Sensitivity, Specificity, and Positive Predictive Value as Measures of Prediction Accuracy: The Case of the Eat-26

Abstract

Recently, investigators have identified several levels of severity among people suffering from eating disorders. One of the more commonly used scales for screening people with eating disorders is the 26-item version of the Eating Attitudes Test (EAT-26). Although decisions such as assigning individuals to treatment and control groups are often determined by the score on the EAT-26, researchers have not examined the sensitivity, specificity, and positive predictive value of the scale. In the present study, 231 female soldiers from a sample of 1,112 were interviewed and studied so as to determine the accuracy of decisions based on the EAT-26. Findings showed that the scale does poorly in identifying the most severe cases of eating disorders, but is better in identifying the milder case. The implications of the findings are discussed.