[R] logistic model diagnostics residuals.lrm {design}, residuals()

[Chaudhari, Bimal]

I am working on applying Sequential Multiple Decision Procedure to
genetic association studies of complex disease.  Province in 2000
(http://www.ncbi.nlm.nih.gov/pubmed/11108641) developed the method for
QTL linkage and Province and Zhang extended it to association with
quantitative traits in 2005.  My interest is in binary outcomes.

The original formulation of the SMDP (Bechhofer, Kiefer and Sobel, 1968)
was predicated on comparing parameters from Koopman-Darmois
distributions.  In that text, they present solutions for normal,
exponential, Bernoulli, poisson and negative binomial distributions.
Province and Zhang exploited the fact that residuals from linear
regression are normally distributed to apply SMDP to the analysis of
these residuals (specifically, they use the case of comparing the
variances of several random variables drawn from normal distributions
with the same known mean, 0, and unknown, unequal variances to identify
the set of random variables/residuals with the smallest variance).

My hope initially had been that I might be able to essentially replicate
this approach with an analogous measure/residual, but this has proven to
be more difficult than anticipated.


-bimal

> -----Original Message-----
> From: Frank E Harrell Jr [mailto:f.harrell at vanderbilt.edu]
> Sent: Thursday, March 11, 2010 12:25 PM
> To: Chaudhari, Bimal
> Cc: r-help at r-project.org
> Subject: Re: [R] logistic model diagnostics residuals.lrm {design},
> residuals()
> 
> Chaudhari, Bimal wrote:
> > I am interested in a model diagnostic for logistic regression which
> is normally distributed (much like the residuals in linear regression
> with are ~ N(0,variance unknown).
> >
> > My understanding is that most (all?) of the residuals returned by
> residuals.lrm {design} either don't have a well defined distribution
or
> are distributed as Chi-Square.
> >
> > Have I overlooked a residual measure or would it be possible to
> transform one of the residual measures into something reasonably
> 'normal' while retaining information from the residual so I could
> compare between models (obviously I could blom transform any of the
> measures, but then I'd always get a standard normal)?
> >
> > Cheers,
> > bimal
> 
> Hi Bimal,
> 
> What would make it necessary for the residuals to have a certain
> distribution?  Why would you expect a categorical Y variable to give
> risk to residuals with a nice distributions?
> 
> You can do residual diagnostics without worrying about the
> distribution.
> 
> Frank
> 
> >
> > Bimal P Chaudhari, MPH
> > MD Candidate, 2011
> > Boston University
> > MS Candidate, 2010
> > Washington University in St Louis
> >
> >
> > 	[[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help at r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide http://www.R-project.org/posting-
> guide.html
> > and provide commented, minimal, self-contained, reproducible code.
> >
> 
> 
> --
> Frank E Harrell Jr   Professor and Chairman        School of Medicine
>                       Department of Biostatistics   Vanderbilt
> University