Unit 2 — Design Surgery

Wednesday 9 September, 12:30–16:00

Diagnosing and repairing flawed experiments

Unit 1 asked, “What is the experimental unit?” This class broadens the question: can an experiment answer the question its investigators think it can?

Students will use a common diagnostic framework to identify problems involving experimental units, replication, pseudoreplication, controls, allocation, confounding, independence, and measurement.

NotePreparation details to be added

Readings and short videos on replication, pseudoreplication, controls, randomisation, and confounding will be added here.

The main activity is Design Surgery:

  1. Establish the intended question, treatment, response, and experimental unit.
  2. Diagnose the most important threats to a valid conclusion.
  3. Repair the design using the smallest defensible changes.
  4. Exchange the proposed repair for a second-opinion peer review.

Detailed cases, worksheets, and timings will be linked when those materials are ready.

Follow-up work will consolidate completely randomised single-factor designs and connect true replication to statistical power.