American Statistical Association Sesquicentennial

I would like to start this presentation by relating to you what John Koren said when he gave his presidential address to the american statistical association on February 13th 1914 on the occasion of the Association 75th anniversary his opening remarks went like this a page of the unpretentious volume containing the first proceeding of our society bears this legend here close the records of the american statistical association for the first quarter of a century indeed there seems to have been something prophetic in the words here closed the records of the american statistical association for the first quarter of a century who wrote them look forward to other milestones in the history of our organization to MU and perhaps ampler records of achievement and we are gathered to bear testimony that his faith did not belie him so spoke John Corran 75 years ago in that same spirit this presentation has been made in celebration of the achievements sustained for a century and a half by the american statistical association on November 27th 1839 five men met at number fifteen cornhill in Boston to found a statistical society the founders were the Honorable Richard Fletcher a US congressman and influential leader of the Boston Bar who became the Society's first president William Cogswell a Protestant minister of great influence in the New England states John dicks Fisher a physician Oliver Peabody a lawyer clergyman and early assistant editor of the North American review and last Lemuel shattuck a publisher a bookseller and a friend of Ralph Waldo Emerson who retired from his business to take up the affairs of city state and nation it is presumed that these men had little formal training in mathematics beyond arithmetic in fact Shattuck the man responsible for gathering this hand-picked group stands alone as the statistician by 1839 shanik had shown a deep interest in the tabulation and interpretation of statistics relating to births marriages mortalities and education in concord and boston and was in correspondence with the leading european statisticians at that time the objectives of the american statistical society as it was called for a brief time after its inception or simple to collect preserve and diffuse statistical information in the different departments of human knowledge the founders said about writing bylaws incorporating the association in Massachusetts establishing a library and beginning a series of publications they agreed to meet quarterly to present reports and to require all members to quote prepare at least one article a year on some statistical subject end quote these early reports and publications focused primarily on enumeration a description of social or demographic data Reverend Joseph felt the Society's first recording secretary and a member of the publication committee took the lead in preparing these early works the collection of these early works into a library was aided by the generous monetary contributions of the second president of the american statistical association George Shattuck who was no relation to Lemuel shattuck early meetings also included the reading of correspondence letters from such honorary members as President Martin Van Buren Florence Nightingale and Belgian scientist Adolphe kettle a one of the leading statisticians of the 19th century kettle a had published extensively in statistics and astronomy he was influential in establishing the statistical Society of London in 1834 later to become the royal statistical society catalase interest in astronomy led him to the works of Legendre Laplace and Gauss in the early part of the 19th century in 1805 Legendre had discovered the method of least squares and by 1810 Laplace and Gauss had synthesized the normal probability distribution and least squares into a coherent framework with many applications in astronomy and geodesy in the mid 19th century cattle a proceeded in great earnest to apply these ideas to the study of human populations his work laid the foundation for quantified investigation in the social sciences and he sought to establish the laws of society of social physics as he called it similar to the physical laws of astronomy that were established in the previous century his writings influenced many throughout Europe especially Francis Galton the English naturalist and first cousin of Charles Darwin like cattle a golden struggled with the problem of describing the variability observed in many social and natural phenomena while astronomers viewed this variability as errors of the measurement process golden recognized the need to explain the distribution of the measurements as we will see shortly his work on these problems became part of the foundation for modern statistics in the United States statistical science of the 19th century continued to deal primarily with data gathering classification and administrative reporting by 1850 the american statistical association had developed a close relationship with the US government for conducting the decennial census Lemuel Shattuck had been successful in conducting a census of Boston in 1845 as a reaction to serious errors in the eighteen 40 US census reports Edward Jarvis a medical doctor was another one of the first to recognize the misrepresentation of blacks and the insane in the 1840 census Jarvis was elected president of a essay in 1852 and consulted on the censuses of 1850 1860 and 1870 Jarvis remained president of a essay for 31 years until he retired in 1882 and Francis Walker the foremost statistician in the United States at the time was elected president Walker a general at the end of the Civil War had been the federal census superintendent in 1870 and again in 1880 while also serving as president of the Massachusetts Institute of Technology he saw the need to increase the awareness