Rationality and Portability: Case Study Research in Practice
Total Page:16
File Type:pdf, Size:1020Kb
Rationality and portability: Case study research in practice Bo Bengtsson & Nils Hertting What general lessons can be learned from the study of single cases? This is one of the most controversial methodological issues in the social sciences. Elsewhere, we have proposed a logic of generalization from case studies using thinly rationalistic ideal-type social mechanisms (cf. Elster 1983) as a conceptual bridge to make the ndings from one case (to some extent) portable to other contexts (Bengtsson & Hertting 2014; cf. also Bengtsson & Ruonavaara 2011; 2017). In this chapter, we recapitulate the general logic behind such `rationalistic generalizations' and apply it to one exemplary and well- known case study: Bent Flyvbjerg's thorough investigation of planning processes in the Danish city of Aalborg, as presented in his book Rationality and Power. Democracy in Practice (1998). Flyvbjerg implies some generalizing ambition when he interprets the object of his study, the so-called Aalborg Project, as `a metaphor of modern politics, modern administration and planning, and modernity itself'. It is, however, unclear what logic of generalization Flyvbjerg has in mind more precisely, and in Bengtsson & Hertting 2014, we actually suggest that it may be in line with our model of generalization based on thinly rationalistic social mechanisms. In this chapter, we attempt to translate Flyvbjerg's argument into such terms. Purpose and outline The main purpose of the chapter is to test the fruitfulness of a logic of generalization based on thinly rationalistic ideal-type social mechanisms by applying it to the analysis of a case study that is largely actor-based but where the generalizing ambitions are not explicitly based on thin rationality. In Bengtsson & Hertting 2014, we argue that our suggested logic is not very original and should not be very controversial either. In that article, we illustrate our argument with examples from our own research on resident co- operation and local network governance, but we also maintain that in practice, many social scientists base their generalizing ambitions on ideas about rationalistic social mechanisms although largely implicitly. For example, we claim that Flyvbjerg's labelling of the Aalborg Project as a `metaphorical', `prototypical' or `paradigmatic' case (Flyvbjerg 2001; 2006) can possibly be interpreted in terms of thin rationality. Thus, the attempt in this chapter to apply our model of generalization to Flyvbjerg's study can be viewed as a rst and preliminary test of our claim in Bengtsson & Hertting 2014 that our thin rationality 15 Rationality and portability: Case study research in practice logic only theoretically makes explicit what is already an implicit methodological practice in case study research. The chapter is organized in ve sections. In the rst section, we briey introduce our logic of generalization based on thinly rationalistic ideal-type social mechanisms. Then, we give a short summary of Flyvbjerg's empirical study and its conclusions. In the third section, we present and elaborate Flyvbjerg's generalizing argument. In the fourth section, we attempt to apply our logic to Flyvbjerg's case study. The nal section discusses the conclusions from our analytical experiment. Generalization by mechanism a thinly rationalistic approach John Gerring has dened a 'case study' as an intensive study of a single unit for the pur- pose of understanding a larger class of (similar) units' (Gerring 2004, 342). Nevertheless, despite the current wave of innovative work on qualitative methods, drawing more general inferences on the basis of observations from single-case studies is still viewed as problem- atic in social science research. Authors such as King, Keohane and Verba (1994, 208212) even claim that generalization without more than one case (or `observation') is impos- sible, and as a solution, they suggest increasing the number of observations. However, in process-oriented research, such as Flyvbjerg's study of Aalborg, this move is seldom possible without mutating the research question from dynamics to comparative statics. Recently, it has been suggested that portability from one empirical case to others can be achieved through analysis based on causal mechanisms with deterministic properties (e.g., Mahoney 2001; Falleti & Lynch 2009; Goertz & Mahoney 2012, chap. 15). How- ever, assumptions of determinism are often too demanding in the social sciences. Other approaches have been suggested, e.g., those related to concepts such as Sartori's 'ladder of abstraction' (Sartori 1970), Yin's `analytical generalization' (Yin 2009) and Flyvbjerg's `paradigmatic cases' the latter model is discussed further below. However, the more