A pre and post training quiz is one quiz, hosted twice
The end-of-day quiz comes back at 81 percent and the room applauds. Then the sponsor asks what the same people would have scored at nine in the morning. Silence. A pre and post training quiz is the fix, and it is less work than it sounds: one quiz, built once, hosted live twice. The first session runs before the first slide. The second runs after the last one. The two Session History overviews, side by side, are the only honest answer to whether the training worked.
We keep the quiz in a shared folder so both sessions use the identical instrument. That matters more than it looks. The usual failure is a trainer who tidies up two questions between the morning and the afternoon, and then compares two different tests. Smart folders hold the one quiz file the whole team hosts from, instead of a copy emailed around and quietly edited.
Why one end-of-day quiz proves nothing
A single score has no baseline. Eighty-one percent could mean the room learned a great deal. It could also mean the room walked in already knowing most of the material. Kirkpatrick Partners define Level 2 of their model as the degree to which "the target audience acquires the intended knowledge, skills, attitude, confidence, and commitment" from the training (The Kirkpatrick Model). The standard way to show acquisition rather than prior knowledge is a measurement before and a measurement after.
There is a second reason to run the quiz twice, and it is about retention rather than measurement. Latimier and colleagues compared pre-testing and post-testing on the same material. Testing after exposure produced a much stronger retention effect seven days later. Post-testing also helped on questions that had never been tested (Does pre-testing promote better retention than post-testing?, npj Science of Learning, 2019). The post session is not just a ruler. It is part of the teaching.
The same instrument rule
Everything that could move the score has to stay fixed between the two sessions:
- Same questions, same wording, same order. Shuffling feels fair and quietly changes difficulty.
- Same answer type per question. If the skill is a set of required actions, use a multiple-correct question in both sessions. We explained why in multiple correct answers stop lucky guessing.
- Same timer per question. Timers run from 5 to 999 seconds and the host sets them per question, so set them once and leave them.
- Same metric. Compare accuracy and the average score percentage, not raw points. Points carry the speed bonus. Speed is a fine thing to reward in an energy round and a confusing thing to measure learning with.
- Same people. The delta only means something for a person who sat both sessions.
The procedure, step by step
- Build one quiz in the shared folder. Eight to twelve questions mapped to the session objectives. Questions have a 250-character limit and options a 100-character limit, which keeps the stems readable on a phone.
- Host the pre session before you teach. Lobby, names required, no explanations. Advance through every question and skip the commentary. Save the teaching for the teaching.
- Teach.
- Host the post session from the same folder item. Same timers, same order. This is the moment people want to skip, because the energy is high and the room is ready to leave. Do not skip it. It is the whole point.
- Open both overviews in Session History. Total Participants, Average Score, Total Questions, Average Completion Time, and the Score Distribution for each session.
- Export both leaderboards. The overview and the leaderboard both export to CSV. Match names across the two files and you have a per-person delta, not just a room average. Our guide to exporting the session, not the screenshot covers the handoff to an LMS.
A worked example with normalized gain
| Session | Participants | Average score | Fat part of the distribution |
|---|---|---|---|
| Pre (9:05) | 24 | 48% | 41–60% |
| Post (16:40) | 24 | 81% | 81–100% |
The raw gain is 33 points. Fine, but incomplete. The better number is the normalized gain that Hake introduced in his survey of six thousand physics students. It is the gain achieved divided by the gain that was available (Interactive-engagement versus traditional methods, American Journal of Physics, 1998). Here that is 33 divided by 52, which is 0.63. Hake's rule of thumb puts anything below 0.3 in the low band, 0.3 to 0.7 in the medium band, and above 0.7 in the high band.
Normalized gain matters because rooms start in different places. A group that walks in at 80 percent cannot gain 33 points. A raw delta would make the trainer look worse for teaching experienced people. A group that starts at 20 percent and ends at 53 has the same raw gain as our example and a normalized gain of 0.41. Same delta, different story.
What the two distributions say that the averages hide
Two averages can hide a split room. If the post session shows one hump in the 81–100 bucket and a second hump down in 21–40, the training landed for half the room and missed the other half. An average of 61 describes nobody. The score distribution chart in Session History shows that shape at a glance. Compare the pre and post shapes before you present the averages, and you will know whether to run a follow-up for the lower hump or move on.
Questions trainers ask us
Should the pre-test and post-test be the same questions?
Yes. The comparison only works on the same instrument. If you worry about people remembering the answers, remember two things. Latimier's results show post-testing is a learning event in its own right, and the pre session ran without explanations. Recalling a question is not the same as having been taught the answer.
How far apart should the two sessions be?
Same day measures acquisition. A week later measures retention, which is the number most sponsors want. If you can only run one post session, run it a week out and tell people in advance that it is coming.
Does the pre-test spoil the session?
Only if you walk through the answers after it. Advance through the questions, let people see their own result, and keep the explanations for the teaching block. The pre session should feel like a diagnostic, not a lesson.
Run it twice, then report the gain
One quiz. One shared folder. Two live sessions, one normalized gain. That is a Level 2 result a sponsor can repeat next quarter with the same instrument. Build the quiz in RiLiFi's live quiz, keep it in a shared folder so the afternoon host runs the identical file, and let the second average do the talking.