Executive Summary

• Creating a data visualisation involves an encoding of data values as visual features.

  For example, numeric counts are encoded as the heights of bars to create a bar plot.

• Reading a data visualisation involves a decoding of visual features to obtain data values.

  For example, the heights of bars in a bar plot are decoded to obtain counts.

• The effectiveness of a data visualisation depends on the effectiveness of the decoding.

  For example, a bar plot is effective for comparing counts because the visual system is effective at comparing the heights of bars.

• More complex encodings produce more complex visual features.

  For example, a scatter plot encodes two numeric variables as the x- and y-locations of points, which produces the position in space of points.

• More complex visual features lead to more complex decodings.

  For example, the position in space of points in a scatter plot is decoded as the correlation between two numeric variables.

• An effective data visualisation encodes data values to visual features that facilitate an effective decoding from visual features back to information about the data.