The logarithmic and square root transformations are typically most effective at normalizing data that have a minimum value near 1 and have a range that is at most a few orders of magnitude. If a variable consists entirely of large positive values, the transformed data do not show improved normality.

For example, if the minimum value of your data is m, you might want to subtract from your data as a first step so that the new minimum value is 1. You can translate (and scale) data by using the **a+b*Y** transformation in the **Scaling/Translation** family. Alternatively, the square root and logarithmic transformations are defined as **log(Y+a)** and **sqrt(Y+a)**, so you can specify negative values for the a parameter in these transformations. An example of this is presented in the next section.