Karen Anderson wrote:

> I've had the most horrible luck you can imagine with "interested and
enthusiastic" >work study students and data entry projects,

indicating that an original file was fouled up beyond belief by work study
students.

        A way around this is to never let anyone except trained staff near
an original file.  For new data entry, duplicate the structure and have the
new information entered in the duplicate and, after proofing, a trained
staff member can append it to the original.  Similarly for corrections to
the originial file:  duplicate, proof, and replace.
        There are too many things that can go wrong with someone not
thoroughly conversant with an application (visualize an attempt to replace
data in a field of a subset, but forgetting to segregate the subset first--X
number of fields with incorrect data and no easy way to retrieve the
original data).
        Which brings up the other absolute requirement in data management:
BACKUP FREQUENTLY!

Arthur H. Harris
Laboratory for Environmental Biology
Centennial Museum
Univ. Texas at El Paso, El Paso, TX  79968-0519
Phone (915) 747-6985
Fax (915) 747-5808
[log in to unmask]