Detection of Faking on the Tennessee Self-Concept Scale

Abstract

The Tennessee Self Concept Scale (TSCS) was administered to a subject of 352 university students under standard administration conditions. Four other subsets responded under conditions designed to induce moderate and maximum faking in both positive and negative directions. Members of a sixth subset ("Sleepers") responded irrelevantly to from 10% to 50% of the TSCS items. Multiple discriminant analysis of the faked responses, using Column and Self-Criticism scores as predictors, supported a two-dimensional interpretation of the TSCS (self-concept and self-criticism). Analysis of classification errors showed the greatest confusion resulted from classification of Fake-Good respondents into the Standard group. Rasch analysis of sleeper responses was relatively insensitive to implausible response subsets. A second study (n = 413) supported conclusions of the first, and exploratory analyses suggested the possibility that a "Faking Score" may be developed using a set of specific item responses rather than subscale scores.