In Memoriam David Freedman (March 5, 1938–Oct 17, 2008) — March 5, 2018 would have been his 80th birthday. I reproduce below an edited version of some memorial remarks I wrote, from
We would like to dedicate this issue to the memory of David Freedman, an outstanding
statistician whose legacy is closely related to the goals we would like to pursue in this journal. One element of this legacy is the importance of undergraduate teaching; attempts to do explain the relevance and importance of statistics to undergraduates transformed the thinking and research of David. Increasing fragmentation of knowledge has led to a situation where specialists have no idea of how their patch of expertise relates to other portions of econometrics, how the whole body of econometrics ties into economics, and how this body of knowledge relates to achieving broader human goals such as eliminating oppression, injustice, poverty and misery, and bringing happiness, joy, wonder and enlightenment into our lives. Teaching undergraduates is a useful antidote to this fragmentation. When we ask them to invest time and effort in learning difficult materials, we must justify this claim by showing them why it is useful in the context of real world examples. When we attempt to do this we will discover that, contrary to the impression created by our specialized education, no one has been there before us. That is, everyone in the knowledge field is a specialist and no one has a broad overview ranging from details of the theory to how these theories are applied in the context of serious real world applications which make a difference to the lives of people.
In his path-breaking undergraduate text Statistics, David documented the disastrous
consequences of this disconnect via many real world examples. A little exploration reveals how often bad theory with faulty assumptions using wrong types of data is used to guide policy. As a consequence, people who acquire the all round knowledge required to bridge the gap between theoretical knowledge and real world applications can make a big difference in changing the world for the better. David’s involvement in consulting and litigation testifies to his deep concern with using his knowledge to improve the lives of people. His more recent textbook Statistical Models explores how models work in the context of real world applications. Richard Berk, one of the commentators on David Freedman’s article in this issue, has provided a similar examination of regression models in his text: Regression Analysis: A Constructive Critique.
We would like to close with some advice to readers and contributors. John Hey summed up his experience of ten years of editorship of the Economic Journal as follows: “Many of the submissions do not appear to be written in order to further economic knowledge. … few economists ask themselves what are the crucial economic problems facing society. If they did so, they might well produce more relevant material.” Our brief moments on this earth are too precious to waste on pursuit of meaningless publications to add to our vitas. It will add meaning to our lives, depth to our knowledge, and create innovative and path-breaking research, if we make a serious attempt to serve humanity by solving the numerous real economic problems facing us globally.
Our first issue of IER is dedicated to the memory of David Freedman. “Limits of
Econometrics” may be his last professional article – he had written it at my request for this first issue. Unfortunately, he did not get a chance to revise it in response to the many
comments received. As a first draft, it may fall short of the standards of excellence set by
David in his numerous works – see his Berkeley website http://www.stat.berkeley.edu/~freedman/ for references – but, under the circumstances, it seems best to print it as is, in tribute to his memory. We are also publishing comments on this article by Arnold Zellner and Richard Berk.
In 2003, David Freedman was awarded the prestigious John J. Carty Award for the Advancement of Science by the National Academy of Sciences, “for his profound contributions to the theory and practice of statistics, including rigorous foundations for
Bayesian inference and trenchant analysis of census adjustment.” More details about these contributions are available from the Wikipedia entry
The obituaries cited below provide a broader perspective on his career and personality:
It is impossible to summarize his contributions, but I will focus on two issues which are
relevant both to the attached article, and to our underlying philosophy for this journal.
Close attention to real world applications, generated by demands of undergraduate teaching
and by consulting and litigation, changed the focus of David Freedman’s research from the
theoretical and abstract, to how these theories work in practice. His landmark text Statistics
(co-authored with Robert Pisani and Roger Purves) is based on serious examples from
economics, epidemiology, medicine, and social science. In his applied work, Freedman
emphasized exposing and checking the assumptions that underlie standard methods, as well as
understanding how those methods behave when the assumptions are false. Zellner’s comment
provides an econometricians perspective on the fundamental methodological issues which
govern the use of models in real life situations.
A central concern of Freedman was the disconnect between the requirements of real world
data analysis and the conditions under which models can produce reasonable answers. The
widespread use of models which make entirely inappropriate assumptions, together with
obliviousness to consequences of these errors was anathema to Freedman. His book on
Statistical Models discusses many professional highly cited articles which draw conclusions
not justified by the data, due to erroneous background assumptions. Because of these
critiques, many commentators remarked that if we take Freedman seriously, we will all be out of work. Richard Berk’s comment on Freedman’s article addresses this issue: how to proceed
if we take Freedman seriously.
It seems appropriate to close with the following quote from his colleague, Phillip Stark: “He
was just an extraordinary person and an extraordinary scientist. He was a truly exceptional
scholar-brilliant, meticulous and committed to truth.” His loss will be deeply felt in the
community of statisticians and econometricians.