I would recommend that, as a general practice, you think in terms
of writing a file of commands that will do each task, so that you
can reproduce it once you have finished. It is useful to test each
bit of the file interactively as you are writing it. The SPSS
syntax window can be used to write the files, or Windows Notepad or
any other editor (e.g., Winedt, Programmers' File Editor, Emacs). If
you have a file which carries out a task (data preparation or
analysis) you automatically have a record of how the task is done,
and you can reproduce or adapt it easily.
© Brendan T. Halpin (e-mail), GNU Free Documentation Licence