Introduction
Durkheim and Weber did not share the same conception of sociology, but they harboured no doubts that it should pursue no other goal than that of a science: the goal, that is, of creating knowledge.
A careful reading of the comments formulated on sociology by philosophers of the social sciences, by historians of sociology, and by sociologists themselves yields the impression that, on the contrary, sociology has never been able to create authentic knowledge, that its goals have been both multiple and indeterminate, and that in ali cases it has been forced to abandon the scientific ambitions of its great founders. In an article published in the prestigious weekly Die Zeit, the commentator went so far as to argue that sociology will not survive unless it officially renounces every scientific ambition.2
Must we be so pessimistic about the past and the future of sociology? Is it necessary to forgo the âpositivistâ goals of the classical sociologists? Whence derives the scepticism of so many sociologists on this issue? These are the questions upon which I shall reflect in what follows.
However, before attempting to answer these questions, clarification is required of what is meant by âscienceâ. The notion, of course, is a complex one, and yet few commentators have sought to define it before asserting the capacity or incapacity of sociology to be a scientific discipline like the others.
1. Positivism or positivisms?
For Durkheim, like many contemporary sociologists, a discipline that claims to be scientific must pursue a positivist programme. Yet âpositivismâ is a set of complex ideas. Like e very influential intellectual movement, it encompasses a wide range of positions, in the same manner as empiricism does, for instance, or marxism or rationalism.
Despite this diversity, like marxism or rationalism, positivism conveys a number of cruciai intuitions. Above ali, it raises a question of essential importance for any discipline that claims to be scientific, that of the criteria with which to distinguish a scientific theory from one that is not, and with regard to scientific theories between those that are valid and those that are not. Durkheim raised this âdemarcationâ issue in Les Règles, where he insisted that sociology should set itself the goal of being a science. This presupposes that a demarcation line can be drawn between scientific and non-scientific. Although it is not certain that this assumption is valid, we cannot evade the issue of the criteria with which a theory may be defined âscientificâ, in an age when scepticism as to sociologyâs ability to produce genuine knowledge affects even sociologists themselves.
Comteâs positivism envisaged science as a superior form of knowledge. But it cannot be reduced to this principle alone, for Comte also contended that the notion of âcauseâ should be replaced by that of iawâ and that science is made up of a finite set of interrelated or distinct disciplines arranged in a complex hierarchy. It follows from these principles, for example, that psychology and economics cannot claim the status of sciences (in that they do not belong among the fundamental disciplines), or that sociology must employ the methods of disciplines lower down in the hierarchy of sciences, like biology or physics or chemistry, while at the same time treating societies as entities with a higher level of complexity than that of organisms.
Durkheim took several of his masterâs lessons on board. He wholly endorsed Comteâs classification of the sciences, as well as his idea that the ontological discontinuity among the orders of phenomena entails a methodological discontinuity among the scientific disciplines that study them (Durkheim, 1972: 106). One condition for sociology to acquire scientific status is, according to Durkheim, that it should respect this discontinuity. It is consequently necessary to exclude psychology from sociological explanation, and to prohibit the interpretation of social facts by means of anything other than social facts. Not to do so would be as unacceptable as claiming that âbiological phenomena can be analytically explained by means of inorganic phenomenaâ (Durkheim, 1972: 102): an idea that Durkheim deemed so obviously mistaken that he used it to sustain by analogy his maxim that social facts should be explained through social facts. However, the idea that underpins modern biology - that the organic can be explained on the basis of the inorganic - was unacceptable to Durkheim because it breached the principle that âa whole is not identical with the sum of its partsâ (Durkheim, 1972:102). From the continuity/discontinuity relation between sociology and biology, he drew the idea that societies must be considered as wholes, although these wholes are more complex in nature than living organisms. Durkheim condemned economics and psychology for other reasons as well, and again in the light of Comteâs theory: economics, he argued, is the product of an arbitrary division among phenomena; by interesting itself in non-observable phenomena, psychology breaches an essential principle of positivism, namely that the scientist must abide by the facts alone.
The academic character of Durkheimâs style contrasts with Comteâs prophetic style; it was Durkheimâs determination to base his analysis on systematically collected data that distinguished him from Comte; indeed, he declared his disagreement with him on a number of significant points. For example, he rejected Comteâs conception of sociological laws, writing (Durkheim, 1972: 125) that he had a âparticular conception of sociological lawsâ. Comte sought laws of historical evolution instead of looking for conditional laws of the kind âif A then Bâ. According to Durkheim, âsociological explanation consists solely in establishing causality relationshipsâ (Durkheim, 1972: 124).
As a new âfusionâ of the theory of science which contained, amongst other things, elements of Comteâs positivism, Durkheimâs positivism was therefore much more complex, and in any case more heterogeneous, than that of Claude Bernard, for example. As latterly for Hempel (1965), so for Bernard science was defined by an objective - that of explanation - which in its turn was defined as the subsuming of phenomena under laws. One gains the impression that Durkheimâ s aim in Le Suicide was to establish laws in sociology that were similar in their form to Boy leâs Law or Mariotteâs Law because they joined variables with a functional form (so that suicide rates are tied to âanomieâ and to âegoismâ by a U-curve). But, although evoking Hempel, he had a restricted conception of causality that modern positivists would reject, in particular when he maintained that the same cause always corresponds to the same effect: âThus, to return to the examples cited above, if suicide depends on more than one cause, this is because there are in reality different types of suicideâ (Durkheim, 1972: 127).
