I agree with Fraser and Reid that the many abuses of p-values in the real world arise not so much because applied researchers have ignored the lessons of statistics, but because in many ways they have learned the lessons of statistics all too well. Thus, to the extent that education from statisticians to practitioners is part of the solution, what is needed is not merely to shout existing messages even louder, or to require scientists to take more of the usual sort of statistics courses, but rather for we, the statistics profession, to think carefully about what messages we are sending and how these messages can be encouraging statistics abuse.