Must sociology be qualitative?

Must Sociology Be Qualitative? EVAN FALES Department of Philosophy University of Iowa ABSTRACT This article addresses the question of whether sociology can in principle become a quantitative science. I distinguish several senses in which a contrast between quantitative and qualitative science might be understood. I focus on the central--and traditional--sense: can sociology become a nomological science, in the way physics is? I argue that it cannot, on the ontological ground that the determinants of human actions cannot be analyzed in purely causal terms. In the article I try to characterize this difference. Is there a distinction in method between quantitative and qualitative sociology? If so, which kind is better? Philosophers and sociologists have debated these questions for a long time; and among philosophers, at least, they are by no means resolved. I wish to examine the alleged distinction and some of its implications. These, I shall argue, embrace not merely methodological issues but also commitments about the nature of society. One might say that each method presupposes a distinct (however vaguely articulated) ontology; and that these ontologies are antagonistic. There is, I think, an important element of truth in this thesis. However, the structure of the issue in question is by no means simple. An effort to sort out what might be meant by "quantitative" methods in sociology as contrasted with "qualitative" ones reveals no neat correspondence between methodology and ontology, that is, between the choice of proper techniques of study and our conception of the subject matter of that study. For one thing, there are different senses in which a method, or the data collected by that method, can Requests for reprints should be sent to Evan Fales, Department of Philosophy, University of Iowa, Iowa City, Iowa 52242. Qualitative Sociology, Vol. 5(2), Summer 1982 0162-0436/82/1400-0089502. 75 89 © 1982Human Sciences Press 90 QUALITATIVE SOCIOLOGY be said to be quantitative or qualitative. For another, data, whether quantitative or qualitative in form, can characteristically be interpreted from the perspective of any of the rival ontologies. This is not to say, however, that there are no essential differences in ontology and methodology. On the contrary, I believe it is crucial to sort out these differences in order to gain a clear view of what is really at stake, and to avoid ways of framing the issues which deflect attention from the fundamental problems. The central purpose of this article therefore is to collect these issues in a way which makes the problems visible. Since both quantitative and qualitative measurements in a science are measurements of the properties of items falling within the domain of that science, it will be appropriate to begin by considering what sorts of distinctions in general might be intended by contrastive uses of "qualitative" and "quantitative," Which, if any, of these distinctions is such that some properties are qualitative, and not quantitative, and vice versa? On one sense of "qualitative," a measurement or datum is qualitative if it is given as a rough estimate rather than precisely specified. But clearly, a qualitative measurement in this sense is in no way incompatible with more precise specification, and the distinction mentioned is hardly relevant to sociology except insofar as many sociologically important properties (including numerically specifiable ones) can at best be estimated roughly. Secondly, the qualitative/quantitative distinction is sometimes associated with a distinction between subjective sensations and the causally responsible objective properties. However, sociologists are concerned with objective facts, so this distinction will hardly be of use to us. A more interesting distinction may be allowed between properties which have a phenomenologically qualitative character in contrast to those which exhibit a phenomenologically quantitative character. It is plausible to say that properties such as distance, duration, and temperature, for example, exhibit themselves to us, however vaguely, as quantities; whereas beauty and odor do not. It is also plausible to argue that a great many sociologically important properties do not present themselves to us as MUST SOCIOLOGY BE QUALITATIVE? 91 having quantitative character. I believe there is merit in attending to this distinction, but I also believe that its ultimate methodological significance remains in doubt so long as (a) the distinction is not a clear one, and (b) the ultimate nonspecifiability of sociologically relevant properties, such as people's emotional characteristics, remains in doubt. But there is in any case a further sense that might be attached to the qualitative/quantitative distinction, and it is this sense which has historically played the most important role in methodological disputes in sociology. One might, among those properties which can be precisely specified, and in particular among those which can be assigned a position in a metric ordering, look for systematic correlations. The importance of such correlations is, of course, as signposts for the existence of laws, in particular, laws associating numerically specifiable ranges of