Another version of positivism, the one developed by Carnap, is much more radicai than Bernardâs. This version contends that, at least in principle, an authentic scientific theory must be amenable to translation into a form that contains only empirical statements - that is, statements which can be immediately verified by observation. There is, of course, a linking thread between Viennese positivism and the early positivism of Comte: for both Carnap and Comte, scientific thought has greater reliability than other forms of thought, and it owes this superiority to the attention paid to the elimination of hidden causes, so that sociology eschews mention of âhuman natureâ, which is a âmetaphysicalâ notion, according to Comte. In general, this determination to adhere as closely as possible to the observable is a principle that appears in diverse forms in the majority of the variants of positivism.
Carnapâ s positivism has been strongly criticised, but his extremist conception of science is to be found in ali the movements of influential ideas (more often denoted by other names). Machâs âempirio-criticismâ in physics, Malinowskiâs âfunctionalismâ, âbehaviourismâ, whose influence on psychology is well known, Milton Friedmanâs âpositivismâ in economics: ali these can be considered to be transformations of Carnapâs âpositivismâ in that they ali identify the notion of scientificity with at least the potential elimination of non-observable phenomena from scientific discourse. Behaviourist psychologists argue that psychological theory in its entirety can be reduced to statements about relations between stimuli and responses, given that both are observable. Malinowski contended that anthropology should restrict itself to studying the relations among observable social entities and to proving, for example, that some social institutions are inter-related while others are mutually exclusive. The same principles are to be found in Murdock. LĂŠvi-Strauss took up the idea and radicalized it: his structuralism maintained that the anthropologist should study the regularities in observable phenomena and in general seek to identify the âstructuresâ that support them. Friedman has defended the idea in a more flexible manner, arguing, for example, that economie actors pursue non-observable interests or preferences. But he treats these latter as âblack boxesâ which should not be opened, or in other words, as axioms that may not be discussed in themselves. He merely asks whether the consequences that they produce in observable phenomena - that is, whether the âpredictionsâ that they permit us to formulate about reality - are confirmed or refuted by reality. Here the non-observable is admitted but it is excluded from discussion.
Popper, as we know, was highly criticai of his predecessor and fellow-countryman Rudolf Carnap, but he may not have distanced himself from Carnap to the extent that he claimed. Popper invariably protested when he was called a âpositivistâ, and he vigorously opposed his âcriticai rationalismâ to positivism. Yet he shared a fundamental âempiricismâ with Carnap. Both of them regarded the congruence between scientific theory and observed reality to be the sole criterion of scientificity, and for both of them a good theory was one that was congruent with the real. Popper undoubtedly denied that a theory can be reduced, even in principle, to statements with the status of protocols of observation. Yet he was uninterested in any discussion of the non-empirical components of scientific theories. What counted for Popper was solely the fact that a theory could be compared against reality and considered to be congruent or otherwise with it. A valid theory was one which was endorsed by the real, or which at least did not appear to be incompatible with it. The fact that Friedman has called his position âpositivistâ while it is actually Popperian, indirectly demonstrates the kinship between âpositivismâ and âcriticai rationalismâ. For Friedman, it is of little importance to know whether the non-empirically verifiable propositions introduced into an economie theory are realistic or not. It is for this reason that the Friedmanian economist (like the sociologist who follows him) believes that s/he can legitimately interpret ali behaviour on the basis of the utilitarian axiomatic of rational choice theory and reject criticisms expressed in the name of realism. According to the Friedmanian positivist, an axiomatic cannot in fact be discussed in itself. Ali that matters is knowing whether it produces results that successfully pass the Popperian test of congruence with reality.
These remarks may suffice to show the diversity of the ideas that have been grouped under the heading of positivism. This historical complexity notwithstanding, it is possible to discern a number of features shared by the main versions of positivism. These common features are the following:
the assertion of science as a higher form of knowledge;
definition of scientificity on the criterion of the elimination of the unobservable (with the two variants described: absolute elimination on principle, or elimination from discussion);
definition of the purpose of science as the study of the relations among observable phenomena.
Although not always to the letter, Durkheim adhered to this distinctive programme of what one might cali âhard positivismâ; a programme to which he added a number of supplementary principles derived from Comte. It is evident that âegoismâ and âanomieâ are not unobservable phenomena, but rather, in Lazarsfeldâs sense, variables which are indirectly measurable by means of âindicatorsâ.
One may counterpose âhardâ positivism with a âmoderateâ or âsoftâ positivism which maintains the first of the above principles but proposes a completely different definition of scientificity insofar as it accepts (contrary to Carnap but in accordance with Popper or Friedman) the introduction of non-observable elements into a scientific theory and discussion of the acceptability of these unobservables (contrary to Popper and Friedman).
Unobservables - that is, non-observable elements and the non-verifiable propositions of a theory - constitute an issue of centrai importance in the debate on the criteria of scientificity. Admitting unobservables is not to claim entitlement to introduce any whatever explanatory theory; this is the preoccupation of the positivist. To the extent that they admit that the axioms of a theory may be discussed in terms of their observable consequences, Popper and Friedman form a link between âhardâ and âsoftâ positivism. An unobservable may therefore be indirectly judged positively or negatively according to whether its introduction gives rise to acceptable theories (that is, congruent with reality). We shall see, however, that it is necessary to go further than Popper and Friedman, and that it is not enough to discuss unobservables solely in terms of their empirical consequences.