A chocolate Bar or a piece of Pie

A couple of years ago, I was assigned a task requiring getting some data and visualising it for a business presentation. Data experts are often asked to present some insight or results, and communication has become a skill businesses require from scientists.

I did what I STILL do to this day. I first got the data and numbers I needed to present. Then I chose the best visualisation that would be simplistic and well-understood.

**Due to confidentiality, I am changing the underlying description of the data.

The data described whether the citizens of a country preferred to eat one of the following:

- pasta

- pizza

- chicken

- other.

Each person is only allowed to choose ONE, so the entire population of a country is split between the four menu items. I chose a PIE chart to show the preferences, which is similar to the one below.

Figure 1: A pie chart.

Upon sharing the visual with the team, one member said that pie charts are horrible for visualisations and useless. I could understand where this may be coming from, but the general negative sentiment was foreign to me. As with most things I discuss in my blog, I will make my peace against this argument by thoroughly explaining the usefulness of visualisation charts.

Perhaps, the most popular argument against pie charts is that without labels, it is difficult to interpret whether a section is 25%, 10%, or so and so. My counterargument is that most charts are useless without labels. To prove myself, I will use the composition of the Universe as an example.

According to the NASA webpage, which was last accessed on the 31st of October 2022; the Universe is made up of:

- 68 % dark energy

- 27 % dark matter

- 5% normal matter.

For those interested, normal matter comprises everything we can see, touch, feel, and hear using our senses or instruments. On the contrary dark matter is believed to be an invisible mass that we can't detect (yet), but we know it's there because its presence can be measured via its effect on other normal matter. Whether normal or dark, matter exerts a gravitational force towards other matter around it. Some go unnoticed or undetected, whilst others can be measured, such as a ball thrown up in the air and it feels gravity pulling it down towards the Earth's centre. On the other hand, dark energy is responsible for the Universe's expansion—a force pushing cosmological objects apart by expanding the space between them. However, experts have yet to determine what dark matter and dark energy are made of. A pie chart is an iconic visualisation to present the Universe's composition. A simple search "composition of the universe" will suffice. For that reason, I was perplexed when I heard "pie charts are useless".

I used the percentages that describe the composition above on a pie chart and a bar chart, which is usually the preferred alternative. I removed the labels from both for fair judgement. Figure 2 shows a pie chart where the pie represents the entire Universe, and each slice shows how much dark matter, normal matter and dark energy there is. Figure 3 shows three bars showing the same statistics.

Figure 2: A pie chart representing the composition of the Universe.

Figure 3: A bar chart representing the composition of the Universe.

Neither visuals (without hovering over the graphs) are clear. If anything, it is evident in a pie chart that the three elements sum up the whole Universe and that roughly dark matter is about 25 %. However, none of this can be deduced from a bar chart.

Now let's add labels!

Figure 4: A labelled pie chart showing the composition of the Universe.

Figure 5: A labelled bar chart showing the composition of the Universe.

Now there is more and similar information on both. I still advocate that a pie chart is better suited for this problem because it is evident that summing up dark matter, dark energy and normal matter make up the whole Universe. On the other hand, a person visualising a bar chart would need to run some calculations in their mind, or a disclaimer should be put up to capture such information.

Whilst pie charts have merit and should be given when due, bar charts offer flexibility when the complete picture becomes overwhelming.

A rather good example is the population of the European Union per country. According to the World Bank, the European Union population has increased in general, but not all countries have reported an increase in the year 2021. I used the population data for 2021 to show differences using a pie chart against a bar chart.

A pie chart shows that the sum of your data must represent a complete picture, a pie! Figure 6 was generated using the % that a country's population makes compared with the whole of the EU. For example, Malta makes up around 0.1% of the EU's population, while France takes 15%. It puts things into perspective, but alas, the distribution of the population across the EU in both Figures highlight why, in this case, a bar chart is preferred over a pie chart.

Figure 6: A labelled pie chart showing each EU country's % population in 2021.

Figure 7: A labelled bar chart showing the % population in 2021 of each country in Europe.

Figure 6 looks chaotic, with many numbers and lines coming out of the chart. It generates a lot of work for someone who wants to communicate this to their audience. They must use many words, disclaimers and additional information for the audience to understand.

Contrarily, Figure 7 is neat! Visualisation Neatness (I may have just coined this term) is crucial in these presentations! The speaker uses a visual presentation because it ought to communicate something complex to the audience while they verbally add information to complete the picture. Personally, I follow ONE rule:

A data chart should be self explanatory.

I take time, and a lot of it sometimes, to ensure that if my chart is shared across different audiences by other people, the message cannot get lost in translation. The best thing about learning how to communicate complex information using visuals is that, sometimes, one does not need an explanation, and your audience is ready for more information! However, visuals occasionally need some added information to avoid different interpretations.

A more informative chart representing the population across EU countries in 2021 is to show not only the percentage of population per country but also its volume. It gives meaning to the percentages indicated. I used both the pie (Figure 8) and bar (Figure 9) charts to capture this, and the neatness of a bar chart is remarkable.

Figure 8: A labelled pie chart showing the population in 2021 of each country in the EU and its relative percentage.

Figure 9: A labelled bar chart showing the population in 2021 of each country in the EU and its relative percentage.

A pie chart is helpful when you have a handful of categories that make up something whole. They are commonly utilised in political election surveys with several parties involved. Whilst a bar chart is also informative; it loses the intuition that the sections represent one thing: a universe, a country, a continent or an organisation.

A bar chart is more suitable when the number of categories increases and the information they contain can be communicated in two ways. It is neater to show counts and percentages with a bar chart. Moreover, a bar chart allows you to split a category, say a country, in two by having more colours put in the graph.

In the end, how data is presented ought to contain enough information that there is little freedom for different interpretations. However, one must remember that a simple, informative chart is preferred over a trendy, complicated visual, which requires more explanation of how the chart should be interpreted than the data itself.

Simplicity is key, and Visualisation Neatness is the Goal!