Reflexive Methodology

New Vistas for Qualitative Research

Mats Alvesson, Kaj Sköldberg

• Publisher: SAGE Publications Ltd
• Year: 2017
• ISBN:
  9781526428851
• Edition: 3
• Topic: Social Sciences
• Subtopic: Social Science Research & Methodology
• Index: Social Sciences

Praise for the Second Edition: "In opposition to most literature on how to conduct good social science research which is either empirically oriented or gives priority to theoretical and philosophical considerations, which tends to make empirical research look odd or irrelevant, this volume on ?Reflexive Methodology? explicitly turns towards a considerationof the perceptual, cognitive, theoretical, linguistic, political and cultural circumstances as backdrop of data interpretation and research design. It showed up to be the most important and informative resource and a source of enlightenmentto my lecture on methodology at our institute. I can highly recommend the volume to lecturers and students alike."

Professor Sabine Troeger, Geography Institute - Library, University of Bonn
Reflexivity is an essential part of the research process. Mats Alvesson and Kaj Sköldberg make explicit the links between techniques used in empirical research and different research traditions, giving a theoretically informed approach to qualitative research. The authors provide balanced reviews and critiques of the major schools of grounded theory, ethnography, hermeneutics, critical theory, postmodernism and poststructuralism, discourse analysis, genealogy and feminism. Useful reading for students and researchers across the social sciences.

The first edition established itself as a ground-breaking success, providing researchers with an invaluable guide to a central problem in research methodology - namely, how to put field research and interpretations in perspective, paying attention to the interpretive, political and rhetorical nature of empirical research. The second edition introduced a new chapter on positivism, social constructionism and critical realism, and offered new conclusions on the applications of methodology.

This third edition of Reflexive Methodology provides further updates on new research, including neorealism, and illustrations and applications of reflexive methodology in formulating research strategies, that build on the acclaimed and successful previous editions

1 Introduction: The Intellectualization of Method

Traditionally research has been conceived as the creation of true, objective knowledge, following a scientific method. From what appears or is presented as data, facts, the unequivocal imprints of ‘reality’, it is possible to acquire a reasonably adequate basis for empirically grounded conclusions and, as a next step, for generalizations and theory-building. So the matter has long been conceived, and no doubt many empirically oriented researchers in the social sciences still conceive it so, irrespective of whether they are examining ‘objective reality’ (social facts), exploring people’s subjective or intersubjective experiential worlds (meanings), or analysing discourse (text).

This view has been subjected to a good deal of criticism, however, much of which appeared towards the end of the 1960s and was directed against ‘positivism’. But there has since been further criticism applying also to diverse variants of the qualitative method, sometimes automatically seen as ‘anti-’ or at least as ‘non-positivist’. For the moment we will mention only such critique that emphasizes the ambiguous, unstable and context-dependent character of language, the dependence of both observations and data on interpretation and theory (interpretation-free, theory-neutral facts do not, in principle, exist), and the political–ideological character of the social sciences. One line of argument here starts from the notion that knowledge cannot be separated from the knower (Steedman, 1991: 53). Data and facts, as we will demonstrate, are the constructions or results of interpretation: we have to do something with our sensory impressions if these are to be comprehensible and meaningful. Alongside this general critique of the objectivist scientific view and the heavy focus on empirical data, more specific criticism is raised about various methodological conceptions and methods. The methodological conceptions and methods of the social sciences have been exposed to such a barrage of objections that one might have expected empiricists to lose their self-confidence and consider turning to some other branch of study instead.¹ But in fact the big risk seems to be that practising researchers stick in the same old rut, either repressing the criticism altogether or remaining more or less unaware of it. Many of the critics, on the other hand, tend to go to the opposite extreme and cut out empirical reality altogether – although exactly how they do this depends on their particular scientific orientation. Not infrequently they study something that carefully avoids statements about anything other than narrative or discourse or social constructions in interviews, thereby ducking many if not most important and interesting aspects of social reality.

In the social sciences – to which we largely limit ourselves in this book – there is a clear division between the great mainstream of empirically oriented research and various currents that are critical of ‘empiricism’ on diverse philosophical or theoretical grounds. To some extent this division overlaps with the dichotomy between scholars who adopt a robust and objectivist ontological approach and those with a consciousness- and experience-oriented, interpretive view of ontology and epistemology (Burrell and Morgan, 1979). But there are certain differences, since some representatives of this second approach are drawn towards the empiricist line – for example, some phenomenologists and other advocates of rigorous qualitative method assume that the very stringency of the method guarantees good research results. The critics of empiricism – ranging from historians of science, sociologists of knowledge, psychologists of science and linguistic scholars to ideological critics and philosophers – claim that culture, language, selective perception, subjective forms of cognition, social conventions, politics, ideology, power and narration all, in a complicated way, permeate scientific activity. These elements leave their mark on the relation between empirical reality and/or attempts to force segments of reality into the research texts, so that the relation between ‘reality’ and ‘text’ (the research results) is at best uncertain and at worst arbitrary or even non-existent. To find support for this thesis we need only consider that, despite the wealth of different theories that exist in most fields in the social sciences, empirical results are generally found to ‘agree’ – at least in part – with the researcher’s own premises, and that most researchers seem disinclined to change their point of view simply because a researcher with another theoretical base has presented empirical ‘data’ which contradict their own point of view.

