One wonders what Professor Seligman’s motivation was for asking 32 year old associate professor of business statistics whether statistics was a “separate science” or a “method”, but since we only have Frederick C. Mills’ response, it is anyone’s speculation for what Seligman was after. I find the letter interesting because Mills names three contemporary textbooks he regarded as “standard” at that time.
Fun fact: At age 22 Frederick C. Mills performed undercover fieldwork to study hobos in the wild. His reports and field notes were compiled, edited, and published as Gregory R. Woirol compiled and edited them as In the Floating Army: F. C. Mills on Itinerant Life in California, 1914 (University of Illinois Press, 1992). And what did you do for summer vacation?
_____________________
Meet the correspondents
Frederick C. Mills (Columbia Ph.D. Dissertation, Contemporary Theories of Unemployment and of Unemployment Relief, published 1917)
Edwin R. A. Seligman (from a 1939 memorial) (and about the same man four decades earlier, 1899)
_____________________
Columbia University
in the City of New York
SCHOOL OF BUSINESS
492 Engle Street,
Englewood, N.J.
Dec. 12, 1924.
Professor Edwin R. A. Seligman,
Kent Hall,
Columbia University, N.Y.
Dear Professor Seligman,
You have requested me to send you a list of references which might serve to show that Statistics is a legitimate science, rather than a method. On thinking the matter over at some length, I am afraid that I can give you no reference which might prove your point. For indicating the content of the usual courses in Statistics any of the standard text-books would serve. I might mention
Yule, G. Udney. An Introduction to the Theory of Statistics.
Kelley, Truman L. Statistical Method.
Bowley, A.L. Elements of Statistics.
In each of these, more evidence might be found for the view that Statistics is essentially a method than for the view that it is a separate science. I am inclined, myself, to consider Statistics as a method, at least in its practical bearing upon our own particular problems. Sciences, I suppose, are differentiated from each other primarily with respect to subject matter. From this viewpoint, the body of Statistics is concerned with methods of dealing with material drawn from many different fields of knowledge, the chief requirement for statistical treatment being that the data are quantitative in character.
Is it not possible to admit as much, and still have a strong plea for adequate treatment of statistical economics? The statistical method is one method, not the only one, by any means, for handling the data of economics and for investigating certain types of relationships. It is a method the application of which calls for a certain type of technical training, and instruction in this technique is essential to a well-rounded curriculum in economics. Statistics is at least a distinct discipline, and as such calls for specialized training. On such grounds, I think, a strong argument for a statistical economist might be based.
Faithfully yours,
[signed]
Frederick C. Mills.
Source: Columbia University Libraries, Manuscript Collections. Seligman, Edwin Robert Anderson Collection, Box 37 (Box 100: Columbia 1924-1930).