The Dirty Dozen: 12 items for three darker traits
Machiavellianism, narcissism and psychopathy are usually measured together because they travel together. The three original scales ran to 91 items between them. Jonason and Webster cut that by 87% in 2010 and still kept enough to score each trait on its own.
Twelve statements, four per trait, each rated on a 1 to 7 agreement scale. Development ran across four studies with a combined 1,085 participants. One item about morality was rewritten after the first three studies because the original used a double negative, and that rewrite lifted the subscale's internal consistency from .63 and .69 to .77.
Three subscales
| Subscale | Items | Range | What the four items ask |
|---|---|---|---|
| Machiavellianism | 4 | 4-28 | Manipulation, deceit, flattery, exploitation |
| Narcissism | 4 | 4-28 | Wanting admiration, special favours, status, attention |
| Psychopathy | 4 | 4-28 | Little remorse, callousness, cynicism, unconcern with morality |
Nothing is reverse-scored. A total across all 12 items runs from 12 to 84, and the authors treat it as usable: a hierarchical model with one general dark factor sitting over the three specific ones fits their data well. Most researchers still report the three separately, because the traits do not move in lockstep. Correlations between them run .30 to .49 for Machiavellianism with psychopathy, .34 to .54 for Machiavellianism with narcissism, and .23 to .31 for psychopathy with narcissism.
Reliability
Internal consistency for the total was .83 in the first two studies and .86 in the fourth. The four-item subscales are the weak point by construction. Machiavellianism ran .72, .77 and .79 across the studies, narcissism .79, .78 and .84, psychopathy .63, .69 and .77. The authors note that alphas in the .60s are acceptable for a scale this short, which is true, and also a reason to hold subscale scores loosely.
Three-week retest in the third study averaged .93 for the total, with the subscales between .74 and .89.
What it correlates with
The Dirty Dozen tracks the longer instruments it replaced, which is the main evidence that the short form measures the same constructs: .33 with the 20-item Mach IV, .51 with the 31-item SRP-III, .41 with the 40-item NPI, and .47 with a 91-item composite of all three. Agreeableness is the Big Five factor that moves with it, between -.24 and -.35. Short-term mating orientation correlates .43 in the first study and .31 in the second. Against an aggression questionnaire the total scored .51, or .45 once sex was controlled.
Cross-cultural work is the strongest support. A study covering 49 countries and 11,488 people across eight world regions fitted the three-factor model in every region, with CFI from .921 to .947 and RMSEA from .062 to .080. Men scored higher on every trait in every region except one: psychopathy showed no significant sex difference in Asia. Comparisons across regions hold at the level of covariances. Full scalar invariance holds only partly, mostly because of the narcissism and psychopathy items.
Where the scale is weak
The psychopathy subscale takes most of the criticism. Four items cannot cover both interpersonal antagonism and disinhibition, and item difficulty varies widely across the set. Set against the 27-item Short Dark Triad, the Dirty Dozen reproduces the original validity network less closely. There is no cutoff score and no clinical threshold. These scores are for comparison and research, not for labelling a person, and this page is not a diagnosis.