U.S. flag

An official website of the United States government, Department of Justice.

NCJRS Virtual Library

The Virtual Library houses over 235,000 criminal justice resources, including all known OJP works.
Click here to search the NCJRS Virtual Library

Scoring Cutoffs -- Picking the Best

NCJ Number
202091
Journal
Polygraph Volume: 32 Issue: 2 Dated: 2003 Pages: 86-96
Author(s)
James R. Wygant
Date Published
2003
Length
11 pages
Annotation
This study applied various polygraph "cutoff" rules (minimum scoring levels that permit conclusions) to 200 polygraph examination scores, based on the results of 2 examiners independently scoring the same 100 verified examinations.
Abstract
Error rate, rate of definite results, false positive results, and false negative results were compared for 13 different "cutoff" decision rules, including several rules developed specifically for this study. All of the rules were applied to the same 200 examination scores to determine whether any 1 rule achieved significantly better results for the 2 examiners than any of the other rules. The rule used by most examiners is the +/-6, under which any total score of +6 or higher indicates truthfulness, and any total score of -6 or lower indicates deception. Other "cutoff" rules used in the study were +/-5, +/-4, full DoDPI (uses +/-6 as minimum total scores but adds some important exceptions), +/-6 or any -3, Senter A, Senter B, +/-6 or any -4, +/-4 or any -3, -6/+4, -6/+4 or any -3, -7/+5, and -5/+6 or any -4. The findings suggest that asymmetric cutting scores, such as -7/+5, may achieve better results than the more traditional +/-6. Under asymmetric cutting scores, a specified total minus score or lower would indicate deception, but a specified plus score or higher would indicate truthfulness. The author advises, however, that among any group of examiners, no matter how large, individual differences in scoring are likely to cause "cutoff" rules to shift up and down any scale of errors and definite results. 6 tables, 1 figure, 3 references, and a 12-item bibliography