Common Statistical Methods for Clinical Research with SAS Examples, Third Edition, 3rd Edition

Appendix F. Data Transformations

• F.1 Introduction 489

• F.2 The Log Transformation 490

F.1 Introduction

Many statistical tests discussed in this book, including the two-sample t-test (Chapter 5) and ANOVA (Chapters 6, 7, and 8) are performed under the assumption of normally distributed data with variance homogeneity among groups. Although such tests are known to be robust against mild deviations to these assumptions, especially with larger sample sizes, clinical data are often encountered that appear to have skewed distributions or larger variability within certain groups. The underlying distribution in such tests might actually be non-normal, or there might be a small proportion of the data called 'outliers' that occur outside the range of usual expectation ...