(d) Analysis, communication and evaluation
- Syllabus
- 2024
- Topic
- —
- Level
- —
Learning objectives
—Analyse data and draw evidence-based conclusionsUse scientific knowledge and understanding to analyse and interpret data and draw conclusions from experimental activities that are consistent with the evidence.—Communicate experimental findingsCommunicate findings from experimental activities using appropriate technical language, relevant calculations and graphs.—Assess experimental reliabilityAssess the reliability of an experimental activity.—Evaluate accuracy and validityEvaluate data and methods, taking into account factors that affect accuracy and validity.
Turn experimental data into a bounded conclusion
Analysing experimental data means identifying the supported pattern, quantifying it where possible, using relevant biological knowledge to interpret it, and limiting the conclusion to what the evidence actually shows.
| Reasoning move | What a defensible statement does |
|---|---|
| read the evidence | compare matched groups, axes, units, time points and sample sizes before selecting a pattern |
| describe the pattern | state direction, range, optimum, plateau or difference and support it with representative values |
| check variation | identify overlap, spread, anomalies or exceptions rather than hiding them |
| interpret biologically | connect the pattern to a relevant mechanism, keeping the causal direction explicit |
| conclude within limits | answer the aim, distinguish association from causation and avoid extending beyond the organisms, conditions or range tested |
If sperm motility falls from about 70% at zero drug to about 5% at 0.4 arbitrary units and remains similar at 0.8, the supported pattern is a steep decrease followed by a plateau. Concentration changes little across the same range, so the drug affects movement much more clearly than sperm number in these mice.
A conclusion is not a restatement of every number, and a biological explanation cannot replace description of the data. 'The treatment works' is too broad unless the measured outcome, comparison, variation and study boundary support that claim.
Communicate experimental findings so they can be checked
Experimental findings are communicated clearly when the calculation, table, graph and written conclusion all describe the same variables, units and comparison using precise biological language.
| Information | Appropriate communication |
|---|---|
| exact readings and repeats | a ruled table with informative headings, units in headings and consistent precision |
| continuous independent variable | a line or scatter graph with a sensible scale and points plotted accurately; add a best-fit line or curve only when justified |
| separate categories | a bar chart with separated bars and clearly named categories |
| calculated change | show the operation, values and unit or percentage so the result can be checked |
| conclusion | name the variables, direction or difference, relevant values and any important qualification |
percentagechange=((finalvalue−initialvalue)/initialvalue)×100
A graph title alone does not define the evidence: both axes need variables and units. Avoid vague phrases such as 'it went up a lot'; state what changed, by how much or across which range, and whether the pattern has an exception.
Judge whether experimental results are reliable
Reliability is the extent to which repeated measurements or independent repeats give consistent results. It is assessed from repetition, sample size and the spread of results—not from whether the result matches an expectation.
| Evidence about reliability | Interpretation |
|---|---|
| repeated readings are close together | low random variation; the result is more repeatable |
| repeated readings are widely spread | high random variation; a mean alone may hide instability |
| only one reading or a very small sample | consistency cannot be judged securely |
| similar result from more samples, people, times or investigators | stronger reproducibility and less dependence on one case |
| an anomalous value | investigate and repeat; exclude it only with a stated evidence-based reason |
Improve reliability by taking independent repeats at every condition, increasing a relevant sample size, using the same defined method and reporting the spread as well as a mean. A longer study can reveal whether an apparent effect persists.
Repeated measurements can agree closely and still all be wrong because of the same calibration error. That result is reliable but not accurate. Repeating a biased or invalid method does not remove the bias or make it test the intended question.
Evaluate accuracy, validity and the conclusion
Evaluation identifies the specific weakness, explains how it could affect the result or conclusion, and proposes a change that addresses that cause. Accuracy concerns closeness to the accepted or true value; validity concerns whether the method and evidence test the intended question.
| Evaluation target | What to inspect | Targeted improvement |
|---|---|---|
| measurement accuracy | calibration, zero error, instrument resolution, parallax and subjective endpoints | calibrate or zero apparatus, use a finer suitable instrument, read correctly, or standardise the endpoint with a colour reference or sensor |
| method validity | control of alternative causes, a fair comparison and whether the dependent measure represents the intended process | control the influential variable, add an appropriate reference group, or measure the process more directly |
| biological validity | sample size and diversity, model organism, laboratory versus natural conditions, and duration | use a larger representative sample, relevant organisms or conditions, and a sufficient observation period |
| conclusion validity | whether the claim exceeds the data range, measured outcome or tested population | narrow the claim and state the evidence boundary |
For every limitation, write a causal link: weakness → effect on measurement or comparison → consequence for the conclusion. For example, judging browning by eye gives a subjective endpoint, so recorded times may differ between observers; a standard colour chart or colorimeter makes the endpoint consistent and more accurate.
Generic statements such as 'use better equipment', 'repeat it' or 'human error' are not evaluations. Name the source of error and show why the proposed change reduces it. Repeats mainly address random variation; they do not automatically correct a systematic error or an invalid comparison.