Monday, August 7, 2023

Very Brief Quote: "Statistics and Common Sense" (1814-2022)

Despite a century of the "statisticization" of medical research, we still have stiff debates about what is proven or not, ranging from FDA Advisory Committees to Medicare's recent concerns within its "TCET Transitional Coverage for Emerging Technologies."

I've always loved a quote that comes out something like:  "Statistics is common sense applied to numbers" - meaning, without the common sense, statistics isn't really anything.   This is sometimes attributed to the mid-twentieth-century British clinical scientist Bradford Hill, but here are connections back to 1897 and 1814.

I ran across the same idea attributed much earlier to WEB Du Bois, in the new book "How Data Happened: Age of Reason to Age of Algorithms," Wiggins & Jones.   The quote here is,

"[An author] has by no means avoided many fallacies of the statistical method.  This method is, after all, nothing but the appplication of logic to counting, and no amount of counting will justify a departure from the rules of correct reasoning."  [In Wiggins, Ch 4; Du Bois, Annals Amer Acad Polit & Soc Sci 9:129, 1897]. 

In a new paper, Hanaki et al., 2022, in their essay "Statistics and Common Sense," trace this idea back to Laplace (1814, Essai philosophique sur les probabilites):  "Probability theory is at its base nothing but common sense, reduced to calculus."