Testing Foreign Policy Belief Structures of the American Public in the Post-Cold War Period: Gross Validations from Two National Surveys

Article excerpt

The basic attitudinal structure underlying the American public's foreign policy preferences is assessed by using both exploratory and confirmatory factor analyses on two major post-cold war surveys containing many similarly worded questions-by the Chicago Council on Foreign Relations (CC, 10/94) and the Times Mirror Center (TM, 9/93). Although previous studies had stressed only two or three primary attitudinal factors and usually ignored factor intercorrelations, our exploratory analyses of these data sets consistently yielded at least four distinct and readily interpretable factors, including two correlated "outward-focused" factors (Global Altruism and U.S. Global Interest), a U.S. Domestic issues factor bearing on foreign policy (e.g., jobs protection), and a Military Security factor. Building upon these results, confirmatory factor analyses using LISREL found that a four-factor model provided probabilistically close fits to both the CC and TM data sets and that accuracy of fits declined with various simpler models.

The American public's views on foreign affairs have been examined extensively during the past two decades. Some studies have attempted to find basic themes or dimensions representing the variety of opinions expressed on different issues. (For recent examples, see Hinckley 1992; Rielly 1995; Times Mirror Center 1993; and Wittkopf 1990.) In contrast to the lack of consistency in attitudes found by Converse (1964), more recent studies have demonstrated that Americans' diverse foreign policy preferences and goals are structured by a relatively small number of distinct attitudinal dimensions (e.g., see Hurwitz and Peffley 1987).

Most previous researchers agree that the public's foreign policy beliefs are multidimensional and cannot be represented adequately by a single "internationalist-isolationist" continuum. However, no consensus exists on the optimum number or nature of the basic attitudinal dimensions. Some researchers have relied on a two-dimensional approach which usually contrasts a militarism dimension with an international involvement dimension (e.g., Wittkopf 1990). Others have documented the need for three or more dimensions to describe adequately the public's foreign policy goals (e.g., Chittick, Billingsley, and Travis 1995).

Differences in identifying useful third and fourth factors presumably stem from differences in sets of questions analyzed as well as from the particular analytic method and interpretation used by the investigators. Most studies of the structure of American attitudes toward foreign affairs have used factor analysis on data obtained from one or more of the quadrennial foreign policy surveys sponsored by the Chicago Council on Foreign Relations (CCFR) between 1974-94. For example, Bardes and Oldendick (1980) analyzed questions from the 1974 and 1978 CCFR surveys-using principal components analysis with orthogonal (varimax) rotation-and found distinct factors representing MILITARISM (including items on military spending and defending our allies' security), a largely altruistic INTERNATIONALISM dimension (e.g., combating world hunger, defending human rights), and AMERICANISM (e.g., protecting jobs of U.S. workers). Howell and Richman (1984) analyzed the 1982 CCFR survey and-using similar analytical techniques-found factors similar to those identified by Bardes and Oldendick, including MILITARY SECURITY, LIBERAL INTERNATIONALISM, and U.S. ECONOMIC SECURITY. Building on the work of Eugene Wittkopf and others, Hurwitz and Peffley (1987) used a confirmatory factor analytic approach to establish three dimensions-MILITARISM, ISOLATIONISM, and ANTICOMMUNISM. Although Hurwitz and Peffley's research used the appropriate analytical approach, their sample involved only a single medium-sized city rather than a representative national sample of the general public.

Chittick, Billingsley, and Travis (1995) have provided the most comprehensive assessment of the relationships of individual questions on different factors generated on national samples. …