The measurement scale of exam scores is ordinal. This ordinal measurement implies that monotonic transformations of published scales convey the same information. Hence, countries should not be ranked according to averaged scores because there may be transformations that would change the mean-based rankings. We suggest alternatives to the mean-based ranking procedure that yield informative and robust rankings. An implementation of these procedures, using PISA 2012 math scores, suggests that previous mean-based country rankings are sensitive to the aforementioned transformations.