1983 年 13 巻 1 号 p. 53-56
Brown-Fereday's reduced variate in a linear structural relationship which is modified to be usable when the error variance is unknown is shown to be distributed asymptotically as a weighted sum of independent chi-square variates each with one degree of freedom. A method of constructing an asymptotic confidence region of the coefficient vector of the structural hyperplane based on this variate may be conceived but proves to be inferior to the maximum likelihood method presented by Okamoto and Isogawa (1981).