Frederick Mosteller: The Architect of Modern Applied Statistics
Frederick Mosteller (1916–2006) was arguably the most influential applied statistician of the 20th century. While many mathematicians retreat into the abstract, Mosteller moved in the opposite direction, bringing the rigor of mathematical proof to the messy realities of social science, medicine, law, and even sports. As the founding chairman of Harvard’s Department of Statistics, he didn’t just teach the field; he expanded its boundaries to touch almost every aspect of modern life.
1. Biography: From Pittsburgh to the Ivy League
Charles Frederick Mosteller was born on December 24, 1916, in Clarksburg, West Virginia, and raised in Pittsburgh. His early aptitude for mathematics led him to the Carnegie Institute of Technology (now Carnegie Mellon University), where he earned his Bachelor’s and Master’s degrees.
In 1939, he moved to Princeton University to pursue a PhD under the legendary Samuel S. Wilks. During World War II, his studies were interrupted by service in the Statistical Research Group (SRG) at Columbia University. This was a "think tank" for the war effort where he worked alongside other luminaries like Milton Friedman and Leonard Jimmie Savage, applying mathematical models to aerial combat and bombing patterns.
Mosteller completed his PhD in 1946 and was immediately recruited by Harvard University. He remained at Harvard for the rest of his career, achieving the rare distinction of chairing four different departments: Statistics (which he founded in 1957), Social Relations, Biostatistics (at the School of Public Health), and Health Policy and Management. He retired in 1987 but remained an active emeritus professor until his death in 2006.
2. Major Contributions: The Mathematics of Evidence
Mosteller’s genius lay in Bayesian inference and data synthesis. His work transformed statistics from a tool for ivory-tower research into a practical engine for public policy.
-
Solving the Federalist Papers Mystery
Perhaps his most famous contribution was using statistical analysis to settle a 175-year-old historical debate: Who wrote the 12 disputed Federalist Papers—James Madison or Alexander Hamilton? Mosteller and David Wallace used Bayesian analysis to study "function words" (like while vs. whilst), proving definitively that Madison was the author.
-
Meta-Analysis
Mosteller was a pioneer of meta-analysis—the statistical process of combining data from multiple independent studies to reach a more reliable conclusion. This is now the "gold standard" in medical research.
-
Stochastic Models for Learning
Collaborating with Robert Bush, he developed mathematical models to describe how humans and animals learn through trial and error, bridging the gap between mathematics and psychology.
-
Public Policy and Health
He advocated for the use of randomized controlled trials (RCTs) in social programs and surgery, arguing that "common sense" was often a poor substitute for empirical data.
3. Notable Publications
Mosteller was a prolific author, known for a writing style that was remarkably clear and free of jargon.
- Inference and Disputed Authorship: The Federalist (1964): Co-authored with David Wallace, this is considered a masterpiece of applied Bayesian statistics.
- Stochastic Models for Learning (1955): With Robert Bush; a foundational text in mathematical psychology.
- Statistics: A Guide to the Unknown (1972): An edited collection of essays designed to show the general public how statistics affects everyday life.
- Data Analysis and Regression (1977): Co-authored with John Tukey, this became a definitive textbook for a generation of statisticians.
- Fifty Challenging Problems in Probability with Solutions (1965): A beloved classic among math students for its elegant puzzles and pedagogical clarity.
4. Awards and Recognition
Mosteller’s peers recognized him as a titan of the field. His accolades include:
- President of the American Statistical Association (1967)
- President of the Institute of Mathematical Statistics (1975)
- President of the American Association for the Advancement of Science (AAAS) (1980)
- The Samuel S. Wilks Award (1970): The highest honor in American statistics.
- Honorary Degrees: He received honorary doctorates from Harvard, Yale, the University of Chicago, and Carnegie Mellon.
- National Academy of Sciences: Elected as a member in 1973.
5. Impact and Legacy: The "Mosteller Principle"
Mosteller’s legacy is found in the rigor of evidence-based practice. Before Mosteller, many medical treatments and social policies were based on anecdote or tradition. He helped establish the Cochrane Collaboration, an international network that summarizes the best evidence from medical research to inform healthcare decisions.
In the classroom, he was a legendary educator. He believed that statistics should be taught through real-world problems rather than abstract proofs. His influence led to the "Mosteller Principle"—the idea that a statistical report should be written so clearly that it can be understood by a reader who is not a specialist in the field.
6. Collaborations: A Scholarly Nexus
Mosteller was famously collaborative, often saying that he:
"never met a data set he didn't like."
- John Tukey: Mosteller’s most significant partnership. Together, they refined the field of "Exploratory Data Analysis" (EDA), emphasizing looking at data to see what it tells us, rather than just testing hypotheses.
- David Wallace: His partner in the Federalist Papers study.
- The "Mosteller Group": He ran a weekly seminar at Harvard that became a breeding ground for future leaders in statistics, including Stephen Fienberg and Howard Wainer.
- The Second Redbook: He worked closely with Daniel Moynihan on the "Moynihan Report" and subsequent evaluations of the Head Start program, bringing statistical rigor to civil rights and poverty debates.
7. Lesser-Known Facts
-
The Magician
Mosteller was an accomplished amateur magician and a long-time member of the International Brotherhood of Magicians. He often used magic tricks in his lectures to demonstrate principles of probability and the fallibility of human perception.
-
TV Star
In 1961, he taught a course called "Probability and Statistics" on Continental Classroom, a nationally televised NBC program. It was one of the first "MOOCs" (Massive Open Online Courses), decades before the internet existed.
-
The "Mosteller" Method of Writing
He was obsessed with clear prose. He would often rewrite a single paragraph a dozen times and was known to ask his secretaries or non-math colleagues to read his drafts; if they couldn't understand a sentence, he would rewrite it until they could.
-
Sports Analytics
Long before Moneyball, Mosteller published papers on the statistics of baseball and the "World Series" outcome probabilities, applying mathematical modeling to America's pastime.