In this paper I study Milton Friedman's statistical education, paying special attention to the different methodological approaches (Fisher, Neyman and Savage) to which he was exposed. I contend that these statistical procedures involved different views as to the evaluation of statistical predictions. In this light, the thesis defended in Friedman's 1953 methodological essay appears substantially ungrounded.
ACKNOWLEDGEMENTS:Thorough comments by D. Gillies, D. Hammond, K. Hoover, U. Mäki, S. Stigler, J. Urrutia and J. Zamora are gratefully acknowledged. This research was funded by the grant BF2003-04372 and the Urrutia Elejalde Foundation. The usual disclaimer on intellectual responsibility applies.