determinate properties in a mathematically formalizable way. The discovery of such laws is, on one view, the paradigm mark of any mature science, a prerequisite for both full explanation and prediction. Here we approach the central issue of this article: is the existence of a systematic correlation between ranges of properties a necessary and sufficient condition for justified assertion of a sociological law--in the sense in which this would be the case, pace problems about induction--in the physical sciences? For, if so, then a primary task for sociology would be to determine whether such correlations exist, though this may be a far from straightforward task if, as one would expect, the underlying laws (perhaps the laws of physics) are related to sociological variables only via some tremendously complicated reduction. But it will be felt by most practicing sociologists, I am sure, that the remoteness of the possibility of completing this task, if that is its nature, is ameliorated by (a) the fact that the existence of some fairly accessible surface correlations may be taken as strong evidence for the existence of underlying laws, and (b) the fact that knowing what is "in principle" possible has a tremendous effect on the nature of our conception of the subject matter, and hence of the nature of the study of it, however unknown the details. I shall call the claim that every event which occurs in a system is governed ultimately by causal laws the thesis of homological depth. 92 QUALITATIVE SOCIOLOGY Now the conviction that laws underlie the pattern of events that constitute the history of human affairs does not need to be sustained by evidence in the form of discoverable regularities at the level of description which the sociologist ordinarily uses. Any "complete" set of sociological laws would obviously have to encompass the description of nonsocial events and properties which influence that history, in addition to making use of distinctively social predicates, as has often been observed. For we have no reason to expect, in a highly complex system which is neither closed under description in purely sociological terms, nor closed in the physicist's sense of immunity to external forces, that exception-free regularities will exist which will enable the sociologist to formulate laws of a distinctively sociological character. Similarly, there are no laws which govern all the turbulences and life cycles in a farm pond "as such." In the latter case, our conviction is in no way dampened that general laws govern every event which transpires in the pond. On the other hand, if there are discoverable patterns in a complex system, even of an imprecise sort, this fact will strongly reinforce our conviction that genuine laws lurk within the woodwork. Such patterns exist within the life of a pond, and they are a stock in trade among social scientists and historians. The discovery of such patterns may serve a number of purposes, of which more will be said presently. But other purposes aside, patterns of an exact or of a statistical nature are surely strong evidence favoring the thesis of nomological depth. Mill (1959), for example, argues that It is found that when one of the features of a society is in a particular state, a state of many other features, more or less precisely determinate, always or usually coexists with it. But the uniformities of co-existence obtaining among phenomena which are effects of causes, must (as we have so often observed) be corollaries from the laws of causation by which these phenomena are really determined. The mutual correlation between the different elements of each state of society is, therefore, a derivative law, resulting from the laws which regulate the succession between one state of society and another; for the proximate cause of every state of society is the state of society immediately preceding it. Mill is arguing here in particular that uniformities of coexistence entail the existence of laws of succession, from MUST SOCIOLOGY BE QUALITATIVE? 93 which they derive; but we may more generally suppose that regularities, whether of coexistence or of succession, are a clue to the existence of underlying laws of much greater, indeed complete, precision. Let us momentarily consider more exactly why surface regularities in a complex system sustain an inference to nomoiogical depth, whereas the converse inference is not permissible. This may be understood in terms of the number of degrees of freedom assignable to a system. A priori, the number of possible degrees of freedom for a complex system is enormous. Were, therefore, a regular pattern in the activity of such a system to be found, the probability that that regularity was the product of random variation in the properties of the system or its parts would surely be miniscule. We can be quite sure that it is only because of some underlying constraints upon the system (or its parts) that such regularities appear. However, the converse is not generally true; as witness the turbulence in a farm pond or the shapes of clouds. Here, the degrees of freedom of the system are so large in spite of the universal constraints operating upon these systems at the microscopic level, that precise repetition of earlier states of the