Extending and building on previous work on the merits of Grace Davie’s theory about ‘believing without belonging’, this paper offers a comparative analysis of changes in the relationships between ‘believing’ and ‘belonging’ across countries. In doing so, two renditions of the theory that co-exist in Grace Davie’s work are distinguished, i.e., the typically foregrounded version about a de-institutionalization of Christianity and its often unnoticed counterpart about a spiritualization of religion. Societal growth curve modelling is applied to the data of the European Values Study for twenty European countries (1981–2008) to test hypotheses derived from both theories. The findings suggest that the typically foregrounded version of a de-institutionalization of Christianity needs to be rejected, while the typically unnoticed version of a spiritualization of religion is supported by the data.

1. 1.

   Davie only made predictions about Britain and Western Europe more generally, hence including former communist countries (Reitsma et al. 2012) or North-American countries (Aarts et al. 2008) seems unnecessary.

2. 2.

   The items about belief that ‘people have a soul’ and belief in ‘the devil’ were only available in the first two waves of the EVS and could therefore not be used for constructing the scale.

3. 3.

   Iceland and Norway have respectively the lowest and highest Eigenvalues (2.124 and 3.256) and proportions explained variance (42.5% and 65.1%). The factor loadings varied from 0.57 to 0.82 (Iceland), and from 0.79 to 0.86 (Norway). Iceland and Norway have respectively the lowest and highest Cronbach’s alpha’s (0.656 and 0.865). Only in Croatia and Malta will the alpha increase slightly if the item about a Personal God is deleted from the scale (from 0.830 to 0.875 in Croatia, and from 0.725 to 0.814 in Malta).

4. 4.

   If we set the minimum to four or five valid answers, we will lose respectively 13% and 27% of the sample. This is due to relatively high proportions of missing values for the items belief in life after death (16.1%), heaven (11.9%), hell (11.5%), and sin (9.4%). Nearly all respondents who have a missing value on one or more of these questions answered them with “I don’t know”.

5. 5.

   This is 91.7% of the original sample.

6. 6.

   The correlation decreased from 0.523 in 1981 to 0.489 in 2008.

7. 7.

   We make use of the continuous variable country-level traditional Christian religiosity 2008 (see Table 5).

8. 8.

   The correlation decreased from 0.538 in 1981 to 0.489 in 2008.

9. 9.

   i.e. Sweden (0.196) and Malta (0.823).

We are grateful to Peter Achterberg, Bart Meuleman, Rudi Laermans, Staf Hellemans, Klaus Friedrich, Erwin Gielens, and Galen Watts for their helpful remarks, and to Hellen Brans for her unconditional support. Draft versions of this paper were presented during Cultural Sociology Lowlands (University of Amsterdam, 20 June 2017) and the Annual Meeting of the Association for the Sociology of Religion (Montreal, 14 August 2017). Final versions of this paper were presented during the KOSMOI symposium (KU Leuven, 12 October 2018) and the SSSR + RRA Annual Meeting (Las Vegas, 27 October 2018). This article is part of the first author’s PhD project on religious decline and religious change in Western Europe (1981–2008), funded by KU Leuven (Grant No. 3H160220).

Affiliations

1. Centre for Sociological Research (CeSO), KU Leuven, Parkstraat 45, 3000, Leuven, Belgium

   Paul Tromp, Anna Pless & Dick Houtman

Corresponding authors

Correspondence to Paul Tromp or Dick Houtman.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Reprints and Permissions