Are Mediums Real? What the Research Actually Shows, and How to Test a Medium Reading Yourself

The short answer. Science has not settled whether mediums are real, and anyone who tells you it has, either way, is overselling. In blinded studies where neither the medium, the sitter nor the raters knew who a reading was for, the Windbridge Research Center reported accuracy above chance (Beischel and Schwartz, 2007; Beischel et al., 2015), and a 2021 meta-analysis of 14 forced-choice studies reported a small but significant effect. Other well-controlled studies found nothing (O’Keeffe and Wiseman, 2005), and critics dispute the statistics behind some positive results. What you can do is test your own reading: book with minimal information, write a sealed list of private facts beforehand, answer only yes or no, and score each statement afterwards. The protocol is below.

“Are mediums real?” is usually not an abstract question. Most people asking it have lost someone, or are thinking about booking a reading, or have just had one that felt astonishing or empty, and want to know what to make of it.

This guide gives you the honest state of the evidence, in plain English, without mocking anyone and without pretending the question is closed. Then it gives you something more useful than an opinion: a way to judge the reading in front of you.

Why So Many People Ask

Contact with the dead is not a fringe experience. In a 2023 Pew Research Center survey, 53% of US adults said they had been visited by a dead family member in a dream or some other form, and about four in ten said they had felt a strong sense that someone who had died was trying to communicate with them (Pew Research Center, 2023; Pew, Spirituality Among Americans).

So when people book a medium, they are often looking for confirmation of something they have already felt. That makes it especially important to know what a good reading looks like, and what a misleading one looks like.

What Mediums Say They Do

A few terms make the research and the advice easier to follow:

• Mental mediumship: the medium reports impressions (images, words, feelings, sensations) they attribute to a person who has died.
• Evidential mediumship: a style that prioritises specific, checkable details (names, causes of death, objects, places, habits) over general comfort. This is what you want to evaluate.
• Sitter: the person receiving the reading.
• Discarnate: the deceased person the reading is about.
• Proxy sitter: someone who sits in for the real sitter and knows little about the deceased, so they cannot give clues.

Mediumship is different from a general psychic reading about your future. If you are unsure which you need, psychic vs medium vs tarot reader explains the difference.

What the Research Found

Mediumship research is a small field with strong views on every side. Here are the studies that matter most, with their designs and the main criticisms.

Study	Design	Result	Main caveat
O’Keeffe and Wiseman (2005), British Journal of Psychology	5 mediums, 5 sitters, 25 readings; mediums and sitters in separate rooms; sitters rated statements from all transcripts blind	No significant evidence of mediumistic ability	Proponents argue the protocol made success unlikely (for example, mediums untested in advance)
Beischel and Schwartz (2007), Explore	Triple-blind; bereaved sitters rated paired transcripts without knowing which was theirs	Significant result (p = 0.007)	Small sample; cannot separate contact with the dead from other explanations
Beischel, Boccuzzi, Biuso and Rock (2015), Explore	20 screened research mediums, 86 phone readings, multiple levels of blinding	All analyses statistically significant	Run by the same research centre; independent replication needed
Rock, Thorsteinsson, Tressoldi and Loi (2021) meta-analysis	14 forced-choice mediumship studies, 2000–2020	Small, statistically significant overall effect	Meta-analyses depend on the quality of the studies included
Silva and Moreira-Almeida (2023), Journal of Scientific Exploration	Review of seven tightly controlled studies with conflicting results	Positive results more likely with mediums who had prior track records, motivated sitters, and scoring of both whole readings and individual items	A review of protocols, not new data

What the positive studies did well

The Windbridge studies were designed specifically to rule out the ordinary explanations. Readings were done by phone with a proxy sitter, so the medium heard no reactions from the real sitter. Mediums were typically given little more than the first name of the deceased. Raters scored transcripts without knowing which reading was meant for them. The research mediums themselves had passed an eight-step screening that included blinded test readings (Psi Encyclopedia: Windbridge Research Center).

What the critics say

• Null results exist under tight controls. The O’Keeffe and Wiseman study is the most cited example (University of Hertfordshire; BPS Research Digest).
• Statistics and researcher choices. A chapter in the skeptical anthology The Myth of an Afterlife (2015) argues that the statistical methods in one major Beischel and Schwartz study were misapplied.
• Who runs the research. Much of the positive work comes from a small number of researchers who openly favour the hypothesis. That does not make it wrong, but independent replication would carry more weight.

