ARENA Working Papers
WP 99/33

Understanding of Theory in the Social Sciences

Lars Mjøset
ARENA, Department of Sociology and Human Geography , University of Oslo

The notion of theory plays a crucial role in social science, as in any science. But few social scientists have discussed the term in a philosophy of science framework. Even fewer have asked whether the term has the same meaning in all social science. Instead, tense debates have raged, with opponents implying — but not explicating — different understandings of theory. Attempting a classification, this overview distinguishes four different understandings of theory, all of which have been influential in the social sciences.

1. The main postwar ideal — the deductive-nomological notion of theory

This notion of theory dominated postwar science. It originated from early 20th century positivism. It generalises the procedures of experimental natural science, physics above all. Two elements are crucial: First, the idea of logical deduction in axiomatic systems: theorems are deduced from a limited number of axioms. Second, a correspondance criterium of truth: empirical sciences depend on correspondance with facts (sense data), a comprehensible external reality (philosophical realism). The deductive system generates hypotheses which are potential universal laws.

Given the laws, it is possible to specify a set of initial conditions under which a predicted result should follow. In an experiment, these conditions are established. If the result confirms the prediction, the hypotheses (laws) are strengthened. The experiment shows that the law represents general knowledge on how the initial conditions cause an outcome. This is also called the covering law model of explanation: The law is the black box which transforms causes into effects. Theory is a set of such laws. Theory is compact knowledge: Many regularities can be subsumed under the same general law, making research systematic and cumulative. Such a notion of theory provides a clearcut criterium of demarcation: testing og law-based deductions. Explanations which do not satisfy this critium are ad hoc explanations and must be avoided.

Theory is thus an ever more complete system of universal laws. At the extreme end lies the vision of a complete unity of science posessing nomological knowledge of reality in its entirety.

There are nuances between various positivist schools. But in the postwar period, the critical rationalist view, first argued by Karl Popper, has dominated: no theory can ever be verified, only falsified. Theories are not just generalizations, they are hypotheses on universal relations which have so far not been falsified.

In later decades, the force of the deductive-nomological notion has been weakened, even in the natural sciences. Philosophers of science now discuss much more sophisticated criteria of demarcation, it is also admitted that the notion tells us little about non-experimental natural sciences like evolutionary biology. However, it seems that in the social sciences, the original and very restrictive notion has retained its status as a regulatory idea.

This is a paradox, since the results of social science's search for universal laws have been disappointing. Merton's (1949, p. 108) statement is as relevant today: �sociologists (including the writer) may discuss the logical criteria of sociological laws without citing a single instance which fully satisfies these criteria.� If an instance is quoted, Durkheim's statement on the relation between social cohesion and suicide is probably the most frequent choice. But it remains contested whether that �law� has served social research into suicide in ways that can be compared the ways laws on gravity, thermodynamics, etc. have served natural science.

This notion of theory may be sustained if social science could be reduced to natural science. We cannot here specify the various forms of reductionism (physicalism, biologism, psychologism), but if the joys, sorrows, memories, identity and free will of humans could in some way be analysed as �no more than the behavior of a vast assembly of nerve cells and their associated molecules� (Crick 1994, p. 3), then social science as such could be dispensed with. Although some scholars hold that such a reduction is in principle possible, very few claim that it can be achieved in practice. This topic, however, will not be discussed further here.

As long as no such reduction has been achieved, we must be aware of the paradoxical influence of the deductive nomological ideal: it has had an enormous influence as a regulative idea, but there is hardly a single example of its successful application in social science. In practice, social scientists adhering to the ideal have applied two different modifications of the ideal: we shall name them the law-oriented and the idealising notions of theory. Opposed to these notions are the constructivist and critical notions of theory, implied by scholars who do not accept the deductive-nomological ideal.

Table 1 — Four notions of theory in social science

These four notions have all been influential and involved in applied social science research in the post-war period. Besides their differing commitment to the deductive-nomological ideal, Table 1 shows that they differ on two other counts: on the question of whether social science should generalise in the same way as the natural sciences, and on the questions of whether ethical foundations matter for theory.