system must be exceedingly rare. And this argument applies even in the special case of a system which is a closed, deterministic one in which every earlier state is a nomologically sufficient one for the states which succeed it. So, we have an apparently asymmetric situation: surface order (even rough order) counts as strong evidence for nomological depth; surface "chaos" does not militate against it. Nevertheless, it will be the burden of this article to try to show that Mill was wrong, that the kind of statistical surface order which quantitative techniques in sociology are so good at uncovering does not imply the thesis of nomological depth. This perhaps surprising denial will, however, be shown not to be equivalent to the claim that statistical regularities are merely fortuitous, nor to the claim that there are no underlying constraints operating upon social systems. It will perhaps be evident that in order to make good the above assertions, to thread our way between nomological depth on the one hand and randomness on the other, we shall have to introduce a distinction, indeed, an ontological distinction, between nomological constraints strictly speaking on the one hand, and some other form of constraint on the other. In this 94 QUALITATIVE SOCIOLOGY way we may vindicate the distinctive place of qualitative methods, properly understood, in the construction of a science of society, without, at the same time, undermining the possibility of such a science by denying one of the preconditions of any science, namely, the existence of some form of order in its domain. The question before us is: are there regulative principles which properly fall within the domain of sociological inquiry, which contribute to the regularities found in social systems (and hence must be appealed to by way of explaining those regularities), but which are not causal laws in the physicist's strict sense? If there are, then we may wish to appropriate the term "qualitative sociology" to designate the study of these principles, although, as we shall see, it is quite misleading to describe these principles as qualitative, by way of providing contrast with the allegedly quantitative nomological principles of the hard sciences, since both may lead to functional relationships expressible in mathematical terms. I believe that there is a critical, ontologically grounded distinction to be made between two principles of order and explanations which appeal to them. It is a distinction which divides the human sciences from the science of inanimate matter. It is, indeed, the familiar but controversial distinction between activity which is governed by causes (strictly speaking) and that governed by reasons. I shall have to summarize--all too briefly--some of the grounds for holding that this alleged distinction is not ultimately genuine, and then my contrary grounds for holding that it is. It will finally be necessary to consider the part played by reason and individual action based upon reasons in the explanation of social phenomena. To prevent confusion, an initial linguistic observation is in order. Both the word "cause" and the word "reason" are used with considerable latitude by speakers of English, and these usages overlap in such a way as to mask the distinction I wish to mark with my own contrastive use of these terms. I shall therefore use "cause" restrictively to apply to physical causation, leaving it an open question whether human actions are physically caused. Indeed, there are a number of reasons for holding (as for instance Donald Da~idson (1963, 1979) does) that the relation which binds together the members of the sequence of events MUST SOCIOLOGY BE QUALITATIVE? 95 leading from the onset of a man's deliberations to the overt performance of the action which we think of as the product of that deliberation, is causal. It seems evident, first of all, that the connections between the events in such a sequence are regular and nonaccidental. If, secondly, we have had some experience of human nature, and we know a man's motives, we can indeed predict his actions with some confidence. But thirdly, although such predictions sometimes misfire, predictions in the physical sciences also misfire sometimes, and the exceptional cases are handled in a formally similar manner. When a prediction misfires in physics, we look for extraneous forces which were not included among our initial boundary conditions; when our action-predictions fail, we characteristically search for additional (sometimes hidden) motives or beliefs held by the actor. Thus in both cases the end result is seen as the product of some (perhaps difficult to specify) totality of interacting precedent events. Finally, the connection between produced event and producing event(s) is in neither case one of logical necessity, though it is nonaccidental. ~ We may, succumbing to a rather barbarous expression, describe these relations as defeasibly determinative. By this I mean that the antecedent events regularly determine that a consequent event of a given kind shall follow, other things being equal. The "other things" which must be "equal" cannot in practice be specified, but we are of the conviction that if a consequent event of the expected kind does not follow, then