A history of the secularization issue

The Search for a New Paradigm: Phillip Hammond and Jeffrey Hadden
The two writers I will discuss in this section are probably the best representatives of a broad movement inside American sociology of religion which seeks to move beyond the secularization paradigm. Hammond's contribution consists mainly in the edition of readers which aim at establishing an agenda for the new approach (Glock and Hammond 1973; Hammond 1985). Hadden's contribution centers more on the political impact of fundamentalist revivals (Hadden 1971; Hadden and Shupe 1986; 1988). The two approaches differ very much in style: Hammond's is rather gentle and constructive; Hadden's is much more aggressive.

Hammond's aim is to establish a new paradigm for the scientific study of religion. According to Hammond, the present paradigm is centered on the notion of secularization. This paradigm, he argues, is not yet ready to be discarded: "While we are [...] not yet ready to replace the secularization paradigm with some other master scheme, we are prepared to look at secularization and the sacred through new lenses" (Hammond 1985, p. 2). Up to this point, it could seem that Hammond's approach is very close to the one I take in the present paper. There is, however, a fundamental difference, which hinges on different definitions of what is to be understood as the "secularization paradigm". In Hammond's terminology, this expression refers to a thesis which, failing to establish a difference between the "sacred" and "religion" (as an institutional expression of the "sacred"), is unable to take notice of the present resurgences of the sacred, simply because they take place outside conventional religion. Furthermore, it refers to a thesis which implies a one-directional process. As we know, this is not the approach I am taking in the present study. What I take to be the "secularization paradigm" is a broader approach which need not be one-directional, and which need not postulate that there are no manifestations of the sacred outside conventional religion. The very notions of generalization and of privatization precisely imply such a survival of the "sacred" even though "conventional religion" is on the wane.

As a matter of fact, Hammond does not differ very much in his approach from what I call the secularization paradigm. The clearest evidence for this comes from a paper which he wrote for a volume devoted to a discussion of Berger's sociology (Hammond 1986). This paper is all the more interesting since it is almost entirely devoted to a discussion of Berger's secularization theory. To assess Hammond's position with regard to this theory, I am not going to rely on the praises with which he opens his paper. Obviously, given the circumstances for which the paper was written (a Festschrift), there is no way of telling the degree to which these praises might be conventional. What I will do instead will be to discuss Hammond's assessment of the theory itself.

This assessment is quite positive. Throughout the whole essay, the only problem Hammond sees with the theory concerns the definition of religion on which it rests. In other words, we are brought back to the problem mentioned before: What Hammond dislikes in secularization theory is the equivalence put forward between the "sacred" and "religion", or the "supernatural": "Where most see the supernatural as a subset of the sacred, Berger sees the sacred as a subset of the supernatural. Put another way, most sociologists of religion operate with a functional definition: Whatever results from encounters with the sacred, if it is systematic and institutionalized, is religion, whether or not it is expressed in supernatural terms" (1986, p. 156). Thus, all that Hammond advocates is a functional, rather than a substantive definition of religion. In doing this, he is at odds with Berger, but not with what I call the secularization paradigm as such: as we know, the latter includes exemplars based on functional, as well as on substantive, definitions of religion.

We must thus clearly differentiate the approach taken by Hammond from the harder approach we will discuss later on, taken by people like Stark and Bainbridge. Hammond argues that we must move beyond secularization theory not because it is wrong, but because it is incomplete. He does not envision a "return of the sacred" in terms similar to Stark's. Indeed, pondering the meaning of the later Berger's predictions of a return of the sacred, Hammond declares: "Is Berger, in contemplating a reversal of secularization, suggesting a reversal in either [pluralization and subjective secularization] or both of these forces? This is unlikely since little or no empirical evidence would support such suggestions; society is not becoming less pluralistic nor is religious explanation of events on the upswing" (1986, p. 155). What, then, is the nature of the revision proposed by Hammond? To use his own words, his aim is to define a framework allowing us "to understand the sacred in societies clearly continuing to secularize" (1986, p. 158). This, I would contend, can easily be done in the framework of the secularization paradigm as I have defined it, and need not await the formulation of a new master scheme.

