What does "ideology" mean? As a preliminary step to answering this muchasked question, I collected what seemed to be the most thoughtful and/or influential definitions circulating within the social sciences in the postwar decades. 1 A quick perusal of these definitions reveals the extent to which ideology remains a highly flexible conceptual tool (see Table 1). One is struck not only by the cumulative number of different attributes that writers find essential, but by their more than occasional contradictions. To some, ideology is dogmatic, while to others it carries connotations of political sophistication; to some it refers to dominant modes of thought, and to others it refers primarily to those most alienated by the status quo (e.g., revolutionary movements and parties). To some it is based in the concrete interests of a social class, while to others it is characterized by an absence of economic self-interest. One could continue, but the point is already apparent: not only is ideology farflung, it also encompasses a good many definitional traits which are directly at odds with one another.
Indeed, it has become customary to begin any discussion of ideology with some observation concerning its semantic promiscuity.2 Few concepts in the social science lexicon have occasioned so much discussion, so much disagreement, and so much selfconscious discussion of the disagreement, as "ideology." Condemned time and again for its semantic excesses, for its bulbous unclarity, the concept of ideology remains, against all odds, a central term of social science discourse.
How, then, are we to understand this semantic confusion, and how are we to deal with it? Five common approaches can be identified among writers in the social sciences: operationalization, terminological reshuffling, intellectual history, etiology and multivocality. In the following section, I outline each of these endeavors and demonstrate their limitations. I then proceed to a new approach which comprehensively maps the meanings of ideology onto a single, reasonably concise, semantic grid. I conclude with a brief discussion of "core" meanings for ideology, and a plea for context-dependent methods of definition.
Among those who study "behavior" in American politics, discussion of ideology has centered on a single empirical question: how ideological is the mass public (compared, that is, with political elites)? There have been a good many twists and turns in this debate since it was introduced by Campbell et al. (1960), McClosky et al. (1960), Converse (1964), and McClosky (1964). But the debate over the ideological proclivities of the mass public does not seem much closer to resolution today than it did in the 1960s.3 The reason for this lack of resolution has something to do with problems of data incommensurability through time and differing methods of operationalizing variables, as generally recognized.
Less often recognized are the various problems of definition inherent in the concept of ideology. Is an "ideological" mode of thought characterized by abstraction, internal consistency, external contrast, endurance through time, rationality, sophistication, a hierarchical ordering of idea-elements, parsimony-or some combination of these characteristics? Is it separate from group affiliation and/or party affiliation? Such questions, which merely scratch the surface of scholarly debate among behavioralists, are "definitional" in the sense that no answer can claim a priori precedence over another. Each definitional attribute may, of course, be operationalized in different ways, raising a second tier of disputes.
Indeed, some writers take the position that definitional tasks are contained within-and rightfully subservient to-tasks of operationalization. "It matters primarily not what you call it, but how you measure it," is the implicit approach of many behavioralists. Although there is surely much to be said for a pragmatic/ empirical approach to concept definition, this has not proven an entirely successful strategy in the instant case. …