Abstract
“Sensitivity,” (percentage correct detections of the number treated) is the most often used index of psychological tests or test battery adequacy, although other factors (stability, reliability, other forms of validity, usability, etc.) are also reported as positive features. This paper reports on a method to improve “specificity” [ratio of persons correctly identified as not treated to total number not treated]. Specificity issues are a great concern, where behavioral testing may be used for managerial or regulatory decisions about workers. In such cases the percentage of workers with appreciable dosages may be < 10% and false positive percentages thereby become important. In two alcohol experiments, we empirically validated different multiple cut-offs seeking good false positive rates. In two additional alcohol studies we cross validated our optimum findings and found false positive rates of 3.6% (96.4% specificity) can be achieved with a combination of 6% decrement on three of four tests while retaining adequate sensitivity.
Get full access to this article
View all access options for this article.