I find the title of this thread to be a bit unsettling. If science is becoming dogmatic, it's no longer science. The scientific method does not always result in a rule or law like Newton's laws of motion. We have the jump from hypothesis to theory when it is impossible to arrive at a rule but there is insurmountable evidence. That's an area where things could get a bit fuzzy. We also have the dismal science of economics where entirely different schools of thought exist. Building most models involves choices of assumption which is another area where someone's world view can influence the result. Perhaps in the fuzzy areas is where dogma comes in.
I think a better way to put it is that many people (creationists, some audiophile, psi advocates, etc) are deeply offended by the fact they don't get a hearing, even though their claims have been tested, and massive quantities of evidence AGAINST their theory are available. A common example is those who demand **ABSOLUTE PROOF** that their notion (for example I'll use "green cheese on the moon") is absolutely 100% false.
There is no such "absolute proof", especially of a universal negative, in science or in logic. Such demands are simply a way to try to continue the argument. Accepted scientific theories are the 'best known explanation', and there are always limits stated. There ***is*** no absolute. There is always an error or probability. ALWAYS. Direct contradictory evidence ALWAYS forces a revision.
See Wegener, or Cuvier (in regard to the Grand Coulee, not everything) for good examples of how a theory that wasn't popular became the dominant theory. What changing scientific viewpoints requires is EVIDENCE. Yes, there will be push back, resentment, and resistance, but evidence wins, eventually.
As a personal example, someone (a leading person in the AES) told me, and I quote, "Nobody will ever accept 4 bit quality" when I first mentioned the idea of perceptual coding. Said that, in fact, in a meeting publicly, to substantial laughter from the crowd. Standard MP3, of course, is 2.9 bits/ sample.
But evidence wins out.