Associative Remote Viewing for Profit: Evaluating the Importance of the Judge and the Investment Instrument

Kruth, J. G. (2021). Associative Remote Viewing for Profit: Evaluating the Importance of the Judge and the Investment Instrument. Journal of Scientific Exploration, 35(1), 13–35. https://doi.org/10.31275/20211395

AI Assessment

A fully pre-specified but unregistered pilot study that reports its judge comparison, its marginal instrument comparison, and its investment losses plainly. Three viewers made weekly precognitive predictions about a volatile stock while, unknown to every other participant, a second judge received identical material and a mock control stock was tracked in parallel. The first judge outperformed the second on the targeted stock (p = .05 one-tailed), the second judge scored significantly below chance on it (1 of 8 correct, p = .02), and both judges were near chance on the control stock. The paper labels itself a pilot with a very small sample and calls for replication.

Provenance

DOI. 10.31275/20211395 · Open access, Creative Commons License CC-BY-NC; article page at the journal site.

Author. John G. Kruth (Rhine Research Center, Durham, North Carolina, USA), sole author. The paper’s primary investigator also served as the second coordinator (C2) for the two blinded protocols. Submitted December 10, 2018; accepted September 29, 2020; published March 15, 2021.

Study type. Prospective pilot study of applied associative remote viewing (ARV) for stock investment: one data-collection protocol with two concurrent protocols, blinded to all participants, testing the effect of the judge (H1) and of the investment instrument (H2).

Funding. The paper contains no funding statement.

Data availability. The paper contains no data-availability statement. An appendix prints the week-by-week data summary, with each judge’s hit, miss, or no-choice result for both stocks.

Source basis. Figures confirmed against the primary article (publisher PDF, Journal of Scientific Exploration, 35(1), 13–35).

What the paper reports

Three experienced remote viewers made weekly predictions about an image they would be shown on the following Friday, across 13 viewing sessions in a 15-week period in 2017, the first session a communication test and 12 planned as investment weeks.¹ Each week a judge compared the viewers’ material against two randomly selected images, one associated with a volatile stock rising and one with it falling, and an investor placed a real trade on the judge’s choice, in a protocol modeled on the silver futures forecasting sessions of Harary and Targ.² Unknown to every participant except the primary investigator, two additional protocols ran concurrently: a second experienced judge (J2) received exactly the same viewing material and targets as the first judge (J1), and a second, mock stock was tracked with no money placed on it. On the targeted stock, J1 was correct in 5 of 10 judged weeks and J2 in 1 of 8; on the control stock, J1 was correct in 6 of 10 and J2 in 5 of 8. J1 outperformed J2 on the targeted stock (p = .05 one-tailed, independent sample t-test), and J2’s targeted-stock performance was significantly below chance (p = .02, effect size d = −1.498). The comparison of the targeted stock against the control stock across both judges (6 of 18 versus 11 of 18 correct) fell just outside the significance threshold at p = .051 one-tailed. The investment itself lost approximately 7.5% of its value because the selected stock turned out not to be available for shorting; the paper reports the loss for informational purposes only and excludes profit from the evaluation of both hypotheses.

The paper describes itself as a pilot study with a very small sample size and a limited number of sessions; it was completely designed and reviewed before data collection but not preregistered, and its strongest single finding is a judge performing significantly below chance.

