R help - Feb 2011 - Interpretation of p-values, coxph

Hi guys,

I have been testing the hypothesis for difference of survival between four
different classes. The p-values provided in coxph output are confusing for me to
interpret.
Here is the output :

sur<-coxph(Surv(SURVIV, status == 1)~factor(A)+cluster(rownames(d)), data =
d, model=TRUE)
Call:
coxph(formula = Surv(SURVIV, status == 1) ~ factor(A) + cluster(rownames(d)),
data = d, model = TRUE)


            coef exp(coef) se(coef) robust se       z      p
A2 -15.438  1.97e-07 3842.826     1.022 -15.112 0.0000
A3   0.471  1.60e+00    0.615     0.564   0.835 0.3000
A4   1.550  4.71e+00    0.547     0.551   2.811 0.0039

Likelihood ratio test=6.73  on 3 df, p=0.0811  n=171

---
In above result, I can see that overall p-value from LR test is non-significant.
But the p-values for two groups(A2 and A4) are significant.
I have three questions: 

Can anyone help me whether this points to any significant interpretation ?

Is it okay to ignore the overall p-value ?

Please also let me know if you have got some more suggestions to improve the
result.

Thanks,
-V







 		 	   		  
	[[alternative HTML version deleted]]

First I have read all the previous post on error bars & segments as well as
the  Lattice book by Sarkar.

I am using xyplot, there are 2 panels, each panel displays the measurements from
5 matrices over time.  The matrices are identified by different colors.  My
error bars show up in the 1st panel and not the second.  Any suggestions would
be greatly appreciated the code follows - thanks.

#Figure 2
data2$Type<-ordered(data2$Type, c("SDM","AcSDM"))
sub2<-table(data2$Matrix, data2$Type)
fill.palette2<-c(rep(palette2, sub2[,1]), rep(palette2, sub2[,2]))
subscripts2<-1:dim(data2)[1]

#Figure 2
xyplot(Response~Time|Type, data=data2, upper=log10(data2$Response+data2$SD), 
	lower=log10(data2$Response-data2$SD),
	groups=fill.palette2, cex=1.35, layout=c(2, 1),
type=c("p","g"), 	col="black", aspect=1, 
col.line="light grey", lty=8, xlim=c(30, 120),
	scales=list(x=list(at=seq(from=30, to=120, by=20)), y=list(log=10)),
	ylab="Concentration in Matrices (ppb)",  
	xlab="Time after last dose (hour)", 
	main="Figure 2.  Depletion for Average Concentration by Matrix over Time 
			(Length of the black vertical line is two standard deviations)",
	panel=function(x, y, groups, subscripts, upper, lower, ...){
		panel.xyplot(x, y, pch=21, fill=groups[subscripts2], ...)
		panel.abline(v=100, col="black", lty=8)	
		panel.text(110, log10(55000), "Post-mortem", cex=0.7)
		panel.text(110, log10(40000), "Measurements", cex=0.7)
		switch(panel.number()
		panel.segments(x[subscripts], lower[subscripts], 
			x[subscripts], upper[subscripts], lty=1, col="black")},
	
	key=list(corner=c(0.02, 0.04), text=list(levels(data2$Matrix)),		
	cex=0.75, columns=1,border=TRUE, 
	points=list(pch=c(21, 21, 21, 21, 21), fill=palette2, 						cex=c(1.5, 1.5,
1.5, 1.5, 1.5)))
	)