統計學程英文網站

Egon Pearson (1895~1980) Egon Pearson was a monumental figure in 20th-century statistics, renowned for his collaborative innovations in inference theory and his pioneering efforts in industrial statistics. His intellectual journey is defined by a commitment to rigorous inquiry, marked by profound academic collaborations and a courageous dedication to scientific integrity. Collaboration with Jerzy Neyman and the Neyman-Pearson Lemma Egon Pearson's most celebrated contribution to statistics is the development of the Neyman-Pearson Lemma. His partnership with Jerzy Neyman, fostered through years of intense and thoughtful correspondence (1928–1933), was built on a foundation of deep friendship and intellectual synergy. While Neyman provided brilliant, original insights into the core of complex problems, Pearson's humble and cautious nature ensured that these methods were rigorously validated. Their collaboration began with a fundamental critique of the Chi-Square Goodness-of-Fit Test, originally proposed by Pearson's father, Karl Pearson. Egon questioned the lack of a standardized criterion for selecting the "best" analytical approach, famously asking: If a dataset yields a non-significant p-value under a standard chi-square test, how can one be certain that no other version of the test would reveal a significant departure from the hypothesized distribution? This question led them to integrate R.A. Fisher's likelihood function concepts, culminating in a series of three classic papers that fundamentally transformed the theory of significance testing. Academic Integrity Amidst Historical Rivalry A compelling chapter in Egon Pearson's career illustrates his professional courage. At the time, his father, Karl Pearson, edited the premier journal Biometrika. However, a bitter feud existed between Karl Pearson and R.A. Fisher—the latter of whom had criticized Karl Pearson’s chi-square test, while Karl Pearson dismissed Fisher's likelihood estimation as absurd. Because Egon Pearson and Jerzy Neyman's revolutionary findings were based on Fisher's likelihood concepts, Egon chose to avoid family conflict by publishing their most significant work in a new, independent journal they founded: Statistical Research Memoirs. This decision highlights his prioritize of scientific truth over the pressures of his academic lineage. Pioneering Industrial Statistics Beyond theoretical statistics, Egon Pearson played a critical role in bringing statistical rigor to the industrial world. After meeting W.A. Shewhart (the father of statistical quality control) at Bell Telephone Laboratories, Pearson became a champion for the application of statistical models in manufacturing. His efforts significantly influenced the Royal Statistical Society, leading to the establishment of the industrial and agricultural sections in 1934. In 1936, he authored B.S. 600, a seminal publication on the statistical methods for industrial standardization and quality control. This work successfully bridge the gap between academic theory and practical engineering, cementing his legacy as a founder of modern industrial statistics.

Sir Ronald Aylmer Fisher (1890~1962) Sir Ronald A. Fisher was a preeminent statistician and geneticist whose work laid the bedrock of modern statistical science. Often described as a genius, his life was marked by both extraordinary intellectual capacity and the personal challenges that frequently accompany such visionaries. A Unique Intellectual Genesis Born with physical frailty and severe vision issues, Fisher was medically prohibited from reading under artificial light during his childhood. As a result, he was taught mathematics and astronomy in the evenings without the aid of visual tools like pen and paper. This forced deprivation fostered an exceptional ability for abstract geometric reasoning and rapid logical deduction. Throughout his career, Fisher could derive complex statistical solutions intuitively—solutions that often required other mathematicians months or years to prove formally. Academic Conflict with Karl Pearson Fisher's career was punctuated by a famous rift with Karl Pearson, the editor of the era's leading journal, Biometrika. During his time at Cambridge, Fisher solved a complex problem regarding the statistical distribution of the Galton correlation coefficient using geometric methods. Although his result was mathematically perfect, Karl Pearson—unable to follow Fisher's abstract derivation—initially sidelined the work. Pearson forced Fisher to reduce the theoretical depth of his paper and eventually buried his result in a footnote, behind pages of computationally derived tables. This incident marked a turning point: Fisher never published in Biometrika again, often choosing obscure journals to disseminate his revolutionary ideas, fueling debates regarding whether he was ostracized by the establishment or simply intolerant of Karl Pearson's professional arrogance. The Lady Tasting Tea: The Essence of Experimental Design Beyond his theoretical contributions, Fisher is best known for the anecdote of the "Lady Tasting Tea," which perfectly illustrates the rigor of scientific experimentation. During an afternoon tea in the 1920s, a lady claimed she could distinguish whether milk or tea had been poured into the cup first. While the claim seemed incredulous to the group, Fisher immediately proposed a controlled experiment. He designed a trial where cups were prepared in different sequences and presented randomly to the lady, ensuring that variables such as temperature, mixing time, and steeping duration were strictly controlled. The lady correctly identified every cup, proving her hypothesis. This simple yet profound exercise of hypothesis formulation, controlled design, and statistical analysis became the cornerstone of Fisher's masterpiece, The Design of Experiments, and remains the gold standard for the logic of statistical inference today.

