NCJ Number
240333
Journal
Polygraph Volume: 41 Issue: 3 Dated: 2012 Pages: 186-198
Date Published
2012
Length
13 pages
Annotation
This study replicated the unweighted Directed Lie Screening Tests (DLST) decision accuracy and inconclusive rates of DLST examinations, as reported in previous studies by the U.S. Department of Defense.
Abstract
Two experienced examiners completed blind scoring tasks on 49 Directed Lie Screening Tests (DLST), also known as the Test for Espionage and Sabotage (TES), conducted by seven inexperienced examiners on eight non-naive examinees who participated in a mock espionage scenario at a forward operating base in Iraq. Seven-position scores, using the U.S. Federal test data analysis (TDA) model, were transformed to three-position and Empirical Scoring System (ESS) scores. Monte Carlo models were used to calculate the distributions of seven-position, three-position, and ESS scores, and the results were analyzed using multivariate ANOVAs. Unweighted decision accuracy and inconclusive rates using the seven-position scores did not differ significantly from previous studies of the TES at the U.S. Department of Defense. Criterion accuracy for the seven-position, three-position, and ESS TDA models was significantly greater than chance. Only the ESS model produced both test sensitivity to deception and test specificity to truth-telling that were significantly greater than chance. The three-position TDA model produced significantly more inconclusive results that were loaded on deceptive cases. The seven-position and ESS scores were found to extract similarly useful diagnostic information from the raw data. Pairwise decision agreement was significantly greater than chance for all models. Results support the criterion validity of the DLST and suggest continued interest in this technique. (Published Abstract)