Many readers judge a research paper by its figures and tables before they read a full paragraph. A clear chart or a well-built table can carry your main finding in seconds, while a crowded or confusing visual can make strong work look careless. Figures and tables are not decoration added at the end of writing; they are part of the argument itself, and they deserve the same care as your methods and discussion.
This guide walks through the whole process in order: planning visuals before you write, choosing the right chart or table for each result, designing for legibility, writing captions that stand alone, and checking every number against the text before submission. Follow these steps and your visuals will help reviewers, editors, and readers understand your work quickly and trust it.
Start with a simple list of your key findings, not with a blank chart. For each finding, write one sentence stating the message: what changed, by how much, and in which group. Then decide whether that message needs a figure, a table, or a sentence in the text. A result that is one or two numbers rarely needs a chart at all, while a pattern across groups or over time usually becomes clearer as a visual.
Sketch each planned visual on paper or in a rough draft before you open your charting software. Note the variables, the groups being compared, the units, and the scale. This short step prevents a common waste of time: building a polished figure only to discover it shows the wrong comparison or duplicates a table.
The right visual follows from the question your data answer. Use a bar chart to compare values across a few groups, a line chart to show change over time or across an ordered variable, and a scatter plot to show the relationship between two continuous variables. Use a table when readers need exact values, when many variables must be shown together, or when the data are mainly categorical and a chart would add little.
Resist fancy charts for simple data. Pie charts work poorly when slices are similar in size or numerous, and three-dimensional effects distort values. If a simpler chart tells the story honestly, it is the better choice. Reviewers judge a figure by what it communicates, not by how impressive it appears.
A good figure can be understood in about ten seconds. Keep only elements that carry meaning: clear axis labels with units, well-labelled groups, and a legend only when direct labelling is impossible. Remove gridlines, shading, and background effects that do not help the reader. Every mark on the figure should earn its place, and white space is not wasted space.
Make text large enough to read at the size the figure will actually be published, not only on your large computer screen. Axis labels, tick labels, and annotations should stay legible when the figure is reduced to a single column width. Check this by shrinking the figure to publication size and reading it without zooming. If you cannot read it comfortably, increase the font size, shorten the labels, or split the figure into two simpler panels.
Every figure and table needs a title that states what it shows, including the population, the variables, and the comparison where relevant. A title such as "Results" tells the reader nothing, while "Mean milk yield by feeding group over twelve weeks" tells the reader what to look for before they study the detail. Keep titles specific, factual, and free of claims the visual does not support.
The caption should let the reader understand the visual without hunting through the main text. State what is plotted, define every symbol, abbreviation, and unit, and explain the sample size and any statistical detail shown, such as error bars. A caption that repeats the whole discussion is too long; one that leaves symbols unexplained is too short.
A good table gives readers exact values they can check, cite, and reuse. Organise it so the eye moves naturally, with the most important comparison easy to find. Align numbers by decimal point, keep the same number of decimal places within a column, and use consistent units throughout. A table that mixes units or decimal places forces readers to do mental arithmetic they should not need to do.
Use footnotes beneath the table for everything that needs explaining: abbreviations, definitions, sample sizes, and the meaning of any symbols or markers. Avoid cramming the table with extra rows that belong in supplementary material, and avoid repeating in the table every number already stated in the text. The text should highlight the key pattern and direct the reader to the table; the table should carry the full detail.
Nothing damages trust faster than numbers that disagree with each other. A mean stated in the text, shown in a table, and plotted in a figure must be the same number, with the same units and the same rounding. Check sample sizes the same way: if the text says forty participants completed the study, the table and figure must not quietly show thirty-eight or forty-two without explanation. Reviewers check these details, and inconsistencies suggest the analysis itself may be unreliable.
Build one checking pass into your final revision that looks only at numbers. Work from your analysis output, not from memory, and tick off each value in the abstract, the results text, every table, and every figure. Pay special attention to percentages and to values copied by hand between software and manuscript. One careful pass here prevents corrections, delays, and embarrassment after publication.
Colour should support meaning, never carry it alone. Many readers have some form of colour vision deficiency, and many figures are still printed or photocopied in greyscale, so a chart that relies only on red versus green will fail for a substantial audience. Combine colour with a second cue, such as different line styles, markers, or direct labels, so the figure survives both colour blindness and black-and-white printing.
Choose a restrained palette with strong contrast, and test the figure in greyscale before submission. Avoid light colours on white, avoid red and green as the only distinction between groups, and keep the same colour meaning across all figures.
Before you export anything, read the journal's author guidelines for figures and tables. Journals specify file formats, minimum resolution, colour mode, and maximum figure sizes, and a figure that ignores these requirements may be returned before review. Vector formats are usually preferred for charts and diagrams because they stay sharp at any size, while photographs and microscopy images need high-resolution raster files. Tables are normally submitted as editable text, not as images, so they can be typeset correctly.
