ABSTRACT
The history of the mainstream of attribution theory and research since Heider has been dominated by covariation and other types of regularity information. Most of the major theories of causal attribution in psychology have involved regularity information of some kind (Cheng and Novick 1990; Einhorn and Hogarth 1986; Forsterling 1989; Heider 1958; Hewstone and Jaspars 1987; Hilton and Slugoski 1986; Jones and Davis 1965-specifically the idea of non-common effects; Kelley 1967, 1972a, 1972b, 1973; Shanks and Dickinson 1987), and large numbers of often-cited studies have looked at the relation between covariation information of some sort and causal attribution (see, for example, Hewstone 1989; Ross and Fletcher 1985; Fincham 1983; White 1988a). So great has this dominance been that Ross and Fletcher (1985) commented that covariation is ‘the sine qua non of causal inference’ (p. 83), in an argument that factors like salience and resemblance depend upon the basic cue of covariation, as far as their role in causal attribution is concerned. The point of this statement is to emphasise, not just that people like to use covariation or regularity information for causal attribution, but that it is basic, essential, to their causal attribution processes.
