Did you know? A plant can finish a classroom experiment taller than its neighbours even though it grew less during the experiment. If we look only at the final photograph, we can tell a perfectly convincing story that the recorded measurements do not support. The missing information may be the height of each plant on day zero.
When seedlings start at different heights, compare each plant’s change from its own recorded starting height before claiming which group grew more. For a specified interval, change in height = final measured height − initial measured height. This tells us how much each plant’s measured height changed during that interval. It does not by itself prove what caused the difference.
This guide uses fictional height readings, hand-checked group averages, a quick diagnosis, guided experiments and a careful conclusion. It is intended for students, parents and teachers learning how to interpret an investigation, not as a report of any actual school class or a promise of examination performance.
Why a final-height photograph is an incomplete answer
Consider a lesson asking whether a different watering routine appears to be associated with faster seedling growth. A taller final plant could simply have started taller. A shorter final plant could have grown more during the week. Both statements can be true without contradiction. The starting measurement is therefore not an administrative detail: it changes what the data can answer.
Real plant investigations require care with the measurement definition. The Huntington’s Access to Light experiment instructs learners to take initial baseline measurements. Science Buddies’ measurement guidance also makes the measurement procedure explicit. Here we use one consistently defined main-stem height, measured in centimetres from a fixed reference position to a clearly identified point on the plant.
Do not assume that total height is the only measure of healthy growth. Some plants branch, grow leaves or change biomass without producing the largest vertical increase. A useful investigation chooses its outcome in advance and does not quietly switch from plant health to height because one chart looks impressive.
A worked classroom example with six invented seedlings
Imagine six fictional seedlings of the same stated species measured on day 0 and again on day 7. Three are in group A and three in group B, which received different imagined watering routines. The planted groups were not randomly balanced by starting height. The numbers below are deliberately chosen to make the interpretation problem visible; they are not real experimental outcomes.
- Group A, seedling A1: day 0 = 12 cm, day 7 = 18 cm, so gain = 18 − 12 = 6 cm.
- Group A, A2: 14 cm → 18 cm, gain = 4 cm.
- Group A, A3: 13 cm → 18 cm, gain = 5 cm.
- Group B, B1: 8 cm → 15 cm, gain = 7 cm.
- Group B, B2: 9 cm → 17 cm, gain = 8 cm.
- Group B, B3: 10 cm → 16 cm, gain = 6 cm.
First compare the final heights
Every plant in group A finishes at 18 cm, so its final-height mean is (18 + 18 + 18) ÷ 3 = 18 cm. Group B ends at 15, 17 and 16 cm, so its mean is (15 + 17 + 16) ÷ 3 = 16 cm. A chart of final heights alone would make group A look taller by 2 cm.
Now compare what actually changed
Group A’s mean gain is (6 + 4 + 5) ÷ 3 = 5 cm over seven days. Group B’s mean gain is (7 + 8 + 6) ÷ 3 = 7 cm over seven days. Thus group B had an average measured gain that was 2 cm greater during the same seven-day period, despite finishing 2 cm shorter.
Check the arithmetic a second way. The average initial height of group A is (12 + 14 + 13) ÷ 3 = 13 cm; 13 + 5 = 18 cm, its average final height. Group B’s average initial height is (8 + 9 + 10) ÷ 3 = 9 cm; 9 + 7 = 16 cm. The means reconcile exactly because these are the same three plants measured at both times with no missing observations.
What would an honest conclusion actually say?
A careful pupil can write: “In this fictional seven-day comparison, group A finished taller on average (18 cm versus 16 cm), but group B had the larger average increase in measured height (7 cm versus 5 cm). The groups began at different average heights, so the final-height comparison alone is misleading. The observations do not establish that the watering routine caused the difference.”
That conclusion separates three different ideas: the final state, the change during observation, and the possible cause of that change. Treating them as one claim is the central learning error.
Does subtracting the starting height make the whole experiment fair?
No. A gain calculation fixes a particular descriptive problem, but it cannot remove every competing explanation. The seedlings may differ in age, condition, genetics, root establishment, light exposure or soil. Groups with different starting sizes may also grow differently even if they receive the same treatment. A comparison of gains is more informative than final heights alone, but it is not automatically a causal experiment.
A fairer planned investigation would choose similar specimens when possible, allocate them using an appropriate method, define the watering treatment, keep other relevant conditions reasonably consistent, and record several independently handled plants per condition. Different seedlings remain biologically variable. A single good-looking pair of plants is weak evidence of a general effect.
An average can conceal important variation too. In this fictional example, group A gains range from 4 to 6 cm and group B from 6 to 8 cm. Their ranges meet at 6 cm. Without more observations, measurement uncertainty and a stronger experimental design, it would be reckless to claim a precise universal improvement caused by one routine.
The diagnostic: can the learner distinguish height from growth?
