BPS3082 Chap.2 Scientific Figures and Evidence
Scientific Figures and Evidence
Scientific communication in this unit begins before writing. The research question determines the protocol, the data structure determines the plot, and the plot supports a bounded claim.
This chapter shows how to choose an honest visual encoding, label units and uncertainty, write a standalone caption, and build a results paragraph that describes direction, magnitude, analysis and scope without turning interpretation into observation.
A time-course question should retain time on the horizontal axis, an association should preserve paired observations in a scatter plot, and a group comparison should reveal the independent values rather than concealing them behind a decorative summary. Axis limits, transformations and colours influence perception, so each choice must make the relevant pattern easier to inspect without exaggerating it.
Captions work as miniature methods: they identify conditions, units, sample size, error-bar meaning and analysis. Results prose then guides the reader from direction to magnitude, uncertainty and the supported conclusion. It does not need to repeat every table cell. The final claim stays inside the measured attribute and laboratory conditions.
Mechanism and broader product implications belong in a clearly signposted interpretation, where alternative explanations and limitations can be considered rather than silently presented as observations.
What this chapter covers
- 01
Question-to-figure alignment
- 02
Axes, units and visual scale
- 03
Independent observations
- 04
Standalone captions
- 05
Results paragraph sequence
- 06
Claim boundaries
- 07
Protocol details that anchor interpretation
- 08
Axes units and honest visual scale
- 09
Independent sample size in captions
- 10
Measured patterns before mechanistic explanation
Worked example · free
Write a bounded dissolution result
- +2State the direction and measured time point without calling one product universally superior.
- +2Report that adjusted comparisons support A versus B and A versus C.
- +2State that the evidence does not support B versus C under the tested conditions.
Key terms
- Visual Encoding
- The graphical use of position, length, colour or shape to represent data.
- Figure Caption
- Standalone text defining what is plotted, how uncertainty is shown and how groups are compared.
- Error Bar
- A graphical interval whose meaning must be labelled, such as SD, SEM or a confidence interval.
- Results Claim
- A statement of the observed pattern limited to the design and measurements.
- Scientific Scope
- The population, conditions and outcome to which evidence can reasonably apply.
Scientific Figures and Evidence FAQ
What information belongs in a figure caption?
Identify the outcome, groups, units, independent sample size, visual symbols, error-bar meaning and analysis where comparison symbols appear. A reader should not need the surrounding paragraph to decode the main evidence. Define any transformation or normalisation that changes the plotted scale.
Should I repeat every table value in the results text?
No. Select the values needed to establish direction, magnitude and the critical comparison. The table stores detail, while the prose explains how that detail answers the scientific question. Repetition should be reserved for quantities essential to the argument.
When does a scientific claim become too broad?
It becomes too broad when it extends beyond the measured outcome, tested conditions or sampled units. An in-vitro product test does not by itself establish clinical benefit, and a correlation does not by itself establish mechanism. Narrow the noun and population until the sentence matches the protocol.
Assessment move
Audit figures without reading the body text. If the comparison, units, sample size or uncertainty cannot be recovered, repair the caption. Then audit each results paragraph without the figure and ask whether its claims are broader than the protocol. Practise a four-part writing sequence: pattern, magnitude and uncertainty, analysis, bounded conclusion. Remove vague adjectives and replace them with the measured attribute.
Keep mechanistic explanation in the discussion unless a task explicitly asks for interpretation beside the result. Rebuild one figure in three possible forms and explain why one encoding best preserves the experiment. Check colour-independent distinctions and read labels at the final page size. Create a caption checklist that includes independent sample size and the definition of error bars.
During editing, highlight every adjective and replace unsupported value judgements with a measured direction or quantity. Read the paragraph aloud to find passive constructions that hide the actor or procedure. Finish by writing the strongest claim the design supports and a tempting stronger claim it does not support; the contrast trains scope control for reports and short answers.
Keep an evidence-chain ledger with four columns: question, protocol feature, display choice and permitted claim. Break one link deliberately, such as removing the independent sample count or changing the plotted summary, and explain how the caption or conclusion must change. Practise spotting axes that begin above zero, unequal intervals and decorative encodings that exaggerate differences.
For a final review, verify that symbols are defined, panels are referenced in order, uncertainty is named consistently and every numerical sentence can be traced to a displayed value or analysis.
