4.2 Pie charts
When to use: Pie charts can be used to show relative parts of a whole, i.e., proportions (or percentages) of responses according to a specific category. These can be used for single-response categorical variables, but not ‘select multiple’ (as the proportions will be greater than 100% of the number of respondents). Pie charts should be used when your data is nominal and not ordinal.
As stated above, our case study asked respondents for the principal drinking water source among their respective households. We previously graphed all of the potential categories through a bar chart, but we now want to understand the proportion of households with ‘improved’ or ‘unimproved’ principal water sources.
‘Improved’ or ‘unimproved’ drinking water would categorize all of the households into these two options. If access to ‘improved’ is considered ‘public tap/standpipe’, ‘handpumps/boreholes’, ‘piped connection to house’, ‘protected spring’ and ‘bottled water, water sachets’, we can re-categorize and answer the research question as follows in a pie chart.
Through the pie-chart, we can visually see a significant difference between the percentage with and without access to drinking water. However, if our ‘target’ value in the project is 90% of the target population having access to improved drinking water sources, we can easily tell that significantly more work required.
Best practices: Keep the pie charts clean and consistent, comparing just a small number of categories. It is also best to avoid tilting, or 3D imagery, as such effects can make relative sizes impossible to read when the viewer is comparing different angles.
When to avoid: It can be very difficult to discern relative sizes of the ‘slices’ of the pie – avoid using a pie chart when proportions of different categories are similar, as this could make differences difficult to visually discern.