ABSTRACT
Estimates :y, p, and p can be calculated as the simultaneous solution of the three equations of(7 .2.15). However, to avoid the complications involved in evaluating the third of these equations, we follow the procedure that was employed in calculating estimates from censored samples. We begin with a first approximation p1 or (a~1 ) = 2/~). We substitute this value into the first two equations of (7 .2.15) and solve for the conditional estimates (approximations) -y(p 1) and ~(p 1 ). These approximations are then substituted directly into the loglikelihood equation (7 .2.12) in search of a maximum. It will be necessary to repeat these calculations for several different values of p and perhaps plot a graph of In L as a function of p, in order to determine the value of p for which In L attains its maximum. Since the number of missing observations is unknown in truncated samples, the modified estimators that were presented for censored samples are not deemed feasible here.
