Wanted outcome: a criterion-specific conclusion without a universal care rule. Prepare: O4; twelve marks/counters optional.
0–3: compare 31 mm/6 alive with 36 mm/4 alive. 3–9 model: P increased 11 mm, Q 16 mm over eight days; P had 6/6 alive, Q 4/6. “More height” and “more alive” answer different questions. These supplied means use all six labelled height records, not survivors only. 9–21: choose a route. 21–27: a reviewer asks whether the recommendation changes with the criterion. 27–30 exit: state a choice for a criterion, or explain why more evidence is needed. If survival called “growth”, separate size and alive-count cells.
- Route 7A · Criterion decision. For fictional keeper Tall (prioritises recorded mean increase), choose Q conditionally; for keeper Alive (prioritises alive-at-check count), choose P conditionally. Cite two measures, period and a confound. Neither is a real care recommendation.
- Route 7B · Denominator repair. Repair “Q has four survivors, so it is four out of four”. Show original six and four alive; compare with P’s six out of six. Explain why a surviving-only denominator hides deaths and why a one-check rate is not lifetime survival.
- Route 7C · Cause investigation. A draft claims “50 mL killed two plants”. Write an alternative explanation question about temperature, initial individual differences or moisture/drainage. Design a repeat comparison with labelled individuals and relevant condition records. Explain that proposing another cause does not prove it either.