A systematic analysis of first-hand accounts from the Hardy Archive

This analysis draws on individual first-hand accounts from the Alister Hardy Religious Experience Research Archive, now held at the University of Wales Trinity Saint David in Lampeter. The Hardy Archive is the largest systematic collection of first-hand religious experience accounts in existence, containing over accounts submitted by members of the public in response to Hardy's original 1969 appeal and subsequent solicitations.

The accounts analyzed here are those that meet our criteria for individual first-hand accounts, meaning accounts submitted by a single person describing their own direct experience, as opposed to letters commenting on the project, secondhand descriptions, or institutional submissions.

The dataset spans five analytical dimensions, each applied systematically across all accounts:

The analysis was conducted through a multi-stage pipeline. The full methodology is documented at methodology.html, including the complete tagging prompt text, inter-rater reliability statistics, and a description of all known limitations.

• Source selection, Accounts from the Hardy Archive meeting the individual first-hand criteria described above. The full archive spans submissions from 1969 to the present; the accounts in this analysis represent the digitized and processed portion available through the RERC.
• Quality filtering, AI-assisted quality review using Claude Sonnet; accounts lacking sufficient content for meaningful coding were excluded before tagging.
• Tagging, Each account was analyzed across all five analytical dimensions using structured prompts with defined vocabulary and explicit coding criteria. Tags were applied by Claude Sonnet.
• Cross-validation, Every account was independently analyzed by a second AI model using equivalent prompts. Disagreements between models were reviewed through an adjudication pass that re-reads the full account and issues a final determination with per-category reasoning. of accounts went through adjudication.
• Validation and export, Automated schema validation confirmed structural integrity across all accounts. Aggregated statistics were generated for the public-facing website.

Unlike the dual broad/strict coding used in our companion site (storiesofawakening.org), the Hardy Archive analysis uses single-pass tagging per category. The accounts are shorter on average ( words per account) and less ambiguous in their coding, the written, reflective format of the Hardy accounts produces clearer categorical signals than transcribed spoken interviews.

The dataset is suited to descriptive and correlational questions about the phenomenology, context, and aftermath of religious experience across a large, heterogeneous sample. Examples:

• What experience types and phenomenological qualities co-occur most frequently, and which tend to appear in isolation?
• Does the trigger of an experience predict the qualities reported? Do crisis-triggered experiences differ phenomenologically from prayer-triggered or spontaneous ones?
• How do lasting effects distribute across tradition framings? Do Christian accounts, secular accounts, and syncretic accounts differ in their reported effects?
• How does the prevalence of veridical claims vary across experience types?
• What is the distribution of experiences across Hardy's original category system, and how does AI-assisted coding compare to prior hand-coded analyses of archive subsets?
• How does the phenomenology of near-death experiences in this archive compare to NDE accounts in interview-based datasets?

The dataset is not suited to: causal claims about what produces religious experience, analysis of narrative structure or linguistic patterns (the available data is categorical, not textual), individual case study work (the archive is held by the RERC and access to original accounts requires separate arrangement), or generalization to non-Western, non-English-speaking populations.

The following limitations are documented here in full and should be considered in any use of this data.

This site provides aggregated statistics, prevalence rates and categorical breakdowns across all five dimensions. Individual account texts are held by the Religious Experience Research Centre (RERC) at the University of Wales Trinity Saint David, Lampeter. Researchers wishing to access original accounts should contact the RERC directly at uwtsd.ac.uk/rerc.

The tagging dataset and pipeline code are not yet publicly released, but collaboration inquiries are welcome. If you are working on research related to religious experience and would like to discuss access or collaboration, please reach out via the Contact page.

For citation purposes, please use: