Title: analyzing the ex9.30step1 data
which has missing data.
Not using Tinterval
analyzing/saving not missing observations
by commenting out missing= and using Useobs
and not using Lagged so that y&1 is not saved
Data:
file = ex9.30step1.dat;
Variable:
names = u z y w time subject;
! don't read y-lag because then all records have missing
usev = z y w;
! missing = all(999);
cluster = subject;
between = z w;
! lagged = y(1);
! Tinterval = time(1);
useobs = y ne 999;
auxiliary = time;
ANALYSIS:
TYPE = twolevel RANDOM;
estimator=bayes;
proc=2;
biter=(1000);
MODEL: ! just apply a simple model:
%WITHIN%
y;
%BETWEEN%
y on w;
z on y;
Output:
tech1 tech8;
Plot:
type = plot3;
Savedata: ! this will be the "real data":
file = ex9.30.dat;