Export each figure at publication size, name files clearly by figure number, and keep the original editable version so corrections are quick. Check every exported file by opening it fresh: confirm fonts rendered correctly, no labels were cut off, and resolution is genuinely high.
If a figure or table comes from another publication, you usually need permission to reuse it, even when you redraw it with only minor changes. Check the licence of the original work: material published under an open licence such as CC BY 4.0 can generally be reused with proper credit, while material under standard copyright normally requires written permission from the rights holder. Allow plenty of time for permission requests, and keep the written permission with your manuscript records.
Credit reused material clearly in the caption, naming the original source and stating that it is reproduced with permission or under its licence, exactly as the permission or licence requires. The same care applies to images from the internet, stock libraries, and colleagues: if you did not create it, say where it came from. Proper credit protects you, respects the original author, and is a basic requirement of publication ethics.
Several mistakes appear again and again in submitted manuscripts. Axes that do not start where the reader expects, or that are stretched to exaggerate a small difference, mislead even when the data are correct. Missing units, undefined abbreviations, and legends that do not match the symbols in the figure force readers to guess. Tables packed with every variable measured, including ones never discussed, bury the findings that matter. Each of these is easy to fix before submission and hard to excuse after a reviewer points it out.
Watch out for two further traps. First, never duplicate the same result in a figure, a table, and a long paragraph of text; choose the clearest form and let the text highlight only the key point. Second, be wary of any service that advertises "guaranteed" publication or a guaranteed outcome for your paper — that is a red flag, not a promise you should rely on, and legitimate journals do not sell acceptance. Honest, well-prepared visuals and a fitting journal choice are what actually support a submission.
Imagine a small study comparing milk yield across three feeding groups. The first bad version is a pie chart of the three group means: the slices look similar, the exact values are invisible, and the comparison the study is about cannot be judged. The second bad version is a bar chart with a truncated axis that makes a modest difference look dramatic, plus a rainbow of colours with no legend. The third bad version is a table listing every measurement for every animal, which buries the group comparison under pages of raw data.
The good version is a simple bar chart of group means with clearly labelled axes, units, and sample sizes, error bars defined in the caption, and a companion table giving the exact means and variability for readers who need the numbers. The text then states the key comparison in one sentence and points to the figure and table. Same result, same data — but only the final version lets a reader grasp the finding at a glance and verify it in detail.
Run this checklist on the final manuscript, not on a draft, and ask a co-author to check it independently. Every figure and table is numbered in the order it is first mentioned, every one is cited in the text, and every title, label, and caption is complete and accurate. Every number matches across text, tables, figures, and abstract, and every symbol, abbreviation, and unit is defined where it appears. Figures are legible at publication size and in greyscale, and files meet the journal's format and resolution requirements.
Finally, confirm permissions and credits for reused material, that tables are editable text rather than images, and that no visual duplicates another one. Save the original files with the submitted version so corrections during review are quick.
| Data or question | Best visual | Why it works |
|---|---|---|
| Comparing a few groups | Bar chart | Heights are easy to compare at a glance |
| Change over time | Line chart | Trends and turning points stand out clearly |
| Relationship between two variables | Scatter plot | Each point shows one observation directly |
| Exact values readers must check | Table | Precise numbers, many variables together |
| Distribution of one variable | Histogram or box plot | Shape, spread, and outliers are visible |
These authoritative external resources explain the standards and tools mentioned in this guide:
Should every result appear in both a figure and a table?
No. Choose the form that communicates that result best and use the text to highlight the key point. Duplicating the same numbers in a figure, a table, and a long paragraph wastes space and invites inconsistencies between versions.
How many figures and tables should a paper have?
As many as the argument needs and no more. Let each visual carry one clear message, and move supporting detail to supplementary material when the journal allows it rather than crowding the main paper.
What resolution should my figure files be?
Follow the target journal's author guidelines, which state the required format and minimum resolution. As a general rule, use vector files for charts and diagrams and high-resolution files for photographs, and always check the exported file at publication size.
Do I need permission to reuse a figure from another paper?
Usually yes, unless the original is published under an open licence such as CC BY 4.0 that permits reuse with credit. Check the licence, request written permission where needed, and credit the source in the caption.
Why do my figures look fine on screen but poor in print?
Screen views are larger and backlit, so small text and pale colours that look acceptable on screen become unreadable in print. Shrink each figure to its publication size and view it in greyscale before submission to catch these problems early.
What should I check last before submitting my figures and tables?
Check that every figure and table is numbered, cited in the text, and captioned; that every number matches the text and abstract; that permissions are in place for reused material; and that all files meet the journal's format requirements.