What nobody has resolved: where the information comes from

Even researchers who accept the positive results note that accuracy alone cannot show the information comes from the dead. An alternative explanation within parapsychology, sometimes called “super-psi”, holds that a medium might be picking up information from living people or the world rather than from the deceased (Psi Encyclopedia: mental mediumship research). The source question is open.

The honest summary

The evidence is mixed and contested. There are controlled studies with positive results and controlled studies with negative ones. Researchers on both sides agree on one thing that matters to you: under blinded conditions, the ordinary tricks stop working, so the way a reading is set up makes an enormous difference to what it can tell you.

How a Reading Can Seem Accurate Without Being Evidential

Knowing these techniques is not about assuming every medium uses them. It is about structuring your sitting so that, if your reading is genuine, the details stand out clearly.

• Cold reading. High-probability statements (“I’m getting a sense of unfinished conversations”) combined with close attention to your reactions. Our guide to spotting a fake psychic covers the techniques in detail.
• The Barnum effect. People rate general descriptions as highly accurate when they believe the description is about them. “She was stubborn but had a soft heart” fits most of us.
• Hot reading. Looking up the sitter or the deceased in advance: obituaries, memorial pages, social media, online guestbooks. A name and a city are often enough to find a great deal.
• The multiple-out. A statement built so that almost any answer counts as a hit: “a J name, or someone connected to a J, or a month starting with J”.
• Fishing and feeding back. A question disguised as a statement (“Who’s the one with the heart condition?”), followed by repeating your own answer back to you later as if it were new.
• Selective memory. Afterwards, sitters remember the hits vividly and forget the misses. This is why a transcript matters more than your impression.

How Readings Can Help (and Hinder) Grief

For a long time, grief theory held that healthy mourning meant “letting go”. That view has changed. The continuing-bonds model, set out in Klass, Silverman and Nickman’s 1996 book Continuing Bonds: New Understandings of Grief, argues that ongoing connection with the dead is common and not a sign of pathology. Keeping a relationship with someone who has died, through memory, ritual or felt presence, is now widely seen as a normal part of grieving.

Researchers at the Windbridge Research Center have argued that readings with mediums may support grief from this continuing-bonds perspective (Beischel, Mosher and Boccuzzi, 2014–15, Omega). Note that the authors run the centre that studies mediums, so the work should be weighed with that in mind.

In practice, people describe readings helping when they:

• bring a sense of peace, permission or completion;
• prompt the sitter to talk about the person and remember them;
• come at a point when the sitter is steady enough to hold the experience.

And hindering when they:

• become the only place the sitter feels close to the person, leading to frequent repeat sessions (see our guide to compulsive reading use);
• deliver messages that frighten, blame or burden the sitter;
• come with demands for money to “help” the deceased.

A reading can sit alongside grief support. It shouldn’t replace it. If grief is making it hard to function months on, a grief counsellor or bereavement group can help in ways a reading can’t.

How to Test a Medium Reading Yourself

This is a home version of the controls researchers use. It works whether you are hopeful or sceptical, and a good medium has nothing to fear from it. You can use all of it or just the parts that suit you.

Before the session

1. Choose the format deliberately. Chat gives you an automatic transcript. Phone removes visual cues. Video lets you see the medium but also lets them see you; if you choose video, keep your reactions neutral.
2. Book with minimal information. Use a first name or nickname and an email address that isn’t linked to your social media. Don’t upload photos of the person. Don’t say who you hope to hear from, how they died or when.
3. Check what is public. Search the person’s name, the obituary and any memorial pages, and note what is findable. Anything on that list will not count as evidence later.
4. Write a sealed list. Write 10 to 20 facts about the person that are not online: a nickname, a private joke, a habit, an object they carried, a phrase they used, a place that mattered, how they spoke. Date it and put it away. Don’t look at it during the session.
5. Consider a proxy sitting. For the strongest test, ask a friend who knows little about the person to book and sit in for you, with only a first name. They cannot give clues they don’t have. This mirrors the research protocol.

During the session

6. Give only short answers. “Yes”, “no” or “I can’t place that”. Don’t explain, correct or add detail. If the medium asks a question, it’s fine to say, “I’d rather not say yet; what are you getting?”
7. Let it run. Don’t help it along by filling silences.
8. Record it. Chat sessions are saved automatically on most platforms. For phone or video, take notes or record if the medium agrees.

