Validity of the positive control polygraph test using the field practice model

Article Abstract:

The validity of polygraph testing remains problematic because of reliance upon unverifiable criteria in field studies and lack of external validity in laboratory studies. This study introduces a model of field polygraph testing that provides examinee choice and commitment in a laboratory setting. Using this model, we tested 38 subjects with one repetition of a four-item Positive Control Test, two repetitions of a four-item Control Question Test, and one repetition of a three-item Guilty Knowledge Test. The Positive Control Test, a new and previously unvalidated test format, obtained average accuracy of 73 percent for the examiner and 78 percent for a "blind" judge of the polygraph record. Similar average accuracy was obtained with the other two tests, but the Positive Control Test was less biased against detection of truth than the Control Question Test and less biased against detection of deception than the Guilty Knowledge Test. Further results indicate that incentive to avoid detection may increase detectability of the deceptive, that a combination of Positive Control and Control Question Tests may be more accurate than either alone, and that number of repetitions of questions may be more important than "chart minutes" in increasing detectability with the Control Question Test. (Reprinted by permission of the publisher.)

Methods

---

Human versus computerized evaluations of polygraph data in a laboratory setting

Article Abstract:

Computer algorithms that process physiological reactions to polygraph test questions and assess the probability that the questions were answered truthfully were evaluated with data obtained in two mock crime experiments. One half of the subjects in each experiment were guilty of committing a mock theft, and one half were innocent. Data from 100 subjects in one experiment (standardization sample) were used to develop a discriminant function of electrodermal, cardiovascular, and respiration measures. The distributions of discriminant scores were used to derive Bayesian assessments of the probability of truthfulness. Data from 48 subjects in another experiment were used to cross validate the computer model (validation sample). Dichotomous computer classifications of subjects in the standardization sample were 93 percent correct. Blind numerical evaluations of the same data by an expert interpreter were 89 percent correct. On cross-validation, computer outcomes were 94 percent correct, and numerical evaluations were 92 percent correct. There were no significant differences between computer and human evaluations. The findings suggest that computer techniques may be developed for applied settings and would perform at least as well as expert human interpreters. (Reprinted by permission of the publisher.)

---

Detection of guilty knowledge with event-related potentials

Article Abstract:

The N400 component of the event-related potential (ERP) is elicited by words that complete sentences falsely. The utility of the N400 in discriminating subjects who have knowledge about a crime from those who do not was examined in this study. Subjects viewed a videotape of either an enacted burglary (guilty condition) or scenes from the city of New York (innocent condition). They then read crime-relevant phrases that had true or false completions but were not required to make any overt response as to the statements' truth. Post hoc analyses indicated that 78% of subjects could be correctly classified as guilty or innocent. Applying the same classifier to two other samples resulted in a correct classification rate of 73.2% With further development, ERPs may become useful auxiliaries to current lie-detection techniques. (Reprinted by permission of the publisher.)

Testing, Truthfulness and falsehood, Guilt (Law), Evoked potentials (Electrophysiology), Evoked potentials

Similar abstracts:

• Abstracts: Validity of the control question polygraph test: the problem of sampling bias. Psychopathy, threat, and polygraph test accuracy
• Abstracts: 1997-98 sexuality education controversies in the United States. Fewer debates about sexuality education as abstinence-only programs take foothold
• Abstracts: Alcohol control laws and the consumption of distilled spirits and beer. Primitive aspects of consumption in modern American society
• Abstracts: Consumer skepticism of advertising claims: testing hypotheses from economics of information. Examining the target of receiver elaborations: rhetorical question effects on source processing and persuasion
• Abstracts: The moderating impact of sex on the equity-satisfaction relationship: a field study. If at first you don't succeed, try, try again: effects of persistence-performance contingencies, ego involvement, and self-esteem on task persistence

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.