Bakker and Wicherts (2011) would like to disagree! Apparently 15 % screw up the calculation of the p-value.