Introduction

Religious affiliation is both the most readily available and least understood indicator of religiosity within the social scientific literature. It is readily available because religious affiliation is regarded as an aspect of personal and social identity (like sex, age and ethnicity), properly included within public enquiries like the national Census. In this sense, ‘religious affiliation’ is regarded as belonging to the public and social domain, in marked contrast to ‘religious beliefs’ and ‘religious practices’ which are generally regarded as belonging to the private and personal domain, properly protected from public scrutiny. It is poorly understood because both conceptually and empirically religious affiliation seems to function quite differently from the ways in which other indicators of religiosity (like beliefs and practices) function. As a consequence, religious affiliation acts as a relatively poor predictor of other religious indicators.

The debate about the usefulness of religious affiliation as an indicator in social research was brought into particular prominence in England and Wales in the six-year period prior to the 2001 national Census, when the introduction of a religious affiliation question within the Census was seriously debated for the first time (Francis, 2003; Weller, 2004). The major argument against accepting religious affiliation as a useful variable in the Census in England and Wales was based on a failure to understand affiliation as a serious social indicator in its own right, but to see it only as a poor predictor of other religious dimensions. Similar debates have occurred in other countries such as New Zealand (Statistics New Zealand, 1998).

For a question on religious affiliation to be included as a valid social indicator in the national Census, affiliation needed to be understood in its own right and not merely as a poor approximation for other dimensions of religion. An important and powerful attempt to rehabilitate self-assigned religious affiliation as a theoretically coherent and socially significant indicator has been advanced by Fane (1999), drawing on Bouma's (1992, p 110) sociological theory of religious identification, according to which religious affiliation is defined as a ‘useful social category giving some indication of the cultural background and general orientating values of a person’. Bouma (1992) then posits a process through which cultural background and general orientating values are acquired and which consists of meaning systems and plausibility structures.