Religious conversion, philosophy, and social science

Abstract

I argue that empirical studies into the phenomenon of religious conversion suffer from conceptual unclarity owing to an absence of philosophical contributions. I examine the relationship between definition and empirical result in the social sciences, and I show that a wide divergence in conceptual approach threatens to undermine the possibility of useful comparative study. I stake out a distinctive role for philosophical treatments of studies into religious conversion. I conclude with the suggestion that use of the terms ‘convert’ and ‘conversion’ may not in fact be conducive to clarity in the present context, and that subsequent studies may improve their precision through replacing them.

Introduction

While the phenomenon of religious conversion has been the subject of considerable empirical study,¹ philosophers have so far thrown precious little light upon the issue.² Of the thirty-nine chapters featured in the Oxford Handbook of Religious Conversion (2014), none are authored by philosophers. In this article I aim to take several steps towards correcting that general neglect. My aim is not to develop a novel definition of conversion, but rather to provide a diagnosis of various conceptual unclarities which, owing to an absence of distinctively philosophical contributions, have so far been obscured from view in the empirical sciences. I examine the complex relationship between choice of definition and empirical result through an engagement with the work of social scientists. I conclude with a suggestion as to the role philosophical work may play in supplying a remedy for the deficiencies identified.

Religious Conversion

In 1983 sociologists David A. Snow and Richard Machalek made the following observation of sociological studies into the causes of conversion³:

Although this research has helped to specify the relative influence of various social, psychological, and situational factors in relation to the conversion process, conversion itself is vaguely conceived. Just how one might identify the convert is never clearly explained. Instead, the characteristics of the convert are typically taken for granted. This is a serious oversight, especially since an understanding of the conversion process presupposes the ability to identify the convert. (1983: 260)

Snow and Machalek identify two distinct questions concerning religious conversion: (i) what causal factors lead to conversion? And (ii) what does religious conversion consist in? They assume, furthermore, that the latter question enjoys a kind of priority over the former. It is a necessary condition, in the context of scientific study,⁴ on one’s ‘identifying a convert’ that one possesses a correct definition of conversion. Possession of such a definition is presupposed by any ‘understanding of the conversion process’. In other words, the possibility of carrying out empirical research into the character of conversion processes depends upon a prior grasp of the relevant concept, according to Snow and Machalek. The suggestion is that one cannot study a phenomenon without first knowing which things are to count as examples of that phenomenon. What this implies is that the choice of definition must precede empirical study in the present context. This, in turn, raises the following question: on what basis does the social scientist choose their definition of conversion? Snow’s and Machalek’s complaint involves the accusation that social scientists have not, prior to their own contribution, explicitly chosen a definition at all. Rather, according to Snow and Machalek, they have largely, and unreflectively, employed a conception of religious conversion insufficiently precise for the purposes of empirical research. Before assessing the positive proposal made by Snow and Machalek, as well as the influence of their position on more contemporary social scientists, I want first to examine the effects a given choice of definition has upon empirical studies of conversion.

The character of any empirical study into, for instance, the causes or prevalence of conversion depends enormously upon the definition of conversion one adopts. The question of who counts as a convert is settled at this early juncture; and it is therefore at this stage that the domain of legitimate subjects of study is determined. It is in this sense that observation of the causes or prevalence of conversion is ‘laden’ by the definition employed. This relationship between choice of definition and empirical research will be made clearer through the treatment of examples. Take, for instance, an influential study conducted by Harvard scholars Robert J. Barro and Jason Hwang, titled ‘Religious Conversion in 40 Countries’ (2007). Barro and Hwang (2007: 28) draw several conclusions concerning the relation of various historical factors and government policies to conversion rates. Fewer conversions occur, they say, in countries with a history of Communism, while more occur in countries with comparatively higher levels of education than in those with lower levels. Barro and Hwang also conclude that per capita GDP is not significantly related to conversion rates.

