The Centre of Biostatistics at the University of Limerick invites you to
its fifth seminar in the autumn series
Prof. Gilbert Mackenzie (University of Limerick)
The R integration in SPSS V16: New Models for Old.
Wednesday November 19th
2:30pm, A2002
In version 16 of the package, SPSS has made available facilities
to call R and so run R code. Lately, SPSS has been perceived,
by the academic statistical community, to have been
rather languishing in the technical statistical doldrums.. However, this
new strategy removes this criticism, at a stroke, since it brings the
existing technical power of R under the aegis of version 16. Moreover,
in R one has a general language with which to develop and fit new
models,
thereby further extending the power of SPSS. The R development is part
of
an overall strategy of "opening up" of the package to other systems.
This talk reviews progress in technical statistical developments in SPSS
since 2001 and
and shows how to run R code in SPSS V16. The context is one of fitting
general
statistical models by maximum likelihood using the nlm function in R.
The examples
are from the survival literature and we use the nlm function to fit the
Generalized
Time -Dependent Logistic Model (MacKenzie, 1996;1997) to a set of lung
cancer
data, contrasting the facilities available in the SPSS non-linear
regression routine CNLR.
We conclude with a discussion of the usefulness of the R environment to
the SPSS user
community and forthcoming developments in SPSS V17.
References
MacKenzie,G (1996): JRSS D, 45, 1, 21-34
MacKenzie G. (1997): SIM, 16, 1831-1843
--
|