Roman Pavelka, Statistical Office of the Slovak Republic, Slovak Republic

Type of article: informative article
Pages: 52 – 67

Abstract

The aim of this article is to present methods for constructing and interpreting linear hypotheses based on linear regression models using the analytical capabilities of selected procedures of the SAS statistical system. The parameters of the examined statistical models will be estimated primarily by solving normal equations using the least squares method. The analytical procedures in the SAS system can not only estimate the parameters of statistical models, but are also equipped with functionality for estimating linear functions of the parameters of the statistical model. Estimates of linear functions of the estimated model parameters thus provide a basis for testing general linear hypotheses, which can range from simple to complex comparisons. The modelling of the examined statistical hypotheses will be implemented using various methods of parameterization of the regression effects of the statistical model, as enabled by the SAS analytical system with its software. Various parameterization rules, and thus different methods of constructing the design matrix of the statistical model, will therefore be applied in the construction and modelling of linear hypotheses. For the implementation of linear hypothesis modelling in the SAS, the analytical procedures REG, GLM and the postestimation procedure PLM will be used.