To analyse observation-per-record data we need to use cross-case functions. Let's say we want to tabulate jbstat at with jbstat at for all six transitions pooled. To do this we need to know for each individual, for waves from 2 to 11 what he/she was doing in waves 1 to 10.
get file 'm:\multilong.sav'. compute laststat=-1. * let it default to a special value. if wave>1 laststat = lag(jbstat). missing values all (-9 thru -1). cro laststat by jbstat.
We see many more cases than we would if we tabulated e.g., ajbstat
and bjbstat in m:\multi.sav
. However, we can still
make wave-specific tables:
temp. sel if wave=2. cro laststat by jbstat.
If you go back to m:\multi.sav
and tabulate ajbstat
and bjbstat you should see the same answers.