Module 1 Worksheet: Statistical Questions and Data Claims
Scenario
A colleague has already written the conclusion they want and is looking for a table that supports it. Your job is to work in the other direction.
Data
Dataset: claim-review-scenarios.csv
| scenario_id | vague_claim | likely_claim_type | safer_question | boundary_note |
|---|---|---|---|---|
| C01 | Students learn better with short lessons | inferential | In this sample do students using short lessons have higher checkpoint scores than students using long lessons? | Can compare groups but cannot prove lesson length caused the difference. |
| C02 | The new button improves signups | causal | In a randomized A/B test did the new button produce a higher signup rate than the old button? | Needs random assignment and practical-size review. |
| C03 | People who study more always score higher | predictive | In this dataset how strongly are study hours associated with score? | Association is not a guarantee for every learner. |
| C04 | The support team was worse in March | descriptive | Did the March support ticket rate per active account increase compared with February? | Counts alone are not enough without active-account denominators. |
| C05 | The training program works for everyone | inferential | Among learners who completed the training what average score change was observed? | Completion bias may make the result too optimistic. |
Your Task
Rewrite three vague claims into statistical questions and add one boundary note for each claim.
Questions
- What claim is being made, and what question would actually test it?
- Who or what is one row, and is that the thing the claim is about?
- Which column would have to change for the claim to be wrong?
- What comparison is missing that would make this interpretable?
- What does the table support, stated as narrowly as you can?
- What part of the original claim is not supported at all?
- What would you need to collect before making the stronger claim?
Write Your Conclusion
Write the claim twice: the version the table supports, and the version you were asked to support. Two to four sentences each.
Self-Check
- Did I write the question before looking for the answer?
- Is my supported claim narrower than the original?
- Did I name what is missing rather than working around it?
- Would the colleague recognise their claim in my version?