Remote Viewing: A 1974-2022 Systematic Review and Meta-Analysis

Tressoldi, P., & Katz, D. L. (2023). Remote Viewing: A 1974-2022 Systematic Review and Meta-Analysis. Journal of Scientific Exploration, 37(3), 467–489. https://doi.org/10.31275/20232931

AI Assessment

The first meta-analysis to pool all remote-viewing studies from 1974 to 2022, reporting a notably large average effect and concluding that remote viewing is the most efficient ESP protocol. Across 36 studies (40 effect sizes, 39 after outlier exclusion) the authors find an average effect size of .34 (95% CI .22 to .45), a raw hit-rate elevation of 19.3% above chance, agreement between frequentist and Bayesian models, no sign of publication bias, and only a minimal decline. The reproducibility is exemplary, the full database and analysis code are openly posted. Two things a reader should weigh: the effect is much larger than other ESP meta-analyses report, and the headline claim that remote viewing beats ganzfeld, forced-choice, and presentiment rests on comparing separate meta-analyses that used different designs and metrics. One author is also affiliated with a remote-viewing advocacy organization. This audit reports what the study found and how; it takes no position on whether remote viewing works.

Provenance

DOI. 10.31275/20232931 · Journal of Scientific Exploration 2023, 37(3), 467 to 489 (Platinum Open Access, CC-BY-NC).

Study type. A systematic review and random-effects meta-analysis of the remote-viewing literature.

Authors. Patrizio Tressoldi (Science of Consciousness Research Group, University of Padua) and Debra Lynne Katz. Katz is affiliated with the International Remote Viewing Association, an organization that promotes remote viewing; this is a relevant context that the paper discloses through its author affiliations.

Data availability. The entire database and analysis code are openly posted on Figshare (doi.org/10.6084/m9.figshare.22298266.v1) for independent reproducibility, an exemplary level of openness.

Source basis. Every figure below is taken from the article’s own Abstract and Results.

What the paper reports

Remote viewing is a free-response ESP protocol in which a percipient tries to describe a distant or hidden target by mental means, a method with a long history in both Cold-War intelligence programs and civilian research.¹ The study’s central aim is to estimate the overall effect size of the remote-viewing literature and to compare it with other ESP protocols, in order to argue that remote viewing is the most efficient experimental approach to ESP.

After more than 50 years of investigation into extrasensory perception, remote-viewing experimental protocols appear to be the most efficient for both experimental and practical applications.

How it was run

• Scope. All remote-viewing studies conducted up to December 2022 that met the inclusion criteria; 36 studies contributing 40 effect sizes were selected.
• Models. Both a frequentist and a Bayesian random-effects meta-analysis, run to test robustness, with outlier detection (the headline estimate is computed after excluding outliers, leaving 39 effect sizes).
• Bias and trend. Publication-bias testing and a decline-effect (trend) analysis.
• Moderators. Comparison of precognitive versus clairvoyance tasks (and outbound-agent designs), plus a comparison of the remote-viewing effect against the effects reported in meta-analyses of other ESP protocols (ganzfeld, forced-choice, presentiment).

Results, as reported

Values are reproduced from the article’s Abstract and Results. The .34 effect size is large for this field, several times the magnitude that recent ganzfeld meta-analyses report, which is both the paper’s central claim and the result most in need of scrutiny.

Eleven-dimension audit

Pre-registration

The meta-analysis does not appear to have been preregistered as a protocol, though it partly compensates with full openness: the database and analysis code are posted, so its choices (inclusion criteria, outlier exclusion, models) can be inspected and re-run by anyone. The outlier exclusion that produces the headline .34 is one such choice a reader can examine directly.

Randomization

Applies at the level of the primary studies (target selection by random procedures) rather than the meta-analysis. The review inherits whatever randomization quality its constituent remote-viewing studies had.

Sensory leakage

This is the historically decisive concern for remote viewing. The original 1970s SRI studies drew well-known critiques that judges could exploit inadvertent cues in transcripts and target lists. A meta-analysis of effect sizes cannot re-adjudicate those primary-study critiques; it can only pool what the studies reported, so leakage control remains a property of the underlying studies rather than something this paper establishes.

Blinding

Judging blindness is likewise a feature of the primary studies. The review does not code or grade each study’s blinding separately, which is the standard limit of a study-level meta-analysis.

Optional stopping

Not applicable at the meta level. The analogous analytic choice is the outlier exclusion (40 effect sizes reduced to 39); because the data are open, the sensitivity of the estimate to that exclusion can be checked.

Outcome measure

A standardized effect size, reported alongside an interpretable raw figure (a 19.3% hit-rate elevation above chance). The dual reporting is helpful, though the standardization of heterogeneous free-response designs into a single metric is itself an assumption.

Effect size

The reported .34 is large by the standards of ESP research, and substantially larger than recent ganzfeld estimates. A large pooled effect from a relatively small set of studies (36) invites the question of whether it reflects a genuinely stronger protocol or a smaller, more selected literature; the paper’s open data allow that question to be pursued.

Multiple comparisons

Several moderator and cross-protocol comparisons are run. The most consequential is the comparison of remote viewing against other protocols, which is not a within-analysis contrast but a comparison across separately conducted meta-analyses; that is a weaker form of evidence than a single harmonized analysis would be.

Internal replication

The agreement between the frequentist and Bayesian models, and the minimal decline effect, provide internal-consistency support. There is no independent replication sample, as is normal for a meta-analysis.

External replication

Remote viewing has a long multi-program history (SRI, later civilian and applied work), so the literature being pooled is itself the accumulated replication record, complete with its long-running methodological disputes.

Transparency

Exemplary on data openness: the full database and code are posted for independent reproduction, which is exactly what a verifiability-first standard asks for. The transparency debits are the absence of a preregistered protocol and the author-affiliation context noted below, both of which the openness of the data helps to offset.

The adversarial record

• An unusually large effect. At .34 the pooled effect is several times what other ESP meta-analyses report. A skeptic will ask whether this reflects a truly superior protocol or a smaller, more heterogeneous, and more selectively assembled literature; the honest answer is that the open data make this checkable, and it should be checked.
• The superiority claim is a cross-study comparison. Concluding that remote viewing beats ganzfeld, forced-choice, and presentiment rests on comparing the results of separate meta-analyses with different designs, metrics, and eras, rather than a single analysis that places all protocols on the same footing. That is an apples-to-oranges comparison and the “most efficient protocol” conclusion should be read with that caveat.
• Author affiliation. One author is affiliated with the International Remote Viewing Association, an organization that promotes the practice. This is disclosed through the affiliations and does not by itself invalidate a transparent, open-data meta-analysis, but it is part of the full context a reader should hold.
• Inherited critiques. Remote viewing’s foundational studies attracted specific and still-cited critiques about sensory cues and judging, which a meta-analysis of effect sizes pools over rather than resolves.
• What is genuinely strong. The complete open deposit of data and code, the use of both frequentist and Bayesian models, and explicit publication-bias and decline testing are real methodological credits, and the openness is what allows every one of the concerns above to be independently pursued rather than merely asserted.

Where this sits. This paper is one study under the microscope. For the whole remote-viewing record at once, see the Discovery Lab answer What is the case for remote viewing?, or the overview at Remote Viewing.