Discovery Lab Answers

Roger D. Nelson, PhD — Experiments and Data

Roger D. Nelson, PhD is a parapsychology researcher profiled at Roger D. Nelson. His work spans several decades, centered on whether human consciousness — individually or collectively — correlates with measurable deviations in random physical systems.

A coverage note before the evidence: the studies below reflect what the ESP-Nexus library currently holds for Nelson’s work; the library’s share of the complete published literature on this question has not been measured. Treat the following as a summary of what the library holds, not a settled account of the full field. The evidence directions are not uniform — most results are positive, but one is null, and the metrics across studies are not comparable to one another and cannot be averaged into a single bottom-line figure.

Experiments

Nelson’s research record divides into three main lines:

1. PEAR Lab REG/RNG experiments. While at the Princeton Engineering Anomalies Research (PEAR) lab — where he served as research coordinator from 1980 — Nelson participated in studies on whether individual operators could intentionally shift the output of random event generators (REGs). The benchmark dataset reported in Jahn (1997) accumulated across 522 experimental series and 2,497,200 trials. Nelson contributed to a large-database analysis of the full PEAR REG mind-machine dataset, covering 108 operators, 1,262 replications, and more than 5.6 million trials.

2. FieldREG experiments. Nelson developed and ran FieldREG studies, in which a portable REG is operated continuously in naturalistic group settings — rituals, concerts, lectures, sacred sites — to test whether coherent group activity correlates with anomalous REG output. Applications included ten initial deployments reported in 1996, and an extended Egypt sacred-sites series reported across multiple publications. The Egypt work included the Pyramid of Khufu’s Subterranean Chamber and a broader series of Egyptian sacred-site visits.

3. Global Consciousness Project (GCP). Nelson founded the GCP in 1997 and directed it through 2026, when active data collection ended; he subsequently became founder and director of GCP 2.0, a successor network. The GCP operates a world-spanning network of physical RNGs running continuously, with pre-registered hypotheses specifying which global events are predicted to correlate with departures from random expectation. Nelson has published cumulative analyses of this formal series at multiple points — 43 events (2001), 109 registry entries (2002), 130+ predictions (2002), 247 sessions (2008), 456 tests (2014), and 500 events (2020/2024). He also conducted sub-analyses examining earthquake anticipation effects, event-category breakdowns, and structural analyses of the GCP data using techniques such as multi-scale entropy.

He has also published a meta-analysis of RNG experiments with Dean Radin, and wrote an obituary of Stephen E. Braude, PhD, reflecting his involvement in the broader Journal of Scientific Exploration community.

Methodology

PEAR REG protocol. The benchmark PEAR experiments used electronic REGs generating binary random sequences; operators set intentions in three conditions (HI, LO, baseline) across counterbalanced runs. Jahn (1997) reports the primary indicator as HI-LO separation. Nelson’s contribution to the large-database analysis examined parameters such as operator gender, spatial and temporal distance between operator and REG, and whether intention targets were operator-selected or randomly assigned.

FieldREG protocol. A portable REG records data continuously during a defined group event. Segments are pre-marked at the start and end of each activity; Z-scores for each segment are computed and combined. To correct for autocorrelation in the continuous data stream, Nelson used permutation-based adjusted chi-square statistics in the Egypt series. The Egypt sacred-site experiments categorized data across multiple site types (pyramids, temples, ritual activities, experimenter-only control periods) and compared their composite effects.

GCP formal experiment. Each GCP event requires a pre-registered hypothesis entry specifying event start and end times and the planned statistical test before data analysis begins. The primary statistic is a Stouffer Z combining across the network of RNG nodes second-by-second. Results are logged in a public hypothesis registry. Nelson has argued that the pre-registration and registry structure guard against post-hoc selection, though critics have disputed this — see the Skeptical Critiques section below.

Data

The evidence table covers 19 result rows across Nelson’s work. The metrics are heterogeneous — z-scores, p-values only, standardized effect sizes (z/√N), and correlation-type effect sizes appear across different studies — and cannot be pooled into a single summary figure. The chart below plots the results reported on the z/√N metric against publication year; no trend line is fitted, because the rows mix effect-size conventions that are not comparable on a single axis.

PEAR and early REG work:

FieldREG:

GCP cumulative formal series:

Unsettled signals — mandatory disclosure:

• Nelson (1998) returns a mixed/borderline result (p =.059) for new exploratory FieldREG applications, not crossing the conventional significance threshold.
• Nelson (2011) reports a null result: the difference in composite Z between Nelson-sourced and other-sourced GCP events was z = 1.143 — not significant — from a recategorization analysis of the formal series. The evidence set breaks 17 positive, 1 mixed (the 1998 exploratory series), and 1 null (this comparison test) and bears directly on questions about experimenter-sourcing effects.
• The GCP effect sizes on the z/√N metric are very small in absolute terms despite high statistical significance, a pattern that emerges from the enormous trial counts accumulated.
• The four studies reportable on a single comparable metric (z/√N) are charted below.

Skeptical Critiques

What critics argue. Scargle (2002) raised concerns about the GCP’s hypothesis-selection procedure, arguing that the freedom to define event windows — start time, end time, and geographic scope — after events are known introduces undisclosed researcher degrees of freedom that inflate apparent significance beyond what pre-registration alone can prevent. Bösch, Steinkamp, and Boller (2006) conducted a formal meta-analysis of micro-PK / RNG experiments that identified significant funnel-plot asymmetry consistent with publication bias, and reported that the highest-quality studies pointed in the direction opposite to intention, with an intention-directed estimate emerging only after the three largest studies were removed — a pattern they argued weakens causal interpretation of the aggregate result.

What the experimental data show. Nelson has addressed the window-specification critique by pointing to the hypothesis registry as the pre-commitment mechanism, with event parameters logged before analysis. The null result in Nelson (2011) — where the composite Z difference between Nelson-sourced and other-sourced events was z = 1.143 — was itself a test of whether experimenter-proximity inflates GCP results; that specific comparison did not reach significance. The GCP cumulative Z across 500 events reached 7.31, and the overall p-values across the formal series are consistent across the multiple publication-year snapshots in the evidence table. Nelson (2024) also examines structural features of the GCP data — including multi-scale entropy analysis and evoked-response patterns — as perspectives independent of the primary event-by-event analysis.

Analysis. The central open question is whether the GCP’s pre-registration procedure fully closes the degrees-of-freedom problem Scargle (2002) identified, or whether residual flexibility in event definition remains. The Bösch et al. (2006) quality-effect-size relationship, if it holds for GCP-scale data, would predict smaller effects as methodological controls tighten — a prediction the growing GCP cumulative Z does not obviously support, but which has not been resolved by an independent replication of the GCP design by a separate research group.

For broader context on Nelson’s work and the GCP network, see Roger D. Nelson and the spoke page Consciousness-Correlated Anomalies: Theory and Evidence.