it is always proper to seek an explanation in terms of some additional intervening factor(s). For we are convinced that there must be some such explanation. These formal considerations weigh heavily in favor of counting people's deliberations as the cause of their action; and indeed there is a common and suitably broad sense of "cause" which implies no more than this. On that usage, people's motives, desires, and beliefs do count as causes of their actions (at least their deliberate actions) insofar as they enter into their deliberations. Of course the fact that such a broad category of causation exists, as defined by the abovementioned formal criteria, in no way implies that there are no further crucial ontological distinctions to be made between the relations which fall within it. It is a common but fatal mistake to pass directly from the fact that deliberation 96 QUALITATIVE SOCIOLOGY produces or causes action in this broad sense, to the conclusion that the causation of deliberate action is to be conceived as of the same kind as the mechanical causation of the physicist. But this inference requires an additional, and dubious, premise: that there is only one kind of relation which is defeasibly determinative, and that that relation is the relation of mechanical causation. There is a particularly crude version of this mechanistic model of action which once had some appeal: according to it, felt desires were to be viewed quite straightforwardly as mechanical pushes and tugs upon the organism's body; the animal moves in the direction of the strongest pull. This picture is clearly hopeless. ~ Hence the more sophisticated strategies of reductive materialism and of epiphenomenalism. According to these, all the complex events of mental life are identical to or nomologically dependent upon the causally determined sequence of events transpiring in our neural systems. I shall not attempt to explore the multitude of issues which these strategies raise. The relevant point here is that desires and motives, though not phenomenologically presented to us as spatially directed tugs and pushes, are on these strategies to be explained in terms of physical forces operating at the more subtle and hidden level of physiology. Now one would be hard pressed to deny that physiological mechanisms are explainable in physicalistic terms, and equally hard pressed to deny that these mechanisms are intimately related to our mental lives. Unfortunately the question of just what this relation may be is one which I shall also have to set aside, if only for reasons of space. We must at the same time admit that so long as it remains unresolved, a crucial element in our understanding of human action is missing. This fact must not, however, prevent us from asking whether there are modes of explaining and understanding human action which rest upon a firm empirical foundation even though we may lack an answer to the above question. I say modes of understanding because human action is multifarious in character. There are ramified distinctions between deliberate action, voluntary action, involuntary action, acting from habit, impulsive action, and so on. There is a whole panoply of characterizations bearing upon explanation which the psychologist and social scientist may have to be sensitive MUST SOCIOLOGY BE QUALITATIVE? 97 to. This discussion will focus on just the crucial case of actions which result from conscious and deliberate choice. I believe that the most centrally interesting feature of the conceptual framework in terms of which we characteristically understand deliberate choice is that this framework entails that such choices are free. For the point of deliberation is precisely to make a choice between what initially are or are believed to be genuine alternative courses of action. If the performance of an action were known by an agent to be causally impossible for him, then that action could not enter his deliberations as a viable option; if all actions but one were known to be causally impossible (and that action were known to be causally necessary) then the point of deliberating would be entirely lost. And yet this conception of free will as action resulting from deliberation is quite compatible with the facts that (a) the desires and beliefs which form the "input" to an agent's deliberation may be antecedently determined by causally governed events, (b) these inputs in some sense determine the outcome of the deliberations (else the deliberations would be pointless for a different reason), and (c) the agent must have control over his actions at least to the extent of being able to determine them on the basis of (the outcome of) his deliberations. Some philosophers have actually used (c) to argue that the truth of determinism is a necesary condition for the exercise of free will (Hobart, 1934). But this does not follow. The crucial question is whether the kind of determination mentioned in (b) is correctly described as causal determination, in the strict sense, and as I see it, it is not. To argue this claim, it is necessary to become somewhat clear about what causal determination is, clear enough so that we can see whether we are familiar with any kind(s) of determination which is not of that sort. The dominant traditional way of understanding causation has been the Humean one in terms of constant conjunction. The glaring defect of this analysis is that it cannot capture the idea of the necessitation of the effect by its cause. In fact, the tradition rejects this idea .