Also, as a colleague pointed out, once you are in the public arena (the professional one, or the more public one), it may be hard to admit the fudges and fixes you do on the data to get your results. At that point your claims are not scientific ones, per se, but either professional or public policy. In both arenas, you could admit your fudges and weaknesses, but then you are less likely or make a big splash. On the other hand, your splash won't come back and get you all wet.
So for example the recent astronomical observations of the polarization that would seem to give further evidence for the inflation scenario had some problematic issues re the background. It may have served the authors longer term interest to have indicated those problems and give a range of estimates of the effects they saw. But perhaps it would also have created less of a splash. Now they may have good reason for the background estimates they used, but others find those reasons convincing?