Thresholds for Attributing Can Affect Factorial Structure

Abstract

Subjects rated four target persons, following a different set of instructions for each. The several instructions specified different criteria for endorsing items. Items were derived for each pole of the set used by Norman (1963). The frequently replicated structure from his items was approximated best under the Definitely Apply instruction for a group of subjects from which those with extreme endorsement frequencies had been eliminated. Since endorsement frequencies and factorial structures varied with instructions, this study provides additional evidence that personality measurements and their relationships are a function of the conditions under which such measurements are obtained.