As we will see, essentially the same conclusion applies to Hadden's approach. There is, however, a substantive difference between Hammond and Hadden. Where Hammond seems to be pondering secularization theory and the ways to go beyond its formulations, Hadden seems to be at war with it. Although the end result is essentially the same, Hadden's approach is much more aggressive, and takes the form of a work of demolition. But what, exactly, does it demolish?

In a paper entitled Toward Desacralizing Secularization Theory, Hadden declares: "Secularization theory has not been subjected to a systematic scrutinity because it is a doctrine more than it is a theory. Its moorings are located in presuppositions that have gone unexamined because they represent a taken-for-granted ideology rather than a systematic set of interrelated propositions" (Hadden 1987, p. 588). What evidence does Hadden present to support this claim? First, he explains, the works of Thomas Kuhn have showed that the thought processes of scholars are shaped by the communities in which they are embedded. Up to this point, I obviously have no quarrel with Hadden. Especially when he writes: "Accounting for the persistence of religion in the midst of the secular takes on the character of 'normal science' in Kuhn's framework - a 'mopping-up' operation wherein scholars fill in the details of the theory and account for the anomalies" (Hadden and Shupe 1986, p. xii).

Where Hadden and I depart, however, is when he considers that the hold of the paradigm is illegitimate, and that the persistence of religion precisely cannot be accounted for by mopping-up operations. We must thus pose the same question again: What evidence does Hadden put forward to further his claim that the secularization paradigm is merely ideological? Curiously, the first evidence he cites is not drawn from the writings of the theorists he is attacking, as one would expect, but from the writings of those who, like himself, are attacking the theory. He thus quotes Lenski: "From its inception [sociology] was committed to the positivist view that religion in the modern world is merely a survival from man's primitive past, and doomed to disappear in an era of science and general enlightenment" (quoted in Hadden 1987, p. 587). He also quotes Stark and Bainbridge: "At least since the Enlightenment, most Western intellectuals have anticipated the death of religion" (quoted in Hadden 1987, p. 587). Leaving aside for a moment the question whether these claims are true, one can wonder why Hadden did not choose to quote one or several of these early scholars. Surely, this would have provided more solid evidence. But the main problem lies elsewhere. As I we know (see chap. 7), these claims are gross oversimplifications. It is true that some scholars did hold these views. But by putting them all into the same bag, Hadden falls prey to the very type of one-sided appraisal he denounces in "linear secularization theory". Surely, if Hadden had read Spencer's section on "The Unknowable" contained in the First Principles (Spencer 1864b, pp. 3-126), he would not have ventured to write that Spencer was "adamantly antagonistic to religion" (Hadden 1967, p. 592). Thus, the first type of evidence presented by Hadden must be dismissed: it is so caricatural as to do more damage to its author than to the scholars attacked.

But this is not all. For even if Hadden were right in his appraisal of XIX^th century evolutionism, this factor would not be very relevant. After all, we are no longer living in the XIX^th century. As Hadden quotes Dobbelaere's trend report (Hadden 1987, p. 599), we must assume that he knows enough about modern secularization theory to leave the "founders of the discipline" out of the debate. So, to pose the question again, what evidence does Hadden put forward to support his claim? He gives four grounds for dismissing secularization theory. First, the theory is "a hodgepodge of loosely employed ideas rather than a systematic theory. Second, existing data simply do not support the theory. Third, the effervescence of new religious movements in the very locations where secularization appears to cut deeply into established institutional religion suggests that religion may really be ubiquitous in human cultures. Fourth, the number of countries in which religion is significantly entangled in reform, rebellion, and revolution is ever-expanding" (1987, p. 598). The fourth claim is the most serious, and we will come back to it at the end of this section. As for the third claim, we sill see in some detail in the section devoted to Stark and Bainbridge why it does not constitute a challenge to secularization theory. The second claim is not wrong, but hardly challenges secularization theory, which does not rest on statistical data, but on a broad historical analysis. Furthermore, as Hadden himself remarks, if secularization theory cannot be proven by the data, "one must also add the caution that neither can the data disconfirm the process" (1987, p. 600). Let us thus for the moment examine more closely the first claim, to the effect that "the theory has not been systematically stated" (1987, p. 599).

