Now let me say up front, this is not meant to dismiss the real pain and suffering you and others have experienced in relationships. It’s not a denial that narcissism itself exists and is often more a spectrum than a true personality disorder according to empirical data. I want to hold space for the pain that you’ve been through at the hands of an underdeveloped personality that you’ve spent years with, but I also want to deal with something deeper. But in no way am I seeking to communicate to anyone that abuse of any kind, whether it comes from a personality disorder or just an unconscientious individual, is acceptable.
There is a thing that happens when people find out what I do for a living, usually somewhere between the second drink and the moment the food arrives, and it has become so regular that I can feel it coming a full sentence before it lands. Somebody leans in. They tell me they were married for eleven years. And then the first actual fact I learn about those eleven years, before I know whether there were children in the house or what either of them did for work or what the fighting was about, is that the other person was a narcissist.
The table nods. Somebody says, “God, yes, mine too.” And then the conversation moves along to parking.
Nobody asked a follow-up question. Not one. If that person had said “we could not stop having the same fight about her mother,” four people would have leaned in and the night would have gone somewhere. But there are no follow-up questions to a diagnosis. A diagnosis is not an opening, it is a closing, and everybody at that table understood it as one, and the map is set.
Some marriages end because one person did terrible things. That happens, it is not as rare as any of us could like it to be. What I am after is the word itself. Where it came from, what it is built to do, and why it has traveled so far past the thing it names that the thing it names has almost disappeared underneath it.
The biggest look anyone has taken at this in the United States is the second wave of the National Epidemiologic Survey on Alcohol and Related Conditions, and Stinson and colleagues published the narcissism numbers from it in 2008. Face-to-face interviews, 34,653 adults, a nationally representative sample. They found a lifetime prevalence of narcissistic personality disorder of 6.2 percent, higher in men at 7.7 percent than in women at 4.8 percent. While six percent sounds like a lot, hold it loosely, because it is the ceiling and not the number you should be focused on.
A systematic review by Dhawan and colleagues in 2010 gathered every community study that had used a real structured interview, seven of them, and found an average prevalence of 1.2 percent, with individual studies landing anywhere from zero to that same 6.2. A later review of five epidemiological studies, the one cited in the professional reference literature alongside the DSM-5-TR criteria, put the median at 1.6 percent. The community range quoted around those criteria still runs from zero to 6.2 percent, which is a polite way of saying the field does not have this pinned down.
So why is the spread that wide, and which end is telling the truth?
Trull and colleagues went back to the same NESARC data in 2010 and changed exactly one rule. The original survey counted a personality disorder when a person endorsed enough criteria and at least one of those criteria caused distress or impairment. Trull’s team required that every criterion counted toward the diagnosis be attached to actual distress or impairment. In their words, it seems difficult to argue that a person has a personality disorder if only one criterion of that disorder is associated with distress or impairment.
That single change took narcissistic personality disorder from 6.2 percent to about one percent. Across all the personality disorders it took the total from 21.5 percent down to 9.1 percent, which brought the American figures in line with what other epidemiological studies in the United States and Britain had been finding.
Nothing about the people changed. What changed was whether we were counting a trait or counting a life the trait was wrecking. Six percent is roughly how many Americans have enough narcissistic features to check the boxes. One percent is how many are impaired by them. That gap is not a rounding error, it is the entire argument, and almost every popular claim about narcissism lives inside it without ever saying so.
The objection I get every time I say this out loud is a good one. “But real narcissists never go to therapy, so of course your numbers are low.” You are counting the ones who show up. But this is a fundamental misunderstanding of how research and science work. That is the whole point of NESARC. It was not a clinic study and it was not a patient sample. Census-trained interviewers went to households across the country and sat down with people, 43,093 of them in the first wave, and asked the questions. Nobody in that sample had to be seeking help, or aware they had a problem, or willing to admit one. The survey went out and found the people who will never in their lives walk into my office, which is exactly why it produces a higher number than clinics do, not a lower one.
