Date: Thu, 8 Jan 2009 05:09:05 -0800
Reply-To: Jonathan Harris <jonathan.harris8@GMAIL.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Jonathan Harris <jonathan.harris8@GMAIL.COM>
Subject: help with justification for programming standards
Content-Type: text/plain; charset=windows-1252
I work in an organization with 12 statisticians, half of whom do the
bulk of the SAS programming; we have no dedicated SAS programmers. Our
work is NIH-funded, but we do contract with drug companies. We do not,
however, do any submissions to FDA. Currently we have little to no
programming standards. I'm trying to get some very basic standards
implemented but I'm getting pushback from some colleagues in a few
areas. There are a couple issues that I find particularly troubling:
(1) One person puts everything into one program: data cleaning,
dataset creation, and analysis. I know that this is bad practice but
I’m having trouble articulating why this is so. I find it to be self-
evident. Clearly there are read/write access issues, danger of
overwriting datasets, plus the need to constantly comment out
sections; what else?
(2) One person does not use permanent format catalogues. To me,
formats and labels are what turn codes into useable information.
I’ve done a lot of research here and other places online to put
together my standards document but I’m having a little bit of trouble
with the “why”. Any help would be appreciated.