The lead author of this study is an affiliate of the National Bureau of Economic Research (NBER), and the paper is a ‘working paper’ of the NBER. The NBER was ranked second most influential domestic think tank in the United States with respect to public policy in 2010.⁵ We must assume, then, that the relevant study was both intended to contribute towards public policy decisions and that it enjoys a platform enabling it to do so. We can therefore conclude that the results of such a study have an important bearing on practical decisions affecting a large number of people. It is important, consequently, to identify the operative definition of ‘convert’ employed. The authors do not, however, provide a definition; of the method employed in this study they write.

[W]e use different retrospective questions from ISSP [International Social Survey Program] 1991 and 1998 and WVS [World Values Survey] 2001 to calculate religious-conversion rates. We use the questions that ask about a person’s current and former religion adherence. (2007: 4)

The conclusions drawn by Barro and Hwang are evidently derived from analyses of data gathered in surveys. If an individual at time $1$ reports an adherence to religion $A$, and at time $2$ reports an adherence to religion $B$, that individual is said to have converted.⁶ Snow and Machalek (1984: 171) though argue that having ‘shifted’ one’s organisational affiliation is not sufficient for having converted. One reason for thinking that mere membership of a new religion does not suffice for one’s having converted to that religion emerges through attending to the distinction between conversion and compliance. This distinction is made by Moscovici (1980: 2011), who describes instances in which one may publicly comply with prevailing views while privately rejecting them, and others in which one may privately accept views while publicly rejecting them. Moscovici suggests that the compliance of those responsible for running concentration camps may count as an example of the former, while participation in secret or persecuted religious societies constitutes an example of the latter.

The survey data drawn upon by Barro and Hwang is arguably too coarse-grained to allow of their making a distinction between ‘genuine’ conversion and mere compliance behaviour. That Barro and Hwang do not draw the relevant distinction is shown in their saying that ‘Historically, religious conversion often resulted involuntarily from conquest or changing preferences of rulers who restricted personal religion choices’ (2007: 1). At least some examples of such historical ‘conversions’ would likely be rejected by Moscovici as instances of compliant behaviour lacking in the requisite private components which attend authentic conversions.

If the criticisms just canvassed of Barro’s and Hwang’s approach are sound, then it is evidently a live possibility that their study of conversion rates fails to count as a study of conversion rates at all. Doubtless some individuals included in the study do count as ‘genuine’ converts, though the proportion of such instances remains obscure and, consequently, the effects of various policies on the prevalence of such instances remains likewise obscure. Whether or not it matters that Barro’s and Hwang’s study more closely tracks a phenomenon distinct from conversion depends upon the authors’ broader aims. At the very least we might reasonably complain of insufficient clarity on their part. A more serious objection involves viewing the correlations they identify between various policies and conversion rates as misleading. We might plausibly suppose that a Moscovician convert is more fervent in their commitment than is someone whose change of religious affiliation is more accurately characterised in terms of mere compliance. We might also suppose that those committed more fervently to a religion behave in ways that those merely compliant with a religion do not. Policy decisions which positively (or negatively) influence the proportion of ‘compliant’ religious adherents in a given population must, on this basis, be treated separately from those which positively influence the proportion of Moscovician converts. The distinction here though is obscured from view on the approach adopted by Barro and Hwang; the relationship between policy choice and conversion of a Moscovician kind does not appear as a discrete object of study. Obviously, the connection just postulated between Moscovician conversion and the exhibition of behaviours not exhibited by merely compliant adherents may fail to hold; whether or not it does hold constitutes a separate empirical question. If it holds, though, it will be useful to know the effect of educational, religious, and economic policies upon the prevalence of conversion so conceived.

Let us examine another contemporary study into religious conversion conducted by Ines W. Jindra (2014). Jindra describes her operative definition:

[…] I define conversions as changes in a person’s religious beliefs that can happen suddenly or gradually. These changes are accompanied by an alternate view of reality and of self, and in general also entail a “reconstruction of one’s biography”. (2014: 10)

Jindra goes on to discuss the relationship between ‘gender related experiences’ and the character of conversion. A gender related experience is, in this context, an experience involving the awareness of one’s own gender role. Jindra’s investigation of this issue takes place against the hypothesis that such experiences contribute towards an individual’s amenability to conversion. Conversion is presented, on this hypothesis, as offering subjects either a route towards the stability of traditional gender roles,⁷ or freedom from the limitations associated with these roles.

