c**********e
发帖数: 2007 | 1 Suppose we have 23 observations, with time t=1,2,...,23. The dataset has 3
variable, t (time), x (independent) and y (dependent). The dependent
variable y is missing for the 3 observations in the middle:
1 x1 y1
... ...
10 x10 y10
11 x11 .
12 x12 .
13 x13 .
14 x14 y14
... ...
23 x23 y23
I run a simple AR(1) model, y on x, i.e., the error term follows an AR(1).
Then I delete two observations with missing dependent y.
Should I get the same regression parameter? | F8
发帖数: 348 | 2 Generally, subjects with missing dependent variable will be deleted in the
mixed model, along with covariates, no matter what var-covar structure used |
|
