On 25/03/2015 17:25, Alan Mead wrote:
On 3/25/2015 10:00 AM, ftr wrote:
Frans,
I don't understand why you don't change the variable width for your
string variables. This seems the most easiest way for everyone.
Cheers,
ftr
If you have dozens of variables in dozens of files, it's not "easy" to
fix this problem.
I find this most often arises when I have used CSV or a Spreadsheet to
hold data for several different instances that later needs to be
integrated. For example, I've been involved in projects where data
was gathered each semester at a few schools. To later join that data
is a nightmare if there are many string variables.
Also, if you know that this is going to be an issue, you can do more
to avoid it. The annoying thing about this issue in SPSS (and PSPP) is
that you find out about the problem when the join fails. If you don't
know to avoid this, it can be maddening to repeatedly run join and
have to fix each issue individually.
-Alan
So this means that the programs that produce the CSV files produce
output with different string variable width ?
This is due to the programs or to the people that use the progs ?
In general, when you import text files you fix the variable width in the
DATA LIST.
Or you use GET DATA/TYPE=
http://www.gnu.org/software/pspp/manual/pspp.html#GET-DATA
And why don't you set FORMAT on each of the separate files before you
integrate them ?
When I worked in a project that sounds similar to yours we did a serious
pre-field work training of the local data producers that succeeded in
making the local projects aware what was on stake (motivation), that
made the local heads control the consistency of data to be sent -
something we could not do because we had no direct access to the local
projects, for which the local heads had better knowledge, and it would
have cost us too much (data control) - and that assured that data were
sent in a coherent format and at time.
Maybe you have to train your local people ?
Just some ideas for local problem solving. I am happy that we have
volunteers doing the programming work so we should not overcharge them
with more work that we can at our side.
Regards,
-ftr
_______________________________________________
Pspp-users mailing list
Pspp-users@gnu.org
https://lists.gnu.org/mailman/listinfo/pspp-users
