### 2.1 Summarising data that are nominal or can be reduced to categories in some way

The number of observations within each category can be represented either as a piechart or a bar-chart. In a pie chart, each category is represented as a segment of a disc (the 'slice' of the 'pie'), whose angle is proportional to the proportion of the overall observations accounted for by the category (Fig. 1).

**Figure 1.** An example pie chart, where each segment is labelled

In a bar chart, the magnitude of each category is represented by the length of a bar (usually vertical but occasionally you will meet horizontal bar charts) (Fig. 2). The value for a category may be the number of observations, or the proportion of the total observations in the category.

**Figure 2.** An example bar chart, showing the magnitude for each category as a vertical bar

A pie chart is not an appropriate choice where there is a large number of categories, say more than six and certainly not more than ten. A bar chart may work better than a pie chart where there are more categories, although the advantages may be minimal and it still becomes difficult to label different categories.

Where there is more than one set of observations, each set can be represented by a separate pie- or bar- chart (Fig. 3). In the case of the pie chart, which can only show proportions as the segment angles, different numbers of observations can be represented by making the overall area of the disc proportional to the number of observations in each dataset.

**Figure 3.** An example of using two pie charts to compare sets of observations, in this case differences in social groups of elephants in the wet- and dry- seasons