Perspectives on Parsimony: How Long Is the Coast of England? A Reply to Park and MacInnis; Schwarz; Petty; and Lynch

Abstract

Park and MacInnis (2006, in this issue) provide the always sound advice to examine the boundaries of the focal construct, in this case attitudes. They wonder whether attitudes are capable of predicting differences in levels of personal investment and commitment to resulting behavior, and they question whether attitude dimensions such as strength and extremity, or additional dimensions and moderators, are rich enough to account for either the range of affect intensity reflected in the degree of emotional attachment or in behavioral commitment. They draw our attention to consumers’ emotional attachment to brands (which they see as inherently linked to one’s self-schema) and argue for a sharper delineation of the boundaries of the attitude construct. Their analysis suggests that emotional attachment is a separate relationship-based construct. Several reactions seem in order. First, we do not suggest that attitude theory can explain emotional attachment and its consequences (see also Cohen [1990], Cohen and Areni [1991], and Pham et al. [2001] for consistent attempts to separate attitudinal and affective processes). In creating the Multiple Pathway Anchoring and Adjustment (MPAA) model, we draw on relevant cognitive and motivational theory and research to better understand both attitude formation and the conditions under which attitudes are likely to guide behavior. Attitudes arise out of a range of human experience extending from direct experience and exposure to internal*Cohen and Reed’s (2006, in this issue) MPAA model describes the attitude formation and reliance process. We thank the notable contributors for their insightful commentaries about our model and attitude theory and research more generally. Since we cannot respond to all the nuances outlined in each of the four separate articles, we primarily address some important differences in the commentators’ views about parsimony. The spirit of our response is intended to convince readers that when it comes to representing psychological processes, there is no one right model. Rather, the construct mapping one uses and related judgments about parsimony should depend on explicit research goals. Joel B. Cohen is Distinguished