The existence of "experiments" to choose from in the first place is already theory-given. As soon as you've formulated a space of such experiments to explore, almost all your theory work is done.

What's more, the existence of data (therefore differentiation of what is and isn't), is theory-laden.