Secularization research and its competitors: A response to my critics

Abstract

This article is a response to the articles published in this issue of Social Compass by François Gauthier, Tobias Müller, David Voas and Sarah Wilkens-Laflamme about the presidential address given by Jörg Stolz at the SISR Congress in July 2019 and titled Secularization theories in the twenty-first century: Ideas, evidence, and problems.

I thank Jean-François Gauthier, Tobias Müller, Sarah Wilkins-Laflamme, and David Voas for their critique of and comments on my presidential address. For better or worse, my address did not leave the critics unmoved. Thus, while Gauthier describes it as suffering from ‘outdated ethnocentrism’ and as ‘[missing] the most fundamental changes that have occurred over the last decades’, Voas feels a sense of ‘vocational renewal’ when reading the ‘proclamation in seven parts’ that has come directly from the ‘Mount Sinai of Secularization’.

These strong reactions are all the more surprising since what I presented was not my own research, but a selection of findings from research on secularization over the last 20 years.

What is open to debate is whether this research is based on sound assumptions and methods, whether its results constitute progress, and whether research on secularization has in fact outstripped its competitors.

Gauthier

Unsurprisingly, I have the strongest disagreements with Gauthier’s (in his own words) ‘radical epistemological, methodological and empirical critique’. In a nutshell, he argues – and he does so with Gauthierian fervour – that there has in fact been no progress in quantitative research on secularization over the last 20 years, this research having merely shown that churched religion in Western countries is shrinking, something that confirms ‘what we already knew’ and what is ‘also supported by qualitative research’. At the same time, such quantitative research has, according to Gauthier, a seriously ‘reductionist’ view of religion (it puts it ‘in a box’) and it ‘misses the most fundamental changes that have occurred over the last decades, and which are qualitative more than quantitative’. If this were true, then indeed I should not have written the address, because there would have been nothing to write about. But let us take the points in order.

• 1. Claiming that quantitative research has made no progress apart from showing that churched religion is shrinking is strange, since my address gives overwhelming evidence of the contrary. I point to very specific research and to some of its key visualizations, which demonstrate, for example, that there is a worldwide correlation between the Human Development Index (HDI) and measures of religiosity (across religions); that secularization in Western countries is mainly a matter of cohort replacement; that natural experiments increasing the length of compulsory education have significantly lowered religiosity both in Canada and in European countries; and that religious pluralism in a given region in the United States tends to depress average religiosity. This is what is new, and not the fact that there is strong secularization in Western countries (which is also true). Did Gauthier know all this 20 years ago? No, he didn’t – because nobody else did, either. Readers who would like to check this are invited to read some of the excellent work that summarized the state of the art of quantitative research on secularization prior to 2000 – for example, Dobbelaere (2002) and Pollack and Pickel (2007). If Gauthier really wanted to argue that there has been no progress, then he would have to show that none of the studies that I presented have added any new knowledge. To be fair, Gauthier does try to attack some of them. For example, he argues that the new findings on the effect on religion of a longer period of compulsory education are ‘contradicted by even the most superficial contextualization’, and that it was the coming to power in Quebec of the Liberal Jean Lesage and not the additional year of schooling that caused the lower level of religiosity there. Now, I do not wish to belittle the possible importance of the Mr. Jean Lesage, but Gauthier’s argument is of course useless, since the Hungerman paper shows the effect for all provinces of Canada and not just Quebec, while another study (Mocan and Pogorelova, 2014) finds precisely the same effect in 14 European countries.

Gauthier’s rebuttals of research on secular competition are even weaker, however. Here, he attacks not the research cited, but a hypothetical example given in the text. When it comes to the secular transition model, he maintains that the model cannot be true because it is based on an ‘evolutionist’ and ‘Western-centric’ ideology.

• 2. According to Gauthier, all the quantitative research that I presented in my address reduces religion to ‘churched’ (or organized) religion, and thus has to be rejected because it forgets to investigate ‘unchurched’ religion. The argument implies that there is a large amount of individualized, lived religion in our societies that has nothing to do with religious groups. The first problem with this argument, at least in Western countries, is that organized religion remains by far the most important part of religious practice we find empirically. It is simply not true that the ‘real action’ in religion happens outside of religious organizations and groups. Both mixed-methods studies in Europe (Stolz et al., 2016) and lived religion studies in the United States (Ammerman, 2014) clearly show that the people who most express religious or spiritual emotions, concerns, and beliefs, are the same people who are involved in organized religion (or who have been so in the past). The second problem with Gauthier’s view is that quantitative indicators actually measure both churched and unchurched religion quite well. Most of what we would want to call unchurched, individualized religion is captured (as, for example, Stolz et al. (2016) show) with items addressing ‘subjective importance of religion’, ‘prayer’, ‘self-assessed religiosity’, and ‘self-assessed spirituality’ as well as item batteries for ‘alternative spirituality’. Such unchurched religiosity is investigated in many of the studies I mention in my address, such as the one by Mocan and Pogorelova (2014) or Voas (2009). Finally, Gauthier accuses quantitative researchers of ‘putting religion in a box’, by which he probably means that they measure religiosity or spirituality with various closed items, thereby failing to capture the true, complex, and contextually influenced nature of religion. If my interpretation of Gauthier’s metaphor is correct, then I would answer that any method – quantitative, qualitative, historical, theoretical – has to draw distinctions and classify phenomena. No method can capture the phenomenon ‘in its entirety’. Indeed, Gauthier himself talks about ‘religion’ in various historical periods, thus assuming that there is ‘something’ (probably ‘in a box’) that is ‘similar’ across these periods.

