CONSTRUCTION OF A FAKING DETECTOR SCALE FOR A BIODATA SURVEY INSTRUMENT

This article describes the development and empirical validation of the Faking Detector scale, a component of the Leadership Effectiveness As sessment Profile (LEAP). Based on an ''unlikely virtues'' model, an in itial pool of 30 items was developed and, after review, formal pretest ing, and piloting, was reduced to 12 items. The final scale was then a dministered as a part of the LEAP instrument to a total of 425 Reserve Officer Training Corps (ROTC) students at five different ROTC summer encampments. A confirmatory factor analysis using the Linear Structura l Relations (LISREL) technique revealed a unidimensional scale account ing for 59% of the total test variance. One month later, the scale was readministered to the ROTC students. Test-retest reliability was calc ulated at r = .65. The covariance structures of the two tests were com pared to see if they were essentially the same. Although the model of equivalent structures was rejected, chi(2)(107, N = 401) = 165.37, p < .01, the adjusted goodness-of-fit index and root-mean-square residual were well within the acceptable range. Finally, correlational analyse s revealed that the Faking Detector scale was uncorrelated with two cr iteria (a supervisor's rating and peer ratings) thought Eo be associat ed with honesty. The potential utility of the scale is discussed, and recommendations for further study are delineated.