A figure in a research paper is any visual element that isn’t a table. That includes graphs, charts, photographs, diagrams, maps, drawings, flowcharts, and infographics. Figures serve a specific purpose: they communicate patterns, relationships, and trends that would be difficult or tedious to describe in words alone. A well-designed figure can convey in seconds what might take an entire paragraph to explain.
What Counts as a Figure
The definition is broader than most people expect. Under APA style, which many disciplines follow, all visual displays other than tables are classified as figures. That means a simple bar graph and a complex microscope image both fall under the same umbrella. Common types include line graphs, bar graphs, pie charts, scatterplots, flowcharts, photographs, maps, plots, and infographics.
Each type serves a different communication goal. Line graphs show trends over time. Bar graphs compare magnitudes between groups. Pie charts display proportions. Scatterplots reveal relationships between two variables. Flowcharts map out processes or decision trees. Photographs and micrographs provide direct visual evidence, like a tissue sample or a geological formation. The choice depends entirely on what the data needs to say.
Figures vs. Tables vs. Text
One of the first decisions a researcher makes is whether information belongs in a figure, a table, or plain text. The general rule: if you’re presenting fewer than about six numbers, a sentence works fine. Tables are better when exact values matter and readers need to scan or compare specific data points across categories. Figures win when the goal is showing a pattern, trend, or relationship that readers wouldn’t spot in a column of numbers.
Tables present precise numerical values but don’t reveal trends easily. Figures sacrifice exact numbers for visual impact. A table might show that sales increased from 12,000 to 15,000 to 22,000 over three years. A line graph of the same data instantly shows the acceleration. Both are valid presentations of the same information, but they communicate differently. The right choice depends on whether the reader needs the exact values or the overall story the data tells.
Parts of a Figure
A complete figure in a research paper has several standard components. In APA format, the figure number appears first, in bold (e.g., Figure 1), numbered in the order figures are mentioned in the text. One line below that sits the title, written in italics with title case capitalization. The title should be brief but descriptive enough that a reader skimming the paper understands what they’re looking at.
The image itself should be as self-explanatory as possible. Ideally, a reader shouldn’t need to hunt through the caption to interpret what’s on screen. Axis labels, data labels, color-coded legends, and clear annotations all reduce the mental effort required. Research on figure presentation has found that even solid data becomes frustrating to analyze without proper labels, while simple additions like text annotations can dramatically improve comprehension.
Below the image, notes provide any additional context the reader needs. These might define abbreviations, explain symbols, provide copyright attribution, or clarify statistical notations. Not every figure needs notes, only those where the image and title alone leave gaps.
When Figures Appear in a Paper
Figures typically show up in the results and discussion sections, though methods sections sometimes include diagrams of experimental setups or flowcharts showing study design. Every figure must be referenced in the text. You won’t find a figure floating in a paper without a sentence somewhere pointing to it and explaining its relevance. The text should highlight what the reader should take away from the figure, not simply repeat everything the figure already shows.
Most journals limit the number of figures a paper can include, often between six and ten for a standard article. Supplementary figures can go in an appendix or online supplement for data that supports the findings but isn’t essential to the main argument.
Technical Requirements for Submission
Journals have strict technical standards for figure quality. Elsevier, one of the largest academic publishers, requires 300 DPI (dots per inch) for photographs, 500 DPI for images that combine photos with text or line elements, and 1,000 DPI for pure line art like graphs and diagrams. These requirements exist because figures that look fine on a computer screen can appear blurry or pixelated in print.
Text within figures should be at least 7 points when printed at final size, with subscripts and superscripts no smaller than 6 points. Common accepted file formats include TIFF and EPS, though requirements vary by journal. Getting these details wrong is one of the most common reasons manuscripts get sent back for revisions before peer review even begins.
Designing Figures for Accessibility
About 8% of men have some form of color vision impairment, and a study of 580 papers in biological sciences found that roughly half contained figures that were completely or partially inaccessible to readers with red-green colorblindness. This is a widespread problem with a straightforward fix.
The most common form of colorblindness makes it difficult to distinguish red from green. Replacing red-green color schemes with blue-orange or blue-red combinations immediately improves accessibility. Rainbow color maps, while visually appealing, are particularly problematic and should be replaced with perceptually uniform palettes like those in the Viridis color scheme. Free tools like Color Oracle simulate how figures appear to colorblind viewers, letting you check your work before submission.
The most robust approach is to avoid relying on color alone. Varying line styles (solid, dashed, dotted), using different marker shapes (circles, squares, triangles), or adding direct text labels ensures that information comes through even in grayscale. This also helps when papers are printed in black and white, which still happens more often than you might think.
Ethics of Figure Preparation
Manipulating figures is one of the most scrutinized areas of research integrity. The basic principle: any adjustment you make should not change what the data shows. Adjusting brightness or contrast uniformly across an entire image is generally acceptable. Selectively altering parts of an image, splicing elements from different experiments, or reusing the same image to represent different conditions crosses into misconduct.
The Committee on Publication Ethics has documented cases where researchers reused gel images across multiple experiments because the originals “weren’t very clear.” Even when the manipulation doesn’t change the statistical conclusions, journals treat this as a serious breach. If original data like gel images or photographs are used only to illustrate a general point rather than present specific experimental results, the paper needs to explicitly state that. Transparency is the dividing line between acceptable image processing and manipulation.