How it was run

• A prospective design, completely designed, specified, and reviewed before any data were collected: 13 viewing sessions, one per week, over a 15-week period; a pilot test the week of August 7–11, 2017, to adjust the protocol, then 12 experimental weeks from August 14 to November 10, 2017, with September 4–8 skipped for a holiday.
• Five participant roles: a coordinator, three viewers, a target selector, a judge, and an investor. The viewers did not know one another and knew the identity of the coordinator only; viewers and judges were very experienced in their roles, and the coordinator and investor roles were structured to minimize their involvement in the portions of the process directly involving psi.
• Weekly cycle: viewers were tasked on Saturday and returned their viewing data to the coordinator by Sunday evening; only then did the target selector pick two target images using the ARV Studio software (a pool of more than 1,000 images with a random selection process), making the viewing precognitive by definition. After sending the targets and viewing data to the judge, the coordinator randomly assigned the up and down meanings to the two images using random.org. The judge reported the best-matching image, or no choice (NC), before Monday at 3 p.m.
• The investor placed the investment on Monday at 3 p.m. EST and resolved it on Friday at 1 p.m. EST; the image matching the actual Friday state of the stock was then shown to the viewers as feedback. Viewers and judges were never told the investment instrument, the weekly investment state, or whether their decisions made or lost money.
• Protocol 2 (H1): a second coordinator (C2, the primary researcher) stripped the email header from the package sent to J1 and forwarded the exact same email to a second experienced judge (J2). Both judges were told there might be a backup judge, but each believed they were the primary judge. No investments were made on J2’s decisions.
• Protocol 3 (H2): a mock stock, randomly selected from the same list of the 50 most volatile stocks and qualified by the same process as the investment stock (volatility within a 5-day period, availability, and reported shortability), was tracked by C2 throughout. Both instruments were fixed before the study began and retained for the whole series regardless of performance.

Results, as reported

The paper reports means, standard deviations, and p values from t-tests on the judges’ hit/miss decisions; the only standardized effect size reported is d = −1.498 for J2’s targeted-stock test, and no confidence intervals are reported for any comparison. The statistical power of that J2 test is printed three different ways in the paper: greater than .80 in the abstract, greater than .95 in Table 4, and greater than .99 in the text.

Eleven-dimension audit

Pre-registration

The paper states plainly: “This study was not preregistered, but it was completely designed, specified, and reviewed before any data were collected.” It is presented as a prospective study, with both hypotheses, the analysis method, and the handling of no-choice decisions defined in advance, and the two investment instruments selected before the study began. The introduction names registries such as the Koestler Parapsychology Unit Study Registry as the route by which prospective studies gain additional credibility; this study did not use one.

Randomization

The weekly target images were selected by a dedicated target selector using the ARV Studio software, which the paper describes as having a target pool of more than 1,000 images and a random selection process, and the up and down meanings were assigned by the coordinator using random.org, described as a truly random process. Both stocks came from a site listing the 50 stocks with the highest volatility and passed the same qualification process (extent of volatility within a 5-day period, availability, and whether the stock could be shorted); the mock control stock is described as randomly selected from that list. The paper reports no randomization audit trail for any of these steps.

Sensory leakage

The viewings were precognitive by construction: the two target images were selected only after the viewing data had been returned, so during the viewing there was no selected target in existence to leak. The viewers did not know one another, knew only the coordinator’s identity, and received only the single feedback image on Friday; the judges were never shown the feedback and were never told which targets were being used for investment purposes. Neither viewers nor judges were told the investment instrument, the weekly state of the investment, or whether their decisions resulted in gains or losses. Material moved by email, and the paper does not describe any monitoring of the viewers’ working conditions.

Blinding

Blinding is the design’s central feature. The coordinator assigned investment meanings to the targets only after the judging material had been sent, a sequence the paper says was designed to maintain maximum blinding. Protocols 2 and 3 were hidden from everyone except the primary investigator: no participant, including the project coordinator, knew there was a second judge or a second investment instrument. Both judges were told there might be an additional backup judge, but each believed they were the primary judge whose decisions drove the investments, and J2 received exactly the same instructions and information as J1. C2 stripped the email header from the package sent to J1 and forwarded the exact same email to J2; the paper itself notes that this intervention may have introduced a psi-enhanced experimenter effect.

Optional stopping

The series was fixed in advance at 13 weekly sessions, the same two stocks were retained throughout regardless of whether the investments succeeded, and the study ran to its planned end despite losing money. No interim analyses are reported. The analyzed sample sizes of 10 and 8 judged weeks come not from stopping but from the predetermined rule that a judge’s no-choice weeks are dropped from the dataset.

Outcome measure

The unit of analysis is the judge’s weekly decision, scored as a hit or a miss against the actual Friday state of each stock, with no-choice weeks excluded. The paper argues that evaluating ARV projects by profit is evidentially flawed, since a stock series can earn the majority of its funds in a single trial while losing money in every other trial, and accordingly reports its own roughly 7.5% loss for informational purposes only. It also notes that there are currently no set standards for evaluating no-choice decisions, citing a 2018 personal communication, and resolves the question by giving the benefit of the doubt to the judge and excluding those weeks. The hypothesis tests are means comparisons using independent sample t-tests, adjusted for unequal variances using Levene’s Test for Equality of Variance.

