ABSTRACT
This chapter argues that a warning against potentially distorted and misleading causal inference, empirical generalisations, and concept-formation and classification. As they are deployed in social-scientific research, retroactive redescriptions and functional concepts can be rather blunt instruments. The inferences, explicit or implicit, that a retroactive redescription suggests, licenses or compels should be valid inferences. The use of abstract functional concepts also reflects and supports a common assumption about retroactive redescription that underlies much orthodox social-scientific procedure: the assumption that past phenomena, originally designated by older concepts, can unproblematically be subsumed under newer concepts. Abstract functional concepts are therefore naturally elastic and omnivorous: they tend to swallow up all manner of motley empirical entities, and are purposely devised so as to cut through contextual diversity. Designed to apply to a broad variety of such contexts, functional concepts are deliberately constructed so as to focus on a very select few attributes, themselves abstractly specified.
