What Sort of Theory Should We Use?

If we accept in principle that more social theory would be a good thing, we can go back to my four tendencies in theory and consider what might be gained from each of them.

There are two obvious difficulties with normative social theory. The first problem is that, while taken in isolation any one such theorist may be inspirational, it is hard to construct rational grounds for choosing between competing visions of the good life when there is more than one of them. For social scientists working within a generally positivist epistemology, the value of any theory is determined by how well it is supported by the available evidence and how well it survives repeated attempts to refute it. Some normative social scientists are pro-cake and pro-the-eating-of-cake. Classically Marxists claimed that their work was both thoroughly scientific and better than bourgeois science because it augmented description and explanation with radical praxis. But the cavalier treatment of evidence in much normative work generally shows disdain for the scientific method and makes it clear that understanding some past or present reality takes second place to promoting extra-social scientific goals or values.

Once one rejects conventional notions of evidence, it is not clear how one chooses between competing normative theories, unless one accepts the theorist's norms, which means one has already chosen. I would like to say that I am withholding his identity to spare his blushes. Actually the author in question had so little impact on the field that I have forgotten his name but I very clearly recall one 'critical criminologist' (critical in both the Marxist and the complaining sense) dismissing the work of the great American criminologist Edwin Sutherland with the rhetorical flourish of asking 'what has Sutherland ever done to promote popular struggles ?'. Like Marx he clearly believed that the purpose of theory was to change the world rather than just explain it. But as any number of us are unlikely to agree on what would count as positive change, political rectitude is no help here.

Rather flippantly, I sometimes think we should judge models of the good life by how well their promoters exemplify the virtues of the good society in their personal lives. The record is not impressive. Karl Marx was clearly a sponger and arguably a hypocrite. Walter Benjamin and Nicos Poulantzas committed suicide. Felix Guattari was a philanderer who bullied his wife into having affairs so that he didn't have to feel guilty about his own infidelities. Louis Althusser did more than bully his wife: he murdered her. These are not people you'd invite round to your house so why look to them for instruction on the nature of the good life? To be more serious: the problem with normative or critical social theory is that it offers no principles that are agreed outside that theory's magic circle for distinguishing between alternatives. Old-fashioned social science as science has agreed rules for choosing between alternatives. Critical theorists reject those rules but cannot agree amongst themselves about alternatives.

My second difficulty is that I cannot see how the majority of the research that is done by sociologists of religion would be much improved by normative social theory. It often seems that normative theory is used more in order to establish one's loyalties, to show affiliation, than to cast new light on empirical material. Even if one thinks there are good political or ethical reasons to take sides, it is hard to see how being on the side of the poor, or of women, or of the subjects of one's fieldwork, helps develop accurate description and plausible explanation.^[1] A case could be made for saying that some degree of sympathy or rapport is required for understanding one's subject or subjects but beyond establishing that research may be hindered by certain attitudes or relationships, it does not offer any guide to the conduct of the descriptive and analytical parts of the research process.

It is even more difficult to see the value of normative social theory for partisan or apologetic studies of religion. If the purpose of one's academic work is to promote a particular religion, there seems little point in wasting time citing some non-religious thinker whose vision of the good life is coincidentally compatible with the value position one is promoting. One adds some apparent authority but the bottom line remains what is always was. In any partisan borrowing from social science, the secular works of people are so obviously trivial compared to the Word of God that there seems little point in mentioning them beyond a little dishonest impression management.

I have as much difficulty seeing the point of Zeitgeist metaphors. It is somewhat ironic, given that the theorist who is in the business of coining some telling new metaphor clearly thinks that he or she is uniquely capturing the essence of modern times, that such metaphors often seem interchangeable. Any reader of academic journals will be familiar with the sort of article I have in mind. The Zeitgeist metaphor gets a big puff in the introduction. There then follows detailed analysis of some small slice of human life which makes no reference to the metaphor. In the conclusion the metaphor re-appears briefly to add some theoretical weight to the substance of the study. In some cases I am not sure whether the novel research supports or refutes the zeitgeist metaphor: that may be a fault of drafting but more often the ambiguity is inherent in the fact that the phenomenon being studied is too small to offer any kind of critical test of a depiction of society as a whole. The lack of close fit between the zeitgeist metaphor theory and what is being studied also runs in the other direction. Just as the original research cannot test the theory, the theory can have so little particular address to the research that one suspects any zeitgeist could be replaced by any other without loss. So the study of the Snibbo sect that is presented as an illustration of liquid modernity could equally well be framed as an example of the network society or the deleterious effects of bowling alone or McDonaldization.