-- incoherent, attempting often to replace it with the notion of natural law--the latter so understood, allegedly, as not to covertly import any notion of causal necessity. 3 According to this picture, we cannot explain the occurrence of any event by appeal to its cause unless we have 98 QUALITATIVE SOCIOLOGY adequate grounds for believing there to exist some universal law or laws which, conjointly with the statement that a preceding event of a certain type occurred (the "cause"), entail the occurrence of the event to be explained. As a model for the explanation of human actions, this must come under suspicion for two reasons: (1) there do not seem to be many (if any) true universal generalizations connecting human agents' desires and beliefs to consequent actions, and (2) it is clearly not the case that in order to understand one's own actions one (in general) needs to know anything at all about universal generalizations regarding human psychology. The philosophical literature has been primarily concerned with the first objection. But I wish to pursue the second one, which seems to me to be more decisive. In fact this second point will lead one directly to seeing the difference between causal determination and determination by reasons. The fact is that, in order to understand a conscious and deliberate action of my own, I need to have nothing over and above the very understanding of my situation and deliberation which led to my decision to act in that way in the first place. 4 Retrospective reflection may of course shed some new light on my action, but it need not. The crucial mode of understanding is the very understanding by virtue of which I was able to deliberate and choose as 1 did. And for me to understand someone else's deliberate action is for me to engage in a similar kind of understanding of those deliberations which were performed by him. This, I take it, is what Weber had in mind when he spoke of Verstehen, and it is what Collingwood (1946) meant by construing historical understanding as a kind of (mental) reenactment. Both Weber and Collingwood were right, I believe, in insisting upon the superfluity of any knowledge of universal laws in sociology. 5 But interestingly enough, one might, if one is willing to reject the Humean analysis of causation, hold a similar view with respect to physical causation. Not that causal relations do not imply the existence of universal laws [pace Anscombe, 1975), but that knowledge of these laws need not be a precondition of our knowing of the existence of causal relations. For if one held that the existence of a causal relation can be ascertained in some single instances of causation (i.e., by direct perception) then one could maintain that knowledge of the causal explanation of an event in such MUST SOCIOLOGY BE QUALITATIVE? 99 instances was independent of the kind of repetitive evidence required to inductively establish the existence of a law. Now the relevant point is that it is open to someone who holds such a view of causation to argue that knowledge of psychological laws is not a prerequisite to causal explanations of human behavior because precisely in those cases where (at least our own) actions are determined by desires and beliefs via deliberation and "choice," we are confronted with a causal relation which we are able to perceive "directly." Since we are in our own case directly acquainted with the causal relation, and can by analogy infer its existence in other rational agents, we do not need to supply our understanding with whatever general law(s) might be entailed. And yet, so the argument would go, the perceived determination of our actions clearly is causal. Or is it? I should agree that we are directly acquainted, in our own case, with the "connection" between input and choice in conscious deliberation; we are acquainted with the way in which our reasons incline us or determine our choice. But to see whether this "way" should be assimilated to physical causation, we shall have to examine it more closely. The first essential point is that the "mechanism" of deliberation hinges upon the propositional content of the premises which go into that stretch of ratiocination. More explicitly, the direction of our reasoning is determined by the logical relations between the propositions which are the objects of our thought. It might be supposed, therefore, that the kind of determination involved in (rational) deliberation is that of logical necessitation or validity--the only species of necessity recognized by Hume. But this is surely incorrect. The propositions in question will indeed enter into logical relationships, independently of whether ! entertain them and recognize those relationships or not. But my having entertained the premises of an argument in no way makes it logically necessary that I shall arrive at and accept a correct conclusion (I might, for example, be killed beforehand). It is this temporally datable event of reaching a conclusion which is to be