The Ability of Police Investigators to Detect False Confessions
Fitzgerald, Kristine, Canadian Journal of Police and Security Services
Kassin, S. M., Meissner, C. A., & Norwick, R. J. (2005). "I'd Know a False Confession if I Saw One": A Comparative Study of College Students and Police Investigators, Law and Human Behavior, 29(2), 211-227.
When police investigators interrogate a suspect, one of the primary goals is to obtain a confession (Joseph, 1995; Kassin, 1997 as cited in Russano, Meissner, Narchet & Kassin, 2005). Consequently, it is vital for police investigators to elicit true confessions when interrogating suspects. However, in some cases, false confessions can be evoked from a suspect. Kassin and Wrightsman (1985) suggest that there are three types of false confessions: (1) voluntary false confessions that are offered in the absence of pressure from police investigators, (2) coerced-compliant false confessions that arise due to coercion from police and a perception of immediate gratification, and (3) coerced-internalized false confessions that are elicited when a suspect actually accepts that they committed the crime in question.
In recent years, media reports have drawn attention to several cases in both Canada and the United States in which a defendant has allegedly given a false confession. Today, when it is suspected that there might be a coerced confession, a judge must be able to determine its admissibility during a pretrial hearing (Lassiter, Geers, Handley, Weiland & Munhall, 2002). Since confession evidence has the potential to be highly influential in court cases, the ability to distinguish between true and false confessions has become a crucial aspect of a police investigator's job. This is especially important given recent research by Kassin and Fong (1999) indicating that police investigators who are trained to detect deception cues are actually less accurate, yet more confident, when distinguishing between truthful and deceptive statements than the average person. In addition, Meissner and Kassin (2002) determined that police investigators were more likely than the average person to believe true confessions to be false, indicating an investigator bias toward determining guilt.
To further substantiate this research Kassin, Meissner, and Norwick (2005) tested the trained police investigator's assumption that "I'd know a false confession if I saw one." This two-phase experiment was conducted to address three major goals. The first goal was to compare the judgment accuracy and confidence of untrained observers to that of police investigators. Second, Kassin et al. (2005) hoped to determine the origin of the investigator response bias and finally, a third goal was to examine if the judging of confessions is influenced by the medium of their presentation (i.e., audiotape or videotape).
Researchers began by recording, both on audiotape and videotape, true and false confessions provided by a group of 17 male prison inmates from a Massachusetts state correctional facility. These confessions were then presented to a sample of 118 observers, 57 federal, state, and local police investigators (47 males and 10 females) and 61 male and female college students, who judged each as either a true or false confession. Each observer also rated their level of confidence in their judgment on a 1-10 point Likert scale (with 1 representing "not at all confident" and 10 representing "very confident"). Both police and college participants were assigned to small groups and randomly assigned to observe the confessions either on audio or videotape.
Consistent with previous research, Kassin et al. (2005) found that overall students were more accurate in their observations than police investigators. Specifically, students who observed confessions in the audiotape condition were the most accurate group while police investigators in the videotape condition were the least accurate in their judgments. The results also indicate that, while students were more accurate in their confession judgments, police investigators were more confident (levels of confidence were not affected by the medium of presentation). …