**Methodologies for studying figurations: an open question about best practices**

In *What is Sociology?* Elias makes the following observation: ‘If sociology has to investigate figurational processes resembling complex games, then statistical aids must be developed which will be suited to this task’ (1978: 132). This statement was originally published in 1970, technologically speaking a long time ago. Much has changed in terms of the development of statistical modelling systems, and other tools for analysis. Sensing an affinity between the figurational processes Elias explains (which we seek to study in our respective research projects: a) understanding and explaining ecological habitus development processes and b) the analysis of path-dependency in processes of overuse) and the uses of sophisticated game models, complexity theory, resilience theory, and complex systems thinking in general, we want to learn more about whether and how sociologists are using these and other methodologies to study figurations.

We would appreciate any input about this topic. We look forward to hearing from you!

Debbie V.S. Kasper, Sweet Briar College, VA, USA dkasper@sbc.edu

Wijnand Boonstra, Swedish University for Agricultural Sciences, Uppsala, Sweden, wijnand.boonstra@sol.slu.se

Following Cas’s quote of Elias’s comments on stats. signif., I have an essay forthcoming in Contemporary Soc. on the misuse of science in soc/behav studies. #76 on my website is the final draft.

Also I have a suggestion for an appropriate methodology that I call part/whole relations. See #75 for a recent version.

Tom

Norbert Elias on the difference between statistical significance and sociological significance (taken from The Established and the Outsiders (1965), chapter 1 ‘Considerations of procedure’:

It often seems to be felt that only statistical analysis can provide the impersonal certainty expected of a sociological enquiry. Statements not based on measurements of quantifiable properties are frequently dismissed as ‘impressionistic’, as ‘merely descriptive’ or as ‘subjective’. Other investigators before must have been troubled by the inadequacy of a conceptualisation which implied that any verbal statement which bears no direct reference to statistical data is necessarily unreliable, imprecise and scientifically suspect, that the only certainties one can have about social phenomena are those based on statements which tell us how much more there is, or was, of phenomenon A than of phenomenon B. Yet statements of this latter type are often not very illuminating unless they are combined with other statements about the mode of connection between A and B – unless procedures aimed at certainties about quantities are enriched by procedures aimed at certainties about figurations.

In actual fact such procedures – figurational analysis and synopsis – form an integral part of many sociological enquiries. They play a part for example in the building of models on the largest as well as the smallest scale – models of bureaucracies as well as villages, of balance of power systems as well as families; they can be found everywhere in the development, making and revising of sociological hypotheses and theories. They play a part, but they are still insufficiently conceptualised as characteristic procedures of a science whose central task is the study of individuals as groups, of figurations of individuals as such. That individuals are to be studied first as isolates, and that the figurations which individuals form with each other are derived from what they are without figurations, is an odd idea which profoundly confuses enquiries into these figurations. The impoverishment of sociology as a science which has resulted from the prevalent evaluation of sociological methods – from the assumption that it is enough to use statistical methods if one wants reliable answers to sociological problems, is obvious enough. It has led to a state of affairs in which wide areas of sociologically relevant problems are either left unexplored or, if explored, may be protected from the slur of being ‘merely descriptive’ (because they are non-statistical) simply by a great name (as is the case with the greater part of the empirical work of Max Weber); or else they are undertaken as non-statistical enquiries simply because they seem fruitful without explicit reflections on the nature of the procedure which makes them so.

Hence, the use of such procedures of figurational analysis and synopsis is still largely confined to the accidence of individual gifts. It does not yet form an integral part of the training of sociologists to learn to observe and to conceptualise systematically how individuals cohere, how and why they form with each other this particular figuration or how and why the figurations they form change and, in some cases, develop. Yet, to overcome the limitations of sociological enquiries centred on statistical procedures is possible only if enquirers trained to perceive and to manipulate single factors or variables join hands with, or are themselves qualified to act as, enquirers trained to perceive and, at least conceptually, to manipulate figurations as such – trained for accurate synopsis as well as for accurate analysis.

Models of figurations, of social patterns or structures can be no less precise and reliable than the results of quantitative measurement of isolated factors or variables. What they lack is the deceptive finality of inferences based on quantitative analysis alone which is often mistaken for precision. Like hypotheses and theories in general they represent extensions, advances or improvements of the existing fund of knowledge, but they cannot pretend to be an absolute terminus in the quest for knowledge which, like the philosopher’s stone, does not exist. Models of figurations, the results of figurational enquiries, form part of a process, of a growing field of enquiries, and in the light of its development they themselves are open to revisions, to criticisms and improvements, the fruits of further enquiries.

The apparent finality of each statistical enquiry [on the one hand] and the openness, and developmental character of figurational enquiries as links in a chain [on the other] are closely connected with certain basic differences between the type of thinking required for a purely statistical and that required for a sociological analysis. In both cases analysing means focusing attention on one element of a figuration at a time – ‘factor ‘, ‘variable ‘, ‘aspect’ or whatever one may call it. But in a purely statistical analysis the study of such elements in isolation is treated as the primary and often as the principle task; ‘factors ‘ or ‘variables’ and their quantitative properties are treated as if they were in actual fact independent of their place and function within that figuration, and statistical correlations, including statistical correlations of relationships, never cease to be correlations of isolates. Sociological analysis is based on the supposition that every element of a figuration and its properties are what they are only by virtue of their position and function within a figuration. In that case, analysis or separation of elements is merely a temporary step in a research operation which requires supplementation by another, by the integration or synopsis of elements, just as the latter requires supplementation by the former; here the dialectic movement between analysis and synopsis has no beginning and no end.

On the assumptions underlying the traditional forms of statistical analysis one might have been justified in thinking that it was enough to determine the numerical size or other quantitative properties of each of the three zones of Winston Parva and, among them, of the minority and majority groups, in order to explain the different part which minorities play in these three zones and in their respective images. The problems with which one was confronted in a figurational analysis and synopsis were such that the finding of quantitative relationships alone, however precise it might have been, could not lead to an adequate answer. These problems were centred on figurations such as ‘working-class-minority-in-middle-class-residential-area,’ ‘middle-class-minority-in-old-working-class-area’, ‘problem-families-in-new-working-class-area’, ‘old-families’-network-in-relation-to-newcomers’, ‘established-power-elites-in-relation-to-outsiders’. However many statistical correlations one might care to establish, they alone could not lead to a clear understanding of the way in which figurations such as these functioned and in which they affected the people who lived there. One could not infer from a mere quantitative analysis, for example, that for the people of a middle-class area, for their manner of life, for the images they had of their own and of other zones the existence in their own zones of a working-class minority was of no significance while for the conditions of life as well as the images of the new working-class area their minority was of the very greatest significance. In some cases quantitative differences and relations were extremely useful as social indices. That rents were generally lower in Zone 3 than in Zone 2 and in Zone 2 than in Zone 1 was certainly suggestive. But the actual figuration, the complex relationship between these three zones could neither be adequately presented nor adequately explained by other than verbal symbols. Without the use of words as instruments of research, figures remain mute. The different roles of minorities in different figurations are an example. In the context of a neighbourhood such as Zone 3, a specific minority played a part quite out of proportion to its numerical size. The present use of statistics often seems to imply that the greater the numerical size, the greater is the significance. In the case of the minorities in Winston Parva as in many other cases, sociological significance was in no way identical with statistical significance. They pointed to a fact known, though perhaps not sufficiently noticed, from other enquiries: that social data can be sociologically significant without having statistical significance and that such data can be statistically significant without having sociological significance.