DoDPI Banner for Research Papers


JANNIRO, M. J., and CESTARO, V. L. Effectiveness of detection of deception examinations using the computer voice stress analyzer. November 1996, Report No. DoDPI96-R-0005. Department of Defense Polygraph Institute, Ft. McClellan, AL 36205.

The accuracy of the Computer Voice Stress Analyzer (CVSA) instrument and associated processes for the detection of deception was assessed using a mock theft scenario. One hundred nine subjects were randomly assigned to two groups and given detection of deception examinations using a CVSA instrument. Subjects in one group were programmed deceptive and participated in taking $100 from a metal box located in a scenario room. The non-deceptive group did not participate in the scenarios nor did they have knowledge of the mock theft. Four trained and certified CVSA examiners conducted the examinations using a CVSA technique called the Modified Zone of Comparison test. Test chart evaluators, who had not taken part in the study and who were blind to subject programming, obtained an overall accuracy of 49.8% (z = -.05, p = .96). Administering examiners correctly identified 53 of the 109 (48.6%) subjects as either deceptive or non-deceptive (z = -.21, p = .84). More deceptive subjects were correctly identified by examiners than non-deceptive subjects (32 of 55 vs. 21 of 54). However, decisions were not significantly different from chance in either case.

Key-words: accuracy, computer voice stress analyzer, voice stress analysis, detection of deception

Director's Foreword

It has long been proposed that changes in human voice characteristics are indicative of deception. The current study is the third in a series of studies completed at the Department of Defense Polygraph Institute to test the latest in a line of instruments and procedures purported to identify deceptive responses through the analysis of verbal responses.

Essentially, the question is whether or not the Computer Voice Stress Analyzer (CVSA) and associated procedures can accurately categorize truthful and deceptive examinees. To answer the question, the authors designed a mock larceny paradigm, and provided it to authorities in the CVSA field for their comments and approval. After receiving approval, the authors acquired the services of four expert examiners identified and selected by proponents of the CVSA. These CVSA experts then conducted and analyzed all tests included in this report.

The results of the study are clear and unambiguous. Police chiefs and attorneys should consider these results and their ramifications when deciding what methods of detection of deception they will employ.

Michael H. Capps