The decision to conduct the 1851 Religious Census reflected a concern that absenteeism from regular worship was a social problem, due in part to inadequacy of provision by the established church in new and rapidly-growing industrial communities. Its clergy could readily point to mismatches between supply and demand as a reason for poor showings. Yet Anglicans equally vociferously criticised the methodology, accusing Nonconformists of making inflated claims. Both things undoubtedly shaped Horace Mann's final report on England and Wales, a summary of which appeared in 1854 to a mixed reception. Yet the two key findings were unlikely to be wrong: if Nonconformist attendance figures were on occasion overstated or reflected unrepresentative attendances drummed up for the occasion, it merely meant that even fewer than half of the adult population went to church on Sundays; and any potential overstatement was probably more than offset by Nonconformist venue omissions, strengthening the view that roughly half of those who did attend chose other than the established church. The exercise was never repeated, but it reinforced the building imperative; and when three-quarters of the adult population shunned the established church, the growing social inclusion of Nonconformity was further assured.

As historians of religion have increasingly accepted the sociological value of the data, they have generally concluded that the returns were made accurately and diligently, with little evidence of the sort of bias of which Nonconformist returners were so loudly accused, then and later. This has led, over the past quarter-century, to the publication of increasing numbers of transcriptions of individual county returns; and some of the data (e.g., chapel sizes and dates of opening) expose patterns unaffected by doubts about attendance aspects. Two recent works have made a step-change in the exploitation of the Census output. First, in 1995 Michael Watts brought together an analysis at registration district level of affiliation across the main denominational groups and of illiteracy from Anglican marriage registers between 1838 and 1841, as well as the county-level analysis of available Nonconformist baptismal registers in decennial steps, considered in the preceding chapter. Five years later, Keith Snell and Paul Ell conducted an exhaustive geographic analysis of patterns of correlation between religious denominations contained within the detailed results. Indeed, the latter work pointed to the strong correlations that existed within the data, which would not have been exhibited if the results had been selectively manipulated.