A study was designed to evaluate the use of summated rating (Likert) scales of agreement, evaluation, and frequency. The subjects, 58 female and 45 male college students, rank ordered the descriptive adjectives for the areas of agreement, evaluation, and frequency on a scale of 1 to 100. They also completed the Personal Report of Communication Apprehension (PRCA) and the Rosenberg self-esteem measure. The analysis of the subjects' ratings of the three sets of descriptive adjectives supported earlier research by P. E. Spector, indicating that the assumptions of equal interval scaling were untenable. Thus, the differences between the "a priori" method of selecting standard scale response values and the actual scales used by each subject produce important differences in the interpretation of social science research. Another finding was that different personality types used the response categories of agreement, evaluation, and frequency in significantly different ways. Subjects with low self-esteem and high apprehension were in less agreement about the meaning of the terms than the subjects with high self-esteem and confidence. The findings also suggested that the center regions of the scales were more stable than the ends, and that the negative ends of the scales might be more unstable than the positive ends. (RL)