Myths of the Chicago School

4. The quantitative tradition at Chicago

4.2 Thomas and the case study versus statistics debate

As has already been demonstrated, Chicago sociologists were not opposed to a 'falsificationist' nomothetic basis for sociological research. Scepticism about statistical approaches thus reflected, not an opposition to the fundamental attempt to construct causal or pseudo-causal relationships in sociology but rather a scepticism that statistical methods could adequately grasp the subjective aspect of interaction (communication), hence the initial concern with life history. However, as indicated above, Thomas was more concerned with the attitude-value relationship rather than with a particular method. Later, as attitude testing became more sophisticated he raised no objection to it provided it could generate the information required.

It is my experience that formal methodological studies are relatively unprofitable. They have tended to represent the standpoint developed in philosophy and the history of philosophy. It is my impression that progress in method is made from point to point by setting up objectives, employing certain techniques, then resetting the problems with the introduction of still other objectives and the modification of technique. For example, Galvani or someone else gets a reaction from a frog's leg ... this may suggest to Pfeffer or Verworn the application of electricity ....
In all of this, there is no formal attention to method but the use of some imagination or mind from point to point. The operator raises the question, at appropriate points , 'What if,' and prepares a set-up to test this query.
Similarly, in our own line, some of us, in connection with some experience, raised a question, 'What would happen if we were able to secure life records of a large number of persons which would show their behavior reactions in connection with their various experiences and social situations?' After some experimentation, yourself [Park], Shaw and others have been interested in the preparation of very systematic and elaborate life-histories. In this connection it is noted that the behavior of young persons is dependent upon their social status and the regions in which they live. Studies are then made from the ecological standpoint. It is discovered that children brought into the juvenile court are predominantly from certain localities in the city. The rate of delinquency is related to gang life and gang life is related to localities. Thrasher then makes a study of the gang from this standpoint. As comparative observations multiply, Shaw undertakes to determine how the cases of boys brought into the juvenile court for stealing are connected with their gang life and determines that 90 per cent of these boys did their stealing in groups of two or more. In the search for causes of delinquency, it then appears that the delinquent and nondelinquent are often very much alike in their behavior reactions. It is then recognized that it is impossible to study the delinquent population without at the same time studying the nondelinquent, and at present we have introduced the plan of using nondelinquent groups as a control in connection with studies of the causation of delinquency.
In all this, also, we move from point to point without necessarily any formidable attempt to rationalize and generalize the process. It is only in fact, so far as sociology is concerned, since we abandoned the search for standardized methods based largely on the work of dead men, that we have made the beginnings which I have indicated. (Thomas, 1928.)

Indeed, Thomas came more and more to accept the possibility of quantitative techniques providing the kind of material life histories provided and this was very much a concern of the Chicago sociologists for two decades; a concern that was shared by the wider sociological community. Throughout the twenties a general debate on the relative merits of case study and statistics centred on the reliability of case study data (Cooley, 1928; Shaw, 1931; Rice 1931) [1].
 
In 1930 two PhD theses at Chicago independently undertook empirical investigations of the extent to which attitude testing schedules were able to provide equivalent data to the self composed life history. Stouffer, in his influential thesis, demonstrated that, for some kinds of attitudes, the administratively easier test instrument was as good as the life history record, although perhaps less subtle in the case of extreme attitudes. Brown (1930) also included an assessment of life history and attitude surveys that suggested that for delicate areas the life history was more accurate, although he deferred to Stouffer (1930), in a footnote, as being a more rigorous study in respect of 'straight forward attitudinal statements'. However, this analysis did not ring the death knell of life history (Cavan, 1983). Stouffer still saw a role for it and Burgess, in particular, was concerned to integrate case study and statistics in a synthesis. Indeed, the personal document was still a central feature of methodological discussion a decade later (Social Science Research Council, 1939).
 
There was no attempt at Chicago to establish a position that gave primacy to case study rather than statistical work, nor was the debate about the two methods indicative of competing camps.

Sutherland ... was a most knowledgeable person in criminology... he was brought up to Chicago and he sort of laughed at these debates about case studies versus statistics. He just plodded right along and got case studies when he damned pleased and wanted them and used statistics and was always trying to get a little more statistical methods too. (Cottrell, 1972)

We had a few sessions on statistics versus case studies and most of us students regarded that as light entertainment because we found both of them useful and didn't think of it as versus. Blumer, however, stayed fairly big with this versus statistics and I think still is today. Blumer did not pick up the tight methods and has sort of gone out on a limb. (Faris, 1972)

 The consolidation of case study and statistics, indeed, seems to have begun early at Chicago. Howard Jensen, a doctoral student in the latter half of the second decade of the century apparently

felt that statistics were important although he was not willing to give them first place in his interest. He was a humanist... and he used statistics only to further that interest. But he was not narrow minded on the subject. He felt that both things were important, the other thing being case study, of course. (Mrs. H. Jensen, 1972).

 Bingham Dai obtained his doctorate in 1937 for his research into opiate addiction in Chicago (Dai, 1937). This research in progress he reported to the summer Institute of 1935 and an article on it in the Bulletin noted that the study consisted of two parts '(i) the analysis of statistical data regarding drug addiction, and (ii) case studies based upon the long interview' (Bulletin of the Society for Social Research, June 1935). [2]
 
Chicagoans were, thus, more likely to adopt methods according to circumstance rather than opt for either side of the case study versus statistics debate.
 
