Qualitative Research is enjoying a new found respectability in medical sociology, derived in part from an increasing willingness to submit to positivist criteria of reliability and validity. Whilst such claims to 'scientific' credibility have raised the status of the approach, this has only been achieved by driving a wedge between ethnographic methods of data-collection and their origins in the phenomenological strands of sociological thought. One consequence of this schism has been to rob qualitative research of its critical potential, transforming it from a means of challenging discursive formations into a mechanism of surveillance.
This paper defines the broad contours of a qualitative methodology synthesised with the perspective of social critique. Positivist arguments are rebutted and validity is re-conceptualised as reflexive management of the relationship between the testimony of informants and a broader process of historical and structural analysis. The process of managing validity is illustrated for each stage of the research cycle.
Qualitative methods of inquiry have often been viewed with ambivalence and a degree of trepidation by researchers concerned with the sociology of health. On one hand such methods offer an important link to some of the main concerns of sociological thought, addressing questions of power, ideology and subjective meaning. Whilst on the other hand, they may be viewed as suspect in terms of their validity and reliability, particularly when compared with the more 'scientific' methods available to the quantitative researcher. Traditionally, scepticism about the value of qualitative methods has been strongest in the sub-discipline of medical sociology where, until quite recently, the dominance of a positivist model of bio-medical research has tended to exclude phenomenological approaches.
Over the last ten years the situation has begun to change, and qualitative research has gradually acquired a new respectability, with even the British Medical Journal, prepared to recognise its worth (Mays & Pope, 1995). But this new found acceptance has been achieved at a cost; most notably it has entailed, if not a complete capitulation to quantitative criteria of validity and reliability, at least a tendency to meet them half-way. Whilst this compromise has undoubtedly raised the status and acceptability of qualitative research it has also weakened the link between the technical process of ethnographic data collection and its basis in sociological theory. One consequence of this schism has been to rob the methodology of its critical content - the assumption appears to be that qualitative research can be valid, or it can be critical, but not both at the same time. The purpose of this paper is to refute that pessimistic conclusion by sketching out the broad contours of a qualitative research model that is not only critical, but also valid and reliable on its own terms, rather than in accordance with the narrow constraints of positivism.
Generally, qualitative research can be characterised as the attempt to obtain an in-depth understanding of the meanings and 'definitions of the situation' presented by informants, rather than the production of a quantitative 'measurement' of their characteristics or behaviour. This concern to reveal the subjective beliefs of those being studied is common to ethnography, participant observation, and the various other strands of qualitative research. For many qualitative researchers the subjective beliefs of the people being studied have explanatory primacy over the theoretical knowledge of the researcher, thus Jorgensen suggests:
While the researcher may have a theoretical interest in being there, exactly what concepts are important, how they are or are not related, and what, therefore, is problematic should remain open and subject to refinement and definition based on what the researcher is able to uncover and observe. (Jorgensen, 1989, p. 18)
The ethnographer's concern to avoid imposing a theoretical framework of meanings and definitions perhaps originates from the anthropological study of low technology tribal cultures in the third world, where the intention was mainly to describe cross-cultural variations in social behaviour and beliefs before they disappeared. Methodologically this entailed detailed observation and interaction by the researcher, in order to see the world 'through the eyes' of the people being studied. The same approach has been adapted to the study of sub-cultures within western society, most notably in the classic studies conducted by the Chicago School and succeeding generations of neo-Chicagoans. As Downes and Rock have noted, the primary imperative for such research is to catalogue and describe a particular worldview without imposing external theoretical schema:
The interactionist takes his job to be the documentation of the social worlds that constitute a society. He methodically plots the connections between communication, meaning, symbolism, and action. He would claim that there is little profit in imposing alien interpretative schemes on a world: people do not build their lives on the logic of sociology or the sensibilities of foreign groups. They have their own methods of doing things together. (Downes & Rock, 1986, p. 143)
Whilst this approach has yielded a diverse and colourful variety of descriptions of everyday life, it poses a number of problems or limitations when viewed from the perspective of critical social research. First, no attempt is made to place the beliefs and behaviour of the people being studied into an historical or structural context; it is considered sufficient to simply describe different forms of consciousness without trying to explain how and why they developed. This leads to a second problem - the tendency to adopt an uncritical attitude to the beliefs and consciousness of informants, without considering their epistemological adequacy or their emancipatory potential. The result is a form of voyeuristic relativism where everyone's testimony is accorded equal status, and no attempt is made either to explain or inform the development of consciousness.
Superficially, such an approach appears to be the epitome of a value-free sociology; rather than passing judgement on the lives of others the researcher becomes an impartial reporter enabling informants to express their own definition of the situation. However, the reluctance to address the processes by which different forms of consciousness are socially and historically constructed, coupled with the absence of any evaluation of the epistemological status and emancipatory potential of a set of beliefs, amounts to little more than a passive legitimation of dominant ideology. Moreover, it is based on an assumed antagonism between social theory and the immediate experience of everyday life. Rather than the means of obtaining a critical consciousness of ideological oppression, social theory is conceptualised as an inevitable part of the dominant ideology - something to be resisted and struggled against, rather than a basis for emancipatory activity. In short, it is assumed that the sociologist has everything to learn from the people she or he studies, but that critical social theory can have no reciprocal role in their emancipation.
