This chapter reviews the Advocacy Coalition Framework (ACF), analysing its account of how ideas transform policy advocacy, sometimes leading to coalescence of opposed forces. It argues that the ACF is a largely descriptive account of belief formation and how those beliefs input into policy change. It examines the hypotheses generated from the ACF and uses the inversion strategy to suggest that most are relatively trivial yet there is a paucity of empirical confirmation. The ACF largely produces proximate descriptions of policy change through historical examination of token cases. This type of informed, qualitative and detailed historical analysis is vital to our understanding of policy change.
This chapter examines whether and in what sense a qualitative case study can test a theory. The author identifies three types of theory: theory as an invariant generalization; theory as an empirical generalization; and theory as a mechanism. Theory is about types. A case study by its very nature is the study of a token. A type is constituted of token examples. A token can be a member of many different types. So this chapter is concerned about how questions about token items contribute to our understanding of questions about types. The author argues that whilst case studies can contribute to tests about invariant and empirical generalizations they can never decisively test such theories. This is so for both logical and pragmatic reasons. Case studies can also contribute to testing theories as understood as mechanisms. Generally speaking, however, case studies test whether or not a particular mechanism applies to this token case rather than whether the mechanism works as it is supposed to. This difference between the two ways in which mechanisms can be tested is often under-appreciated perhaps because many historical events are thought to be unique. The author discusses the sense in which uniqueness matters here and argues that even for unique events, the actual case is a still a token example of the type. The actual reason that some outcome occurred in a token historical case might be unlikely for the type as a whole. Social scientists tend to be more interested in types, historians in tokens. Social scientists tend to be more interested in likely outcomes, historians in unlikely ones. In part this comes about because social scientists and historians are interested in descriptions of events at different levels of granularity (or detail). The type–token distinction is a key distinction for social science explanation that is under-appreciated in the discipline.