November 2006
Welcome to our latest VSNi newsletter, I hope you find the articles informative and useful.
Throughout the newsletter there are links to the web site to give you the opportunity to
seek further information on these and other aspects of our business. We are lucky to be
supported by a committed and loyal user base and I am sure that our newsletters and web
site will receive much comment from you as we roll out significant enhancements to the
web site over the coming months.
VSNi, as ever, is in a period of on-going change, improvement and development.
As one of the fastest growing data analysis vendors this will come as no surprise.
Once again we have our annual release of GenStat, now 9th Edition, firmly in the market
place and receiving good reviews. ASReml too has seen the launch of version 2 on the
wide variety of platforms.
2007 will see VSNi very much out and about. We have just returned from the SASA
conference in Stellenbosch and now prepare for the Australasian GenStat conference
in December. Next year we plan a world-wide road show of events, so if you would
like us to come to you, then drop me an e-mail. Over the coming months our timetables
will be broadcast, so come along and meet with us.
Stewart Andrews, CEO
|
Hierarchical Generalized Linear Models HGLMs have been a feature of GenStat's advanced statistics for several editions
now, but until recently their use seems to have been restricted to a small group
of specialists. This year, however, that looks set to change.
The book Generalized Linear Models with Random Effects: Unified Analysis by H-likelihood
by Youngjo Lee, John Nelder & Yudi Pawitan has been published by Chapman & Hall
to give a full account of the theory. John and Youngjo's HGLM algorithms have
been reimplemented in a more efficient form, with the assistance of Roger Payne,
in the 9th Edition of GenStat for Windows. There is also a menu in the 9th Edition
allowing you to access and run many of the examples from the book.
Finally, Youngjo and Roger have been running workshops on HGLMs at locations ranging from Spain
to South Africa. So, if you need to allow for several
sources of random variation but your data are not from a Normal distribution,
why not give them a try?
Roger Payne, Chief Science & Technology Officer
|