NCJ Number
240331
Journal
Polygraph Volume: 41 Issue: 3 Dated: 2012 Pages: 156-169
Date Published
2012
Length
14 pages
Annotation
This study used computer scoring and comparisons between the same relevant and control questions as performed by human scorers.
Abstract
The Army and Air Force variants of the Modified General Question Test (MGQT) and Federal Zone Comparison Test (ZCT) have been widely used for investigations in private, governmental, and police work. The Army MGQT has had the reputation as being a deceptive test. Previous review of field testing has shown the Army MGQT test protocol may be one of the causes for this problem, a problem not seen with the Federal Zone when total scores are used to classify examination as deceptive or non-deceptive. When positive spot scores are required to classify ZCT examination as non-deceptive the false positive rate again increases. This study uses computer scoring and comparisons between the same relevant and control questions as performed by human scorers. The frequency of negative scores at each relevant question was compared to ground truth known deceptive and non-deceptive examinees, and there was a notable deceptive bias. The analysis reveals serious difficulties for the non-deceptive examinee with traditional Army MGQT. The Air Force MGQT suffered the same problem to a much lesser extent. Additionally, when comparison questions preceded relevant questions in test design and when the more tracing comparisons were available per test decision, results improved. This appears to support the observed balance and strength of the Federal Zone and very similar Utah formats. (Published Abstract)