John T. Wixted and Gary L. Wells published this article in Mar. 2017 finding that eyewitness confidence and accuracy are strongly related in pristine lineup conditions. Where lineup conditions are non-pristine, accuracy of even a high confidence suspect ID is seriously compromised. Pristine line-up procedures include using only one suspect per lineup; the suspect should not stand out in the lineup; a caution should be given that the offender might not be in the lineup; testing should be double-blind; and a confidence statement should be taken at the time of the identification. Full text of the article can be downloaded.