A latent structure model for answering true-false questionnaire items is shown to predict a circular ordering among some MMPI scales that were used in a recent study by Wiggins and Lovell (1965). A circular ordering can be demonstrated, but certain systematic departures from the predicted effects also occur. The data suggest the operation of three latent traits: self-descriptive accuracy, favorable bias, and a tendency to endorse items that are neutral in the sense of being unrelated to the other two traits. Alternative interpretations of the results are discussed together with suggestions for further research.
