n his article entitled Logic of Mere Exposure: A Reinterpretation . . . , Timothy B. Heath (1990) provides a useful tutorial on difference between moderation and (Baron and Kenny 1986; Brown 1989). Heath illustrates distinction as it pertains to research on mere exposure effects via reinterpretation of a study reported by Anand, Holbrook, and Stephens (1988). By means of discussion, he claims to show Anand et al.'s empirical work unwittingly supported a conclusion runs contrary to intended. Specifically, he argues study by Anand et al. is consistent with an independence hypothesis (in which may occur without cognition) than with a cognitive-affective model (in which depends on some kind of mediation). When Anand et al. (1988, p. 386) say they seek evidence for in formation of affect, Heath assumes such cognitive must involve a process wherein objective familiarity (OF) -* cognition or subjective familiarity (SF) -* (as shown, e.g., in his Table 1). Thus, drawing on his Figure 1, Heath represents Anand et al. as concluding that OF causes cognition (SF), which in turn causes affect (1990, p. 242, repeated verbatim on p. 243) and proceeds to point out this reasoning . . . suffers from two distinct (p. 242). The second of these shortcomings hinges on a claim authors confuse tests of moderation (OF by SF interaction) with those of (OF -* SF -* evaluation) (p. 243). We share concern for difference between and moderation, and we acknowledge Heath has identified an unfortunate looseness in our use of term cognitive However, though Heath's critique does distinguish correctly between moderation (as in ANOVA) and (as in path analysis), it fails to note Anand et al. used term cognitive in a more general sense to refer to processes of cognition underlie formation of affect. We now regret somewhat casual use of term mediation. However, we did not anticipate readers might construe our ANOVA approach from viewpoint of path analysis. The reason we failed to anticipate possibility is it makes no sense to suggest a straightforward ANOVA such as ours, in which some dependent variable (C) is affected by two factors (A and B) and their (A X B), can permit any sort of meaningful interpretation as a path analysis in which A --BC or A -A X BC. For reason, it never occurred to us to try to interpret our results via a path-analytic framework. Yet Heath represents our analysis as concluding OF -* SF -* affect. In actuality, we never intended any such thing. Rather, we clearly indicated the objective familiarity effect was [i.e., qualified] by a significant objective X subjective familiarity interaction (1988, p. 389). Our explicit use of term moderated (meaning qualified) should have discouraged any critique based on mediation in path-analytic sense. Indeed, we never claimed a path-analytic OF-SF-affect of type Heath describes (and could not have done so without committing gross errors in logic). If one were to insist on adopting a path-analytic view of our study (a perspective we definitely do not recommend), one would need to construct a model of following form:
Search from 250 M+ research papersSearch from 250 M+ research papers
Open Access
Dec 30, 2006
Belén Altuna 
