>Perhaps some of you can lend a hand with the problem I keep running up against. I am sure some of you must work in corporate environments and had this happen to you.
>
>
>The problem:
>Aftering send the reports, our clerical and medical staff can go back and alter data that was entered incorrectly, missed or for some other reason. At a later point Finance was come back to use and ask for much more detailed data (like race, city and other). Inevitable the numbers don't match and they get their nose out of joint. Finally, this has come to a head and they are jumping up and down and demand and answer why the numbers don't match. The Finance dept has powerful friends.
>
>A possible solution:
>- make it more difficult to change historical data after accounting periods. In fact they will need to come to me. I will then resend any reports to Finance
>- put a date stamp on records to tell if they have been modified after their original creation date.
>
>Suggestions:
>What do others do in cases like this? I am more looking for a concept more than a technical solution. It is sort of like we want to "close the books" then have an audit trail to track any changes.
>We certainly don't want to discourage anyone for correcting data since clinically we want to know what "actually" happened.


I've run into the same sorts of problems. I developed a bunch of analysis reports that can be run on the estimates the techs write up. The same reports can be run on the actual sales. The problem we had is that some of the information could change, people leave fields blank, etc. That means the reports wouldn't really match.

I didn't really come up with an elegant solution to this (sorry!). I ended putting a note on the report when the information was changed (and why). It stopped the questions, at least. Forcing people to not change the information didn't work very well for us - for exactly the reasons you mentioned. We want the correct information.

I kind of looked at the reports as though they were a Balance Sheet. A snapshot of a moment in time. If you run it later, the information changes. It sounds like they're not using the information that way. So, I guess technically they're using the information incorrectly - your program isn't wrong. In addition to the "changed" message, I added a small comment at the bottom of the reports explaining exactly what the information was saying and the types of things they could and couldn't infer from them.