Random bit of data from PISA, looking specifically at how immigrant students do in various countries: "immigrant children in some OECD countries lag more than two years behind their native counterparts in school performance". OECD concludes (my emphasis): "In some countries, such as Canada and Australia, immigrant children perform as well as their native counterparts. But in other countries, notably those with highly tracked education systems, they do substantially less well."

PISA reports consistently note that tracking is bad for kids - but look at what tracking means: "They tend to be directed to schools with lower performance expectations, often characterised by disadvantaged student intakes and, in some countries, disruptive class-room conditions. In all but four countries under review, at least 25% of second-generation immigrant children attend schools where immigrants make up more than 50% of the roll-call."

Which goes back to - the problem isn't the grouping, it's what we do to kids once they are grouped.

http://www.oecd.org/general/oecdeducationsystemsleavemanyimmigrantchildrenflounderingreportshows.htm