A re‐analysis that supports our main results: A reply to Levine et al.

Levine et al. (2018) criticized our meta‐analysis, but their conclusion was the same as ours: The cognitive approach to lie detection results in a modest improvement. We address and dismiss Levine et al.'s (2018) three criticisms. Regarding the ‘confound’, in our meta‐analysis we averaged the results of two cells on statistical grounds, which does not constitute a confound in statistical terms. Regarding ‘aberrant controls’, that depends entirely on the benchmarks selected and type of statistical test and meta‐analysis used. Regarding ‘unreliable data’, the claim that there is a positive relationship between ‘unreliable’ data and total accuracy in the cognitive lie detection conditions is not even supported by their own data (p = .16). We conclude with a request to Levine et al. to focus on our shared aim: to develop interview protocols that enable lie detection.     

Detecting children's false allegations and recantations of a crime

This study examined adults’ abilities to detect the veracity of children's (ages 6–11) initial disclosures and their later recantations about a crime. Children (N?=?32) were asked to make a false denial or a false accusation of an alleged theft, while some were asked to tell the truth. Afterwards, children recanted their initial statements in a second interview; thus, children who initially denied the theft accused a researcher of the transgression in the second interview, and vice-versa. Adult raters (ages 18–25; N?=?108) watched both interviews and completed a questionnaire that required them to determine the veracity (i.e. whether the report was true or false) and credibility of the disclosure. Adults accurately detected the veracity of children's reports 53% of the time (55% of original reports, 50% of recantations). Raters were more accurate when detecting false denials than false accusations in the children's original and recanted reports. Despite being more difficult to detect, children's recanted denials that became accusations were rated as the least credible. Furthermore, self-reported level of experience with children and ratings confidence were not significant predictors of truth/lie detection accuracy.     

When the long road is the shortcut: a comparison between two coding methods for content-based lie-detection tools

When using a content-based lie detection tool, a decision regarding veracity is made by evaluating the presence of specific content criteria within the interviewee’s account. This evaluation can be achieved either by counting the frequency of occurrence of the criteria (frequency counts; FC) or by using a scale to rate the intensity at which they appear in the entire text (scale rates; SR). In the current study, we compared these two coding methods with respect to their accuracy in determining veracity, and their intercoder and test–retest reliabilities. Fourteen coders coded the presence of perceptual and contextual details in true and false statements, each used the FC method for one set of 30 statements and the SR method for another set of 30 statements. One month later, eight of the coders recoded 28 statements. Results showed a significant advantage for the FC method over the SR method. While the coders perceived the SR method as less time-consuming than the FC method, accuracy level as well as intercoder and test–retest reliabilities were higher for the FC than for the SR method. These findings suggest that when using a content-based lie detection tool, FC coding should be preferred over SR coding.     

Effect of two types of control questions and two question formats on the outcomes of polygraph examinations

Abstract: Two major variations of polygraph “Control Question” testing, the Zone Comparison (ZoC) and the Modified General Question Test (MGQT) were evaluated. Within each, the type of control question, Exclusive or “time bar” (e.g., “Before you were 21, did you ever...”) and Nonexclusive or “no time bar”(e.g., “Did you ever….?”) was manipulated in a mock theft scenario, with 80 male and 40 female subjects randomly assigned to be either innocent or guilty. Polygraphic data collected by experienced field examiners were numerically scored by an evaluator blind to all aspects of the study. Decision accuracy was not related to the type of procedure (ZoC/MGQT) used or the subject’s sex. Accuracy was significantly related to the type of control question [χ²(2) = 11.46, p = 0.003; τc = 0.29]. Nonexclusive control questions produced greater accuracy than Exclusive control questions on both innocent and guilty subjects. These results and subjects’ self‐reports support the general “theory” on which control question (CQ) testing is based. The need for better empirical support of accepted dogma and current field practices is strongly indicated by these findings.     

‘Gross inefficiency and criminal negligence’: the Services Reconnaissance Department in Timor in 1943–45 and the Darwin war crimes trials in 1946

The post-World War II Australian military war crimes trials of Japanese from 1945–51 have been criticised for using a rule of evidence considerably relaxed from the ordinary requirements of a criminal trial, one that did not require witnesses to give evidence in person. Circumstantial evidence suggests that, in relation to a trial held in Darwin in March 1946 for war crimes committed in Timor, the secretive Special Operations Australia, otherwise known as the Services Reconnaissance Department (SRD), took advantage of the rule. This article argues that the SRD did not allow their members to give evidence in person in an attempt to control and limit the dissemination of information about their operational and security failures in Timor from 1943–45. The SRD operation was adjudged by its own official historian as displaying ‘gross inefficiency and criminal negligence’. While the SRD’s failures were known to select personnel at the time, access restrictions to archival records in the post-war period, including the war crimes trials, meant that the extent of its failures and how it appeared to manage knowledge of them has not been widely known.     

Mechanisms underlying response bias in deception detection

Abstract

The present research aims to explore the mechanisms underlying response bias in detection of deception. In addition to examining the predictive value of generalized communicative suspicion (GCS), age, and professional experience, the present approach also investigates the role of error weighting by testing the hypothesis that a greater concern about Misses is associated with a lie bias. In Study 1, we analyzed samples of (a) students, (b) police trainees, and (c) police officers. Results revealed an asymmetrical error weighting as the strongest predictor of response bias. Supporting our hypothesis, participants who were more concerned about False Alarms were particularly truth biased, whereas a lie bias was observed among those who were more concerned about Misses. In Study 2, we manipulated the error weighting in order to test its causal relation to response bias. Results again show the predictive value of asymmetrical error weighting for response bias in deception detection, and indicate that the effects of the manipulation are moderated by individuals' habitual error weighting.     

To act truthfully: Nonverbal behaviour and strategies during a police interrogation

Abstract

In an experiment, nonverbal indicators of deception in police interrogations of mock crimes were examined. Both vocal and nonvocal behaviours were scored. Thirty participants were subjected to long interrogations (over 9 minutes) conducted by 30 experienced police officers, asked to interrogate as they normally do. Although the liars reported being significantly more nervous, and found the task more strenuous than the truth tellers, no differences in the nonverbal behaviours scored were found. In an analysis of the strategies employed, both truth tellers and liars were found to try to not make excess movements. The principal verbal content strategy for the liars was to keep the story simple, and for truth tellers to keep the story real. The reasons why, in this demanding situation, the truth tellers and liars could not be distinguished by their nonverbal behaviour are discussed.