This study examined the notion that selection of semantic differential scales by reference to the table of factor loading of C. E. Osgood, et al., may be of doubtful validity if concept-scale interactions produce wide variations in factor structure between concepts. Factor analyses were carried out with respect to the measurement of prisoners' attitudes for each of 12 concepts and for the concepts as a whole. The overall analysis yielded a cognitive-evaluative factor, an affective-evaluative factor, and a 3rd factor in which the 2 activity scales featured strongly. That the 1st 2 factors appeared in many of the individual concept analyses was interpreted as providing some justification for the use of the overall analysis as a basis for the selection of scales. However, in order to construct the most sensitive measures of attitudes to particular concepts, selection of scales from the individual concept analyses here reported is recommended.
British Journal of Social and Clinical Psychology,
Vol. 12, no. 1 (Feb 1973), pp. 73-77