Chapter 3. Methodology
II. Linking discourses and practices: interpretative process tracing
1. Interpretative process tracing
i. Interpretative process tracing versus positivist process tracing
Process tracing is a method that has become increasingly popular in political science following a relatively recent focus on ‘process’ and ‘causal mechanisms’ to explain social phenomena as an alternative to statistical analyses, correlations and formal theorising. Process tracing enables the researcher to take into account both the institutional and social structures as well as the agency of actors (Bennett and Checkel 2011, 1). The objective is to proceed to an in-depth examination of the different elements coming together to form an event or phenomenon following a process or ‘social mechanism’ that the researcher tries to uncover. The guidelines to proceed to process tracing in the literature point to looking at qualitative data such as press articles, expert surveys, interviews, official documents, etc. to trace the interactions and links between different variables leading to a particular outcome (Checkel 2008, 115). Bennett and Checkel (2011, 8) define the data as ‘evidence from within the temporal, spatial, or topical domain defined as a case. This can include a great deal of evidence on contextual or background factors that influence how we measure and interpret the variables within a case.’ They also advise for a mix of induction and deduction (Bennett and Checkel 2011, 22).
The process tracing employed in this research shares most of these guidelines to the extent that it looks at qualitative data with the aim to trace the link between discourses and practices. The data is therefore gathered from the context (historical, geographical, etc.) – for instance, through the discourse analysis – as well as from the direct context of the social practices studied. My method of proceeding is both deductive and inductive, or abductive as promoted by Pouliot (2007): going back and forth from the hypotheses to the empirical data. While starting from several non-
constraining hypotheses on the regionalisation process and the actors involved – together with an analytical model based on discourses and practices – the empirical data guided most of my analysis, which led to a constant revision and refining of my hypotheses. According to Wagenaar (2011, 244) this is what characterises interpretative work: ‘the key heuristic moment consists of making sense of raw empirical data. This process of sense making always consists of entering into a dialogue between the preconceptions we bring to the study and the empirical data we have collected.’ The heuristic strategy of confronting our theories with our data is what helps us create ‘the conditions for novelty, for surprise’ in our research that distinguishes good interpretive work (Wagenaar 2011, 243).
However, interpretative process tracing differentiates itself from positivist process tracing in that, instead of hypothesising ‘causal mechanisms’, it looks at constitutive mechanisms. Indeed, there is a consensus in the mostly positivist literature on process tracing that it should be used to trace the causality between independent and dependent variables: ‘we understand causal mechanism to mean the intervening processes through which causes exert their effects’ (Goertz and Mahoney 2010, 24). In a more elaborate definition Bennet and Checkel (2011, 17) explain it ‘as the analysis of evidence on processes, sequences, and conjunctures of events within a case for the purposes of either developing or testing hypotheses about causal mechanisms that might explain the case.’ They recognise, nevertheless, that process tracing could be used in interpretative work even in the case where agents and structures are considered as so inherently mutually constitutive ‘that it is not possible to separate events into discrete moves in which either the agent or the structure is primarily driving the process’ (Bennet and Checkel 2011, 17). Hence, while it is impossible to find out which variable is the cause or consequence of another, it is
possible to break down events and steps, using a bracketing strategy, between the effect of structures on agents and the way agents respond to it (Bennet and Checkel 2011, 17).
This is precisely the method used in this research. My argument is constitutive rather than causal inasmuch the structures of meaning within language constitute – open the possibility – for discursive and social practices but do not deterministically cause them. Moreover, the practices also (re)produce these discursive structures, which makes these two processes inseparable. A bracketing strategy is thus used to be able to analyse this interplay: first with the analysis of the discourses categorised into ‘basic discourses’ as was detailed in the first section of this chapter; then with the tracing of these discourse(s) to social practices. I look at how social practices relate to the dominant or contesting discourse, why they were established in a particular context, and their meaning for the wider process of regionalisation. The aim is also to understand how the (discursive) structures interact with the agency of the actors adapting to their context and how this may feed back and transform the discourse. It is heuristically useful to separate the discursive structures from the practices to understand the effect of this interplay on the regionalisation process. This method gives importance to agency to be able to conceptualise change; but is at the same time strong on the structural context which is not the case of positivist process tracing; a weakness which is acknowledged by its main proponents (Checkel 2008, 116; Bennet and Checkel 2011, 29).
