I have defined 'sociology' in pretty broad terms here. There is much discussion about the boundaries of the subject, and the distinctions and overlaps with other social sciences such as psychology or economics. Sociology also shares boundaries with much more general approaches to social life, general theories or philosophies. The material under 'non - empirical' approaches gives some clue to the debates that go on here. Although this material is classified into broad groupings, the intention is to not patrol these boundaries too tightly, and not define sociology too narrowly. The focus on methods makes this broad focus particularly relevant, since there are clearly some techniques for gathering data which are shared by different social sciences such as sociology and psychology.
One general division which we reproduce in this collection is between quantitative and qualitative approaches. Again, for some texts, this is a really deep and significant division. It was a major issue in the 1970s and 1980s, for example. Quantitative approaches were seen as completely different from qualitative ones, on a number of important axes or dimensions. It was usual to talk about major epistemological and ontological differences between quantitative and qualitative work. These issues are still prominent in some of the older texts.
Drawing upon some major philosophical debates, often enough German ones, (such as the positivist dispute or the 'methodenstreit') it was possible to suggest that the distinction worked with quite different assumptions about what counted as proper knowledge and social reality (epistemology and ontology respectively). There are several ways to address this issue, but a fundamental aspect of the problem was whether human beings could be studied with the same kind of scientific knowledge as was common in the natural sciences.
Early social sciences, including the work of J S Mill, suggested that human behaviour could indeed be understood as having been caused by a variety of external events, just as, say, the trajectory of a billiard ball is the result of complex combinations of forces. In one of the classic formulations of positivism, Comte suggested that sociologists should attempt to build up a body of knowledge about which events caused which behaviours, and eventually even develop general laws from these observations. Social science would then emerge with all the power and clarity of natural sciences.
Moving from the individual to social and group phenomena, Durkheim's work seemed to agree, especially in his much-quoted notion that social events are caused by social facts which the sociologist should study empirically (see link for a brief intro) . There is even a version of Marxism, of course, which talks about the general laws of history, including the one that predicts that the rate of profit must fall as capitalism advances. More generally, Marx and Engels once clearly thought that they were on the verge of a new materialist social science which could predict social and political formations on the basis of what was known about economic development.
For those on the other side of the epistemological divide, these formulations, and terms like 'cause', 'law' or 'fact' could only be metaphors at best. Human beings were not like billiard balls because they had a level of consciousness that made them aware of the world in a unique way. They interpreted events impinging on them, and were able to define them linguistically in ways which permitted communication and discussion among themselves. Similarly, at the group level, 'social facts' have to be interpreted, agreed and understood before they become socially effective, and political action is far more a matter of consciousness and agreement than some sort of automatic response to economic changes.
It is clear that a number of implications spin off from this basic argument, certainly for methods. If human beings are responding to external events without necessarily being aware of them, social science becomes a matter of trying to uncover social events and social processes and measure their effects. The classic way to do this to study social patterns: if the rate of suicide rises in particular urban conditions, to cite Durkheim's classic work, then there is something about those urban conditions which is predisposing people to suicide irrespective of their will. Durkheim suggested that we might literally map areas with high suicide rates, and see if we could compare those to maps of particular urban conditions. In modern social science, a whole range of research techniques has developed to try to indicate social patterns and then to explain them, classically using social surveys and statistical analysis -- the familiar quantitative methods that we are going to chart. Few quantitative researchers would want to reconstruct the big project to find social causes and laws, and are usually content to talk about probabilities and correlations.
If, however, human beings have to become aware of these events before they can have a social or individual effect, then there is a need to somehow study human consciousness and how it works in particular individuals or groups, and to abandon the formal vocabulary of science with all its talk of causes and variables. I say 'somehow' because studying human consciousness is almost by definition deeply difficult and paradoxical. It cannot be observed directly, for example, and must be studied through external manifestations such as words or actions. Similarly, if human consciousness is central to understanding, then the researcher must also be centrally engaged in interpretation and cannot pretend be objectively describing events from the outside. One much debated implication is whether sociology can be 'value-free'.
Qualitative methods therefore characteristically take the form of engaging in conversations with human beings or observing them, attempting to understand. Actual techniques have ranged from the attempt to build abstract models (e.g. 'Economic Man') to predict the understandings and actions of others, through various interviewing and ethnographic techniques which aim to see the world as the participants do thermselves, to personal disclosures and reflection which attempt to show in detail how the researcher's consciousness works itself. Again, the material indicates a range of these techniques.
I have very briefly sketched some of the basics of the debates that have raged in sociology for well over a century. Although I am no expert, they seem to have raged in psychology as well, with very similar debates establishing boundaries between positivistic behavioural approaches (behaviourism, biologistic psychology, evolutionary psychology) and more humanist approaches (like those developed by Freud or Rogers).
