Prasanta -

It sounds like that depends upon what you are trying to estimate, and to whom you want it to apply.  Would you want to know the result unconditionally, or for it to be conditional upon being a user?  Perhaps you might also consider whether or not your independent variables make sense outside of the user population. 

Cheers - Jim

Hello,

I do not know what you are looking at so this is only a suggestion: non users could be considered as a control group.

Best Wishes,

Lydie

Both the ways are OK

Leonid

Thanks Lydie a comparative study can also be done i guess..

Thanks for your views Prof.James.