Re: Means of levels of a random term
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Means of levels of a random term



Dear Nick,

The ASREML list is still not getting through to me.
There is an addressing problem we still have not sorted.

Your .pin file problem is a bug which will occur if you have only
1 or 2 fixed effects in the model.  I've now fixed it in my code
which should be available on the website later in the week.

The problem is that it thinks there is an extra set of solutions in the
DENSE equations but they are not there and so the values are zeros.

There is no work around until I fix it.

Note that ASREML does not give the correct SE's for predicted values
when some of the effects are in the SPARSE part of the model.
You need then to specify !DENSE to put them in the !DENSE part
but that only works for some random effects and also seems to be broken at the 
moment.

Arthur

> 	We (Dadan Kusnandar and myself) are trying to get means for levels of a
> random term using a .pin file to combine the parameter estimates in the .sln
> file.   The means given are correct except for the first level of the random
> factor, which is being taken to have an effect of zero, and hence is giving
> the grand mean.   This is despite the fact that the .sln file does NOT give
> the effect of this level as zero.

> 	As you will see from our .pin file (appended), we have tried various > 
ways
> of specifying the effects we are after - shorthand, longhand, and the
> effects of each level rather than the means.   In every case we get the
> result we expect except for Level 1.

> 	I also append the .pvs file and the first few lines of the .sln file.

> 	What's going on?

> 	The examples in the manual that we're following are for the means of 
levels
> of a fixed-effect factor, and we suspect that this is related to the
problem.

> Best wishes,

> Nick Galwey
___________________________________
.pin file

# Obtain provenance means
# Provmeans 1 [2:9]
Grandmean 1
Effect1 2
Effect2 3
Provmean1 1 2
Provmean2 1 3
Provmean3 1 4
Provmean4 1 5
Provmean5 1 6
Provmean6 1 7
Provmean7 1 8
Provmean8 1 9

___________________________________
.pvs file

 'Grandmean           '     1.00000     0.03076     0.15650
 'Effect1             '     1.00000     0.00000     0.00000
 'Effect2             '     1.00000     0.35460     0.78177
 'Provmean1           '     1.00000     0.03076     0.15650
 'Provmean2           '     1.00000     0.38536     0.79728
 'Provmean3           '     1.00000    -0.34114     0.76116
 'Provmean4           '     1.00000     0.17046     0.75600
 'Provmean5           '     1.00000     0.23476     0.79814
 'Provmean6           '     1.00000    -0.58524     0.76202
 'Provmean7           '     1.00000    -0.39054     0.75987
 'Provmean8           '     1.00000     0.47146     0.75558

___________________________________
.sln file (first few lines)

  mu                              1          0.3076E-01  0.1565
  prov                            1          0.2701      0.1704
  prov                            2          0.3546      0.1781
  prov                            3         -0.3719      0.1697
  prov                            4          0.1397      0.1685
  prov                            5          0.2040      0.1783
  prov                            6         -0.6160      0.1699
  prov                            7         -0.4213      0.1694
  prov                            8          0.4407      0.1684
  rep                             1         -0.2503E-01  0.4021E-01
  rep                             2          0.1654E-01  0.4053E-01
  rep                             3          0.8483E-02  0.4036E-01
  rep.block                1.0001           -0.9539E-01  0.8190E-01
  rep.block                1.0002           -0.1372E-01  0.8434E-01
  rep.block                1.0003            0.1655E-01  0.8210E-01

--

<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>
Arthur Gilmour PhD                  mailto:Arthur.Gilmour@agric.nsw.gov.au
Senior Research Scientist (Biometrics)               fax: <61> 2 6391 3899
NSW Agriculture                                           <61> 2 6391 3922
Orange Agricultural Institute             telephone work: <61> 2 6391 3815
Forest Rd, ORANGE, 2800, AUSTRALIA                  home: <61> 2 6362 0046

ASREML is still free by anonymous ftp from pub/aar on ftp.res.bbsrc.ac.uk
    or point your web browser at ftp://ftp.res.bbsrc.ac.uk/pub/aar/ 

To join the asreml discussion list, send the message  
     subscribe
mailto:asreml-request@chiswick.anprod.CSIRO.au

To send messages to the list, mailto:asreml@chiswick.anprod.CSIRO.au

Asreml list archive: http://www.chiswick.anprod.csiro.au/lists/asreml

                        <> <> <> <> <> <> <>
"Christ Jesus came into the world to save sinners"   I Timothy 1:15.
<><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>


--
Asreml mailinglist archive: http://www.chiswick.anprod.csiro.au/lists/asreml