I want to examine whether the students of School A have done significantly better than the students of School B in science subjects. In Math, School A does moderately better (p-value: 0.06) than school B. We observe similar and consistent results for physics (p-value:0.09) and chemistry (p-value: 0.15). Therefore, I suspect that school A is doing consistently and perhaps significantly better than school B in the science subjects collectively. How to get a combined p-value for that?

Please note: the results from physics, chemistry and math for each student cannot be directly added up (because of the range of score, grading etc) and them compared between the groups.

I know of Fisher's method for combined p-value or Stouffers method for combining Z-scores (with same or different weight), but please advice if those would be appropriate (or perhaps there is an established method for it).

Please note: one might expect that the performance in the subjects (e.g. physics and math) might be partly correlated, if that affects the choice of test.