The Equal-Interval Nature of Semantic Differential Scales: An Empirical Investigation Using Fiedler'S Least Preferred Coworker (Lpc) Scale And Magnitude Estimation and Case Iii Scaling Procedures

Abstract

An assumption made in using semantic differential measures is that the adjective pairs used are truly bipolar; otherwise, statistical analyses may yield distorted findings. This article examines the 18 bipolar adjective pairs used in Fiedler's Least Preferred Coworker (LPC) instrument by using two scaling techniques and a sample of 113 respondents. To replicate and extend previous research, the 36 LPC adjectives were ranked according to their relative interpersonal favorability. The same sample also scaled the 36 adjectives using Stevens's magnitude estimation technique (MET). Both pair-comparison treatment of complete ranks scaling (under Thurstone's Case III) and the magnitude estimation procedure indicated that true bipolarity did not characterize most of the paired adjectives. Implications for future LPC and semantic differential research are discussed.