Amongst scientists, there is a gaping chasm that splits the experimentalists from the theoreticians. Of course there are big resentments between the two - experimentalists are better funded, whilst theoreticians are more acclaimed. There is one burning question, that as a card-carrying theoretician, I ask myself all the time about crossing the chasm: when constructing my theories, how much should I pay attention to experimental facts?
To answer that question, here's two quotations by Albert Einstein, the greatest theoretical physicist of the 20th century. The first quotation, dated 1920, has Einstein admitting that his Theory of General Relativity is only as good as the possible experimental fit:
"If the observations of the red shift in the spectra of massive stars don't come out quantatively in accordance with the principles of general relativity, then my theory will be dust and ashes."
~ Einstein, 1920
The second quote is a much later exchange, in 1952, between Max Born and Albert Einstein. Once again, the subject is Einstein's Theory of General Relativity, where Born is alerting Einstein of an experimental refutation of The Theory of General Relativity:
Born to Einstein: "It really looks as if your formula is not quite correct. It looks even worse in the case of the red shift; this is much smaller than the theoretical value towards the centre of the sun's disk, and much larger at the edges. Could this be a hint of non-linearity?"
Einstein to Born: "Freundlich does not move me in the slightest. Even if the deflection of light, the perihelial movement or line shift were unknown, the gravitation equations would still be convincing because they avoid the inertial syststem. It is really strange that human beings are normally deaf to the strongest arguments while they are always inclined to overestimate measuring accuracies."
~ private correspondence, 1952
In 1952, 30 years after the first quote, Einstein simply dismisses the experimental result. He even admonishes Max Born for not treating his theory seriously enough.