The undercounting objection predicts that community surveys will find more narcissism than treatment settings find. They do. Six point two percent is that prediction coming true. And it still tops out at six percent, and it still collapses to one when you ask whether the traits are actually ruining anything.
So run it on your own life. At one percent, if you put a hundred people in a room and every one of them told the story about their ex, roughly one would be describing narcissistic personality disorder and ninety-nine would be describing something else. Take somebody with three exes they have all called narcissists, which is not an unusual number of exes and is a very common number of narcissists, and at that rate the odds that all three met criteria are about one in a million. Insist on the ceiling, on the full 6.2, and three in a row still lands somewhere around one in four thousand.
Those are not odds. That is a vocabulary.
There is a deeper problem underneath the numbers. In 2018 Aslinger and colleagues went looking for the line. They took five separate datasets, a clinical sample, a mixed sample, a community sample, an undergraduate sample, and the whole 34,653-person NESARC file, and tested whether narcissistic personality disorder behaves like a category people fall into or a dimension people sit somewhere on. A single dimension won in four of the five and was the best interpretable model in the fifth. Foster and Campbell had found the same shape eleven years earlier in the Narcissistic Personality Inventory, with no point anywhere along the scale where a person stops being normal and becomes a narcissist.
There is no line. There is a dial, and every human being reading this is on it, including me, including the person you left. Which means that narcissist as a noun, a kind of person, a species you either are or are not, is not a finding. It is a habit of speech. The adjective is real and the noun is a story we tell with the adjective.
So why did the noun win?
Follow it to where it lives now. In March of 2022, MindSite News counted 1.9 billion views on the NarcTok tag and 1.6 billion on narcissism, on a platform discussing a condition that affects about one person in a hundred. Three years later a team led by Hudon published an analysis in the Journal of Medical Internet Research of a thousand TikTok mental health videos across three languages. Personality disorder content was the single largest topic in the sample, 207 videos out of the thousand, and 207 out of the thousand cited a source of any kind. Their read on the content was direct, that these videos frequently portrayed traits of borderline or narcissistic personality disorder using inaccurate or stigmatizing descriptions, often lacking any clinical basis. Sixty-three of the thousand carried outright disinformation and another 157 carried misinformation.
That is the supply. The demand is not stupidity. People reach for this word because it works, and it works because of what a label does to distance. This intersects with my main thesis about mental health and the relationships that we are all trying to navigate, because the map is not the territory. At the bedrock of that concept is that when we define someone as a label, we have flattened the dimensions of a real breathing human being into a category.
This makes sense from a protective standpoint. Sometimes it relieves us from having to try any longer in a relationship that is too painful, too abusive, or where the other person is just not willing to try and grow. It is also often the easier, unless difficult, thing to assign the label and therefore not have to deal with the multi-dimensional factors that make humans human.
There is a good study on this that has nothing to do with romance. Angermeyer and Matschinger ran a representative survey of just over five thousand German adults, handing them a case description, and for half of them attaching a psychiatric label to it. The label did not change the described behavior. It changed the readers. Attaching it raised the perception that the person was unpredictable and dangerous, and that perception raised the desire for social distance. The word did the separating on its own, independent of the facts.
That is not a bug in the word. That is the word’s function. And if what you needed at the end of a marriage was distance, then it delivered exactly what you came for, faster and cleaner than any account of what actually happened between two people could have.
There is a second mechanism stacked on top of it. Bertram Malle ran a meta-analysis in 2006 across 173 studies of the actor-observer asymmetry, the old idea that we explain our own behavior by our circumstances and other people’s by their character. He found the overall effect close to nothing, which was a small scandal at the time. But it held for negative events, and it held when the actor and the observer were intimates.
Read that last clause again. The people most likely to explain your behavior by what kind of person you are, and their own by what kind of week they were having, are the people who know you best, and only when the behavior is bad. That is not a moral failing. It is the standard operating condition of intimacy under stress, it runs in both directions, and it is running in your ex right now, about you.
