Skip to main content

Confirmation bias for degraded forensic audio evidence

Resource type
Thesis type
(Thesis) Ph.D.
Date created
As recording devices such as cell phones and body cams become more accessible, audio recordings are increasingly being used as evidence in legal cases. The goal of this research was to test the (1) applied circumstances under which people misinterpret audio recordings; (2) mechanisms that contribute to confirmation bias in these cases; and (3) methods for mitigating confirmation bias for degraded audio recordings. In Study 1, I found that incriminating contextual information biased participants' interpretations of innocuous degraded audio recordings. Specifically, participants who learned that they were listening to wiretapped conversations with criminal suspects or that they were listening to wiretapped conversations with criminal suspects in cases also involving eyewitness evidence made more incriminating misinterpretations than participants who learned no context about the recordings. In Study 2, I tested whether fluency misattribution is a mechanism for confirmation bias by including reaction time as an independent measure of fluency. I manipulated perceptual fluency within-subject by presenting non-degraded recordings, minimally degraded recordings, and moderately degraded recordings. I manipulated conceptual fluency between-subject by varying the amount of context participants received about the recordings: (1) no context; (2) the recordings came from criminal suspects' conversations (criminal suspect); (3) incriminating written transcripts. Across degradation levels, participants in the incriminating written transcripts condition made more incriminating misinterpretations than participants in the criminal suspect condition, who in turn made more incriminating misinterpretations than participants in the no context condition. Additionally, reaction time partially mediated the effect of context on incriminating misinterpretations for minimally and moderately degraded recordings, suggesting that fluency misattribution contributes to confirmation bias. In Study 3, I tested whether the Canadian Model Jury Instructions for audio recordings and written transcripts effectively reduce participants' tendency to make incriminating misinterpretations after reading incriminating transcripts. When there was no delay between participants' exposure to incriminating transcripts and when they interpreted the recordings, participants who received instructions made fewer incriminating misinterpretations than participants who did not receive instructions. However, when there was a one-week delay, there were no differences between those who received instructions and those who did not. These studies have methodological, theoretical, and applied implications.
148 pages.
Copyright statement
Copyright is held by the author(s).
This thesis may be printed or downloaded for non-commercial research and scholarly purposes.
Supervisor or Senior Supervisor
Thesis advisor: Connolly, Deborah
Member of collection
Download file Size
etd22171.pdf 1.35 MB

Views & downloads - as of June 2023

Views: 91
Downloads: 1