To test how soil type affects plant growth, all other variables must be controlled so any difference you observe can be attributed to soil alone. Keep constant: cup size, amount of soil, seed type, water volume, light exposure, temperature, fertilizer, and measurement schedule. If these vary, results become confounded—you won’t know whether poor growth came from the soil or from less light/water. Controlling variables improves reliability (repeatable outcomes), validity (the experiment measures what it intends to measure), and allows you to make a causal conclusion: “This soil type leads to better growth.”