We handed you this. Not TikTok. Us, the professionals, first. My field spent a generation teaching the public to name things, we called the naming insight, we let insight stand in for treatment because insight is what fits inside a fifty-minute hour and inside a book, and we never taught the second half. So people learned the vocabulary beautifully and got nothing to do with it. You can diagnose the man you married with real fluency and you still cannot say what happened between the two of you on an ordinary Tuesday, and that gap is ours and not yours.
What was it, then, if it was not a personality disorder?
Usually it was avoidance. Mickelson, Kessler and Shaver ran attachment questions inside the National Comorbidity Survey in 1997, roughly eight thousand adults, nationally representative, and found about 59 percent secure, about 25 percent avoidant, about 11 percent anxious. One in four. Depending on which end of the prevalence range you take, avoidance is somewhere between four and twenty-five times more common than the disorder, and it produces almost every surface behavior the word narcissist gets used to describe. The going quiet. The leaving the room. The reasonableness that lands like a wall. The unbearable calm while you are coming apart. Horney named the same movement in 1945, that out of the chaos three lines crystallize, and a child can move toward people, against them, or away from them. Away is not a character. It is what somebody learned to do when getting closer stopped being safe, usually long before you met them.
Or it was rupture with no repair, the same injury opened and reopened for years because neither of you had ever once seen it done differently. Or two people so fused that neither could hold a position without the other reading it as an attack. Or a nervous system that flooded faster than either of you could track it. Or, sometimes, and this one is almost never said out loud, two decent people who did not fit and stayed anyway.
The secret to stopping the dance is to recognize that no one has to be the bad guy. The accuse/accuse pattern itself is the villain here, and the partners are the victims.
Sue Johnson
Everything above can be misread by exactly the wrong person, or used by someone to excuse their poor behavior. That is not the intention of this article.
If you were hit, or threatened, or controlled financially, or monitored, or cut off from the people who loved you, or frightened of what would happen when you said the wrong thing, or coerced sexually, or systematically told that things you remember clearly never happened, then the diagnosis is irrelevant. It was never the point. The harm is the fact, it needs no clinical name to be real, and nothing above applies to it. In the United States the National Domestic Violence Hotline is 1-800-799-7233, it is free, and somebody answers it at any hour.
That paragraph exists because narcissist is sometimes the only word a person has ever been given for a situation that has a much more precise one. If that is you, use the precise one.
For everyone else, the word costs something.
Korzybski wrote in 1933 that a map is not the territory it represents, but, if correct, it has a similar structure to the territory, which accounts for its usefulness. Read the conditional. A map earns its keep by sharing structure with the ground. Diagnosis is a genuinely good map, one of the better ones we have, and it was drawn for a specific job, which is deciding what treatment a suffering person should receive. Pointed at somebody who is not your patient, inside a relationship that is already over, it shares structure with nothing. It is a map of a country you are not standing in.
And it does something worse than being inaccurate. It occupies the space. Once the word is applied, every memory reorganizes around it, every ambiguous moment resolves in its favor, and the actual relationship, the sequence of who did what and what each of you learned to expect, quietly disappears underneath the verdict. Analysis can make someone feel labeled rather than seen. It does the same thing to a memory.
We have never had more language for suffering than we have right now, and never had less of the thing the language was supposed to buy us. More words, less understanding. That is not a paradox, it is a trade. Every word that arrives finished is a word that ends an inquiry, and a hundred finished words will end a hundred inquiries and leave you fluent and no closer to the ground than you were the day you started.
The second cost shows up later, in the next relationship. Certainty stops data collection. The moment you know what kind of person you were married to, you stop gathering, and the version of you that walks into the next thing is carrying a template where a pair of eyes should be. That is how somebody ends up with three narcissists. Not by repeatedly selecting the disordered one percent, but because the file got written after the first one and everybody since has been read into it.
The last cost is the one I care about, because it is the only one that closes.