of statistical science and to support statistical education his a SI presidential address of 1897 stated firmly the United States have spent millions and tens of millions upon the collection compilation and publication of statistics and yet they have never spent perhaps ten thousand dollars certainly the government has never spent anything in training and preparing the men who should conduct the statistical service of the country now while our government has been thus delinquent our colleges and universities have done almost nothing under Walker's leadership and the assistance of Davis Dewey a young professor at MIT and Secretary of a sa a quarterly publication now known as the journal of the american statistical association or Jassa was founded in 1888 Walker also made a concerted effort to include academic in the a-si which up to then consisted largely of lawyers public officials and those in the Social and Health Sciences membership included such notable figures as Herman Hollerith Andrew Carnegie and Alexander Graham Bell and grew to more than 500 during Walker's presidency which lasted until 1897 Walker also wanted to make a essay and acknowledged National Association and in 1896 aasa began to hold me in other cities initially in Washington DC under the leadership of Jarvis and then Walker the a-si vigorously supported the use of demographic statistics by the federal government their efforts helped lead to the establishment of the bureau of labor statistics in eighteen eighty five the bureau's first director carol right ultimately followed Walker as president of a sa from 1898 2 19 09 the a-si also provided tremendous support for the establishment of a permanent Census Bureau in 1902 simon north the first census bureau director followed right as president of a sa in 1910 and was the first president to serve a single year term unlike the Americans the leading English scholars and statistics took a more mathematical route in the 1870s and 1880s Francis Galton began what was to be a long legacy of statistical development in Britain when he formulated his notions of correlation and regression as he studied heredity and sweet peas and human populations fellow English scholar Francis edgeworth supplied some of the more rigorous mathematical formulations of correlation and regression Karl Pearson an applied mathematician and philosopher of science brought the techniques into prominence with the assistance of Anu you'll by teaching these techniques at University College London in the late 1890s throughout the decade of the 1890s while Francis Walker was lamenting the state of statistical science in the US English scholars were making statistics a discipline worthy of study in its own right in 1891 Florence Nightingale wrote two golden pleading that he uses influence to obtain a statistical professorship or readership at the University of Oxford for the teaching of social physics the ultimate goal for such a position was to teach people how to use statistics in order to legislate for and to administer our national life while that professorship did not materialize karl pearson did take up the challenge of teaching statistical science along with walter Weldon a cambridge zoologists pearson founded the biometrical school at University College London which was to become a center of statistical learning and research for many years Pearson was committed to make statistics a branch of applied mathematics with a technique and nomenclature of its own to train statisticians as men of science and in general to convert statistics into a serious branch of science which no man could attempt to use effectively without adequate training Pearson was an enormously energetic man whose contributions to statistical theory methods and applications include more than 300 publications besides his formulation on the sample product-moment correlation coefficient Pearson published major works on multivariate normal correlation multiple linear and polynomial regression general measures of association for categorical data and large sample estimation and distribution theory one of his most famous works the chi-squared test for goodness of fit was published in nineteen hundred this comparison between what is observed in a collection of data and what is hypothesized by scientific theory is a fundamental theme of statistical analysis and one of the great scientific developments of the 20th century and weldon suggestion pearson founded the journal biometrika in 1901 and was its principal editor until his death in 1936 although his contributions to science span many disciplines including mechanics eugenics and philosophy karl pearson is most notable as the founder of statistical science the mathematical approach to statistics continued unabated in England as utne you'll further elaborated the mathematics of correlation and regression in the early 1900's retrieving the method of least squares to define the new technique to be called regression analysis you'll also wrote the first authoritative texts in mathematical statistics and introduction to the theory of statistics first published in 1911 this book went through 10 editions and remained the definitive theoretical statistics treatise through the late 1930s when Maurice Kendall became a co-author and then extended the work to his own book the advanced theory of statistics although yewll's work reflected Pearson's interest in studies based upon examination of statistically large numbers of specimens others began taking interest in data collected from small-scale laboratory experiments and from agricultural field trials one such individual was William Gossett who had studied at Pearson's biometrical laboratory in 1908 Gossett under the pseudonym student extended the comparison of two means two small samples developing the t test to compare varieties of barley used