ABSTRACT
We consider a progressively censored sample of size N from a Type I distribution of least extreme values that is censored at points ln Tj, j = 1, . . . , k, where at each point of censoring cj items are removed (censored) from further observation. The number of complete (full-term) observations then is n = N - IJ= 1 cj. The likelihood function for a random sample as described is L = K [fi _!_e<x,-f.t)la exp ( -e<x,-f.t)/a] [n {exp ( -e0"T;-f.t)lay,], (8.2.20)
~ (In 0-J.L) L.J cj exp j=l a + const. (8.2.21) Maximum likelihood estimating equations, a In L/af.L in this case follow as
- - + - 2: exp --n 1 " (x; - J.L) a ai=l a
(8.2.22)
Note that estimating equations (8.2.17) for singly censored samples can be obtained as a special case of these equations with j = 1.
