Development of Parallel Forms of a Generalized Scale of Attitudes toward Speakers

Abstract

Potential advantages of the synthesis of generalized attitude scales and empirically scaled attitude statements were indicated in previous research ( 1 ) . A generalized attitude scale is one that has as its attitude object a general class of attitude objects rather than a specific attitude object ( 4 ) , e.g., attitudes toward any state ar, opposed to attitudes toward Indiana, or Idaho, etc. Empirically scaled attitude statements have obvious advantages over a priori statements (2, 3, 5 ) . The purpose of this research was to ( 1 ) scale empirically attitude statements toward speakers and ( 2 ) develop rwo parallel forms of a generalized scale of artirudes toward speakers. The primary intent of the scale is as a criterion in the evaluation of effectiveness of speakers (regardless of their forum, e.g., instructors, students, debate team members, professional lectuers, faculty candidates, assessees, executives, etc.). A group of 34 junior and senior business administration students held conferences concerning the variables, the content, which ought to be covered by such an attitude scale, e.g., voice quality, apparent preparation, knowledge of subject, etc. The group produced 240 statements which were edited by the project director. The students then rated statements as to favorability on an 11-point scale, as per Thurstone. Scale and agreement values were obtained for each starement. Two parallel 22-item forms were assembled by considering ( 1 ) scale and agreement values and ( 2 ) statement content. Scale values range from 1 to 11 (median value, 6 ) and agreement values range from 0.0 to 1.5 (median value, 0.5 ) . The two forms ( A and B) were administered, counterbalanced, to an independent group of 89 junior and senior business administration students. Their task was to use the forms to evaluate a guest speaker they had just heard. Completion time of the two forms ranged from 3 to 10 min. Odd-even, corrected, reliability estimates were obtained ( A , .85; B, 3 4 ) ; and a parallel-forms reliability coefficient was obtained (.86). Adequate reliability has been demonstrated. Content validity is implicit. Meaningful measures of attirudes toward speakers may be generated inexpensively, reliably, and rapidly. This synthesis of generalized attitude scales and empirically scaled attitude statements can be used to meet other management information needs, e.g., attitudes toward political figures. The list of classes of attirude objects is limited only by the needs of the organization.