A Type-II vulnerability in parapsychology is a methodological feature that makes genuine effects harder to detect, thereby inflating false-negative rates. Experimenter effects are particularly acute because they operate at the level of the experimental session itself: if a skeptical experimenter’s presence or demeanor suppresses psi performance, then well-powered studies run by skeptics will systematically underestimate effect sizes. Schlitz has noted that this creates a structural problem for replication: a skeptic who cannot replicate a proponent’s result has not necessarily disconfirmed the effect, they may have introduced a moderating variable. She has described this as one of the field’s core challenges alongside funding and the small researcher pool.⁴ The Schlitz–Wiseman collaboration was designed explicitly to test whether experimenter identity was such a moderating variable by holding all other conditions constant.

In the joint studies, participants were connected to EDA recording equipment. A video camera captured a live image of the participant and transmitted it to a monitor in a separate room where the experimenter sat. Each session comprised 32 thirty-second periods, half randomly allocated to a staring condition and half to a non-staring condition. During staring periods the experimenter looked directly at the monitor image; during non-staring periods the experimenter looked away and directed attention elsewhere. The two experimenters ran their respective participant sets at the same time and in the same facility, eliminating site-level confounds. Schlitz’s data showed statistically significant EDA differences between stare and non-stare conditions; Wiseman’s data did not reach significance. The design was intended to rule out equipment artifacts (same equipment), participant sampling bias (same pool), and procedural variation (same protocol), leaving experimenter identity as the primary candidate explanation for the divergence.¹

The third study, reported in Schlitz, Wiseman, Watt, and Radin (2006), used the same basic EDA-staring paradigm as the earlier collaborations. The paper explicitly frames two competing interpretations of the overall three-study pattern: (1) the first two studies captured a genuine psi effect that was disrupted in the third study by some unidentified aspect of the new design or context; or (2) the first two studies represented chance findings or undetected subtle artifacts, and the third study’s null result more accurately reflects the absence of a remote staring effect. The paper does not adjudicate between these interpretations. The authors note that the collaborative design itself, a skeptic and proponent working jointly, offers a model for resolving disagreements in other controversial areas of psychology, regardless of the ultimate verdict on the psi question.²

The subtle-artifact interpretation centers on demand characteristics: participants assigned to Schlitz may have received implicit cues, through pre-session interaction, tone of voice, or framing of the task, that primed them toward greater autonomic reactivity, independently of any remote staring. This artifact was partially addressed by the joint-study design (same location, same equipment, same pool), which eliminated many obvious confounds. However, the design did not fully address experimenter–participant interaction during the consent and setup phases, which occurred separately for each experimenter’s participants. Schlitz and Wiseman acknowledge in the 2006 paper that subtle interaction differences during setup remain an unresolved potential confound, mitigated but not eliminated by the simultaneous same-site design. The psi interpretation, by contrast, proposes that the experimenter’s beliefs or intentions constitute a causal variable in the production of the effect, a claim that, if true, would have significant implications for how replication is understood across the field. The researcher’s preferred interpretation, as stated in the 2006 paper, is that the data are genuinely ambiguous and that further collaborative research is needed.²

A separate line of evidence bearing on experimenter and participant expectancy comes from a large international multi-experimenter replication of Daryl Bem’s time-reversed priming task, coordinated by Schlitz and involving researchers from more than a dozen institutions. The preregistered confirmatory hypothesis, that response times to incongruent stimuli would be longer than to congruent stimuli even before the prime appeared, was not supported in Experiment 1 (N not specified in the extract; international multi-site sample, fixed-N stopping rule, preregistered).³ Exploratory analyses in Experiment 1 found that participants completing the English-language version showed a significant effect, while those using translated versions did not, a language-moderator pattern that was present in the original Bem study.³ Experiment 2 tested whether reading a pro-psi versus anti-psi statement at the outset would modulate performance. The primary psi hypothesis was not supported. However, exploratory analyses found that participants who received the pro-psi statement showed a larger psi score than those who received the anti-psi statement, a pattern consistent with expectancy moderating performance, though not reaching the threshold for the confirmatory hypothesis.³ Neither experimenters’ nor participants’ beliefs were consistently associated with the dependent measure across the full dataset. The personality variable Sensation Seeking, a component of extraversion, emerged as a correlate of psi performance in exploratory analyses.³

The Schlitz–Wiseman program is notable not only for its findings but for its design logic. By having a proponent and a skeptic run simultaneous sessions with the same equipment, same participant pool, and same procedures, the collaboration operationalized experimenter identity as an independent variable. This design addresses the multiple-comparisons artifact (both experimenters’ data are reported together, preventing selective reporting of only the significant arm), partially addresses demand characteristics (same setup procedures, though pre-session interaction was not fully standardized), and eliminates equipment and site confounds. The design does not address optional stopping (session counts were not preregistered in the early studies), and the third study’s failure to replicate introduces the possibility that the earlier positive results reflected chance fluctuations, an interpretation the authors explicitly acknowledge. Schlitz has described the collaborative model as potentially applicable to other contested areas of psychology where researcher belief may function as an uncontrolled moderator.⁴

In her autobiographical account of coming of age in parapsychology, Schlitz describes experimenter effects as one of the field’s defining methodological puzzles, one she encountered early in her career and that has shaped her research agenda across institutions from the Rhine Research Center to the Institute of Noetic Sciences. She frames the Schlitz–Wiseman collaboration as a deliberate attempt to make the experimenter’s role explicit and measurable rather than treating it as noise to be averaged away. She also notes that the field’s small size and funding constraints make it difficult to accumulate the large, multi-lab datasets that would be needed to fully characterize experimenter effects as a moderating variable.⁴

Cells marked “not-disclosed” reflect genuine absence from the source paper (typical of pre-2015 parapsychology corpus where preregistration + stopping rules were not standard methodological practice). Per the ESP-Nexus honesty mandate, the gap is surfaced rather than filled with inference.