NCJ Number
191921
Journal
Polygraph Volume: 30 Issue: 3 Dated: 2001 Pages: 172-181
Date Published
2001
Length
10 pages
Annotation
This article replicates the polygraphy study by Honts and Driscoll (1987) and extends it with field data.
Abstract
The discipline of polygraphy has expended no small effort in the exploration and development of analytical methods for scoring data, with the aim of maximizing decision accuracy. Some have advocated the use of rank order scoring systems as one method of minimizing subjectivity. To date, only two ranking methods have been tested empirically with traditional polygraph recordings on single-issue examinations utilizing the Comparison Question Technique (CQT), the Rank Order Scoring System (ROSS), and the Horizontal Scoring System. Honts and Driscoll (1987) suggested that the ROSS, with its use of simpler and more objective scoring rules, should make this system easier to teach polygraph students, as well as convey to a lay public. This method is based on the ranking of individual physiological responses, and determining algebraically from those ranks which category of question, relevant or comparison, elicited more physiological arousal, thereby permitting inferences of truthfulness or deceptiveness. Honts and Driscoll tested the method with laboratory cases, and finding it very effective, urged more research on the ROSS. This study was a replication and extension of the work of Honts and Driscoll, using field cases instead of mock crime cases. Results showed that accuracy in the first cross validation, using 150 nondeceptive and 150 deceptive cases, was 80.3 perecent with inconclusives, and 91.6 percent without them. A subset of that sample, consisting of 100 cases, was used in a second cross validation where ROSS was directly compared to traditional 7-point manual scoring. Traditional scoring was 69.7 percent accurate including inconclusives, and 90.5 percent correct without them. ROSS produced an average of 75 percent correct decisions with inconclusives, and 88.2 percent without them. There were no significant differences in accuracy or rates of inconclusive results. These results suggested that the ROSS might be a valuable alternative to traditional 7-point scoring, especially in contested evidentiary cases. Because ROSS uses measured features, it holds the potential for exceptional inter-scorer reliability. References