Discovery Lab Answers

Dean I. Radin, PhD — Data Summary

Corpus coverage note (required): The evidence below reflects studies currently in the ESP-Nexus library. The library’s share of the published literature on Radin’s work has not been measured, so treat this as a summary of what the library holds rather than a settled account of his full output or the broader field.

The evidence table contains 66 studies and 81 result rows, spanning multiple research lines — Global Consciousness Project analyses, double-slit optical experiments, water pH studies, remote viewing, and more. Because the rows report different, non-comparable metrics (standardized effect sizes, z-scores, hit rates, raw correlations, and p-values only), they cannot be pooled into a single overall figure.

Reported directions are mixed: of the 81 result rows, 9 report null or below-chance outcomes — these are genuine parts of the record and are flagged throughout.

Experiments

Radin’s work in the retrieved sources spans several distinct research programs:

• Global Consciousness Project (GCP) analyses: Radin examined whether the network of random number generators maintained by the GCP showed anomalous deviations around New Year’s Eve midnight transitions across time zones (1998–2025), treating it as a case study in what he calls “experimental metaphysics.”

• Double-slit optical experiments: The sources reference a cumulative body of work across 30 experiments at 5 labs, testing whether human intention influences the interference pattern in a double-slit optical apparatus.

• Water pH “excess correlation” study: Radin conducted an independent replication of a claimed effect in which isolated beakers of water show correlated pH changes under magnetic field stimulation — testing a physically unorthodox “excess correlation” hypothesis.

• EEG/electrophysiology studies (rows 8, 24): Multiple experiments examined whether brain activity (EH3/EH4 trend analyses, RLO vs. RRO slope comparisons) differs during observation vs. non-observation conditions.

• Collaborative work: Several result rows involve co-authored studies — Wahbeh et al., Kauffman et al., King et al., Yu et al., and Alibalaei et al. — placing Radin within a broader collaborative research group at the Institute of Noetic Sciences.

Methodology

Across the retrieved studies, several design features recur:

• Preregistration: The 2026 study (row 1) is explicitly flagged as a preregistered primary test, a methodological step Radin has adopted for at least some recent work.
• Nonparametric and bootstrap methods: The GCP New Year’s Eve analyses use permutation-based comparisons — each midnight transition benchmarked against all other midnight transitions of the year — and bias-corrected accelerated (BCa) bootstrap procedures to handle autocorrelation concerns.
• Autocorrelation correction: In the water pH study, the Durbin-Watson statistic flagged excess autocorrelation in the ΔpH measures, and BCa bootstrap correction was applied; this correction reduced one interaction from significant to non-significant (p = 0.111), a methodological transparency point Radin reports explicitly.
• PCA composite scores: The GCP analysis used principal component analysis across multiple complexity/entropy measures; the corrigendum corrected a variance-inflation error in that PCA, revising the reported composite p-value.
• Double-slit paradigm: The optical mind-matter studies use physical interference-pattern measurement, with the logic that conscious observation should alter fringe visibility in a way detectable by the apparatus.

Data

Because the rows span incompatible metrics, figures are presented by study rather than pooled.

Unsettled signals that must be stated plainly:

• Nine result rows show null or below-chance outcomes. The below-chance result in Radin (2023) — a single-participant experiment where the session z = −3.0 — is flagged as “psi-missing” rather than a null, meaning performance went in the direction opposite to prediction.
• The water pH study illustrates how autocorrelation correction can shift a result from significant to non-significant: the Phase × Condition interaction that appeared significant in uncorrected analysis became p = 0.111 after BCa bootstrap correction, even while the overall Condition effect remained significant (p = 0.006).
• The corrigendum corrects a variance-inflation error in the GCP PCA analysis: the originally reported figure was revised to z = −2.97, p = 0.0013 — still the most extreme of the 100 time points tested, but weaker than first published.

Skeptical critiques

Kauffman et al. (2023), in reporting the cumulative double-slit result across 30 experiments at 5 laboratories, note that independent replication outside the original research group remains a live methodological question — the multi-lab result is cumulative across groups collaborating with or following the original paradigm, not a replication by fully independent investigators starting from scratch.

Radin himself (2026) published a corrigendum acknowledging that the original GCP New Year’s Eve analysis contained a variance-inflation error in the PCA composite: the inflated variance made the composite appear more significant than the corrected figure warranted. The corrected p = 0.0013 (z = −2.97) is a downward revision from the original. This self-correction is methodologically notable — Radin identified and published the error — but the revision also illustrates the sensitivity of PCA-based composite statistics to correlation structure among component variables.

Radin (2025) explicitly reports that autocorrelation in the ΔpH time series, detected by the Durbin-Watson statistic, required re-analysis: after BCa bootstrap correction, the Phase × Condition interaction that originally appeared significant dropped to p = 0.111. The paper reports this transparently, but it means the primary interaction test in that experiment did not survive standard autocorrelation correction. Radin characterizes the overall Condition effect (p = 0.006) and the combined Fisher p (0.00046) as the headline findings, but the non-significant interaction is part of the record.

The null and below-chance results in the collaborative Wahbeh et al. studies (rows 17–18, 24) — including a below-chance Hedges’ g = −0.16 on extended perception and an overall null on the Quick Remote Viewing task — indicate that not all paradigms tested within this research group produce significant positive outcomes, even in samples of hundreds of participants.

For Radin’s full profile, including biographical detail and the complete set of linked study pages, see Dean I. Radin, PhD on ESP-Nexus.