If data analysis is to be well done, much of it must be a matter of judgment, and "theory" whether statistical or non-statistical, will have to guide, not command.
Annals of Mathematical Statistics, The Future of Data Analysis, Volume 33, Number 1, March 1962 (p. 10)