Having just devoted 60 pages to demonstrate that there indeed is a coherent and systematic set of propositions (if not a theory) in the secularization paradigm, I find it emotionally hard to agree with Hadden. But strictly speaking, Hadden is right: although there is a paradigm, there is no unified theory. "Secularization theory" as a collective enterprise does not really exist. This, however, does not provide sufficient evidence to assert that there is "no theory". For it is nowhere said that the theory to which Hadden refers to must be a collective construction; it can also be an individual construction. The question that has to be answered is: is there such a thing as a single, or several different coherent secularization theories put forward by individual theorists? As noted above, Hadden has read Dobbelaere's trend report³²⁶. He thus knows of the existence of the theories of Berger, Wilson, and Fenn. These are the sociologists who should have been discussed to support the claim that the theory has not been stated systematically. Unfortunately, none of them is cited, nor even mentioned, in the paper. Not even Berger, whose Sacred Canopy Hadden has read and commented upon rather positively a few years before (Hadden 1971, pp. 6-8). Hadden thus (apparently purposefully) presents a totally misleading picture of secularization theory. By failing to discuss the most explicit and systematic of the theories which he knows to exits, he disqualifies himself, leaving his claim unsupported.

We are thus left with the last of the challenges, concerning the role of religion in political movements all around the world. Strictly speaking, in the framework of the present study, I should not discuss this claim. As stated in the introduction, I have bracketed the question of the empirical validity of the secularization theories. However, I will here make an exception since, being the prime animus of Hadden's criticism³²⁷, this item is particularly interesting. Broadly speaking, there are three types of evidence of the entanglement of religion and politics: 1) religious strifes, as in Ireland or in India; 2) fundamentalist revolutions, as in Iran; 3) the rise of fundamentalist religion and its involvement in politics in the United States. As we know, Martin's theory accounts for the first case: Whenever religion becomes entangled in problems of national identity, the pattern of secularization is very likely to become altered. The second case does not really constitute a challenge for secularization theory; although some imprudent previsions have been made by certain carriers of the CISR paradigm, the theory itself clearly links secularization to industrialization, and is thus hardly applicable outside the western world. We are thus left with the last case. It is indeed one of the central tenets of secularization theory that politics become independent from religion. If this were proven not to be the case in the United States, the paradigm would enter into a crisis, and an intellectual revolution would become inevitable. But to what extent is religion really becoming an important factor in politics in the United States?

I would not like to appear to belittle this issue. It seems to me that, of all the challenges faced by the paradigm, this is one of the most serious. I am not going to argue that it can simply be brushed off. All that I am going to argue is that the available evidence does not as yet allow us to decide of the severity of the challenge. On the one hand, the sudden political impact of fundamentalist movements in the United States in the 70s and 80s really did come as a surprise. Obviously, no one - least of all secularization theorists - expected this to happen. On the other hand, by the accounts given in the readers published by Hammond (Johnson 1985) and by Hadden and Shupe (Roof 1986) themselves, one is left with the impression that the impact of the new fundamentalism was not as strong as it seemed at first sight. For one thing, contrary to expectations, "follow-up studies suggest that evangelical votes were not a decisive factor in the [1980] election" (Johnson 1985, p. 310). For another, this new fundamentalism appears remarkably "tame"; it is not opposed to "the world", but set on a course of accommodation (Roof 1986, p. 30). In other words, Luckmann's solution to the "American paradox" seems to apply again: Fundamentalism gained a foothold in public life only to the extent that it had become "internally secularized".

It should also be remarked that the new political religions - which represent the only evidence that could constitute a challenge to the paradigm - do not carry the brunt of Hammond and Hadden's arguments. This topic is far outweighed, in their readers, by the problems of religion in the Third World. Finally, even if the new fundamentalism turned out to be a really important factor in American politics - which might after all be the case - it remains to be seen what the nature of the challenge posed to the paradigm would be. As we know, the paradigm leaves much room for non-linear developments. It allows for a complex development of secularization, and leads to no single evolutionary outcome. Furthermore, the notion of generalization implies that a certain form of religiosity (perhaps not unlike the new "accommodated fundamentalism" just discussed) will always hold a certain sway on public life. Thus, even if the final evaluation of the importance of political religion in American must be left open, the charge brought against the paradigm seems very exaggerated.