In the light of such criticisms a new generation of 'critical' ethnographers (see Hammersley, 1992) have attempted to synthesise the traditional focus on the meanings and definitions of those involved in a social phenomenon with the insights gained from social critique. The objective is still to access the subjective beliefs of the people being studied, but rather than accepting such beliefs at face value they are examined critically in the context of a broader historical and structural analysis. Whilst Hammersley is keen to distance himself from critical ethnography, he recognises the value of this shift in emphasis:
...we have no grounds for dismissing the validity of participant understandings outright: indeed they are a crucial source of knowledge, deriving as they do from experience of the social world. However, they are certainly not immune to assessment, nor to explanation. They must be treated in exactly the same manner as social scientific accounts. (Hammersley & Atkinson, 1983, p. 234)
Crucially, appraisal of the testimony of respondents amounts to much more than simply checking that they are telling the truth, it entails looking at the processes that shaped their views and assessing the extent to which they may be distorted by ideology. The intention is not primarily to 'weed-out' unreliable testimony, but to use historical analysis and sociological theory to get beneath the surface of everyday 'common-sense' assumptions in order to arrive at a deeper level of understanding that will not only be of academic interest to the researcher, but will also contribute to the development of critical consciousness amongst oppressed groups. However, the synthesis of ethnography and critical social research is an uneasy one, not least because it necessitates a re-conceptualisation of validity, in a form quite different to that adopted by traditional ethnography or positivist empiricism. The purpose here is to commence this reconceptualisation, and in so doing reclaim qualitative methods for a critical sociology.
Critical inquiry is diverse and flexible and only takes on a specific form when applied to the study of a particular phenomenon. Even so, the approach has several essential elements, and the intention is to summarise them in order to arrive at a deeper understanding of the perspective, and its consequences for ethnographic validity.
At the core of a genuinely critical methodology lies the application of dialectical logic. Dialectical reasoning is complex, and only a brief account can be given here. Essentially, the approach addresses the relationship between objects and events in the material world and their subjective representation in human consciousness. The dialectical character of this relationship was addressed by G. W. F. Hegel, but he took as his point of departure the realm of ideas, or human spirit, and was concerned with the way in which human thought became materialised through practical activity. From this perspective the individual's knowledge of the world was essentially knowledge of materialised, and therefore alienated, human thought. Through practical activity the individual could overcome his/her alienation and re-appropriate the human spirit as it had been externalised by previous generations. Ilyenkov gives an eloquent account of Hegel's approach:
This world was the materialised thought of humanity, realised in the product, was alienated thought in general; and the individual had to de-objectify, and arrogate to himself, the modes of activity that were realised in it, and it was in that the process of his education properly consisted. In the trained mind categories actually functioned as active forms of thought activity, forms of processing the material of sense impressions into the form of a concept. When the individual had them in his experience, and made them forms of his own activity, he also possessed them, and knew and realised them, as thought-forms. Otherwise they remained only general forms of the things given in contemplation and representation, and counterposed to thought as a reality existing outside it and independently of it. (Ilyenkov, 1977, p. 208)
Thus Hegel shifted the focus of logic towards consideration of sensuous human activity, and indicated the extent to which the products of human thought and intervention in nature take on a naturalised and ahistorical appearance. Much of Hegel's logic is retained in modern dialectical thought, but for one fundamental difference. By starting in the realm of ideas and examining their impact on the material world, Hegel was unable to account for the origins of those thoughts, and was obliged to rely on an essentialist and metaphysical account of the 'human spirit'. Vulgar materialists, like Ludwig Feuerbach, noted that this amounted to little more than religious belief, and sought to ground human consciousness in experience of the material world. Thus for Feuerbach it was the real world that became imprinted on the physical human brain, rather than the unfolding of the human spirit that gave rise to the world.
Although Feuerbach's account of human consciousness dispenses with the idealism of Hegel's approach, it only does so at the cost of perceiving humanity as the passive recipient of data from the material world. It took Marx to combine the materialist basis of consciousness identified by Feuerbach, with the dialectical insights of Hegel. For Marx, consciousness came from the individual's experience of the real world, but this experience was one of practical activity, of conscious intervention to adapt nature to meet human needs. From this perspective phenomena could still be identified as socially constructed, that is, as the product of human will, without resorting to a metaphysical account of 'human spirit'. Ilyenkov has noted the advance that this represents over vulgar materialism:
Materialism in this case does not consist at all in identifying the ideal with the material processes taking place in the head. Materialism is expressed here in understanding that the ideal, as a socially determined form of the activity of man creating an object in one form or another, is engendered and exists not in the head but with the help of the head in the real objective activity (activity on things) of man as the active agent of social production. (Ilyenkov, 1977, p. 261)
Importantly, the human head is not seen as a purely physical entity, but as a social head, full of socially constructed information on how to understand and act in the world, for example, language, concepts and categories. The focus on conscious human activity (production in its broadest sense), essential to the materialist conception of dialectics, re-introduces Hegel's problematic, but sets it on a materialist footing - effectively turning Hegel on his head, as Marx put it. The problem resides in the gap between socially constructed phenomena as they exist in the real world, and the equally socially constructed representations of those phenomena in the consciousness of the individual. The key to this problem is the continuing process of change over time. First, the world in which we find ourselves is in a constant state of flux, requiring us to constantly act upon it, and constantly represent that process in consciousness. Unfortunately, as we know from Hegel, the phenomenal forms through which we grasp reality are constructed over generations, and may therefore present themselves to the individual as static and immutable. The application of dialectical logic enables us to recognise the historical specificity and social construction of prevailing phenomenal forms, in order that we can act to consciously transform them, and better satisfy our needs and wants.
