TL;DR. Usually because one dark trait is doing all the work and there's nothing underneath it. Narcissism, Machiavellianism and psychopathy were named as a set by Delroy Paulhus and Kevin Williams in 2002, and the finding that matters for writers is that the three overlap without collapsing into each other. The fix on the page is to pick one as the engine, write the other two weaker, and give the character something they're genuinely good at. The fix that matters more is remembering these are normal-range research constructs, not diagnoses.

What did the 2002 study actually find?

Paulhus and Williams published "The Dark Triad of Personality" in the Journal of Research in Personality, volume 36, pages 556 to 563. They ran 245 undergraduates through the Narcissistic Personality Inventory, the Mach-IV and a self-report psychopathy scale, alongside the Big Five, an IQ test and the Over-Claiming Questionnaire, which catches people claiming familiarity with things that don't exist.

The correlations topped out around .50, and corrected for unreliability the highest pair still reached only .61. Related, not interchangeable. Low agreeableness was the only Big Five trait all three shared. Then they split: psychopathy with low neuroticism, psychopathy and Machiavellianism with low conscientiousness, narcissism with a small positive association to cognitive ability.

The one I keep coming back to is the over-claiming result. Narcissism predicted it strongly, psychopathy somewhat, Machiavellianism not at all. That's a dialogue note disguised as a statistic. A narcissist inflates his own competence in front of you; a Machiavellian won't, because inflating is a tactical error.

How does each one read in dialogue?

Narcissism sounds like a person managing an audience. The subject returns to himself, compliments get fished for, and the register changes the moment admiration stops arriving. Write the collapse, not the boasting.

Machiavellianism sounds like somebody asking questions. It's the only one of the three that mostly talks about other people. It lets a flattering misunderstanding stand, and never claims more than it can deliver, because the plan depends on being underestimated.

Psychopathy sounds flat under pressure, which is the tell: everyone else's register shifts when the stakes rise and this one doesn't. Worth knowing the construct also carries high impulsivity and thrill-seeking, not just low fear and low remorse. That's the half writers drop, and dropping it gives you the chilly chess-playing killer, a genre convention rather than a finding.

Which engine is running the character?

Amy Dunne is the Machiavellian case, in Gone Girl, written by Gillian Flynn from her own novel and directed by David Fincher. The plan runs for years, it's documented, and it correctly predicts how everyone else behaves.

Nurse Mildred Ratched is the same engine in a different costume, in One Flew Over the Cuckoo's Nest, screenplay by Lawrence Hauben and Bo Goldman from Ken Kesey's novel, directed by Milos Forman. She's never personally frightening, because the ward's procedure does it for her: cigarettes confiscated and rationed, card privileges suspended, the threat to tell Billy Bibbit's mother. After Billy's suicide she insists the daily routine continue as normal, and that insistence is the character. Forman said the Communist Party was his own Nurse Ratched, so the engine was institutional before it was personal.

Tom Ripley is the narcissism case, from Patricia Highsmith's 1955 novel, adapted and directed by Anthony Minghella, who built him as an ordinary envious man aspiring to a perfection he can see. Every lie serves that wound.

Patrick Bateman is the same engine with the wound reversed, in American Psycho, written by Mary Harron and Guinevere Turner from Bret Easton Ellis's novel, directed by Harron. Ripley wants to be somebody else; Bateman wants acknowledgment as what he already believes he is, which is why the scene everyone remembers is four men comparing business cards. Harron refused him a psychological explanation and called him a monster.

The psychopathy engine with no vanity in it is Todd Alquist in Breaking Bad, created by Vince Gilligan. He enters in "Hazard Pay," written by Peter Gould, and two episodes later, at the end of "Dead Freight," written and directed by George Mastras, he waves at a boy on a dirt bike and then shoots him. Jesse is destroyed. Todd isn't. Gilligan called him "the most weirdly likeable sociopath" he had ever seen, and Jesse Plemons described him as an ambush predator who "likes having jobs to do, likes tasks." That last phrase is the whole characterization. A narcissist wants you watching. Todd doesn't care if you remember his name.

One honest wrinkle: the research links Machiavellianism to low conscientiousness, so the meticulous long-game schemer is largely a fictional convenience. Amy Dunne's filing system isn't what the literature predicts, which is fine if you know you're writing a type.

