[R] Non-normal residuals.

[Ben Bolker]

  [Taking the liberty of posting back to r-help]

  I'd be curious what the particular objections were.
  I have one of those annoying "it depends" answers.

   In general, transformation can

  (1) change [(de)stabilize] variance across groups/gradients;
  (2) (non)normalize residuals;
  (3) change [(non)linearize] relationships with gradients or time;
  (4) modify interaction terms;
  (5) change the interpretation of responses.

 The problem is that sometimes these goals conflict.
"AVAS" (acepack, Hmisc packages) attempts to do #1 and #3 at the
same time.

  If transforming your data brings you closer to satisfying
the assumptions of your analytic methods and having a sensible
analysis, then that's good.  If it makes things worse, that's bad.
Other choices, depending on the situation, include robust methods
(for "outlier" problems); generalized linear models etc. (for
discrete data from standard distributions); models using t- instead
of normally distributed residuals; generalized estimating equations;
etc etc etc ...  transformation (if it works) is simple and (sometimes)
interpretable.

rkevinburton at charter.net wrote:
> That seems to be a general consensus to transform the data through
> sqrt, log, diff, etc. I was particularly intrigued when I considered
> Box-Cox transformation but there were other time-series gurus that
> recommended against it. Particularly with seasonal or data with a
> trend. Would you have any reservations?
> 
> Thank you.
> 
> Kevin
> 
> ---- Ben Bolker <bolker at ufl.edu> wrote:
>> 
>> 
>> rkevinburton wrote:
>>> Hello,
>>> 
>>> I asked a question about what the most likely process to follow
>>> if after a time-series fit is performed the residuals are found
>>> to be non-normal. One peron responded and offered to help if I
>>> supplied a sample data set. Unfortunately now that I have a
>>> sample I have lost the emai addressl. If you are that person or
>>> have some ideas please email me back at rkevinburton at charter.net.
>>> 
>>> 
>>> 
>> It wasn't me, but ... transform the data? See e.g. ?MASS::boxcox
>> 
>> 
>> -- View this message in context:
>> http://www.nabble.com/Non-normal-residuals.-tp26083746p26084836.html
>>  Sent from the R help mailing list archive at Nabble.com.
>> 
>> ______________________________________________ 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.
> 


-- 
Ben Bolker
Associate professor, Biology Dep't, Univ. of Florida
bolker at ufl.edu / www.zoology.ufl.edu/bolker
GPG key: www.zoology.ufl.edu/bolker/benbolker-publickey.asc

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 261 bytes
Desc: OpenPGP digital signature
URL: <https://stat.ethz.ch/pipermail/r-help/attachments/20091027/f0fd2bf8/attachment-0002.bin>