2. The law-oriented notion of theory

This notion retains a nomological focus at the cost of deductive strength. This solution can be related to an ambiguity in Popper's original discussion. Despite his support for the covering law model, Popper was critical of the unity of science programme. He rather accepted the distinction between generalising sciences and history. Causal explanations in history relate to singular events, and universal laws here prove empty. �For the theories or universal laws of generalising science introduce unity as well as a `point of view'; they create, for every generalising science, its problems, and its centres of interest as well as of research, of logical construction, and of presentation. But in history, we have no such unifying theories; or, rather, the host of trivial universal laws we use are taken for granted; they are practically without interest, and totally unable to bring order into the subject matter� (Popper 1943, p. 264). Popper notes particularly that such trivial laws do not define the initial conditions.

Popper mostly classified social science together with the generalising natural sciences. But at times he also regarded social sciences as specifying. The law-oriented notion of theory emerged as a response to this dilemma. The deductive-nomological ideal is retained: social science is seen as generalising, distinct from the historians' focus on singular processes. But it is admitted that universal laws in the social sciences have proven trivial. Social science must still generalize, but given the absence of truly universal laws, we should focus instead on regularities that apply only within specific contexts. Such a notion of theory as a collection of �lawlike regularities� or �quasilaws� is less strict that then deductive-nomological ideal.

In this understanding, the difference between a theoretical hypothesis and inductive generalization evaporates, as does the distinction between �theoretically based� and �ad hoc� explanations. Lawlike or bounded regularities are difficult to separate from boundary conditions, since these conditions themselves enter into the definition of the lawlike regularities. It is thus hard to devise decisive experiments, even in cases with no ethical obstables to experiments involving fellow humans. If a theory (a set of quasilaws) does not explain (is falsified), it will always be possible to argue that slightly different initial conditions applied. In addition, such bounded regularities are also more easy reconcilable with statistical generalisations than the strictly determinist notion of a covering law is.

Merton (1949), one of those who sense the gap between his own methodological ideals and the ways in which social scientists really proceed when they explain something, coined the notion of middle range theories. These were legitimated by reference to the deductive-nomological ideal, but allowed contextual explanations unrelated to laws. This was the main methodological conviction in all branches of variables-oriented social science, dominating Western research in the postwar period.

The dissonance between the deductive-nomological ideal and real-life explanatory strategies Merton solved by treating middle range theories as temporary. He claimed that social science was still too young. When they matured, as physics already had, the body of middle range theories would converge into a system of universal laws on the model of experimental natural sciences. Untill social science matured, he concluded, it should give priority to middle range theories, deliberately avoiding universal theory. On this notion, the social sciences would be weak on deductive rigour and parsimoniousness, but strong on contextual explanations.

3. The idealising notion of theory

All through its history, the social sceinces have been haunted by debates on methods, always returning to the question of whether the subject-matter of the social sciences — human action, intention and meaning — require a specific approach to theory. This line of argument made possible another modification of the deductive-nomological ideal, one which retained deductive rigour by replacing law-orientation with idealisation.

As a human being, the social scientist has the capacity to understand motives involved in actions and interaction. But these motives cannot be conceived as universal laws without exceptions. Even the simplest economic regularity — that demand increases when the price decreases — has its exceptions. Trying to transcend the dichotomy between nomothetic and ideographic approaches in the German �Methodenstreit�, both Max Weber and Georg Simmel emphasized that �laws� in the social sciences are ideal types, idealizations of human motives.

These idealizations are often formalized in axiomatic systems. In the law-oriented notion of theory, emphasis was on the positivist experimental criterium of correspondance to facts. In the idealising notion, the emphasis is rather on the other feature of the positivist model: deduction in axiomatic systems. While law-oriented researchers want to retain a notion of laws by specifying all its conditions, idealising researchers hold that it is impossible to describe �all the sets of circumstances� (Boudon 1986, p. 65) in which an idealisation applies.

They thus focus on the conditions which establish the ideal situation. Under such stylized conditions, human interaction can be understood via mathematical modelling of parametric and strategic rational action (game theory) (Elster 1983). This is the way the most prestigous postwar branch of social science, neoclassical economics, has always gone about. The programme of rational choice theory — particularly influential in political science and sociology since the 1980s — strives to generalize such procedures in the other social sciences. This is often formulated as a programme for �metodological individualism�, according to which the human actor is the smallest unit in social science explanations.

In these idealised models, the researcher is herself in extensive command over the initial conditions, notably the preferences that are plugged into the models. These idealised conditions are �theoretically interesting�, but they do not occur frequently (Boudon 1991, p. 67). It is thus crucial to distinguish social science models from natural science laws: the former yields predictions in an idealised world (classical economists often regarded this as a �natural order�), but only in a very limited number of real situations. In contrast, the law of gravity applies very widely and in cases of flying objects such as birds or airplanes, the physisist can readily define the conditions for this deviance. �A law applies to all cases, a model to ideal, that is, to particular cases� (Boudon 1991, p. 68).