at the Guinness brewery in that same year Gossett also derived the small sample distribution for the sample correlation coefficient arising from a bivariate normal distribution with correlation equal to zero several years later Gossett's work was read by Ronald Fisher who had graduated from Cambridge in 1912 one of Ronald Fisher's earliest contributions to mathematical statistics was the extension of Gossett's work to the distribution of the sample correlation coefficient for any sample size and any degree of correlation Fischer's work completed while he was teaching mathematics and physics in English public schools was quickly recognized by Pearson who readily published it in biometrika in 1915 in 1919 Fisher took a position involved with agricultural research at the Rothamsted Experimental Station an event which accelerated the development of statistical science in his fourteen years at rothamsted Fisher's contributions to mathematical statistics and statistical methods were many and varied Fisher either developed or laid the foundation for much of today's statistical theory and methodology some of his major contributions were maximum likelihood which had its seed in his 1915 work on correlation sampling distribution theory consistency and efficiency of estimators sufficient statistics randomization and experimental design perhaps the greatest legacy to statistics or two signs in general from fishers days at rothamsted is his work on experimental design as exemplified by his 1926 paper the arrangement of field experiments and his 1935 book that sign of experiments his 1925 book statistical methods for research workers has served as a guide for generations of statisticians in 1933 Pearson resigned his post at University College London and Fisher was named his successor as the second galten professor of eugenics Pearson's son Egon was appointed head of the statistics department at University College London Frank Yates who had worked at rothamsted since nineteen thirty one followed in Fishers footsteps at rothamsted contributing significantly to experimental design during the second decade of this century the need for an interest in mathematical statistics was growing in the United States the earliest work of substance on mathematical statistics in the u.s. was due largely to Henry Reese Reese received his PhD in 1902 from Cornell University taught for a time at the University of Illinois and settled at the University of Iowa in 1918 he was well informed about the conceptual developments in mathematical statistics in Europe and actively work to transplant these ideas into the United States Harry Carver was another major proponent of mathematical statistics in those early years while World War one raged in Europe Carver was asked to teach a course on mathematical statistics at the University of Michigan at the time no American writers covered what Carver envisioned as mathematical statistics so he developed a new more theoretical outline for the course using yewll's text and papers from Pearson gossip and Fischer undoubtedly Carver's greatest contribution was the founding of the annals of mathematical statistics first published in February 1930 and presently published as the annals of statistics and the annals of probability at Carver's prompting on September 12 1935 several persons gathered at Michigan to form an organization known as the Institute of mathematical statistics with reits as president walter shoe hard as vice president and allen craig as secretary-treasurer with the annals of mathematical statistics as its primary publication the institute gave new focus to research on the mathematical theory of statistics by the 1930s the Atlantic Ocean was no longer a barrier to the free flow of ideas in mathematical statistics Jerzy Neyman and Egon Pearson developed a rigorous theory for testing statistical hypotheses based on the Neyman Pearson lemma other early contributors to the mathematical theory of statistics included William Cochran who proved that certain quadratic forms possess a chi-squared distribution RC Bose who developed much of the theory for partially balanced experimental designs and PC mahalanobis harold hotelling Samuel Wilkes John wishard and SN Roy who extended many results to multivariate data as part of the centenary celebration of a essay in 1939 Davis Dewey who had labored long and hard with Francis Walker on behalf of the a-si brought these developments into focus when he recalled the changes he had seen since the 1880s statistics as I knew it a half-century ago was a humble worker in the society of arts and sciences which was trying to explain the happenings of this complicated world statistics at that time was wedded to another humble toiler arithmetic since then statistics has grown in many ways and has taken unto itself algebra geometry and calculus a rapid growth and applied statistics accompanied the expansion of statistical theory and provided fertile ground for continued theoretical research and development for example quality control now so visible in our high technology society had its beginnings in the 1920s when Walter shoe heart of Western Electric and Bell Laboratories developed a simple workable control chart in 1929 Harold dodged the originator of the operating characteristic curve for sampling inspection procedures published with Harry Romig a classic paper entitled a method of sampling inspection in the Bell System technical journal this led to courses sponsored by the War Department in the late nineteen thirty taught by dodge by the advent of World War 2 W Edwards Deming also had initiated a program that included a series of ten day courses on quality control given by notable statisticians such as Samuel Wilkes Walter schuchardt Holbrook working Eugene grant Boyd harshbarger and deming himself these developments starting in the 