After the session

9. Split the transcript into individual statements. One line per claim.
10. Score each statement using the sheet below.
11. Compare with your sealed list. How many items appeared without prompting?

The scoring sheet

Question to ask of each statement	Score
Is it specific (a name, a cause, an object, a date, a phrase) or general (a personality trait, a feeling, “unfinished business”)?	Specific = 1, General = 0
Is it correct, incorrect or can’t verify?	Correct = 1, Incorrect = −1, Can’t verify = 0
Was it private (not online, not something you said earlier) or public / given away?	Private = 1, Public or given away = 0

A statement scores as strong evidence only when it is specific, correct and private (3 points). Count those. Then count the misses separately.

A rough guide to reading your results:

• Several strong hits, few misses, including items from your sealed list: that is an evidential reading by any reasonable standard, whatever explanation you prefer.
• Mostly general statements that felt accurate: a comforting reading, which may still have value to you, but not evidence.
• Hits only on public details or things you said: discount them; the reading didn’t tell you anything new.
• Many misses with a few vague hits: a weak reading.

A worked example

Here is how scoring works on a short, illustrative excerpt (invented for this guide, not from a real session). The sitter’s grandmother has died; the sitter booked by chat with a first name only.

Statement from the reading	Specific?	Correct?	Private?	Points	Note
”She was a strong woman who put others first.”	0	1	0	1	True of most grandmothers: Barnum
”I’m hearing the name Margaret, or Maggie.”	1	1	0	2	Correct, but her name is in the online obituary
”There’s a problem with the chest or lungs at the end.”	1	1	0	2	Correct, but common cause of death and in the obituary
”She’s showing me a tin with buttons in it, a biscuit tin.”	1	1	1	3	On the sealed list, not online: a strong hit
”She lived near the sea.”	1	−1	0	0	Incorrect
”She called you ‘chicken’.“	1	1	1	3	On the sealed list: a strong hit
”There’s a man with her, a husband or brother.”	0	1	0	1	Multiple-out: could be anyone

Two strong hits and one clear miss in seven statements. On its own, that would be an interesting reading worth noting, not proof. If the full transcript held several more items like the button tin, and few misses, it would be a genuinely evidential reading by the standard used in research. If it held only lines like the first and last, it would be comforting but not evidential.

Questions you can ask without giving anything away

If you want to engage rather than sit silently, these keep the information flowing from the medium to you:

• “What else are you getting about how they looked?”
• “Is there anything about where they lived?”
• “Can you tell me more about that object?”
• “What do they want me to know?”

Avoid questions that contain the answer (“Is it my grandmother?”), and avoid confirming more than you are asked (“Yes, and she had a dog called Bramble”). Generous sitters make every medium look better, and you deserve to know what the medium actually brought.

Optional: the blind comparison

If you want to go further, ask a friend to have a reading too (about their own loved one) with the same medium or a different one. Swap anonymised transcripts and see whether each of you can pick out your own. If you can do so reliably, the readings contain something specific to each of you. This is a simplified version of the design used in the blinded studies.

Choosing a Medium

• Look for evidential focus. Profiles and reviews that mention specific details (names, causes, objects, phrases) are more informative than reviews that only say “so comforting”.
• Look for consistency over time. Years of reviews that describe similar kinds of evidence carry more weight than a handful of glowing ones.
• Value honesty about limits. A medium who says “I’m not getting a clear connection today” is showing integrity, not weakness.
• One price, no add-ons. You should pay for the session and nothing else.

If you want to compare options, best mediums on Kasamba and evidential mediums for connecting with a spouse profile specific readers, and our Kasamba review explains how one platform shows reviews and keeps chat transcripts.

Red Flags Specific to Mediumship

• “Your loved one is trapped” or “can’t cross over” and you need to pay for help. This is a scam pattern, closely related to curse-removal fraud.
• Messages that ask for money, or say the person “wants you to keep coming back”.
• Messages that blame or frighten you, such as suggesting the death was your fault.
• Requests for detailed information before the reading (full name, date of death, photos) with no reason given.
• Pressure to move off the platform to private payment.
• Guarantees of contact with a specific person.

