[sb2nov]

Can the exam be the confounder here, where the exam is just not good at differentiating the different in educational experience between the universities.

Second the job interview and exam might just be looking for skills on different dimensions ?

[zwaps]

Yes that is a possibility in this sort of research design. The exit-test scores are obviously taken as proxy for educational quality and the study cannot test for any "secret sauce" beyond that.

On the other hand, there's only so much one can do with observational data doing these sort of causal analyses. I'd say its still an interesting outcome.