precise rationales behind these concepts are not always clear. Lacking a convincing logic of generalization, qualitative studies of social and political processes are, too often and too modestly, framed as `explorative' or `pilot studies' or as `plausibility probes' (Eck- stein 1975) aimed at generating hypotheses for subsequent, and allegedly more conclusive, large-n studies. Unfortunately, however, it is often tricky to design relevant large-n studies of complex societal processes. In Bengtsson & Hertting 2014, we have suggested a logic of generalization based on thinly rationalistic social mechanisms. Assuming `thin rationality' (roughly, that actors normally do things for a reason), we can derive ideal-type social mechanisms from em- pirical observations of social action and interaction in one empirical case. Based on the same assumption, we can also expect the same ideal-type mechanisms to be applicable in similar actor constellations in other contexts. Thus, thinly rationalistic mechanisms can be used as a generalizing bridge between contexts. From this perspective, generalization from single-case studies produces expectations about similar patterns of thinly rational action and interaction in similar contexts. In general terms, mechanisms are regular patterns of actions and interactions, patterns that are causally productive, which means that they bring about certain outcomes. To be 16 Rationality and portability: Case study research in practice Figure 1. Coleman's macro-micro model (From: Hedström & Swedberg 1998, 22.) generalizable, a mechanism also implies portability, that is, that the pattern discovered in one context is also relevant in others (cf. Bengtsson & Ruonavaara 2011; 2016). Jon Elster suggests that mechanisms are `frequently occurring and easily recognizable causal patterns that are triggered under generally unknown circumstances or with indeterminate consequences' (Elster 1998, 45). We largely adhere to that denition, although we ad- mit that the elements `under generally unknown circumstances` and `with indeterminate consequences' both may be contested and are not essential to the concept (Hedström & Ylikoski 2010, 56). To be portable, social mechanisms must be based on actors with intentions. We share this point of departure with Hedström & Swedberg (1998), who relate their discussion to James Coleman's 'macro-micro-macro' model of collective ac- tion (the so-called `Coleman's boat') that, in the spirit of methodological individualism, connects macro-level structural explanation to micro-level individual agency. Starting from Coleman's model, Hedström and Swedberg present a typology with three dier- ent types of mechanisms: (1) from societal macro-conditions to actors' perceptions and values ('situational mechanism'); (2) from actors' perceptions to their individual action (`action formation mechanism'), typically an individual `calculus' in a wide sense and often implicit and intuitive; and (3) from individual action to changes on the macro- level (`transformational mechanism'), e.g., via a game of collective action (Hedström & Swedberg 1998; cf. Coleman 1990, Chap. 1). The model is summarized in gure 1. Similar to Hedström and Swedberg, we view social mechanisms as micro-level links, but we go further in developing a logic of generalization that is explicitly based on thin rationality and ideal-type analysis. This perspective allows a particular form of portability and generalization in terms of reasonable expectations of similar mechanisms operating in similar contexts. The rst general idea behind our approach is an assumption of rationality in a wide 17 Rationality and portability: Case study research in practice sense, i.e., that actors typically do things for a reason. A wide notion of rationality, we argue, is an analytical element that elucidates why we gain a sense of generalized understanding when identifying mechanisms in individual cases. Making this method- ological idea more explicit in qualitative analysis also makes the logic of generalization from single-case studies more convincing. This argument is less controversial than it may seem at a rst glance. As we view it, an assumption about some type of rationality underlies most micro-level mechanisms proposed by authors in the eld. Even actor-related mechanisms that are expressed in more structural terms, e.g., power dependencies, network relations and social norms, can often be deduced from assumptions about more or less rational actors who are attempting to reach their goals in a particular context within the limits of their perceived scope of action. Thus, our approach is based on `weak' or `structural' methodological individualism, where not only individual agents but also relations and relational structures are emphasized (cf. Hedström & Ylikoski 2010, p. 60). We view