Judging the Credibility of Criminal Suspect Statements: Does Mode of Presentation Matter?

摘要

For a study of modality differences in deception detection accuracy, groups of graduate students judged segments selected from videotapes of criminal confessions. Twenty brief utterances were presented in four ways: content only transcript, verbatim transcript, audio only, and audio/videotape. No modality difference in unbiased truth hit rate was found, but unbiased lie hit rate varied by modality, with judges of transcripts stripped of pause indications, word repeats, and umms and uhhs less accurate than verbatim transcript judges, audio judges, and audio/video judges. The 62% overall accuracy and 61% lie detection accuracy of audio judges was highest and, in contrast to other judges, audio judgments did not display a response bias. The results remain consistent with the presence of valid visual cues but suggest that at least in some situations focus on valid vocal cues may offer more accuracy.