As someone working in assessments (behavioral, academic), the theory is based on decent enough science, I guess, re: what kind of data is needed to make a prediction of likelihood in that setting, but seems to be taken way too far in how it's extended here. I'm not sure 20 hours is close to...