Project overview
Work in your existing group of three to analyse a supplied dataset and use what you learn to plan a new experiment. Choose a focused biological question, justify your decisions and explain what the evidence supports.
1. Choose a focused question
Choose one of our four supplied cases: calf growth, broiler feed efficiency, pig diarrhoea or horse shelter use. Your instructor-provided Posit Cloud project contains the data, study context, data dictionary and starter code to import and inspect the observations. See group projects and teaching datasets for the cases.
You do not need to collect or find data. Another existing dataset is possible after an instructor check of its design, documentation and scope. Keep a supplied case available until an alternative is confirmed.
Focus on one main outcome and one main comparison or relationship. Read enough about the biological context and design to explain why your question matters and whether the data can answer it.
2. Inspect the data and submit an analysis plan
You can inspect the data, make exploratory plots and check the variables before submitting your plan. Submit the plan before fitting the model for your project analysis in step 3, then receive instructor feedback.
Explain your question and what you want to estimate, including its units, comparison, time period and the animals or groups it concerns. Describe how treatments were assigned, or how animals or groups were selected for observation. Identify repeated measurements and animals sharing a group. For example, when diets are assigned to whole pens, the pens provide separate treatment assignments; the birds within each pen do not.
Propose a model, explain its assumptions and identify checks that could lead you to revise it. Describe one reasonable alternative analysis. Use the analysis-plan.qmd in your case folder or the equivalent Class 10 worksheet; submit one initial plan.
3. Analyse the data and explain the result
Fit your chosen model, check its assumptions and compare the result with your alternative analysis. This sensitivity comparison asks whether a reasonable change in the analysis changes your answer.
Present one clear main figure and estimate your main comparison or relationship with uncertainty. Explain its biological meaning and the limits imposed by the study design. Record and justify changes to your plan.
4. Design a follow-up experiment and calculate power
Propose a feasible new experiment: describe the treatments, how animals or groups would be assigned to them, and what you would measure. Choose a biologically meaningful effect to detect, supported by reasoning or relevant evidence. Do not simply copy the effect estimated in step 3.
Use simulation to produce power curves under at least two justified sets of assumptions. Power is the probability that your planned analysis detects the specified effect when that effect is present. For each proposed study size, simulate repeated experiments and apply the planned analysis.
- Vary sample size across a range that shows how power changes. Count separately allocated animals or groups: for a pen-level experiment, vary the number of pens per treatment.
- Compare two plausible assumption sets, such as lower and higher variability. Normally keep the meaningful effect fixed so the curves show how uncertainty about other assumptions affects the design.
- Recommend a feasible sample size from the curves. Explain how the recommendation depends on your assumptions and practical constraints.
Represent the proposed allocation, outcome and planned analysis in your simulation. Explain simplifications, failed model fits and uncertainty from using a limited number of simulated experiments. Use class examples and instructor support to develop the simulation.
5. Write and present the project
Produce a short reproducible report, with R code that allows your work to be checked and repeated. Explain your question, design, analysis, conclusions and proposed experiment. Every member should contribute to the presentation, understand the whole project and be ready to explain the group’s choices.
What to produce
- A one-page scoping brief, developed during calendar week 42 and revised after feedback.
- An analysis plan submitted before fitting, with later revisions recorded.
- A 3–5-page reproducible report, including the main figure, with supporting R code.
- A follow-up experiment proposal, power curves and sample-size recommendation.
- An 8-minute presentation followed by 5 minutes of questions.
Milestones and support
Confirm your question and dataset by 4 November. Submit your analysis plan on 11 November (teaching Unit 10) before fitting the project model. Later workshops support analysis, checks, interpretation, power analysis, peer review and rehearsal. The detailed project brief sets out the milestones.
We provide teaching examples, feedback and technical help. Your group is responsible for understanding and explaining its choices. Acknowledge sources, code assistance and any previously accessed worked solutions.
The supplied datasets are synthetic teaching data. Label them accordingly in your report and figures; your results are not empirical findings about the animals or source papers. For observational data, distinguish an association from the effect of assigning a treatment.
These are the educational project expectations. Official assessment requirements, final submission deadlines and the presentation date remain to be confirmed. See the examination page for confirmed assessment information when available.