Your friend took your quiz and got 22 out of 30. Is that good, or just average? On its own, a score doesn't tell you much. You need something to compare it with.
So we counted 231,366 plays of quizzes on Rikaido's English version. From each play we used only two things: how many questions the quiz had, and how many the player got right. We didn't look at names or anything else.
Most plays are on 30-question quizzes
When you make a quiz in the main edition, you choose 30, 60, 90, 120 or 150 questions. The Love edition offers 30 or 50. Of all the plays we counted, 93.6% were on 30-question quizzes, so that's what this article looks at: 216,646 plays.
One thing to keep in mind: the play logs don't record which edition a quiz came from. So the 30-question plays mix main-edition quizzes (five questions from each of the six categories) with Love-edition quizzes (love questions only).
The average is 21.3 out of 30
Here's how the 30-question plays came out:
- The average score was 21.3 (70.9% correct), and the median, the score right in the middle, was 21.
- 61.1% of plays scored from 18 to 24.
- A perfect 30 came up in 3.5% of plays, and a zero in 0.1%.
Scores on 30-question quizzes (216,646 plays)
The peak sits at 18 to 23 correct, above the halfway mark. With two options per question, blind guessing would land around 15 on average. Most plays score well above that.
Look closely at the top end, though: 8,322 plays scored 26, then 6,843 scored 27, 6,530 scored 28, 6,682 scored 29, and 7,558 got all 30 right. From 27 up, instead of thinning out, the numbers level off, and more plays got all 30 right than scored 27, 28 or 29. Some of that might be people playing again after they'd already seen the answers. Our How to Play & FAQ points out that nothing stops someone from doing that under another name, and the logs can't tell those plays apart from the rest.
What the titles mean
After each play, Rikaido gives the player a title based on the percentage of questions they got right, rounded to a whole number: Soulmate at 90% or more, Best friend at 73% or more, Friend at 50% or more, Acquaintance at 27% or more, and Stranger below that. On a 30-question quiz, that works out to 27 or more for Soulmate, 22 to 26 for Best friend, 15 to 21 for Friend, 8 to 14 for Acquaintance and 7 or fewer for Stranger.
Here's how often each title came up in 30-question plays:
- Soulmate: 12.7%
- Best friend: 32.5%
- Friend: 50.6%
- Acquaintance: 3.9%
- Stranger: 0.3%
What if you just bet on the popular answer?
To read a score, it helps to have a yardstick. Here's one: how many would someone get right if they knew nothing about you and simply picked the more popular option every time? By "more popular", we mean the option that got more of quiz creators' own answers.
Most 2-choice questions on Rikaido don't split evenly. Across all the answers we counted, 67.1% went to the more popular option on their question. So, averaged over many questions, a bet on the popular side would be right about that often.
We used those shares to simulate 200,000 quizzes built like a 30-question main-edition quiz: five questions from each of the six categories, drawn at random (in the simulation, questions with more answers were drawn more often, to match how often each one ends up in finished quizzes). Every guess went to the popular side and counted as right with a probability equal to that side's share.
Here's how the two compare:
- Betting on the popular answer every time: an average of 20.1 and a median of 20. It reached 24 or more in 9% of simulated quizzes, and a perfect 30 almost never (0% after rounding)
- Real 30-question plays: an average of 21.3 and a median of 21. They reached 24 or more in 27.9% of plays, and a perfect 30 in 3.5%
The averages are close. The gap shows up at the top: real plays reach 24 or more far more often than the popular-answer bet does, and they get perfect scores that the simulation almost never produced.
Tora has a point. The logs can't tell whether a high score came from knowing the creator well, from a second try, or from luck. Still, the popular-answer yardstick makes your friend's score easier to read.
How to read your friend's score
Putting the numbers together, here's a rough guide for a 30-question quiz:
- Friend (15 to 21): 50.6% of real plays. The popular-answer bet averaged 20.1, inside this band, so a score in this band can't, on its own, show whether your friend knows you or just bet on the popular side.
- Best friend (22 to 26): 32.5% of real plays, and above the real-play average of 21.3. Betting on the popular side lands in this band in 29.6% of simulated quizzes, but mostly at the low end: it reached 24 or more in only 9%.
- Soulmate (27 to 30): 12.7% of real plays. Betting on the popular side got this far in only 0.4% of simulated quizzes, but one score still isn't proof of how well someone knows you.
- Acquaintance or Stranger (14 or fewer): 3.9% and 0.3% of real plays. A low score might also mean you picked a lot of less popular answers. Then betting on the popular side backfires, so even a close friend who leans on it can score low.
Questions are drawn at random for each quiz, so one quiz can simply end up with more questions where you're in the minority than another. For what else the results screen shows, see our guide to reading your results. To see which questions split evenly and which get nearly the same answer every time, see where answers split 50/50.
About this data
Popular-answer shares: calculated from the answers quiz creators chose for themselves, counted at the same time as the scores. Each answer counts, so these are numbers of answers, not people, and players' guesses aren't part of them. We don't know whether any player actually bet on the popular side; the simulation is only a yardstick for reading scores.