For clinicians
Tell Your Clients What You Don’t Know
A generation of therapists confidently told clients about catharsis — hit the pillow, let it out, get it up and out of the body. The idea is roughly a hundred and thirty years old, it has been tested repeatedly, it consistently fails, and it frequently makes aggression worse rather than better.
Those therapists were not careless. They were reading the books they were handed, in which findings and clinical hunches were printed in the same typeface, with no way to tell which was which.
That is not a stylistic complaint. It is the mechanism by which a discredited idea survives for a century inside a helping profession.
The four tiers
A discipline that fits on a page and changes what you say out loud.
• Tier 1 — Solid. Replicated across laboratories or samples; would survive a hostile review. State it to a client as a fact about people and put it in a chart note. Acute stress degrades the cognitive functions this conversation requires, and it recovers over roughly twenty minutes once the stressor stops.
• Tier 2 — Reasonable. Real evidence, but smaller than advertised, or narrower than the use you are putting it to, or drawn from populations unlike your client. State it as a finding and name the limit in the same breath.
• Tier 3 — Clinical. Yours, or your field’s. A map somebody drew that has proved useful in rooms. No data. Offer it as a way of looking, never as a finding, and say so to the client.
• Tier 4 — Contested or rejected. Circulated widely; unsupported or contradicted. Worth knowing so you recognise it when a supervisee, a client, or a podcast brings it to you.
The tier that does the work is the third one, because it is where most of what we actually use lives — and because naming it costs something.
Tiering your own models
It is easy to apply this to other people’s ideas. The test is whether you apply it to your own.
Every named model I use in my own work sits at Tier 3. The typology, the sequence, the two questions underneath most recurring fights — all of it. There is no norming sample, no reliability coefficient, no cut score. These are sorting devices that have proved useful, and the book says so on the page where each one appears, in colour.
This is not modesty. It is the only position that survives contact with a client who tries the model and finds it does not fit. If you presented it as a finding, their misfit is a problem with them. If you presented it as a map, their misfit is information about the map — and you get to keep both the client and the working alliance.
How to say a Tier 3 claim out loud
Not: research shows there are four types. There is no such research, and saying it costs you something you cannot buy back.
“Can I offer you a way of looking at this? It might not fit. If it doesn’t, say so — that’s useful too.”
Then the model. Then: where does that land? Anywhere, or nowhere?
Four seconds. What you get in exchange is a client who is permitted to disagree with you, which is the only condition under which their agreement is worth anything.
Why this matters more with attachment material than almost anywhere else
Attachment language has escaped into the culture. Your clients arrive already holding it — usually in the form that lets them file a complaint about their partner. He’s avoidant. She’s anxious. That’s my attachment wound.
If your own grip on what is established and what is a useful metaphor is loose, you will not be able to tell those apart either, and the session becomes two people trading unfalsifiable descriptions of each other with a professional refereeing. Nothing said in that room can be wrong, which means nothing said in it can be useful.
There is also a duty-of-care argument, and it is the one I would put first. When a practice is untested, the people most likely to be hurt are the ones who conclude, when it does not work, that the failure was theirs.
A short list worth checking
Not to embarrass anyone — most of us have said at least one of these.
• Catharsis and the punching bag. Tested, fails, often counterproductive.
• Fixed attachment style as a stable trait that explains a person. The measures describe strategy under stress and are considerably less categorical than the popular version.
• Specific-sounding statistics about divorce prediction, delivered without their sample, their base rate, or their prospective-versus-retrospective status.
• Neuroscience vocabulary used as explanation rather than description. If removing the brain region from the sentence leaves the clinical content unchanged, the region was decoration.
The practice
For one week, tag every substantive claim you make in session with a tier in your own head before you say it. Most clinicians find two things. A surprising amount of what they say is Tier 3. And saying so out loud does not diminish their authority in the room — it increases it, because clients can tell the difference between someone who is certain and someone who is trustworthy.
What to do when a client brings you a Tier 4 idea
This happens weekly now. A client arrives with something from a podcast, a book, or an algorithm, and it is either wrong or wildly overstated, and they have organised something important around it.
Correcting it head-on rarely works and costs you standing. What works better is to take the underlying observation seriously and let the framework go.
“The thing you noticed is real — you do go somewhere when this happens, and it is fast. I am less sure about the explanation than the people who made that video are. Can we look at what actually happens, and worry about what to call it later?”
You have kept the client’s observation, which is usually accurate, and quietly declined the theory. Very few people defend a theory they were not attached to in the first place; what they were attached to was being right that something was happening.
Tiering in the chart
A short note on documentation, because this is where imprecision becomes a liability rather than an inelegance.
• A structured clinical impression is not a test result. If a score from an unvalidated instrument appears in a chart, write what it is: a structured way of asking, not a measurement.
• Do not use diagnostic-sounding language for constructs that have no diagnostic status. Attachment classifications are not diagnoses and do not belong in a record as though they were.
• If you would not be able to name the basis for a claim in front of a board, do not write it in a way that implies you could.
None of this is defensive practice for its own sake. It is the same discipline as the tiers, applied to the document that outlives the session.
From the book: “Who Are You Fighting With?” carries a tier on every substantive claim in all thirty chapters, including the author’s own models, with the full inventory in Appendix A. It is the discipline the book is organised around.
Approx. 1164 words in the body above.
Where this goes next
Who Are You Fighting With? — the Evidence Ladder and Appendix A.