Anscombe's Quartet: The Four Datasets Built to Fool the Printout

Introduces Francis Anscombe -- Cambridge-trained, a WWII Ministry of Supply calculator, Rothamsted alumnus, founding chair of Yale's Statistics Department, and John Tukey's brother-in-law -- and his 1973 paper 'Graphs in Statistical Analysis' (The American Statistician, Vol. 27, No. 1, Feb. 1973, pp. 17-21), correcting the common misremembering of it as a 'mid-1970s' paper: it appeared at the front edge of the decade, four years before Tukey's Exploratory Data Analysis. Walks through the four datasets Anscombe built -- each sharing, to two or three decimal places, the same n, mean, variance, correlation, regression line, and R^2 (independently recomputed here, not just quoted) -- and what each panel actually demonstrates: a genuine linear fit, a mismatched nonlinear relationship, a clean line wrecked by one outlier, and a correlation manufactured entirely by a single high-leverage point. Closes with the quartet's legacy (Chatterjee & Firat's 2007 general construction method, the 2017 Datasaurus Dozen) and why the warning about trusting summary statistics over a plot applies with even more force to automated dashboards and ML pipelines than it did to 1973 regression printouts.