Abstract:
One of the purposes of this research is to study four method of estimating variance component in two way classification with unbalanced data, which is factorial experiment in completely randomize design. These four methods are Maximum Likelihood method (ML). MINQUE method, Iterative MINQUE method and Restricted Maximum Likelihood method (REML) The estimates of both fixed affects and variance components obtained by the ML method are those which make the likelihood function maximized. The MINQUE method give the estimators which have good properties about norm, unbiasedness and variance. The estimators will have minimum variances only if the prior values of variance components are the true values. But that kind of values are not easily to obtain. One way to solve this problem is using the iterative MINQUE method. The estimators obtained from this method may be negative. If we only restricted to the non-negative estimates, the estimated are the same as those obtained from the REML method. The REML method is the one which applies the maximum likelihood technique to the part of the likelihood function which entirely free of fixed effects. Comparisons are also made among the estimators obtained from these four method both in computation and their properties. It can be pointed out that the Iterative MINQUE with non-negative estimates method is the best among the four.