Designing a Controlled Experiment

Designing a controlled experiment means isolating one variable, predicting its effect, and comparing the result against a matched group that never received it.

Return to the wilting tomato plants. The hypothesis is that compacted soil blocks water and oxygen from the roots. The prediction: loosening the soil should stop the daily wilting within a week. The design needs two plants alike in every other respect — same variety, pot size, sun exposure, watering schedule, planting day. One plant, the experimental group, gets its soil loosened around the base. The other, the control group, is left compacted. Soil looseness is the one variable that differs between them. After a week, the gardener records whether each plant wilts at noon. If the loosened plant recovers and the compacted one still wilts, the result supports the hypothesis. If both recover, something else was responsible — perhaps the week merely turned cooler — and the control catches that, since a yard-wide change would have shown up in the untouched plant too. If neither recovers, a different explanation is needed.