Based on the U.S. Department of Education's description of the sampling process[1], it seems like the representativeness is dependent on how the school framing

Based on the U.S. Department of Education's description of the sampling process[1], it seems like the representativeness is dependent on how the school framing is done. It's not clear from that description what happens with individual students at the selected schools. You could still get distortions if groups of students within the selected schools are over or under represented. [1] http://nces.ed.gov/surveys/pisa/faq.asp#4 http://nces.ed.gov/surveys/pisa/faq.asp#4