How the South Park test works
This page explains exactly how the South Park test works, what it measures, and where its limits are. We'd rather you knew the method than trusted a black box.
The six dimensions
These are the axes we measure. They were chosen because they're the ones that actually separate this particular cast — not borrowed from a generic personality inventory.
Heart
Selfish ↔ Selfless
Whether other people count. High Heart gives up the last slice and sits with the kid nobody talks to. Low Heart treats people as either useful or in the way.
Chaos
Order ↔ Mayhem
How much disruption you generate and how much you need things to be predictable. High Chaos improvises, escalates, and treats a plan as a suggestion.
Authoritah
Defiant ↔ Deferential
Your relationship with rules and whoever is holding the clipboard. High Authoritah respects the chain of command and wants to be near the top of it. Low Authoritah treats every rule as an opening bid.
Cynicism
Idealist ↔ Jaded
How much faith you still have in people and in the ending of things. High Cynicism has seen this episode before and can recite the speech.
Hustle
Content ↔ Driven
Appetite for winning, building, scheming and being seen to have won. Low Hustle is genuinely fine, which the rest of the town finds suspicious.
Voltage
Chill ↔ Anxious
Your baseline arousal level. High Voltage is braced for the worst at all times. Low Voltage is unbothered to a degree that occasionally alarms people.
The scoring, step by step
- Each answer you pick awards affinity points to specific characters. The result is the character whose share of its own maximum possible affinity is highest.
- Because each character's score is divided by its own ceiling, a character that appears in fewer questions isn't penalised. Every character can reach 100% by picking only its own answers — we test that for all 16.
- The six dimensions are a second, explanatory layer. They power the bars on your result page and are never used to pick the winner.
- The displayed match percentage is an affinity strength, not a statistical probability. It is not a confidence score and it is not a diagnosis.
There are 18 questions. Each option awards points to specific characters — a primary character worth more than a secondary one. Your result is the candidate with the highest share of its own ceiling. Ties are broken deterministically by editorial priority, so the same answers always produce the same character.
What the result does not mean
- This is an entertainment quiz, not a validated psychometric instrument. There is no normative sample, no reliability study, and no clinical meaning.
- Results reflect how you answered in one sitting, which can shift with mood and context.
- The characters are fictional and the descriptions are written to be funny and a little unkind. Don't read your fate into them.
- We did not use any third-party personality dataset. The dimension model, the questions and the character profiles were all built for this site.
The secret result
One result is hidden. It is not a competing character — it's a variant that can only appear if you match one specific character and your dimension profile is unusually extreme in a particular direction. Its unlock thresholds were calibrated against simulated answer sets so that it stays rare but genuinely reachable: roughly one in eleven of the relevant results unlocks it. We test that range as part of the build.
How we test the model
The scoring code is a pure function with no randomness and no I/O, which makes it testable. The build runs a suite that asserts, among other things: every character can reach 100% by picking its own answers; every character is reachable across 300,000 simulated answer sets; no character exceeds a 15% share; the four "always pick the same position" paths give four different results; and a simulated coherent user gets their intended character in the top two at least 90% of the time.
The palette is verified the same way — the background colours are checked for perceptual distance and WCAG contrast, because the show's own palette collapses at thumbnail size (two of its reds differ by an almost imperceptible margin).
Questions about the method
Is the South Park test scientifically validated?
No, and we will not pretend otherwise. It is an entertainment quiz. There is no normative sample, no reliability study and no clinical meaning. What we can promise is that the scoring is deterministic and the method is documented on this page.
Is everyone going to get Cartman?
No. This is the classic failure mode of character quizzes and we tested against it explicitly. Over 300,000 simulated answer sets, no single character wins more than about 10% of the time, and all 16 are reachable.
Why do I get a different result to my friend?
Because you answered differently. The same answers always produce the same result — there is no randomness in the scoring. Your result page shows which of your answers contributed most.
Can a character ever be impossible to get?
It can, if the model is built carelessly — and our first version had exactly that bug. Two characters were unreachable and we only found it by simulating 200,000 answer sets. A regression test now fails the build if any character becomes unreachable.
What does the match percentage actually mean?
It is an affinity strength: how much of that character's maximum possible signal your answers produced. It is not a probability, not a confidence score, and not a diagnosis. Two different characters can both be in the 80s.
Do you store my answers?
No. There is no server-side storage at all — the test runs entirely in your browser and your progress is kept in your own browser's local storage so you can refresh safely.