Attitudes, Ideology, and the Factor Model

Abstract

Recent discussions of the ideological basis of attitudes utilize both ordinary and confirmatory factor analysis as analytical representations of ideology. We examine the theoretical implications of employing this modeling style, and conclude that it imposes severe limitations on how ideology can be conceptualized and that it can lead to unreasonable claims about the sources of attitudes. The counting of as the dimensionality of ideology is an exercise doomed to triviality by its own basic logic. Error variables may have substantial item spans and may embody numerous higher-order factors. Additional complications arise because there is no consensus on whether to count factors whose effects bypass intervening levels, or whether directed effects between the conceptualfactors alter the rulesfor counting. Discussions of ideology are appropriate only after one has controlledfor the contribution that mundane causes make to consistencies among the attitudinal items. Structural equation modeling, as opposed to mere confirmatory factor analysis, has much to offer in terms of refocusing these issues and respecifying currently loose conceptualizations of the ideological basis of attitudes.