The dialectical logic revealed by Marx reflected the largely unconscious technique used by other scientists, for example, Darwin had been able to reveal an organic world in a constant state of flux, where minute quantitative changes over time led to the qualitative transformation from one species to another. That we bother to assign fixed labels to things which are constantly changing is a matter of practicality rather than precision - we name things in order to understand them, so that we might use them to meet our needs. However, the fixity of meaning implied in the act of naming cannot keep up with a world that is constantly changing, leading us to continually revise our knowledge of the world. In this sense a body of knowledge is always historically specific. This does not imply the adoption of a relativist epistemology, because at any given moment some truth claims will grasp reality more adequately than others. The importance of dialectical logic is that it enables us to choose between alternative truth claims without losing sight of their historical specificity and transitoriness.
The adoption of dialectical logic has a series of methodological consequences for critical social research. First, it is essential to study the historical development of a phenomenon to reveal changes in the way it has been conceptualised over time. The purpose of studying a phenomenon over time is not simply to record changes in its appearance or phenomenal form, but to reveal the nature of the relationship between the phenomenon's appearance and its underlying essence. We noted above that the production of knowledge involves abstraction from the material world to the theoretical world, in order to better inform our practical activity. The dialectical approach problematises this relationship between objective reality and our attempts to represent it in knowledge. Part of the problem is that objective reality is in a constant state of flux and our attempts to grasp it through categorisation and definition must inevitably become out-dated or inadequate over time. The purpose of studying a phenomenon over time is, therefore, to reveal the historical specificity of phenomenal forms and the extent to which they are socially constructed.
The relationship between essence and appearance is not only problematic because phenomenal forms become outdated in the face of constant changes in the material world, but also because the historically specific categories through which we grasp the material world also have a political dimension in that they enable powerful groups to exercise domination over less powerful groups. Thus, the second element of social critique is the deconstruction of categories or phenomenal forms. This does not simply entail the production of a detailed description of the material contents of a given category, but an attempt to reveal the extent to which the existence of a category depends upon a series of relationships with other phenomena in the social and economic totality. For example, an uncritical definition of the category 'working class' might produce a list of occupations, or an income band, or cultural characteristics such as educational attainment or 'lifestyle choices'. A critical account, by contrast, would attempt to locate the category in a series of social and economic relations.
Viewing a phenomenon as the product of a whole network of social and economic relations, does not mean that all of those social relations carry equal explanatory potential. It may be possible to find a key relationship or 'overriding moment'; for example, with the category 'working class' the key relationship might be that workers are obliged to sell their labour power to the private owners of capital, thereby forfeiting their right to exercise conscious control over the production and distribution of goods and services. This process of deconstructing a phenomenal form or explanatory category is fundamentally empirical; it is grounded in observation of phenomena in the real world at a specific location in time and space, and cannot be dogmatically generalised to a different context.
Critique of categories has a series of effects. First, it shifts explanatory emphasis from the categories themselves to the social relations that underpin them. This makes the categories derived from such analysis more enduring over time. For example, if the working class is defined in terms of particular occupations or cultural attributes, then the term must inevitably become redundant as the labour market changes and different cultural patterns emerge. For instance, it is often remarked that the working class has declined as a result of the re-structuring of manufacturing industry (Gorz, 1982), however, if the category is defined in terms of the relationship between labour and capital, then the category continues to be of value while ever that relationship endures, even though its demographic content may change over time.
A second effect of deconstruction is that in laying bare the essence of a phenomenon by locating its conditions of existence in a specific network of social and economic relations it also reveals political factors that cannot be grasped from its surface appearance. The classic example is Marx's critique of the phenomenal forms of bourgeois political economy (Marx, 1887/1983), which revealed the essentially exploitative and coercive relations that lie behind the apparent freedom and equity of commodity production. The hidden political nature of phenomenal forms is not only to be found in class relations, but anywhere that knowledge is implicated in the domination of one group by another, for example, around gender and race. The objective of critical social research is to make such oppressive structures overt in order that they might be challenged.