1920s and continuing through World War two spurred the application of statistics to engineering problems many other areas of statistical application grew rapidly during the 1930s the value of well-designed sampling plans was given a tremendous boost with a successful completion of the 1937 unemployment census which was designed largely by Calvert Diedrich and Morris Hanson of the Census Bureau sampling research and application accelerated with the addition of ed Deming William Hurwitz Phil Hauser and William Matt out to the Census Bureau opinion polls such as the Gallup poll also emerged at this time other government statisticians provided leadership and applications of statistics including jack newton and churchill eisenhardt at the national bureau of standards jerome cornfield at the National Institutes of Health and William calendar of the crop reporting board of the US Department of Agriculture biometry and medical statistics grew rapidly at this time Joseph berkson provided leadership at the Mayo Clinic Chester bliss and David Finney developed statistical methods in bioassay and Bradford Hill applied Fischer's ideas of randomization to clinical trials of antibiotics the a-si responded to the growing interest in biological applications of statistics by chartering the biometrics section in 1938 as the first section of a essay as theory and application pushed forward the need for university courses and programs grew although Richmond Mayo Smith had taught from his applied statistics text at Columbia from the early 1890s the department of biostatistics at the johns hopkins university found in 1918 by Raymond pearl was the first in the United States with statistics in the title however the strongest leadership on the academic front in those early years came from George Snedeker at Iowa State University driven by the application of statistical methods to agricultural problems he offered a course in statistics at Iowa State as early as 1914 in 1924 snedeker and Henry Wallace who later became vice president of the United States organized a series of saturday afternoon seminars for agricultural research workers this effort led to establishing the mathematical statistical service in 1927 which developed into the statistical laboratory in 1933 Snedeker was instrumental in transporting the work of Ronald Fisher to the United States and at the invitation of Snedeker Fisher spent the summer of nineteen thirty-one at Iowa State snedeker had a tremendous knack for teaching applied statistics and he made the application of statistical methods understandable for research workers in many fields Snedeker's text statistical methods first published in 1937 has been through numerous editions it continues to be one of the most cited publications in the science Citation Index clear testimony to its impact on the development of science and technology as karl pearson had done earlier in the century at University College London snedeker established Iowa State as a focal point for research on statistical theory and methods William Cochrane and Charles Windsor were on the faculty there and Gertrude Cox was the first master of science graduate in statistics from Iowa State University in 1931 Cox went on to organize and head the department of experimental statistics at North Carolina State University in 1940 in 1944 she organized the institute of statistics which was expanded in 1946 to include a department of mathematical statistics and expanded again in 1949 to include a department of biostatistics both at the university of north carolina the department of experimental statistics remained at north carolina State University the statistical efforts at Iowa State culminated in the formation of the statistics department in 1947 and in 1950 Theodore Bancroft was named the first permanent head of the department throughout the United States academic statisticians were forming statistical laboratories consulting services and separate apartments key figures were Raymond pearl at Johns Hopkins George snedeker at Iowa State holbrook working at Stanford harold hotelling and later Abraham Wald and Jacob Wolfowitz at Columbia Frank y de at George Washington Samuel Wilkes at Princeton Jerzy Neyman at Berkeley Gertrude cocks at North Carolina State Boyd harshbarger at Virginia Tech Walter Wilcox at Cornell and Chester bliss at Connecticut and later at Yale the rapid expansion of statistical theory and application in government and industry began shortly after World War one and grew rapidly through World War two the a-si adapted to the growing needs of its members during this vibrant era Jassa became more flexible and gradually shifted its emphasis away from descriptive articles and toward more analytical work in order to meet the needs of smaller groups of statisticians the formation of local chapters began in the late 1920s in los angeles washington d.c boston and other industrial cities of the Upper Midwest as previously mentioned sections were also formed starting with the biometrics section in 1938 followed by the section on statistical education in 1944 in 1935 aasa moved its headquarters to Washington DC and in 1936 a sa appointed its first full-time paid executive director and secretary Frederick's to fan thus solidifying its commitment to the long-term vitality of a SA and its members in 1944 the a-si continued its progressive movement by a lectin Helen Walker its first woman president by the 1940s the statistical community would growing at an astonishing rate with World War two as the catalyst statistical science had come to the fore following the war many statisticians turned their efforts to a great variety of human endeavors although a great number of statistical researchers are worthy of note here future generations will undoubtedly extol their contributions as they withstand the test of time these