If that conclusion seems jaundiced, consider the short career enjoyed by such metaphors. Were it the case that fleshly-attractive depictions of the modern condition were of lasting benefit to empirical research, they should surely endure. But they do not. Twenty years ago Roy Wallis and I wrote a general review of the British contribution to the sociology of religion. In it we took issue with two esteemed colleagues, both of whom I am very pleased to say are still with us, literally as well as metaphorically. Bryan Turner's Religion and Social Theory (which can still be read with great profit) criticized the sociology of religion for not being involved in 'any major theoretical debate in modern sociology'. In particular we had failed to engage 'in neo-Marxist debates about modes of production and ideology, French structuralist discussions of subjectivity and power, and critical theory's discussion of knowledge, the state and legitimacy'

(1983: 3). James Beckford made the same point in his Religion and Advanced Industrial Society when he said that 'the sociology of religion has been intellectually isolated against, and socially isolated from, many of the theoretical debates which have invigorated other fields of modern sociology' (1989: 13). The important theorists we should have been attending to were, according to Turner, Louis Althusser, Nicos Poulantzas, Jurgen Habermas, Barry Hindess and Paul Hirst. Althusser and Habermas also appears on Bedford's list, which also contains Michel Foucault, Antonio Gramsci, Alberto Melucci, Claus Offe and Alain Touraine. Twenty years later, probably only Habermas and Foucault are still read and then not often.

If we could be confident that the decline in popularity of theorists was a consequence of refutation, it could be claimed as a mark of the health of a discipline. However, I suspect that Zeitgeist metaphors are, like butterflies, doomed to a short life irrespective of their intrinsic merits because their success depends precisely on their novelty. Like every other sort of fashion, changes in this realm of social theory seems driven not by testing through some Popperian process of conjecture and refutation but by cohort replacement. Each generation of students is unduly attracted to something hip and novel and each generation thus dooms their heroes and heroines to be disdained by the next generation.

As with the value of normative social theory, we can add a secondary observation about the use of zeitgeist metaphors by partisan apologists for religion. As it seems likely that the writer's religious faith comes first and determines the selection of, and attitude to, social theorists, claiming that some secular image of the age confirms (or at least does not clash with) the faith that one is punting is simply a rhetorical device; little different to Billy Graham's fondness for decorating his sermons with references to this or that bit of recent scientific research which apparently showed that southern Baptist evangelical Protestantism was superior to all other religions.

The value of agenda-setting theory is clearer. Even if we do not believe that scholars are unable to transcend either the ideological limits of their upbringing or the current interests of their class, gender, national, ethnic or professional identity, we can accept that we can become blinkered. And here the outsider has an advantage. There is always value in asking new questions and a new set of interests will often cast new light on old problems. However, I would caution against two common dangers of agenda-setting theory: exaggerating novelty and exaggerating the defects of what is old.

As an example of exaggerating distinctiveness, I offer ethnomethodology, especially in its sub-field of conversation analysis or 'CA'. Harold Garfinkel and his students (particularly Harvey Sachs) made a great fuss of criticizing all previous sociology for relying on unproven assumptions about motives (Garfinkel 1967; Sacks 1963). Ethnomethodology was superior because it did not make contentious guesses about motives and intentions. And yet it seems clear to those who are interested in the social organization of talk, but are not CA partisans, that the most basic tools of conversation analysis do rely on guesses about actor motives and intentions. For example, a question and a request are distinguished not by their grammatical form but by the assumed intention of the first speaker in the adjacency pair. So the person who replies to 'Do you know the time ?' with 'Yes' will almost certainly have either made a mistake or be teasing, as will become clear when one or other repairs the fracture. Either the first speaker adds 'No, I mean "what time is it ?"' or the second speaker adds 'Ha. Only kidding. It's 3.30'. One could argue that the sorts of motives and intentions assumed in ethnomethodology are considerably narrower and more mundane than the causal connections assumed in conventional sociology (where, for example, we might interpret an interest in esoteric magic as evidence of status anxiety) and are thus less likely to be mistaken. But it remains the case that the research done within the ethnomethodology canon now seems far less novel than the paradigm shift advertised in the original programmatic statements.

