Abstract: Two types of diagnostics are presented for the transformation of explanatory variables in regression. One is based on the likelihood displacement proposed by Cook and Weisberg (1982) for assessing the influence of individual cases on the maximum likelihood estimate of a transformation parameter. The other is based on the local influence theory proposed by Cook (1986) for assessing the influence of small perturbations on the parameter estimates. Computations are performed on two data sets to illustrate the usefulness of these diagnostics.
Key words and phrases: Case deletions, likelihood displacement, local influence, perturbations.