Utts has articulated that her evaluation of the Stargate evidence applied the same criterion she would use for any scientific claim: whether the effect size is large enough, and the replication record consistent enough, to warrant taking the phenomenon seriously as a research object. She explicitly declined to treat parapsychology as requiring a higher evidential bar than other sciences.¹ Her position is methodological rather than metaphysical, the data either meet conventional standards or they do not, and her assessment was that they did, while acknowledging that the mechanism remains entirely unknown.

Utts and Hyman agreed on the statistical analysis itself, both acknowledged that the hit rates in the Stargate corpus exceeded chance expectation at statistically significant levels. Their disagreement was interpretive: Utts held that the effect sizes were comparable to those accepted as real in other scientific domains and that the replication record was sufficient to conclude the effect was genuine; Hyman held that methodological inadequacies in individual studies and the absence of a satisfactory non-psi explanation meant the anomalous cognition interpretation was premature.⁶ Utts’s response to Hyman’s report specifically addressed his methodological objections, arguing that the cumulative evidence across multiple independent studies could not be dismissed on the grounds of any single study’s flaws.⁶

The Stargate program encompassed remote viewing research conducted across multiple government-sponsored laboratories over more than two decades. Utts’s evaluation drew on the cumulative record rather than individual studies, treating the program as a whole as the unit of analysis. Her published evaluation appeared in the Journal of Parapsychology and addressed both the laboratory experimental record and the operational applications of remote viewing within the program.¹ The evaluation also appears in a later Elsevier volume co-authored with Edwin May, which situates the Stargate findings within the broader literature on non-sensory information access.⁷

The SRI International review covered psychoenergetic research conducted from 1973 through 1988, representing a substantial body of experimental trials across multiple viewer populations and target types. The review by May, Utts, Trask, Luke, Frivold, and Humphrey documented the statistical outcomes of this corpus as an internal SRI report, providing the primary quantitative record of the program’s experimental history.² A subsequent published analysis in the Journal of Parapsychology extended this work, examining advances in remote viewing analysis methodology and the consistency of effects across the corpus.⁸

May, Utts, Luke, Frivold, and Trask’s 1990 paper in the Journal of Parapsychology examined methodological advances in how remote viewing data were analyzed, addressing the challenge of converting qualitative viewer descriptions into quantifiable outcomes. The paper considered the properties of different scoring and judging systems and their implications for statistical inference, a methodological concern that runs through all of Utts’s remote viewing work: that the statistical conclusions are only as valid as the measurement and judging procedures that generate the numbers.⁸

Hansen, Utts, and Markwick’s 1992 Journal of Parapsychology paper and the companion 1991 Research in Parapsychology proceedings paper identified several categories of problems in the PEAR remote viewing experiments.³⁹ These included: (1) judging contamination, ambiguities in whether judges were adequately blind to target identity during the rating process, partially addressed by procedural description but not fully eliminated; (2) randomization documentation, inadequate documentation of how targets were selected, leaving open the possibility of non-random target assignment that could produce spurious matches; (3) multiple comparisons, the use of multiple dependent measures without appropriate correction, inflating the probability of false positives; and (4) response bias, the possibility that viewers’ descriptions were systematically biased toward certain target categories in ways that could interact with non-random target selection. The critique concluded that these problems were serious enough that the PEAR remote viewing results could not be taken as clean evidence for anomalous cognition, regardless of their statistical significance, an assessment that, given documented flaws, yields no conclusion in either direction per the symmetric standard Utts applies.

Humphries, May, and Utts’s 1988 paper presented at the 31st Annual Convention of the Parapsychological Association explored the application of fuzzy set theory to the problem of scoring remote viewing sessions. The core challenge is that viewer descriptions are rarely exact matches or complete misses, they are partial, graded correspondences between a verbal or drawn description and a physical target. Classical binary scoring (hit/miss) discards information; rank-order judging preserves ordinal information but loses the magnitude of match quality. Fuzzy set technology offers a mathematical framework for representing graded membership, a description can be 0.7 consistent with target A and 0.3 consistent with decoy B, potentially increasing the sensitivity of the scoring procedure and reducing the information loss inherent in cruder methods.⁴ The specific artifact this addresses is judging insensitivity: coarse scoring systems may fail to detect genuine partial correspondences, contributing to Type-II error (false negatives) in individual studies.

May, Utts, and Spottiswoode’s 1995 paper in the Journal of Parapsychology introduced Decision Augmentation Theory as a formal model of anomalous mental phenomena.⁵ DAT proposes that anomalous cognition operates not by physically influencing random systems (the classical psychokinesis model) but by influencing the timing of decisions about when to sample from a random process, so that the observer’s anomalous access to future information leads them to initiate measurements at moments that happen to produce favorable outcomes. This is a mechanistic interpretation, not a data claim; the researcher’s preferred interpretation is that DAT better fits the statistical signature of the existing data than force-based models. The proposed mechanism remains contested and has not achieved consensus. A companion paper applied DAT specifically to random number generator data, deriving testable predictions about the relationship between effect size and sample size that differ from what a genuine psychokinetic force model would predict.¹⁰ Alternative interpretations include conventional psychokinesis, publication bias, and methodological artifact, none of which DAT’s statistical predictions fully rule out in the existing literature.

May, Utts, and Spottiswoode’s 1996 Journal of Scientific Exploration paper applied Decision Augmentation Theory to the random number generator (RNG) literature, testing whether the statistical signature of reported effects was more consistent with a genuine force-based influence on physical systems or with the decision-timing mechanism DAT proposes.¹⁰ The specific competing non-psi explanation tested was optional stopping, the possibility that experimenters unconsciously or consciously stopped data collection when results were favorable, producing inflated effect sizes. DAT’s predictions about the relationship between effect size and sample size were used as a discriminating criterion: if effects are genuine physical influences, effect size should be independent of sample size; if effects arise from decision timing, a specific pattern of size-sample covariation is predicted. The paper reported that the RNG data were more consistent with the DAT prediction than with a force model, though this finding is exploratory and the analysis has not been independently replicated with a preregistered protocol.