13-03 · Final project preparation

Scope and time

Use 1–2 hours total to revise the report, prepare slides, rehearse and check rendering. Choose durations within that cap: report and figure 30–45 minutes, slides 20–35, rehearsal 10–20, final check 10–20. No new model-fitting assignment or additional reading is required.

Preparation slides Rehearsal and actions handout · Peer-review checklist · Recap

Suggested report

A 3–5-page narrative, supported by reproducible source and required inputs: biological question, documented design, one main outcome and comparison, justified distribution used to describe the outcome and model structure, a plausible alternative and reasoned revisions from your saved plan, main figure and estimate with uncertainty, a diagnostic decision, one sensitivity comparison, limitations, and a proposed future experiment with prospective simulation under two assumptions. Cite background papers; clearly identify synthetic teaching data. Keep project scope from the launch.

Suggested talk

Eight minutes followed by five minutes of questions. Use approximately one minute for question/context; 1½ for design and analysis; two for the main figure and practical interpretation; one for diagnostic/sensitivity; 1½ for follow-up design and simulation; one for conclusion and limits. Share speaking sensibly. Every student should explain the allocation, model rationale, a plausible alternative and main estimate, and answer a directed individual question.

Final checks

Render from the report folder using required relative-path inputs. Check which group is subtracted from which, outcome units, interval target, time and population against the main figure and prose. Preserve raw data, record software and unresolved assumptions, and state whether computation was actually rerun. Prepare answers about the number of animals or groups assigned to each treatment or selected for observation, meaningful effects and what would change your conclusion.

Presentations are on Wednesday 2 December, 12:30–16:00. Formal assessment arrangements will be shared separately.

For future power, use simglm; the worked pen-trial script shows the direct calls. Adapt the treatment assignments, outcome, assumptions and fitted analysis to your proposed study. Report failed fits and uncertainty caused by using a limited number of simulated experiments. Use lme4 for mixed models, performance for checks, and marginaleffects for predictions and useful comparisons.