Mediumship – How Is It Studied?
Can a medium provide accurate information about a deceased person without receiving any clues? This is one of the questions researchers attempt to investigate through controlled experiments.
Mediumship research uses methods including double-blind and triple-blind protocols to minimize the risk of information being passed to the medium, consciously or unconsciously. In double-blind testing, both the medium and the person conducting the test lack certain information that could influence the results. Triple-blind testing introduces an additional level of blinding, for example by having the deceased person’s relative assess different readings without knowing which one was intended for their deceased loved one.
Blinding is important for reducing the possibility of cold reading, in which a medium may use clues from a relative’s words or reactions. By limiting contact and controlling the flow of information, researchers can better investigate whether accurate details are provided without conventional information transfer.
In mediumship research, studies using triple-blind protocols have been published in scientific journals after undergoing peer review. This means that independent experts assess the studies’ methods, analyses, and conclusions before publication. Peer review is widely regarded as an important mark of scientific quality, because it helps identify weaknesses and strengthen the credibility of research. However, it does not guarantee that the findings are correct.
How Does an Experiment Work?
In the methodological paper Contemporary Methods Used in Laboratory-Based Mediumship Research (Beischel, 2007), the author describes how researchers can organize such experiments.
A medium attempts to describe a deceased person without knowing their identity. There is no direct contact with the deceased person’s relative, known as the sitter. Instead, an intermediary is used who does not know relevant details about the deceased person.
The protocol includes several levels of blinding:
- The medium has no information about the person being described.
- The intermediary does not know the person’s identity or life history.
- The relative assessing the statements compares an intended reading with a control reading without knowing which is which.
In some experiments, the readings concern deceased people of the same sex, preventing sex from revealing which reading belongs to whom. The relative then assesses how well the descriptions match. The comparison shows whether the information fits the intended person better than someone else. The results can be statistically compared with what would be expected by chance.
What Have the Studies Found?
Beischel and Schwartz’s 2007 study examined whether mediums could provide accurate information about deceased people without receiving information from their relatives. Eight mediums took part in the experiment. The relatives then compared two anonymized readings: one intended for their deceased loved one and a control reading concerning someone else. On average, the relatives rated the intended readings more highly than the control readings. The difference was statistically significant, meaning it was difficult to explain by chance alone.
A study published in 2021 also used a triple-blind protocol. Nine mediums completed a total of 38 readings. When the relatives compared the readings, they identified the intended reading in 65.8% of cases, versus the 50% expected from random choices between two options. The researchers also found that the intended readings contained, on average, 29.5% more accurate information than the control readings. The results therefore indicated differences both in the relatives’ overall assessments and in the evaluation of individual statements.
A study published in 2022 examined 100 readings from 28 mediums, again using triple blinding. The relatives identified the correct reading in 65% of cases. The intended readings also received higher overall ratings than the control readings. The researchers additionally compared the ratio of accurate to inaccurate statements and found a statistically significant difference favoring the intended readings. These findings were therefore consistent with earlier positive results.
A meta-analysis published in 2020 combined results from 18 experiments reported in 14 scientific articles. A meta-analysis brings together findings from multiple studies to investigate whether an overall pattern emerges. The researchers found a combined positive effect that was statistically above chance. They also reported that the findings passed two tests for possible publication bias—the risk that studies with positive results are more likely to be published than studies without such results.
Taken together, these investigations show that some controlled experiments have reported positive findings even when researchers have tried to limit clues and ordinary information transfer. The findings raise questions about how the mediums obtained the information, but they do not establish that it came from deceased people. Further methodological scrutiny and independent replications are needed to assess the overall reliability of the research.