Jindra presents the interesting case⁸ of a woman who, raised in a Mormon community, eventually became a Unitarian Universalist. The subject of this study describes Mormonism as conservative, and Unitarian Universalism as contrastively liberal. The difference between these organisations is especially prominent, according to the subject, where issues relating to gender are concerned. The absence, for instance, of female figures from theological discussion, clergy membership, and positions of authority in the Mormon community is lamented. The subject describes these and other features of Mormonism as causing her considerable disquiet. She says, in an interview.

All the time I’ve been sitting there in church and speakers would be talking in the Mormon church, and I think, ‘no, no, no, I don’t agree with that; I don’t agree with it.’ So I would be protesting, protesting, protesting. Every Sunday I went home with a migraine – every Sunday evening – because I’d be having this argument in my head, and I couldn’t say it without being ostracized. And the times that I did I was given positions that... pure silence, assigned. You don’t get to choose what you do, and so I was progressively given assignments that had less and less responsibility, less and less, insolent to the people, less and less contact where I would badly influence somebody else. So I guess that is why I have felt so at home in the Unitarian community […]. (2014: 173)

The subject of this study also reports developing an interest in feminism, which she views as inconsistent with Mormon teaching, at around the age of nine,⁹ having been introduced to it by her long-suffering mother. The discomfort with Mormon doctrine expressed in the just-quoted passage appears, therefore, to have been in place at a relatively early stage. Of her initial encounters with Unitarian Universalism, she says.

There was the counterpoint to my idea of this very elitist religion. Here was a popular religion that was consistent with my more democratic thinking, and it felt so comfortable to finally be affirmed in my belief and to have my behaviour and my ethics consistent with my religion. (2014: 171)

What is striking is the extent to which the subject views this new organisation as fitting with her manner of thinking, behaviour, and ethical inclinations. The subject does not, in other words, appear to have discovered substantially new material with which to furnish her beliefs. Rather, the views of the subject, with respect to the issues which are in her view centrally important, appear to have remained relatively stable. Does the subject count as a convert by the lights of Jindra’s definition? There is some reason for answering this question in the negative. It does not seem that the subject of this study has changed their view of reality or self. The subject was possessed of views regarding herself and others which she perceived to be in conflict with Mormonism long before her eventual shift in organisational affiliation. What has changed in the present case is more accurately characterised in terms of institutional membership¹⁰ than through reference to doxastic considerations. Jindra though clearly does view the relevant subject as a convert, for an investigation into that subject’s biography figures as an important case study in a treatment of the relationship between gender and conversion. If Jindra’s definition appears to rule out the subject we have examined from counting as a convert, then we must conclude that some alternative, and inexplicit, definition of conversion is in fact employed in the determination of appropriate study subjects.¹¹ Those who place great weight on the presence of doxastic change will perhaps agree with Jindra’s stated definition of conversion, and consequently disagree with her description of the subject so far discussed as a convert. Other critics who view the supposed phenomenological character of religious conversion as essential¹² will similarly object that the change in institutional affiliation which attends this subject’s conversion is not sufficient for their having ‘truly’ converted. Still others who argue that conversion necessarily involves moral change¹³ will find the required features missing from the case under consideration. Jindra views the present case as an example in which gender related experiences do play a role in the conversion of an individual. If Jindra is wrong to count the subject of her study as a convert, then this conclusion is unwarranted.

Let us take stock. I have examined two contemporary studies of conversion in the social sciences and offered a number of objections to those studies through appeal to competing definitions of conversion. I have not sought to show that either the definitions employed in those studies or those offered by competitors are correct. Rather, my aim has been to show just how much depends upon the definition one adopts, and to indicate in an inexhaustive fashion something of the definitional variety presently available. Given the discussion above, the question of just what considerations are relevant when choosing a definition of conversion is a pressing one. In other words, we have yet to answer the question asked at the outset of this article: on what basis does the social scientist choose their definition of conversion? In the next section I shall outline a method of answering this question suggested by social scientists which I argue is inadequate.

