[R] log-linear analysis - calculate treatment effects

[Prof Brian Ripley]

Use multinom or glm instead.  The IFP algorithm used in loglin is not 
designed to find parameter estimates let alone standard errors.

Fitting log-linear models is discussed in all good books on R/S, e.g.
MASS (see the FAQ for full details).


On Tue, 17 Jan 2006, Becker Cordula wrote:

> Dear all,
>
> I have run a hierarchical log-linear analysis using loglin {stats}
> and came up with a specific model. loglin returns me the parameter
> estimates giving me an idea in which direction the observed
> frequencies deviate from the expected ones for my different factors.
> To assess the significance of any such effects I would need to
> calculate the ratio of the log-linear parameter estimate to its
> standard error (as explained in Howell: Statistical Methods for
> Psychology, 4th Edition, p. 637ff). Unfortunately, I have no idea and
> couldn't find out what this standard error refers to and how to
> calculate it or the ratio using R (well, the ratio is no problem if I
> have the standard error).
>
> I would very much appreciate any help!
>
> Thanks and best wishes, Cordula Becker
>
> ----------------------------------------------------
> Dr. Cordula Becker
>
> Department of Experimental Psychology
> University of Cambridge
> Downing Site
> Cambridge  CB2 3EB
> United Kingdom
>
> Tel.: +44 (0)1223 339715
> Email: cb486 at cam.ac.uk
>
> http://vision.psychol.cam.ac.uk/cbecker/
>
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595