When Not to Book

• In the rawest weeks of grief, unless you feel steady enough to judge the reading. Many mediums and grief professionals suggest waiting until the most acute stage has passed.
• If you hope the reading will stop you grieving. It won’t, and that expectation sets up disappointment.
• If you are in crisis. If you are thinking about joining the person who died, or can’t keep yourself safe, contact a doctor, a crisis line (988 in the US, Samaritans on 116 123 in the UK and Ireland) or emergency services first.

FAQs

Are mediums real?

Science has not settled it. Blinded laboratory studies by the Windbridge Research Center have reported accuracy above chance, and a 2021 meta-analysis of forced-choice studies reported a small but statistically significant effect. Other carefully controlled studies found no effect, and critics argue some positive results reflect statistical choices. Even positive results cannot show where the information comes from.

How can I tell if a medium is real?

Test the reading, not the reputation. Book with minimal information, write down beforehand a sealed list of facts about the person that are not online, answer only yes or no during the session, and afterwards score each statement as specific or general, correct or incorrect, and private or publicly findable. Specific, correct, private details are what count.

How do fake mediums know things?

Mostly through cold reading (high-probability statements and reading your reactions), hot reading (looking up obituaries, memorial pages and social media), and the multiple-out, where a vague statement can be matched to almost anyone. The sitter’s memory then keeps the hits and forgets the misses.

Can a medium reading help with grief?

For some people, yes. Grief research has moved away from the idea that healthy mourning means cutting ties with the dead; the continuing-bonds model treats ongoing connection as normal. A reading should sit alongside grief support, not replace it, and a medium who demands money to help a loved one cross over is running a scam.

What is a proxy sitting?

A reading in which someone else books and sits in for you, knowing little or nothing about the person you hope to hear from. Because the proxy cannot give away information through their reactions, any accurate detail is much harder to explain by cold reading.

How soon after a death should you see a medium?

There is no fixed rule, but many mediums and grief professionals suggest waiting until the most acute stage of grief has passed, often several months. If you are in crisis, reach out to a doctor, grief counsellor or crisis line first.

Is video, phone or chat best for a medium reading?

Each has a trade-off. Chat leaves a transcript you can score. Phone removes visual cues. Video can feel more personal but gives the medium more cues from your face. For testing, chat or a phone reading with a proxy sitter is strongest. Our comparison of chat, phone and video readings covers the costs.

Sources

Controlled studies and reviews

1. O’Keeffe, C. and Wiseman, R. (2005). Testing alleged mediumship: Methods and results. British Journal of Psychology, 96(2), 165–179. University of Hertfordshire
2. Beischel, J. and Schwartz, G. E. (2007). Anomalous information reception by research mediums demonstrated using a novel triple-blind protocol. Explore, 3(1), 23–27. ScienceDirect
3. Beischel, J., Boccuzzi, M., Biuso, M. and Rock, A. J. (2015). Anomalous information reception by research mediums under blinded conditions II: Replication and extension. Explore, 11(2), 136–142. PubMed 25666383
4. Rock, A. J., Thorsteinsson, E. B., Tressoldi, P. E. and Loi, N. M. (2021). A meta-analysis of anomalous information reception by mediums: Assessing the forced-choice design in mediumship research, 2000–2020. University of New England repository
5. Silva, J. and Moreira-Almeida, A. (2023). Controlled experiments involving anomalous information reception with mediums. Journal of Scientific Exploration, 36(4). DOAJ
6. Augustine, K. and Martin, M. (eds.) (2015). The Myth of an Afterlife. Rowman and Littlefield (chapter “Madness in the Method: Fatal Flaws in Recent Mediumship Experiments”). PhilPapers
7. Society for Psychical Research. Psi Encyclopedia: Windbridge Research Center; Modern mental mediumship research. psi-encyclopedia.spr.ac.uk

Grief and experience

8. Klass, D., Silverman, P. R. and Nickman, S. L. (eds.) (1996). Continuing Bonds: New Understandings of Grief. Taylor and Francis. Routledge
9. Beischel, J., Mosher, C. and Boccuzzi, M. (2014–15). The possible effects on bereavement of assisted after-death communication during readings with psychic mediums: A continuing bonds perspective. Omega, 70(2), 169–194. doi:10.2190/OM.70.2.b
10. Pew Research Center (2023). Many Americans report interacting with dead relatives in dreams or other ways. pewresearch.org
11. Pew Research Center (2023). Spirituality Among Americans. pewresearch.org