A Guideline of Transformations in Linear and Linear Mixed Regression Models

Published in , 1900

Linear and linear mixed models typically rely on the fulfillment of a set of conditions which are not always met by empirical data. Transformations attempt to correct violations of the assumptions underlying these models. In particular, non-normality, heteroscedasticity and non-linearity. In this work, we review the current state of the literature, develop a guideline for the use of transformations and put our guideline to test with data from the National Household Income and Expenditure Survey.