In numbers we trust? Scientists go to great lengths to ensure that data are collected and analysed properly, so why, asks Eve Marder, do they apply different standards to data about the number of times research papers have been cited and viewed?