Social Research Glossary

Social Research Glossary

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z Home

Citation reference: Harvey, L., 2012–14, Social Research Glossary, Quality Research International, http://www.qualityresearchinternational.com/socialresearch/

This is a dynamic glossary and the author welcomes e-mail suggestions for additions or amendments. Page updated 11 November, 2022 , © Lee Harvey 2004–14

A fast-paced novel of conjecture and surprises

_________________________________________________________________

Howard Becker and Blanche Geer on Participant Observation as a Sociological Method

additional material

Introduction

The series of articles produced by Geer and Becker in the 1960s constitute an episode in the qualitative-quantitative debate in which the ethnographers were under pressure and felt it necessary to validate their methodology.

There were several varieties of symbolic interactionism and some were more inclined than others to incorporate the demands of the quantitative practitioners for validity and reliability.

Becker and Geer were among those who took on these concerns and could be regarded as the 'loyal opposition' (Mullins, 1973) to the more dominant quantitative tradition in the 1950s and 1960s. Others, such as Goffman, were less concerned with codifying methodological procedures and less inclined to engage in methodological debates. Nonetheless, it is not unreasonable to argue that the symbolic interactionists in general offered a 'loyal' opposition to the structural-functionalist based dominant variable analysis approach in as much as there was no fundamental critique of positivist epistemology.

The aim of the articles produced by Becker and by Geer around 1960 was, effectively, to legitimate participant observation as a 'valid' data collection process. These articles will be examined in some depth below, but some general points need to be outlined initially.

First, these articles are part of the literature of the time, which attempt to show that participant observation could and should be a systematic technique. Second, they are involved in the argument that participant observation is not simply an exploratory tool of social research, and that, third, participant observation can generate and test theory and thereby conform to the taken for granted standards of 'scientific research'. Fourth, these articles can be seen as part of the debate that sees participant observation as an alternative approach to sociological research, which exposes and confronts some of the limitations of scheduled interviewing and quantitative techniques.

It is this latter point that has the potential for a critique of the prevailing positivistic view of scientific sociology. However, this point is developed in three ways and only one of them constitutes a critique of positivism. The first line of development is 'triangulation'. Triangulation is the notion that various different techniques can be combined to give a more balanced view. This presupposes a common aim, and tends to take for granted the positivistic tenets embodied in variable analysis.

The second approach is to argue for the validity of participant observation as a technique that underpins all other methods and is the method by whose standards all other methods should be judged. Thus, large-scale quantitative surveys are seen as adequate based on the degree to which they fulfil the potential of participant observation. In its weakest form this view suggests that in some areas (e.g. deviancy studies) participant observation is the only valid or viable technique. The strongest development of this argument is that participant observation reveals the 'reality' underlying what people say.

This position is close to the third line of development that, in arguing that what people say and what they do are different, offers the basis for a critique of prevailing positivist views. This was, essentially, a position advocated by Blumer from 1930, although with decreasing conviction over time. It was a position developed in a variety of ways by the ethnomethodologists from the mid-1960s. It is interesting that the legitimation embodied in Becker and Geer's methodological articles (which supposedly reflect the Chicago opposition to structural functionalism) only hint at this critique.

These articles will now be examined.

Problems of proof

In 'Problems of Inference and Proof in Participant Observation', Becker outlines the methodic principles of participant observation. He lays out the analytic operations involved and shows how the method constitutes a systematic approach. He addresses himself to the issues of validity and proof and argues that it is only in respect of presentation rather than in any substantive way that participant observation differs from qualitative sociology on the matter of proof and inference.

Becker creates some confusion when he comments that sociologists usually use the method

'when they are especially interested in understanding a particular organization or substantive problem rather than demonstrating relations between abstractly defined variables.' (Becker, 1958, p. 399)

For some commentators this implies that participant observation is the method of understanding rather than of explanation.

For others, it implies that the scope of participant observation is limited to that of exploratory method. This latter view is reinforced when Becker goes on to say that such sociologists do not presume they know enough about the organisation a priori to identify relevant problems and hypotheses and that they must discover these through the research. This constitutes a weak reassertion of Blumer's critique of variable analysis in which he argued for sensitizing rather than definitive concepts in sociology.

