Error bars

Error bars are used to indicate the estimated error in a measurement. In other words, an error bar indicates the uncertainty in a value. Data might not always be precise, for example, in measurement data error margins might exist. When there is a need to visualize the uncertainty in the data, you can use error bars to indicate statistical probabilities of errors, or actual errors. The error bars represent upper and lower limits of the data relative to a value, and they can be added to visualization items such as bars in bar charts, and markers in line charts and scatter plots.

In all these three types of visualizations, upper and lower vertical error bars can be drawn. In scatter plots, the error bars can also be drawn horizontally.

The image below shows all four possible error bars on a scatter plot marker.

The degree of how uncertain an item value is, is indicated by the length of the error bar. For example, for the values that the bars above represent, a long error bar means that the value is uncertain. A short error bar means that the value is more certain.

If more than one column is selected on the axis selector, you can specify different error bars for each of the columns. You can choose to show only one of the error bars, or any combination of them. However, all the defined vertical error bars will share the same settings for Color, Show end caps, and Include error bars in axis range. For example, you cannot set one vertical error bar to be blue, and another vertical error bar to be yellow in the same visualization. This also applies to using multiple measures on a Y-axis with a single scale.

Note: Error bars are drawn relative to a visualization item. For example, in a bar chart, an upper error bar will be drawn with the top of a bar as starting-point. If you use reversed scales in a visualization, or change orientation of the bars in a bar chart, the error bars will also be reversed or change orientation respectively.

When specifying what the length of the error bars should indicate, you have different options. You can

let the length represent the actual values in a data column that contains absolute error figures.
for aggregated visualization items, use any of the pre-defined aggregation methods, for example, standard error (StdErr) or standard deviation (StdDev).
define your own method by writing an expression.

This means that you can use error bars also for other purposes than indication of data uncertainty.

Note: When you work with error bars in bar charts, make sure that the bar chart is shown using side-by-side bars.

For more information on how to use the error bars, see Adding error bars.