The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century

Rating:
Summary: Pleasing intro to statisctics for lay(wo)men
Review: An intriguing story based introduction to the fast field of
statistics. No formulas but still plenty of math terms explained
as easily as possible. The life stories of many statisticians
are combinded with the history of certain statistical problems.
This book showed me how huge the field of stastics is.
Statistics and Probability seem now to be scientific issues
on not just ways for politicians to cheat the public. In
everyday life, any mention of a statistic result causes at best
a compasionate smile. But this book changed that for me and I'd
like to learn more about this topic.

Rating:
Summary: Humane biographic sketches of numeric people
Review: David Salsburg has written gracefully engaging and humane sketches regarding some people who made diversely valuable contributions to methods for analyzing data.

I knew a number of people profiled in this book. A prior tenant of the house in which I grew up was biometrician Chester Bliss; from Salsburg, I learned that Bliss lost his job during the Depression, lived with R. A. Fisher for a few months in England before finding a job at the Leningrad Plant Institute. He barely escaped Russia in advance of a bloody Stalinist purge. I never would have suspected that this good-natured, elderly statistican had had such an eventful life.

Another chapter concerns Princeton Professor Samuel Wilks. Unbeknowst to Salsburg, since this has been otherwise unreported, at the time of his death in 1964, Wilks headed the Science Advisory Board for the U.S. National Security Agency. (I learned this from looking in the archive of Wilks' professional papers.) A small number of people within the Princeton mathematics department contributed quietly to U.S.-British efforts to read German codes during World War II and continued this activity during the Cold War. Wilks recruited my father to Princeton in the early 1950s. I was glad to read the chapter on Sam, who died when I was eight.

I also enjoyed the chapter on my uncle, John W. Tukey. Likewise, I enjoyed a profile of the English cryptologist and Bayesian statistician I. J. "Jack" Good. One of Tukey's more noted contributions to the advancement of the information age was the Fast Fourier Transform algorithim, which enabled digital computers to solve certain problems that formerly required analog computers. Tukey's 1965 FFT paper, with IBM programmer J. Cooley, draws from a paper by Good.

I enjoyed the mention of diverging opinions among statisticians during the 1950s regarding the potential health impacts of smoking. Fisher and Tukey's friend Mayo Institute biometrician Joseph Berkson were among those skeptical of studies which argued that smoking increased incidence of cancer. Salsburg suggests that a 1959 paper by another friend of Tukey's, Jerome Cornfield, was influential in shaping this debate.

Salsburg is a fine, highly readable writer. I believe that for years he was involved in a program at the University of Connecticut to record interviews with statisticians for the history of science. From these and other sources, he has painted pictures, going behind the numbers and mathematics in professional writings to capture something of the flavor of the authors as people. In so doing, he has done their memories kind service and helped explain, in accessible, non-technical ways, why interpreting data is important, in diverse ways.







Rating:
Summary: A laidback "Men of Mathematics" for statisticians
Review: David Salsburg's book "The Lady Tasting Tea: How Statistics Revolutionized Science in the Twentieth Century" (W.H. Freeman & Co., 340 pp., $23.95) celebrates the lives of two dozen great statisticians.

Short biographies of statistical innovators -- such as Francis Galton, Karl Pearson, Edward Deming, John Tukey and the most important of all, Ronald A. Fisher -- might seem of limited interest. Yet, over the past century, statisticians probably have done more to help us understand the real world than philosophers, who are endlessly profiled in countless books.

When discussing what has helped him in his work, Nobel Laureate physicist Stephen Weinberg has undiplomatically referred to "the unexpected uselessness of philosophy," while praising the "unexpected usefulness of mathematics."

The fecklessness of philosophy stems in part from the anti-statistical bias of the central tradition in European philosophy. Going back to Plato, philosophers have tended to assume that reality is based on abstract essences that could be described by geometry or words. In truth, though, the natural and human worlds appear to be probabilistic affairs. Statistics have thus proven crucial for describing subjects as commonplace as differences in human intelligence, as esoteric as quantum mechanics, and as life-or-death as the testing of new medicines.

