Winter holiday is time to read the books that do not fit easily into the research and teaching agenda yet promise some new insights. This year’s reading was no exception: Pierre Bourdieu’s “Esquisse pour une auto-analyse” which made me think through a number of questions about data collection for conflict analysis. This autobiographic non-autobiography is another tentative of the French sociologist to explain his approach to social sciences, this time by referring to his intellectual and academic trajectory. He notably explains in length his early (intellectual) struggle as young “normalien” (graduate of France’s prestigious grande école Ecole Normale Supérieure) against the grand authorities of the time, and particularly his ambiguous relationship to Claude Lévi-Strauss’ works which he admires for their pioneering character but also sharply criticizes for their epistemological and, hence, methodological premises.
Commonly, Bourdieu’s criticism is understood to have been directed against Lévi-Strauss’ scientistic naturalism for its rigid structuralist thought…and, yes, in this sense Bourdieu can certainly be called a “post-structuralist”. Yet, it is less the naturalism of Lévi-Strauss that is at stake but the ahistorical and unreflective take on societies against which Bourdieu argues. Bourdieu most certainly does not share any so-called post-modern arguments about the utter contigency of society which leaves us with pure phenomenologist thunder and aw. He does argue that social behaviour follows patterns and rules, yet these are historically specific and need to be analysed empirically. He upholds this epistemological position for two reasons: one, because he ascertains that any social situation is fundamentally shaped by power and the particularity of power is exactly that it shapes, determines, limits and enables human behaviour — the sociologist’s task is to analyse these shapes, determinations, limits and abilities and to do so we need to know also the subjective side of power, how power is perceived (or not), used (or not), expanded (or not), diminished (or not) and the effects these power games have on body and mind of individuals, groups adn entire societies.
Second, assuming in a positivist manner fundamental laws of society poses a major epistemological problem, namely the question how we, as researchers and observers, can know, understand, think and talk about these laws to which we would be, logically, also subject. We would have to be able to step through the looking glass and make us as observer disappear in another world (which we know from Alice is a paradox by itself) OR we assume that WE are fundamentally different from THEM. Bourdieu argues that Lévi-Strauss had chosen the latter option, hence, “a vision of the social world based on the denial of the social” as Bourdieu puts it (Esquisse pour une auto-analyse, p. 62) by representing his objects of analysis as preserved in a historical, social, political and cultural vacuum, unconscious of the world around them and ready-made aesthetic, museal objects. The counter argument Bourdieu makes is that the world is not stuck in an eternal variation of the same theme (like we would be running up and down Esher’s staircase) but evolving, changing and moving because individuals, groups and societies do, subjectively, deal with those objective structures of which they are part. Once we admit this, we also have to admit that we, the observers, are part of this “game”. Instead of denying our integration into the subjective living of objective structures, we should rather use this awareness as source of understanding of and knowledge about the social world.
If Bourdieu’s argument would be taken for granted on this very basic level of thought, one could misread him as pleading for an empiricist sociology. Yet, Bourdieu draws on a huge philosophical fund when he negotiates the relationship between the empirical and theoretical, the subjective and objective, his main reference being the French philosopher Pascal (Méditations pascaliennes) and the German “idealist” Emmanuel Kant. Put in a nutshell, he refuses to accept the distinction between the empirical and ideational world and challenges the common argument that one cannot analyse both at the same time (Loic Wacquant has nicely written about the ways Bourdieu bridges the empirical/theoretical and objective/subjective divide here). For the analysis of armed and violent conflict this throws up a row of interesting challenges, and it does so first of all for the questions what exactly should be empirically observed and this question does not only concern the problem whether large or small phenomena should be observed (already discussed in this post) but also what about the conflict needs to be observed.
Taking Bourdieu’s critique seriously one will stumble and fall when trying to identify “causal mechanisms”. Whatever the mutual constitution and influence of agent and structure is, it will hardly be a linear one of an independent variable A having an effect on a dependent variable B, maybe (or not) transformed by intervening variable Z. Not only does the idea of linear causal mechanisms exclude any possibility of reflexive “loops”, it also ignores the idea of mutual constitution (so the acceptance of the fact that one cannot know whether the hen or the egg were first), it denies the freedom of subjective alteration and transformation, it disregards the historicity of structures and it denies the freedom of subjective alteration and transformation of these. Yet, much of the current conflict analysis is still preoccupied with “causal mechanisms” despite the fact that research of the past ten years has shown that there are too many, that they are too unspecified and that there are barely “provable”.
This is particularly evident in the behaviouralist approaches to the influence of economic conditions on war which has been particularly prone to arguing in terms of “causal mechanisms”. Yet, it is the proliferation of hypothetical causal mechanisms that has made this research area one of the most frustrating in civil war analysis. Despite increasing efforts of collecting data, the major neglect of context and complex causalities has produced an erratic variety of “maybe” explanations. Ross for instance identifies five causal mechanisms which could explain the relationship between resource wealth and onset of war ; Humphreys indentifies six “families” of causal mechanisms how natural resources set off civil wars, and another seven families for causal mechanisms how natural resources impact on the duration of conflicts. None of these bundles of causal mechanisms has ever been systematically tested, probably because they are much too complex for linear regression models; particularly if more than two variables have to be assumed of influencing each other dynamically.
