My basic problem with interviewing is you are observing behavior in one sort of situation, and on that basis trying to predict behavior in a very different sort

My basic problem with interviewing is you are observing behavior in one sort of situation, and on that basis trying to predict behavior in a very different sort of situation, namely job performance, which is actually a whole bundle of different types of situations. It seems like it would be much better to instead put the job prospect in situations that model the sorts that would come up at work.