This ignorance of statistics also plagues our public life. Veteran pundit James J. Kilpatrick has rightly argued that young journalists absolutely ought to study statistics in college. For instance, the press is constantly fouling up stories on topics as important as health or race because reporters don't understand that when a scientist says that "A correlates with B," he does not necessarily mean "A causes B." The other three possibilities are: 1. "B causes A." 2. "Something else causes both A and B." Or, 3. "A and B aren't actually related, they just looked that way because of random luck or a mistake in our study."

The founder of modern nursing, Florence Nightingale, said, "To understand God's thoughts, we must study statistics, for these are the measure of His purpose." As the inventor of the pie chart, which she used to show that bad medical care was killing more British soldiers than enemy bullets, she makes a brief appearance in Salsburg's engaging "The Lady Tasting Tea."

The whimsical title refers to a Cambridge University tea party at which a lady insisted, "Tea tasted different depending upon whether the tea was poured into the milk or whether the milk was poured into the tea." Most of the scientists attending thought this nonsense, but the great R.A. Fisher immediately devised a careful experiment that was largely capable of ruling out the effect of random luck. In Fisher's experiment, the lady correctly identified each cup.

Fisher published two crucial books in 1925 and 1935 that showed scientists for the first time how to design experiments that would produce statistically valid results.

To avoid scaring off readers, Salsburg left out all mathematical formulas, but that's a little like a history of art without pictures. Still, for anyone somewhat familiar with the main statistical techniques, this is a pleasant introduction to the men and women behind them.

Of course, statisticians generally try not to lead lives of lurid drama.

Yet, quite a few were persecuted by Hitler, Mussolini, and Stalin.

For example, a brilliant agricultural statistician named Chester Bliss couldn't find a job in America during the Depression, so Fisher landed him a post at the Leningrad Plant Institute. One day, his Russian girlfriend told him that the Communist Party had decided he was an American spy.

As his inquisition began, Bliss immediately went on the offensive, denouncing the communist experts for bad statistical techniques. He also called communism "the gospel according to Saint Mark and Saint Lenin." Astonished, Stalin's minions decided he was too honest to be a spy. So, the communists left him alone for months until they eventually realized that while he wasn't a spy, he was an anti-communist. He had to flee for his life.

The Stalinists were even more offended by the discipline of statistics than were the Nazis and Fascists. Salsburg describes why in a passage of black comedy:

"The mathematical concept of a 'random variable' lies at the heart of statistical methods. The Russian translation for 'random variable' is 'accidental magnitude.' To the central planners and theoreticians, this was an insult. All industrial and social activity in the Soviet Union was planned according to the theories of Marx and Lenin. Nothing could occur by accident. ... The applications of mathematical statistics were quickly stifled."

Salsburg makes clear that the early statisticians were largely interested in developing techniques for studying the inheritance of intelligence, an inquiry that continues to attract furious denunciations even today.

Francis Galton -- who invented fingerprinting, the weather map, and the silent dog whistle -- was the smarter half-cousin of Charles Darwin. Their common grandparent was the near-genius Erasmus Darwin, who had proposed his own version of a theory of evolution. Not surprisingly, Galton was fascinated by how intelligence tends to run in families. In 1869, Galton wrote the first book on the subject, "Hereditary Genius."

To aid his research, he invented the correlation coefficient and the concept of "regression to the mean," which explained why smart parents tend to have less smart children. Galton invented the term "eugenics" to describe the now highly unfashionable field of studying how to improve the human genetic stock. He suggested encouraging the finest young men and women to marry.

Fisher, in fact, was such an enthusiast for eugenics that during World War II he was falsely accused of being a fascist and blocked from helping with Britain's war effort. Fisher's belief in the value of eugenics led him to become perhaps the leading mathematical geneticist of his generation.

