When the chips are down : attribution in the context of computer failure and repair.
Cognitive attribution theories provide convincing and empirically robust models of attribution. However, critiques include the scarcity of empirical research in naturalistic settings and the failure of cognitive attribution theorists to account for why, when and how much people engage in attributional activity. The present study draws data from naturalistic recordings of the common experience of computer failure and repair. A simple content analysis explores the extent to which everyday attributional talk is modelled by the cognitive theories of attribution. It is found that everyday talk matches the cognitive theories of attribution reasonably well for socially safe operative information about the problem, but poorly for socially unsafe inspective information about the agents and their actions. The second part of the analysis makes sense of this empirical pattern by using conversation and discourse analysis to explore the social functions of observed attributional talk. Participants use attributional talk to achieve two broad social goals: to negotiate and manage the social engagement and to construct and defend positions of competence and expertise.