Dear All:
I have some questions about the output of coxph.
Below is the input and output:
----------------------------------------
> coxph(formula = Surv(futime, fustat) ~ age + rx + ecog.ps, data +
ovarian, x = TRUE)
Call:
coxph(formula = Surv(futime, fustat) ~ age + rx + ecog.ps, data ovarian, x =
TRUE)
coef exp(coef) se(coef) z p
age 0.147 1.158 0.0463 3.17 0.0015
rx -0.815 0.443 0.6342 -1.28 0.2000
ecog.ps 0.103 1.109 0.6064 0.17 0.8600
Likelihood ratio test=15.9 on 3 df, p=0.00118 n= 26
---------------------------------------
Question One:
As I know, the p-value of "age" is the significance level. However
what
is the exact meaning of the parameter, and how do we calculate the
parameter? If the sample size is small (20~40), is this estimation still
reliable?
Question Two:
the p-value in the last line (Likelihood ratio test=15.9 on 3 df,
p=0.00118) is asymptotically equivalent tests of the omnibus null
hypothesis that all of the 佄伈佲檚 are zero, according to John Fox's "Cox
Proportional-Hazards Regression for Survival Data"
(http://cran.r-project.org/doc/contrib/Fox-Companion/appendix-cox-regression.pdf)
Can anybody explain that why this true? (As I know, the p-value is
obtained by 1-pchisq(2*log Likelihood ratio), and this is because
2*log(likelihood ratio) is approximately chi-square for nested models.)
Thank you very much.
Sincerely,
Alan
2005-11-27
Dear All:
I have some questions about the output of coxph.
Below is the input and output:
----------------------------------------
> coxph(formula = Surv(futime, fustat) ~ age + rx + ecog.ps, data +
ovarian, x = TRUE)
Call:
coxph(formula = Surv(futime, fustat) ~ age + rx + ecog.ps, data ovarian, x =
TRUE)
coef exp(coef) se(coef) z p
age 0.147 1.158 0.0463 3.17 0.0015
rx -0.815 0.443 0.6342 -1.28 0.2000
ecog.ps 0.103 1.109 0.6064 0.17 0.8600
Likelihood ratio test=15.9 on 3 df, p=0.00118 n= 26
---------------------------------------
Question One:
As I know, the p-value of "age" is the significance level. However
what
is the exact meaning of the parameter, and how do we calculate the
parameter? If the sample size is small (20~40), is this estimation still
reliable?
Question Two:
the p-value in the last line (Likelihood ratio test=15.9 on 3 df,
p=0.00118) is asymptotically equivalent tests of the omnibus null
hypothesis that all of the 佄伈佲檚 are zero, according to John Fox's "Cox
Proportional-Hazards Regression for Survival Data"
(http://cran.r-project.org/doc/contrib/Fox-Companion/appendix-cox-regression.pdf)
Can anybody explain that why this true? (As I know, the p-value is
obtained by 1-pchisq(2*log Likelihood ratio), and this is because
2*log(likelihood ratio) is approximately chi-square for nested models.)
Thank you very much.
Sincerely,
Alan
2005-11-27