Fertilizer Trial with Confounded Variables
Two botany students at a college in Cuttack test the hypothesis 'A new organic fertilizer increases tomato plant height.' They take 20 seedlings of the same variety. Ten are potted in soil mixed with 5 g of fertilizer and placed on a sunny windowsill; ten receive no fertilizer and are placed on a shaded rack. All pots get 200 mL of water daily. After 28 days, the fertilized plants average 18 cm and the unfertilized average 13 cm. One fertilized pot develops wilt and is excluded from the average. The students conclude that the fertilizer caused a 5 cm increase and recommend it to local farmers. Their teacher points out that the design does not isolate the fertilizer's effect and asks them to revise the hypothesis, the prediction, and the experimental design before presenting at the district science fair, which requires a controlled comparison and an operational definition of growth.
1. Identify the confounded variables in the experiment and explain how each could produce the observed 5 cm difference.
Read the worked answer
The main confound is light. The fertilized pots are on a sunny windowsill and the unfertilized pots are on a shaded rack. Light is a well-known influence on plant height, so the 5 cm difference could be due to light alone, or to light and fertilizer together, rather than to fertilizer alone.
A second confound is position: windowsill versus rack may differ in temperature, airflow and exposure, not only in light. A third problem is the post hoc exclusion of one fertilized pot that developed wilt. If wilt stunts growth, removing that pot after the fact raises the fertilized group's average and biases the comparison. There is also no randomization: the seedlings were not randomly assigned to the two treatments, so initial differences between seedlings could affect the result. Finally, 10 pots per group is a small sample, so one or two unusual plants can shift the mean noticeably.
2. Rewrite the hypothesis as a testable causal hypothesis and state a corresponding prediction with an operational definition of 'growth'.
Read the worked answer
A revised causal hypothesis could be:
'Applying 5 g of the new organic fertilizer to tomato seedlings at potting, while holding light, water, soil, pot size and temperature constant, increases mean plant height after 28 days compared with seedlings that receive no fertilizer.'
A corresponding prediction is:
'After 28 days, the mean height of the fertilized group will be greater than the mean height of the control group by at least 2 cm.'
The phrase 'growth' must be operationalized. For example: height in centimetres from the soil surface to the highest point of the main stem, measured with a ruler at 9 a.m. on day 28, with two readings per plant averaged by the same observer. This makes the outcome measurable and repeatable. The hypothesis is causal because it claims the fertilizer produces the difference; the prediction is the observable result that would support it.
3. Describe a revised design that would isolate the fertilizer effect, and state one result that would count against the hypothesis.
Read the worked answer
A stronger design would use at least 20 seedlings, randomly assign 10 to the fertilizer treatment and 10 to the control, and keep all pots on the same bench under the same light and temperature. Pots should be rotated periodically to reduce position effects. Both groups should receive the same soil, pot size and 200 mL of water daily; the only intended difference should be the 5 g of fertilizer. Height should be measured blind - the measurer should not know which pot received fertilizer - and no pot should be excluded after the results are seen unless a pre-stated rule applies. The trial should be repeated, preferably over more than one batch.
A result that would count against the hypothesis is: the fertilized group's mean height is not greater than the control group's mean height, or the difference is no larger than the measurement error. If the control grows as well as or better than the fertilized group under identical conditions, the claim that the fertilizer causes increased height is not supported. A wilt outbreak confined to the fertilized group would also raise a serious doubt rather than being quietly excluded.