# Group project: your own analysis plan

Group, case, date and version: ____________________

Complete this short plan in the Week 10 workshop **before fitting the case model**. Save the first version before instructor feedback; retain revisions with reasons. This sheet and the case folder’s `analysis-plan.qmd` serve the same checkpoint: submit one plan, not both. Follow-up simulation assumptions can remain preliminary at this stage. You may inspect the raw data and use syntax examples from other teaching datasets.

1. **Question and target:** name one outcome, comparison direction, scale, period and intended population. Explain its biological relevance.
2. **Known design:** distinguish allocation/sampling units from rows; draw grouping or repeats. Cite design metadata. Mark unknown facts without guessing.
3. **Outcome and family:** describe possible values, units, denominators/exposure and missing values. Propose a distribution or descriptive approach and explain why it could be appropriate.
4. **Model structure:** propose mean terms and how shared units/dependence should enter. State what the model estimates. A formula may help, but words and reasons are required.
5. **Checks and assumptions:** name two checks that could change your trust or require a revision; explain the consequences of failure.
6. **Alternative:** describe one plausible alternative addressing a specific scientific/design assumption. State how its target or interpretation may differ.
7. **Next action:** identify essential information gaps, fitting steps and responsibility for each.
8. **Assistance and exposure:** record sources, advice, code adaptations and whether complete case solutions or reference figures were previously accessed. Distinguish your reasoning from supplied choices.

Evaluation focuses on a defensible link between the question, known design, model and interpretation. More than one model can be appropriate when assumptions and consequences are explained. A narrow interval or a favourable result does not establish that a model is appropriate.

If computing fails, retain this plan and a labelled descriptive display and ask for help with execution. Acknowledge code assistance and any previously accessed worked solutions; distinguish supplied choices from your own reasoning and be ready to explain them individually.