(Here, critics of my approach could argue that the CISR paradigm is so flexible that it cannot be falsified. But this criticism, although true, would miss the point. A good theory can be falsified, but a paradigm is not a theory, and it is by definition not amenable to falsification. It is precisely the main originality of Kuhn's approach that paradigm becomes obsolete not through an intellectual operation [falsification], but through a social movement inside the scholarly community [revolution]. The accumulation of anomalies in research carried out in the framework of the paradigm must be such as to convince scholars to abandon it. I am not arguing that the CISR paradigm should not be abandoned - maybe it should [although I do not see by what it would be replaced] - all I am arguing is that this has not as yet happened.)

To conclude, Hadden's dismissal of secularization theory is very superficial, and can be understood only from a Skinnerian perspective, by taking into account his own intentions and interests. Being very interested in the entanglement of religion and politics - which is quite a legitimate posture - Hadden mounted a make-believe attack on the dominant paradigm to increase the reader's interest and to orient the sub-discipline as a whole to this new topic³²⁸. The best evidence that the attack is only make-believe is that Hadden himself believes in secularization theory. Witness the following paragraph drawn from the very paper which seeks to demolish secularization theory: "I used to think it terribly important that every significant figure in the founding generation of social scientists wrote at length about religion. But when one recognizes the fact that they had lived in a world that had been dominated by religion for centuries, one can readily ask how could they have written a theory of society without considering religion" (Hadden 1987, p. 591). Apparently, Hadden thinks that modern society is no longer "dominated by religion", since he precisely laments the fact that contemporary social scientists tend to neglect it!
New Religious Movements: Rodney Stark and William Sims Bainbridge
Stark and Bainbridge are among the most well-known and vocal students of new religious movements³²⁹. From the emergence of these movements in the late 60s, they draw the conclusion that "secularization theory" is wrong - or at least partially wrong. However, as we will see, this challenge to the paradigm again misses its target. Stark and Bainbridge's failure to take into account what the secularization paradigm is really saying is prototypical of the attitude of countless other sociologists who consider the emergence of new religious movements a falsification of the "secularization thesis". In other words, they are the most typical representatives of the unjustified "mood of triumphalism" (Robbins 1988, p. 53) that the rise of the new religious movements raised among certain sociologists of religion. In order to illustrate the problem raised by this type of misreading of the paradigm, I will concentrate my analysis on the discrepancies between Stark and Bainbridge on one hand, and Wilson on the other. I will present the arguments of the two sides in a schematic and simplified way, in the sense that I will deal with only two points: 1) the definition of secularization; 2) the evaluation of the impact of the new religious movements on secularization.

1) Definitions of secularization. For Stark & Bainbridge, secularization is a decline in belief, generally located at the individual level - although they are not always very clear about this point³³⁰. For Wilson, secularization is a structural process situated at the social level³³¹.

2) Impact of the new religious movements. For Stark and Bainbridge, new religious movements provide evidence that the secularization process is self-limiting (1985, pp. 429-30). In Wilson's view, new religious movements are a mere by-product of the secularization process, and do not run counter to it³³².

So far, so good. There is no problem in sociologists' taking different views on the secularization process. The problem arises, however, when Stark and Bainbridge, after having defined their own position, go on to describe their opponent's view. For although they list all of Wilson's works which we will be using here in their bibliography, the account they give of Wilson's position is not faithful - or maybe the confusion merely stems from the fact that they do not always specify to which author they refer when they attack "secularization theory". Thus for example, on page 438 (1985), they write: "[we] can test our theory through a geographic hypothesis: Cults will abound where conventional churches are weakest. To follow the logic of the dominant view of secularization leads to a contradictory prediction. Secularization is seen as an unstoppable trend. If we mean by secularization a decline in the credibility of all systems of thought that postulate the existence of the supernatural, then it follows that secularization produces people who resist supernatural explanations. To the extent that that occurs, there will not exist a clientele for religion, whether new or old, or for magic. [...] This leads to a hypothesis wholly opposite to the one we have advanced: Cults will be weakest where conventional churches are weakest".

