STEM Dopers, how Do you View Social Sciences?

[Dr.Strangelove]

@Sam_Stone

It is entirely possible to do good science in all of those fields.

I’m not so sure. What if some of these problems are just too hard?

A significant chunk of experimental physics is done with blinded data analysis. For example, the results from a particle accelerator might be offset or scaled in a way such that the people performing the analysis can’t know if a spike at so many electron volts is the one they were hoping to see (based on theory), an outlier, or something else entirely. That way, they can’t have a bias toward their expectation, perhaps (subconsciously) massaging the data in a way that leads toward that result.

Blinding would have fixed the post-Millikan problems that Feynman described. Maybe it’s one of the tricks he had in mind when he wrote that passage.

At any rate–if even the hardest of the hard sciences have to use state-of-the-art anti-self-delusion tricks like these to get reliable results, what hope do the fuzzier sciences have? There are only so many of these tricks available. What if it still isn’t good enough?

[Chronos]

@Dr.Strangelove

It takes only a single contradictory experiment to invalidate a theory, so the ones that persist are highly resilient.

With the caveat that if a single experiment does seem to invalidate a well-accepted theory, you want to make darned sure that your experiment is working correctly. Every so often you’ll read reports of CERN generating a particle beam that appears to move faster than light, or the like, and it generates a lot of buzz for a while, until someone finds the flaw in the experiment that lead to the erroneous measurement.