### 3.1 Average and variability for a single parametric dataset

Parametric data are scale data (not ordinal or nominal) that are at least approximately normally distributed. For such data, the mean is the most appropriate measure of the population average, and variability is typically represented by the 95% confidence interval (CI). These items of information are shown graphically as an error plot (Fig. 5), where the mean is shown as a symbol against an appropriate vertical scale, and the 95% CI is indicated by a vertical line (or 'error bar') passing through the symbol for the mean. In some instances, you may see error plots where the line indicating the value of the 95% CI is terminated with horizontal bars. The caption to the illustration must indicate what measure of variability is used to calculate the length of the error bar.

**Figure 5.** Examples of two forms of error bars, both indicating the 95% confidence interval about the mean