Check your work: Classify fictional causal claims by design and alternatives

Staged checking material

Answer key

1. C1: proposed cause = umbrella presence/count; outcome = path wetness on the recorded day; design = observation with cause and outcome recorded together. Wet-path rates are 30/30 versus 0/30, a 100-percentage-point association. Compatible alternatives are rain causing both umbrella use and wetness, and rain-linked drainage or weather conditions affecting wetness independently of umbrellas. A discriminating benign test would compare equally rainy, otherwise matched path sections with and without umbrella drip exposure. Classification: umbrella causation unsupported.
2. C2: cause = random assignment to blue rather than clear filter; outcome = mean seedling height after four weeks; design = randomised intervention with cause preceding outcome. The difference is 18.4 - 16.1 = 2.3 cm. Compatible alternatives are random sampling variation and position or measurement effects within the one greenhouse. Repeat randomised blocks across greenhouse positions with blinded height measurement to discriminate them. Classification: better supported for these pots, not universal.
3. C3: cause = self-selected quiet-room use; outcome = pages completed in the same hour; design = observational self-selection with room choice preceding the count. The association is 42 - 30 = 12 pages. Compatible alternatives include book difficulty, prior reading habit and reason for choosing the room. Randomly assign willing readers to rooms while holding text difficulty constant, or predeclare and adjust measured selection factors. Classification: suggestive association with causality unresolved.
4. C4: cause = random assignment of the reminder card; outcome = completion within the fixed window; design = randomised fictional intervention. Completion is 40/50 = 80% versus 30/50 = 60%, a 20-percentage-point difference. Compatible alternatives are random variation and an unrecorded implementation or device difference. Repeat the randomisation with allocation checks and identical delivery logs to test them. Classification: better supported within this fixture, not broad human effectiveness.

Scoring rule

For each case award 1 point for operational variables, time order, correct design, two plausible alternatives, a discriminating test, mechanism status, confidence update and bounded classification: 8 per case, 32 total. Mechanism without measured pathway evidence cannot be called shown. Any move from a fictional row to real blame, diagnosis or policy earns STOP.

Authority boundary: A governed step-by-step lesson may be completed independently inside its stated limits.