Aspect ratio

Aspect ratio #

When you make a plot in software, it will always default to some aspect ratio. In Excel, it is a rectangle of some kind, in Matlab, it is a square.

But, these default aspect ratios are an arbitrary choice. The odds are slim that it is the best one for your plot. You should always choose your aspect ratio intentionally, not just accept the default, because the aspect ratio of your plot is a powerful storytelling tool. Here’s a simple workflow for how to choose.

Option #1: Let Your Data Pick the Shape #

The best way to choose an aspect ratio is to let your data’s trend determine it.

Research shows that the human eye is best at perceiving slopes and changes when the line on a graph runs at 45 degrees. This concept is often called “banking to 45 degrees.”

You can choose your aspect ratio to “bank” your key trend to this 45-degree angle. Consider the $\ce{[CO2]}$ data from Mauna Loa. This data has two trends: a long-term rise and a yearly oscillation.

  • To emphasize the long-term rise, we’d make the plot nearly square to bank that data trend to 45°.

  • To emphasize the yearly oscillation, we’d make the plot very wide to bank the more rapid changes to 45°.

Aspect ratio can also be used to make a change feel more or less dramatic. A very narrow plot can make the same rise in $\ce{[CO2]}$ look terrifying.

The point is that the aspect ratio is a tool. Use it to make the trends underlying your story—whether it’s the long-term trend or the seasonal change—as clear as possible.

Option #2: When in Doubt, Use one of many Standard Rectangles #

What if your data doesn’t have one clear trend you want to emphasize? For instance, in science, many spectra don’t have specific trends, but multiple features that are easy to see across many aspect ratios. In such cases, the best choice is to use a “standard” non-square ratio that people are comfortable with.

  • The Golden Ratio (~1.61:1): Used in art and architecture for millennia. A classic, proven choice.

  • Paper Standards (e.g., US Letter at ~1.29:1 or A4 at ~1.41:1): We are conditioned to find these shapes “normal” from a lifetime of reading.

  • Video Standards (e.g., 4:3 or 16:9): These are also deeply familiar

When I publish, I often match the aspect ratio of the medium I am publishing in. So, in American journals, I will match US letter size and in European journals, I will match A4, unless the data dictates a different aspect ratio (see above and below).

The Exception: When to Use a Square #

Above, I stated that one should default to some sort of rectangle, as opposed to a square. The psychological justification for this is given below, but here, I want to discuss that there is a powerful use case for square aspect ratios: when you are comparing two variables that share the same scale.

A “correlation plot”—like comparing homework scores to exam scores that both run from 0 to 100—is the perfect example. A square plot with equal axes makes the 45-degree line of perfect agreement intuitive.

However, unless your data demands this kind of direct 1:1 comparison, you should avoid a square.

The Psychology: Why Humans Prefer Rectangles #

The advice to avoid squares isn’t just a design “rule”; it’s grounded in human psychology and biology.

The simple reason: our binocular field of vision is a wide horizontal rectangle, not a square.

It’s no surprise, then, that the media we choose to look at is almost always rectangular:

  • Phones & TVs: Wide or tall rectangles.
    Image of standard cell phones from 2024. These are tablet in form and shaped as rectangles, where the height is greater than 2 times the width.
  • Movie Screens: Extreme wide rectangles.
    Stuff
  • Books & Photographs: Standardized rectangles.
  • Paintings & Windows: Overwhelmingly rectangular.

When we create objects for visual consumption, we instinctively default to non-squares. In fact, artists that produce square painting are often making a point. Your data visualizations should be no different.

Conclusions #

Choosing your aspect ratio is an active, data-driven decision.

You have an infinite number of options, but your choice should be simple:

  1. First, try to let your data decide. Bank your main trend to 45 degrees to make it as clear as possible.
  2. If there’s no clear trend, default to a standard, familiar rectangle (like the Golden Ratio or a paper size).
  3. Only use a square when your data (like a correlation plot) specifically demands it.

Ditching the square default is one of the easiest ways to make your plots look more professional and feel more comfortable to your audience.