A variety of ideas about how social reality is constructed – not just how it is represented – by the researcher will be investigated in this book. We believe, with the ‘anti-empiricists’, that empirical social science is very much less certain and more problematic than common sense or conventional methodological textbooks would have us think. The great array of books on the ‘qualitative method’ does not differ decisively from the quantitative literature on this score. Nor as a rule are the former ‘qualitative enough’, in the sense of being sufficiently open to the ambiguity of empirical material and the complexity of interpretations. The focus on procedures and techniques implies an imitation of the quantitative methodology textbooks, and draws attention away from fundamental problems associated with such things as the role of language, interpretation and selectivity in research work, thus underrating the need for reflection. On the other hand, there are also certain risks involved in a too strong emphasis on this need. By problematizing research, we may come to overrate its difficulties, which leads in the long run to a defeatist reaction, and perhaps even to asking ourselves whether empirical social science has any reasonable function at all.

But we do not give up so easily, despite our ambition to take account of doubts about the ability of empirical material (data) to provide crucial input into research. We are not convinced that the opposite pole to methodological textbook wisdom – where it is claimed in a spirit of postmodernism or poststructuralism, for instance, that empirical reality can be ignored altogether – is in any way preferable. Nor is the phobia of empirical matters that characterizes much hermeneutic and critical theory to be recommended. It is our experience that the study of a confusing and contradictory, but often surprising and inspiring, empirical material has much to offer. It is precisely this combination of inspiration from the philosophy of science and empirical interests that provides this book with its raison d’être and, we believe, makes it unique. Most of the literature in the relevant field – broadly defined as ideas about how to conduct good social science research – is either empirically oriented or gives unequivocal priority to theoretical and philosophical considerations, which tends to make empirical research look odd, irrelevant, naive or even feeble-minded. We try instead to manoeuvre between these two conventional – and safe – positions, which appear to us rather as a kind of methodological Scylla and Charybdis.

In our dealings with empiricism – broadly defined here as all research in which ‘pure data’ or uninterpreted ‘facts’ are the solid bedrock of research – we try to take account of the objections which have been raised by hermeneuticians, critical theorists, poststructuralists, linguistic philosophers, discourse analysts, feminists, constructivists, reflectivists and other troublemakers who render life difficult for the supporters of either quantitative or mainstream qualitative methods. Against these troublemakers – who explicitly or implicitly leave their readers despairing and irresolute vis-à-vis empirical research – we stubbornly claim that it is pragmatically fruitful to assume the existence of a reality beyond the researcher’s egocentricity and the ethnocentricity of the research community (paradigms, consciousness, text, rhetorical manoeuvring), and that we as researchers should be able to say something insightful about this reality. This claim is consistent with a belief that social reality is not external to the consciousness and language of people – members of a society as well as researchers (who, of course, also are members of a society).

Before proceeding with our distinct approach to methodology, we relate it to and ground it in a broadly accepted thesis in philosophy of science: that how we interpret phenomena is always perspectival and that so-called facts are always theory-laden.

Ways of explanation and understanding

In explanatory models, it is usual to distinguish between induction and deduction.² An inductive approach proceeds from a number of single cases and assumes that a connection that has been observed in all these is also generally valid. This approach thus involves a risky leap from a collection of single facts to a general truth. Consider, for example, ‘there have never been any rocks on the bottom so far when I have dived into the water; therefore there are probably not any this time either’. The weakness is, it appears, that the underlying structure or situation is not included in the picture, but only a mechanical, external connection. The method, as it were, distils a general rule from a set of observations; what comes out then becomes merely a concentrate of what is already included in the observations themselves.

A deductive approach, on the contrary, proceeds from a general rule and asserts that this rule explains a single case. This approach is less risky – at the price of seeming to presuppose what is to be explained: that the general rule always holds true, hence also in the current case. Moreover, it does not really appear to explain anything, but rather avoids explanation through authoritarian statements, rather as a parent under stress might answer an inquisitive child: ‘Why do butterflies have wings?’ ‘Because all butterflies have wings, dear.’ Thus, in deduction, too, we see a lack of underlying patterns and tendencies, which makes the model flat, bordering on the empty.

These two models are usually regarded as exclusive alternatives, but it would be difficult to force all research into them, if they are not to serve as a Procrustean bed. There are in fact other possibilities, and we will now present one of them.

Abduction is probably the method used in real practice in many case-study-based research processes (Sköldberg 1991). It has also been recommended more generally as an innovative approach to theory-driven empirical research (Meyer and Lunnay 2013). In abduction, an (often surprising) single case is interpreted from a hypothetic overarching pattern, which, if it were true, would explain the case in question. The interpretation should then be strengthened by new observations (new cases). The method has some characteristics of both induction and deduction, but it is very important to keep in mind that abduction neither formally (see note 5) nor informally is any simple ‘mix’ of these, nor can it be reduced to these; it adds new, specific elements. During the process, the empirical area of application is successively developed, and the theory (the proposed overarching pattern) is also adjusted and refined. In its focus on underlying patterns, abduction also differs advantageously from the two other, shallower models of explanation. The difference is, in other words, that it includes understanding as well.