explained. It seems once again, therefore, that we should invoke a (defeasible) nomological relation between my entertainment of premises and my acceptance of a conclusion. Nevertheless, I think this would be a mistake, for the defeasib[e determination in question is neither logical nor 100 QUALITATIVE SOCIOLOGY nomological. The above appeal to the propositional content of thoughts does, however, contain a crucial insight. This crucial insight concerns the intrinsically intentional character of thought--and in our case, of deliberation. To understand one's own deliberations or someone else's, it would not be enough to know of some universal law connecting the input with the conclusion reached. For such a law could be imagined to connect any input with any conclusion, without damage (if it were true) to its explanatory force. The explanatory burden of a law is carried by its universality, not by the propositional content of input and conclusion, where these are the relata. But it is precisely in terms of the intentional objects of our thoughts, the propositions that are thought, that we understand the passing from input to conclusion. This passage is not logically necessary, as the making of mistakes in reasoning shows, but it must be understood in terms of a recognition of logical relations. This r e c o g n i t i o n is the intrinsic, and phenomenologically accessible, feature of reasoning which drives it non-randomly forward. It is not like physical causation? Because reasoning relies upon the recognition of logical relations between its objects, it is appropriate to apply to it a whole panoply of normative terms which cannot be applied to causally produced phenomena. We speak of a conclusion, but not an effect, as correct or incorrect; we speak of reasoning, but not causation as good or bad, valid or invalid. On the Humean account, we never have any experience of causal necessitation. This by itself would, if true, be sufficient to distinguish causal relations from the defeasibly determinative process of deliberation if, as I maintain, we are directly acquainted with that. However, I am not convinced by the Humean position. I shall not of course launch a fullscale siege upon it here. But I think I can point, as other philosophers have (Mace, Stout, Ewing, 1935), to that element of direct experience which seems to me to be constitutive of our phenomenal awareness of causation. It is our experience of physical force upon ourselves. That experience seems to me to bear a number of features which together yield the defeasibly determinative character of causal relations. The central feature, perhaps, is the vectorial character of our experience of force: it has a direction in space, and a magnitude. These characteristics are inextricably bound to MUST SOCIOLOGY BE QUALITATIVE? 101 our apprehension of forces as determining movement. The effects of forces are not spatially random but directed in the direction of the forces themselves. The magnitude of a force is similarly connected to the magnitude of the effect. This is not, I should maintain, the result of an induction from repeated experience of force. But a further feature of felt forces is crucial to our account--namely the property they have (which we can directly experience) of adding together in accordance with a vector algebra, if simultaneously appliedF It is this which tallies with the defeasibility of causal laws. For a force-law which relates a force to its effect presumes that, as we say, "other things are equal," that is, that there are no other forces unaccounted for in the initial conditions. But any actual application of such a law is generally subject to the possibility that additional forces may be present, and will have to be added in. It is from felt forces on our bodies, I suspect, that we extrapolate to physical interactions at large. We can now state the point of this digression into the phenomenological character of our experiences of certain singular causal relations. It is simply that these experiences display a completely different phenomenological character from our experience of the way reasons determine action. In the end, we perhaps cannot really describe either kind of experience but must simply point to it. But I think it is enough to show that our understanding of the experience of physical force is closely linked to the notion of spatial vectors, whereas our experience of deliberation is not associated with this but with the notion of logical relations obtaining between the propositions contemplated. These, then, are my reasons for maintaining that causation and reasoning, while both species of defeasible determination, are distinct. It will by now seem that we have strayed very far from the everyday concerns of the working sociologist. Not so, as I hope to show in some brief concluding remarks. We may take as our point of departure a suggestion made by Rescher and Joynt (196t) that, although there may be no universal laws of human history, we can expect to find laws which apply restrictively to given cultures or societies. Essentially the idea is, when in Rome, Romans do as the Romans do. More formally, we should look for laws of the form: Any member of culture C who finds himself in situation S will