Teacher note: these are low-risk, print-first activities. Use classroom materials only; do not collect student health, identity, or personal device data.
Mystery Box: model what you cannot see
Your job is not to guess the hidden object. Your job is to make a model that explains the observations and survives a fair test.
Record three direct observations. Avoid “it is probably…”
Sketch Model A. Label the feature that explains your strongest observation.
Write one test that could distinguish Model A from a different model.
What would count as evidence against your model? Name the result before you test.
Paper Helicopter: one variable, five drops
Build two helicopters. Change exactly one feature. Drop from the same height and release without a push. Time from release to floor in seconds.
| Question | Your answer |
|---|---|
| Testable question | |
| Independent variable: what changes? | |
| Dependent variable: what is measured? | |
| Controlled variables: what stays the same? | |
| Prediction with because |
| Design | Trial 1 | Trial 2 | Trial 3 | Trial 4 | Trial 5 |
|---|---|---|---|---|---|
| Design A · baseline | |||||
| Design B · changed one factor |
For each design, calculate the mean, median, and range. Which design has the larger typical time?
Which design is more consistent? Use range or mean absolute deviation (MAD) as evidence.
Write a cautious claim. Include one limitation or alternative explanation.
Graph Repair Shop + Claim Clinic
A graph is part of an argument. Repair the representation before you trust the conclusion.
A bar chart starts its y-axis at 14 cm, making 15.2 cm look much taller than 14.8 cm. Name the trick and the repair.
A line graph says “average speed” but has no axis labels or units. What question must the reader ask?
A pie chart shows 40%, 35%, 35%, and 20%. What is impossible? What should be fixed?
Audit this claim: “4 out of 5 students prefer our study app.” Ask about sample size, comparison, wording, and who benefits.
Sample Detective: how much can one sample tell?
Use a cup of colored tiles or beads as the population. Draw without looking, record, replace, and repeat. Compare small samples with pooled samples.
| Sample | Total drawn | Target color | Relative frequency | Representative? Why? |
|---|---|---|---|---|
| Group 1 | ||||
| Group 2 | ||||
| Pooled class |
Why can two small samples disagree without either group making a mistake?
What changes when the sample grows? What does not become guaranteed?
Mini Investigation Conference
Propose a safe, bounded investigation. Your partner is a reviewer, not a judge of whether your result is “right.”
| Prompt | Your plan |
|---|---|
| Question | |
| Prediction | |
| Independent + dependent variables | |
| Control / comparison | |
| Trials, measurement, and safety limit |
Claim, evidence, reasoning: make the argument after the data are collected.
Peer review: What is one strength? What is one question about evidence or design? What will be revised?
Answer key + teacher moves
Teacher moves
Ask students to name the measurement before they name the answer. Praise a clear limitation. When a model or claim is wrong, ask which observation or data point forced the revision.
Paper Helicopter
Mean is the sum divided by 5. Median is the middle value after sorting. Range is maximum minus minimum. MAD is the average distance from the mean. A higher mean time means longer average time aloft, but it does not by itself prove the design caused the difference; release consistency, air movement, and small sample size remain possible explanations.
Graph Repair + Claim Clinic
Repair a truncated axis by showing a truthful scale; label both axes and units; a single pie must total 100%; and a claim needs its sample size, comparison, wording, and possible sponsor or selection bias.
Sample Detective
Small samples naturally vary. Larger pooled samples often give a steadier estimate, but no sample automatically becomes representative or proves a cause. Students should use “in this sample” language when the design does not support broader inference.
Conference rubric
Look for: measurable question; one changed variable; clear control/comparison; repeated, recorded measurements; graph or summary that matches the data; CER reasoning; honest uncertainty; and a concrete revision after peer review.