William Gosset (1876~1937) William Sealy Gosset, widely known in the statistical community by his pseudonym "Student," is one of the most intriguing figures in the history of science. Despite being employed as a chemist at a brewery, his pioneering work on small-sample statistics transformed the field of modern inference, giving rise to the ubiquitous "Student's t-test." From Guinness Brewery to Statistical Innovation Gosset studied mathematics and chemistry at Oxford. Upon graduation, he joined the Guinness brewery in Dublin to improve brewing quality. His initial statistical breakthrough stemmed from a practical industrial problem: accurately estimating the concentration of yeast in brewing vats. By analyzing yeast counts under a microscope, Gosset identified that the distribution followed a Poisson distribution. His ability to design an estimation formula based on this distribution allowed Guinness to standardize yeast addition, ensuring consistent beer quality. This practical necessity was the crucible for his future theoretical contributions. The Birth of "Student" To protect commercial secrets, Guinness strictly prohibited employees from publishing research. However, with the encouragement of his mentor Karl Pearson, Gosset began publishing his findings under the pseudonym "Student." It was not until after his sudden death from a heart attack in 1937—when friends attempted to compile his works—that the Guinness company finally discovered that their chemist was the brilliant mind behind over 30 influential papers that had shaped the discipline of statistics. The Student's t-Distribution Gosset's most significant theoretical contribution arose from the limitations of his time. He observed that many scientific experiments relied on small sample sizes, whereas Pearson's existing methods required large datasets to accurately estimate population parameters. Gosset discovered that by calculating the ratio of the sample mean to the sample standard deviation, one could derive a distribution—the t-distribution—that remained valid even with limited data. This revolutionized the reliability of statistical inference in small-sample scientific research. A Bridge Between Giants In addition to his theoretical work, Gosset played a unique social role in the history of statistics as the mediator between the two greatest masters of his era: Karl Pearson and R.A. Fisher. Despite the intense rivalry and personal friction between Pearson and Fisher, Gosset maintained deep friendships with both. He navigated their complex academic relationship with grace, often acting as a bridge between their diverging statistical philosophies.

