A Multivariate Approach to a Mixed Linear Model
混合线性模型的多变量方法
The usual treatment of the two-way mixed model parallels that of the fixed models, with the usual sums of squares appearing in the F ratios. In this article, we argue in favor of a multivariate approach. We propose a new criterion for testing a general hypothesis under general conditions and offer an expression for the asymptotic expansion of the distribution function.