604 Chapter 14*:Life Testing
Method 2:This method uses the fact that
E[−log(1−F(X(i))]=
1
n
+
1
n− 1
+
1
n− 2
+ ···+
1
n−i+ 1
(14.5.7)
and then approximates−log(1−F(x(i))) by the foregoing. Thus, this second method
calls for setting
yi=log
[
1
n
+
1
(n−1)
+ ···+
1
(n−i+1)
]
(14.5.8)
REMARKS
(a) It is not, at present, clear which method provides superior estimates of the param-
eters of the Weibull distribution, and extensive simulation studies will be necessary
to determine this.
(b)Proofs of equalities 14.5.5 and 14.5.7 [which hold wheneverX(i)is theith small-
est of a sample of sizenfrom any continuous distributionF] are outlined in
Problems 28–30.
Problems..........................................................
- A random variable whose distribution function is given by
F(t)= 1 −exp{−αtβ}, t≥ 0
is said to have a Weibull distribution with parametersα,β. Compute its failure
rate function.
- IfXandYare independent random variables having failure rate functionsλx(t)
andλy(t), show that the failure rate function ofZ=min(X,Y)is
λz(t)=λx(t)+λy(t)
- The lung cancer rate of at-year-old male smoker,λ(t), is such that
λ(t)=.027+.025
(
t− 40
10
) 4
, t≥ 40
Assuming that a 40-year-old male smoker survives all other hazards, what is
the probability that he survives to(a)age 50,(b)age 60, without contracting
lung cancer? In the foregoing we are assuming that he remains a smoker throughout
his life.