Parsimoniousness and deductive power here comes at the cost of a gap between models and the real world. Constructing idealised models, social scientists use their skills in calculating, infering, idealising and guessing with respect to human intentionality. If the model is to be fully idealised, perfect knowledge must be assumed. The scientist assumes perfect knowledge of the knowledge possessed by the actors of her model world. Both the scientist and the actors in the model world are omnipotent. A dilemma in this approach to theory is thus that these very strong assumptions must be relaxed if the model is to claim any empirical relevance. This relaxation, however, makes the models highly complex.

This gap between model and explanation seems even harder to bridge than the theory/explanation gap of the law-oriented approach. But scholars who are concerned about explanatory relevance try to bridge it. They then end up in a methodological compromise quite similar to that of middle range theories (cf. recent discussions on the concept of mechanisms). The move to the middle range level mostly also imply relaxed assumptions about knowledge, sometimes bringing them quite close to the constructivist position, which we shall present below.

The two modifications of the deductive-nomological ideal — law-orientation and idealisation — has different comparative advantages. But very often, the distinction between them is not made, and scholars strive to have it both ways, that is to discover lawlike regularities through idealised models. Such attempts (e.g. Elster 1983, Goldthorpe 1996) should be regarded as attempts at synthesis, and be judged with reference to the differences between generalisation and idealisation, as traced above. In addition, other differences between the two modifications should be taken into account: they also differ with respect to their implied notions of causality. Models of parametric and strategic rationality are mostly simultaneous equation models. While it may be claimed that any equation implies a causal relation, such general interdependence is a looser notion than the mechanical notion of causality —requiring temporal asymmetry — implied in the law-oriented understanding of theory.

4. The constructivist notion of theory

The term constructivism has many meanings. In philosophy, it indicates a long tradition of scepticism, in our century known under labels such as pragmatism and anti-realist epistemological constructivism. In the social sciences, more special schools are called constructivist. For the purpose of this survey, we define a constructivist notion of theory with reference to its questionning of any foundation for the social sciences: Not only is the deductive-nomological ideal rejected, it is also implied that no alternative ethical foundations can be found (Table 1). It should be noted that this constructivist position has not been systematized to the same extent as the deductive-nomological notion of theory. What follows, then, is a synthetic presentation based on several statements that may differ with respect to important details.

Compared to the law-oriented and idealising notions of theory, the constructivist approach entails a very different notion of knowledge. The deductive-nomological ideal provides a strict separation between scientific knowledge and everyday knowledge. According to the constructivist position, in contrast, the line of demarcation between science and everyday knowledge is fuzzy. Social science theory is not in principle different from everyday knowledge! It may differ in degrees, but since scientists are themselves human beings, also their knowledge is grounded in their everyday life. This is because also science is a human collective, research groups are communities, their knowledge is the bounded knowledge attained by interacting human beings. Any demarcation of social science is seen as a result of social processes, and thus relative.

Originating in scepticist epistemology, such a notion of knowledge has been implied by many schools in the sociology of knowledge. More recently, interdisciplinary cognitive science has also been a major inspiration to constructivism. Such a notion of theory does not draw its main notions from the routines of experimental natural science, but from the study of knowledge, communication and information in society at large. It also has a strong relationship to applied studies of social problems, which study how social processes define the problems and risk perceptions that create demands for social science research.

Science cannot be easily demarcated, so social processes define (construct) certain realms of knowledge as science. Attempts to apply strict demarcation criteria implies a focus on testing of theories, while constructivists mainly focus on theory formation, studies of the disciplines and professions into which knowledge production is organized, and of continuing struggles to define what science (theory) is. The scientific community is influenced by relations of trust, authority, tacit knowledge, processes of exclusion, and rhethorical strategies. As Kuhn (1970, p. 21) emphasized, the explanation of scientific discoveries must �be psychological or sociological [...], a description of a value system, an ideology, together with an analysis of the institutions through which that system is transmitted and enforced.�

Certainly scientists work in a social subsystem which encourages analytic rigour, systematization, quest for consistence, democratic access to research results so that they can be replicated, etc. But still science is not qualitatively different from everyday knowledge, not something normatively sacred pursued independently of society at large. A sociologist of knowledge may study the deductive-nomological ideal of theory as a powerful rhethorical figure in the contest on how to define science.

