The
main points to note when interpreting REML outputs are (a) the validity
of the Wald test depends on the size of the sample, and (b) that the
Wald test is more liberal than the F test, with the significance levels
of the two becoming closer as the sample size increases.
Some
statistical packages apply an F test to the Wald/d.f. value rather than
a Chi-square test to the Wald statistic. Nevertheless, the above
comments still apply and the user needs to take care in calculating
significance values for F-tests in a mixed model analysis.
|
|
*** Table of effects for YEAR ***
YEAR |
91.00 |
92.00 |
93.00 |
94.00 |
95.00 |
96.00 |
|
0.000 |
-1.566 |
-1.096 |
-2.833 |
-3.228 |
-2.351 |
Standard
error of differences: |
Average |
0.3373 |
|
Maximum |
0.3898 |
|
Minimum |
0.2753 |
Average
variance of differences: |
0.1151 |
***
Table of effects for SEX ***
|
Standard error of differences: 0.1695
** Table of effects for AGEWEAN ***
0.07022 Standard error: 0.008856
*** Table of effects for DL ***
2.726 Standard error: 0.3150
*** Table of effects for DQ ***
-0.2689 Standard error: 0.03401
** Table of effects for RAM_BRD ***
RAM_BRD D R
0.0000
-0.4429
Standard error of differences: 0.1728
** Table of effects for EWE_BRD ***
EWE_BRD D R
0.0000
-0.5855
Standard error of differences: 0.2366
|
|