Verba et al., 1967). Many of these studies have been genuinely insightful in their clarification of the patterns of opinion on certain issues. But despite some noteworthy exceptions (Gaining, 1964;Wright, 1972; and Mueller, 1973), few have been theoretically engaging as well; rather, the predominant tendency has been more toward detailed descriptions of attitude configurations than toward building and testing theoretical models of the linkage between personal characteristics and foreign policy opinions.A major exception to this prevailing tendency towards descriptive narrative rather than theory development can be found in Gamson and Modiglianf s (1966) work on models of the relationship between Abstract This paper tests Gamson and Modiglianfs enlightenment, mainstream, and cognitive consistency models of the linkage between knowledge and foreign policy opinions. Data from a survey of Lexington, Kentucky residents shortly after the seizure of the American embassy in Teheran, Iran in late 1979 indicate that although the mainstream model performs somewhat better than its two competitors, none of the Gamson-Modigliani models adequately explains public opinion concerning American policy options. Downloaded from admitted to the United Nations, whether the U.S. should trade with the Soviet Union, and how much military strength the U.S. should have. 2 The belief systems were measured in terms of three ordinal categories relating to how serious people considered the worldwide Communist threat. at Carleton University on June 23, 2015 http://poq.oxfordjournals.org/ Downloaded from ' Again, collapsing was done in order to maintain comparability with the Gamson-Modigliani study, and to avoid having unreasonably small numbers of respondents in any category. Again, too, the percentages exclude respondents from interviews which terminated before one or more of these questions were asked. at Carleton University on June 23, 2015 http://poq.oxfordjournals.org/ Downloaded from * The first and third options were not asked about after it was announced that the Shah had departed for Panama. Approval of each of the six options as a percentage of all respondents who expressed an opinion is summarized in Table I. 7 Refusal and "don't know" responses were omitted from the analysis, thus allowing for the formation of the approve-disapprove dichotomies.• Had Somers' d, an alternative ordinal coefficient, been selected the results would have been virtually identical to those reported here.