We have seen that there was a particularly strong connection between economics and the idealising notion of theory. The constructivist notion of theory has a particularly strong affinity to anthropology. It is critical of methodological individualism, emphasizing the crucial role of shared knowledge. A strong version of the constructivist position argues that any normative definition of �science� is in principle not different from the myths found e.g. in tribal societies studied by anthropologists.

In our everyday interaction we agree on a view of reality. Such is the case among scientists too: physicists, for instance, use a number of concepts which cannot be �seen� as facts in reality (force, field, atoms, electrons, quarks). These entities are studied through complex experimental equipment, e.g. particle accelerators. But scientists agree on the reality of these entities. Thereby, they retain their culture, just like other human communities. In tribal societies, people believe in other kinds of entities, but the claim that these entities are real, is quite parallell (Barnes, Bloor, Henry 1996, p. 84). Even in the hardest of sciences, the process of discovery imply visions, analogies and other features that are also found in religious or mythical world views.

Kuhn's famous notion of paradigms is in line with such a view: paradigms are exemplars, synthesizing in them theoretical discussion, experimental practice and instrumentation. It is a �disciplinary matrix� with three main components: a specific symbolism which allows formalization, specific models for connections in the research field (from heuristic analogies to causal claims) and master examples, exemplary solutions derived from experimentation and research. The members of specific research communities are socialized into such a matrix via education and recruitment procedures (Kuhn 1974).

Theory is here understood in continuity with the formation of everyday knowledge. Observing something unfamiliar, we try to understand it in the light of what we already know. We use analogies, extending these to new areas. Such an extension always also imply adjustments. A famous case from the history of science is Darwin, who formulated his theory of natural selection by extending an analogy based on a view of the market mechanism implied in Malthus' theory of population.

Social science has often seen debates on which analogies to allow: The idealising notion of theory prefers game-theoretical and mathematical analogies, and have been quite sceptical of functionalist theories which extend analogies based on the biological organism. Constructivists rather emphaisizes that a much larger diversity of analogies exists to be extended in the service of better explanations of social affairs. Interestingly, even Merton notes that middle range theories start from some notion and its �associated imagry�.

The idealising notion of theory emphasized the similarities between social science and history. In the constructivist notion, not only history, but all of the humanities are seen as highly relevant to theory formation in the social sciences. History is the specifying science of singular events, tracing complex causal chains involving both intentions and social structural forces. These chains are reconstructed as narrative sequences, unpredictable in advance but understandable in retrospect (Ricoeur 1986). The historian's craftwork is to interpret the motives of the actors in their historical context, that is to imagine the everyday reality of historical actors. Similarly, the majority of antropologists define their task as �thick description� (Geertz 1973) of how everyday action is embedded in more or less local cultures. The various phenomenological and symbolic interactionist (Glaser & Strauss 1967) strands within sociology share such a perspective.

Also most post-structuralist approaches imply a constructivist notion of theory: the basic structure of everyday knowledge is here conceived as discourses, which are analysed by means of analogies drawn from linguistics. Social science theories are embedded in such discourses. As parts of such discourses, the social sciences influence the practices of institutions, media, professions and disciplines of our modern society.

Constructivists agree with the idealising approach that social science study meaningsful human action. Meaning is irreducibly linked to human interaction, which is crucially mediated by symbols and language. The social sciences thus face a challenge which does not bother natural science: how to access the symbolically mediated meaning, motives, preferences, identities of interacting human beings. The idealising position tackles this challenge by making strict assumptions about knowledge. In contrast, the constructivist focus on everyday knowledge implies that knowledge is always limited, and so rationality is always bounded. The notion of theory is not linked to the ideal of perfect knowledge. Theories do not take the form of axiomatic models, but of contextual understanding of interacting motives (as e.g. in Goffman's notion of an interaction order). Social science theories are social products, they are sometimes drawn into complex social constellations where other individual or collective actors pursue their own strategies. Theories may influence real developments, but they scarcely predict!

