Chapter 4: Analysing the Data |

## NormalisingA common requirement for parametric tests is that the population of scores from which the sample observations came should be normally distributed. While many variables are close enough to a normal distribution and many of the tests that we will encounter are quite robust to moderate departures, occasionally there is a need to transform a variable so that the requirement of normality is better met. Essentially this means transforming the distribution such that the Howell describes transformations on pages 323-329. You only need to understand the principles behind such transformation and the names of some of the main ones such as the logarithmic, reciprocal, and square root transformations. Assignment I asks you to look more closely at some transformations. |

© Copyright 2000 University of New England, Armidale, NSW, 2351. All rights reserved Maintained by Dr Ian Price |