Road Salt and Bean Germination
Five identical trays test the effect of road salt on bean germination. Each tray holds 20 bean seeds in the same potting soil, receives 15 mL of solution every second day, and sits on the same shelf under a grow light. The solutions are 0, 2, 5, 10 and 20 g/L.
On day 2, the 10 g/L tray is moved to a windowsill to free space under the light. On day 4, mould appears in the 5 g/L tray and the teacher removes the 5 affected seeds. On day 7 the class counts germinated seeds:
- 0 g/L: 16 of 20 (80%)
- 2 g/L: 16 of 20 (80%)
- 5 g/L: 9 of the 15 remaining seeds
- 10 g/L: 4 of 20 (20%)
A group writes: « Salt clearly harms germination. The 10 g/L tray proves that the more salt there is, the worse the plants grow. »
1. State the strongest conclusion these data support, and one conclusion they do not support.
Read the worked answer
Supported: In this setup, germination was about 80% at 0 and 2 g/L and clearly lower at 5 g/L and above. The data suggest a threshold between 2 and 5 g/L rather than a steady decline, because 0 and 2 g/L are indistinguishable at this sample size.
Not supported: That « the more salt there is, the worse the plants grow. » Two problems:
- The variable measured is germination, not growth. Nothing here describes plant height, leaf number, or biomass after sprouting.
- There is no smooth trend to extrapolate. The data jump from 80% to roughly 60% to 20%; with n = 20, each seed is worth 5 percentage points, so the "trend" rests on very few data points and is not a demonstrated dose-response curve.
A defensible phrasing: « Concentrations of 5 g/L and above reduced germination compared with 0–2 g/L in this classroom setup. »
2. Identify two design flaws and explain what each one does to the interpretation of the results.
Read the worked answer
Flaw 1 - the 10 g/L tray was moved to a windowsill. Light intensity and temperature now differ from the other four trays, so the 20% result cannot be attributed to salt alone; low light or a temperature swing is an equally good explanation. This is a confounding variable, and it undermines the single most dramatic data point.
Flaw 2 - five mouldy seeds were removed from the 5 g/L tray.
- The sample is no longer equal (n = 15 vs n = 20), so the percentages are computed on different bases.
- The removal was not random: mould may track moisture or the treatment itself, so the remaining seeds are not a fair sample of the original 20.
- The 60% figure is therefore uncertain in a way the other figures are not.
Fixes: keep all trays under identical light and rotate positions systematically; run at least three replicate trays per concentration so one accident does not destroy a condition; if contamination occurs, discard and report the whole tray rather than individual seeds.
3. The group predicts that 20 g/L will give « about 0% germination. » Evaluate this prediction.
Read the worked answer
It is plausible but over-stated.
In its favour: the observed direction is consistent - 80%, 80%, ~60%, 20% - so a further drop at a higher concentration is a reasonable expectation.
Against it:
- No concentration between 10 and 20 g/L was tested, so the shape of the curve there is unknown; the drop could level off rather than continue to zero.
- The 10 g/L value is itself compromised by the windowsill move, so the group is extrapolating from an unreliable anchor.
- « About 0% » is a precision the design cannot deliver: with 20 seeds, one seed is 5 percentage points, and 0% is a claim of complete inhibition.
Better prediction: « Germination at 20 g/L will be lower than at 10 g/L, possibly near zero. » It is directional, testable, and honest about uncertainty - which is what a prediction should be.