Jerzy Neyman (1894~1981) Jerzy Neyman was a visionary Polish mathematician and statistician whose work redefined the foundations of modern statistical inference. Known for his profound influence on experimental design and quality control, Neyman is equally remembered for his intellectual rigor and his warm, respectful approach toward his colleagues. Early Life and Mathematical Pursuit Born into a Polish noble family, Neyman demonstrated early linguistic versatility, mastering five languages. Despite receiving an education in Kharkov that lacked modern mathematical resources, his determination led him to discover the works of Henri Lebesgue. His early encounter with Lebesgue—which resulted in a cold rejection by the master—left a lasting impression on Neyman. He transformed this negative experience into a life-long philosophy of treating everyone he met with genuine care, respect, and attentive listening. The Neyman-Pearson Partnership After earning his Ph.D. in Warsaw, Neyman traveled to London in 1925 on a Rockefeller fellowship to work with Karl Pearson. While he found Karl Pearson's grasp of modern mathematics lacking, he forged a lifelong friendship with Egon Pearson. Their collaboration, conducted largely through extensive correspondence between Warsaw and London, flourished into one of the most productive partnerships in statistical history. Their joint efforts led to the development of the Neyman-Pearson Lemma, which provided the theoretical framework for modern hypothesis testing. Building the Berkeley School of Statistics In 1938, Neyman moved to the University of California, Berkeley. At the time, statistics was merely a small laboratory within the mathematics department. Neyman fought tirelessly against departmental opposition to establish statistics as a distinct academic discipline. In 1955, his vision materialized with the founding of the Department of Statistics at Berkeley, which became the premier cradle for elite statisticians in the United States. Diverse Contributions to Science Neyman's impact extended far beyond theoretical statistics: Contagious Distributions: He developed models to describe the spatial distribution of biological species, which became standard tools in ecological research. Sampling Theory: His pioneering work on Stratified Sampling and purposive selection fundamentally changed methodology in sociology and the social sciences. Regulatory Standards: His research on experimental design and statistical methodology deeply influenced the U.S. Food and Drug Administration (FDA), shaping the rigorous protocols used today for pharmaceutical testing and approval.

Andreyii Nikolaevich Kolmogorov (1903 - 1987) Andrey Kolmogorov was a legendary Russian polymath whose contributions spanned mathematics, physics, biology, history, and philosophy. Often described as a brilliant comet that illuminated 20th-century science, his work laid the modern foundations for probability theory and stochastic processes. His intellectual breadth was matched only by his insatiable curiosity and legendary physical vitality. A Prodigy in Formative Years Born in Tambov, Russia, Kolmogorov lost his mother during childbirth and was raised by his aunts, who nurtured his early genius. His first mathematical discoveries were published in the family's homemade journal, Spring Swallow. By age 14, he was already teaching himself advanced mathematics and proving theorems that were left unproven in his textbooks. His academic prowess was so immense that, upon requesting to take his high school graduation exams early, the faculty simply granted him his diploma without examination, claiming he had already surpassed the school's requirements. Mathematical Ascendance At Moscow State University, Kolmogorov's genius became legendary. To complete his graduate studies, he was required to pass 14 core examinations or present an original paper for each. While a single original paper was considered an achievement for an average student, Kolmogorov submitted 14 original papers, each in a different field of mathematics. This level of prolific originality defined his entire career, culminating in his masterpiece, the Axiomatization of Probability, which transformed probability from a collection of intuitive puzzles into a rigorous branch of mathematics. A Devoted Educator Beyond his research, Kolmogorov was a passionate educator. He dedicated years to designing curricula and personally mentoring classes of gifted children in the Moscow region. He emphasized the development of a "complete and natural personality" over pure academic output. He believed that if a child's horizon remained broad and their curiosity remained active, they would find fulfillment in any profession they chose, whether as mathematicians or otherwise. A Life of Legend and Vitality Kolmogorov's life was as colorful as his work. He was known for his physical vigor, often priding himself on his athletic abilities. Stories of his life are filled with legendary anecdotes—from his youthful bouts in train stations to his eccentric late-life decision to swim in the icy Moscow River at the age of 70. These stories reflect a man who lived with the same intensity he applied to his research. In his later years, he grappled with the philosophical depth of statistics—questioning the very meaning of probability in real life—a challenge that continues to inspire and intrigue scholars today.

