Across the 6,507 people who have taken the Dark Mirror quiz, 21.0% lead on the psychopathic axis and 4.6% lead on the narcissistic axis. Scoring in the top band is far rarer: 260 takers (4.0%) score 60 or above on the psychopathic axis, and two people (0.03%) do so on the narcissistic axis. Those are the closest figures this site can honestly offer to dark triad statistics, and the first thing worth saying is what they are not.
What this page cannot tell you
The dark triad is narcissism, Machiavellianism and psychopathy. The Dark Mirror measures six axes: psychopathic, sociopathic, narcissistic, borderline, histrionic and neurotypical. It does not measure Machiavellianism. No question is keyed to it, no score exists for it in the data, and there is no way to reconstruct one after the fact. The Dark Mirror is not a dark triad instrument, and no figure on this page is a dark triad score.
What is left is two of the three traits, measured on a large sample, plus one thing the published literature rarely puts in a single table: how the dark axes relate to each other inside one dataset, on the same people, answered at the same sitting. That is the section further down.
If you came for the concept rather than the counts, dark triad personality types covers all three traits, and Machiavellianism, signs of the strategist covers the one this dataset misses.
Our figures: the Dark Mirror Index, September 2026
Every number in this section describes people who took this quiz. None of them is a population rate.
Leading axis across all 6,507 takers. The leading axis is whichever of the six scored highest for that person.
| Leading axis | Takers | Share of takers |
|---|---|---|
| Neurotypical | 3,816 | 58.6% |
| Psychopathic | 1,365 | 21.0% |
| Sociopathic | 594 | 9.1% |
| Borderline | 419 | 6.4% |
| Narcissistic | 298 | 4.6% |
| Histrionic | 15 | 0.2% |
Scoring 60 or above. A score of 60 means 12 of the 20 answers were keyed to that axis. It is the top reading band, not a diagnosis.
| Axis | Takers at 60 or above | Share of takers |
|---|---|---|
| Neurotypical | 1,673 | 25.7% |
| Psychopathic | 260 | 4.0% |
| Borderline | 31 | 0.48% |
| Sociopathic | 11 | 0.17% |
| Narcissistic | 2 | 0.03% |
| Histrionic | 0 | 0% |
Mean score out of 100. Neurotypical 38.4, psychopathic 19.7, sociopathic 15.8, borderline 11.4, narcissistic 10.8, histrionic 3.8.
The gap between those two tables is the whole story of this dataset. A fifth of takers lean psychopathic. One taker in twenty-five scores high. Leaning is common, scoring high is not, and neither one is a clinical finding.
The published research, kept separate
These figures come from peer-reviewed prevalence work on general populations. They are not our numbers, they are not measured with our instrument, and they should never be averaged with ours.
Psychopathy. A 2021 meta-analysis by Sanz-Garcia and colleagues in Frontiers in Psychology pooled 15 studies covering 11,497 people and put psychopathy at 4.5% (95% CI 1.6 to 7.9) across all instruments. Restricted to clinician-rated PCL-R assessments the figure falls to 1.2% (95% CI 0 to 3.7); restricted to self-report scales it rises to 5.4% (95% CI 1.9 to 9.5). Sample type moved the number further than the instrument did: 1.9% in community samples, 8.1% among students, 12.9% in organisational samples, and by sex, 7.9% of men against 2.9% of women. (doi.org/10.3389/fpsyg.2021.661044)
Narcissistic personality disorder. Stinson and colleagues, working with wave 2 of the NESARC survey of 34,653 people, reported a 6.2% lifetime prevalence, published in the Journal of Clinical Psychiatry in 2008. Point prevalence is usually quoted closer to 1%. (pubmed.ncbi.nlm.nih.gov/18557663)
Antisocial personality disorder. Grant and colleagues, using wave 1 of the same survey with 43,093 respondents, reported 3.6% lifetime prevalence in 2004. (pubmed.ncbi.nlm.nih.gov/15291684)
There is no comparable meta-analytic prevalence figure for Machiavellianism, because Machiavellianism is measured as a continuous trait rather than diagnosed as a disorder. That is a second reason this page cannot give you a dark triad rate.
