Introduction

There is general agreement among academics, religious leaders and the media that religion is becoming less important in British society (see, for example, Brierley, 1991, 1995; Smith, 1992; Bruce, 1995a, 1995b, 1995c, 2002; Jenkins, 1996; Brown, 1997, 2001; Gill et al, 1998; Bruce and Glendinning, 2002a, 2002b; Denholm, 2002; Kerevan, 2002; Reid, 2002a, 2002b; Swanson, 2002; Wormsley, 2002). Most measures of ‘religiosity’, such as church membership, church attendance and religious attitudes, are trending downwards, with the decline being particularly sharp since the early 1960s. However, there is less agreement concerning what factors are responsible for this decline. A recent review (Bourque and Wright, 2002) of empirical studies that have attempted to model such factors reached three main conclusions. The first is that the rate of decline varies by observed characteristics such as age, gender, socioeconomic status, marital status, presence of dependent children, denomination and geographic location. The second is that research in this area has been hampered by a paucity of high quality data and by the application of relatively unsophisticated modelling techniques. The third is that on aggregate this empirical research has not added greatly to the understanding of the causes of the decline.

It is our view that much can be learned about the factors causing this decline by comparing England with the devolved territories of Scotland and Wales. In this chapter, we restrict the focus to a comparison between Scotland and England using data from the British Household Panel Survey (BHPS). The remainder of this chapter is organised as follows. The second part describes what variables relating to religion are included in the BHPS. The third part examines Scottish–English differences in these variables. In the fourth part of this chapter, logit regression models are estimated in an attempt to control for observed and unobserved factors that might ‘explain’ what appears to be a large difference in church attendance between Scotland and England. A brief concluding comment then follows.

Religion variables

The BHPS is a panel survey based on annual observations or ‘waves’, where the same people are interviewed year on year. As a result of attrition, some individuals leave the survey each year and new members replace these individuals. The BHPS can be used for both traditional cross-sectional analysis and for longitudinal analysis (that is, following the same individuals through time).