School doesn't make much of a difference beyond the abre minimum.

From It doesn't matter whether refugees are in the same classroom by Emil O. W. Kirkegaard.  

Reporting on the results from three separate international studies on individual performance in different types of classroom environment.  For those opposed to high levels of immigration, one of the arguments is that high percentages of immigrant students (especially if illegal or refugee) degrade the education experience for native born citizens, reducing the quality of their educational outcomes.

From the main study, Do refugee children impair the academic performance of native children in the school? Informative null results from Danish Register Data by S. Hassan, C. Hvidtfeldt, L.H. Andersen,  & R.O. Udsen.

Our findings show that refugees tend to cluster in schools that had poorer performance even prior to the refugees’ arrival. When we take this selection pattern into account, the effect of receiving refugees on the academic performance trajectory of natives is both statistically insignificant and substantially unimportant.

It invites the obvious follow-up question.  If a similar proportion of refugees entered classes in a high performing school, would we expect to see a similar null effect?  It doesn't seem likely.  I have to wonder whether their might legitimate but irrelevant.  In other words, if the refugee children are placed in a dysfunctional school with children from dysfunctional families, is it all that surprising not to see much change in overall (already bad) performance?  

Kirkegaard is Danish and deeply familiar with their scientific literature and seems to hold this study in high regard.  I accept that it is likely that there was a null-result in the conditions actually experienced.

I am much less confident that there would be no effect were the same percent of troubled refugee children introduced into highly functioning schools with highly functioning children.

Kirkegaard comes with a particular perspective based on his past work:

Many people will claim that teachers or peers or other classroom-related factors like pupils per teacher are important. If these are important, we expect that twins who are in the same classroom or go to the same school are more similar than those who don't. It turns out they are no different. This result isn't based on small samples and lack of precision, but on the combined data of 13,000+ twins from 6 countries. They looked at not just school performance, but also various non-cognitive or personality traits like desire to learn.

I suspect he is correct about the larger picture.  It has been quite obvious in the US for at least two decades that school performance is overwhelmingly the consequence of the cognitive quality of the student body.  Take high cog students from stable high-functioning families (rich or poor, native or immigrant) and you will get great academic results independent of class size, teacher credentials, budgets, etc.  The stuff we want to pour money into doesn't make a difference.  High quality input gives you high quality output.

And correspondingly, low cognitive capability kids with fragile or dysfunctional home lives just aren't going to do particularly well, regardless of how you spend on which initiatives.  It doesn't make a difference.  At the population level.  There is always the random individual student who is beneficially affected.  As is the random individual student who is negatively affected.

Teacher's unions don't like this message because it suggests teacher qualification doesn't matter.  In a multicultural environment, there are further challenges when different groups perform differentially, even if they are of the same race (but different ethnicity).  

Eventually we will get around to focusing on helping individual students become all that they can be even if what they can be is more limited than others.  Much better to have incremental real improvements than chasing an unrealizable dream.