Reductionism can be powerful.
Through careful study, a component of a system can be identified and its role in the function of the system defined. Manipulation of that component can be shown to affect the system in a predictable way. It’s often possible to generalize- the heart is a pump in mice, cats and elephants. At a molecular level in the brain, the established role of CREB in aplysia neuronal function predicts role of CREB in mouse hippocampus. Human hippocampus? Well we can’t know because the experiments can’t be done, but the body of available evidence generates a strong belief that it does. The scientific fact that KREB is involved in human learning and memory
These scientific theories are not facts, they are statements with a probability of being true and a complementary probability of being false. Scientists recognize this at least implicitly because the method of science is to collect additional data that will either falsify or support the theory. This data will either increase or decrease belief in the truth of the theory. Sometimes theories are completely abandoned. Our earlier belief turns out to have been unwarrented.This is straightforward pragmatism.
It’s a major mistake to ignore the probability element in scientific theory. We don’t even need to consider how reliable the data really is, the nature of these complex systems makes their significance uncertain. The influence of one system component may vary in hard to predict ways because of changes in other components.
And because statements about complex systems are not true or false, we must make decisions based on belief, the probability of the statement being true. Since just about everything we deal with is a complex system, this impossibility of knowing the future is everywhere. Uncertainty can be found in the very structure of the world.
I know of no published paper with field data that does not have error bars. It seems many in the CAS domain have forgotten this.
I think that the problem is that we look at p<.05 as a standard of universal truth rather than a hypothesis about a particular data set collected under particular conditions. In complex systems like effects of drugs on the body, we don't get to truth with a clinical trials, just evidence that leads us to believe the drug is useful. The better the evidence, the stronger the belief. Replication strengthens belief, contrary or negative studies decreases belief. Of course with publication bias toward positive studies, so I start reading every clinical report with the jaundiced eye of one who wonders how many negative studies remain unpublished.