PART ONE
Orientations and Techniques
Elizabeth Colson
The Intensive Study
of Small Sample Communities
THE last few years have seen a resurgence of interest in the comparative method among social anthropologists both in Europe and America. Perhaps the greatest recent contribution to comparative research has been that of Murdock (1949), who used information drawn from 250 societies to test the degree to which various social factors were correlated with kinship terminologies. The recent interest in the comparative method undoubtedly reflects the fact that anthropology has reached a new stage in its struggle to become a āscienceā. With the use of the comparative method comes the application of statistical techniques, although much of the information that is available at the present time was not collected with this end in view.
The new techniques of analysis, as well as the new problems with which anthropologists are engaged, have already affected the fieldwork which is being done. Fieldworkers are producing more and more meticulous descriptions with a good deal of quantitative material to back up generalized statements, and at the same time they are narrowing their areas of investigation. This, I think, reflects a general and important trend in anthropology and in the social sciences in general.
It is towards a consideration of these new demands upon the fieldworker that this paper is directed. Kluckhohn (1939) some years ago pointed out: āIf we are to deal with any problem (such as that of the acquisition of culture by individuals) in a way which is reducible to actual human behaviors, generalizations must be given a quantitative basisā (p. 6). More recently Driver (1953) pressed this point home when he wrote, āIf we are going to use more mathematics, we must organize fieldwork with that in mind. We must obtain more quantification of every kind wherever it is possible to do so. . . . If one of the goals of ethnology is to arrive at patterns, configurations, or structures of cultures, these must be determined inductively from adequate numbers of actual facts if they are to satisfy the standards of scienceā (p. 53).
The worker in the comparative field is hampered very badly because he does not have āadequate numbers of actual factsā upon which to base comparisons. Instead he must rely upon generalized statements about social institutions and compare these on an all-or-none basis, though the generalized descriptions may be incomparable in actual fact. Thus, in her article on the Arapesh which appears in Cooperation and Competition among Primitive Peoples (1937), Margaret Mead follows the description of the formal organization of Arapesh society with the comment that the actual functioning of the society is āof a very different order than that implied in their structural arrangementsā (p. 27). She further says that the formal structure is not observed in practice, unlike āmany primitive societies, notably Dobuā (p. 27). If we had the same type of material for other small-scale societies that Mead has for the Arapesh, it might well turn out that the formal structural arrangement of the society offers only a very general guide to behaviour which does not jibe with practice to any exact degree, and that those societies where formal structure and actual practice largely coincide are the exceptions. At present we simply do not know, and the type of comparative work which Murdock and others have attempted to do becomes questionable when viewed against this background. Murdock uses the Arapesh as one of his sample societies. Presumably he works with the formal structure of Arapesh society as he does with the formal structure of the other societies in his sample. As things stand, we do not know whether or not he is comparing comparable phenomena. It is certainly not suggested here that Murdock is unaware of this problem, witness his discussion of the criteria to be used in classifying a given society as polygamous or monogamous (1949, pp. 24ā28). But much of the information available to him is inadequate for the purpose to which it is put.
I suspect, therefore, that the comparative method will contribute little further to our understanding of social organization and the field of cultural phenomena in general until we shift from the all-or-none classifications so largely used at present to a method based on the comparison of rates con-structed from quantitative information collected in a systematic fashion. For this we need new standards of fieldwork. Counting noses is a tedious and uninspiring job, but perhaps the new depths of insight into human organization to which it should lead will at least partially compensate even the fieldworker for the loss of some of his old freedom.
We want, then, quantitative material to back up qualitative statements: an analysis of actual residence to accompany descriptive statements about whether residence is virilocal, uxorilocal, avunculocal, neolocal, or what have you; of actual inheritance and successions to back up stated rules of inheritance; of the number of cross-cousin marriages expressed as a proportion of total marriages; and so on through the whole gamut of social facts used to characterize a society and thus to contrast it with others.
This development should enable us to test a large number of hypotheses which have been put forward over the years about the nature of social interaction. Monica Wilson (1951) has some extremely suggestive things to say about the importance of residence patterns in determining the nature and direction of witchcraft accusations. To test her hypothesis, we need some cross-cultural information, showing the actual composition of residence units and the number, type, and direction of witchcraft accusations made by members of these units. Does cross-cousin marriage resolve some of the stresses that occur in a matrilineal society, as suggested by Malinowski? Some quantitative data on divorce frequencies for cross-cousin marriages as against non-cross-cousin marriages, as well as some better information on the incidence of quarrels between those united by a cross-cousin marriage tie, might help us to check this hypothesis. Does the joking relationship serve as a safety valve for people who are in some way in an ambivalent position with respect to each other? Radcliffe-Brown has made this suggestion, but nobody to my knowledge has actually tested it in a systematic fashion.
Nadel (1947), on the basis of his work among the various Nuba groups of the Sudan, has suggested that an intrinsic and logical correlation exists between social rigidity or inclusiveness and the incidence of suicide (p. 480). In this, of course, he is following in the footsteps of Durkheim. He points out that his information on the incidence of suicide for the different Nuba groups is inadequate. He usually cites the number of cases which his informants listed for him, or remarks that it was unknown or denied by informants of some particular group. This is enough, perhaps, to point the problem. But since he does not give the size of the population or the time period to which the information applies, it is impossible to construct a suicide rate for comparisons with other societies. I suspect on the basis of my own fieldwork among the Plateau Tonga of Northern Rhodesia (Zambia) that their suicide rate is high, and I think that the frequency may be related to the diffuse nature of the social network in which the Tonga finds himself, where he lacks the security that a more rigid formulation of rights and obligations might provide. On the basis of my material, I may be able to work out a very crude rate, with number of suicides, size of population, and period of time taken into account. But until other anthropologists produce the same sort of material for the peoples with whom they are working, I am in no position to be sure whether the Tonga takes their lives with any greater or lesser frequency than the majority of other pre-literate peoples. The Nuba material as it stands is suggestive, but useless for comparative purposes. What is needed is material of a type that allows of a direct comparison of rates. This, of course, will give the information for only one of the factors involved in Nadelās hypothesis. To test the hypothesis itself, we shall also need some means of stating objectively the degree of diffuseness or rigidity of social relationships, and to state this in such a way that it will be possible again to collect the information that will allow of the construction of rates for cross-cultural comparisons.
Gluckman (1950) has suggested that frequency of divorce is correlated with social structures of a particular type, but with the crude information at his disposal he is able to do little more than state the problem. To test his hypothesis we need divorce rates, worked out in a comparable fashion, for a series of societies with different forms of social structure. Barnes (1949; and see below, pp. 61ā64) has attempted to develop a method of computing a divorce rate for non-literate societies where no records are kept of either marriages or divorces, and he has combed the literature for quantitative information in divorce. Other anthropologists working with various societies in Central Africa, under the auspices of the Rhodes-Livingstone Institute, are collecting quantitative information on divorce in a standard fashion and will use the method worked out by Barnes for presenting their results. We shall soon then be in a better position to test Gluckmanās hypothesis, and to refine it (cf. p. 23).
I have suggested (1951) that population movements may be related to strains within the social structure, and have characterized the Plateau Tonga as highly mobile, i.e. as making frequent changes of residence from one village to another. Although, in the paper in which I put this forward, I gave a certain amoun...