Effect size

The paper reports one standardized effect size: d = −1.498 for J2’s below-chance performance on the targeted stock. The statistical power of that same test is printed inconsistently: greater than .80 in the abstract, greater than .95 in Table 4, and greater than .99 in the text. No effect sizes and no confidence intervals are reported for the H1 judge comparison or the H2 instrument comparison, and the judged samples are 10 and 8 weeks.

Multiple comparisons

Each hypothesis carried one pre-specified test: an independent sample t-test comparing the two judges on the targeted stock (H1) and one comparing the targeted stock against the control stock (H2). The post hoc section adds several more analyses: the streak probabilities for J1 and J2, a one-sample t-test of J2 against chance, and re-analyses restricted to weeks in which the stocks changed more than 1% and more than 2%. The paper labels these analyses as post hoc and reports no correction for multiple comparisons across them.

Internal replication

The study is a single pilot series with no internal replication. Its built-in cross-check is the two judges working from identical material, and they diverged sharply: 5 of 10 against 1 of 8 correct on the targeted stock, with J1 at one point correct 4 weeks in a row and J2 incorrect 9 weeks in a row. On the control stock both judges performed near chance (6 of 10 and 5 of 8). The author states that the study should be replicated and that additional evaluations of ARV projects should be completed to confirm the findings.

External replication

The judge-comparison design has no reported direct replication; the paper positions itself against the ARV investment literature it reviews, and that record is mixed by the paper’s own account. Harary and Targ raised more than $100,000 in 9 weeks before a modified second series failed after a few unsuccessful sessions; Kolodziejzyk’s 13-year series of 5,677 trials reported 52.65% correct investment predictions (z = 4.0) and a profit of more than $146,000;³ a meta-analysis by Bierman and Rabeyron of 17 sufficiently designed and well-reported ARV projects found a 63% success rate over more than 550 trials, with their own automated casino-style followup at 56%;⁴ Smith, Laham, and Moddel reported seven correct predictions out of seven (p < .01) and a $16,000 profit for two investors;⁵ and the Project Firefly FOREX effort reported by Katz, Grgic, and Fendley, with more than 60 viewers and 177 predictions over 14 months, lost nearly all of its seed capital.⁶ The paper’s concluding advice to that literature is to qualify judges for accuracy just as viewers are qualified.

Transparency

The article is open access under CC-BY-NC. The paper prints a week-by-week appendix with each judge’s hit, miss, or no-choice result for both stocks, discloses the shorting complication and the resulting loss without hedging, states its lack of preregistration in a dedicated note, and flags its own possible experimenter effect. By stated design it withholds the viewers’ session material and per-viewer success rates, leaving viewer performance to other researchers. It does not name the stocks or the participants, contains no funding statement and no data-availability statement, and prints the power of its central post hoc test three different ways across the abstract, Table 4, and the text.

The adversarial record

• The protocol descends from the 1982 silver futures sessions of Harary and Targ, which raised more than $100,000 in 9 weeks and then terminated after a few unsuccessful sessions under a modified second protocol that prevented adequate viewer feedback; Harary himself later published a criticism of psi-based silver futures forecasting.⁷ The paper calls the protocol iconic and uses both the success and the failure as design context.
• The two hypotheses formalize claims from informal remote viewing discussion: that the judge may matter more than the viewers, and that the choice of investment target matters. The paper concludes that judge selection can have a significant effect, while cautioning that it would be incorrect to conclude that one judging method is superior to another from this study (J1 used a quick intuitive method; J2 used a more procedure-based method built on the Targ Scale).
• Reading against the study: it is a pilot with 12 experimental weeks and judged samples of 10 and 8; it was internally specified but not preregistered; the H2 instrument comparison fell just outside significance (p = .051); the headline judge effect is carried by one judge’s significantly below-chance run, which the paper says could be interpreted as psi missing or an exceptional string of bad luck; the hypothesis tests apply independent sample t-tests to binary hit/miss data; the power of the key post hoc test is printed three different ways; and the investment series lost money, so the revenue goal the sessions were organized around was not met.