Philosophy and Social Science

Having identified a vagueness in the operative definition of conversion employed in empirical studies, Snow and Machalek go on to offer a corrective to that vagueness through the presentation of a more precise proposal. They offer four formal properties of a convert, where emphasis on the formal character of these properties serves to rule in the possibility of conversion which is not religious in character. These four properties are related to the ‘talk and reasoning’ (1983: 266) of the convert. Converts may be identified, according to the authors, as individuals exhibiting biographical reconstruction, the adoption of a master attribution scheme, the suspension of analogical reasoning, and the embracement of a master role. The first of these features points to changes in the way an individual narrates their personal history. The second involves the adoption of one interpretive scheme which informs all causal attributions. The suspension of analogical reasoning reflects the disinclination of a convert to view their novel circumstances as comparable to others. And finally, the embracement of a ‘master role’ involves the viewing of one’s convert status as a role prior to the others which one inhabits.

I am less interested in the nature of these formal properties than in the question of how Snow and Machalek have chosen them. The authors inform us that, ‘This line of analysis was suggested during the course of an ethnographic study of the Nichiren Shoshu Buddhist movement in America’ (1983: 260). The approach to definition then, appears strongly influenced by empirical findings. That Snow and Machalek adopt this strategy is confirmed by their arguments in favour of the definition they propose. In support of their suggestion that the suspension of analogical reasoning is partially constitutive of conversion they write.

Consider the following example of a Nichiren Shoshu recruit's use of analogical metaphor to understand the movement's recruitment practice referred to as Shakubuku: “Doing Shakubuku as a follower of Nichiren Shoshu is just like witnessing as a follower of Jesus Christ. Shakubuku is just like proselytizing; it's just another word for what the Hare Krishna and Jesus people do in the streets.” Upon hearing this, a Nichiren Shoshu convert of several years turned around and exclaimed: “Shakubuku and proselytizing aren't the same! Shakubuku is to tell somebody about Nam-Myoho-Renge-Kyo! It is a great act of mercy and compassion, whereas to proselytize is to put pressure on people and force them to come to meetings. The two aren't the same.” By denying the validity of the novice's analogy, the convert laid claim to a certain incomparability regarding Nichiren Shoshu. (1983: 274, emphasis original)

Snow and Machalek report an observation here in which the suspension of analogical reasoning is apparently exhibited by a convert and not by a ‘recruit’. Given that the authors cite empirical study as influencing their choice of definition, we must assume that such observations constitute, at least in part, the basis upon which that definition is formulated. Here though a ready objection is available. We have seen, in section one, that the choice of definition determines the domain of subjects upon which research is subsequently conducted. In the present case, though, the formulation of a precise definition of conversion cannot play the role of determining such a domain but instead depends upon that determination having already been made. In other words, Snow and Machalek seek to confirm their definition through empirical means, where the observation of behaviour exhibited by converts is thought to lend that definition support. Observing converts though, such that reports of their behaviour count in favour of some definition, requires that converts be identified for study. The character of an empirical study conducted with a view to precisifying one’s definition of the subjects studied will clearly be at the mercy of whatever definition has already been assumed in the service of identifying those subjects. A remark of Peter Winch’s, made in a separate but related context, serves to make the point perspicuous:

But the force of the philosophical question cannot be grasped in terms of the preconceptions of experimental science. It cannot be answered by generalizing from particular instances since a particular answer to the philosophical question is already implied in the acceptance of those instances […] (1958: 9, emphasis added)

While Winch’s ‘philosophical question’ is not, in this context, my own, his description of the relationship between philosophy and science nicely summarises the dialectical territory I have described. In the present case the priority of definition over empirical study is manifested in the distinction made between recruit and convert. This distinction is made in order that observations of the convert be contrasted with those of the recruit; the distinction is therefore also drawn in advance of those observations. The empirical findings acquired on the basis of that distinction are then employed in the subsequent effort to formulate a definition of greater precision.

