I guess it is only because I am an empiricist that I do.
I have, unfortunately, seen it happen, more than once.
The most egregious example I recall was the discovery of a protein (i.e., gene product) that was judged to be capable of inducing apoptosis (so called "programmed cell death" -- a very orderly and non-necrotic way for a cell to die
when it is supposed to). Apoptotic agents are/were attractive cancer therapeutics if they can, e.g., be targeted to neatly and predictably
"kill" cancer cells specifically. Novel gene, novel sequence, great IP position.
The "apoptosis inducing" protein was purified and shown to have activity based on an assay for the characteristic 200 bp DNA "laddering" that is one of the hallmarks of apoptotic DNA degradation. To cut a long, and very expensive story, short -- the active principle was not the purified gene product
per se, but a contaminating endonuclease. The whole endeavor was based on a poor choice of "gold standard" surrogate assay for apoptosis coupled with poorly designed (incompletely controlled) experiments, coupled with a whole lot of wishful thinking &, of course, good old fashioned greed.
EDIT: and, as a peer reviewer, I've read numerous submitted manuscripts that either misinterpreted or (more typically) overinterpreted data based on poor (I might say
wishful) experimental design.