Show pupils only the final-height means, 18 cm and 16 cm. Ask which group grew more. The correct diagnostic response is: “We cannot tell from final heights alone; we need the starting heights or another valid growth record.” Then reveal the day-zero table. A pupil who changes the answer after the new evidence is demonstrating scientific reasoning, not inconsistency.
- Result mistaken for change: “A is 18 cm, so A grew 18 cm.” The plant did not begin at zero centimetres.
- Group mistaken for specimen: treating one unusually strong plant as a whole group’s result.
- Units dropped: writing “7 faster” rather than “7 cm measured growth over seven days.”
- Time omitted: comparing a five-day gain against a seven-day gain as if the exposure periods matched.
- Cause overstated: claiming one watering routine was superior merely because the group average changed.
- Measurement shifted: measuring from the soil one day and the pot rim the next without a recorded correction.
Progressive practice: explain, calculate, then challenge the design
Guided comparison with two plants
A first seedling begins at 6 cm and ends at 11 cm. A second begins at 10 cm and ends at 13 cm. Both are observed over five days using the same reference point. The gains are 11 − 6 = 5 cm and 13 − 10 = 3 cm. The second finishes taller, while the first has a greater height increase during the interval. Have the learner say both facts in one sentence.
Paired comparison with a measurement problem
A pupil reports: “Pot C measured 12 cm on Monday and 17 cm the following Monday, so it grew 5 cm.” Ask what must be checked before accepting the difference. Partners should name a consistent stem point, measurement origin, same plant identification, suitable resolution, dated entries and whether the pot’s soil or stem position changed. The subtraction is correct under the readings, but measurement consistency still needs evaluation.
Independent comparison with uneven intervals
An invented plant gains 4 cm over four days while another gains 6 cm over eight days. If the question is about average change per day over those intervals, calculate 4 ÷ 4 = 1 cm/day and 6 ÷ 8 = 0.75 cm/day. But the average over unequal intervals still does not prove that either plant maintained a constant growth speed every day. A more rigorous comparison uses matched intervals and additional observations.
How teachers can help pupils move from a formula to independent judgement
First model one before–after subtraction. Next let pupils compute changes for matched plants while the teacher checks units and labels. Then withhold the initial readings from a final-height graph and ask what cannot be concluded. Finally present the complete data with a change in observation interval or a different starting-size pattern. Progress is the pupil asking for the missing baseline without being prompted.
Parents can ask, “Are those the same plants in both readings? Where did you measure from? What else changed?” A teacher can ask students to distinguish measured difference, reasonable interpretation and causal explanation in three separate sentences. A student can maintain an honest raw record with dates, units, original values and notes about damaged or missing specimens.
If one plant dies or one measurement is missing, never fill the blank with zero growth as though it were observed. Record the missingness and explain how any average was formed. Missing plants may not be random; dropping them silently can make the surviving group look healthier than the whole starting group.
Limitations of the classroom model
Height increase is not the same as plant mass gain, leaf area, survival or overall biological health. A percentage change, such as 5 ÷ 13 or 7 ÷ 9, answers a different question and can be especially sensitive to a small baseline. A tiny or non-representative group and an unmatched starting condition constrain inference. Calculating gains accurately cannot convert an observational comparison into a controlled experiment.
No particular watering quantity, fertiliser schedule or local school syllabus requirement is asserted here. The general inquiry principles apply only insofar as the variables, measurements and design match the real question. We should never infer an approved school experiment or learning outcome from our invented dataset.
Frequently asked questions
Should I draw a graph of final heights or height gains?
Draw whichever graph directly answers your stated question, label the axes and include the units. For a question about change during the observation week, gains are relevant; for a question about how tall plants ended, final heights are relevant. A pair of clearly labelled graphs can show why the answers differ.
What if every plant starts at the same height?
Starting heights then no longer explain the difference between those specific final heights, assuming consistent measurement. But other relevant factors and natural variation may still differ; starting equally is not an automatic certificate of a fair test.
Is the group with the larger mean gain definitely better?
Not without defining “better” and establishing that the groups are comparable. Here we know only a seven-day average difference in measured height. A causal judgement about the treatment requires stronger controls and appropriate evidence.
Can I repair an experiment after discovering unequal starting heights?
Keep the original records, report gain calculations as a clearer description, disclose the unequal baselines and limits, and propose a better-designed repeat. Never rewrite day-zero heights to make the groups match.
Next reading and a stronger habit of checking
For a complementary explanation about why repeated readings are not always independent experiments, see repeated measurements versus trials. For a different way a condition can shift unnoticed, read controlled-variable drift. The eduKateSG Education Hub and the immutable Clementi Mathematics teaching reference show how diagnosis, explanation, guided practice and independent checks support transferable learning.
The central achievement is simple and powerful: a pupil who sees the taller plant and asks “But how tall was it at the start?” is using evidence rather than appearance to think scientifically.