• 3. Gauthier accuses quantitative research of not contextualizing religion and of not recognizing how religion is ‘affected by wider social logics’. But it is again precisely in this domain where the research that I pointed to in my address has made significant progress. For example, Olson has used multi-level models to show how meso-level context (religious pluralism) influences the religiosity of individuals, and Höllinger and Muckenhuber have investigated the extent to which religiosity is dependent on deprivation at the country (wider context) and individual (individual context) level. In contrast to Gauthier, this kind of research does not only talk about context in a rather vague manner (‘wider social logics’); it actually tests whether specific contextual factors have an effect or not.

I do not believe that quantitative observational methods are the only ones that we should use. Many questions in the social sciences need qualitative or historical methods (I myself have taught mixed methods for years). Neither do I believe that all the results that I reported are valid. This is why I consistently pointed to the new problems and questions that the recent findings raise. What I do believe, however, is that the studies that I flagged up as constituting ‘progress’ are indeed new, important, and should be known by every researcher working on secularization, whether she works with quantitative, qualitative, or historical methods.

Müller

Tobias Müller argues that a decolonial perspective raises concerns, and he points to three blind spots in the progress made by the quantitative research on secularization that I presented in my address: namely, its definition of religion and secularity, the issue of Euro- or Western-centrism, and the issue of gender. ¹

Regarding the first point, the definition of religion, Müller follows Asad (1993) and others, reminding us that religion and secularity are inherently difficult to define and ‘imbricated with . . . epistemic and political hierarchies’; that the term religion and the items used in quantitative research to measure religiosity are based on the Christian prototype, and may therefore be ill suited to other religious phenomena; and that the term secularity is often only used as a synonym for ‘non-religion’, and not as something in its own right. Finally, according to Müller, the very practice of investigating with items constructed by the researcher herself may only reproduce her assumptions about what religion is and ‘ignore how people (re)conceptualize their symbolic worlds as immanent or transcendent’. To a certain extent, I accept these points (except, perhaps, the point that we should define secularity independently of religion). Religion is indeed rather difficult to define; historically, it is a relatively new term; and, in its current understanding, it has been coined in a predominantly Christian environment. Research on secularization is indeed still Christo-centric, and we will still have to change and refine our indicators to better capture non-Christian religions and religiosities.

My difference with Müller is that he seems to think that these are epistemological problems that might call into question the usefulness of quantitative research per se. In my view, however, they are practical problems that can, at least in principle, be solved. For example, finding an all-encompassing definition of religion is a quite difficult (according to Asad: impossible) task when looked at epistemologically. For quantitative researchers, though, deciding on a (working) definition of religion is a necessary step in the research process. Knowing, with Asad, that all our theoretical terms have a history and that the meaning of these terms itself depends on power structures in society is all very well – but it does not absolve us from deciding on a definition, because we would otherwise not know what to observe. Or, to give another example, it is true that quantitative research measures religion with items constructed by the researcher herself, and therefore runs the risk of not capturing what religion really means to the people who are studied. But this is not an insurmountable epistemological, but a practical problem that the researcher can solve by becoming intimately acquainted with the research object – its language, rules, representations, and so on, hence the importance of qualitative and historical research as a precondition of good quantitative studies.

While I believe that researchers will still have to adapt their definition and instruments to better investigate religions in non-Western countries, I also maintain that our definition of religion and our items used to measure religiosity for Western countries work quite well. It is of course true that, if we change the definition of the object of our research, then the results will also change. But it is then up to the critics to actually prove that using a different definition leads to studies with different and more plausible results. For the Western countries, this is exactly what research in the ‘individualization tradition’ has attempted to do in what I called the ‘contentious phase’. Here, scholars have used wider (e.g. functional) definitions of religion, and argued that, instead of secularization, we see only religious change. In my opinion, these scholars have not been successful, and the emerging consensus is that individualization is not an alternative to secularization, but actually accompanies it (see more on this below).