In 'Problems of proof...' Becker is concerned with the role of, participant observation in both discovering hypotheses and testing them. This, he claims, is what typically participant observation is concerned to do. The problem for Becker is that given the vast amount of 'rich' but varied data, how does one analyse it systematically and present conclusions convincingly?

Becker reasserts the cumulative model view of the development of sociological knowledge by suggesting that participant observation research is sequential and reflects the technique of analytic induction. He points to three distinct stages of the fieldwork and a final analytic stage once the fieldwork is completed. The stages are: first, the selection and definition of problems, concepts and indicies; second, the checking of the frequency and distribution of phenomena; third, the construction of social system models; and fourth, the post-fieldwork stage of final analysis and presentation of results.

These stages are differentiated not only sequentially but also by the kinds of conclusions each generates and by the different kinds of criteria used to assess evidence and reach these interim conclusions.

In the first stage, the researcher seeks out problems and concepts that will maximise understanding of the organisation under study. Indicators for less ostensive 'facts' are assessed. Typically, this stage does nothing more than conclude that particular events occurred, that a given phenomenon exists, that two phenomena have been observed as related in at least one instance, and that a certain statement appears to be a useful indicator for a less tangible attitude. For example, Becker suggests that in the study of medical students (Becker et al., 1961) complaints of weekend work provide an indicator of students' perception of workload.

This first stage is a speculative stage and these provisional conclusions and the hypotheses that are constructed from them may be discarded. However, this stage is not mere guesswork nor should it be an arbitrary and impressionistic exercise overdetermined either by preconceptions or initial observations. Certain 'problems of evidence' must be confronted by the researcher at this stage if 'false leads' are to be reduced to a minimum.

Becker would like to have had an established set of scientific guiding principles, or canons, for systematic assessment of evidence. He regrets their absence but suggests some 'commonly used tests' as surrogate canons. These tests concern establishing the credibility of informants; differentiation of volunteered and directed statements; addressing the context of statements (private/individual or public/group); and reflecting on the researcher's own role in the group.

The second stage, checking on frequency and distribution of phenomena, is essentially a quantitative exercise. Incidents of phenomena are logged (and counted) and interrelations between them are noted. Becker suggests that, typically, participant observation cannot fulfil the theoretical requirements for statistical analysis and that the collected data are quasi-statistics. The numerical results, then, are not, and do not need to be, precise. What is sought are suggestions of prevailing tendencies.

In effect, Becker is overemphasising the quantitative role in participant observation. Becker may simply have been attempting to offer a sop to quantitative researchers, or more likely just reflecting the prevailing positivistic view of the scientific method that had become so entrenched in sociology (Lundberg; 1936) in the United States. However, Becker overstates his case by his references to probabilistic analysis. He suggests that, in drawing up quasi-statistics, the participant observer is not attempting to establish the truth or falsity of a conclusion but, taking a cue from statisticians, merely its likelihood.

The parallel drawn here with statistical analysis is entirely spurious. Statistical probability is dependent upon theoretical considerations in no way matched by the generation of quasi-statistics. So, Becker's assertion that in constructing some kind of assessment of the likelihood of the relationship between observed phenomena the participant observer reflects quantitative techniques is misleading. What his pronouncements do show is that he fails to provide any solid critique of the prevailing 'definitive concept—cumulative development model'.

The third stage in the field is the construction of social systems models, this is the incorporation of individual findings into a generalised model of the social system or organisation under study, or of some part of it. Typical of the conclusions at this stage are statements about a set of complicated interrelations among many variables. Such conclusions relate to the necessary and sufficient conditions for the existence of a given phenomenon, to the establishment of some phenomena as basic elements and to the link between these empirically validated phenomena and more abstract constructs in established sociological theory. These kinds of projected conclusion clearly reflect Becker's acceptance of the explanatory cumulative model. He explicitly accepts that the kind of participant observation he is discussing takes on board the notion of social system as a basic intellectual tool (and in his development of the technique it is clear that this conceptual assimilation involves no critical edge) and that the aim is to 'explain particular social facts by explicit reference to their involvement in a complex of interconnected variables'.

Becker almost apologises to quantitative practitioners for the tendency for ethnographically derived models to be descriptive rather than mathematical and suggests that while factor analysis may have a role to play here, that current available statistical techniques are inadequate.