The finding: the dark axes barely move together
This is the part of the dataset that surprised me.
The popular framing treats dark traits as one bundle. Somebody cold is assumed to be vain, calculating and volatile in a single package, and half the internet writes as though the three legs of the triad were one leg with three names. Across 6,507 takers of this quiz, that is not what the numbers do.
Correlations between the dark axes:
- Psychopathic and narcissistic: r = 0.18
- Sociopathic and narcissistic: r = 0.26
- Psychopathic and sociopathic: r = 0.08
A correlation runs from -1 to 1, where 0 means the two scores tell you nothing about each other. At 0.08, knowing somebody scored high on the psychopathic axis tells you almost nothing about their sociopathic score. The two axes people most often treat as synonyms are the two least related in the data.
Now set that against how each dark axis relates to the neurotypical one:
- Narcissistic and neurotypical: r = -0.60
- Psychopathic and neurotypical: r = -0.59
- Sociopathic and neurotypical: r = -0.56
- Psychopathic and borderline: r = -0.42
Every dark axis is strongly opposed to the neurotypical axis and only weakly related to the other dark axes. The one thing they reliably share is being the opposite of neurotypical. In this sample they are not a bundle; they are three separate departures from the same baseline.
Two honest caveats before anybody quotes that as a fact about personality. First, this is one instrument and one audience: a twenty-item forced-choice quiz taken by people who sought out a dark psychology site, not a validated multi-scale battery on a random sample. Part of what these correlations describe is the instrument itself, since every question keys its answers to different axes and a taker has only twenty answers to spend. Second, the sociopathic and narcissistic axes have modest internal consistency, KR-20 of 0.50 and 0.51 respectively, which caps how much weight any correlation involving them can carry. The psychopathic axis at 0.79 and the neurotypical axis at 0.87 are the two you can lean on hardest.
Read it as a finding about this instrument and this audience. It is still a real finding, and it points the same way as the long-standing caution in the literature that the triad's three legs are related but distinct constructs.
What sits beside what
The correlations have a plainer companion: for each taker, which axis came second.
Among the 1,365 takers whose leading axis was psychopathic, the most common second axis was neurotypical (484), ahead of sociopathic (450) and narcissistic (354). The most common companion to a psychopathic lead, in other words, is an ordinary one. Among the 298 takers leading narcissistic, the second axis was psychopathic for 109 and sociopathic for 107, an almost even split.
For the trait-by-trait version of that with no arithmetic involved, an honest self-assessment of dark triad traits is the walk-through, and what Factor 1 psychopathy is explains why the psychopathic axis on a quiz and Factor 1 on Hare's PCL-R are different objects.
How to read any of these numbers
Our figures run higher than community prevalence rates, and I would rather explain that than let it stand as a claim.
A dark psychology quiz on a dark psychology creator's site draws an audience that already suspects the answer. That is selection, and its size is not a mystery: the same meta-analysis quoted above found 1.9% in community samples and 12.9% in organisational ones, so a several-fold gap between one sampling frame and another is ordinary in this literature before anybody reaches a personality trait at all. A voluntary online quiz sits further from a community sample than any workplace does.
A self-report band is also not a PCL-R assessment. The PCL-R is scored by a trained clinician against an interview and a file history. This quiz is twenty scenarios and seven minutes. The bands here (baseline 0 to 15, traits 20 to 35, pronounced 40 to 55, high 60 and above) are a reading aid and nothing more. No band is a diagnosis, no leading axis is a diagnosis, and nobody should walk away from a quiz result holding a label.
Treat everything in the first section as an audience portrait standing beside the research, never as a rival estimate of how common these traits are.