This relationship between critical social research and political activity is two-fold. First, for the critical social researcher the distinction between the material world and its abstract representation in knowledge is not absolute. As well as having an existence as text or as ideas, knowledge is also externalised through our conscious manipulation of material objects, for example, even a very basic manufactured object like a spoon has more than a purely physical existence; as well as being a piece of metal it is also a consciously designed instrument that can be used in a specific way to solve a specific problem. This process of externalising conscious knowledge to give it a material existence extends beyond the production of basic tools to include the built environment, social and economic institutions, et cetera. Thus, the production of an alternative body of knowledge entails much more than abstract re-conceptualisation of the material world, or the production of a text, it entails the transformation of the phenomenal forms of knowledge, that is the material objects, institutions and processes in which previous knowledge is embedded. As Lee Harvey has noted:
Knowledge changes not simply as a result of reflection but as a result of activity too. Knowledge changes as a result of praxis. Similarly what we know informs praxis. Knowledge is dynamic, not because we uncover more grains of sand for the bucket but because of a process of fundamental reconceptualisation which is only possible as a result of direct engagement with the processes and structures which generate knowledge. (Harvey, 1990, p. 23)
This unity between the subjective and objective world underlines the political basis of critical social research. The point is not simply to reveal the oppressive aspects of existing phenomenal forms as an end in itself, but to embed this knowledge in the consciousness of the oppressed in order that they might engage in practical activity to emancipate themselves. Füredi suggests that:
The power of Marxist theory is derived, not from the elegance of its arguments, but from its capacity to make conscious the unconscious forces driving towards social change. (Füredi, 1990, p. xxiii)
We noted above that the relationship between theory and practice is not uni-linear. Whilst the objective of critical social research is to inform conscious activity, it also derives its validity from active involvement in political struggle. From this perspective the production of knowledge is deeply embedded in the process of social transformation; both informing, and derived from, the struggle to consciously change the material world.
To summarise, although critical social research is diverse and constantly developing the following characteristics are essential to the approach: the application of dialectical logic which views the material and social world as in a constant state of flux; the study of phenomena over time to reveal their historical specificity; the critique or deconstruction of existing phenomenal forms and analytical categories that delves beneath the superficial appearances available to unaided common sense to reveal the network of social and economic relations that are the essential conditions of existence for a phenomenon; the exposure of previously hidden oppressive structures; and a praxiological orientation in which knowledge is considered to be inseparable from conscious practical activity. But the question still remains as to whether this approach can be synthesised with ethnography and still constitute a valid methodology.
Not all critical ethnographers would agree with the above definition of social critique, although, most would presumably agree that critical ethnography entails synthesising the subjective testimony of informants with a broader historical and structural analysis. But this synthesis of the insights that traditional ethnography provides into the subjective experience of everyday life, with the historical and structural insights offered by social critique, is by no means easy to maintain. Not least because it entails combining two quite separate and possibly incompatible formulations of validity. The possibility always remains that the analysis will slide into either a top-down deductive approach in which a pre-existing theory is simply legitimated by the selective and biased use of ethnographic data, or else into a superficial and particularistic account of the views of respondents. Giving explanatory primacy to the testimony of informants would appear to undermine the validity of social critique by contravening the injunction to always look beneath the surface of everyday appearances; whilst a broader historical and structural analysis might contradict the traditional ethnographer's claim that valid research does not impose a priori theoretical constructs.
The solution to this dilemma lies in ensuring that the analysis is informed by both strands of inquiry, for example, that issues emerging from participant observation or ethnographic data can be placed in an historical and structural context, and that problems identified in the academic literature can influence the direction of the ethnographic study. As such, critical ethnography entails a constant inter-weaving of inductive and deductive logic. The researcher does not set out to test a pre-conceived hypothesis, nor is an entirely open-ended approach adopted, instead the researcher begins by observing the field of study, both as a participant observer and as a reviewer of academic literature. From the synthesis of these sources a research agenda emerges that can be pursued, again, by a mixture of observation and theoretical work.
The key to managing this unstable dialectical relationship between ethnographic observation and social critique is to re-conceptualise validity in terms of reflexive practice. Reflexivity refers to the researcher's conscious self-understanding of the research process (Hammersley & Atkinson, 1983), or more specifically, to a sceptical approach to the testimony of respondents (i.e., Are they telling me what I want to hear?), and to the development of theoretical schema (i.e., Am I seeing what I want to see?). The purpose of reflexivity is not to produce an objective or value-free account of the phenomenon, because qualitative research of this kind does not yield standardised results, as Janet Ward-Schofield has suggested:
...at the heart of the qualitative approach is the assumption that a piece of qualitative research is very much influenced by the researcher's individual attributes and perspectives. The goal is not to produce a standardised set of results that any other careful researcher in the same situation or studying the same issues would have produced. Rather it is to produce a coherent and illuminating description of and perspective on a situation that is based on and consistent with detailed study of the situation. (Ward-Schofield, 1993, p. 202)
Thus, reflexivity is not primarily a means of demonstrating the validity of research to an audience, but rather a personal strategy by which the researcher can manage the analytical oscillation between observation and theory in a way which is valid to him or herself. Of course, this will be anathema to the positivist. But is it really so different to the process of establishing validity in quantitative research? Random sampling and statistical testing may appear to make the assessment of validity transparent to a third party, but such techniques are not immune to manipulation by an unscrupulous researcher. In fact, the validity of particular research findings, be they qualitative or quantitative, ultimately depends upon trust in the researchers integrity, at least until the research is replicated. Validity therefore refers to the techniques employed by the researcher to indulge a Socratic distaste for self-deception, and in critical ethnography this is achieved by reflexivity.
