Estimation of linear parametric functions in multiresponse models

.

Consider an incomplete multiresponse experiment with a classification of the observations into u classes. The class Si contains ti treatments with ni observations. Thus, the experiment can be described by the set {Y1,Y2,...,Yu}, where Yi are ni×ti random matrices of observations from the class Si. For the expectation of each Yi a linear model with main and nuisance parameters is proposed. The problem of (linear) estimation of the main parameters is considered using results of R. Zmyślony [in Mathematical statistics and probability theory (Wisła, 1978), 365–373, Lecture Notes in Statist., 2, Springer, New York, 1980; MR0577292] and J. K. Baksalary [Math. Operationsforsch. Statist. Ser. Statist. 15 (1984), no. 1, 3–35; MR0729609].