Carl Snyder b #OTD 23 Apr 1869 (d 15 Feb 1946) ASA Fellow 1923, ASA President 1928. Beginning as a journalist & science writer he somehow ended up as Chief Statistician for the Federal Reserve Bank, & later an economic policy leader, especially for his work on long-term economic growth & capitalism

Also #OTD 22 Apr 1916 Lee Cronbach b (d 1 Oct 2001) ASA Fellow 1956, Fellow AAAS; known for Cronbach's α, he did enormously influential work on measurement theory, test performance quantification, program evaluation, & concepts of validity (which have extended far beyond psychology)

5/ This may have been related to Eddington’s claim that the best estimator of the standard error of a normal distribution was the mean of the absolute deviations, which Fisher disproved in 1920.

4/ Apparently what Eddington had to say about inference ‘horrified’ them both, so they “promised not to write any more rude things about each other”.

3/ To the day of his death, Jeffreys strongly opposed theory of continental drift. Fisher said “He was wrong about probability & that makes me sure he is wrong about continental drift.” However, Joan Box has it that both Fisher & Jeffreys reconciled over a disagreement with Sir Arthur Eddington.

2/ David Finney adds that it didn't help that Jeffreys was a terrible lecturer who couldn’t make ideas clear or keep students. Fisher flamed his book The Theory of Probability: “He makes a logical mistake on the first page [i.e Bayes’ postulates] which invalidates all the 395 formulae in his book”.

#OTD 1891 Sir Harold Jeffreys b (d 18 Mar 1989) Guy Gold Medal 1962. A geophysicist known for his ideas on Bayesian theories of probability and Jeffrey’s prior. Bayesian statistics were overwhelmed by the dominance of the Fisher-Neyman-Pearson school (which he opposed) 1/5🧵

4/
8. Statistics seen as “an honourable and populous profession"
9. "The man in the train who begins his remarks with 'Of course, I am not a statistician' does so with a sense of humility and not, as at present, with an air of quiet satisfaction at freedom from a serious mental disfigurement” 🤣

3/
4. Royal Statistical Society should increase its role to provide advice to government entities
5. National adoption of the metric system
6. Statistics will be taught in schools
7. Every sizeable business firm will include a statistician among its retinue

1. Statisticians will take over every branch of science "like Colorado potato beetles".
2. There will be more “new types” of statistician, from general practitioners to pure mathematicians
3. The rise of ‘statistical machines’ i.e. electronic computers.

#OTD 1942 Sir Maurice Kendall ASA Fellow 1950, reads his paper ‘On the Future of Statistics’ before the Royal Statistical Society. Eighty-four years later, how many of his predictions pan out? 1/4🧵

#OTD 1923 Jacob Cohen b (d 20 Jan 1998) ASA Fellow 1983, Psychologist & statistician, developer of Cohen’s kappa, d, and h for effect size, & critic of misapplications of null hypothesis significance testing ["The Earth is Round (p < .05)" is a classic].

4/ Finally he introduces new terms and definitions including key criteria of consistency, efficiency, sufficiency, estimation, likelihood & information
Fisher RA 1922. Phil Trans Royal Soc. A 1922: 309-368

3/ He clarifies the distinction between ‘parameters’ (the objects of estimation) from ‘statistics’ (used to estimate the parameters), hopelessly confused in the literature up to that time.

2/ The paper was first read to the Royal Statistical Society in Nov 1921. Fisher being Fisher he begins with a jab at Karl Pearson. He then proposes a unifying theory of statistics, with the focus on problems of estimation & distribution.

Ronald Fisher again. Today is the 104th anniversary of The One Ring to Rule Them All: #OTD 1922 Fisher publishes his paper "On the Mathematical Foundations of Theoretical Statistics". It ushers in modern mathematical statistics which until then lacked a unifying structure. 1/3

In April 1912, as a third-year undergraduate student at Cambridge, RA Fisher publishes his first mathematical paper ‘On an Absolute Criterion for Fitting Frequency Curves’. This is the earliest iteration of what he would later call the method of maximum likelihood.

In April 1959 Nathan Mantel and William Haenzel published a method for effective use of heterogeneous retrospective study data, one of the earliest examples of analytical rather than descriptive epidemiology. The Mantel-Haenszel test is now key for analysing stratified or matched categorical data.

3/ Ronald Fisher at first welcomed Neyman as an ally against Bayesian statistics. However, relations soon deteriorated, to say the least. Neyman later said that Fisher’s ‘torrents of derogatory remarks, insults, & polemics’ were ‘probably unparalleled in the history of science’.

2/ Joint work with Egon Pearson had been carried out by letter (1928-33). He had gone to UCL on a research fellowship with Karl Pearson in 1925 but found the work ‘old-fashioned’ & Pearson ‘surprisingly ignorant of modern mathematics’ so he went to Paris. He returned in 1934 after Karl retired.

#OTD 1894 Jerzy Neyman b (d 5 Aug 1981). ASA Fellow 1942; RSS Guy Gold Medal 1966; ASA Wilks Award 1968; US National Medal of Science 1969. Codeveloper with Egon Pearson of the ‘hypothesis testing’ paradigm (Neyman–Pearson lemma 1933) 1/3

4/ Although nearly blind during the last years of his life he produced at least one maths paper per week. On the day of his death “He gave a mathematics lesson to one of his grandchildren, calculated the motion of balloons; then discussed with Lexell and Fuss the recently discovered planet Uranus”

3/ He did pioneering work in graph theory, topology, astronomy, fluid mechanics, music theory, analytic geometry, trigonometry, calculus & number theory.

2/He used (& invented) consistent notation and terminology (e.g summation Σ, Euler's number e = 2.71828, i for √(-1) ). He pioneered the concept of functions f(x), beta and gamma functions, and promoted use of exponential functions and logarithms in analytic proofs

#OTD 1707 Leonhard Euler b (d 18 Sep 1783) Laplace called him the “Master of us all”. Polymath & prolific in every sense of the word (>860 publications, 13 children) he created modern mathematical thinking 1/4 🧵

#OTD 1912 RSS Titanic sank after an iceberg strike, killing 1502 of 2224 people on board. In 1996 Robert Dawson published the first public database, used ever since for teaching contingency analysis, risk, logistic regression, prediction modelling, machine learning & data visualization.

5/ In 1837 he published an explanation of his "numerical method" in English in the London Medical Gazette
www.jameslindlibrary.org/louis-pca-18...

4/ In an 1828 study using this method, he analysed 77 pneumonia cases to evaluate claims of efficacy for early vs delayed bloodletting. Apparently early bloodletting reduced disease duration but increased short-term mortality (“the operation was a success, but the patient died”)

3/ His methods were precursors to modern case-control studies: he grouped cases by similarities in age, sex, diet, disease severity, & onset. He used meticulously documented case records & standardized patient histories.

2/He argued that there would be no progress in medicine if each case was assessed individually & efficacy based on individual subjective judgement. Instead, he promoted ‘population’ comparisons, with appropriate comparators.