Don’t let color distort the meaning of the underlying data

Brewer color palette

Brewer palette

Effective color choices for effective data visualization

How do you select the colors for your graphs or any other kind of data visualization? Chances are that you do it quite randomly and pick the colors that you like. I’m not sure if this is the most effective way.

As scientists we are very careful with our data. We know that our data were generated with calibrated methods including appropriate controls. It assures correct interpretation and it is at the core of good research. However, the moment we start visualizing the data, most scientists don’t have a calibrated method anymore. This might sound unimportant, but the truth is that it often results in misinterpretation of the data.

The reason is simple, colors aren’t perceptual uniform. Some colors are perceived as more dominant than others. In data visualization this knowledge is very important. You don’t want your reader to be attracted to insignificant data or have them look to one data set more than to the other. Poor color choices might result in unconscious and erroneous  grouping of data sets too.

In most cases, the data is purely informative and we need to present it that way for correct interpretation. This seemingly simple task turns out to be a very difficult one. We’ve never really learned what a perceptual uniform color space is, right? In essence it would be a color space where colors are defined on how we perceive them. Simply put, colors that are further apart in this space will be perceived as very different, while colors close to each other will be perceived as very similar.

Most people use RGB and CMYK color models to choose their colors. These color spaces are not perceptual uniform and easily lead to poor color choices for data visualization. Without going in further detail about the technical aspects I’ll just provide you with a simple solution to tackle this problem:

Brewer palettes

These color palettes are hand-picked by Cynthia Brewer and originally designed for cartography, but quickly picked up in the field of data visualization. Martin Krzywinski wrote an excellent piece about it on his blog. He also offers Brewer palettes as Adobe swatch files for easy integration in Adobe Photoshop and Adobe Illustrator. I highly recommend to check it out.

If you like what you read, feel free to subscribe to our newsletter and mark the box “Tips and Tricks to ‘Visually’ Succeed in Scientific Publishing”. This section of our newsletter is especially meant for publishing authors, PhD students, postdoc and research scientists.

Please send me info about