The second danger of agenda-setting theory is the wrong solution to the baby-and-bathwater problem. Feminist social scientists are quite right to point out that until the 1970s much social science paid little overt attention to gender. That tells us what new work needs to be done but it does not tell us what to do about the old stuff. Mary Jo Neitz is clear that work done within the secularization paradigm is of little value: 'in telling the story of secularization in terms of decline, we tell a partial story of elite white men' (2008: 222). To the extent that we are influenced by any agenda-setting theory we can dismiss everything that has gone before or we can treat the effects of omission as a research problem in its own right and ask to what extent would a different set of questions have changed the research and its conclusions. I would argue that much of the explanatory part of the secularization paradigm stands up pretty well to a gender-minded re-examination. For example, the social-psychological claim that being exposed to a range of different religious perspectives creates problems of certainty and conviction for the believer is a properly universal claim. It rests on universal assumptions about the difficulty of believing one's own views to be uniquely correct when one enjoys pleasant and rewarding interaction with people who hold very different views. It may be wrong but, unless we are to suppose that women are naturally more (or less) anti-social, obstinate or dogmatic than men, it is not wrong because it fails to distinguish between the sexes. The same can be said for the proposition that effective technologies displace occasions for resort to religious solutions and thus reduce the presence of religious ideas. The illustrations that we could draw from the life experiences of women and men are different (as are the examples we would draw from different social classes) but the key principle - that religion loses authority as the range of events for which it appears to offers the best solution decreases - seems genuinely universal. Furthermore, gender differences in outcomes may be perfectly well explained by universal propositions. The universal proposition that rewarding social interaction with people of different faiths weakens dogmatism fits well with the observation that generally male church adherence declines before that of women: in many settings gender differences in domestic roles and in employment patterns meant that men were more likely than women to interact with strangers and to perform jobs that required the sublimation of private preferences to religiously-neutral public roles.

To summarize, there is obvious value in agenda-setting theory which alerts us to omissions and identifies new research problems but we should beware of exaggerating both the virtues of the new and the vices of the old.

Finally I come to sociological explanation. Not surprisingly, as a sociologist who just happens to study religion, it is in general sociological explanation that I find a body of knowledge from which students of religious phenomena can borrow with advantage. Before introducing an example, I will introduce a caution which takes us back to my introductory point about the downside of inter-disciplinary bricolage.

Within any discipline, concepts and observations generally have a history of elaboration and testing that means that, in borrowing some notion from an unfamiliar field, one may impute to an authority which it does not have within its home discipline and hence which arguably it does not deserve. Theologians will doubtless be horrified by the mess I have frequently made of using theological concepts. But my concern here is with arts and humanities scholars borrowing from the social sciences. The point is best made with a detailed illustration. In their deservedly well-known study of alternative spirituality and conventional religion in the small English town of Kendal, Paul Heelas and Linda Woodhead (2005) explain the appeal of New Age spirituality as a reaction to the frustrations caused by what Max Weber called 'the iron cage of rationality': a popular headline idea. That explanation suffers the flaw of there being a major discrepancy of scale between those of us whom Weber thinks are trapped in the iron cage of rationality (which is pretty well everyone) and those of us interested in alternative spirituality (which, by their own data, is less than 1 per cent of the population). But arguably it also suffers from giving too much weight to an idea that sociologists have long since questioned. Garfinkel (1967) on good reasons for bad organizational records, Donald Roy (1959) on time and task management on a production line, Melville Dalton (1959) on personal relationships within complex organizations, and Jason Ditton (1977) on workplace fiddling; these and many others have raised important questions about the nature of bureaucratic rationality. We now know that even in the most straitened and bureaucratized circumstances, even relatively powerless people find creative ways of manipulating rules and procedures to serve their own local and immediate interests. And one of those interests is the re-gaining of the autonomy supposed lost to the iron cage. One may be skeptical of the possibility of social science research emulating the natural sciences in the gradual improvement of knowledge through theory-testing research but this example is offered as an illustration of the general observation that inter-disciplinary borrowing will be most beneficial when an idea is followed through its natural history rather than simply lifted out of a disciplinary context.