Advances in the Human Genome Project, genetic engineering, and sperm and egg selection are now beginning to make it feasible for couples to choose some of their child's genes. So, the controversies over eugenics are beginning all over again. But pro or con, anyone attempting to understand the coming impact of the new genetic technologies will need to use the statistical techniques invented by Galton and Fisher. -- Steve Sailer

Rating:
Summary: A light read on modern statistics
Review: David Salsburg's goal is to tell a story of how statistics changed the philosophy and practice of science in the 20th century. He starts with an anecdote in the late 1920s about designing an experiment to find out if a person can tell the difference between a hot drink made by pouring milk into tea or pouring tea into milk (hence the book's title). The first eight chapters describe the introduction of statistics and the concept of designing experiments in Great Britain from the start of the 20th century up to the 1930s by Karl Pearson and R. A. Fisher. Chapters 9 to 19 tells of the spread of the use of statistics around the world and the efforts to formalize statistics and probability. Chapters 20 to 28 covers the modern application of statistics in science and industry. The last chapter discusses the meaning and utility of statistics and probability. The book includes a timeline of events and people, and an annotated bibliography.

The earlier chapters on Pearson and Fisher are more coherent than the middle ones which dip into the work and lives of different mathematicians, and the final chapters can be read individually. Salsburg should have had narrowed his scope rather than try to cover so many topics and people in such a short book. There are some annoying repeated remarks and relatively unimportant characters, for example, Henri Lebesgue's slight of Jerzy Neyman is mentioned thrice and Churchill Eisenhart appears twice for no other reason than as the person who did not meet Karl Pearson.

The book is an easy and light read about mathematicians influential in the development and application of modern statistics in the 20th century. Readers trained in statistics and with an interest in the history of its development would enjoy it most.

Rating:
Summary: Yes, But What do the Numbers Mean?
Review: I am a high school math teacher who has long thought that statistics is more important in the curriculum than calculus. This book is one I can now recommend to colleagues and parents who question the importance of the subject. (Usually these adults had miserable experience with statistics courses in the past where formulas and computation, rather than understanding, drove the class.)
This book accomplishes two things. First, it conveys the development of statistics in the 20th century as the science of science - i.e. how experiments and surveys form the basis for knowledge and how to evaluate that knowledge. Second, it puts a human face on those who contributed to the field. The author's stories of Fisher vs. Neyman are wonderful. I especially appreciated how Salsburg relates the role of women in the field. They were often would-be mathematicians who were directed into statistics as a more "appropriate" field for women. Fortunately, as government use of statistics expanded, women civil servants were often already in place to provide quality analysis.
This book will probably not be widely read, but it should be...especially by scientists, journalists, and teachers.

Rating:
Summary: Good reading - but might be lacking an appropriate audience
Review: I am pleased that someone in the area of statistics has followed the example of those authors who have attempted to outline the history - and to some extent the philosophy - of math (Eric Bell) and physics (Timothy Ferris). While statistical methods form the basis for so many areas of cutting-edge research today, when it comes to the general public, statistics is probably one of the most unappreciated sciences (read: lies, damned lies, and statistics). Additionally, those in the statistical sciences (I'm a graduate student in statistics) many times overlook the development and hence the philosophy and thinking behind many of the methods they use. Unfortunately, statistical methods are thus sometimes put to use by someone who is intent on using a tool without considering the implications or liabilities of using that tool.

Hence, the topic and motivation of this book are needed. But I wonder whether the book has an appropriate audience, especially one outside of the statistical sciences. As I read through the book, I had to wonder whether I would have understood a single thing had I not had the background in statistics. True, the author avoids use of mathematical formulas. But telling the reader that so and so developed goodness of fit and that so and so developed maximum likelihood is not going to be very helpful if that reader doesn't know conceptually what these things are. To be fair, the author does go into more detail than that, but I still wondered whether the general public would get any of it.

As a statistician, I did enjoy reading the book and I obtained a good amount of cohesive information that would have been very difficult to find elsewhere. My only complaint about the book is that it was not more mathematically or philosophically rigorous.

But then I don't think such a book can provide both rigor to one audience and ease to another. I'm just not quite sure which audience the book was intended for.

