Discovery Lab Answers

Julie Milton: Experiments and Data

Julie Milton is a parapsychologist whose work centers on meta-analytic synthesis and methodological critique of ESP research, conducted largely in collaboration with Richard Wiseman. The sources retrieved for this question cover several distinct lines of work.

The Post-Communiqué Ganzfeld Meta-Analysis (Milton & Wiseman, 1999)

The most cited product of Milton’s empirical work is the meta-analysis published in Psychological Bulletin examining 30 ganzfeld ESP studies conducted after Hyman and Honorton’s 1985 methodological guidelines — the so-called “Joint Communiqué” — all from seven independent laboratories ^[4]. The key findings from that paper:

The result was non-significant: the 30 post-Communiqué studies, drawn from 14 papers by 10 different principal authors, failed to replicate the above-chance hit rate found in the earlier PRL autoganzfeld work reported by Bem and Honorton (1994) ^[4]. Milton noted that the mean effect size in the recent studies was less than one-seventeenth of that found in the PRL work, and a post-hoc comparison showed the difference was statistically significant ^[6].

One important nuance Milton raised: a single study with a highly significant outcome (p = 7.2 × 10⁻⁸) was sufficient, on its own, to pull the entire null database into overall statistical significance — but without identifying what variables produced that result, only a small number of experimenters would be positioned to replicate it ^[2].

Of the five internal effects Bem and Honorton had identified as statistically significant, three were subjected to replication attempts in the new studies; only one was confirmed ^[4].

Internal Effect: Dynamic vs. Static Targets

One internal variable Milton’s analysis was able to examine was the use of dynamic versus static targets. This was the one Bem and Honorton variable for which the post-Communiqué studies had sufficient reporting to permit comparison ^[6]. The broader set of variables Bem and Honorton flagged — including participant type, sender presence, and personality measures — were either not measured or not reported in enough of the 30 studies to permit a proper replication test ^[6].

Free-Response ESP in Ordinary Waking States

Separately from the ganzfeld work, Milton conducted a meta-analysis of free-response ESP studies performed in ordinary waking states of consciousness — not altered states such as ganzfeld, hypnosis, or sleep — synthesizing data across multiple experiments to assess whether ESP effects appear in standard laboratory conditions. The ESP-Nexus page on this work notes that the analysis found effect sizes consistent with meta-analytic synthesis, though heterogeneity across studies remained significant. The 78 free-response studies Milton had analyzed by 1997 are referenced in the SAIC remote viewing critique ^[5] as the nearest comparison group for Experiment One of that program.

Remote Viewing: SAIC Experiment One Critique

Wiseman and Milton published a critical re-evaluation of Experiment One of the SAIC remote viewing program, identifying potential information-leakage pathways ^[5]. In a reply to Edwin May’s response, they maintained their position on those pathways, noting that evidence of cued receivers obtaining an 89% hitting rate in 100 trials (where 50% would be expected by chance) in a separate study demonstrated that such cues can be effective — and that chance performance by other receivers does not license a general inference that cues are ineffective ^[5].

Psychic Pet Research

Wiseman, Smith, and Milton conducted experimental tests of the claim that a dog (Jaytee) could anticipate its owner’s return — a claim associated with Rupert Sheldrake’s research ^[3]. Their reply to Sheldrake’s criticism of that work appeared in the Journal of the Society for Psychical Research ^[3].

Methodological Contributions

Beyond specific experiments, Milton’s work includes:

• Consensus-building on methodology: A questionnaire survey of both parapsychological experimenters and critics to establish consensus-based methodological guidelines for ESP studies.
• Participant behavior in forced-choice tasks: Research on how guessing strategies and confidence-call criteria systematically influence performance outcomes.
• Quality-coding critique: Milton flagged that quality coding in parapsychology meta-analyses has almost always been conducted non-blind, making it difficult to rule out coder bias; and that applying quality weights to only a subset of studies — as Storm and Ertel (2001) did — distorts the overall picture ^[1][2].
• Publication and funding pathways: Survey-based documentation of barriers parapsychology researchers face in publishing in mainstream venues.

Skeptical Critiques

What critics argued. Storm and Ertel (2001) challenged the Milton & Wiseman (1999) null result directly in Psychological Bulletin, constructing a quality scale and applying it to weight effect sizes — arguing that when study quality is properly accounted for, the post-1986 ganzfeld database supports the psi hypothesis. They also applied Timm’s (1983) statistic, which tests deviation from mean chance expectation regardless of direction, obtaining p = .027 for the Milton-Wiseman database, and argued this result “supports the psi hypothesis” ^[2].

What the data show. Milton and Wiseman replied that Storm and Ertel’s quality scale excluded important safeguards such as duplicate target sets, and that the scale was applied as a weight to only 11 of the 79 studies in the combined database — leaving the other 68, including heavily criticized earlier studies, unassessed and unweighted ^[1][2]. They also noted that Storm and Ertel contained bibliographic errors in study attribution ^[2]. On the Timm statistic point, Milton and Wiseman observed that no previous parapsychological meta-analysis had examined data for deviation from chance regardless of direction, making the interpretation of that result difficult to contextualize ^[2].

Analysis. The exchange is a direct methodological dispute about which studies belong in the database, how quality should be weighted (and across which studies), and what the appropriate statistical test is. Storm and Ertel (2001) used a different database composition and a different quality-weighting scheme; Milton and Wiseman’s objection is that applying weights selectively across a subset of studies is not a justifiable method regardless of the resulting direction of effect. Neither party disputes the raw non-significance of the 30 post-Communiqué studies on the primary measure; the disagreement is about how to interpret the broader corpus and whether Storm and Ertel’s quality scale is defensible.

For deeper context on where the Milton & Wiseman (1999) result sits within the longer arc of ganzfeld replication attempts, the Replication in Psi Research page and the Julie Milton scientist hub both carry relevant synthesis.