Nonetheless there is no question here of challenging the prevailing scientistic view. The identification of necessary and sufficient conditions reflect the, then, contemporary view of causal analysis. The reductionism inherent in isolating basic phenomena, the attempts to build up and elaborate existing sociological theory, and the taken-for granted view of the social system, all reflect the dominant approach of middle-range theorising.

The third stage also reflects a wider metascientific concern that underpins the positivistic view of scientific sociology in post-war United States. Middle-range theorising, which seeks out explanations is underpinned by a falsificationist view of scientific proof. In this respect it confronted the naive inductivist view and the problems of complete proof that such a perspective embodied.

Falsificationism, as developed by Popper, is a sophistication of inductivism and one that underlies quantitative empiricist analysis. Its call for clear concepts and bold conjectures and refutations is essential to middle-range theorising. The construction and testing of hypotheses in probabalistic terms is the social sciences accommodation of the falsificationist principle Popper proposed for singular statements. This accommodation took account of the non-experimental methodology applicable to the social sciences and, incidentally, a pragmatic approach to Popper's normative prescriptions.

Becker, in his suggested third stage, similarly incorporates a falsificationist principle. He suggests that, after constructing a model specifying the relationships among various elements the model is successively refined by searching for negative cases. He suggests that

'While a processural model may be shown to be defective by a negative instance which crops up unexpectedly in the course of the fieldwork, the observer may infer what kinds of evidence would be likely to support or to refute his model and may make an intensive search for such evidence'. (Becker, 1958, p. 108).

The construction of models of parts of the social organisation goes on simultaneously and some areas will be more developed than others. As the field work progresses, tentative links between separate substructures are made and these are more fully assessed and developed in the final, post fieldwork, stage.

This final stage is systematic and involves checking and rebuilding models with as many safeguards as the data will allow, notably by cross classifying all items so that checks can be made as complete as possible.

This approach can, as Becker suggests,

'profit from the observation of Lazarsfeld and Barton that the "analysis of 'quasi-statistical data' can probably be made more systematic than it has been in the past, if the logical structure of quantitative research at least is kept in mind to give general warnings and directions to the qualitative observer."'(Becker, 1958, p. 409)

Becker, then, presents a legitimation for, and an elaboration of, participant observation as a technique wholly within a prevailing middle-range theorising framework. This is reflected in the problems he poses for himself (and is unable to answer) concerning the specification of proof. He notes that

'readers of qualitative research reports commonly and justifiably complain that they are told little or nothing about the evidence for conclusions or the operations by which the evidence has been assessed.' (Becker, 1958, p. 410).

The problem, for Becker, is that qualitative data are difficult to present adequately whereas the quantitative analysis can utilize tabular form to summarize data that have been collected for a few fixed categories and assigned a limited range of values within those categories. Statistical shorthand, both inferential and descriptive measures, may then be used. Participant observation data cannot be so summarised, nor can all the data be published or even made available in archives. Even if the latter were possible it would not obviate the problem of how to present proof.

Becker offers a tentative solution through his suggestion that a 'natural history' of the process by which the research reached its conclusion should be presented. Such a natural history would require the presentation of how the data evolved, thus revealing the inferences made at each stage. He suggests that this would provide the reader with the opportunity to assess the method and criticise the way conclusions were made as statistical analyses currently do. This ignores the adequacy of the operationalisation procedures in both types of research.

The above analysis of Becker's seminal article reveals the essentially positivistic nature of the participant observation he advocates. He argues that participant observation can be a systematic approach to data collection, that it generates and tests hypotheses and is more than an exploratory tool. Participant observation provides a framework for elaborating theory.

Becker provides no strong rationale for participant observation in itself vis á vis quantitative research, and in no way attempts to establish an independent epistemological grounding for participant observation. Becker makes two incidental points about the rationale for participant observation. First, participant observation is a method suitable to practitioners who have no clear a priori concepts. Second, participant observation is suitable for close scrutiny of organisational structure. A view tended to emerge in the 1960s that participant observation was, indeed, an exploratory technique and that it was suitable for small group analysis of otherwise inaccessible organisations or groups.

