The quiz score distribution describes the room. The podium describes three people.
The trainer photographs the podium, reports an average of 61 percent, and the sponsor nods. Nobody in that exchange knows whether 61 means thirty people who each got about six in ten, or fifteen people at 90 and fifteen at 30. The quiz score distribution is the chart that answers it. In RiLiFi's Session History, the overview for every live quiz shows Total Participants, Average Score, Total Questions, and Average Completion Time. It also shows a score distribution bar chart with five buckets: 0–20%, 21–40%, 41–60%, 61–80%, and 81–100%. Five bars. That chart tells a trainer what to do next. The podium tells them who won a mug.
We built the archive this way because we kept seeing the same thing at the end of training days: a leaderboard screenshot in the wrap-up email and no idea whether the material landed. The winner is the least informative person in the room. The shape of the bars is the information.
Five shapes and what each one asks you to do
| Shape | What it looks like | Likely cause | Next move |
|---|---|---|---|
| Right-heavy | Most people in 81–100 | The material landed, or the quiz was easy | Move on; raise difficulty next time |
| Fat middle | A bulge in 41–60 | People know the vocabulary, not the procedure | Reteach the procedure, not the terms |
| Two humps | One cluster high, one low, little in between | Two audiences sat in one session | Split the follow-up by cohort |
| Left-heavy | Most people in 0–40 | Missing prerequisite, or a broken question | Check the items first, then the prerequisite |
| Flat | Roughly equal bars | Guessing, or options that do not discriminate | Rewrite the questions |
Two of those shapes get misread constantly, so they get their own sections.
The fat middle, in detail
A bulge in the 41–60 bucket is the most common shape we see after a half-day product or compliance session. It is also the most misunderstood. Trainers read it as "roughly half the room got it." It usually means something more specific: everyone recognised the terms and almost nobody could execute the sequence. Recognition questions got answered. Procedure questions got guessed.
The fix is in the question design, not the slides. Single-answer questions reward elimination, so a person who knows the vocabulary can clear the easy items and coin-flip the hard ones straight into the middle bucket. Multiple-correct questions, where the answer is a set of required actions, pull the half-knowers down and the real knowers up. The bulge splits into something you can act on. We covered the mechanics in multiple correct answers stop lucky guessing.
Two humps means two audiences
A bimodal result is a gap, not an average. A 2022 study of a university biochemistry course describes the shape. The top quartile is skilful in the subject matter while the bottom quartile is not retaining the material. The authors treated it as a missed opportunity rather than a curve to grade on (Changing Bimodal Grade Distributions: A Missed Opportunity?, International Journal of Higher Education, 2022). In a corporate room the two humps are usually two cohorts: the team that already uses the tool and the team that was told to attend. Averaging them produces a number that describes nobody. Export the leaderboard, sort by score, and you will usually find the cohort line in the names.
Use the distribution in the room, not only after it
Physics instructors have run a version of this for more than thirty years with clickers. In Crouch and Mazur's decade of Peer Instruction, the share of students answering correctly decides what happens next (Peer Instruction: Ten Years of Experience and Results, American Journal of Physics, 2001). Well below a third correct means re-teach the concept. Roughly a third to seventy percent means have students argue it out with a neighbour before a second vote. Above seventy percent means explain briefly and move on. Beatty and colleagues make a related point about the tabulated answer counts from a classroom response system (Designing Effective Questions for Classroom Response System Teaching). The counts are themselves a teaching tool, because the wrong options people pick reveal the misconception behind them.
A live quiz gives a trainer the same lever. Session History keeps who selected which option on every question. After the session, a trainer can see not just that the room missed question seven but which wrong answer it chose. During the session the host controls the pace. When a question splits the room, with groans on one side and nods on the other at the reveal, stop and discuss it before advancing. Our note on timer pacing covers how to leave room for that pause.
Read the shape against completion time
The overview includes Average Completion Time, and it changes how you read the middle bucket.
- Fast and middle: people answered quickly and landed around 50 percent. That is guessing with confidence. Check the options; they are probably too similar or too easy to eliminate down to two.
- Slow and middle: people took the full timer and still landed around 50 percent. That is thinking without a method. The procedure was not taught in a way that can be recalled under time pressure.
- Fast and right-heavy: the quiz was too easy for this room. Save the questions for the onboarding cohort.
- Slow and right-heavy: the material landed and it was hard. That is the shape you want, and it is worth telling the room so.
Then run a poll, not another quiz
A fat middle or a split room is a question for the audience, not a verdict on it. The fastest follow-up is a live poll in the same workspace: "Which step of the procedure is least clear?" with the steps as options. One RiLiFi workspace covers the live quiz, the live poll and Q&A, and the spinning wheel. The trainer opens the poll from the same folder as the quiz instead of stacking a second vendor for a two-minute pulse check. Anonymous mode, with "Ask name" unchecked, gets a more honest answer than a show of hands. Our live poll page shows the live bar chart that comes back.
Questions trainers ask about score distributions
What does a bimodal quiz score distribution mean?
Two populations took the same quiz. Find the cohort line in the exported leaderboard and plan two follow-ups, not one.
Why is the average score misleading?
Because it collapses the shape. An average of 61 can be a tight cluster or two distant humps, and those need opposite responses.
What should we do when most scores sit in the middle?
Look at the question types and the completion time first. Then reteach the procedure, not the vocabulary. Run the same quiz again after the follow-up to see whether the bulge moved.
Report the shape, not the winner
Next time the wrap-up email wants a screenshot, send the five bars instead of the podium. The live quiz archive keeps the overview and the per-question detail for every session, and the CSV export carries the leaderboard into whatever system has to hold the record. The podium can go on the slide. The distribution goes in the plan.