa B. Russano, Roger Williams University
Aria D. Amrom and Johanna Hellgren, John Jay College of Criminal Justice of the City University of New York
Jeff Kukucka, Towson University
Victoria Z. Lawson, Institute for State and Local Governance of the City University of New York

In partnership with a small city police department, we randomly informed or did not inform 122 crime suspects that their interrogations were being video-recorded. Coding of all sessions indicated that camera-informed suspects spoke as often and as much as did those who were not informed; they were as likely to waive Miranda at the outset and later; they were as likely to make admissions and confessions, not just denials; and they were perceived no differently by detectives on a range of dimensions. Looking at distal outcomes, we observed no differences in ultimate case dispositions. In terms of policy and practice, results did not support the hypothesis that recording—even when transparent, as required in 2-party consent states—inhibits suspects or alters case dispositions. At least for now, this conclusion is empirically limited to situations in which cameras are concealed and to interrogations that do not involve juveniles, homicides, or drug crimes, which we a priori excluded from our sample.

police interviews, interrogations, confessions

“In recent years, numerous wrongful conviction cases in which false confessions were a contributing factor and research on the causes and consequences of false confessions have inspired calls for reform. Some proposals seek ways to protect vulnerable suspects such as juveniles and people with intellectual or emotional impairments; others call for the reform of police interrogation practices, banning the use of certain confrontational tactics in favor of investigative interviewing. Perhaps the most significant proposed safeguard is to require the electronic recording of interrogations—the entire process, not just the confession. As stated in a American Psychology-Law Society white paper: ‘Without equivocation, our most essential recommendation is to lift the veil of secrecy from the interrogation process in favor of the principle of transparency’” (p. 46).

“At present, 25 states and the District of Columbia require the full recording of custodial interrogations; 25 states do not. This split betrays a history of debate regarding what constitutes good policy and practice. On the one hand, various professional organizations have favored mandatory recording and surveys of individual law enforcement investigators have yielded supportive results. For example, Sullivan, Vail, and Anderson (2008) interviewed police from hundreds of departments that recorded custodial interrogations and found that they cited numerous benefits of recording (e.g., it allows detectives to attend to the suspect without concurrent note-taking; it allows them to review the sessions for any incriminating comments initially unnoticed; it reduces the need for detectives to defend their interrogation practices in court). Yet others have opposed recording either on pragmatic grounds (e.g., citing the scope of such a requirement; financial costs; the evidentiary consequences of a failure to comply; and issues of consent, especially in two-party consent states) or out of concern for how recording will alter the behavior of both police and suspects during interrogation and the subsequent decision-making of judges and juries” (p. 46)

“To test the hypothesis that recording will alter a suspect’s behavior and decision-making, we observed real suspects caught up in real investigations. With cooperation from a small city police department in the U.S. Northeast, a unique opportunity presented itself. Consistent with statewide practice, this department records all suspect interrogations involving capital felony cases (i.e., those in which the suspect faces the possibility of life in prison; as a matter of departmental practice, many but not all other suspect interviews are also recorded). Operating within a one-party consent state, this department has the option by law to inform or not inform suspects of this practice. This combination—a police department that records, can do so without a suspect’s consent, and was interested in partnering on the present study—enabled the first fully randomized field experiment of police interrogations” (p. 46).

“With a total of 122 suspects brought into custody for questioning, we randomly assigned some but not others to be informed that their session was to be recorded by a concealed camera. We sought to guard against experimenter bias, a form of reactivity in which experimenters’ expectations can unwittingly influence the behavior of the participants with whom they interact. Hence, our experimental protocol called for the lead detective to be blinded regarding condition” (p. 47).

