Three Fertilizer Investigations, Three Broken Designs
Three fourth-grade groups are testing the claim "Fertilizer X makes bean plants grow best." Group 1 puts three cups with 5 g of fertilizer on the windowsill and three cups with none in a dark closet. Group 2 puts all six cups on the windowsill; the three with fertilizer get 50 mL of water daily and the three without get 100 mL. Group 3 uses pots of different sizes, measures only the tallest plant in each pot after four weeks, and plans to report that height equals "best." All three groups will present a conclusion Friday, and Group 3 has already watered for two weeks and does not want to start over.
1. For each group, identify the specific flaw, state whether any comparison is still valid, and name the variable that should have been controlled.
Read the worked answer
Group 1: light is confounded with fertilizer. Any difference could be caused by light, not fertilizer. No valid comparison. Light should be held constant.
Group 2: water amount is confounded with fertilizer - and in the direction that favors the control, so the fertilizer group is handicapped. No valid comparison. Water should be held constant; only fertilizer varies.
Group 3: three problems at once - different pot sizes (root space confounds), measuring only the tallest plant (ignores the other plants and rewards a single lucky seedling), and defining "best" as height alone (a tall, spindly, pale plant is not thriving). No valid comparison of fertilizer, though the pots do produce data about pot size if reported as such.
In every case the independent variable is fertilizer and everything else - light, water, soil, pot size, seed type, planting depth - must be identical.
2. Write a single corrected procedure the class could run, including an operational definition of "best" and a measurement plan.
Read the worked answer
Question: Does adding Fertilizer X increase bean plant growth?
Procedure: 12 identical pots, same soil and seed type, 3 seeds each, all on the same windowsill, all watered 50 mL every other day. Six pots get 5 g of Fertilizer X; six get none. Assignment to the two sets is random.
Operational definition: "Best" will not be a single number. Measure at day 28: (a) height in cm from soil to highest leaf, (b) number of leaves, (c) a leaf-color rating against a 1–4 greenness chart, (d) total plant mass after harvest. Report means for each group.
Measurement plan: every plant measured, not just the tallest; same ruler, same observer pair, same time of day; record in a table; graph group means with the range so students see spread. Conclusion is only about this fertilizer, at this dose, with these plants.
3. Group 3 has two weeks of data and won't restart. How do you salvage the investigation without pretending the design was sound?
Read the worked answer
Turn the flaw into the lesson. Have Group 3 report what their data can support and what it cannot.
Can support: comparisons within a single pot size - e.g., pot 4's fertilized plant versus pot 4's unfertilized plant, if pot size is the only difference left. That is a smaller but real fair test.
Cannot support: any claim about fertilizer across the whole set, because pot size and fertilizer move together.
Activities:
- Have the group graph height versus pot size and see whether pot size alone tracks the results - students often discover the confound themselves.
- Ask them to write a "limitations" paragraph: what a skeptical reader would object to.
- Have them redesign on paper and run the corrected version with fast-germinating seeds (radish) so they still get hands-on data before the unit ends.
Grading should reward the accurate limitation statement at least as much as a tidy result.