The elusiveness of representativeness in general population surveys for alcohol: Commentary on Rehm et al

For alcohol researchers, it is increasingly difficult to mount a rational argument for stoically pressing on with well-worn probabilistic survey approaches and analysis of survey data as if representativeness remains assured, particularly when negative consequences for science, society and individuals are foreseeable. Some of the most striking examples of biases in action come from large surveys of marginalised populations that use approaches presupposed to be gold standard. Remarkably ill-suited to their populations of interest, the conspicuous failings of these surveys have, for many years, demanded we find better ways. Work in this area has begun to demonstrate the utility of alternative approaches and solutions that, more likely than not, are transferable to general population surveys.