## ABSTRACT

In the maximum likelihood approach, estimating the population parameters from sample statistics is the central concern. A procedure is sought for generalizing from a sample of individuals to the population of individuals. (This is, it should be noted, a definite departure from principal factors where most derivations assume that population parameters were involved.) There are numerous ways in which estimates of the population parameters could be made. For the procedure derived to maximize the likelihood function, the method of estimating the popu lation parameters must have two valuable characteristics. First, a maximum likelihood estimate will have the highest probability of converging to the popula tion parameter as the size of the sample increases toward that of the population. Second, the estimated parameters will be the most consistent with the smallest variance across samples. Even though maximum likelihood estimates may occa sionally lead to biased solutions, their other advantages often outweigh this disadvantage.