Müller’s second, and closely related, point concerns the possible Western-centrism of research on secularization. Müller argues that such research is ‘Western’, ‘hegemonic’, and ‘parochial’, in that it uses ‘modernist, catching-up narratives that cast non-Western countries as historically lagging behind’. This might have the problematic effect that we are not able to capture the diversity of possible religious changes, such as religious decentralization, or the rise of fundamentalisms. As I noted in my address, such bias evidently exists because quantitative sociology of religion started out in Western countries. But it has made progress in precisely this area in the last 20 years, decreasing the bias by extending its reach to non-Western societies and religion(s). The progress made by Norris and Inglehart (2012[2004]) in using the world value surveys lies in the fact that they measured religiosity in 81 countries, in Western and non-Western areas of the world, in all continents, and across all major religions. The point is not that they eliminated all Western-centrism, but that they made a start in addressing the issue – with new and surprising results. Regarding Müller’s comment on the possible rise of fundamentalism, at least for Western countries, we know that fundamentalist movements have not been able to stop secularization – if they had, this would have shown up in our data.

There is another angle of Müller’s Western-centrism critique that I find problematic. Müller seems to think that the theory of secular transition should be rejected because it is ‘Western-centric’, and a ‘modernist, catching-up narrative’. In my view, however, empirically falsifiable theories should be judged not on ideological, but on empirical grounds. In other words, if the model of secular transition has to be discarded, then it should be discarded not because we may dislike its suspected underlying ideology, but because it is empirically wrong. The paper that I have written with Detlef Pollack and Nan Dirk de Graaf shows that East Germany does not fit the Voas model empirically. If many more such papers are published, then the Voas model may have to be discarded. Conversely, if a great deal of corroborating evidence is produced, then the model may carry the day – even if we don’t like the fact that there may indeed be something like ‘social evolution’. Adopting such an empirical criterion, my view is that there is accumulating evidence that modernization indeed leads to secularization all over the world, in Western as well as in non-Western and in Christian as well as in non-Christian societies, and that some societies have moved farther along this process than others (Norris and Inglehart, 2012[2004]; Pew RC, 2018). I also suspect that this process does not always follow the strict Voas model, but can take different routes.

Müller’s third criticism is that my address did not touch on the link between secularization and gender. I agree that this issue is important. But it is not an area that recent research on secularization has left untouched. On the contrary, while I had gender on my list of areas where progress has been made, I decided not to include it because the results on the link between gender and secularization, as well as the explanation of the gender gap in religion, are still very mixed and need further research. One major advance in quantitative research in the last few decades has been for me the publication of the Pew RC (2016) report on the gender gap in religion around the world, which showed that this gap (women are more religious than men) is not constant across religions, that it is differently gendered depending on the religion in question, and that it is at its widest in the Christian world.

Wilkins-Laflamme

I agree with Wilkins-Laflamme’s general outlook when she discusses my address. Her mind-set is precisely the mind-set that has produced the progress in research on secularization that I identified. I have two comments, though.

First, she suggests that my address should have given more space to a competitor of secularization theory: namely, individualization theory. She writes, ‘There has been very limited quantitative research so far testing the secularization and individualization frameworks against one another’. It is true that my address sought to present progress in research on secularization only, and did not aim to describe and evaluate progress in competing research programmes. So what about the individualization thesis? Has there been similar progress? In my view, we can make two main generalizations.

1. One way of seeing the individualization thesis is to regard it as a competing alternative to secularization theory. First outlined by Luckmann (1967), this thesis states that ‘religion and spirituality are changing, rather than declining in stages’ (Wilkins-Laflamme), and that the decline of institutional religiosity is completely compensated for by new forms of individualized religion, such as ‘believing without belonging’, ‘invisible religion’, ‘implicit religion’, ‘quasi-religion’, and ‘alternative spirituality’. Unless researchers use such a wide definition of religion as to render secularization impossible in principle (as Luckmann in effect did), the individualization thesis has been falsified every time it has been tested in Western countries (De Graaf and Te Grotenhuis, 2008; Pollack and Pickel, 2007; Pollack and Rosta, 2017; Stolz et al., 2016; Voas and Bruce, 2007). True, there has indeed been a significant upswing in individualized forms of religion since the 1960s, and alternative spirituality has had a certain success since the 1970s. But the former has turned out to be a transient phenomenon, an intermediate position that morphs into secularity over time, while the latter has always remained a fringe phenomenon.

