Milan Stehlík Department of Statistics and Mathematics, University of Economics Vienna, Augasse 2-6, A-1090 Vienna, Austria DPS, FMFI, Comenius University, Bratislava,

DARWin Programming Language - Language Definition and User’s Manual.

Factorial experimental plans and their analysis Karel Kupka, Trilobyte Statistical Software

Karel Kupka, Jaroslava Hálová Advanced mathematical and statistical models help to plan and improve properties of food and pharmaceuticals.

Certification procedure Computational algorithms in the software QCExpert were verified with certified values provided by National Institute of Standards and Technology, Mariland, USA (NIST).

Jan Ámos Víšek The present paper continues in considerations started in Víšek (2000 b) and (2002 c) and focus mainly on a discussion of the underlying philosophy of regression, or even more generally, on the philosophy of modelling. An example of wrong estimation of regression model in the case when we neglect heteroscedasticity is given. The paradigm of point estimation inspired by robust estimating regression model is recalled together with the proposal of the Least Weighted Squares (LWS). Their properties together with some report of results and early-expected results concludes the paper.

QC-Expert 3.3 User's Manual.

Some concepts in evaluation and improvement of quality Qualimetrics and Statistical Software Karel Kupka, Trilobyte Statistical Software

Jan Ámos Víšek The paper recalls ideas of robust statistics, of the Cragg improvement of the OLS-estimator under heteroscedasticity and explains GMM approach in the point estimation. The problem of heteroscedasticity is demonstrated on the numerical example and the correction of conclusions about the significance of explanatory variables by employment of White's estimator of the covariance matrix of the estimates of the regression coecients is presented and discussed. Then, a version of the least weighted squares resistent to heteroscedasticity is given. A proposal of the weighted GMM estimation, safeguarded against contamination of data concludes the paper.