What these considerations demonstrate is the extent to which the choice of definition must precede empirical study. Here, it may be argued, is a place for the distinctively conceptual work familiar to philosophers.¹⁴ What is required, according to this argument, is an analysis of the concept conversion which is capable of providing clarity with respect to a domain of subjects. Such an analysis may proceed in the familiar way, through the assessment of proposals against intuitions in given cases. The analysis will not, therefore, be carried out in ignorance of the world, though it will not consist in an attempt to supplant conceptual work with empirical research.

My view is not, however, that philosophy will serve the social sciences best through accurately describing the individually necessary and jointly sufficient conditions which must be met in order that someone count as having converted. Such a project may well be of philosophical value, if executable. There are well-known Wittgensteinian reasons for thinking that it is not executable; though I aim to remain neutral on this issue. In fact, I do not think it is necessary to carry out an analysis of that kind in order that philosophy help to supply conceptual clarity to the social sciences with respect to the relevant phenomenon. My suggestion is, rather, that philosophy may contribute to supplying the requisite clarity through an engagement with particular studies into religious conversion. The philosophical task on this proposal will be to identify, for any given empirical investigation, precisely what has been studied. Social scientists employ the word ‘convert’ in a number of different ways, depending on both their pre-theoretical assumptions as well as the methodological norms which define their discipline. This much should be clear from the discussion so far. Conceptual diversity here through threatens to obstruct successful inter-disciplinary engagement. It is inappropriate, for instance, to compare a causal model aimed at explaining belief change with one constructed on the assumption that changes in organisational affiliation constitute conversions. Use of one and the same word ‘convert’ obscures the inappropriateness of such comparisons. We may facilitate the making of apt comparisons, and concomitantly the recognition of other comparisons as inapt, through an analysis of the concept employed by social scientists in the context of particular studies. This exercise will necessarily involve a case-by-case treatment of empirical studies, and it will not have as its aim the illumination of conversion’s essential nature. If successful in the proposed endeavour, philosophy will have aided in making meaningful comparative studies in the social sciences more likely to displace those in which distinct phenomena are mistakenly subsumed under one label.

My suggestion here is not that philosophers, considered as a distinctive professional group, are uniquely able to assess empirical studies in the suggested way. Social scientists themselves are well-placed to discern the definitions they have either implicitly or explicitly adopted,¹⁵ though they do not, as in the case of Jindra, appear always to keep their operative definition in view. Rather, it is my contention that such work is distinctively philosophical, for it consists in identifying, for a given domain of subjects, which concepts have precisely that domain as their extension. This is conceptual, rather than empirical work. Given a concept, it is indeed an empirical matter what domain is its extension; we may only find out which things are red, for instance, by examining the world. Given a domain, though, one cannot appeal to observation in order that the concepts which possess that domain as an extension be determined. The relationship between concept and extension is, as it were, of an empirical character in one direction, while non-empirical in the reverse. It is not my aim in this article to delegate the proposed work to specific groups of people, but rather to accurately describe the character of the work in question as conceptual in character rather than empirical. Suffice it to say at this stage that collaboration between philosophers and social scientists is likely conducive to the best results here. The former group are presumably familiar with the methods by which it is determined which concept a given domain is an extension of. Philosophers routinely assess the extensional adequacy of definitions, where doing so involves judging whether or not some target domain is indeed the extension of the concept(s) employed in the proposed definition. The latter group plausibly have a deeper familiarity with the domain(s) of subjects relevant to the exercise than do philosophers, and also with those concepts peculiar to their area of inquiry.

It may in fact transpire that once it has been identified exactly what has been studied in empirical treatments of conversion, the words ‘convert’ and ‘conversion’ are found to be unhelpful for future discussions. Assume that we have identified the operative definition of a convert in some study as, for example, that of a subject exhibiting the four formal properties described by Snow and Machalek. It is not conducive to clarity that the causal model (dis)confirmed in such a study be subsequently characterised in terms of the word ‘conversion’, given the variety we have observed in use of that term. This is not to suggest that we adopt an eliminativist stance of ontological significance with respect to the relevant phenomenon,¹⁶ but only to appreciate the heterogenous use made of the offending word. Where such a term may, in the context of empirical study, be replaced with others less heterogeneously used, the demand for clarity dictates that those replacements be made.