The Toilet Paper

Would you prefer pie or cupcakes?

For Pi Day we take a look at best practices for data visualisations from the perspective of design professionals and normal people.

A happy cupcake, a happy pie and a happy chocolate bar
Bars and pies are pretty sweet (and so are cupcakes)

Data visualisations like bar and pie charts can be used to convey large amounts of quantitative information in an accurate and efficient way. When such visualisations are shared with wide audiences via printed mass media or the internet, they are often primarily designed to be attractive and aesthetically pleasing. It can be hard to find the right balance between these two sets of qualities.

Why it matters


Design is not a science: it’s the result of intuition and experience rather than explicit knowledge. However, designers can use scientific insights to create designs that are more effective at conveying the right information.

These insights have given us guidelines and principles for the design of data visualisations which typically emphasise the importance of clarity, efficiency and accuracy. Acclaimed information designer Edward Tufte in particular strongly believes in high data-ink ratios in visualisations and considers anything that does not convey information to be chartjunk.

On the other hand, visualisations that are made for larger audience need to be able to grab and retain attention of readers, and . It’s probably not a coincidence that most visualisations in design literature primarily focusses on visual pleasure.

Another prominent designer, Nigel Holmes, who is well-known for his information graphics in Time magazine, claims that visual embellishments are necessary to grab and hold the attention of readers. This claim is in stark contrast with what we see in practice however: most still appear to prefer graphs that are conservative in style.

So do we really know what works best?

How the study was conducted


The authors wish to study whether professional designers (who make the visualisations) and laypeople (whom the visualisations are made for) perceive data visualisations differently.

To this end, they study two attributes of data visualisations:

  • The construction type, which can be standard or non-standard: Standard constructions, like bar charts, are the most efficient way to present data. Non-standard constructions are charts that are not as efficient, like pie charts and donut charts. In this study the authors choose to treat pie charts as standard constructions though, due to their ubiquity in mass media and ability to visualise proportions.

  • The expression mode, which can be abstract or pictorial: Abstract visualisations are simple, but still include some decoration. Pictorial visualisations are those that contain graphics that depict recognisable objects or scenes.

An experiment was conducted with 30 graphic design students and 41 students with communication or economic majors. Participants carried out several tasks related to 20 data visualisations with the aforementioned attributes. Each participant was asked to rate the clarity, attractiveness and quality of visualisations, and to select their three most and least favourite visualisations.

What discoveries were made


Most of the results were not very surprising.

For instance, both design professionals and laypeople believe that than visualisations that are non-standard or pictorial. The response times that were measured for the tasks confirm these beliefs, as the times were lower (i.e. better) for standard and abstract visualisations than for non-standard and pictorial visualisations.

Nevertheless, there are some interesting differences between the two groups.

As one would expect, laypeople generally gave higher attractiveness and overall ratings to visualisations than design professionals – probably because they are less critical. Design professionals also were more positive about non-standard constructions and pictorial visualisations, whereas laypeople preferred standard constructions and abstract visualisations.

The difference between the two groups can be explained by the fact that professionals tend to assign more value to attractiveness and originality, whereas laypeople primarily value clarity of visualisations.

These results cast doubt on the ability of graphic designers to effectively bridge the gap between usability and aesthetics in data visualisation. The authors therefore urge designers to test their designs among laypeople if those designs should be tailored to the needs of their audiences.


  1. Abstract, standard visualisations, like bar charts, are easier to read than visualisations that are “original”

  2. Unlike designers, laypeople prefer easy-to-understand visualisations over creative and non-standard visualisations

  3. Designers should not only test their designs among fellow designers, but also among laypeople