The effect of the proportion of mismatching trials and task orientation on the confidence-accuracy relationship in unfamiliar face matching
Files
(Accepted version)
Date
2017
Authors
Stephens, R.
Semmler, C.
Sauer, J.
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Journal article
Citation
Journal of Experimental Psychology: Applied, 2017; 23(3):336-353
Statement of Responsibility
Rachel G. Stephens, Carolyn Semmler, James D. Sauer
Conference Name
Abstract
Unfamiliar, one-to-one face matching has been shown to be error-prone. However, it is unknown whether there is a strong relationship between confidence and accuracy in this task. If there is, then confidence could be used as an indicator of accuracy in real-world face matching settings such as border security, where the objectively correct decision is typically unknown. Two experiments examined the overall confidence-accuracy relationship, as well as the relationship for positive (match) and negative (mismatch) decisions. Furthermore, they tested whether these relationships were affected by factors relevant to applied face matching settings: the proportion of mismatching trials (PMT), and the task orientation of the decision-maker (look for matches, or look for mismatches). Both calibration analyses and signal detection methods were applied to assess performance. The results showed that confidence can have a high correspondence with accuracy overall, regardless of task orientation but with small effects of PMT. Thus, confidence is promising as an indicator of accuracy in face matching. However, PMT systematicallyproduces large detrimental effects on the confidence-accuracy relationships for positive and negative decisions, when considered separately. Signal detection measures help with understanding these effects and proposing future research directions for improving the relationships.
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
© 2017 American Psychological Association