Sir Austin Hill (1897 ~ 1991) Sir Austin Bradford Hill was a distinguished British public health scientist and a self-taught statistician who revolutionized the integration of statistical rigor into medical and public health research. His work, particularly in establishing the causal link between smoking and lung cancer, remains a landmark achievement in epidemiology. From War and Illness to Academic Success Hill's early life was marked by resilience. His medical studies were interrupted by World War I, during which he served in the Royal Naval Air Service. In 1917, he contracted tuberculosis—then considered a terminal illness—and was sent home to recover. Following a successful artificial pneumothorax procedure, he transitioned to economics at the University of London. Despite spending much of his time studying from his sickbed, he graduated with first-class honors, finishing second in his class. A Pioneer in Medical Statistics Hill eventually found his calling at the Medical Research Council, investigating regional mortality rates. It was here that he encountered Karl Pearson, whose passion for statistical philosophy deeply influenced Hill's approach. Hill successfully bridged the gap between academic theory and practical medicine, becoming a leader in the Medical Research Council's statistical unit and a professor at the London School of Hygiene and Tropical Medicine. Transforming Medical Research Hill is widely regarded as having a greater impact on medical research than many Nobel laureates. His contributions are threefold: Principles of Medical Statistics: In 1937, he published a series of articles in The Lancet, later compiled into the seminal book Principles of Medical Statistics, which became the definitive guide for medical practitioners. Experimental Design: Hill was a tireless advocate for applying R.A. Fisher's principles of experimental design to medical trials. He single-handedly convinced the British medical community to adopt rigorous, evidence-based research methods. Randomized Controlled Trials: He was instrumental in introducing randomized treatment methods into clinical trials, fundamentally altering how we evaluate medical efficacy today. His legendary collaboration with Richard Doll remains the gold standard for epidemiological research, providing the definitive evidence that smoking causes lung cancer and establishing the "Hill's Criteria" for causality that are still taught in public health curricula worldwide.

Chester Ittner Bliss (1899 ~ 1979) Chester Ittner Bliss was an American entomologist and statistician whose development of Probit Analysis revolutionized pharmacology and toxicology. His work provided the scientific community with the methodology to quantify the effects of chemicals and drugs, most notably through the estimation of the LD-50 (Median Lethal Dose). From Entomology to Statistical Innovation Bliss began his career as an entomologist for the U.S. Department of Agriculture. Influenced by R.A. Fisher's Statistical Methods for Research Workers, Bliss became fascinated by the challenge of modeling insecticide effects. He observed that mortality responses in insect populations did not follow simple linear patterns. By utilizing statistical distributions to link dosage levels with mortality probability, he developed Probit Analysis, a method that remains a standard tool for determining the potency of substances. A Perilous Journey In 1933, Bliss's laboratory was closed due to bureaucratic indifference to laboratory-based insect research. During the Great Depression, finding work was nearly impossible. With R.A. Fisher's assistance, he eventually accepted a position at the Institute of Plant Protection in Leningrad, USSR. His time in the Soviet Union during the Stalin era was marked by high tension; many of his colleagues disappeared or were executed. Thanks to the intercession of a young American student who befriended him—and a government committee that concluded Bliss was too "transparent" to be a spy—he managed to survive. He was eventually urged by his friend to flee the country abruptly, leaving his experiments unfinished. Legacy in Statistics Upon returning to the United States, Bliss pursued a distinguished academic career. His contributions bridged the gap between biological observation and mathematical modeling. His invention of the Probit (Probability Unit) transformation allows researchers to translate the non-linear relationship between dosage and survival probability into a linear one, making the calculation of the LD-50 mathematically robust. His life, characterized by both scientific brilliance and extraordinary survival in a volatile historical period, remains a powerful testament to the impact of rigorous statistical inquiry on practical health and safety standards.

Sir Francis Galton (1822~1911) Sir Francis Galton was a British polymath whose pioneering work in statistics, heredity, and human variation fundamentally shaped modern biology and social science. As a cousin of Charles Darwin, Galton applied quantitative rigor to the study of evolution and human traits, leaving behind a legacy that includes the invention of fingerprint identification and the statistical concept of regression. From Anthropometry to Regression Galton's pursuit of understanding heredity led him to establish an anthropometric laboratory in London in 1884, where he collected vast amounts of data on physical traits such as height and bone length. Through his analysis of height data, he observed the phenomenon of "regression to the mean": extremely tall parents tend to have children who are shorter than themselves, while extremely short parents tend to have children who are taller. To quantify this relationship, Galton developed the concept of the coefficient of correlation. This foundational work was later expanded upon by Karl Pearson, evolving into the robust statistical method now universally known as regression analysis. Pioneering Statistical Theory Beyond regression, Galton's contributions were profound: Branching Processes: His 1874 collaboration with Reverend H.W. Watson on the extinction of family names is recognized as the origin of the statistical theory of branching processes. Normal Distributions: He identified that a linear combination of normally distributed random variables remains normally distributed—a discovery cited by historian S.M. Stigler as one of the most significant breakthroughs in late 19th-century statistical development. A Patron of Science: The Birth of Biometrika Galton's commitment to the field was not merely intellectual but also institutional. In 1901, he used the Biometrika Trust to fund the creation of the journal Biometrika, edited by Karl Pearson and W.F.R. Weldon. Galton's generous support ensured the journal set the gold standard for scientific publishing, featuring high-quality paper, intricate illustrations, and a commitment to displaying the most complex mathematical formulas of the time.