Rating:
Summary: Excellent description of how statistics was founded
Review: I have taken courses in statistics, taught it many times and solved several statistical problems that have appeared in journals. But until I read this book, I never really thought about it in so deep and philosophical a manner. Which is most unusual, in that it is a book written to a popular audience. Some of the very deep and critical problems raised consider questions such as, "How do you deal with outliers?" An outlier is a data point that differs from the others by a great deal. The fact that it is a data point means that it is part of the sample, but the large differences from the others means that there are valid reasons to consider it flawed. Given these differences, including or excluding an outlier can lead to substantial changes in the results.
Other issues concern the accuracy of measurement, for example, when can specific tests be applied and what consequences can be associated with the results. We saw an example of such complexity in the 2000 presidential election in the United States. The vote was essentially a tie, with the differences being well within all possible measures of sampling error. As some of the wiser news commentators pointed out, it is impossible to count every vote, an election is only an approximation of the true, unknown value. No statistician could have said it better.
Given the context, Plato's idea of the abstract form appears in this history of the development of statistics as a discipline separate from mathematics. A statistical sample is only an estimate of a value that will never be known. The key is to get an approximation that is close enough to be usable in whatever the current context is. In this respect, statistics is like engineering, where the interest is in getting usable, rather than precise information.
The author also describes many details of the historical environments that the principal early statisticians worked in. Repressive governments such as...Germany, ...Italy and the communist Soviet Union provided the backdrop of the actions of many of the people who founded statistics. While the sentiments of the author are clear, he does a good job in avoiding overt political statements.
What I liked best about the book was the clear description of the life and career of Ronald Aylmer Fisher, a man whose genius is rarely spoken of in histories of science. And yet, some of the ideas that he expounded are the basis for many of the decisions that are made in our modern society. All new medications must pass rigorous statistical tests for efficacy and safety, and virtually every scientist must subject their data to some form of statistical analysis.
This is the most interesting book on statistics that I have ever read. It caused me to think about the underlying philosophy of statistics in ways that I had never done so before. Furthermore, it is written at a level where non-mathematicians/statisticians can understand it. I soundly recommend it for personal enjoyment as well as for any course in the history/philosophy of science or statistics.

Published in Journal of Recreational Mathematics, reprinted with permission.

Rating:
Summary: I don't like math but this book kept me reading
Review: I thought this was a great book. The author uses interesting examples for the use of statistics in every day life and we discover how this complicated science has many everyday applications

Rating:
Summary: Noble effort, and entertaining.
Review: It should come as no surprise to any reader that a 300 page collection of anecdotes might fall a bit short in realizing the implied goal in Salsburg's subtitle. He attempts to explain the paradigmatic shift in science from a Newtonian determinism to a probabilistic worldview by focusing on the statisticians themselves. The reader is often left with a desire for more - either more explanation of the paradigm shift or more anecdotes.

Nonetheless, I found this volume entertaining. I was fascinated by the newness in this field. Certainly nothing in my education led me to believe that virtually every aspect of social science research and statistical analysis is a 20th century invention. Who would have thought that the essence of 21st century social science research would be so well-anchored in agricultural studies and, perhaps most importantly, in the quality control efforts by master brewers at Guinness?

Salsburg intends to write to a non-statistical audience in language that can be understood without mathematic symbols. In this he is only partly successful. He does avoid technical symbols and most technical jargon, but in doing so he is often too vague to make his point clear. Even with three years of graduate statistics (from a social science perspective), I often found myself unsure of his explanations.

In the final analysis, Salsburg's description of the "statistical revolution" in science is really more of a sketch than a portrait. The significances of a shift from certainty to probability cannot be easily explained, but I will give him credit for trying to do so. That he is able to deal with this shift without explicitly commenting on the implications of this shift for religion, values, meaning, and justice is perhaps one of this book's major strengths.

Unfortunately, Salsburg concludes with a critique of the statistical revolution that may weaken the impact of his stories. Those desperately holding onto a Newtonian worldview could use this critique to discount 20th century science, especially social science. If, as Salsburg suggests, we are on the cusp of another paradigm shift, any post-statistical revolution is unlikely to be advanced by those continuing to resist the statistical one.

Rating:
Summary: Recommended for leisure browsing as well as study
Review: Lady Tasting Tea is an unusual guide which explains how the statistical revolution in science came about, examining how statistical modeling examples were developed and used. Unique to Lady Tasting Tea is an exploration which includes no math formulas and assumes no prior grounding in math concepts, statistics or math history; making it quite accessible to lay readers, and recommended for leisure browsing as well as study.