Using normal quantile plot to select an appropriate transformation to achieve normality

Abstract

The normal quantile plot is a popular and useful tool for assessing the normality of a data set. A nonlinear plot is used to infer evidence that the data did not come from a normal population. The curvature of a plot is exploited to suggest a transformation required to normalize the data, if it exists. This is done by comparing the curvature of a plot against a series of reference curves which correspond to different transformations. Unlike the maximum likelihood method, this technique is robust to outliers. © 2003 Elsevier B.V. All rights reserved.