The Greenhouse Trial That Would Not Reproduce
A student team tests whether a new liquid fertilizer increases bean growth. They plant 30 bean seeds in 30 identical pots using the same potting soil. Fifteen pots receive 50 mL of fertilizer solution every two days; the other fifteen receive 50 mL of plain water. All pots sit on the same bench in a greenhouse. After three weeks, the fertilized plants average 18.4 cm and the watered plants average 17.9 cm. The team concludes the fertilizer has no effect. A classmate notices that the fertilized pots were placed on the sun-facing side of the bench, where afternoon temperatures reach 29 °C, while the watered pots sit in shade at 24 °C. She also points out that the fertilizer pots were watered only on days when someone was present, and two weekends were missed entirely.
1. Identify the most serious flaw in the experimental design and explain how it undermines the conclusion.
Read the worked answer
The most serious flaw is a confounding variable: pot position is tied to treatment. All fertilized pots are on the hot, sunny side and all control pots are in shade, so light and temperature differ systematically between groups. Any growth difference could be caused by warmth or light rather than fertilizer, and any absence of difference could hide a real fertilizer effect offset by heat stress. Because the two variables move together, the experiment cannot separate them. Randomizing pot positions (or rotating pots regularly) would break the link between treatment and environment.
2. The missed watering weekends affected only the fertilizer group. Explain how this affects interpretation of the 0.5 cm difference and what the team should do next.
Read the worked answer
The fertilizer group was stressed by drought that the control group did not experience, so the 0.5 cm difference is not a clean measure of fertilizer effect. The fertilizer may have helped growth while the missed water suppressed it, meaning the true effect could be larger than observed. The difference is also small relative to typical plant-to-plant variation, so it may not be meaningful. Next, the team should repeat the trial with randomized positions, identical watering schedules for both groups, and more plants per group. They should report means with a measure of variability (such as standard deviation) and only claim an effect if the difference exceeds normal variation.
3. The classmate claims the results 'prove the fertilizer does nothing.' Evaluate this claim using the idea of experimental error.
Read the worked answer
The claim is too strong. A single trial with confounded placement and unequal watering cannot establish that the fertilizer has no effect; it can only fail to show an effect under flawed conditions. Absence of evidence is not evidence of absence. To support a no-effect claim, the team needs a well-controlled, repeated experiment with sufficient sample size and a variability measure. Until then, the honest conclusion is that the current data are inconclusive.