R devel - Jul 2007 - anova tables in survreg (PR#9806)

Full_Name: Andrew Manners
Version: 2.5.1
OS: windows xp prof 2003
Submission from: (NULL) (130.102.0.177)



To whom it may concern,

I'm trying to get an ANOVA table within survreg but it always produces
NA's in
the p-value, regardless of the data set. The data set below comes from Tableman
and Kim 2004. I had the same problem on a number of my own data sets. I searched
the R site for others that have had this problem and it appears that at least
one has: Ronaldo Reis Junior <chrysopa> wrote a couple of messages but
there
were never replies (search using "survreg" and "anova"
brings up the two
relevant threads).

If this problem is unlikely to be fixed soon can you please let me know how to
install R 2.3.1? I've downloaded the zip but can't seem to get it
installed...I'm sure it's a matter of knowing which button to press,
nevertheless I can't find that button.

Thanks for your help,

Andrew
> motor<-read.table("motorette.csv",header=TRUE,
sep=",",na.strings="NA",dec=".",strip.white=TRUE)
> 
> motor
    X time status temp    x
1  11 1764      1  170 2.26
2  12 2772      1  170 2.26
3  13 3444      1  170 2.26
4  14 3542      1  170 2.26
5  15 3780      1  170 2.26
6  16 4860      1  170 2.26
7  17 5196      1  170 2.26
8  18 5448      0  170 2.26
9  19 5448      0  170 2.26
10 20 5448      0  170 2.26
11 21  408      1  190 2.16
12 22  408      1  190 2.16
13 23 1344      1  190 2.16
14 24 1344      1  190 2.16
15 25 1440      1  190 2.16
16 26 1680      0  190 2.16
17 27 1680      0  190 2.16
18 28 1680      0  190 2.16
19 29 1680      0  190 2.16
20 30 1680      0  190 2.16
21 31  408      1  220 2.03
22 32  408      1  220 2.03
23 33  504      1  220 2.03
24 34  504      1  220 2.03
25 35  504      1  220 2.03
26 36  528      0  220 2.03
27 37  528      0  220 2.03
28 38  528      0  220 2.03
29 39  528      0  220 2.03
30 40  528      0  220 2.03
> 
>
mot.sreg<-survreg(Surv(time,status)~x,dist="weibull",data=motor)
> anova(mot.sreg)
     Df Deviance Resid. Df    -2*LL P(>|Chi|)
NULL NA       NA        28 311.3634        NA 
x    -1 22.77348        27 288.5900        NA #this is the
problem...
> summary(mot.sreg)
Call:
survreg(formula = Surv(time, status) ~ x, data = motor, dist =
"weibull")
             Value Std. Error     z        p
(Intercept) -11.83      1.957 -6.04 1.50e-09
x             9.00      0.901  9.99 1.73e-23  #works fine
Log(scale)   -1.02      0.221 -4.64 3.49e-06

Scale= 0.359 

Weibull distribution
Loglik(model)= -144.3   Loglik(intercept only)= -155.7
        Chisq= 22.77 on 1 degrees of freedom, p= 1.8e-06 
Number of Newton-Raphson Iterations: 7 
n= 30

<a.manners <at> sib.uq.edu.au> writes:
> To whom it may concern,
> 
> I'm trying to get an ANOVA table within survreg but it always 
> produces NA's in
> the p-value, regardless of the data set. 
  [snip]

   The problem is fairly clear, but not knowing the larger context
I don't know the best way to fix it.

  survival:::anova.survreg passes the following table to stat.anova():

     Df Deviance Resid. Df   -2*LL
NULL                    28 311.363
x    -1   22.773        27 288.590

stat.anova figures out where to look for the deviance
(dev.col is 2 here) and pulls out the difference
in degrees of freedom (dfs = c(NA,-1))
It then uses

vals <- table[, dev.col]/scale * sign(dfs)

  which comes out negative because sign(dfs)
is negative but the deviance is calculated as positive

  therefore

pchisq(vals, abs(dfs), lower.tail = FALSE))

  comes out as NA since vals is < 0.

  In the short term you could do:

a1 = anova(mot.sreg)
pchisq(abs(diff(a1$"-2*LL")),df=abs(a1$Df[2]),lower.tail=FALSE)

[1] 1.822632e-06

  In the longer term it looks like someone (??) should
figure out why anova.survreg and stat.anova are disagreeing
about the appropriate signs of df and deviance, and how to fix it ...

  Ben Bolker