A relational wound does not heal by being correctly named. It heals in relationship, which is an inconvenient fact and the most reliable one in my field. The flinch you still get when somebody goes quiet on you is not an argument you can win, it is a prediction your body made out of evidence, and predictions do not update through explanation. They update through repeated contrary evidence, delivered by another person, over ordinary time, on unremarkable days. A label cannot give you that. Nothing pointed at an absent person can. It can only tell you where the wound came from, and knowing where a wound came from has never once closed one.
You do not have to turn every thought into a verdict. You can simply describe the experience.
So describe it. Not what he was. What happened. What he did when you asked for something, what you did when he did that, where it always ended, and what you stopped saying somewhere in year three because saying it had stopped being worth the week that followed. That is the territory. It is slower than the word, it is much less satisfying at a dinner table, and it is the only version of that marriage that has anything left in it for you.
I keep asking myself what I stopped being able to see about people once I had a vocabulary good enough to file them with. I do not have a clean answer. What I notice is that the years I understood my own marriage least were the years I could explain it best, with labels. But labels kept me lost.
Ask yourself what you actually know about your ex, as distinct from what you have concluded. And then see whether you can write one true sentence about those eleven years that does not require them to be a kind of person.
As always, stay curious.
References
Stinson, F. S., Dawson, D. A., Goldstein, R. B., Chou, S. P., Huang, B., Smith, S. M., Ruan, W. J., Pulay, A. J., Saha, T. D., Pickering, R. P., and Grant, B. F. (2008). Prevalence, correlates, disability, and comorbidity of DSM-IV narcissistic personality disorder: results from the Wave 2 National Epidemiologic Survey on Alcohol and Related Conditions. Journal of Clinical Psychiatry, 69(7), 1033-1045.
Dhawan, N., Kunik, M. E., Oldham, J., and Coverdale, J. (2010). Prevalence and treatment of narcissistic personality disorder in the community: a systematic review. Comprehensive Psychiatry, 51(4), 333-339.
Trull, T. J., Jahng, S., Tomko, R. L., Wood, P. K., and Sher, K. J. (2010). Revised NESARC personality disorder diagnoses: gender, prevalence, and comorbidity with substance dependence disorders. Journal of Personality Disorders, 24(4), 412-426.
Aslinger, E. N., Manuck, S. B., Pilkonis, P. A., Simms, L. J., and Wright, A. G. C. (2018). Narcissist or narcissistic? Evaluation of the latent structure of narcissistic personality disorder. Journal of Abnormal Psychology, 127(5), 496-502.
Foster, J. D., and Campbell, W. K. (2007). Are there such things as “Narcissists” in social psychology? A taxometric analysis of the Narcissistic Personality Inventory. Personality and Individual Differences, 43(6), 1321-1332.
Angermeyer, M. C., and Matschinger, H. (2003). The stigma of mental illness: effects of labelling on public attitudes towards people with mental disorder. Acta Psychiatrica Scandinavica, 108(4), 304-309.
Malle, B. F. (2006). The actor-observer asymmetry in attribution: a (surprising) meta-analysis. Psychological Bulletin, 132(6), 895-919.
Mickelson, K. D., Kessler, R. C., and Shaver, P. R. (1997). Adult attachment in a nationally representative sample. Journal of Personality and Social Psychology, 73(5), 1092-1106.
Hudon, A., et al. (2025). Navigating the maze of social media disinformation on psychiatric illness. Journal of Medical Internet Research, 27, e64225.
Kapp, D. (2022, March 20). What’s behind the narcissism obsession on TikTok? MindSite News.
Morgan, T. A., and Zimmerman, M. (2018). Epidemiology of personality disorders. In Handbook of Personality Disorders (2nd ed.).
American Psychiatric Association. (2022). Diagnostic and Statistical Manual of Mental Disorders, 5th ed., Text Revision.
Korzybski, A. (1933). Science and Sanity, Book II ch. IV, p. 58.
Horney, K. (1945). Our Inner Conflicts, ch. 2, p. 42.
Johnson, S. (2008). Hold Me Tight, ch. 4.
