But no one seems to spend time on trying to figure out how to do it right or even how well the current methods perform.

Studying people is much harder than doing math proofs or experimenting with chemicals.

I meant that corporations should be trying to figure this stuff out, not academics.

They have the same problem as academics do, only without the benefit of being able to consistently apply the scientific method and good experimental practices (the only hope of cracking that nut).

The difference is they have lots of funding and a large monetary incentive to do the research.

Their development teams have funding to build products, rarely do they have funding/resources/expertise to do "meta" research on processes.

Obviously you wouldn't fund it through the development team budget, it would be a corporate level initiative.