Now let me come to the positive. There is a great deal of sociological theory that can greatly improve our studies of religious phenomena but it is mostly to be found, not in the sweeping generalizations that appear in the social theory textbooks, but in next layer down: in the work of empirical sociologists who present their research findings in general and comparative terms that invite extension to other fields. What used to be called 'theories of the middle range' and is now called 'analytical sociology' (Demeulenaere 2011) or 'social mechanisms theory' (Hedstrom and Swedborg 1998) encompasses a very large body of pertinent sociological theorizing. To illustrate the claim I will describe one body of social scientific work which, I would argue, has been neglected by students of religion. To be more precise, there is no doubt about its neglect; what needs to be argued is its value for students of religion.

In stark contrast to those social theorists who are frequently cited but who have actually made little difference, Robert K. Merton is little known outside the profession but he was responsible for coining a surprisingly large number of terms and phrases that we now take for granted because they proved useful in research: the self-fulfilling prophecy, the reference group, and role strain are just three reminders of a long and influential career. Merton was introduced to communication research in the 1940s by Paul Lazarsfeld and colleagues, who were working on voter selection in a presidential election (Lazarsfeld et al. 1944). Merton researched a US war bonds radio sales campaign and the results were published as Mass Persuasion (1946). That slim volume included a thought-provoking section on the creation of 'pseudo-gemeinschaft' that was later much anthologized. In association with Elihu Katz, Lazarsfeld went on to develop the 'two-step flow of communication' thesis (Katz 1957; Katz and Lagerfeld 1955), which argued for the importance of personal relationships in the mediation of effects of the mass media. The rural sociologist Everett M. Rogers studied the spread of new seed varieties and developed that work into The Diffusion of Innovations (1962), which remains one of the most cited works of social science. What Merton, Katz, Lazarsfeld and Rogers had in common was an interest in how people are persuaded to change their ideas and practices and what is particularly valuable about their work is that it does not radically separate the social structural elements of persuasion from attitudes to the ideas themselves. A persistent problem of much sociological treatment of religious change is that it tends to present causal explanations of conversion in a way that implies either that people are dopes (they convert because some social force makes them) or that they are cynics (they convert because they wish to achieve latent or secondary functions such as finding a new 'family' or achieving social advance). The great advantage of the two-step flow of communication and the diffusion of innovation work was that it treated the 'plausibility' of new ideas and practices as complex function of the interaction of social forces, individual preferences, interests, and the use value of innovations. The implications for the study of religion conversion and, in the wider sense, the growth and spread of new religions would seem obvious but I am hard pushed to recall any student of religion using it. The same could be said for the sociology of science: a body of research in which British academics have achieved international prominence and which is concerned with many of the questions which feature in accounts of contemporary religious change, but which is so rarely cited I cannot think of an example.

To summarize my main point, we can simplify my opening four types of social theory to one axis: at the high end we have abstract theorizing about the nature of society that is so removed from empirical support that it comes close to sealed room speculation; at the low end we have detailed sociological explanations of areas of human life. While the former can offer some inspiration and even orientation, it is the latter which offers the greatest benefits to the empirical study of religion. The best sources of explanations of the social structural elements of religious change are not texts which aim to characterize the nature of modernity in a few phrases; they are detailed studies of putatively similar secular behaviour.

• [1] For the debate in the sociology of deviance, see Howard Becker (1967) and the debate it occasioned. Hammersley (2000) offers sensible reflections on the debate.