87
ʌθe
ʌle
Φ0(g) - Φ0(‰)
θ ^min
⅛) - Φo(0)
@max θ
(5.5)
(5.6)
5.2.2 Iterations
During the iteration, in the first round, estimates of cumulative hazard functions can
be obtained through Equations 3.13 and 3.14. Later on, they can be obtained through
Equations 3.19 and 3.20.
The variance of λ0 and Λ1 can be estimated by the bootstrap method. The
corresponding 95 percent confidence interval will be λo ÷ 1.96 × SE(Xq) and ʌɪ ±
1.96 × SE(λι), respectively. For λ0c and ʌɪɑ, we have 95 percent confidence intervals
λ0c ± 1.96 × SE(Xqc) and λlc ± 1.96 × SE(Xic).
In order to test the hypothesis H0 : Xi = λ0; Hi : ʌɪ ≠ λ0, we have the fol-
lowing ilzχ statistic” = jʌɪ- ., which follows a normal (0,1) distribution. If the
bE{λι~ Aq) x 7
p-value is small enough (with certain significance level), we reject the null hypothesis.
Otherwise, we fail to reject the null hypothesis ʌɪ = λ0∙