Discovery Lab Answers

Jessica M. Utts, PhD

Jessica M. Utts, PhD is a statistician and Professor Emerita at the University of California, Irvine, whose engagement with parapsychology is methodological rather than experimental — she brings academic statistics to bear on existing data rather than running her own laboratory protocols. She served as President of the American Statistical Association in 2016 and has argued consistently that the same evidential standards applied elsewhere in science should be applied to psi research, no more and no less.

A note on coverage: the studies in the ESP-Nexus library associated with Utts’s work represent what the library currently holds, not the complete published literature on this question. The library’s share of that literature has not been measured, so the picture below is a summary of what the library holds rather than a settled account of the field.

Experiments

Utts’s contributions are primarily evaluative and analytical rather than experimental in the laboratory sense. The major lines of work the sources document are:

Government-commissioned remote viewing evaluation. Utts authored a 1995 report assessing roughly two decades of government-sponsored remote viewing research conducted at SRI International and Science Applications International Corporation (SAIC). The report examined whether the phenomenon had been scientifically established and whether it was useful for government purposes. The same 1995 AIR review paired her assessment with an independent skeptical evaluation by Ray Hyman.

Ganzfeld statistical analysis. Utts contributed statistical analysis to the autoganzfeld literature. Her 1991 analysis examined eleven series from Honorton et al. (1990). She also published a 1999 combined analysis of ganzfeld and remote viewing results across all labs, drawing on data from multiple sites.

Predictive anticipatory activity (PAA) meta-analyses. Utts co-authored two meta-analyses with Julia Mossbridge and Patrizio Tressoldi — a 2012 paper and a 2014 follow-up — examining physiological responses preceding unpredictable stimuli across 26 studies.

DMILS (Direct Mental Interaction with Living Systems). The Schmidt (2004) row in the evidence table is associated with Utts’s work on this paradigm, covering 36 studies and 1,015 sessions.

Bayesian analysis of ESP studies. Utts collaborated on Bayesian reanalyses of ganzfeld and other psi datasets, including a 2011 paper with Bem and Johnson responding to frequentist critiques of Bem’s precognition studies.

SRI overall analysis. The SRI program analysis (May, Utts, Trask, Luke, Frivold & Humphrey, 1988/1989) is a technical report for which Utts was a named co-author, covering 154 studies and approximately 26,000 trials with 227 participants. See the Data section for how this figure must be read.

Methodology

Utts’s methodological signature is the application of rigorous frequentist and Bayesian statistics to parapsychology data, combined with persistent critique of how significance testing is misapplied on both sides of the psi debate.

Statistical critique of ganzfeld analysis. Her 1986 statistician’s perspective on the ganzfeld debate identified methodological weaknesses in both pro-psi and skeptical analyses of the early ganzfeld literature, arguing neither side had applied fully adequate statistical reasoning.

Bayesian inference. A 1999 Bayesian analysis co-authored by Utts provided posterior probability assessments for ganzfeld studies, framing the evidential question in terms of prior beliefs and updating rather than binary significance thresholds.

Power and effect size. Utts has argued that individual psi studies are typically underpowered given the expected effect size, so null results in single studies carry limited evidential weight; the appropriate vehicle is meta-analysis. She has illustrated this with power curves for typical sample sizes in the ganzfeld literature.

Meta-analytic standards. The PAA meta-analyses with Mossbridge and Tressoldi applied quality-weighting and examined moderator variables. The Schmidt (2004) DMILS analysis weighted effect sizes by overall study quality across 36 studies.

Government evaluation framework. The 1995 AIR report applied conventional scientific standards — the same criteria used in other domains — to assess the SRI/SAIC remote viewing program, explicitly framing the question as whether the evidence met those standards rather than invoking special criteria for or against psi.

Data

The evidence table below your answer carries the full figures; the patterns across rows are summarized here.

All six rows in the library’s holdings for Utts’s work point in a positive direction (dir=positive across every row), but they span three different, non-comparable metrics — standardized effect sizes (Cohen’s d or Cohen’s h), raw hit-rate differences, and a p-value-only result — so they cannot be averaged or pooled into a single bottom-line figure.

Important qualification on the SRI row: the figures in the May, Utts, Trask, Luke, Frivold & Humphrey (1988/1989) SRI technical report are that report’s own findings, not Utts’s independent result. They are relayed and re-derived inside Utts (1996/the 1995 AIR report); that holding paper documents those figures as the program’s own analysis. Attribute them to the original SRI technical report, not to Utts as independent analyst.

No chart is rendered for this question because the rows carry three different, non-comparable metrics and the data do not support a single-metric trend plot across publication years.

Skeptical critiques

What critics argue. No published critique of this work appears in the sources retrieved for this question.

Wagenmakers, Wetzels, Borsboom, and van der Maas (2011) published a Bayesian reanalysis of Bem’s precognition studies in Journal of Personality and Social Psychology that argued the data did not support the anomalous cognition interpretation when appropriate priors were applied. Utts, Bem, and Johnson (2011) responded in the same journal, arguing that the choice of priors in the Wagenmakers et al. reanalysis was unreasonably skeptical and that a more neutral prior would yield different conclusions.

What the experimental data show. The 1999 combined ganzfeld and remote viewing analysis documented a hit rate of 0.34 across 2,097 sessions at all labs, with a raw effect-size difference of 0.09. The autoganzfeld series examined in Utts (1991) returned a hit rate of 0.344 across 355 trials and 11 series (Cohen’s h = 0.20, p = 0.00005). The PAA meta-analysis by Mossbridge, Tressoldi, and Utts (2012) found a pooled Cohen’s d of 0.21 across 26 studies (p < 2.7 × 10⁻¹²). These are the figures the library holds; whether they survive the methodological objections Hyman and others raised is the contested question the paired 1995 review left open.

Analysis. Hyman (1995) and Utts (1995) examined the same SRI/SAIC database and published their divergent conclusions simultaneously, an unusual structure that makes the disagreement a matter of documented record rather than subsequent commentary. The core dispute — whether the statistical patterns in the remote viewing and ganzfeld databases reflect a genuine anomaly or uncontrolled artifacts — has not been resolved by a mutually agreed independent replication outside the original program. The Bem/Utts/Johnson and Wagenmakers et al. exchange documents that the same disagreement recurs in Bayesian form: which prior probability to assign psi before examining the data drives the posterior conclusion as much as the data themselves do.

For a fuller treatment of Utts’s work, see her profile at Jessica M. Utts — ESP-Nexus, including the spoke pages on ganzfeld methodology and statistical analysis and statistical literacy and education in science.