In this notion of theory, the gap between theory and explanation, between models and their empirical application, is minimized. Since all knowledge is limited, any theory is �middle range�! That label is unnecessary since there is no assumption of a complete, full range theory. Since we do not possess coherent deductive theories accusations of ad hoc theorizing are misplaced. Theories are processes (Glaser & Strauss 1967), they cannot be accumulated as a stock of knowledge, they are not specific enough for direct application and they change over time. They do not provide clearcut accounts of an external reality, since — even if we assume that such a reality is �out there� — any such alledgedly unequivocal representations will eventually be contested.

The deductive-nomological ideal, in this view, leads social science astray by implying that science should converge towards complete knowledge of the laws governing an external reality. Quite the contrary, social science has never seen anything like e.g. the theory of relativity, social scientists tend to use the modest label �a theory of�... Constructivists argue that this plurality of theories should not be regretted, more theories will lead to better explanations (Glaser & Strauss 1967).

The constructivist approach — which we have here defined very broadly — does not generate very parsimonious theories, but by resisting any kind of decontextualisation, it often proves strong on explanation. The lesson is that careful consideration of context and motives — structure and action — can never be dispensed with.

5. Critical theory

The constructivist notion of theory is frequently accused of relativism: by devaluing the correspondance to an external reality, constructivism alledgedly implies that the most influential groups in the social system define what scientific theories are. This view is a provocation to the positivist tradition, whose notion of theory was actually designed to expose as pseudo-science fascist theories about such things as the �fate of a people�. But with the use of science as legitimation of techocratic rule and the role of natural science in the postwar nuclear arms race, criticism rather turned against the positivist position.

In this context, J. Habermas (1970) and others revitalized the interwar (Frankfurt School) programme of critical theory. This position agrees to the rejection of the deductive-nomological approach, but insists that constructivism must be cured of its relativism by reference to an ethical foundation defined by communicative ethics. This internal linkage of theoretical and ethical reflection sets critical theory apart from the other three notions of theory.

The social sciences are related to human interaction, involving understanding based on language. Since it should in principle be possible to agree peacefully in disputes between human beings, a communicative ethics is implied in all social interaction, the social sciences included. The evolution of modern society has enlightened the humans, equipping mankind with a set of human rights. Wherever these are not respected, social movements may form, legitimating their strategies with reference to solid ethical principles. Habermas regards this consensus-criterium of truth as an alternative to the positivist criterium of correspondance. Social science is defined not by its stock of universal laws, but by its commitment to universal ethical principles. Since the reality studied by social science consist of interacting humans, the thruth of its theories cannot be determined by correspondance to the present state of affairs, since social mobilisation may prove this state of affairs to be unjust. Since also the social scientist is a participating actor, she cannot refrain from making her own judgement on this question.

The mobilisation of social movements creates increased public attention to the life situation of certain social groups, supporting claims of equality (universal suffrage, equal social rights, etc). The standard example is the labour movement, which according to classical 19th century Marxism was the �critical subject� whose interests coincided with the universal interests of mankind. Marx himself, in fact, combined a sociology of knowledge with a law-oriented approach: He exposed �bourgeois� economists by claiming that their theories were constructions based on the special, non-universal interests of their class. Only scientists who related to universal interests could provide an �objective� understanding of the manmade world, of the �laws of motion� of contemporary society. This idea has been taken up by later Marxist philosophers of science and others who try to fuse inspiration from the sociology of knowledge with anti-constuctivist realism (Bhaskar 1978, Bourdieu 1990). In this survey, however, we emphasize the features that unite critical theory and constructivism.

In the postwar 20th century several feminist theorists claimed that the womens' movement at that time played a role similar to that of the 19th century labour movement. Fusing such an emphasis on women's �standpoint� with a phenomenological focus on everyday knowledge, Smith (1990) critisized the male bias of all established sociological theories, using the common female experience of housework and childrearing as a standard.

This position was countered by feminist theorists who link up with poststructuralism (Scott 1988), claiming that the idea of a standpoint serving as a basic foundation of theory is in itself a construction, as relative as others. More generally, the whole dispute between modernists and postmodernists in contemporary social theory — dating at least back to Habermas' (1985) attack on the post-structuralists, and on Foucault in particular — can be seen as a controversy between critical and constructivist notions of theory.

Critical theory has focused more on ethical foundations and less on concrete paradigms involved in the explanatory efforts of applied social science. Habermas (1968) distinguishes the action sciences from empirical-analytical (natural) sciences and historical-hermeneutic sciences. These are driven by different �knowledge interests�: technical dominance (natural science), maintainence and extension of intersubjective understanding (historical sciences) and liberation. The latter, critical knowledge interest implies that to the extent that social sciences discover lawlike regularities, these are evaluated through collective reflection on their legitimacy.

