I will stop believing that what is contradicted by scientific data is in fact wrong as soon as someone shows me just one killer counterexample, that is, where the truth is exactly what carefully checked and cross-correlated scientific data are contradicting.
If I mistakenly believe that A*A + B*B = 2*C*C, all I have to do is draw one reasonably accurate right triangle in a plane to disprove it, and by drawing just three right triangles on planar surfaces, and modeling the data, I would inevitably come up with Pythagoras’ theorem, which is easily provable mathematically, especially from the law of cosines. The beauty of the empirical approach is that I do not have to understand logic or mathematical proof or even deductive reasoning. I just have to make experiments and model the results, in this case with a mathematical formula. This empirical procedure is universally applicable, and while mathematical proof may be applicable to all problems, I have yet to realize that it is. Plato believed that mathematical proof is indeed universally applicable, and maybe he was just smart enough to see it. I am not that smart, but am just smart enough to see that I do not have to rely on mathematical proof to know at least when something is just plain wrong, as most assertions unfortunately are.
Now interpretations that scientists make – I dare not call them “scientific interpretations” because they are often dead wrong- are a different matter entirely. Philosophically shallow scientists (e.g. ones who believe in causation) screw up interpretations as a matter of course. Scientists with political agendas royally screw up interpretations that are clearly contradicted by the data that they have so carefully filtered out.
After subtracting the many bogus interpretations from the body of science, what remains are cross-checked, reliable scientific data (often buried in the “literature” of less read journals), the theories/models that the reliable data flat out contradict, and a handful of tentatively believable scientific models that are consistent with the reliable data. The former theories/models are wrong and have already been proven to be inadequate by the reliable data.
For example, I see so much data contradicting religious assertions, particularly the most comforting ones, like the ideas of a “chosen people” or the ideas that a god who loves us created everything. If I am simply misinterpreting religious assertions, then perhaps with the correct interpretations, I would see that all data actually affirm religious assertions.