Frank Wilcoxon (1892~1965) Frank Wilcoxon was an American chemist and statistician whose pioneering work in the 1940s fundamentally challenged the dominance of parametric statistics. His development of rank-based methods laid the groundwork for modern nonparametric statistics, providing robust alternatives to classical tests when data is skewed or contains outliers. The Industrial Chemist's Dilemma While working as a chemist at the American Cyanamid company, Wilcoxon encountered a recurring problem: the "Student's t-test" and Fisher's analysis of variance (ANOVA) were frequently failing him. His chemical experiments often produced outliers—sometimes caused by fluctuating reaction temperatures or catalyst variability—which drastically skewed his results, rendering standard parametric tests unreliable. Seeking a way to handle these extreme values without resorting to arbitrary data deletion, Wilcoxon turned to the literature but found no existing statistical tools to address his needs. The Birth of Nonparametric Statistics Driven by practical necessity, Wilcoxon developed a method that bypassed traditional formulas altogether. Instead of relying on population parameters, he devised a procedure based on the ranking of data points, using combinatorial comparisons. Though he initially doubted the originality of his approach, assuming that a professional mathematician must have already solved such a simple problem, he submitted his work to the journal Biometrics. To his surprise, neither the reviewers nor the editor knew of any similar method. His paper, published in 1945, introduced what are now known as the Wilcoxon two-sample rank sum statistic and the one-sample signed rank statistic. A Paradigm Shift in Statistical Inference Before Wilcoxon's 1945 paper, statistical testing was almost exclusively built upon estimating parameters from specific probability distributions. Wilcoxon's breakthrough shifted the focus away from parameter estimation and toward the comparison of distributions based on data ranks. By assessing whether observed data differed from a random distribution, he opened a new frontier in statistics: nonparametric testing. His work marked a major milestone in the evolution of statistics after the Pearson era, offering scientists a powerful, distribution-free method that remains a cornerstone of data analysis today.

George Snedecor (1881~1974) George Snedecor was a seminal figure in the development of statistical education and institutional infrastructure. While he did not focus on developing original mathematical theory, his unparalleled ability to bridge the gap between abstract statistics and practical application—and his vision in building academic departments—left an indelible mark on the field. A Fortuitous Career and Institutional Leadership Snedecor's career began with a geographical misunderstanding: in 1913, while attempting to interview for a position at the University of Iowa, he inadvertently ended up at Iowa State College in Ames. Despite the school's lack of a vacancy in mathematics, he was hired to teach algebra and eventually remained there, establishing the first statistical research center and, subsequently, the first department of statistics in the United States. His commitment to the discipline extended to mentorship, most notably in his guidance of Gertrude Cox, whom he famously recommended as the "best" candidate to lead the statistical center at North Carolina State University, surpassing his own list of established experts. Spreading Statistical Literacy Snedecor's most enduring legacy is his textbook, Statistical Methods, co-authored with William G. Cochran and first published in 1937. Modeled after R.A. Fisher's Statistical Methods for Research Workers, the book was designed to be a practical guide. It integrated basic mathematical derivations, organized related themes, and provided essential statistical tables, allowing researchers to easily determine p-values and confidence intervals. The impact of this work was immense: it became the mandatory textbook for generations of scientists and, by the 1970s, was the most frequently cited reference in scientific journals across all disciplines. Continuously updated and revised for decades, it remains a testament to Snedecor's success in demystifying statistics and making rigorous analytical methods accessible to the global research community.