Unsurprisingly, this strand of research has not produced any conclusive insights about how economic structures shape the likelihood of collective violence, rebellion or war. Michael Ross’ work is examplary for this: In his early works, swimming in the streamline of Collier and Hoeffler’s greed model, he found a significant relationship between resources that can easily be looted so which excluded for instance oil. Two years later, he finds that oil wealth is correlated with a risk of war as is wealth in diamonds and gas if a different regression model and different data is used. He then, in his most recent book, again belittles the risk of oil as triggering factor for civil wars alltogether, stating “When oil-producing states fall prey to civil war, oil is never the only factor; it is sometimes not even the most important factor” (145). In this strand of research, whether oil is important for the onset or the duration of conflicts does not depend on what people make out (or don’t make out) of oil wealth but on the data rows the researcher uses.
Yet, asking these questions is, again, not enough as I have argued in my post on Weinstein’s and Humphreys’ ill conceived survey of former combattants in Sierra Leone. It is how you ask questions. Peters summarizes this very concisely in his study of young fighters in Sierra Leone when he asserts the necessity for an empathetic encounter which takes the subjective understandings, thoughts and feelings of the object of analysis seriously. However, Peters’ book also epitomizes the practical and methodological difficulties of such research: it requires extremely good knowledge of the society under investigation, including language skills, and access to the population that is observed over a longer period and based on trust and, at least in parts, intimate knowledge of what Charles Tilly called “local scripts”. Most of the literature that provides deep insights into civil wars like Elliotts “Vietnamese War”, Wood’s “Insurgent collective action” or Mats Utas’ “Sweet Battlefields” are the result of years if not decades of work within the communities. Of course these difficulties exist for all deep sociological work, but in cases of collective violence they are aggravated by the sheer brutality of conflicts, the strain their observation puts on the observer who might become witness to extreme cases of violence and who, in any case, will have to take note of the devastation of wars.
Instances of collective violence are furthermore particularly difficult to define and delimit (see my post on Syria) as these rarely take place in all the territory and covering the totality of the population. As Charles Tilly already noted in 1969, collective violence is a particularly ill-bounded social phenomenon, and all tentatives to establish clear definitional boundaries to the categories of violent events observed necessarily allows the influx of normative theorizing about political authority, its legitimacy and the legitimacy of contesting and protesting against it. The risk of being thwarted by the phenomenon observed is clearly recognizable in Carolyn Nordstrom’s work which presents masses of empirical materials, asks extremely well formulated and challenging questions but gives only very little answers or conclusions.
Furthermore, going into the field cannot and must not be the only way to collect data as this would make all historical research futile. Bourdieu himself was, indeed, very critical of history as science (although one might say that this has been ascerbated by the French media/academic context and that most of these debates are rather personal feuds). When dealing with the reported experience and sources, reflexivitiy becomes not only an ethical requirement but an epistemological necessity. There is no language, no experience and no concept that is not shaped by social domination and none has a meaning per se. Only if we ask ourselves what we understand by the words we use and how we understand how others use them, are we able to decipher those social structures of meaning that “make” the world. Empathy is essential but not sufficient; reflexive empathy is necessary if we want to grasp the full meaning of people’s thoughts, motivations, actions and words.
This said, the idea of “authentic” voices, sources or, more generally, data becomes critical. In fact, such “authentic” data does not exist per se; it is interpreted as such by the observer (and then, of course, not “authentic” anymore). In order to show the own meaning that subjects concede to their world, the observer has to render their words “authentically” (e.g. verbatim, as Bourdieu chose to do in La misère du monde) yet these same words also need contextualisation, analysis, dissection and critical examination (in the Kantian sense of “critique” as proof, test, check etc.). Writing about these experiences, rendering subjective thoughts and critically discussing them in the light of objective regularities becomes a challenge of scientific inquiry in its own right.
For conflict analysis this means that we do not forcibly need “more” and “new” data and the tendency, that can be observed particularly in US research, to accumulate more and more interview hours, surveys, datasets etc. is actually detracting the observer’s critical sense from a deep analysis of meaning. Yet, for meaning we need a deep understanding of contexts and this is rather often lacking. We can find for instance dozens and dozens of surveys on about any aspect of the life of Bosnians during and after the war, but we have very little critical, contextual and reflexive-empathetical literature on the lives of Yugoslavs before the war. The same can be said for other societies up to the point that a colleague who is a specialist on Algeria had the one 2013 new year’s resolution of “breaking the 1962 barrier” (i.e. wanting to write a history of Algeria AFTER 1962).
A critical and reflexive-empathetic treatment of data also opens another way of data collection as much more than the politically spoken and written word can become a source once it is reasonably interpretable as expression of meaning. Even the observer’s own experiences may become a source of understanding if they are systematically, critically and empathetically reflected in the context of the analysis. Participant observation hence obtains an epistemological importance that positivist approaches cannot grant.