That's definitely relevant to the question of why it's difficult to do such a study. However, it's not relevant to the question of whether such a study is necessary to make strong inferences about the population as a whole. The difficulty of making the right study does not change our ability to draw inferences from the wrong study. (We can't.)