researchers produced a tremendous diversification and expansion of statistical science that is difficult to capture in detail but several trends which have their roots in the distant past are worthy of note from the early work on probability theory by Jacob Bernoulli and abraham de moivre in the late 1600s and early 1700s to the profound work of Henry chef a Harold Kramer david blackwell Jack Kieffer and many others that followed mathematical statistics continues to be the foundation for statistical thinking from the origin of statistical graphics in the 1786 book commercial and political atlas by William play fair to Francis Galton's 1863 representation of multivariate data in weather maps and to Francis walkers innovative graphical techniques in the first u.s. atlas in 1870 graphical statistical methods are an important part of exploratory data analysis and have grown into a field of study of their own John groans 1662 work natural and political observations upon Bills of mortality in which he studied the plague in London anticipated the use of modern-day survival techniques such as the kaplan-meier estimator and proportional hazards models certainly these have played an important role in the health and economic welfare of society and Thomas bayes's work an essay towards solving a problem in the doctrine of chances which appeared in 1764 several years after his death has been revived in the form of Beijing statistical inference advanced by Jimmy savage and other notable statisticians in the 1950s and 1960s the utility of Beijing methods remains a popular topic in statistical research and application today in 1710 John Arbuthnot's si an argument for divine providence taken from a constant regularity observed in the births of both sexes in which he used the equivalent of a sign test was a precursor to Charles spearman's 1904 publication on rank correlation and Frank Wilcox ins 1945 publication on rank tests statisticians have continued to develop the theory and application of what is now known as nonparametric statistical inference although many more developments can be traced from modest beginnings hundreds of years ago to modern fields of statistical interest perhaps the most dramatic advances in statistical science have been aided by advances in computer technology there is a long history of computing machinery and devices but the most important breakthrough in computing came from Herman Hollerith of the census office following the 1880 united states census which took nearly eight years to tabulate hollerith and others realized that tabulation of the 1890 senses would be an insurmountable task Hollerith left the census office to form his own company which evolved into IBM his computing equipment was used for the 1890 census and was a tremendous success Karl Pearson and later Egon pearson and h/o Hartley published the biometric at tables for statisticians which represented a tremendous effort in tabulating critical values of various sampling distributions moments of order statistics and many other valuable statistics florence david also contributed heavily to the tabulation of statistics associated with the sampling distribution of correlation coefficients and symmetric functions however with the advent of computers many of these laborious calculations are made routinely now sophisticated computations analyses of large data sets computer simulations and computer intensive methods involving resampling and permutation tests are now at the fingertips of statisticians for theoretical developments and practical applications the a-si has kept pace with the acceleration of developments related to statistical science today a.s.a is more involved and committed to the growth of statistics than ever before in its 150-year history from its small beginning the a-si has evolved into an international organization of more than 15,000 members with numerous sections and local chapters to meet the interest and needs of its members a central headquarters in Alexandria Virginia and annual meetings attracting thousands of members the men who founded the a-si in 1839 may have dreamed of such a result in the first bound publication in 1843 one of the concluding remarks of the preface states that the members of this newly formed association hope that by increasing progressively in efficiency and usefulness they shall eventually realize many of the beneficial effects which they anticipate and they also hope that the society will hold a permanent place among the important institutions which are steadily contributing to the welfare and improvement of this and other countries their hopes are now reality the dedicated men and women of a SI have helped transform the science of statistics from the avocation of a small group of men to a scientific field with scholarly pursuits in research and applications with a solid foundation based on 150 years of experience the a-si is poised for continued growth well into the next century and beyond this presentation was coordinated by stephen j rue berg and was sponsored by the council of chapters for the american statistical association sesquicentennial celebration major funding for this project was provided by the following imsl which is pleased to be a major contributor to the filming of this historic video and to offer our congratulations and continued support to the american statistical association on the occasion of their sesquicentennial and by bmd p a respected leader in the statistical software field for the past 30 years and by beranger mannheim new technologies and diagnostics biochemicals orthopaedics and now pharmaceuticals and by fiddy afar misu t'kul corporation innovation in neuroscience and excellence in clinical research other significant contributors to this project were the following this has been a presentation of the american statistical association