do A. 102 QUALITATIVE SOCIOLOGY Or, more realistically, for statistical or probabilistic laws of the form: If x is a member of C and in situation S, then the chances are P that he will do A, It seems entirely reasonable to suppose that the sociologist might discover such statistical generalizations as these. Indeed Joynt and Rescher (1959) suggest that the statutory laws of a society will produce such regularities. I can see no reason why such statistical regularities might not be of the greatest interest in informing us of the character of a society which is under study. In many cases displaying data in some well-chosen statistical form may make relationships leap to the eye which would otherwise not have been apparent. But the question before us is whether, as Mill (1959) thought, the existence of such society-relative statistical regularities can be taken as evidence for underlying causal laws. As will perhaps be clear by now, I think it cannot, or not always. Such regularities are indeed evidence for something. They are evidence for the existence of processes underlying the course of social events, processes which are not purely random, which are, in a word, defeasibly determinate. But as we have seen, this is not sufficient to establish the existence of causal laws. For as we have seen, there is at least one other candidate for such a defeasibly determinative relation, and it is clear that this alternative must be relevant to the explanation of most social statistics. 8 There is no space here to discuss the interesting question whether human rationality and strictly causal laws are together sufficient to explain all the phenomena met within history and sociology. But clearly both these kinds of underlying mechanisms are at least relevant. Because people's reasons determine, in a defeasible way, what their actions will be, we can, knowing their reasons (and their character), forecast their actions with a fair, often very high, degree of probability. Conversely, it is precisely this relation between reason and action which enables us to explain actions by appeal to reasons. Given a statistically large sample of people with similar reasons, this explains why we may expect statistical uniformity in their behavior, provided MUST SOCIOLOGY BE QUALITATIVE? 103 that other relevant variables (character, cultural traditions, intelligence, etc.) are either held fixed or suitably randomized within their respective ranges. An obvious example which satisfies these conditions is the sort of case appealed to by Joynt and Rescher (1959): the behavior of a nation's citizens vis'a vis its laws. Assuming cultural uniformity, we have a situation in which that variable is held fixed, intelligence and intrinsic temperament are averaged over, the risks of detection and sanctions imposed by the legal system provide reasons against law-breaking, and whatever gains are to be had from illegal activity provide incentive for it. To the extent that risks outweigh benefits, we may expect people to be reasonably law abiding--provided that both the risks and benefits considered are sufficiently obvious. The same reasoning applies to the sorts of statistics Mill cites in support of his historical determinism. The number of murders, suicides, and marriages each year (assuming a constant population) will all be predictably influenced by the way in which the social climate imposes or tends to impose incentives and stumbling-blocks with respect to the acts in question~ But it is (in most cases) a chain of reasoning, I maintain, which leads from the imposed situation a person finds himself confronted with to action. In this sense, his actions, if deliberate, are not caused, though they can contribute to statistical regularities. Thus, if by qualitative sociology we mean the study of those aspects of social phenomena which are the distinctive products of human consciousness and thought, then I believe we can sustain the claim that social science requires distinctive methods of investigation, and a mode of understanding not found in the physical sciences. But this is in no way incompatible with the use of statistical or other techniques which make use of mathematical analysis. Insofar as they are applied to human consciousness and its products, the information these techniques produce is amenable to being understood and explained in terms of the categories of Verstehen. This means, if I am right, that it is the proper business of sociology to uncover the reasons why people think and act as they do, insofar as this is relevant to the formation and perpetuation of social arrangements. Statistical uniformities can reveal much about such patterns of reasoning (among other things) and the extent of their influence. No single agent's reasoning will encompass the entire range of circum- 104 QUALITATIVE SOCIOLOGY stances relevant to a given problem requiring social action, so much that occurs will involve factors unknown to him or to anyone and beyond control. Sociology can make us more aware of such factors, but it cannot hope to ultimately capture its subject matter in a net of laws. Nevertheless, this does not mean that sociology must forfeit the project of explaining what goes on, nor that of forecasting the future. My