This reflexive management of the research process in the pursuit of validity applies to each stage of the research process, from establishing relations in the field to writing up the conclusions. Rather than presenting a taxonomy of qualitative techniques, participant observation and in-depth interviewing are presented as the main exemplars of the ethnographic approach.
Choosing an appropriate site to study, and forging a relationship with its participant members, is a key issue for all ethnographic studies. Ward-Schofield (1993) has explored the consequences of site selection for validity and generalisability, and suggests that both can be maximised either by selecting a 'typical' site or else conducting a multi-site study. However, there are problems with either option; first, how can one identify what constitutes a typical site without conducting at least a basic reconnaissance of all potential sites? Secondly, given that most qualitative studies are conducted by single researchers or small teams, is it viable to study multiple sites to the depth required by qualitative analysis?
Other commentators set less rigorous standards for site selection, for example, Jorgensen (1989) refers to a number of qualitative studies derived from everyday life experiences, where the researcher literally found him or herself in a fortuitous location to study a phenomenon. There is more to recommend Jorgensen's argument than the dictates of pragmatism; rather it constitutes a refusal to be bound by quantitative criteria of reliability, particularly the claim that the informants in a qualitative study should be representative of a broader population. Clearly, if the need to identify a representative site is accepted, then qualitative research must always appear to be the poor relation of quantitative methods where random sampling can be applied. By contrast the ethnographer is more concerned with the validity of the data she or he collects, that is, with whether or not the data express the considered and authentic views of the informant, with minimal interference or distortion by the research process. It is this criteria of validity (i.e., the potential to access the authentic views of the informants) that guides the ethnographer's selection of a site, rather than the largely unattainable goal of representativeness. Again this process needs to be reflexively managed according to the specific context, potential considerations include: ease of access to the informants, whether data can be adequately recorded (either by field notes or tape-recorder), and crucially, whether there are any characteristics of the site that might adversely influence an informant's testimony, for example, the close proximity of his/her employer or spouse. As the context for qualitative research is infinitely variable the characteristics of an ideal site cannot be prescribed in advance; hence the need for reflexive management by the researcher.
Having chosen a site, the researcher must establish contact with the informants. Again the character of this relationship varies, for example, from the relatively brief contact of a single in-depth interview, to participant observation which might entail a close relationship between researcher and informants lasting several weeks or months. Again, quantitative criteria of validity cannot be imposed on this aspect of qualitative research. Although the ethnographer must take care not to influence the informants in ways which might distort their behaviour or testimony, the necessity of establishing a rapport in order to gain access to what might be sensitive or closely guarded information means that the degree of detachment associated with quantitative research may not be viable or desirable. Instead, close proximity to the phenomenon being studied can be viewed as an advantage, for example, some feminist writers (Stanley, 1990) have gone so far as suggesting that personal experience can be a vital source of inside information to be tapped as part of the research process. Nigel Fielding has also suggested that there may be advantages as well as disadvantages:
One is participating in order to get detailed data, not to provide the group with a new member. One must maintain a certain detachment in order to take that data and interpret it [sic]). But it is also important to note that another problem is much less remarked in the literature, though it may be more common. This is the problem of 'not getting close enough', of adopting an approach which is too superficial and which merely provides a veneer of plausibility for an analysis to which the researcher is obviously committed. (Fielding, 1993, p. 158)
Managing the relationship with informants, or 'front end management', is an important aspect of the validity of any qualitative study, but again it cannot be prescribed as a specific procedure, and its adequacy or effectiveness is unlikely to be immediately transparent to a third party. Whilst a conscious and reflexive attempt to establish valid relations in the field can be reported by the researcher, and may go some way to reassuring reviewers and readers, any post-festum assessment of the validity of this process is ultimately dependent upon trust.
Qualitative researchers have a number of techniques at their disposal for data collection, including non-participant and participant observation, focus groups, and in-depth interviewing. Each of these techniques raises particular concerns about validity, for example, participant observers may have difficulty in taking adequate field notes. However, there are techniques for overcoming such difficulties, and as with site selection and front-end management the greater problem resides in demonstrating validity rather than achieving it.
