The article deals with the use of various types of contrasts, especially in logistic regression. Contrasts were originally developed within the framework of the analysis of variance. They gradually expanded into other statistical methods, for example, into logistic regression and loglinear or logit models. They are also used in linear regression when there are categorical variables among the explaining variables. Contrasts represent a method for working with variables that are not numerical but categorical. The article refers to the well-known types, such as indicator, simple, deviation, repeated, Helmert, difference and polynomial contrasts. Several others are also proposed. Contrasts are classified according to the appropriateness of their use for different types of categorical variables (nominal, ordinal). Their meaning and effect on the interpretation of odds ratios are explained on the basis of examples created using real data.
Financed by the National Centre for Research and Development under grant No. SP/I/1/77065/10 by the strategic scientific research and experimental development program:
SYNAT - “Interdisciplinary System for Interactive Scientific and Scientific-Technical Information”.