ASReml prediction problem

Dear ASRemlers,

Given two groups, n1 (x,y) pairs in one group, n2 in the other, is this
the way to predict the difference between regression lines? Is there a
more straightforward way? Why does ASReml think these predictions are
non-estimable? (I want to generalize to higher degree polynomials, hence
use of pol).

 group 2
 x
 y
test.dat
y ~ p(x,1) group.p(x,1)
predict line 2 p(x,1) !onlyuse group.p(x,1) !printall

test.dat looks like this:

1 1 10.01
1 2 10.8
1 3 14.5
1 4 14.6
1 5 16.58
1 6 15.25
1 7 17.89
1 8 18.97
1 9 18.65
1 10 19.95
2 1 2.39
2 2 5.72
2 3 4.16
2 4 7.24
2 5 9.28
2 6 10.5
2 7 10.67
2 8 14.27
2 9 16
2 10 15.56

======================================
I.White
University of Edinburgh
Ashworth Laboratories, West Mains Road
Edinburgh EH9 3JT
Tel: 0131 650 5490 Fax: 0131 650 6564
E-mail: iwhite_at_staffmail.ed.ac.uk
======================================
Received on Tue Feb 09 2005 - 10:32:11 EST