RESUMO
OBJECTIVE: To explore the accuracy rate of the comparison question test (CQT) by Binomial Distribution way on field cases in Chengdu area of China. In detail, the study revealed the accuracy rate of detecting guilty and innocent examinees, and the rate of False Positive and False Negative. METHOD: The study was built on the field cases with only two examinees that held the opposite opinions to the same single case. The original evaluators and four experienced polygraph specialists independently chose and scored 148 field criminal cases into the final study, among which 111 paired-records were concluded as indicative records and the rest, 37 ones, were ruled as non-indicative. RESULT: Judging from all records, the accuracy rate of detecting guilty examinees was 0.836 with a 0.164 False Negative rate. In addition, the rate to detect innocents was 0.822 with a 0.178 False Positive rate. Judging from indicative cases only excluding the non-indicative, the accuracy rate rose to 0.958 for perceiving guilty examinees with a 0.042 False Negative rate. To innocents, the accuracy rate was 0.859 with a 0.141 False Positive rate. CONCLUSION: The study with non-standard methodology was not limited to mock cases and the final confessions. It was able to estimate validity of other lie-detection ways, GKT, POT, etc., with this method being applied. With high accuracy rate and validity, polygraph examinations were able to give better decisions to real-life investigations.