Now, whose position is this second hypothesis supposed to represent? Although the authors do not give any names in this passage, the argument they develop is clearly directed against "the prevailing wisdom, best expressed by Bryan Wilson [1975] and Richard Fenn [1978]" (citation located less than two pages before Stark & Bainbridge's hypotheses, 1985, p. 436). However, the approach to secularization which they use in this passage is not Wilson's (nor - this is an even more fantastic assertion - Fenn's!). We may even say that it is directly contradictory to Wilson's position. Stark and Bainbridge do not seem to be aware (or are not interested in acknowledging) that there are two radically different views on the secularization process. To use a distinction made by Thomas Robbins, there is a theory of "institutional or systematic secularization" on one hand, and there is an "atomistic, non-institutional" theory on the other (Robbins 1988, pp. 56 and 57). I would contend that Stark and Bainbridge have done a good job in defining their opponent's point of view if, by "conventional wisdom", they mean the second type of secularization theory (but this kind of theory is no longer conventional wisdom in sociology, except for some uninformed sociologists who, for some reason, did not keep abreast of the developments which occurred in the 60s and 70s), which is well represented by Anthony Wallace, whom they cite on page 430.

However, Stark and Bainbridge have misrepresented their opponent's point of view if, by "conventional wisdom", they mean a view which is close to Wilson's³³³. Indeed, if we follow the logic of Wilson's argument, we will be led to predict, exactly as Stark and Bainbridge do, that cults will be strongest where conventional religion is weakest. For if, as Wilson maintains, church involvement can be used as an index of secularization (Wilson 1966, p. 2; 1975, p. 79), and if, as Wilson also maintains, cults are a consequence of secularization (1975, pp. 81-82), it follows quite logically that cults will flourish where secularization is most advanced, that is, where churches are weakest. (I am not here interested in the question of whether Wilson is right or wrong, but only in the logic of his argumentation.) Therefore, the test designed by Stark and Bainbridge - whatever its intrinsic validity, which we need not discuss here - does not falsify Wilson's hypotheses, but only their own view of Wilson's hypothesis, in other words: the ghost of XIX^th century evolutionsim. As for Wilson's hypothesis, it is clearly confirmed along with Stark and Bainbridge's, for there is no substantial difference between them.

To summarize. Although the opponents appear to disagree, in reality, they agree as to the available evidence: 1) institutional religion has lost its dominant social position³³⁴, although 2) faith remains very important in many individuals' lives³³⁵; 3) cults emerge when conventional religions decline, but are unable to fill the gap³³⁶; 4) which shows that there is a correlation between secularization and the emergence of cults.

Now, what is the difference between these authors? Stark and Bainbridge call the correlation set forth in 4) a "reaction" (Stark and Bainbridge 1985, p. 437), whereas Wilson calls it a "response" (Wilson 1975, p. 82). The only difference between these two words is that the first might have connotation implying that cults will ultimately be able to assert themselves against secularization. And this is exactly what differentiates these authors: Stark and Bainbridge are confident that cults will eventually become very powerful³³⁷, whereas Wilson is convinced that they will not³³⁸. Clearly, these convictions do not belong to the domain of sociology, but to prophecy. And this is exactly the point which I would like to make: this is no longer a sociological divergence of views, but a clash of opposite beliefs. Opponents in reality agree as to the facts - although they deny it³³⁹ - but have a different evaluation of the facts.

