The D-factor: one trait under the dark side, and what 16 items can measure
Machiavellianism, psychopathy, narcissism, sadism, spite. They look like separate personalities, and for decades they were measured separately. Moshagen, Hilbig and Zettler argued in a 2018 Psychological Review paper that they share a common core, and that the core is simpler than any of the labels: the tendency to put your own interests first even when it costs other people.
They called it D, the dark factor of personality. The 2020 follow-up built the instruments, and the 16-item version (D16) is what you answer here. Sixteen statements, rated 1 to 5, averaged into a single score.
One design detail matters more than it looks. Eight of the sixteen items are reverse-scored: 1, 3, 5, 7, 9, 13, 14 and 16. Those are the ones that sound decent, like finding it hard to watch someone suffer. Agree with everything and you land near the middle rather than at the top, which is exactly the point of writing them that way.
What the items cover
Bader and colleagues mapped five themes inside the longer D70 in 2021: callousness, deceitfulness, narcissistic entitlement, sadism and vindictiveness. They are useful for describing what the scale touches, but they are not subscales you should read separately. The evidence says so directly: the themes fail the thresholds for independent use (FD under 0.90, H under 0.80). D is one dimension, and the themes are internal structure, not scores.
Scoring and what the bands mean
Reverse the eight keyed items (1 becomes 5 and so on), then average all sixteen. The result runs from 1 to 5.
| Average | Band |
|---|---|
| 1.0-2.0 | Lower: you tend to weigh other people's welfare in your decisions |
| 2.0-3.0 | Moderate: the most common range, self-interest traded off against others |
| 3.0-4.0 | Higher: self-interest more often wins, with beliefs that justify it |
| 4.0-5.0 | Very high: strong self-orientation, little regard for others' welfare |
The instrument has no official cutoff, and D is a continuous trait rather than a type. These bands describe where you sit relative to other people, not a category you belong to. A low score is not naivety, and a high score is not a diagnosis.
Psychometrics
Pooled internal consistency for the three lengths in the 2020 paper: D70 at .970, D35 at .950, D16 at .906. Shorter means less reliable, which is the ordinary trade. Swedish validation put D70 at .94 and D16 at .84; German samples ranged from 0.80 to 0.93. Retest over roughly 34 days came out at .95 for D70, .93 for D35 and .90 for D16.
Structure holds up. The D16 single-factor model fits acceptably (RMSEA 0.058, SRMR 0.056, CFI 0.899) with loadings between 0.29 and 0.67, averaging 0.51. The longer D70 correlates with D16 at r = 0.91, so the short form tracks the long one closely. Item response analysis gives D16 discriminations from 0.59 to 1.81, averaging 1.26, with every item contributing at least moderate information.
What it correlates with
D sits close to the dark traits you would expect and, interestingly, not to all of them. Correlations with Machiavellianism run 0.63 to 0.64 and with psychopathy 0.71 to 0.80. Narcissism is the exception: 0.09 and 0.06, neither significant. On the bright side of the ledger, D correlates negatively with agreeableness (around -0.58 to -0.60) and honesty-humility (-0.55 to -0.62). The interesting finding is that D predicts unethical behaviour beyond what honesty-humility alone accounts for, which is the main argument for measuring it at all.
The Chinese version
Wang and colleagues adapted the scale in 2022 with 762 Chinese university students: 251 men and 511 women, average age 19.99. Three items came out for cultural fit, on the reasoning that some content does not read as clearly negative in a Chinese context. Four more came out for low discrimination. What remains is a 28-item version with an alpha of 0.90 and marginal reliability of 0.92. A short form of 15 items keeps 62.8% of the test information. Gender differences were negligible, with one item showing a trivial effect of 0.006.
Reading your own number honestly
Two caveats from the item response work. First, the scale measures more precisely above the average D level than below it, and at the low extreme the marginal reliability drops to 0.77. So if you score low, the number is fuzzier than the same distance above the mean would be. Second, D predicts a higher likelihood of exploitative behaviour in studies; it does not determine what any individual will do. Beliefs and justifications sit between the trait and the act, which is where the 2022 work on aversive personality points.
If a high score bothers you, that reaction is worth more than the number. This test is not a diagnosis and does not replace talking to a professional.