As noted above, a more fundamental difficulty arises in reconciling the 'bottom-up' approach of qualitative research with the structural and historical perspective adopted in critical social inquiry. For example, for many ethnographers the in-depth interview should be entirely open ended, with at most a series of topics to be discussed, but certainly no pre-conceived questions, as this would entail the researcher imposing his/her own definition of the situation rather than enabling the respondents to structure the research. Thus, Sue Jones suggests that pre-conceived interview schedules are unacceptable because:
...the interviewers have already predicted, in detail, what is relevant and meaningful to their respondents about the research topic; and in doing this they have significantly prestructured the direction of enquiry within their own frame of reference in ways that give little time and space for their respondents to elaborate their own. (Jones, 1985, p. 46)
Whilst this may be a valid approach for traditional ethnography, or where the interview is the researcher's first contact with the respondent, critical ethnography may entail a much more focused approach to interviewing, in which questions are asked about specific issues derived from the broader social critique. This is particularly the case where extensive participant observation has already revealed issues to be examined in greater depth in the interviews. Whilst critical ethnographers are keen not to ask leading questions and to enable informants to express their views fully, the research agenda and scope of the study are not primarily determined by the informants. Rather, a dialectical approach is adopted, allowing the researcher to oscillate between the world view of the informant, (e.g., by departing from the interview schedule to pursue an interesting line of inquiry), and the insights offered by the historical and structural analysis, which may enable the constructs and categories employed by the informant to be actively deconstructed during the course of the interview.
The validity of this tendency to structure the research agenda according to themes raised by social critique is more likely to be questioned by traditional ethnographers than by quantitative researchers who are more or less obliged to commence data collection with a pre-conceived agenda. For the critical ethnographer validity depends upon getting beneath the surface appearances of everyday life to reveal the extent to which they are constituted by ideology or discourse. Thus, rather than commencing the process of data collection with an 'empty head' the critical ethnographer is pre-armed with insights gleaned from social critique.
Nigel Fielding (1993, p. 163) has summarised a common approach to ethnographic data analysis in the following model:
or cut up
Harvey (1990) refers to the same process as 'pile-building', in which the ethnographic data are first read 'vertically', usually in chronological order, to identify common themes and relations which are then coded. The data are then literally cut up and re-ordered into 'piles' that reflect the key themes. Specialist software is available to facilitate this process, but the 'cut and paste' utility of a word-processor is likely to be adequate. The re-ordered data are then re-read, enabling a sequential argument to be constructed, and illustrative quotations from the transcripts to be selected.
Harvey also suggests that critical ethnography differs from traditional forms of qualitative data analysis, by bringing the broader critique of social relations to bear on the structuring of analytical themes. Hence the final analysis is not derived exclusively from the ethnographic data but from an oscillation between that and the social critique. As Sue Jones has noted, this inevitably entails going beyond the concepts and understandings of the respondents:
I know I cannot empathise with the research participants completely. I also know that I am likely at some points to set my understanding of their 'concrete' concepts - those which they use to organise, interpret, and construct their own world - within my own and/or an audience's concepts and frameworks that are different from theirs. When I do this, however, I try to be clear that I am doing this and why, and to ensure that this 'second level' of meaning retains some link with the constructions of the research participants. (Jones, 1985, pp. 56-57)
The identification of themes and the selection of quotations to illustrate them also raises a fundamental issue about the validity of qualitative research; as David Silverman has noted:
The various forms of ethnography, through which attempts are made to describe social processes, share a single defect. The critical reader is forced to ponder whether the researcher has selected only those fragments of data which support his argument. (Silverman, 1985, p. 140)
Silverman's preferred solution to this dilemma is to introduce simple counting procedures into the analysis, for example to identify how many people referred to a specific theme. But this begs the question of how many people must refer to a theme before it is deemed significant; moreover with a small quantitative study it is unlikely that a view shared by a majority of respondents could be claimed as representative of the views of a broader population. In fact, the application of quantitative criteria of validity to qualitative data is inappropriate. The rationale for conducting in-depth interviews is that people involved in a phenomenon may have insights that would not otherwise be available to the researcher, and it is the quality of the insight that is important, rather than the number of respondents that share it. This is what Hammersley meant (in the above quotation) when he suggested that ethnographic data should be treated in the same manner as social scientific accounts - when we quote the work of a particular social scientist we do so because of its explanatory power, not because it represents a commonly held view, and the same logic applies to qualitative data.
Mays and Pope (1995) have addressed the validity issue by recommending that qualitative researchers pass their ethnographic data to independent researchers to see if they arrive at the same analysis. But again this is unsatisfactory. How are we to select an alternative researcher, and how many should we consult before we can conclude that the analysis is valid? Again the mistake lies in applying quantitative criteria of validity to qualitative data. Whilst two statisticians applying the same test of statistical significance to the same quantitative data-set might stand a good chance of arriving at the same result, it is extremely unlikely that two ethnographers would produce the same reading of a case study, for the same reasons that a group of students reading the same text books would be unlikely to produce the same essay.
As noted above, the researcher can influence the validity of an ethnographic analysis by adopting a reflexive perspective on his or her work, the problem resides in demonstrating this validity to the reader. Whilst the quantitative researcher may have more convincing ways of demonstrating validity, such as, random sampling and statistical inference, even here there remains the need for the reader to trust in the integrity of the researcher not to knowingly engage in deception, and this is doubly true of qualitative research. Beyond this, the qualitative ethnographer can strive to demonstrate the validity of the analysis by providing a 'thick' description of the case study, and including sufficient ethnographic data for an alternative reading to be constructed.
