Sentence examples for measures of eyewitness from inspiring English sources

Fig. 4 Bayesian measures of eyewitness identification decisions.

More appropriate measures of eyewitness identification accuracy can be obtained from receiver operating characteristic curves (ROCs) based on confidence ratings.

Bayesian measures of eyewitness identification decisions have also been proposed (Wells & Lindsay, 1980) and are argued to offer quite specific quantitative information about the probability that a suspect is guilty.

In the third section, we evaluate Bayesian measures of eyewitness performance and show that they reflect a complex mixture of true discriminability of guilty from innocent suspects, response bias, and the probability that a guilty suspect is presented to the witness.

For this reason, diagnosticity is not an appropriate measure of eyewitness identification accuracy (nor for any other task of which we are aware; see Swets, 1986a).

One side of the argument (e.g., Wells, Yang, & Smalarz, 2015a) is that the probative measure that has been used for decades, namely the ratio of correct identifications of guilty suspects to false identifications of innocent suspects (i.e., hit rate/false alarm rate = H/F; Wells & Lindsay, 1980, Equation 6), provides the best measure of eyewitness accuracy.

As a consequence, his simulations seem to suggest that the AUC fails to provide a good measure of eyewitnesses' ability to discriminate guilty from innocent suspects.

The best measure of eyewitnesses' ability to discriminate guilty from innocent suspects is one that does not conflate decision accuracy and response bias.

How should the accuracy of eyewitness identification decisions be measured, so that best practices for identification can be determined?

Receiver operating characteristic (ROC) analysis has long been used in applied fields to measure discriminability, but it was only recently introduced to the field of eyewitness identification.

The criticisms of ROC-based interpretations of eyewitness identifications are in many cases simply wrong: The area under the curve really does measure discrimination accuracy (d′) and does not provide information that is redundant with diagnosticity.