Re: Heterogeneous variance model in ASREML

Dear Subash
Arthur wrote
The corresponding ASReml code is

fruit_set_mean ~ mu pedigree location pedigree.location !r rep.location
3 1
<n1>
<n2>
<n3>

ASReml requires the data be sorted on location and that you have
counted the numbers of records in each location
[which you could do with
  tabulate fruit_set_mean ~ location !COUNT
given as a command line before the model line,
or by using the !SUMMARY qualifier on the data line.]

In this code, 3 is the number of separate locations; equivalently
written as
location 1

<n1> is the number of plots in location 1
<n2> is the number of plots in location 2
<n3> is the number of plots in location 3
--------------

An alternate specification is
fruit_set_mean ~ mu pedigree location pedigree.location ,
 !r rep.location uni(location,1) !GU uni(location,2) !GU

wherein the uni(location,1) model term picks up the difference in
residual variance between loc 1 and loc 3; uni(location,2) picks up the
difference in residual variance between loc 2 and loc 3; and the
reported residual variance will be the variance for location 3.

################################
 your question - is it easier to code in ASReml-R ?
answer
no easier just different syntax
 in an R session you would type
(i assume you have a data frame with columns coded correctly as well as
the variate y = fruit_set_mean (dont use '_' in R as a rule)

>require(asreml)
> your.asr <-
asreml(y~pedigree+location+pedigree:location,random=~rep:location,
rcov=~at(location):units,data=your.data)

thats it

your.data is the name of the data frame - the rest is straight forward I
hope
This is not in most cases an optimal analysis as there is most likely
heterogeneity, in rep:location which is easily fitted with
> your.asr <-
asreml(y~pedigree+location+pedigree:location,random=~rep:diag(location),
rcov=~at(location):units,data=your.data)

warm regards

Brian Cullis|Research Leader, Biometrics &
Senior Principal Research Scientist | Primary Industries
Industry & Investment NSW | Level 6 | 201 Elizabeth Street | Sydney | NSW
2001
T: 02 8281 7777 | F: 02 8281 7799 | M: 0439 448 591 | E:
brian.cullis_at_industry.nsw.gov.au
W: www.industry.nsw.gov.au | www.dpi.nsw.gov.au

Visiting Professorial Fellow
School of Mathematics and Applied Statistics
Faculty of Informatics
University of Wollongong

Professor,
Faculty of Agriculture, Food & Natural Resources
The University of Sydney

Adjunct Professor
School of Computing and Mathematics
Charles Sturt University
 
Received on Wed Aug 25 2009 - 07:34:15 EST