Estimation of the Intercept of a Linear Regression Function.

Experiments performed at the levels x1, x2, ... lead to the obtaining of observations yiabxii1,2,..., the synbol being the independent NO,2 random variables, and a and B the unknown linear regression coefficients. Designs of experiments for estimating are given which are based on certain optimum criteria in both the nonsequential and sequential cases. Properties of both the nonsequential and sequential estimates tN are discussed, including in the nonsequential case of approximation to the distribution of tN and a confidence interval for 0. Examples were constructed to illustrate the character of the designs.