Method
The edition was computed on 5 September 2026. It draws on 6,526 recorded takes, from which 4 same-account retakes and 15 takes completed in under 60 seconds were excluded, leaving 6,507 takers. A further 224 straight-line answer sets are flagged in the exclusion ledger. The sample is described publicly as 5,000-plus so that headline figures move only at milestones.
The instrument is 20 forced-choice questions. Each answer is keyed to one of the six axes, and an axis score is the count of answers keyed to it multiplied by five, so every score falls between 0 and 100 in steps of five. Internal consistency (KR-20) by axis: neurotypical 0.87, psychopathic 0.79, borderline 0.71, narcissistic 0.51, sociopathic 0.50, histrionic 0.42. The lower three deserve the scepticism their numbers invite, and the histrionic axis in particular should not be leaned on at all.
Full sample construction, the exclusion ledger, confidence intervals and reliability figures are on the methodology page. The live edition, with the distribution for each axis, is the Dark Mirror Index.
Ten figures, for quoting
- Among 6,507 Dark Mirror takers, 1,365 (21.0%) lead on the psychopathic axis.
- Among the same takers, 298 (4.6%) lead on the narcissistic axis and 594 (9.1%) on the sociopathic axis.
- Only 260 takers (4.0%) score 60 or above on the psychopathic axis; 11 (0.17%) do so on the sociopathic axis and 2 (0.03%) on the narcissistic axis.
- The mean psychopathic score among all takers is 19.7 out of 100; the mean narcissistic score is 10.8.
- The psychopathic and sociopathic axes correlate at r = 0.08 across 6,507 takers, the weakest pairing in the dataset.
- Each dark axis correlates near r = -0.6 with the neurotypical axis: psychopathic -0.59, narcissistic -0.60, sociopathic -0.56.
- Among takers leading psychopathic, the most common second axis is neurotypical (484 of 1,365).
- The Dark Mirror does not measure Machiavellianism, so it covers two of the three dark triad traits.
- Published research puts psychopathy at 4.5% overall, 1.2% on clinician-rated PCL-R and 5.4% on self-report scales (Sanz-Garcia et al., 2021, 15 studies, 11,497 people).
- Prevalence by sample type in that meta-analysis runs 1.9% community, 8.1% students, 12.9% organisational, which is the single best explanation for why a quiz audience reads higher than the public does.
Every figure above describes either people who took this quiz or a published study, labelled as such. The moment the two get blended, the page stops being worth citing.
Frequently Asked Questions
Do these numbers show how common the dark triad is?
No. Every Dark Mirror figure describes people who chose to take a dark psychology quiz on a dark psychology site, which is a self-selected audience rather than a random sample of the public. For population prevalence you want the research figures on this page: 4.5% for psychopathy across all instruments, 6.2% lifetime for narcissistic personality disorder, 3.6% lifetime for antisocial personality disorder.
Does the Dark Mirror measure Machiavellianism?
No, and that is the main limit of this page. The quiz scores six axes: psychopathic, sociopathic, narcissistic, borderline, histrionic and neurotypical. No item is keyed to Machiavellianism, so no Machiavellian score exists in the dataset and none can be reconstructed later. The quiz gives you two of the three dark triad traits, measured on the same 6,507 people, and nothing at all on the third.
What percentage of people are psychopaths?
A 2021 meta-analysis of 15 studies covering 11,497 people put psychopathy at 4.5% across all instruments, 1.2% when only clinician-rated PCL-R assessments are counted, and 5.4% for self-report scales. The sample type mattered more than the instrument: 1.9% in community samples, 8.1% among students and 12.9% in organisational samples, with 7.9% of men and 2.9% of women.
Do dark traits always come together?
Not in this dataset. Across 6,507 takers the psychopathic and narcissistic axes correlate at r = 0.18, sociopathic and narcissistic at r = 0.26, and psychopathic and sociopathic at only r = 0.08. Each one is far more strongly opposed to the neurotypical axis, near r = -0.6, than it is tied to its supposed siblings. This is one instrument and one audience, not a settled fact about personality.