conclusion is simple. It is that qualitative sociology (in the last relevant sense distinguished at the outset) and statistics (properly understood) can live together peacefully, even happily ever after. NOTES ~Some philosophers, e.g., A.I. Melden (1961) have argued that the connection does involve logical necessity, since an action cannot be relevantly identified except in terms of the prior intentions of the actor. Was it a left-turn signal? Or merely a pointing to a butterfly? A physical description of the motion of the arm does not suffice to determine which. Nevertheless, there is no logical contradiction involved in asserting that A intended to do (an action of type) X and that A did not do X. For it is possible--not merely logically but physically possible--to have the intention but fail to do X. One might {accidently) do Y instead. Or one might fail to perform any action as a result of one's intention (a bullet in the brain might intervene). So intending to do X is neither a necessary nor a sufficient condition of someone's doing X. 2Balaam's ass, as the facetious medieval story had it, starved to death when placed midway between two equally succulent bales of hay. By extension, if he had been placed elsewhere on the perpendicular bisector of the line connecting the bales, he would have become a damped harmonic oscillator, moving along that perpendicular and expiring in the same spot as before. 3There is some consensus now among philosophers that no analysis yet offered of the concept of a law of nature which abjures the notion of nomological necessity has been successful. ~Thus the knowledge that my behavior conforms to some universal behavior pattern is immaterial to personal understanding of my action. This is somewhat imprecise. One could, for example, ask about the physical details of the process whereby the decision to act is translated into action--i.e, into bodily movement. But to understand the making of the decision itself we need nothing more than is alleged, at least in unproblematic cases not involving irrationality or other abberration, SWeber's (1977) central argument for the distinction between Verstehen and the causal description of nature is, however, a very bad one in my opinion. Weber's argument is based upon the idea that any concrete event or event sequence is infinitely complex, that no finite description can exhaust all of its features. Laws, however, relate events only insofar as the,,, satisfy some finite characterization, and hence laws inevitably involve abstracting away from full concreteness. Weber sees the method of Verstehen as moving toward grasping particular event-sequences in human affairs in their full concreteness, which involves a movement away from abstraction. This is correct in its denial that repetition is the key to understanding human affairs. But the causal determinist can and would easily deny that causal laws fall short of determining concrete events, no matter how wide in complexity. It is just a matter of finding all the constituent forces and component particulars relevant to answering any question that might be asked about a system. MUST SOCIOLOGY BE QUALITATIVE? 105 6This object-directedness of thought explains why, in my own case, I will not appeal, as I might in explaining another's action, to the contingent fact that I had such and such thoughts. Rather, in explaining {and justifying) my action, I point to the content of the propositions themselves. Someone who did not understand these propositions and their logical relations would fail to understand my action as I understand it, whether he possessed knowledge of some suitable laws or not. 70f course sophisticated inferences are required to obtain the general mathematical form of this algebra. ~Which is, of course, not to deny that causal necessities may also be relevant. If there are certain regularities in the behavior of people in times of famine, these will inevitably require, as part of their explanation, those causal laws in accordance with which lack of food leads (statistically) to feelings of hunger, poor health, and enervation. But human rationality (and irrationality) will also play a role. REFERENCES Anscombe, G.E.M. 1975 "Causality and determination." Pp. 63-81 in E. Sosa (ed.), Causation and Conditionals. New York: Oxford University Press. CoIlingwood, R.G. 1946 The Idea of History. Oxford: Clarendon Press. Davidson, D. 1963 "Actions, reasons and causes." Journal of Philosophy 60:685-700. 1979 "Mental events.*' Pp. 218-238 in T. Honderich and M. Burnyeat (eds.), Philosophy As It Is. New York: Penguin. Hobart, R.B, 1934 "Free will as involving determination and inconceivable without." Mind 63; 1-27, Joynt, C.B. and N. Rescher 1959 "On explanation in history." Mind 68:383-87. Mace, C.A,, G.F. Stout, A.C. Ewing 1935 "Mechanical and teleological causation." Proceedings of the Aristotelian Society. Supp. Vol. 14:22-82. Melden, A.I, 1961 Free Action. New York: Humanities Press. Mill, j.S. 1959 "A System of logic." Pp. 85-86 in P. Gardiner (ed.), Theories of History, New York: Free Press Rescher, N. and C.B ]oynt 1961 "The Problem of uniqueness in history." History and Theory 1:150-62. Weber, Max t977 Critique of Stammler, translated by Guy Oakes. New York: Free Press.