Abstract
One methodological concern which exists with regard to the use of semantic differential measures is whether or not they meet the assumption of true bipolarity; if they do not, the use of interval-scale statistical analyses may distort data-analytic results and produce artifactual findings. This paper presents two tests examining the psychological midpoints of the 18 bipolar adjective pairs that make up the most current version of Fiedler's Least Preferred Coworker (LPC) instrument. In the weaker test, 131 subjects assessed scale midpoints by bisecting 18 five-inch lines which were anchored at their ends by Fiedler's adjective pairs. The results indicated that there were nonsignificant differences between the true scale midpoints and the midpoints identified by the subjects. In the stronger test, the adjectives were presented in three sets and the same 131 subjects ranked them based on the favorableness of each as an interpersonal descriptor. Using paired comparison treatment of the rank data (under Thurstone's Case III), the investigators found that six of the LPC adjective pairs deviated significantly from true bipolarity. Implications for future LPC use and for future research are considered.
