Locking a dataset in a stat program


Ninja say what!?!
I'm hoping one of the brilliant minds here will be able to help me with this.

First some background
I'm in the Epi / Biostat field and currently in grad school. My interests currently lie in intervention trials. Recently, we covered different aspects of randomizing.

My question
Obviously, to add more credibility to clinical trials, you want to blind as many of the parties as possible. So I was thinking, I can get SAS to assign treatment randomly. However, I won't be blind to it unless I can hide the randomization process calculations from myself. What I'm wondering is if there's a way for STAT software to hide or lock the results from the randomization calculations until a specified date or until some conditions are satisfied.


Ambassador to the humans
What exactly do you mean? I mean I guess you could get the data and then you could blind yourself by having a function relabel the treatments and you could output what it randomized things to and from into a file without ever seeing what changed to what. In essence you would no longer know which treatment is which when you're running the analysis.

I have a feeling this isn't what you mean at all though. Would you care to elaborate a little more?


Super Moderator
Just a thought Link:

could YOU run the randomisations, hand them to a collegue who then randomises the treatments back to you? (esstentailly double blinding the trails).


Ninja say what!?!
Thanks for the results guys. What I had in mind is a way to randomize without knowledge of the results, but to do it without the need of another persons assistance.

I suppose I can have the results printed and stored without looking at them. However, what if I was doing it for someone else? It would be nice if the results were exported to a data set or some other file and locked until a certain date, time, or certain conditions were met.


Super Moderator
If you are using R, could you have the results automatically export to a locked folder on your desktop, or wherever?