Reporting
By
comparing with the earlier model without random effects for ram and ewe
it can be seen that the standard errors for year and sex have changed
little. These are fixed effects associated with the lowest layer in the
fitted model.
With the reduction in residual variance one might have
expected standard errors to be reduced too. This will often be so.
However, there is considerable imbalance in the way that ewes are
distributed among genotype groups and years, as illustrated in Case Study 3, and this is probably reflected in the calculation of
the standard errors.
|
|
|
Variable |
No.
|
Meana
|
Meanb
|
|
Ram
breed |
|
|
|
Dorper |
433
|
10.95
|
11.05
|
Red
Maasai |
439
|
10.54
|
10.61
|
S.E.D. |
|
0.18
|
0.17
|
Ewe
breed |
|
|
|
Dorper |
544
|
10.97
|
11.12
|
Red
Maasai |
338
|
10.51
|
10.54
|
S.E.D |
|
0.27
|
0.24
|
Year |
|
|
|
1991 |
144
|
12.64
|
12.67
|
1992 |
109
|
11.07
|
11.11
|
1993 |
168
|
11.56
|
11.58
|
1994 |
79
|
9.64
|
9.84
|
1995 |
107
|
9.35
|
9.45
|
1996 |
93
|
10.19
|
10.32
|
S.E.D. |
|
0.32
|
0.33
|
Sex |
|
|
|
Female |
323
|
10.54
|
10.59
|
Male |
377
|
10.94
|
11.07
|
S.E.D. |
|
0.16
|
0.17
|
|
a - without random effects
b - with random effects
|