If by "collecting data" you meant conducting a controlled experiment that can potentially falsify the picture of reality, then I'm with you. But to the extent that it might mean gathering existing data I must disagree.
In the latter case forming a hypothesis actually does put this in reverse. At that time the scientist comes up with an idea based at least in part on existing data as well as previous theory.
Approaching this from another angle: the scientist certainly can not and does not test his entire picture of reality, but only a specific subset of it. Being a mortal, there are some things he simply presumes and does not test.
All concepts begin with sensation.