
improve one failure point in our key, retest it and use data to compare version 1 with version 2.
Phase 4: Build
Do Now. A team changes three steps of its key at once. The key works better. Do they know which change helped?
What is the problem with changing many steps at once?
Junior chemists, today you revise. Pick the failure point from your critique notes. Change that one step. Then retest version 2 the same way you tested version 1.
Test in different situations. A key might work for the person who wrote it but fail for a new reader. Test it with a teammate, then with someone from another team. Test it with a fresh sample, too.
Worked example 1: Malik’s fix. Version 1 used the magnifier to split sugar from salt. 3 of 6 testers got it right. Version 2 used the bulb tester instead. 6 of 6 testers got it right. Malik’s data showed version 2 met the “identifies all 5” criterion much better. These are sample numbers.
Worked example 2: Sofia’s trade-off. Sofia’s version 2 added a second check for flour: a sticky dough test. Accuracy went up, but the full test now took 12 minutes. That breaks our 10-minute constraint. Her team kept the dough test but cut a slow step somewhere else. Then they timed it again: 9 minutes.
| Check | Version 1 | Version 2 |
|---|---|---|
| Correct IDs (out of 5 known powders) | ||
| Testers who finished without help | ||
| Time for a full test (minutes) | ||
| Supplies still fit in a shoebox? | ||
| The one step we changed |
Revise, retest and record. Use these checks along the way.
Make the case for your revision.
Use your data to compare version 1 and version 2. Which one better meets the criteria and constraints? Give numbers.
Write your test plan for Test Day: one test for each criterion or constraint, what you will measure, and what you will keep the same.
Your key is getting stronger. Next, we meet real people who test unknown powders for a living.
Exit ticket: what one change made the biggest difference to your key? How do you know?
Materials: kits and key cards, known powder samples, fresh index cards for version 2, timers, the optional bulb or LED tester.
Setup: 10 minutes to revise, 15 to retest with at least two testers, 5 to compare. Pair teams that had similar failure points so they can share fixes.
Listen for: teams rewriting the whole key; teams declaring version 2 “better” without numbers.
Support / stretch: pre-mark the one step to revise for teams that need it. Stretch: run a third version and graph accuracy across all three.