Mediums Under Scrutiny: Messages From Beyond?
On this page
Can mediums really communicate with the deceased?
Imagine sitting in a quiet room while a stranger tells you specific details about your deceased grandmother—details they couldn't possibly know. In this controlled study, researchers tested whether mediums could actually provide accurate information about dead people to grieving relatives. They used a clever setup: mediums gave readings about specific deceased persons to 'proxy sitters' (stand-ins), while the real family members later rated these readings blindly alongside fake control readings. The results from their larger study were statistically striking, but they raise as many questions as they answer.
Two studies of mediums showed mixed results - one failed, one succeeded significantly.
With growing public interest in mediumship, especially among grieving people seeking contact with deceased loved ones, researchers wanted to test whether mediums can actually provide accurate information about the dead. Two controlled studies were conducted to see if mediums could deliver specific details they couldn't have known through normal means.
In a controlled experiment, mediums provided information about deceased persons that recipients could distinguish from random control readings with statistical significance.
Key Findings
- The results were contradictory: the first study showed no evidence that people could pick out their intended readings better than chance.
- However, the second study - which was much larger - found highly significant results, with people successfully identifying their readings at rates far above what would be expected by random guessing.
What Is This About?
The researchers set up a clever test using 'proxy sitters' - people who stood in for the real clients during medium readings about specific deceased persons. After the mediums gave their readings, the actual bereaved persons (who hadn't been present) were given their intended reading mixed in with several control readings from other sessions. The real sitters then rated all the readings without knowing which one was supposedly meant for them, to see if they could identify their own reading as more accurate or meaningful.
Mediums gave readings about deceased persons to proxy sitters, then real sitters blindly rated their intended reading alongside control readings to test accuracy.
First study showed no significant results, but the larger second study found highly significant evidence that sitters could identify readings intended for them above chance levels.
How Good Is the Evidence?
The second study achieved statistical significance with p < 0.0001, meaning there's less than a 1 in 10,000 chance these results occurred by random luck alone - a very strong statistical result that's much more convincing than the typical p < 0.05 threshold used in psychology research.
Supporters argue these results suggest genuine mediumistic abilities, pointing to the strong statistical significance and the controlled design that eliminated normal sensory cues. Skeptics counter that even with controls, subtle biases could influence how people rate readings, and note that the authors themselves acknowledged two potential weaknesses in their successful study. The mixed results - one study failing, one succeeding - also raise questions about reliability and what conditions might be necessary for any genuine effect.
Mainstream: The mixed results and acknowledged weaknesses suggest these findings are likely due to methodological artifacts rather than genuine mediumistic communication. Moderate: While intriguing, the contradictory results indicate we need more rigorous replication before drawing conclusions about mediumship abilities. Frontier: The highly significant results in the larger study provide compelling evidence for genuine mediumistic phenomena that deserves serious scientific attention.
Many people think mediumship research just involves asking if the medium got details right or wrong. Actually, the strongest studies use blind testing where the real clients never meet the medium and must pick their reading from a lineup, eliminating the possibility of the medium reading body language or getting feedback during the session.
To settle this question, we'd need multiple independent replications using the same methodology, larger sample sizes, and addressing the acknowledged weaknesses in the successful study. Pre-registration of analysis plans and public data sharing would also strengthen confidence. This study meets some criteria by using blind evaluation and control readings, but the mixed results and acknowledged limitations mean more rigorous follow-up studies are needed.
In the first study, the results were not significant. In the second, much larger study the results were highly significant (z = -3.89, p < 0.0001, 2-tailed).
Stance: Mixed
What Does It Mean?
The statistical significance was remarkably strong (p < 0.0001), meaning there's less than a 1 in 10,000 chance these results happened by pure coincidence. What's particularly intriguing is that real family members could consistently pick out readings meant for them from a lineup of decoy readings—as if the information had an unmistakable personal fingerprint.
It's like testing whether a fortune teller can really 'read' you by having them give readings to a stand-in, then seeing if you can pick out which reading was supposedly about you from a lineup - except here, people were trying to identify readings about their deceased loved ones.
If these results hold up under further scrutiny, they would suggest that some mediums can access information through unknown means—challenging our understanding of consciousness, memory, and the nature of death itself. This could revolutionize how we think about the continuity of personal identity and whether some aspect of human consciousness persists beyond physical death. It would also validate the experiences of countless bereaved individuals who report meaningful connections with deceased loved ones.
This study demonstrates the importance of replication in science - when one study fails and another succeeds using similar methods, it suggests we need more research rather than accepting either result as definitive.
Understanding Terms
What This Study Claims
Findings
The second, larger study showed highly significant results (z = -3.89, p < 0.0001)
moderateThe first mediumship study produced non-significant results
moderateInterpretations
Mediums may be able to produce specific and accurate information to which they have had no normal access
weakLimitations
The successful study had two possible weaknesses that the authors acknowledged
moderateImplications
Controlled studies of mediums are needed to provide criteria for judging mediums and to determine if they can produce accurate information without normal access
weakThis summary is for general information about current research. It does not constitute medical advice. The scientific interpretation of these results is debated among researchers. If personally affected, please consult qualified professionals.