But let us now abandon this rather pathetic "debate", and turn to Stark and Bainbridge's own secularization theory, as it is exposed in their recent Theory of Religion (Stark and Bainbridge 1987). Until now, we have been working our way through a nicely organized (if imaginary) antagonistic field. On one side, we have found explicit advocates of the secularization thesis, who all worked with a narrow, substantive definition of religion. On the other side, we have found critics of the secularization thesis, who worked with a broad, functional definition of religion. The equivalence between narrow definition and secularization on the one hand, and between broad definition and refusal of secularization on the other hand, has appeared almost axiomatical. Moreover, advocates of the secularization thesis have tended to be either theologically liberal or agnostic, whereas those who criticized the thesis tended to stress the perennial and universal quality of belief. With Stark and Bainbridge, this nice pattern is completely subverted. Not only are Stark and Bainbridge the first to seek to refute the secularization thesis with a very narrow definition of religion; they are also the first to do so while admitting to be "incapable of religious faith" (1987, p. 23). But the reversal of the situation is even more spectacular. For decades, scholars sympathetic to religion have complained that most sociologists tended to discard religion because of their positivistic, reductionistic views. But today, Stark and Bainbridge claim to demonstrate the continuing significance of religion through the use of the most reductionistic and positivistic perspective ever.

Stark and Bainbridge's Theory of Religion is a highly formalized deductive theory, designed for the sole purpose of empirical testing and falsification. It is based on seven axioms, which spell out the fundamental theoretical assumptions. In order to insure the rigor of the deductive logic, the theory also contains 104 very brief and clear definitions. Finally, a set of 344 propositions, all rigorously deduced from the axioms, form the working basis for future empirical testing. It is not necessary, for our purposes, to go into a detailed presentation of the theory. A few lines will suffice to outline its main features. The most fundamental axiom is that "Humans seek what they perceive to be rewards and avoid what they perceive to be costs" (1987, p. 325). The theory further states that, whenever people are unable to obtain rewards which they think exist, they will accept instead "compensators"³⁴⁰. "A compensator is the belief that a reward will be obtained in the distant future or in some other context which cannot be immediately verified" (1985, p. 6). Compensators can be specific (if they substitute for specific rewards), or general (if they substitute for a cluster of rewards, or for rewards of greater scope). Religion, as defined in the theory, precisely deals with such general compensators. As is stated in Definition 22: "Religion refers to systems of general compensators based on supernatural assumptions". The theory unfolds starting from the individual, then moves on to groups, to end with the most macro-sociological aspect: the dialectic of secularization and revival. To this aspect, we now turn.

Stark and Bainbridge's definition of secularization is consonant with the one given in the CISR paradigm. Definition 104: "Secularization is the progressive loss of power by religious organizations". Starting from the beginning of the theory, the first and only fundamental divergence with the CISR paradigm must be acknowledged: "The chief causes of secularization are to be found outside religion itself" (1987, p. 280). The rest of the theory is nothing but a rewording of the paradigm. (I do not want to give the impression that I do not appreciate the significance of the theory. I understand that its originality rests not so much on the formulation of new ideas, but on their very systematic organization. But from our own vantage point, this aspect happens to be irrelevant.) The whole argument rests on the notion that there are three main types of cultural system: religion, politics, and science. Religion emerges first. It is the first form of explanation which allows humans to satisfy their basic need: obtain rewards or compensators. Religion can emerge early because it "does not have to wait until progress discovers objectively correct explanations" (1987, p. 281). Politics is the second sphere to emerge. Later on, it is followed by science. One thing, in this familiar description of the process of differentiation³⁴¹, seems not completely clear. At some moments, it seems as though the three spheres each emerged out of a primordial common cultural ground (through the actions of individuals evolving cultural specializations). But other passages leave the reader with the impression that politics and science emerged out of religion. Thus for instance, "the fact that politicians traditionally drew on religion to cap their political theories with general explanations shows that politicians were unable to find efficacious, distinctively political, general explanations. Reliance on religion to complete a political ideology proves that the politics has not yet itself evolved into a cultural system" (1987, p. 282).

Be that as it may, once differentiation has occurred, adverse effects on religion make themselves felt. In other words, "progress [Definition 81: Progress is the gradual improvement in the human ability to achieve desired rewards] has a direct corrosive effect on religion" (1987, p. 280). The reason for this is that "religious traditions [...] tend to contain many explanations vulnerable to disconfirmation by science and competition from politics". Logically, "people will abandon compensators for the corresponding reward when it is made available to them" (1987, p. 284). This view brings together the two other core exemplars of the CISR paradigm: rationalization* (remember the definition of "progress") and worldliness* (Proposition 18: "Humans prefer rewards to compensators and attempt to exchange compensators for rewards").