2. Another way of stating the individualization thesis is to see it as a form of secularization. This is a view that has much more support in empirical research. In the course of secularization, new generations seem to opt increasingly for religious positions that are more nominal than practicing, that give more choices of varied practices and beliefs, that are more open to doubt and relativism, and that are flagged up as ‘spiritual’. But these new forms of religiosity or spirituality turn out to allow individuals to spend less time and effort on the practices and on transmitting them to the next generations. In effect, they seem to be forms of secularization. Thus, there has indeed been an important individualization of Western societies, and it has had an effect on religion. But it goes hand in hand with secularization, and in no way stands in opposition to it. In our mixed-methods study on religion and spirituality in Switzerland (Stolz et al., 2016), for example, we found exactly this: religious individualization (an importance given to the ‘I’ and its needs) was very strong across the entire religious field; alternative spirituality had appeared, but remained a rather minor phenomenon; and all this was embedded in a very strong intergenerational process of secularization overall.

Second, Wilkins-Laflamme also suggests that I should have included another approach, ‘nonreligion’, that is, ‘studies that examine how religion is translated into or replaced by new values and worldviews in more secular societies [and] the content of these forms of nonreligion’. I do acknowledge that there has been valuable research in this area that can be seen as progress. Let me point out, however, that I do not see the point of trying to find the attributes of the ‘secular’ without making reference to religion. Imagine that for whatever reason people suddenly stop practicing sports. They run, cycle, and swim less and less. Should we then open up a new field of nonsport studies, whose function it would be to study the specific ways of competing and moving that have nothing to do with sports? That would not be very useful because such nonsport competitions and activities are just what people do in their everyday life and are already addressed by general sociology.

Voas

The comments made by Voas can be read as complementing my address in several respects. He draws attention to the fact that some of the research that I presented concerns the ‘how’ (e.g. secularization takes the form of cohort replacement), while other research concerns the ‘why’ (e.g. is it education that mainly causes secularization?). He argues that advances have been more impressive in the ‘how’ than the ‘why’, but that it is ultimately the ‘why’ questions that interest us most. Furthermore, he compares the programme of research on secularization to its main competitors in what I called the ‘contentious phase’: namely, to individualization and rational-choice theory. His view is that the competing theories have not led to significant advances in the field. Finally, he places my claim of progress within the context of the debate in the philosophy of science, assessing the claim with respect to the Lakatos model of research programmes. He comes to the conclusion that my claim that there has been progress in research on secularization seems ‘right’, although ‘less clear-cut’ when judged by the standards of the Lakatos MSRP (Methodology of Scientific Research Programmes). This is because, for him, progress is more about the ‘how’ than the ‘why’.

I basically agree with these points, but would add the following two comments. First, I am not sure whether progress in the ‘why’ area is really more important than progress in the ‘how’ area. In the social sciences, all objects requiring explanation are bound to socio-historical context, and social evolution and history rapidly alter causal mechanisms. In such a situation, causal explanations (answering the ‘why questions’) are more difficult than in the natural sciences, and it may be that in many areas, all we can do is to provide a sound description of how things happened. Going from unclear or misleading to clear and robust descriptions therefore seems to constitute real progress in its own right. Second, I agree that it is not so clear whether the Lakatos model really captures the progress that I identified in research on secularization. The model speaks of progress where theories in the programme make new predictions that are then verified. In the case of secularization theory, there have not been many clear-cut and specific new predictions (except, perhaps, concerning the model of secular transition). Instead, it was mainly advances in methodology and the massive expansion of data (geographically and longitudinally) that allowed researchers to test the model’s original predictions in wider contexts. The results then overwhelmingly supported the secularization model (more education, more pluralism, and more modernization lead to lower religiosity). There is also another aspect that leaves me doubtful whether the Lakatos model applies. In the model, researchers hold firmly to their ‘core beliefs’ and change the auxiliary assumptions and hypotheses in the ‘protective belt’. But, when we look at the research that I highlighted in my address, we find that not all of the researchers who contributed to these studies were really committed to hard-core secularization beliefs. Norris and Inglehart, Mocan and Pogorelova, and Hungerman probably just felt that the data available allowed them to conduct an interesting test. Had they found the opposite of what they actually found, they would have happily reported it and changed their mind on the secularization issue (at least in this specific case). Progress in research programmes therefore seems possible even where researchers are not committed to the hard core of the research programme.

Looking at our discussion from a wider perspective, I am struck by the fact that all four critics mention the issue of the theoretical competitors to the secularization paradigm. While my address focused mainly on progress in quantitative research in the secularization tradition, my critics mainly argue that we should also compare the possible progress of secularization theory to other theories of religious change. This leads to the question of whether similar papers to mine could be written about the individualization or rational-choice approach, or even about other competitors. It would be very interesting to see what such papers could come up with.

As our discussion shows, there are still very different opinions around, and perhaps we have just entered another period in the ‘contentious phase’, as Müller suggests. In any case, I hope that our discussion has helped move the debate forward and has inspired other researchers to continue working on these issues. I again thank the four critics for engaging in this lively discussion.