As well as deriving the long format from the wide format in SPSS, we can also generate the long format directly. First, save temporary files, one for each wave, creating a wave variable and removing the wave prefix from the variables of interest:
get file "s:\bhps\spss\aindresp.sav"/keep=pid, ajbstat. compute wave=1. rename var ajbstat=jbstat. save out = "c:\temp\tempa.sav". get file "s:\bhps\spss\bindresp.sav"/keep=pid, bjbstat. compute wave=2. rename var bjbstat=jbstat. save out = "c:\temp\tempb.sav". . . . get file "s:\bhps\spss\kindresp.sav"/keep=pid, kjbstat. compute wave=11. rename var kjbstat=jbstat. save out = "c:\temp\tempk.sav".
Then simply add
the temporary files together, sorting by
pid and wave:
add files file = "c:\temp\tempa.sav" /file = "c:\temp\tempb.sav" /file = "c:\temp\tempc.sav" /file = "c:\temp\tempd.sav" /file = "c:\temp\tempe.sav" /file = "c:\temp\tempf.sav" /file = "c:\temp\tempg.sav" /file = "c:\temp\temph.sav" /file = "c:\temp\tempj.sav" /file = "c:\temp\tempk.sav". sort cases by pid, wave.
One important difference: this format has records only for those waves for which the individual was in the sample.