Hi everyone
I have a data set that consists of quality attributes for wheat from
9 countries. There are 2 years of data. I have used the following
model :
y ~ mu c(year) c(country) !r line year.country year.line !f mv
The blups in the .sln file for line are extremely small, absolute
values of the magnitude 0.1 e-5 to 0.1 e-19. So when I calculate
the predicted means for lines the values are the same for lines within
a country.
Out of curiousity I tried
y ~ mu c(year) c(country) !r c(line) year.country year.line !f mv
and the blups look more realistic, in the e-1, e-3 range.
I'd like to know why these two models behave as they do.
Thanks for your time
Carole