The diagnosis nobody may make from a distance

In the election year 1964 the American magazine Fact printed a survey of psychiatrists about Barry Goldwater, the Republican candidate for the presidency. The issue was titled “The Unconscious of a Conservative”. None of those polled had ever examined Goldwater. A good many declared him unfit regardless, and supplied the reasoning in the vocabulary of their profession.

Goldwater sued the publisher, Ralph Ginzburg, for defamation and won: in 1969 a court awarded him $75,000. Four years later the American Psychiatric Association wrote the lesson into its code of ethics, the Principles of Medical Ethics With Annotations Especially Applicable to Psychiatry, section 7.3. It has been named ever since after the man who was argued over: the Goldwater rule.

What the Goldwater rule requires

The wording is short:

Two conditions, and both must be met: your own examination, and consent. The rule does not forbid speaking publicly about mental disorders; it forbids attributing them to a particular person who has not been examined.

Half a century later the rule came under pressure, and from within. In 2017 the American Psychiatric Association reaffirmed it expressly, after professionals had begun to comment publicly on a sitting president. The counter-position invoked a duty to warn: where a person holds power over life and death, silence weighs heavier than the breach. The association replied that precisely this weighing hollows the rule out, because it makes the exception depend on urgency — and urgency is not an examination. The quarrel is not settled. But it shows what the rule is actually about: it protects not the powerful from the profession but the profession from the temptation to deploy its authority where it has no data.

The American Psychological Association — the body for psychologists, not for psychiatrists — has no identically worded provision. The same conclusion is nonetheless drawn from its ethics code: no diagnosis in the media of a living public figure you have not examined.

The rule binds members of a profession, not the public. Anyone who belongs to no professional body breaks no ethical guideline by declaring a celebrity or their own ex-partner narcissistic. The methodological objection stands all the same, because it depends not on professional standing but on the available evidence: what can be observed from outside is behaviour, not its unambiguous cause. A person may deflect criticism out of entitlement, out of fear, out of depression, out of trauma, out of an interest in power, or out of ordinary immaturity. Remote diagnosis explains too much and tests too little.

Two narcissisms that hardly look alike

Even within research the term is no single thing. It separates at least grandiose from vulnerable forms. Grandiose narcissism stresses status, dominance, entitlement and visible self-enhancement. Vulnerable narcissism combines self-focus with distrust, sensitivity to shame and withdrawal. Both share antagonistic tendencies but look very different in daily life — which explains how two people can carry the same label and would not recognise each other in it.

Research has a model for this relationship, the so-called trifurcated model. It breaks the term into three components: a shared core and two additions. The core is interpersonal antagonism, in the language of the five factors low agreeableness. It is present in both forms and is the reason they bear the same name at all. What is added is what distinguishes them: in the grandiose form an assertive, extraverted component, in the vulnerable form an emotionally labile, inward-turned one. The common denominator, then, is not self-love but the way of dealing with others.

That the popular picture nevertheless knows almost only the grandiose form also has a reason of measurement. The most widely used instrument captures chiefly that form; the vulnerable one it barely represents. A field that has worked for decades predominantly with this one questionnaire necessarily produces a literature in which narcissism looks loud, visible and self-assured. The quiet half of the term hardly appears in the public imagination — not because it is rare but because it has been measured less often.

The popular picture knows almost only the first variant. Whoever acts self-centred, unfaithful, controlling or unfeeling gets a diagnosis in retrospect. That can give people who have been harmed a language for their experience; it can also make very different problems disappear beneath an apparently deep label.

What a questionnaire score measures and what it does not

The most widely used instrument in narcissism research is explicitly not a diagnostic one. Robert Raskin and Calvin Hall introduced the Narcissistic Personality Inventory, or NPI, in 1979; they had derived it from the criteria for narcissistic personality disorder in the DSM-III. The original version had 54 items. Raskin and Howard Terry presented the 40-item standard version together with a components analysis in 1988, describing seven facets, among them authority, superiority, entitlement and exploitativeness.

The format explains a great deal. The NPI works by forced choice: each item offers two statements, and you pick the one that applies more. The result is a position on a distribution — the operationalisation of a normal personality trait in a non-clinical population. There is no threshold at which a score tips over into a disorder, because the instrument was never built for that. Someone who scores high on the NPI has selected a series of statements; they do not have a diagnosis.

This distinction is regularly flattened, and usually in exactly one direction. A number from a self-description form travels into a newspaper article and arrives there as a clinical finding. A questionnaire can say how someone describes themselves relative to others. Whether a pattern is enduring, inflexible and impairing, it cannot say.

Trait, state, disorder

Narcissistic thoughts and behaviours can occur as traits on a continuum and fluctuate with the situation. Narcissistic personality disorder, by contrast, is a clinical judgement. The pattern is enduring, inflexible, appears across several areas of life, and leads to substantial impairment or distress. Diagnosis needs conversation, a course over time and differential assessment — three things a remote diagnosis has none of.

This does not mean playing down harmful behaviour. Boundaries need no diagnosis. Repeated devaluation, control, threats or exploitation are reasons to act, whatever label fits. More precise language actually strengthens the decision: “this person crosses my boundaries” is observable; “they have NPD” can turn into an endless argument over evidence.

Observing relationship patterns more closely

Course and function yield more than labels. How does someone respond to substantive criticism: with questions, withdrawal, counterattack or contempt? Is admiration sought in order to regulate insecurity, or are others systematically belittled? After an injury to their standing, can the person take up another perspective again and repair the damage?

Questions like these distinguish passing self-absorption from rigid patterns. They also guard against confirmation bias: anyone already expecting “narcissism” reads charm as manipulation and vulnerability as tactics. A fair analysis always states which observation would refute its own hypothesis.

What remains

The Goldwater rule is not language policing but a statement about the quality of evidence. It records that between a public appearance and a finding lies a procedure that cannot be skipped: examination, course over time, consent. That it binds only members of the profession does not make it wrong for everyone else — merely not binding. The difference between an expert’s judgement and a spectator’s is not the choice of words but access to the person.

In everyday life the error is recognisable by one simple sign: the label comes first and then explains everything. Friendliness becomes a façade, remorse a ploy, withdrawal a punishment — every observation confirms the hypothesis, none can refute it. Put the term at the end instead, and you first have to say which pattern is stable, who reports it, in which situations it appears, and what would change it. Narcissism research helps to understand the dynamics of status, rivalry, admiration and injury. It should not divide people into good empaths and evil narcissists. Diagnosed people, too, are heterogeneous and cannot be reduced to a meme; treatment is possible, if often demanding.