When you apply some sort of calculation to data items in order to transform them into a form more suitable for analysis. For example, if you have a 5 point Likert scale you might transform this into {{binary data} by mapping the top two values (4 and 5) into 'satisfied' and the rest (1, 2, and 3) into 'not satisfied'. Temporal data is often asymmetric and furthermore larger items typically have greater variance. However, if you take the logarithm of the times, the resulting data is often approximately Normal. This last transformation is so common that data such as reaction times are termed log-Normal data.
Used on pages 52, 53
Also known as transformation