How to Create Good Charts

Data visualisations is really useful. I don’t need to tell you why.

There are so many terrible charts out there though - in academia, in the development sector, in corporate. There are entire websites dedicated to collecting hilarious examples of bad charts.

When trying to make evidence-based decision-making happen, we can follow a few rules of thumb to ensure that we are safe on the charts front.

What is a Good Chart?

A good chart is simple.

A good chart:

1. Does not look cluttered
2. Has easily readable text
3. Can be understood with little effort

That’s it. That’s the rule of thumb. Just keep repeating these three points over and over in your head when you create a chart the next time.

In most cases, this can be achieved by restricting yourself to pie, line or bar charts. (Definitely not the 3D kind)

In the context of evidence-based decision-making, if your chart requires a “how to read this” explanation, it’s not a good chart.

How Do You Create a Good Chart?

To make a good chart:

1. Use the right type of chart
2. Provide a title, legend and labels
3. Use fonts, color and contrast effectively
4. Label the axes and data points clearly
5. Use the appropriate scale for the axis

I provide examples for a good pie chart, line chart, and bar chart below.

A Good Pie Chart

A Pie Chart divides a circle into segments.

• All segments must add up to 100%
• Comparing segments gives only an approximate idea
• Using more than 3-4 segments can make the chart difficult to read
• Using a 3D chart can change the perception of its size
• In most cases, it is best to use a bar chart instead

A Good Line Chart

A Line Chart shows data against a continuous axis (usually time).

• X-axis should be continuous
• It can be used to compare multiple lines
• Too many lines make it difficult to read the chart
• Changing the scale of Y-axis can change the meaning of the chart

Instead of a line chart, a scatterplot or a bar chart can be used as well.

A Good Bar Chart

A Bar Chart compares discrete values across categories.

• X-axis shows categories; Y-axis shows values
• It can be used to compare multiple categories
• Values of bars should be clearly mentioned
• Changing the scale of Y-axis can change the meaning of the chart

If there are less than four bars, a pie chart or a stacked bar chart may be used, so long as the percentages add up to a 100.

BONUS: ONE HACK TO INSTANTLY MAKE GOOD CHARTS

Use Datawrapper.¹

Just use it. It’ll save you from 9 out of 10 sins of bad data vis automatically.

Another hack: Show your chart to a colleague or friend and ask them - “Hey, what do you think this chart means?”

So, go on, make good charts. Make a great chart even. Something that you are proud of, like this one that I quite like –

Again, this page is opinions, not gospel. Use your better judgement.

Relevant:

• Friends Don’t Let Friends Make Bad Graphs
• The Data Visualisation Catalogue
• Data Vis Do’s and Don’ts by Datawrapper
• Step-by-step tutorial of improving the readability of a line chart by Yan Holtz
• How to Spot Visualization Lies by Flowing Data