Lesson 44 of 54 in this sequence

81%

Not completed yet

Saved on this device only — not synced to an account yet.

Learning path

Step 44 of 54

Continue the sequence

ResumeReturn to this step ReviewRevisit the core idea RemediateGet repair practice Next stepChoose the next lesson manually

Completion

Review objectives and instruction
Complete the activity with manual learner confirmation
Answer every understanding-check item
Show completion feedback and reward only after the learner finishes the check

Progress

Step 44/54
15 content types
AI help source context available
Auto-advance blocked

Review objectives and instruction

AI help

Available only with source grounding, privacy limits, age controls, and measurement.

AI source context

Testing Designs by Changing Only One Thing · The Design Process · 3-5-ETS1-3 · fair test

Evidence: local lesson AI-help metadata only; source-grounding, privacy, age-safety, and measurement hooks do not prove real model quality, production telemetry, child AI approval, or AI tutor production readiness

Production AI readiness not claimed

AI source context

Assessment shape

Local lesson assessment shape is complete.

Local shape only; production answer-attempt telemetry, live ownership/RLS evidence, mastery interpretation, release, and production readiness are not claimed here.

Objectives: 4
Activity: Prompt ready
Quiz items: 3 items, 3 answer keys, 3 explanations
Reward: Completion reward ready

Assessment progress stays manual: do not auto-advance through uncertain answers, failed writes, consent, media, payment, learner choice, or AI-assisted checks.

Learning-loop evidence: Content-map contract only; persistence and production learner-progress writes require route/API evidence.

Manual policy: assessment, media, consent, payment, learner choice, failed writes, and AI-uncertain help stay under learner control.

Prerequisite: Complete or review Strong, Light, or Cheap: Every Material Trades Off; Use with parent or guardian controls for child learners

AI guardrails

Privacy

Do not send raw child free text without consent
Use source excerpts and non-identifying progress state only
Asset is not registry-gated

Age safety

Apply kids audience controls before AI-help claims
Keep child learner AI paths behind guardian/product review evidence before readiness claims

Measurement

Log AI help usage by surface and asset slug
Evaluate answer quality against source-grounded rubric before production-readiness claims
Tie remediation prompts to completion, retry, or review outcomes

Local AI-help guardrails only; they do not prove guardian approval, production telemetry, model evaluation, release, deployment, or AI tutor production readiness.

Source and rights

Source path: koydo-app/scripts/curriculum-authoring
License: ai-generated-koydo
Attribution: Attribution decision needs owner review
AI provenance: claude-opus-4-8

Source review gates: attribution_missing

Local source evidence only; this does not prove publication rights, attribution approval, AI provenance clearance, release, deployment, or production readiness.

Route review

Current route: /library/kids/engineering_making/en/lesson/fair-test-one-variable
Route source: Fallback route needs owner canonical review
Candidate route: /library/kids/engineering_making/en/lesson/fair-test-one-variable

Local route evidence only; fallback candidates need owner-approved canonical route policy before route writes, production discoverability, or readiness claims.

Content gates: canonical_target_web_missing_uses_library_fallback

lessonslessonslesson objectivesmedia scenesdrillspractice setsquizzesfeedbackcharacterscitationsconcept mapsprogressresumereviewremediationnext steps

Testing Designs by Changing Only One Thing

with Atlas

Atlas stands at a wooden workbench cluttered with paper bridges, small weights, and sticky notes, carefully lowering a penny onto one bridge while the other three identical bridges wait in a row behind him.

Explain why engineers change only one thing at a time when testing a design.
Identify which part of a test is being changed and which parts are being kept the same.
Compare two test results and explain what the difference tells you.
Predict what would happen to your results if you changed two things at once.

Key terms

fair test: a test that changes only one thing
variable: the one thing you change on purpose
controlled variable: a thing you keep the same
result: what happens at the end of a test

Change Just One Thing

A fair test means you change exactly one thing and keep everything else the same. The one thing you change is called the variable. If you change two things at once, you cannot tell which one mattered. Say one bridge is wider and also folded differently. If it holds more weight, was it the width or the folds? You will never know. Changing one thing keeps the test honest and clear.

Keep the Rest the Same

The parts you keep the same are called controlled variables. You control them, which means you hold them steady. If you test if thick rubber bands launch farther, you keep the same arm, the same angle, and the same ball. Only the rubber band changes. Now any change you see must come from the rubber band. That is how you know your answer is true and not a lucky guess.

Worked examples

Mia swaps a thicker rubber band but keeps everything else the same. Is it fair?

Check what she changed: just the rubber band.
Check what stayed the same: arm, angle, ball.
Only one thing changed.

Answer: Yes, it is a fair test.

Plane A is longer AND has wider wings than Plane B. What is wrong?

Two things are different at once.
You cannot tell which one helped.

Answer: It is not fair, because two things changed together.

Hey there, engineer! I'm Atlas, and today I want to show you one of the most important moves in all of engineering: the fair test. Imagine you built two paper bridges. One is wide, and the other is narrow — AND it is also folded differently. You test both and the narrow one holds more weight. Great, right? But wait — was it the width that helped, or the folding? You changed TWO things, so now you cannot tell which one actually mattered. That is the sneaky problem. A fair test means you change exactly ONE thing — called a variable — and keep everything else the same. The one part you are changing on purpose is what you are testing. Everything you keep the same is called a controlled variable. Because only one thing is different, any change you see in the results must have been caused by that one thing. Here is a hint: before you test, make a list of everything about your design — materials, size, shape, number of folds. Pick ONE to change. Put a circle around it. Put a check next to everything else to remind yourself to keep those the same. Now your test is fair, and your results will actually tell you something true.

Activity

Sort each test plan into Fair Test or Not a Fair Test, then explain your choice.

Practice

Decide if changing two ramp things at once is fair.

Plan a fair test for which paper folds the strongest.

Common mistakes to avoid

Change many things to save timeChanging many things hides which one caused the result.
Testing once makes it fairTesting once is fine, but you still change only one thing.

Check your understanding

Mia wants to find out if a thicker rubber band makes a catapult launch farther. She uses the same catapult arm, the same launch angle, and the same ball, but swaps in a thicker rubber band. What makes this a fair test?

Jaylen tested two paper planes. Plane A was longer AND had wider wings than Plane B. Plane A flew farther. What is the PROBLEM with Jaylen's test?

An engineer says, 'I will change the height of the ramp AND the weight of the car at the same time to save time.' What is most likely to go wrong?

Recap

A fair test changes only one thing, called the variable, and keeps everything else the same. The parts you hold steady are controlled variables. Then any change in the result must come from the one thing you changed.

Reflect

What one thing would you change in your next fair test?