Data constraints, bias and the (mis-)use of scientometrics for predicting academic success: A comment on van Dijk et al.