Several other exemplars are contained in the subsequent unfolding of the theory. The first is pluralism. "Proposition 67: The more cosmopolitan [Definition 36: Cosmopolitan refers to the existence of plural cultures within a society] a society with respect to religious culture, the lower the market value of any given general compensator" (1987, p. 290). As in Berger's theory, pluralism results in a form of collapse. Of course, as Stark and Bainbridge's theory is based on a view of man as an economic animal, in contrast to Berger's, which rests on a view of man as a cognitive animal, there is a difference at this level. For Stark and Bainbridge, in a situation of pluralism, "one will be forced to conclude that the proffered compensators are uncertain in value, perhaps not very valuable at all" (1987, p. 290). In spite of the economist slant, the argument is familiar: When not founded on a stable social basis (in Stark and Bainbridge's terminology: when the people on whose testimony we rely in evaluating general compensators disagree), belief in the truth of a given religion becomes problematic. Another exemplar contained in the theory is the decline of practice*. "Proposition 322: Cosmopolitan societies tend to have rates of religious affiliation significantly lower than the maximum permitted by other characteristics of the societies". It should be noted at this point that, exactly as in the CISR paradigm, this exemplar has rapidly been weakened (and, what is more surprising, by one of the authors of the theory himself!). In a 1988 paper written together with Roger Finke (Stark and Finke 1988), Stark reported that, in one empirical case at least, religious pluralism did not result in a decline in religious affiliation, but in an increase³⁴².

Up to this point, it is fair to say that even Stark and Bainbridge would probably not contend that their theory differs from the CISR paradigm. Where they see themselves as disagreeing is in the second part of the theory, which bears on religious revival. The first part of the theory deals only with "low-tension" religion (that is, religious groups in a low state of tension with their socio-cultural environment). As for "high-tension" groups, for example sects (Definition 57: A sect movement is a deviant religious organization within traditional beliefs and practices) and cults (Definition 67: Cults are social enterprises primarily engaged in the generation and exchange of novel compensators), the outcome is different. Stark and Bainbridge's argument is based on the following idea: "The fates of low-tension and high-tension religious groups are different at the stage when churchly power begins to weaken. Indeed, the slow collapse of the churches gives sects a new market to exploit" (1987, p. 301). Thus, sects (which appear first) and cults (which appear at a later stage) are a product of secular society, if only because, "the very cosmopolitanism of secularizing societies gives people greater freedom to switch affiliations" (1987, p. 302).

Now in all this, there is still nothing incompatible with the CISR paradigm. But what are the more general consequences of this evolution? Here, curiously, Stark and Bainbridge lose much of their analytico-deductive rigor, and indulge in rather impressionistic assessments: "While we cannot yet specify the mathematical functions that would describe precisely the contrary trends, we can confidently observe that secularization is self-limiting. The level of affiliation in a cosmopolitan society may be lower than in culturally-monolithic society with tolerance for sects. But it may be high compared with societies that prescribe religious diversity. A few lines later, the authors add: "One striking consequence of secularization is a flowering of religion, not its demise" (1987, p. 304). These passages are quite remarkable. First, witness the vagueness of the terminology: The level of affiliation "may" be lower, but it is "substantial". It must be remembered that P337, like the 344 other Propositions, was specifically designed for empirical testing. But how could anyone possibly falsify the proposition that the level of affiliation is "substantial"? Even more remarkable than the vagueness in this supposedly rigorous theory are the semantic shifts. Secularization is "self-limiting". In Stark and Bainbridge's theoretical framework, this expression sometimes refers to the idea that secularization will some day be reversed. But strictly speaking, the notion of self-limitation only suggests that the process cannot become complete because it produces antibodies. As long as even the tiniest minority resists secularization, it is limited. I think that hardly anyone would contest this idea. But then, the affirmation shifts: The affiliation rate becomes "substantial". This suggests that the tiny minority must, after all, be a "substantial" minority. What if 2.6 million people in the United States turned out to be "deeply religious"? Surely, this would be a "substantial" amount (it is roughly equivalent to the "new underclass" the researchers on poverty tell us about, a strata which is not considered unsubstantial). On the other hand, it would represent hardly 1% of the population. From this uncertain middle ground, another, much stronger assertion suddenly emerges: religion is "flowering". Surely, when one considers the number of cults that emerged in the last two decades, the image may seem adequate. But it also connotes something else. Secularization is not only self-limiting; religion is not only of "substantial" importance: it becomes an irresistible force.

