@jsk well, it's not how I'd have answered it, but it's possibly a take on the question that might suit the OP. Did you have a specific objection you wanted me to understand?
I accept your concerns in the comments, but your concerns are already there for anyone to read.
@jsk I think a good option is to write a better answer.
@Glen_b The part that I object to is Michael claiming that subracting a constant will allow the OP to make his desired conclusion of setX exceeding setY by a particular %! I guess I wanted to make I wasn't the one misunderstanding something. Of course the OP will be happy if he is unknowingly misled! I'm not sure I have an answer at the moment that would allow the OP to make the desired conclusion.
@jsk You make good points in your comments. I found the one about the variance of the difference to be quite convincing. The entire question raises interesting problems: for instance, what can be said when the data means are negative or zero?