George Box (1919 ~ 2013) George Box was a pioneering statistician whose career path—from a self-taught chemistry student to the founder of a major statistics department—is one of the most remarkable stories in the history of science. His work on experimental design, time series analysis, and the Box-Cox transformation remains fundamental to modern statistical practice. From Chemical Warfare Research to Statistical Mastery Box's entry into statistics was born of necessity during World War II. While serving in a research unit focused on chemical defense, his superiors struggled to analyze vast amounts of complex data. Box volunteered to take on the task, relying on his self-study of R.A. Fisher's Statistical Methods for Research Workers. Faced with problems that existing methods could not solve—such as controlling for inter-arm correlations in skin-irritation tests—Box intuitively applied Fisher's principles of experimental design to derive new models from first principles. A Champion of Scientific Rigor Box's commitment to rigorous experimentation was evidenced by his work on a claimed "antidote" for toxic gas. Upon reviewing the flawed animal-study data provided by an American researcher, Box identified clear confounding factors. He redesigned the experiment, demonstrating that the antidote was, in fact, ineffective. This experience solidified his reputation; when his statistical appendices were initially rejected by military reviewers for being "too difficult," his mentor, Sir Joseph Barcroft, famously interrupted a review meeting to insist that Box's work be included, declaring it exceptional. Institutional Impact and Theoretical Legacy Following the war, Box formalized his statistical education at the University of London, earning his doctorate. His professional journey took him from the Imperial Chemical Industries (ICI) in the UK to Princeton University and finally to the University of Wisconsin, where he founded the Department of Statistics. Box's legacy is defined by his ability to make statistics an active, diagnostic tool for industrial and scientific progress. Alongside his extensive theoretical papers, his collaboration with David Cox resulted in the Box-Cox transformation, a vital technique used to stabilize variance and normalize data. His life serves as a testament to the idea that statistics is not just a branch of mathematics, but an essential language for uncovering truth in experimental science.

Wang Chueh-Ming 汪厥明 (1897~1978) Wang Chueh-Ming was a distinguished agronomist, plant breeder, and the widely recognized founder of biostatistics in Taiwan. His tireless dedication to institutionalizing statistical education and his development of field experimentation techniques fundamentally transformed agricultural research in the region. Academic Resilience and Wartime Research Wang's academic journey was defined by profound resilience. After earning his Master's degree in Agriculture from Tokyo Imperial University in 1924, he held prominent teaching roles in Beijing. During the Second Sino-Japanese War, despite being displaced by conflict and living in extreme poverty in the mountains of Guangxi, Wang remained committed to his scholarly pursuits. With the help of his students, who transported his research data into the wilderness, he managed to complete the original manuscript for his seminal work, Moments, New Moments, Product Moments and Their Interrelations. Building a Statistical Legacy in Taiwan In 1946, Wang moved to Taiwan to join National Taiwan University (NTU), where he served as a professor and head of the Department of Agronomy. To honor his contributions, prominent scholars initiated a fundraising campaign that led to the establishment of the Chueh-Ming Biostatistics Research Laboratory in 1947, which was later elevated to the Chueh-Ming Institute of Biostatistics. Wang's work established the foundation for modern inferential statistics in Taiwan, and his influence was formally recognized through the Ministry of Education's Academic Award in 1955 and his election as an Academician of Academia Sinica in 1963. Theoretical Contributions to Agricultural Statistics Wang was a prolific author with over 50 major works, most notably Theory of Field Experiment Errors and Their Estimation, which provided the theoretical basis for agricultural experimental techniques and error estimation. His other core contributions—such as Estimating Methods of Error Degrees of Freedom and Premises and Significance Testing of Variance Component Analysis—remain essential texts for understanding the development of modern statistics. His life's work serves as a testament to the power of integrating rigorous theoretical frameworks with practical agricultural applications.