However, this interpretative process tracing cannot be assessed in terms of verifiability or reliability which are standards used for positivist process tracing. Indeed, ‘constitutive mechanisms’ cannot be tested against an objective, mind- independent reality. As mentioned, the analytical framework of this dissertation is
monist, it does not recognise ‘causal mechanisms’ as an ontological reality, separated from the mind, which can be discovered by the researcher, as positivist process- tracers put it (Bennet and Checkel 2011, 14). Consequently, the recommendation to test one’s analysis against other theories to assess which one explains better this reality is fruitless here (Bennet and Checkel 2011, 23). The validity of this analysis resides in a clear and structured methodology and enough empirical evidence to support the arguments and conclusions. It has to be coherent and strong enough to convince the readers of its empirical value and shed light on the regionalisation process. Against criticisms from positivist researchers qualifying interpretative work as a soft or unsystematic way of doing research, Wagenaar (2011, 251) aptly responds that interpretative methods are ‘systematic, methodical, empirically driven activities that, when done well, set up conditions for a generative, critical confrontation of theory and the empirical world.’
Interpretative process tracing also stands aside from positivist process tracing works to the extent that it refutes the possibility of finding a mix of variables that deterministically cause the same outcome – one of the main weaknesses of the IR literature on regional projects as shown in chapter one. The variables used in this research – the discourses and practices – are not objective variables; they are heuristic tools and too context and meaning-dependent to produce the same outcome in a variety of cases. Geertz (1973, 26) summarises this clearly about interpretative work: ‘the essential task of theory building (…) is not to codify abstract regularities but to make thick description possible, not to generalize across cases but to generalize within them.’ Generalisability is thus not possible in this type of work even though structural (but unstable) features can shape similarities across cases.
ii. The case studies
Two case studies were chosen to study the social practices of regionalisation: the first case study concerns the practice of regional intervention, which can take various forms including political action, electoral monitoring, humanitarian or military intervention in a state of the region or outside of the region. The second case study concerns the regional management of transnational security issues (drug trafficking, organised criminality, terrorism, piracy, etc.). These two types of regional practices are telling cases about the regionalisation process, and community and identity building inasmuch as they raise very sensitive issues for the states in respect of borders, sovereignty, non-interference, etc. The establishment of these regional practices depends greatly on the articulation of the nation and the region by the respective states’ elites. Hence, the evolution of these practices provides important insights into the political project for the region carried by the relevant actors. These case studies show how discourses frame social practices, how they are put-in-use and how the main representations are confronted by ‘events’, ‘facts’, or practices that can trigger change by making apparent the existing tensions in the dominant discourse.
It should be noted that pragmatic choices had to be made concerning the level of analysis chosen in this research which was constrained by the time frame of the fieldwork. It would have been ideal to interview a much greater number of actors involved in the practices of regionalisation of foreign and security policies to grasp all the micro-practices of these actors that give daily meaning to the process. However, this possibility was limited in this research which aims at getting a better understanding of the regionalisation process through a focus on the macro level of social structures. Both levels should be analysed in-depth but this would require much more extensive fieldwork. The solution to this problem was therefore a pragmatic
one: first, delimiting only two case studies of social practices; secondly, attempting to interview key actors within the main institutions to make sense of these practices and understand how they relate to more structural representations. The decision to choose only two case studies of social practices was motivated by Geertz’s (1973, 23) warning of the specificness and circumstantiality of ethnographic research:
‘It is with the kind of material produced by long-term, mainly (though not exclusively) qualitative, highly participative, and almost obsessively fine- comb field study in confined contexts that the mega-concepts with contemporary social science is afflicted – legitimacy, modernization, integration, conflict, charisma, structure,…meaning – can be given the sort of sensible actuality that makes it possible to think not only realistically and concretely about them, but, what is more important, creatively and imaginatively with them.’
It is indeed through the detailed study of specific social practices that the meaning and path of regionalisation in South America and West Africa can better be grasped. However, in any case, ‘Cultural analysis is intrinsically incomplete. And worse than that, the more deeply it goes the less complete it is’ (Geertz 1973, 29).