A school psychologist wants to study the impact of a new teaching method on student engagement. They decide to use random sampling by assigning a number to every student in the school and then using a random number generator to select 100 students. What is a potential limitation of this approach for generalisability?