In many accounts of the research process 'writing up' receives little attention, or else it is treated as an independent activity from data analysis. Hammersley and Atkinson (1983) however, claim that writing-up is inevitably a part of the analytical process, suggesting that the structure of the report influences the type of analysis, or at least the way it is understood by the reader. They identify four types of report: the 'natural history' (in which the report reflects the different stages of the research process as they progressed over time), the 'chronology' (also temporally organised, but reflecting the development or 'career' of the phenomenon being studied, rather than the research process), 'narrowing and expanding the focus' (in which the analysis moves backwards and forwards between specific observation and consideration of broader structural issues), and 'separating narration and analysis' (in which the ethnographic data are presented first before theoretical issues are addressed).
The 'natural history' approach is a very common format, particularly for presenting quantitative findings, but it does not fit well with qualitative research. In fact it can lead to a form of dishonesty, giving the impression that that the research followed a tight structure of background reading, hypothesis construction, research design, data collection and analysis, and discussion of results. The qualitative research process is less well ordered. Background reading is essential, but which texts are relevant, and therefore, worth including in a report or publication, only becomes apparent towards the end of the research process, and the literature review should continue throughout the project as the ethnography raises new themes for analysis. Similarly, as the above discussion demonstrates, the sequence of hypothesis - data collection - analysis, is not clear cut or linear, but an ongoing and dialectical process.
The other three formats for writing-up are more relevant, but should be seen as different aspects of the process, rather than discrete types. Whilst the organisation of a report or publication cannot in itself confer validity, it can make the research process more transparent to the reader and allow validity to be more clearly assessed. It is important to use the report format to illustrate the oscillation between micro and macro analysis that comes from combining the methodologies of ethnography and critical social research; looking in detail at the informants' testimony, but broadening this out to a consideration of structural and historical issues.
An important question to ask at the end of this discussion of methodology is the extent to which the methods employed in critical ethnography enable research findings to be generalised to other situations. In a quantitative study generalisability is largely determined by random sampling and statistical inference, obviously such techniques are not usually relevant to qualitative research, making generalisation more of a problem. In many respects, the way in which generalisation is conceptualised in quantitative studies is alien to both ethnography and critical social research. For the ethnographer what matters most is gaining an in-depth understanding of the attitudes, beliefs and behaviour of the people s/he studies; the assumption is that this worldview will be context specific, and that generalisation to others will therefore be extremely limited. Similarly, critical social research starts from the assumption that society is in a constant state of flux, that the social world and our understanding of it are constantly changing, again limiting the value of generalisation.
However, although ethnography and critical research may question positivist/quantitative assumptions about generalisability, both approaches aim to produce findings that have relevance beyond the immediate context of the study. Whilst the production of laws of behaviour is eschewed, there remains an often almost hidden claim that the behaviour found in the study will shed some light on the behaviour of others, even if this explanatory range is limited in time and space. As Janet Ward-Schofield (1993) has suggested this claim entails a re-conceptualisation of generalisability in terms appropriate to qualitative research. She prefers the terms 'fittingness', 'comparability', or 'translatability', reflecting the process of detailed description of the content and context of a study, so that it can be generalised to examples that match it closely.
The use of 'thick description' to boost the generalisability of a qualitative study, is important, but generalisability depends not just upon detailed description of a phenomenon, but on revealing the social relations that underpin it. For instance, a qualitative study might reveal that 'patient empowerment' is an expression of a specific set of relations between the state, healthcare professionals and their clients, indeed that these relations are the condition of existence for patient empowerment.
It might also be concluded that these relations limit the emancipatory potential of patient empowerment, and that this limitation will apply wherever those conditions prevail. To move beyond those conditions would be to move beyond something that was recognisable as patient empowerment.
Conceptualising a phenomenon in terms of its conditions of existence and the social relations that characterise it, is a sounder basis for generalisation than the simple description of immediate appearances. For example, the number of people involved in a patient empowerment initiative might change, as might the designation of the healthcare professionals, the geographical location, even its claimed aims and objectives, but whilst the conditions of existence and social relations that characterise the phenomenon remain constant, then the conclusion that its emancipatory potential is extremely limited would remain valid. By the same token, a patient empowerment initiative might retain its outward appearance, in terms of the people participating, its geographical location, et cetera, but be transformed into a completely different phenomenon (with a different degree of emancipatory potential), if change occurred in the social relations that underpinned it. In either instance, defining the phenomenon in terms of social relations reveals whether or not generalisation is valid.
Over the last decade qualitative research has gained a new respectability in the sociology of health, even in the more conservative sub-discipline of medical sociology. But legitimacy has been bought at the cost of embracing positivist criteria of reliability and validity. This compromise has effectively burned the bridge between ethnographic techniques of data collection and their origins in phenomenological thought. One important consequence of transforming qualitative research into another weapon in the positivist arsenal has been to rob the approach of its critical potential. Rather than being a means by which researchers and informants can jointly travel from the experiential reality of everyday life towards a critical understanding of discourse formation and the exercise of power, qualitative research is rapidly becoming a mechanism of surveillance; an instrument for measuring subordination to discourse instead of a means of countering it.
