The Effects of Expert Conclusions and Additional Forensic Evidence on Listener Evaluations in a Voice Comparison Task

Little is known about how lay people understand and evaluate forensic evidence. Yet in many jurisdictions, lay people make up juries that are responsible for determining the innocence or guilt of the accused. In this paper, we examine how lay people's evaluations of voice comparison evidence are affected by likelihood ratio (LR)-based conclusions of different magnitudes, expressed in the form of numerical and verbal statements, as well as the suggestion of the existence of additional forensic evidence in a mock case. We use a novel “jury-of-the-future” game to elicit participant judgments about whether pairs of voices belong to the same or different speakers. We then assess how those judgments are affected by expert conclusions and additional forensic evidence. Results show that verbal LRs shifted participants’ responses in a consistent way relative to the strength of the conclusion presented (i.e., people thought the voices were more likely to be the same if the LR supported the prosecution). This effect was stronger for voices from the North East of England rather than Standard Southern British English (sometimes referred to as ‘Received Pronunciation’). Less consistent patterns were found for numerical LRs, suggesting that participants struggled to interpret numerical values, particularly those of the highest magnitude. When primed with the suggestion of additional forensic evidence (DNA, footprint, fingerprint), participant responses shifted toward the voices being more likely to be the same, for both same- and different-speaker pairs, with the effect being strongest for footprint and fingerprint evidence. Here we discuss the implications of these findings for the presentation of conclusions in forensic voice comparison cases, and for forensic science more generally.