Beans, Windowsills, and a Confounded Experiment
Sixth graders designed an investigation: does fertilizer affect bean plant height? Team A put four cups of planted beans on the classroom windowsill and watered them every other day with 50 mL of fertilizer solution. Team B put four cups inside a supply cabinet and watered them every other day with 50 mL of tap water. After three weeks, Team A's heights were 18.4, 17.9, 19.1, and 16.2 cm. Team B's were 9.2, 8.8, 10.1, and 9.6 cm. One cup in Team A never sprouted and was left out of the mean. On Day 10 a Team A student measured in inches and converted by multiplying by 2, then switched back to centimeters; nobody recorded which plants were measured which way. The team's poster claims fertilizer triples plant growth. During the gallery walk, a visiting class asks how the team knows the fertilizer caused the difference.
1. Identify every flaw that undermines the causal claim, and state the direction in which each flaw pushes the result.
Read the worked answer
Confounded location. Team A was on a windowsill and Team B in a dark cabinet. Light, temperature, and airflow all differ, and any of them could produce the gap. This flaw alone makes the causal claim unsupportable.
Confounded watering content is not the problem; the absence of a location control is. Fertilizer solution versus tap water is the intended treatment. The design needed both groups in the same light, with only the liquid differing.
Excluding the non-sprouting cup biases upward. A cup that never sprouted would have contributed a very low or zero height. Dropping it raises Team A's mean, and it was dropped from the group that already looks better. Germination should be reported separately or the cup counted as zero under a pre-stated rule.
The unit conversion is wrong and inconsistent. Multiplying inches by 2 instead of 2.54 understates those Day 10 values by about 21%, and mixing converted and direct centimeter measurements corrupts any growth calculation.
'Triples' is arithmetically wrong regardless. 17.9 divided by 9.4 is about 1.9, roughly double. The poster overstates its own data.
2. Redesign the investigation so that a causal claim would be justified, using the same materials and a three-week window.
Read the worked answer
One location. All cups on the same shelf, away from a draft, with positions shuffled every few days so no cup sits permanently at the sunniest edge. Light and temperature become constants rather than variables.
One variable. Fertilizer solution versus plain water, same volume, same schedule, same water source, same soil, same seed lot, same cup type. Everything else identical.
More replicates and a stated rule. Eight cups per group rather than four, because four is too few to see past one odd seed. Decide before planting how non-sprouting cups are handled: count them as zero height and report germination rate for each group separately.
One measurement protocol. Measure in centimeters from the soil line to the highest leaf tip, with the same ruler, on the same days, recorded by two students who agree. Ban mid-study unit changes; if inches are used, convert by 2.54 once at the end.
Pre-stated prediction and analysis. Write down what result would support the claim and what would refute it before planting. With the redesign, a consistent difference across eight cups per group in the same light supports a causal claim about fertilizer; the original design supports none.
3. The visiting class asks how the team knows fertilizer caused the difference. Write the response the teacher should coach, and name the misconception the students would be practicing if they defended the original claim.
Read the worked answer
Coached response. 'We cannot tell from this experiment. Our fertilized plants were also the only ones in the sunlight, so the difference could come from light or from fertilizer. Both changed at once, so we cannot separate them. We are repeating it with all the plants in the same light so that the only difference is the fertilizer.'
That response names the confound, states why it blocks the inference, and describes the fix. It also credits the observation without endorsing the conclusion, which is the scientific posture the standard is asking for.
Misconception being practiced if they defend the original claim. Post hoc reasoning: two things happened together or in sequence, so one caused the other. The specific form here is treating a single comparison as if all other variables were held constant when they were not. Left uncorrected, students carry forward the habit of accepting any difference between two groups as evidence of a cause.
Follow-up question to pose. If the repeated experiment shows the same gap with all cups in the same light, what would still be needed before saying fertilizer causes taller plants in other classrooms or other soils? That opens replication and generalization without abandoning the claim.