Seminar 3 – Figures and tables

Figures and tables

PhD Seminar · Chemical Engineering

A figure that supports a precise claim

14 October 2026 12:00–13:00
Soorathep Kheawhom

The scientific question determines the display

Temperature dependence: a response plot

Independent variation: raw points with a summary

Prediction quality: parity and residual plots

Exact conditions or values: a table

The same means can hide the evidence

Intentionally incomplete teaching example. Synthetic data.

Independent batches make variation visible

Synthetic conversion data at 350 °C. Points: independent batches. Summary: mean ± sample SD.

SD, SE and confidence intervals answer different questions

Synthetic data. 95% t intervals assume independent, approximately normal observations. n = 3 per group.

Krzywinski and Altman: Error bars

A comparison needs matched conditions

Synthetic data. Mean ± SD, n = 3 independent batches per catalyst at each temperature. Lines guide the eye.

Conversion and yield support different claims

Metric at 350 °C	Baseline	Modified
Conversion (%)	64	75
Selectivity to P (%)	89	87
Mean yield to P (%)	57.0	65.2

Synthetic paired run data. Yield = conversion × selectivity / 100 for the stated 1:1 molar basis.

A high correlation can coexist with biased predictions

Synthetic model predictions. Residual = predicted − observed conversion. Dashed lines show equality and zero.

A caption that defines the evidence

Conversion at 350 °C for baseline and modified catalysts under matched feed and flow conditions.

Points show three independent catalyst batches per condition. Large markers show means. Error bars show ±1 SD.

The comparison covers one feed composition. The experiment does not establish a reaction mechanism.

Synthetic teaching data. Report the remaining operating conditions in the final caption or linked methods.

A caption should support independent reading

What system and comparison does the figure show?

Which conditions, units and normalization apply?

What is n, and what counts as an independent replicate?

What do points, lines and uncertainty marks mean?

Workshop: redesign one figure

10minutes

Pairs, 10 minutes. Use Worksheet 2A.

Choose a scientific claim and a suitable display.

Sketch the axes, units, raw observations and uncertainty.

Write a caption and identify one unsupported interpretation.

Tables preserve exact values and conditions

Put units in the headers.

Use precision that the measurement supports.

Distinguish missing, not measured and not applicable.

Keep each row’s basis of comparison explicit.

Figure integrity and reproducibility

Keep raw data unchanged and document exclusions.

Generate the figure from a saved script.

Use consistent scales for panels meant for comparison.

Retain source files, processing steps and permissions.

Peer review of the revised figure

6minutes

Can you identify the claim within 20 seconds?

Can you reconstruct the comparison from the caption?

Does the display show variation and limitations?

What would you ask the author to change?

Technical repeats do not create new batches

Design A: one catalyst batch, measured nine times.

Design B: three independently prepared batches, each measured three times.

For between-batch variation, Design B has n = 3 independent batches.

A readable export is part of the evidence

Check the figure at its intended publication size.

Use the lab palette and redundant markers or labels.

Export vector files or raster files at ≥600 dpi.

Check embedded PDF fonts and exclude Type 3 fonts.

Exit ticket & anonymous feedback

3 minutes

One concept check

One evidence-based reflection

One point that is still unclear

QR code for session 2 exit ticket and anonymous feedback

seminar3-cheme-feedback.soorathep-k.chatgpt.site/session/2

Exit ticket: student ID optional. Feedback: no name or student ID.

Reading and policy sources

Krzywinski and Altman: Error bars

I&EC Research: Author Guidelines

Krzywinski and Altman: Error bars I&EC Research: Author Guidelines