Dear Megan:

I can tell you a bit about our experience with multiple users in different countries we have had.

We use one original set:

original.qes original.rec original.chk

Once this is definitively agreed upon, we use Copy structure (including the check file) to get:

a.rec a.chk

b.rec b.chk

..

as many as we need and give one pair to each user (actually in duplicate for double entry). They enter the data at their leisure and then email the validated *.rec files. Because these files are so small, this can be done from even the slowest email connection (which matters to us as we work in some of the poorest countries).

In EpiData Analysis, we simply do:

close erase "abcd.rec" read "a.rec" append "b.rec" append "c.rec" append "d.rec" savedata "abcd.rec" close read "abcd.rec"

People in the periphery simply continue to enter data into a.rec or b.rec and send so and so often the updated file, and the coordinator runs the append.pgm again. We did this with 30 locations in Zimbabwe collaborating with 31 locations in Mongolia, with a similar set-up for Moldowa collaborating with Uganda, and we never had a need for a central server. Email did just fine. A file with 58,000 records and and 9 fields took just 2 MB.

I hope this helps a bit with that part of your question.

Regards,

Hans Rieder