ABSTRACT
Estimation of parameters of the remaining (p - 1) variates and their correlation coefficients follows essentially the same pattern as that involved in the estimation of parameters of the variate y in the bivariate distribution. A complete derivation of maximum likelihood estimators for parameters of the multivariate normal distribution was given by Cohen (1957). Applicable estimators in this case are
' 2 1 - A. (1 - flj) 1 - A. (12.5.1)
}. = 1 (12.5.2) Since by definition fu == 1 , the last equation of ( 12.5. l) for i = 1 becomes
(12.5.3)
Note that the bivariate estimators of (12.2.9) can be obtained as special cases of (12.5.1) with p = 2.