Thus one of the difficulties of Stark and Bainbridge's theory is to determine what exactly they are saying. On empirical grounds, the notion that cults could "reverse" the secularization trend (if this really is what Stark and Bainbridge are asserting) seems unsupported. As Wallis remarks, "what is important [...] is not how many cults there are per million of the population and the like, but the ratio of those recruited to new religious movements relative to the total lost by the old faiths. Outside the Bay Area of California, the evidence suggests that the new faiths have made negligible inroads into the mass of the unchurched" (Wallis 1984, p. 61). Other sociologists have reached the same conclusions (for an overview, see Robbins 1988, pp. 58-59). But the disagreement does not seem to rest so much on empirical criteria than on a subjective assessment of the future importance of these movements³⁴³. This, I would argue, is immaterial for our concerns. Questions of measurement and semantic shifts set apart, the only question that is relevant with respect to the CISR paradigm is the following: Have Stark and Bainbridge proposed an alternative view? We are thus brought back to the conclusion put forward above: Their view is not an alternative; as a matter of fact, it is the same as Wilson's: New religious movements are a product of secular society, and help alleviate some of the problems it brings in its wake.

As noted above, the general compensators offered by traditional religions are very vulnerable to progress. But "other compensators may exist that progress has not yet touched. Most importantly, it is not difficult to invent brand new compensators designed to harmonize with the most recent political and scientific developments. [...] New cults may easily be invented to avoid the anarchic and primitive notions of the world and crude explanations invented by ancient religions" (1987, p. 308). Again, as with the problem of pluralism, the only bone of contention with respect to Wilson's view is the different underlying theoretical assumption: Wilson considers that religion first and foremost brings an actual reward: community life. In contrast, Stark and Bainbridge start from the idea that "the ultimate source of religion is the fact that humans greatly desire rewards which are not to be found in this material world of scarcity, frustration and death" (1987, p. 312). As both community life and compensation for death are very highly demanded "goods" in secular society, both theories are compatible with the same exemplar.

One last apparent divergence between Stark and Bainbridge's theory and the CISR paradigm must be disposed of. Stark and Bainbridge sometimes seem to advocate a cyclical view of the secularization process. Secularization is "constantly going on in all societies" (Stark and Bainbridge 1985, p. 2), and is always followed by its reversal, religious revival. One reading of this formulation could be that there is no linear development at all. Surely, this is not what Stark and Bainbridge mean. For they also assert that, with progress, magic progressively became eliminated from religion (P311). Moreover, they clearly indicate that one of the results of secularization was to durably decrease the appeal of low-tension religions. Thus, they clearly acknowledge some measure of linearity: The process may be circular, but the circles themselves move forward along a spiral.

There remains one troubling question. If revival and self-limitation really are compatible with the CISR paradigm, why have I not included an exemplar to that effect in its description? The answer is simple, but might appear embarrassing: There is no such exemplar in the paradigm. Of course, as we know, Wilson does argue that sects are a product of secularization. But he mentions this en passant. As he is not interested in showing that sects are important, and does not make this point very systematically³⁴⁴. Not systematically enough, at any rate, for his statements to qualify as an exemplar by the standards I use in this study.

But why not simply include Stark and Bainbridge in the paradigm? After all, they would not be the first to have been enlisted in it against their will! I cannot include them in the paradigm because, strictly speaking, I am not engaged in paradigm-building (which is by definition a collective enterprise), but in the description of an existing paradigm. (Of course, I am aware that, by the very fact of naming it, I contribute to the building of the CISR paradigm. This, I cannot avoid; but I can avoid engaging in any other operation beyond this simple ostension³⁴⁵.) As we remember, I have identified the carriers of the paradigm through a historical analysis. Even though I consider that revival is logically part of that paradigm, it is not in my power as an analyst to include this exemplar in it.