In sociology at least, as well as the methodological issues, there are clearly theoretical issues and political issues involved as well. Occasionally, these have clustered together to produce deep and apparently permanent divisions between the approaches, so much so that some commentators have talked about 'two sociologies'. A-level Sociology texts in the UK still operate with two or three major and apparently irreconcilable 'perspectives'
In one such episode of division, in the 1970s and 1980s, many sociologists would line up quite clearly against the use of anything that smacked of 'positivism' in social sciences. This was partly due to the emergence of new qualitative methodological perspectives which had become codified. At first, these different approaches were simply collected together, from German theorists especially Weber, and from a fairly different American perspective, usually called American pragmatism, or, specifically, symbolic interactionism. Sociology itself was emerging as a popular discipline, with a considerable readership in schools and universities, and, as is the way of textbooks, divisions were hardened and simplified in order to engage relatively inexperienced readers. At the same time, sociology was in the forefront of a number of radicalized university subjects. Positivism was rejected as offering an uncritical support to the status quo. At its most extreme, positivist sociologists and economists had been employed directly by the US government in order to gather information that would help them destabilize various left-wing regimes, as in the notorious 'Project Camelot' (Horowitz 1967). Positivism was also seen as insufficiently critical of the burgeoning commercial and capitalist society developing in the US and Europe (Marcuse 1968). In one tradition, Marxist thought, once re-read, (and later, radical feminist thought) was a more suitable alternative, but qualitative approaches were also seen as offering critical insights into the resourcefulness and centrality of human subjectivity: a humanist politics, especially one celebrating 'the underdog' (Becker's claim in Becker 1973) seemed preferable.
The division had hardened in that period, but a kind of truce seems to be much more characteristic of the last decade or so. The radicalized generation has retired. A scholastic drift has rendered political implications far less relevant to academic business as aociology has become much more professionalized. Empirical research as a professional activity has increased in importance. Funded research in particular has encouraged a much more pragmatic compromising approach, which uses a variety of methods to claw back the most data available for the money. There is also a belief that major funding bodies are much more likely to fund quantitative research which produces impressive tables of 'objective' data and statistics. A kind of division of labour seems to have become fairly accepted in much contemporary research, with large-scale quantitative work supplemented by small-scale qualitative detail.
Of course, there have been theoretical and methodological reasons for this new multi-method approach as well. A range of social theorists have proposed various theoretical models which claim to draw the best from the different classical theories on offer. Of more immediate relevance, however, is the realization that different methods have different strengths, and even that they share the same dilemmas and procedures, whatever the differences on the surface. To take the most obvious examples, discussed very well in Bryman (2001), no sociological study offers just pure statistical description without an attempt to explain the social or personal significance of the findings, to discuss the meaning of the results. Similarly, it is almost impossible, even in the purest of ethnographic studies to avoid generalization, or even basic forms of quantification such as 'most respondents found that...'
There has even been some rallying around what might be called a weak positivist approach, in the face of serious challenges to the whole discipline coming from a variety of philosophical and cultural positions, most notoriously and recently postmodernism. Sociologists are recovering and remembering some of the earliest arguments from Mill and Comte in favour of positive knowledge, as opposed to endless airy speculation, of finding out things, as opposed to sitting in in armchairs producing witty commentaries on television programmes. The same spirit produces a tendency to crack on with developing actual research programmes rather than worrying too much about 'paradigms' ( for a discussion from Kuhn himself, who introduced the term into modern debate, try this link)
For me, these points lead to a general remark about these divisions and others found in discussions of sociology and other social sciences. My general point is that none of them are absolute but necesarily refer to their opposite, and they all take place in a context of scholarly debate, but with a keen eye to political social and professional relevance. What energises the debate among qualitative researchers is their attempt to distinguish themselves from quantitative ones rather than to found some pure new 'paradigm' -- and vice versa. Empirical work is useful way to stave off the claims of philosophers. Philosophical sophistication is a useful way to mark out a particular research programme and to claim high-status for it. In this sense at least, sociology is definitely similar to natural science.
A Brief Annotated Bibliography
As well as the techniques discussed and illustrated here, there are many examples of excellent research methods texts which pursue these issues in much more depth. Some of them are reviewed briefly below. In the manner of these things, this list is predicted to grow steadily, so check back a lot.
Balnaves, M., and Caputi, P. (2001) Introduction to Quantitative Research Methods: an investigative approach, London: Sage.
A useful discussion, not particularly maintaining rigid boundaries around quantitative methods, for example in emphasizing the creative diagnostic side of research. A rather irritating running metaphor of detective stories makes this point too. One unusual feature is to add the creative process of abduction to the usual discussions of deduction and induction. There is a basic checklist approach to guide practical work, and a very clear tutorial on SPSS on an accompanying CD Rom. One example is of the Australian study of taste, based on Bourdieu, which was the forerunner of the larger UK ESRC project.
Bryman, A. (2001) Social Research Methods, Oxford: Oxford University Press.
My current favourite, with its comprehensive discussion of methods and also some of the epistemological and ontological issues I have briefly outlined above. Thus the excellent 'practical' emphasis leads to excellent discussion in part 4 on the differences and similarities of quantitative and qualitative approaches. Bryman is particularly clear on the limits of thinking about different approaches as 'paradigms'. The excellent examples include discussions of sport and exercise.
Seale, C. (2004) (ed) Social Research Methods. A Reader, London: Routledge.
A very useful collection of some of the foundational debates, including theoretical and philosophical arguments as well as more immediate methodological discussion. Thus postmodernism appears, as does structuralism and more poetic interpretive approaches. There is a strong representation of linguistic and conversational analytic techniques. There are chapters on general issues of validity and on the actual history of research methods . There is a useful discussion of the paradigm issue. Actual examples cover the usual issues found in sociology, including gender, identity and health.
Becker, H. (1973) Outsiders: studies in the sociology of deviance,
Horowitz, I (ed.) (1967) The Rise and Fall of Project Camelot , MIT Press: Cambridge, Mass.
Marcuse, H. (1968) One-Dimensional Man, London: Sphere Books