“All recordings were then made available for transcription and coding, allowing us to compare the two groups on a number of objective behaviors—including the frequency with which suspects gazed in the direction of the camera (a behavioral measure of self-consciousness) as well as indicators of possible inhibition such as the number of suspects who invoked and/or waived their Miranda rights, the average total duration of interrogation, the length—in time and word count—of suspects’ responses to questions, the frequency with which suspects made full or partial admissions and confessions, and the nature and level of detail in their statements. After each session, the lead detective rated the suspect on a number of relevant dimensions (i.e., forthcoming, cooperative, truthful, inhibited, anxious, talkative, relaxed, and self-conscious). After all data were collected, we tracked case dispositions for both groups” (p. 47).

“Results in general did not support the claim that recording would have inhibitory effects. To begin with, suspects exhibited little awareness or concern about the presence of a camera. Not a single suspect refused or expressed a reluctance to proceed because of being recorded. This simplest of results is important in light of the contrasting fact that 22 of 25 mandatory recording statutes across the United States contain a recording exception in the event that a suspect refuses to speak if recorded. Nor did we observe significant differences in other extraneous behaviors, such as whether suspects talked to themselves when left alone or cried during their interrogation. From a historical perspective, the finding that no suspects refused to proceed on camera resembles early experiences with the videotaping of confessions” (p. 52).

“Consistent with the actual behavior of suspects, detectives’ perceptions of their suspects were also not influenced by the camera condition. After each session, the primary detective rated the suspect on a brief questionnaire. On average, they saw the suspects as generally talkative, cooperative, and forthcoming, even while anxious; as moderately truthful and relaxed; and as not particularly inhibited. No condition differences were obtained on any of these measures. On the most direct question of all concerning the effect of the camera, detectives did not rate suspects in either group as seeming self-conscious about the possibility of being recorded. At 1.80 on a 10-point scale, this rating was the lowest of all ratings in both conditions” (p. 53).

“Turning from proximal to distal outcomes, we tracked case dispositions 14 months after all data were collected. In what is arguably the most important data point, we found no differences in the extent to which the cases of camera-informed and -uninformed suspects were resolved—and of those that were resolved, no differences in the extent to which suspects were convicted” (p. 53).

“At present, 25 states in the United States mandate the recording of interrogations, at least for serious crimes. Among those that do, 22 contain exceptions in the event that a suspect refuses to speak if recorded. Some of these states have one-party consent laws that enable police to record interrogations covertly. Others are two party states that require the suspect’s knowledge and consent (some make exceptions for custodial police interrogations). Among states that do not have a recording requirement, a concern often expressed is that the real or imagined presence of a camera will unduly inhibit suspects, causing them to “clam up,” become tense, and refuse to speak, thereby obstructing law enforcement efforts to investigate crimes” (p. 54).

“It is interesting that the same argument was made by early opponents of Miranda v. Arizona (1966). President Nixon at that time commented that Miranda constituted a victory for the forces of crime. Yet after 50 years, especially in light of high waiver rates, most researchers have concluded that these fears have proved unfounded— that, if anything, Miranda has functioned as a ‘safe harbor’ for police” (p. 54).

“Regarding the recording of interrogations, Sullivan and colleagues interviewed investigators in police and sheriff’s departments and found that these fears seemed unfounded in their own experience. By observing and coding the behavior, decision-making, and case dispositions of suspects who were randomly informed or not informed, the present study supports these prior self-report results. In terms of policy and practice, these results should allay fears in two-party consent states that are reluctant to mandate recording and offer guidance to one-party states that can inform or not at their discretion” (p. 54).

“Although no significant differences were found on admissions of guilt or confessions, it is worth noting that—in response to a hypothesis suggested by detectives—subsequent coding and analyses indicated that although few suspects openly expressed a reluctance to implicate someone else during their interrogations, camera-informed suspects were somewhat less likely to actually incriminate another person. It is important to caution that this tendency was not significant. Still, we think this hypothesis deserves additional testing, perhaps in a larger sample that contains more homicide investigations and drug crimes, which may involve gang-related activities that invite the possibility of implicating others” (p. 53).

As always, please join the discussion below if you have thoughts or comments to add!