Re: Contrast

Dear Seyed,

Iknow nothing about SAS

Your request seems very odd to me in that your estimate of X1 and X2
double counts the 'intercept' and the 'YS' effects with different weights.

Given the YS by Line means are
 Line YS Predicted_Value Standard_Error Ecode
       1.0000 1.0000 5.6000 0.6060 E
       1.0000 2.0000 7.5000 0.6775 E
       1.0000 3.0000 8.0000 1.3550 E
       2.0000 1.0000 6.6667 0.7823 E
       2.0000 2.0000 6.3333 0.7823 E
       2.0000 3.0000 4.5000 0.9582 E

Its easy to calcul;aten the Line means from this and the SED

viz
 Line Predicted_Value Standard_Error Ecode
       1.0000 7.0333 0.5439 E
       2.0000 5.8333 0.4879 E

 Predicted values with SED(PV)
    7.03333
    5.83333 0.730638

But you are adding to this difference, the unweighted (by YS) differences
between lines (6.6 and 6.0) to get
  (6.6 + 3x7.0333) = 27.7 and (6.0 + 3x5.83333)=23.5
to get a difference of 4.2

So you have a weighted sum of line differences from two different models,
and ASReml cannot easily compute the SE of that difference.

The .as file I used is

SED
 Line *
 YS *
 Animal *
 TR
line.asd !skip 1
TR ~ Line.YS
predict Line !SED
predict Line YS !SED

with data file
 Line YS Animal TR
1 1 1 5
1 1 1 6
1 1 2 5
1 2 2 8
1 2 3 7
1 2 3 9
1 3 3 8
1 1 4 7
1 2 4 6
1 1 5 5
2 1 6 6
2 2 6 5
2 3 7 4
2 1 7 9
2 2 8 8
2 1 8 5
2 2 9 6
2 3 10 5

------------------------
Arthur Gilmour

Retired Principal Research Scientist (Biometrics)

-------------------- m2f --------------------

Sent using Mail2Forum (http://www.mail2forum.com).

Read this topic online here:
http://www.vsni.co.uk/forum/viewtopic.php?p=1009#1009

-------------------- m2f --------------------
Received on Wed Nov 10 2009 - 22:56:02 EST