Drawing 28 questions from an 84-question bank: how I score a deadly sins quiz
AI-assisted: I used an AI model to help draft and tidy this post. The design decisions and the site are mine.
Most online personality quizzes end the same way: a single label. As a developer that always bothered me, because a label throws away almost everything the answers contained. When I built a quiz around the traditional seven deadly sins (pride, greed, lust, envy, gluttony, wrath and sloth) I wanted the opposite: seven numbers, a formula you can check by hand, and a result that doesn't claim more than 28 answers can support.
To be clear about the topic, this is the classical list of vices, not the characters from the Seven Deadly Sins anime or manga. It is for entertainment and self-reflection, not a diagnosis.
You can try it at 7 Deadly Sins Test. Here is how the question selection and scoring work, and why.
An 84-question bank, 28 questions per run
The bank has 84 statements, 12 per sin. Each run draws 28 of them: four per sin. Within a sin, the draw takes two forward statements and two reverse ones, and picks them from different facets of that sin, so the four items don't just rephrase one idea. The 28 are then shuffled with one constraint: the same sin never appears twice in a row. Every new test and every retake draws again, so a retake isn't a replay.
Why mix forward and reverse items? A forward statement means agreeing shows more of the sin ("I put tasks off until the last possible moment"). A reverse statement means agreeing shows less of it (keeping your cool when someone annoys you). If every item were forward, someone who simply agrees with everything would get a high score on all seven sins. With two of each, that habit cancels out. It is a standard trick in questionnaire design, and cheap to implement.
The scoring formula
Each answer is 1 (strongly disagree) to 5 (strongly agree), and each item is worth 0 to 4 points:
- Forward item: points = answer − 1.
- Reverse item: points = 5 − answer.
Add the four items for a sin to get 0 to 16 points. The score is points ÷ 16 × 100, rounded, so 0 to 100 percent. The method page walks through the rules.
A worked example: on one sin you answer 5 and 4 on the two forward items and 2 and 3 on the two reverse items. That is 4 + 3 + 3 + 2 = 12 points, or 75 percent. Answer 3 on everything and every sin lands on exactly 50 percent. I like that property as a sanity check: a perfectly neutral person gets a perfectly flat profile.
There is no weighting and no hidden model, which means anyone can recompute their own result.
Ties and flat profiles are results too
A single label forces a winner even when there isn't one. With seven scores I wrote two explicit rules:
- Any sin within 3 percentage points of the top score shares the lead, so a tie is never settled by picking one.
- If the highest and lowest scores are 10 points apart or less, the result is called balanced and says that no sin clearly leads.
Both rules matter because of the "neutral all the way down" case. Someone who answers 3 to everything shouldn't be told they are definitely one sin. The quiz has nine cartoon characters: one for each sin, one for scoring low on everything, and one for having no leader.
The result is a pure function of the seven scores, and the result link only carries those seven. The free result also includes a share card in two sizes, 1080×1920 for stories and 1080×1350 for a 4:5 post, and a compare view for putting two profiles side by side, with no account.
Where the answers go
The free result is computed in your browser and isn't sent to the server. The one exception is the optional paid deeper report: if you buy it, your answers are sent so the report can be generated. I'd rather say that plainly than claim the answers never leave your device.
What the number does not mean
A score like "80%" looks scientific, and it isn't. The site says it is not a validated psychological instrument, hasn't been tested on a large sample, and has no reliability figures. The percentage is not compared with other people: 80 percent only means you agreed strongly with most of that sin's four statements. The page on how to read your result suggests reading your top two sins as a pair, looking at your lowest score too, and treating it as a snapshot.
I kept the writing separate from the formula so it can change without touching it:
- The seven deadly sins meaning covers where the list comes from.
- The deadly sins personality types describe how each leading sin tends to show up at work, with friends and under stress.
- The seven heavenly virtues page pairs each sin with its traditional opposite (humility for pride, charity for greed, chastity for lust, kindness for envy, temperance for gluttony, patience for wrath, diligence for sloth).
The lust statements are about desire, craving and being swept away by excitement, nothing explicit.
What I'd do differently
Drawing from a bank has a cost I haven't measured. Two people with the same habits, or the same person on two days, answer different items, so part of the difference between two runs comes from the draw and not from the person. Four items per sin keeps the quiz to about four minutes, but it also makes each score noisy. Before claiming anything stronger, I'd want to measure how much a retake moves the scores, and I haven't yet. For now the honest position is the one on the site: a prompt for reflection, not a measurement.
If you've built a scored quiz or assessment, how did you decide what the number was allowed to claim?
Disclosure: it's my site. 7 Deadly Sins Test is independent and not affiliated with any church, publisher or The Seven Deadly Sins (Nanatsu no Taizai) manga or anime. The free test and result cost nothing; there is an optional paid deeper report.