Abduction is the method used in medical diagnosis and in diagnosing errors in technical systems; the interpretation of poetry is another field. It has had increasing impact in many areas of linguistics and the social sciences. Abduction is close to hermeneutics (Eco, 1990; cf. Chapter 4 in this book).

Induction has its point of departure in empirical data and deduction in theory. Abduction starts from an empirical basis, just like induction, but does not reject theoretical preconceptions and is in that respect closer to deduction. The analysis of the empirical fact(s) may very well be combined with, or preceded by, studies of previous theory in the literature, not as a mechanical application on single cases, but as a source of inspiration for the discovery of patterns that bring understanding. The research process, therefore, alternates between (previous) theory and empirical facts (or clues) whereby both are successively reinterpreted in the light of each other. In comparison, induction and deduction appear more one-sided and unrealistic, if we take into consideration how research is actually carried out; in other words, those who follow them too strictly risk putting a straitjacket on their research. Theory is poetry over facts, it has been said (Erslev, 1961). Maybe, but then as much as poetry in and through facts. Even though ‘facts’ are the surface of friction necessary to generate theory, theory is not a simple summary or description of ‘empirical facts’ as in natural history. The theory must also transcend ‘facts’ in order to achieve scope. ‘Facts’ thus serve to occasion the theory, while continually playing the role of critical tuning instrument and fount of new ideas for the theory.

Glaser and Strauss’s induction from theory-free facts (Chapter 3 below) can be regarded as a counter-picture to Popper’s long-dominating and one-sided thesis of deduction from fact-free theory (e.g. Popper, 1963).³ In the latter case, it is a question of a kind of scientific virginal birth which should be as rare or miraculous for the practical researcher as its obstetric counterpart. Since Popper’s influence has been so strong, Glaser and Strauss’s thesis can perhaps be viewed as a polemically understandable one-sidedness. We argue, though, that there is a way beyond this polarization between induction and deduction.

Here it is fitting to quote what Whitehead says about induction and, implicitly, about abduction:

This collapse of the method of rigid empiricism … occurs whenever we seek the larger generalities. In natural science this rigid method is the Baconian method of induction, a method which, if consistently pursued, would have left science where it found it. What Bacon omitted was the play of a free imagination, controlled by the requirements of coherence and logic. The true method of discovery is like the flight of an aeroplane. It starts from the ground of particular observation; it makes a flight in the thin air of imaginative generalization; and it again lands for renewed observation rendered acute by rational interpretation. …

[T]his construction must have its origin in the generalization of particular factors discerned in particular topics of human interest. In this way the prime requisite, that anyhow there shall be some important application, is secured. The success of the imaginative experiment is always to be tested by the applicability of its results beyond the restricted locus from which it originated. In default of such extended application, a generalization … remains merely an alternative expression of notions [already] applicable. (1929: 4ff.)

Also, a theoretician of science like Bunge never tires of pointing out that it is not possible to generate theory by just condensing empirical data (see, for instance, Bunge, 1967; cf. also Toulmin, 1953).

Let us for the sake of simplicity illustrate this with the traditional example of positivism – swans and their colours. Deduction would start by postulating that if a bird is a swan, it is white, and then draw the conclusion that if we meet an individual swan, it is white. Induction first meets one white swan, then another, then yet another … and finally draws the conclusion that all swans are white. Abduction would at first observe a swan with a certain colour, and then show how, for example, the bird’s genetic structure might generate a certain colouring. This underlying pattern then explains the individual case.

Neither induction nor abduction are logically necessary – that is, they allow mistakes – yet we could not do without them, any more than without deduction, which is logically necessary at the price of empirical emptiness (it does not say more than its premisses). Through induction, we draw, for instance, as children the conclusion that objects a, b, and so on fall to the ground if they are dropped, and therefore probably also all other objects. Abduction can, as was indicated above, be illustrated by diagnostics and also with the interpretation of poetry. In the former case, we observe a symptom and from this draw the conclusion of an underlying pattern – that is, a disease. In the interpretation of poetry, we see a certain pattern as an indication of a hidden but underlying pattern in the text. Since abduction is not logically necessary, it must be controlled against more cases. The physician must, for instance, compare with more symptoms (or patients); the interpreter of lyrics with more expressions, verses (or poems). A research process may rather be compared with a series of flights, such as was described in the Whitehead quotation, rather than with a single one (or even better with one long air trip with several intermediate landings). In other words, what is needed is a repeated process of alternating between (empirically laden) theory and (theory-laden) empirical ‘facts’.⁴ This means a hermeneutic process during which the researcher, as it were, eats into the empirical matter with the help of theoretical preconceptions, and also keeps developing and elabo...