Deception Detection Accuracy

Deception detection accuracy concerns the ability of instruments or humans to distinguish truthful from deceptive communication. Accuracy is often described in terms of signal detection theory (hits, the percentage of actual deceptive utterances that are accurately classified as deceit; misses, the percentage of actual deception that is erroneously misclassified as truthful; false alarms, the percentage of truthful cases erroneously identified as deceptive; and true negatives, the percentage of truthful cases correctly identified as truthful). Whereas the polygraph achieves accuracy near 90% for specific-incident diagnostic examinations, and other instruments return a mix of accuracies, human detection hovers near chance. Multiple meta-analyses estimate that humans accurately classify truth 61% of the time and deception 47% of the time, for an overall average of 54% detection accuracy. False alarms are less frequent than misses due to a robust truth bias. Disputed is whether professionals are more accurate than novices or merely more confident and lie-biased. Training appears to confer some advantage, but its effect depends on such factors as feedback, repeated sessions, computer aids, and training in telltale content, language, or nonverbal behaviors. Factors moderating detection accuracy include base rate, deceiver communication skills, availability of context, extent of exposure, and questioning strategies.