Pre-attentive attributes play an important role in data visualization because it highlights significant data points. As a pre-attentive attribute, length makes it very easy to compare items at a glance. That is why bar and column charts are so commonly used. However, could a pie chart have pre-attentive attributes of its own?
There is certainly a lot of debate about whether pie charts are helpful at all. Notable names in the world of data visualizations such as Cole Nussbaumer Knaflic have strong opinions against it for fair reasons.
I am here, however, to make the case that pie charts can be a strong tool for information dissemination if used appropriately.
How we read pie charts
It is difficult to find the largest share in a pie chart, but suppose what we really want to know is how much of the whole is taken up by the red share.
The pie chart in Fig 2 highlights only one section and compares it to the rest of the whole. Our brain is taking in three aspects of the red part, the area, the arc length and the angle. A recent study conducted by Robert Kosara and Drew Skau, found that we are more accurate in guessing the share of a pie because our brains can calculate the area (possibly in conjunction with arc length) better than the angle. When shown only an angled line, participants are less accurate and even less sure about their guesses. Whereas, they are more likely to be correct and confident of their answers when shown a pie chart with a filled-in sector. Treemaps have probably become popular charts for the same reason (apart from the fact that they are hierarchical).
Area as a pre-attentive attribute
Our brains can already divide a circle into quarters using multiples of 90-degree angles. We are taught how to read time in the same way; the first 90 degrees clockwise is an area within quarter-of-an-hour etc.
We know the following and can process it very quickly:
0–90 deg = 1/4
90–180 deg = 1/2
180–270 deg = 3/4
So imagine you want to show that in 2016 only 22% of households in South Sudan had access to electricity (World Bank, 2019). When turned into a pie chart, it would fall under the area where the angle would be within 0 and 90 degrees and our brain would immediately process that as less than a quarter. This visual cue could be more impactful and memorable than the actual statistic.
Pie charts work best only when comparing one item to the whole. If different sections are to be compared to each other, I would refer you back to a bar or column chart. If the data in such a chart is in percentages then the axes should still go up to 100%. This has a two-pronged display. Not only are you comparing each item to each other, but you are also comparing each item to the whole (100%) in the same chart. This only works if the values are in percentages and large enough to be distinguishable when put against an axis that goes up to 100%.
The following is a chart created by Code for Africa that uses this principle. We learn that most of the population accesses water through shared boreholes. Additionally, we can see that less than half of the population has their own private borehole.
Takeaway point: Use bar charts when comparing items to each other, use pie charts when comparing one item to the whole.
There are some caveats with pie charts. Sections that are not clearly marked or highlighted can cause confusion. Here are some tips that can improve the efficiency of a pie chart:
- Use Hue and contrast to make the distinction very clear. Consider turning down the opacity and saturation of the sections you don’t want to highlight ie. make them closer to grey.
- Highlight the main section starting from the top and then in a clockwise direction because that is how our brain is conditioned to read degrees on a circle.
Whenever we have categorical data that adds up to 100%, most of us would first think of turning it into a pie chart. However, it takes a lot of work to make it efficient as a tool and perhaps that is why it is becoming increasingly unpopular. Nevertheless, if used correctly, pie charts can add variety to your infographics and make data interesting. Our foremost goal is to bring attention to the data and it is only a few tips and tricks away.