Habermas argued that adaptations of the deductive-nomological ideal of universal laws by social engineering-oriented social scientists turned science and technology into ideologies. The lawlike regularities are reinterpreted: the very point that social science laws can never be universal is not regretted, but given an �offensive� meaning: their non-universal nature means that they can be changed by social action. Later, Habermas (1981) linked technocratic world views and also mainstream economics to the systemic imperatives of modern societies, while critical theories we connected to the defence of �Lebenswelt� against the system's �colonizing efforts�.

Unique to critical theory is its explicit reflection on the ethical foundations of social science theory. Like constructivism, critical theory is weak on parisimonious theories, but in addition it is mostly also weak on applied explanations. In the latter respect, it may at times converge with the constructivist focus on unique historical sequences, at other times with the laworiented approach, provided that the quasilaws found are shown to be changeable.

6. Conclusion

The discussion of understandings of theory in the social sciences have brought us around a vast rectangular field. Two varieties of a very strict, formal, and demarcation-focused notion occupy the corners at the one end. A rather loose, substantial, and descriptive notion of theory is found in a third corner, and an ethically conscious version in the last corner. Representatives of each side and even of the third and fourth corners inbetween them, are mostly so mutually suspicious that possibilties of coexistence and even interaction are seldom considered. Still, it is striking that the first approach focus on theory testing, the second on modelling, the third on theory formation and the fourth on ethical reflection. Possibilities of crossferilization cannot, however, be specified here. While one must not gloss over fundamental differences, such possibilities should be explored.

We can conclude that there are clear ties between substantial theoretical perspectives and notions of theory in the social sciences. A number of the fundamental debates on social science — between constructivists and positivists, between critical theorists and constructivists, between methodological individualism and �collectivism�, between rationalists and empiricists — in fact imply debates between different understandings of theory and of knowledge more generally. This should not surprise us: social science research is after all in itself a social activity.

Bibliography

Barnes B Bloor D Henry J 1996 Scientific Knowledge. A Sociological Analysis. Athlone, London

Bhaskar R 1978 A Realist Theory of Science. Harvester, Brighton

Boudon R 1991 Theories of social change. Press, Cambridge

Bourdieu P 1990 Outline of a Theory of Practice. Polity, London

Crick F 1994 The astonishing hypothesis. Schribner's, New York

Elster J 1983 Explaining technical change. Norwegian University Press, Oslo

Geertz C 1973 The interpretation of cultures. Basic Books, New York

Glaser B G & Strauss A L 1967 The Discovery of Grounded Theory. Aldine De Gruyter, New York

Goldthorpe J 1996 The Quantitative Analysis of Large-Scale Data-sets and Rational Action Theory: For a Sociological Alliance. European Sociological Review 12: 109-126

Habermas J 1968 Technik und Wissenschaft als `Ideologie'. Suhrkamp, Frankfurt am Main

Habermas J 1970 Zur Logik der Sozialwissenschaften. Suhrkamp, Frankfurt am Main

Habermas J 1981 Theorie des kommunikativen Handelns. Suhrkamp, Frankfurt am Main

Habermas J 1985 Der Philosophische Diskurs der Moderne. Suhrkamp, Frankfurt am Main

Kuhn T S 1970 Logic of discovery or psychology of research. In: Lakatos I & Musgrave A (ed.) Criticism and the Growth of Knowledge Cambridge University Press, Cambridge

Kuhn T S 1974 Second Thoughts on Paradigms. In: Suppe F (ed.) The Structure of Scientific Theories, Urbana, Ill

Merton R 1949 Social Theory and Social Structure. Free Press, Glencoe, Ill

Hempel, C G 1965 Aspects of Scientific Explanation. Free Press, New York

Popper, K R 1942 The Open Society and its enemies. Vol. II. Routledge, London

Ricoeur P 1986 Du texte � l'action: essais d'herm�neutique 2. �ditions du Seuil, Paris

Scott J 1988 Deconstructing equality — versus difference: the Uses of Post-structural Theory for Feminism. Feminist Studies 14: 33-50

Smith D 1990 The Conceptual Practices of Power. A Feminist Sociology of Knowledge. Northeastern University Press, Boston

[Date of publication in the ARENA Working Paper series: 11.01.2000]