Becker made no attempt to critique the epistemological underpinnings of prevailing scientific sociology nor took into account the reservations of (his supposed mentor) Herbert Blumer (1956)

First days in the Field

In 'First Days in the Field', an account of ongoing fieldwork practice (published in Hammond (1964), but written around 1960), Geer reasserts the primary concerns of what has become labelled the later Chicago symbolic interactionist approach to participant observation. (These practitioners would be the fourth generation of Chicagoans). Many of the issues she addresses are covered by Becker (1958) as outlined above, although the focus of her analysis is on the early stages of observational fieldwork.

Geer, however, does suggest a slightly altered orientation to that advanced by Becker (1958). Geer's purpose was to provide insights into the relationship between fieldwork experiences and both preconceptions and final analysis. She attempted to reveal the extent to which strategies and concepts change during the early days of field research, what mechanisms were operating to effect changes and what impact this had on later fieldwork and how much of these early experiences get into the final publication.

Geer argued for value free or neutral observational research. The researcher should not contaminate the research environment by appearing to take sides. [See also Becker, 'Whose side are we on?'] Neutrality, for Geer, is avoidance of adopting 'out-group' identification and also mannerism that would lead to being regarded as one of the subject group. For Geer, the researcher should be a passive data gatherer, a role in principle like that of the scheduled interviewer. For Geer, the researcher should avoid 'wising up' the informants. [This is the opposite of later feminist positions, as for example in the work of Ann Oakley]

Geer's discussion considered methodological problems that may be seen as more germane to an interpretive approach. When, for example, she referred to the problem of empathy, it was about confronting prejudices, thus: 'to study, as we propose to do, the experiences of going to college from the point of view of the student necessarily entails at least recognition of personal bias in order to achieve empathy with the informant group.' (Geer, 1964, p. 328)

This led to some conflict with the view that one avoids taking sides, and while she argued that she 'doesn't pretend to be one of them', she admitted that during her initial period of fieldwork she soon started to take the subjects' side despite being unfavourably disposed towards them prior to the commencement of the fieldwork.

Another hint of the interpretive role that participant observation could play for Geer emerges from her comment that the fieldwork reduces problems rather than increases the problems of conceptualisation:

'researchers underestimate people's trust in our neutrality, their lack of interest.... And we project theoretical problems into the field. Because the process of group formation is difficult to conceptualize we suppose it will be difficult to observe. We expect ephemeral, unstructured situations ... to be incoherent. Perhaps such mistakes are a necessary part of our efforts to design the study in advance.' (Geer, 1964, p. 331).

The irony is that having identified the interpretive context for participant observation, Geer delimits the scope for interpretive enquiry by arguing for pre-design of the research endeavour. This pre-design includes the designation of a series of potentially relevant variables.

Geer, then, proposes the sequential model outlined by Becker (1958) although she reframes as the generation and testing of working hypotheses and their combination into compound propositions. Again, like Becker, she sees the first operation as consisting of the testing of 'crude yes - or - no propositions', the second stage is 'seeking negative cases' or setting out 'deliberately to accumulate positive ones'. One disconfirming instance, she argues, forces modification. A simplistic falsificationist model is reaffirmed, such that confirmation of 'what is' is accomplished by eliminating 'what is not'. The third stage is elaborated, by Geer, into a proto-path analytic model, following the suggestions of George Polya (1954)

Geer's conclusion is that the first days in the field do radically affect strategies and concepts that one brings to the field. However, the development of empathy with subjects is not seen as a long-term problem for the exposition of a 'value free' research. Indeed, very soon after leaving the field the researcher looses the empathy and is able once again to be a neutral observer. Discussion with co-workers is useful in rapidly restoring a balance, she argues.

Despite concluding that the first days in the field may transform a study out of recognition, Geer merely reflects Becker's earlier comments, failing to take the opportunity to propose an interpretive ethnography with distinct epistemological possibilities.

Geer reasserts the systematic nature of participant observation and its explanatory potential. She makes no attempt to argue a role for it as a technique able to develop social theory nor does she provide a rationale for its use, making no attempt at all to suggest that it may be useful to dig beneath the surface.

Participant observation and scheduled interviewing

Becker and Geer's joint article, which compares participant observation and scheduled interviewing essentially argued that the former acted as a yardstick by which to measure the effectiveness of structured interviewing.

related areas