[R] split and rlm

[Peter Dalgaard]

Stuart Luppescu <s-luppescu at uchicago.edu> writes:

> Hello, I'm trying to do a little rlm of some data that looks like this:
> 
> UNIT    COHORT     perdo     adjodds
>  1010      96      0.39890    1.06894
>  1010      97      0.48113    1.57500
>  1010      98      0.36328    1.21498
>  1010      99      0.44391    1.38608
> 
> It works fine like this: rlm(perdo ~ COHORT, psi=psisquare)
> But the problem is that I have about 100 UNITs, and I want to do a
> separate rlm for each one. I tried to use split and lapply but it didn't
> work at all. Is this possible?

How about by()?

 
> In addition, I'm trying to extract the t statistic for the slope
> coefficient and the degrees of freedom so I can put them into dt() to

pt(), I would hope....

> get the p-value. I can get the t from coef(summary(u))[2,3] (where u is
> my rlm object), but u$df.residual gives me NULL. Also, the help for
> summary.lm says it returns coefficients, which contains a 4x4 matrix
> including the p-values, but when I do summary(u)$coefficients I get:
> 
>  summary(u)$coefficients
>                                        Value Std. Error    t value
> (Intercept)                      0.151756859 3.00972988 0.05042209
> drops$COHORT[drops$UNIT == unit] 0.002769108 0.03086700 0.08971097
> 
> Any help with this, and on getting the degrees of freedom or the p-value
> would be much appreciated. 


As Brian usually puts it: This is support software for a book. You
might want to read it...

Looking at the code, it seems to explicitly set the residual DF to NA,
which - along with the absense of p values - usually suggests that the
author(s) do not trust them and I think I can see why. Apparently the
lm() methods which get invoked through inheritance have ways of
sticking the residual DF back in, but that doesn't make them any more
sensible. If you must, have a look inside print.summary.lm and see
where "rdf" comes from, but I'd suspect that you're just as well off
by using the normal approximation.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907