This transformation is legitimated by the bogus belief that positivist criteria of validity confer a degree of authenticity upon research findings that is immediately transparent to a third party. However, even the powerful tests of validity which are available to the quantitative researcher, such as random sampling or statistical inference, are not immune to manipulation by disreputable researchers. Rather than a clearly discernible hallmark of authenticity, the techniques employed in the pursuit of validity comprise a means by which the researcher can minimise the risk of self-deception. Whilst these techniques can be reported, their acceptance by a third party must ultimately entail a degree of trust in the diligence and integrity of the researcher. This does not mean that all research findings should be blindly accepted at face value; the onus is always on the researcher to persuade his or her audience that the research findings are valid. The appropriate perspective for the reader of research should be, to borrow from Antonio Gramsci, 'pessimism of the intellect, optimism of the will', and this applies equally to quantitative and qualitative research.
The importation of positivist criteria of validity into the qualitative research process is not only unjustified on the grounds of scientificity, it is also grossly inappropriate for the type of knowledge produced by such a perspective. The aim of the qualitative researcher is not to produce a representative and unbiased measurement of the views of a population, but to deepen his or her understanding of a social phenomenon by conducting an in-depth and sensitive analysis of the articulated consciousness of actors involved in that phenomenon. Interview transcripts and field notes are read by the ethnographer, for the same purpose that academic texts are also considered, that is, in the hope of finding fresh insights and new ways of understanding a particular phenomenon. Positivist criteria of validity are quite inappropriate to this process, their imposition reduces ethnography to the status of a 'poor relation' of quantitative research - a means of gathering sensitive information, but ultimately less valid.
I have argued in this paper for a re-conceptualisation of rigour and validity in qualitative research, based on a rejection of positivist criteria, in favour of the insights offered by social critique. From this perspective validity can be re-couched in terms of reflexively managing the relationship between the testimony of informants and a broader process of structural and historical analysis. This is an uneasy and in some senses contradictory combination that requires careful management at each stage of the research process. However, it does provide an opportunity to get beneath the surface of everyday appearances, to produce theoretically informed accounts of social phenomena that are grounded in people's experience of everyday life, but which take a critical approach to the categories and forms through which everyday life is experienced.
Downes, D., & Rock, P. (1982). Understanding deviance: A guide to the sociology of crime and rule breaking. Oxford: Oxford University Press.
Fielding, N. (1993). Ethnography. In N. Gilbert (Ed.), Researching social life (pp. 154-171). London: Sage.
Füredi, F. (1990). Introduction. In F. Jakubowski, Ideology and superstructure in historical materialism (pp. vii-xxxi). London: Pluto.
Gorz, A. (1982). Farewell to the working class. London: Pluto Press.
Hammersley, M. (1992). What's wrong with ethnography? London: Routledge.
Hammersley, M., & Atkinson, P. (1983). Ethnography: Principles in practice. London: Tavistock Publications.
Harvey, L. (1990). Critical social research. London: Unwin Hyman.
Ilyenkov, E. V., (1977). Dialectical logic: Essays in its history and theory (H. Campbell Creighton, Trans.). Moscow: Progress Publishers.
Jones, S. (1985). The analysis of depth interviews. In R. Walker (Ed.), Applied qualitative research (pp. 56-70). Aldershot (Hants.): Gower.
Jorgensen, D. L. (1989). Participant observation: A methodology for human studies. London: Sage.
Marx, K. (1983). Capital (Vol. 1). London: Lawrence & Wishart. (Original work published 1887)
Mays, N., & Pope, C. (1995). Rigour and qualitative research. British Medical Journal, 311, 109-112.
Silverman, D. (1985). Qualitative methodology and sociology. Aldershot (Hants.): Gower.
Stanley, L. (Ed.). (1990). Feminist praxis: Research, theory and epistemology in feminist sociology. London: Routledge.
Ward-Schofield, J. (1993). Increasing the generalisability of qualitative research. In M. Hammersley (Ed.), Social research: Philosophy, politics & practice (pp. 200-225). London: Open University/Sage.
+David Wainwright, Ph.D. is a Research Fellow at the Centre for Health Services Studies, University of Kent at Canterbury. Prior to this he worked for a London health authority. He received his B.A. in Sociology from Kent University in 1988, followed by an M.Sc. in Social Policy & Planning at the London School of Economics, and a Ph.D. in Sociology also from the University of Kent. Dr. Wainwright is currently conducting research into the health consequences of job insecurity. Other interests include: the medicalisation of social problems, the politics of health promotion, and the construction of professional self-identity. Recent publications include: 'The Political Transformation of the Health Inequalities Debate, Critical Social Policy, 16(4), 67-82 (1996), and 'The Best of Both Worlds', The Health Service Journal, 106(5487), 30-31, (1996). His email address is D.M.Wainwright@ukc.ac.uk.
Wainwright, D. (1997, July). Can sociological research be qualitative, critical and valid? [69 paragraphs]. The Qualitative Report [On-line serial], 3(2). Available: http://www.nova.edu/ssss/QR/QR3-2/wain.html
June 2, 1997 copyright