"how does someone in her position know they're right and the research is wrong, and not the reverse?"

This is where you pull out the rusty ol' tools of Figuring Things Out From Evidence, which mostly boil down to:

come up with your own theories and listen to theories from others

think really hard about which one explains historical evidence the most consistently.

think about experiments to make sure theories aren't overfit to historical evidence

conduct those experiments, and "penalize" theories which don't explain the experimental results

these tools are strange things because they (noisily, but surprisingly reliably when applied truly honestly and openly) predict which theories are predictive of future experimental results.

and that's all you want out of a theory, really.