Subsequently, the case study-statistics debate shifted emphasis from a concern with the efficacy of statistics in the collection of attitudinal data to a concern with the definitive nature of concepts. Statistical analysis of schedules required that concepts be definitive and that they be predetermined by the interviewer. The debate on 'operationalisation' was indicative of this reorientation of the case study-statistics debate. It was not one concerned with establishing the primacy of quantitative approaches over qualitative ones per se but rather of the possibility of a falsificationist science requiring conceptual explication and accurate measurement (Lundberg, 1936; Waller, 1936).
 
Blumer responded vigorously to the movement towards opera­tionalisation of concepts by arguing that concepts in sociology were primarily 'sensitizing' and not definitive (Blumer, 1931) [3]. This line of debate was not concerned with the efficacy of statistics in assessing attitudes but cast doubt on the possibility of formulating concepts to a degree that measurement would be at all meaningful. Case study and other ethnographic techniques, it was argued, offered a sounder way of generating 'sensitizing' concepts.

These two phases of the debate as to the efficacy of statistics and the possibility of definitive concepts were engaged in as fully at Chicago as elsewhere and there was no 'Chicago' view that, as the myth suggests, saw the 'Chicago School' as defenders of case study and opposed to statistics. The Chicagoans were more likely to adopt work methods to suit circumstances rather opt for either side of the case study versus statistics debate.

Notes

[ 1] There are various notions about the wars and battles fought between the quantitative and qualitative traditions and the exponents of particular positions reconstruct these encounters to project their side in the most favourable light. This leads to contradictions and confusions. Some wars are forgotten, other battles given exceptional prominence, and so on. Thus Blumer is seen as a major standard bearer of qualitative sociology by some historians of sociology, the early Chicago School, by others. Yet others do not seem to distinguish between the two. Howard Becker is sometimes portrayed as the principle mover in a late rear-guard action mounted by qualitative sociologists that became, eventually, necessary following the coup in the American Sociological Society in 1935. Others regard this as superficial posturing, the war having already been lost in the case study versus statistics and life history versus attitude scale debates of the 1929-1931 period. Ethnomethodologists, on the contrary, see the real battle beginning only in the late 1960s when, for the first time, the focus of qualitative sociological enquiry was radically questioned. In short, battle has been joined since sociology became an empirically orientated pursuit in the United States and will probably continue to be joined while a nomological prescription informs all spheres of science.
   The reconstruction of the qualitative-quantitative debate has frequently been in combative terms but this tends to exaggerate the division. While certain elements in American sociology have conflicted, there is an enormous middle ground that has tended to avoid such conflict. Eclecticism of method, and a lack of epistemological dogmatism, has prevailed rather than rigid adherence to singular orientations. This has been, possibly, at least in part, a function of the distance American sociology (and social psychology) has maintained from its European counterpart. American sociology has not developed a full-fledged phenomenological, structural, hermeneutic or critical-dialectical sociology and has maintained a nomological orientation. Return

[ 2] Dai, had reported on his research to the Society for Social Research in April and the minutes record that it was a study based on 'a mass of statistical data' and 'repeated, or protracted interviews with addicts', fifty of whom have been interviewed once a fortnight for a year. In addition

statistical data is being collected for a five year period from 1928-34 on 1219 cases from the Narcotic Bureau and 326 pedlars from the same source, 834 cases from psychopathic hospital, 429 cases from the city police records, 118 cases from the Women's Reformatory at Dwight, 193 cases from the Probation Office, 70 cases from the Municipal Psychiatric Clinic and a few cases from a Behavioral Clinic for a ten year period from 1923 to 1934 and 359 cases from the Keeley Institute (Minutes of the Society for Social Research, April 8th 1935)

 The types of conclusion drawn from the data are simple correlational, e.g. addicts live primarily in the zones of transition, very few crimes of violence are committed by drug addicts, most addicts start between the ages of 20 and 25, the majority of Chicago addicts were born in other states, addiction is negatively correlated to level of formal educational qualification. In the same year, Lindesmith (1937) also received a doctorate for research on opiate addiction, a study that was more dependent on in-depth interviewing. Return

[ 3] Blumer actually used these terms in a paper of 1954. However, he developed the thesis much earlier in Blumer (1931) and later specifically addressed social psychology (Blumer, 1940). In the paper of 1931 he concluded 'What I would declare, then, is that to use concepts in science as natural ultimates instead of tentative convenient conceptions, or to be uncritical or unreflective as to their import, is not likely to lead to genuine understanding and control' (Blumer, 1931, p. 170).

It was to this sentiment, that Lundberg responded in 1936 with his comments on operationalisation, which Blumer (1940, p. 182) discounted arguing that:

The improvement in judgement, in observation, and in concept will be in the future, as I suspect it has been in the past, a slow maturing process. During the process the concept will continue to remain imprecise, but it should remain less so as observation becomes grounded in fuller experience and in new perspectives. Even though imprecise, the concept will serve, as it does at present, to help direct the line of observation and to help guide the forming of judgements involved in that observation. Return

Next 4.3 Park's approach to quantification