"The fact that an opinion has been widely held is no evidence whatever that it is not utterly absurd" (Bertrand Russell)
The real problem for this subject is the way the vast majority of scientists chose to present themselves to society at large. Some of them even believe their own message, though start picking at the seams over dinner and many (often begrudgingly) agree that "yes, it's complicated". The problem is far more evident with the Social "Sciences" but the same issue affects all "Sciences".
Science is almost exclusively presented to the lay-person as a process of discovering the "True Nature Of The Universe" - what scientists *actually* do is create models. Models can be very useful indeed, like when they enable us to make and do wonderful things, like create computer processors, split the atom, cure certain diseases or go to the moon. When the models are for massively complex systems like humans, human interaction, or the climate, it tends to be easy to find counter-examples. And according to the pillars of the Western scientific tradition, when we find phenomena that sit outside the model, the only honest thing to do is to reject the hypothesis. But after you've spent a considerable amount of your professional career, often decades, working on something, it can be tempting to call into question certain measurements, and maybe even exclude them from your data. After all, you'll often have an "explanation" for why so-and-so's team's data is not applicable, and should be excluded. If everyone else is doing it and for you to get funding for the next 4 years, you must, well, you must. Particularly because Mr Politician doesn't want to pay for "half-cocked theories that are right half the time". He wants The Truth, dammit!
The process of creating models is mainly a process of simplifying a particular phenomenon down to "the essentials", to let us accurately predict what is going to happen in a similar situation to the ones observed while the model is elaborated. It enables us to do all manner of useful things. The problems come when the future situations we want to predict (or "control" if you prefer) have many other influences that make our ability to accurately predict, within the confines of the model, impossible. Sometimes modelling "the real world" is hard...
So what? Should we stop funding science and simply go back to consulting the Good Book? That's one option... Not one I'm particularly enamoured with but why not. Or we could stop believing there is even a True Nature Of The Universe in the first place, and just get on with making models that are useful.
"Essentially, all models are wrong, but some are useful" (George Box)