How do you write the charm so it works on the reader?

Hans Landa, in Inglourious Basterds, written and directed by Quentin Tarantino. Tarantino has called the opening farmhouse interrogation his favorite thing he's ever written, and thought the part unplayable until Christoph Waltz read it.

The mechanism is worth stealing. Landa's charm isn't warmth. It's competence, performed at length by a man enjoying his own good work. We lean in because he's extraordinarily good at a job we don't want him to succeed at, and the tension comes from watching a skill operate rather than from waiting for violence.

So: charm first, damage second, and the charm has to be made of something, a skill or a fluency. Then let somebody be right about them early and get ignored. Billy Bibbit's arc is being right about Ratched with no standing to say so.

What are the failure modes?

Four, all diagnosable in a read.

One trait and no behavior. If "manipulative" covers the character and nothing contradicts it, you've written a label.

All three at full strength. A character who is maximally vain, strategic and fearless has no priority, so there's no scene where two of his traits disagree. The engine exists to create that scene.

Charm with no payload. Charm that isn't competence at anything reads as a writer asking us to find somebody charming, which we won't.

Visible on page ten. If the reader clocks it early you've written a warning sign. The person who gets close enough to do damage got close because something about them worked.

Where does the research stop and the cartoon start?

Further than most people assume, so be exact.

No psychiatric or psychological body recognizes a diagnosis called psychopathy. The DSM has antisocial personality disorder, the ICD has dissocial personality disorder, and psychopathy appears only as a specifier in the alternative model in the DSM's Section III. Robert Hare's Psychopathy Checklist-Revised was built for research, not clinical diagnosis. Machiavellianism has no clinical counterpart at all; it came from Christie's work turning lines of Machiavelli into a scale.

The scales measure normal-range traits by self-report, often in undergraduate samples, and Paulhus and Williams called them subclinical constructs. The literature argues with itself, too. McHoskey, Worzel and Szyarto contended in 1998 that the practical differences are slight, Muris and colleagues published a critical meta-analysis in 2017, and measurement work keeps struggling to separate Machiavellianism from psychopathy.

There's a fourth trait in play too. Erin Buckels, Daniel Jones and Paulhus argued for sadism in Psychological Science in 2013, in a study where participants chose among unpleasant tasks including grinding live pill bugs in a modified coffee grinder. Sadism predicted who volunteered, and predicted unprovoked noise blasts at a non-retaliating opponent even controlling for the other three. The triad treats harm as instrumental; sadism treats it as the point.

Before you call it a character, run this checklist

Which of the three is the engine, and can I name the scene where the other two go visibly weaker?

What is this person genuinely good at, and do we watch them do it before we watch them harm anyone?

Does their register change under pressure, and if not, was that a choice?

Would they inflate their own competence to a stranger, and does that match the engine I picked?

Who is right about them early, and what exactly stops that person being believed?

Is the charm made of a skill, or am I asserting that people find them charming?

Is there a moment where two of their traits want different things?

Does a label anywhere in my pages do work the behavior should be doing?

The label is a tool for the desk, not a line in the script, and the specific person is always stranger than the construct. Open the file and find the scene where your villain is best at something, and if there isn't one, that's your next page.

---

Sources. Delroy L. Paulhus and Kevin M. Williams, "The Dark Triad of Personality: Narcissism, Machiavellianism, and Psychopathy," Journal of Research in Personality 36 (2002), 556 to 563, read in the authors' own PDF at psych.ubc.ca, which supplies the sample, instruments, correlations, Big Five patterns and over-claiming results. The tetrad study is Buckels, Jones and Paulhus, "Behavioral Confirmation of Everyday Sadism," Psychological Science 24(11), 2013, also at psych.ubc.ca. The clinical position (no sanctioned diagnosis called psychopathy, the relation to antisocial and dissocial personality disorder, the Section III specifier, Hare's PCL-R) is from the Psychopathy entry on Wikipedia. The McHoskey, Worzel and Szyarto critique, Christie's Mach-IV and the trouble separating Machiavellianism from psychopathy in brief scales come from EBSCO's research starter on the dark triad; the 2017 meta-analysis is Muris, Merckelbach, Otgaar and Meijer in Perspectives on Psychological Science. All film and television credits, scene detail and quotes are from the respective Wikipedia entries.