Trust in Numbers: The Pursuit of Objectivity in Science and Public Life (Princeton Paperbacks) Kindle Edition

Theodore Porter makes the stimulating observation that "objectivity" is more prized in a democratic political culture based on competing interests than in autocratic cultures. He backs this up with a great deal of evidence -- perhaps too much -- by comparing two bureaucracies, the Corps des Ponts et Chaussees (CPS) in 19th Century France, and the Army Corps of Engineers (ACE)in 20th Century America.The CPS was secure in its elite status; no one outside their community was second-guessing their judgment. Their rigorous mathematical education served more as evidence of a general level of culture, like knowledge of Greek or Latin, than as their main lens for viewing the world. By focusing on CPS debates about the appropriate placement of rail lines in France, Porter shows that CPS engineers used quantitative arguments as only one persuasive technique among many, and as only one type of input out of many that helped them to formulate their professional judgment. OTOH, the ACE's decisions about dam placements were under constant attack from various industry groups, other Federal bureaucracies, and the Congressmen who represented those other interests. As a result, the ACE gradually took refuge in cost-benefit analyses that were "objective" in the sense that they followed fixed and published rules (even if there was still a lot of opportunity to fiddle the figures). Ultimately, the ACE became proud to ignore "intangibles" -- the stuff that doesn't fit so easily into the rules -- even though, it is hinted, these should have been some of the most important considerations for the dam-placement decisions. This put the ACE in contrast not only to the CPS but even to 19th Century actuaries, whom Porter shows to have relied on their statistical tables as only one input among many other, less quantifiable judgments in making decisions about whom to insure (a quite surprising and interesting revelation).Porter relied on many arcane and imaginative primary sources for his discussions of CPS and ACE, including actuarial trade journals, memoirs of the construction of rather discrete railway segments, and even 19th Century French farces (to illustrate popular attitudes toward statisticians). However, the extreme length and depth of his descriptions of the debates about rail lines and dams were overkill to make his case, since the substance of the decisions was less important than the bureaucracies' style of argumentation. These passages are also often excruciating to read, since the book doesn't include a single map, diagram or illustration of any kind. I really wished he'd omitted many of the details or at least relegated them to an appendix.At the end of the book, Porter expands his thesis to include the social construction of quantities in social science and laboratory science, especially physical science. That's led to this book having been cited hundreds of times since its publication in 1994. (It was a cite in such a context that led me to read the book, in fact.) But his presentation of the expanded thesis, while suggestive and intriguing, is relatively rushed. It relies mainly on secondary sources and occupies only the last 30 pages out of 230 pages of text. Here is where more detail would have been a blessing.As for style, it's quite heavy and stiff throughout the book. Unlike Philip Mirowski, many of whose themes and concerns are similar to Porter's, but who manages to be funny even when he's being pompous, Porter always sounds here as if he takes his subject too seriously. (Maybe he's lightened up since writing this work?) Compounding the stylistic problem, or maybe causing it, is that Porter seems never to have met a cultural theorist he didn't like. There's even an unnecessary, but politically correct, digression on multiculturalism, complete with reference to Michel Foucault. And expect to be puzzled by the book's concluding sentence unless you know the difference between Gesellschaft and Gemeinschaft (in German, no less), which is nowhere explained or even mentioned prior to this culminating point. That Porter ends with such an erudite reference at the very moment when he should be hitting the ball out of the park is an unfortunate epitome of the book's mandarin approach.

The bulk of this book consists of detailed historical accounts of the influence of the French Corps des Ponts et Chaussées and the US Army Corps of Engineers on public policy. However, despite the length, these are anecdotal accounts, not true histories. They are intended as case study illustrations of the author's main contentions laid out at the beginning of the book, and transformed into recommendations at the end.A textbook account of the scientific method stresses hypothesis formation and testing, independent replication of results, formal peer-reviewed publications and consensus. The author argues that scientific progress is instead driven by informal networks of personal influence, and that the textbook method is something invoked when findings are attacked as unscientific. Of course, experiments are essential to science, but seldom in the strict hypothesis-testing framework taught in statistics courses. These methods were introduced not to make experiments more rigorous, but when experimental work was parceled out to less-skilled and less-trusted workers. For example, when astronomers or trusted assistants made observations directly, they used judgment to throw out anomalous observations made under bad conditions and to aggregate multiple measurements. Much later, when hundreds of workers were hired to examine thousands of photographic plates for specific data, strict statistical procedures were developed for classification of outliers and for combining measurements. This was industrial quality control, not scientific theory.The author argues that replication is seldom anything like independent. Researchers send assistants to work in other labs, and share reagents and measuring equipment and biological samples, among other things. Strictly independent replication from scratch is far too troublesome to be a routine part of science. Peer-reviewed articles appear long after important results are disseminated among working professionals, or in other cases, long after everyone has dismissed or ignored the findings. Claims of "consensus" are the last refuge of scoundrels. When true consensus exists, no one has to rely on it for support, and not incidentally, it is often wrong.As a revisionist account of the scientific method, the book presents a provocative picture that will deepen understanding of the philosophy of science, but is overstated on its own. It is more on target as an account of how claims of scientific impartiality and rationality can be used and abused to affect public policy, and also how quantification can either clarify or cloud political debate.The biggest value of this book is to shake up sloppy and thoughtless opinions about the role of science in politics, and to provide extensive practical examples for consideration. The author fails in his attempt to make a specific coherent case, but succeeds in demonstrating the incoherence of other viewpoints. It is a book to help begin a search for answers, not a book with all the answers. Two other excellent and relevant books in this area are  The Politics of Large Numbers  and  Uncontrolled .

Theodore Porter is perhaps the most qualified person today to speak about the increasing mathematization and staticization of our pubic policy debates. He explains how and why we have let math and statistics come to dominate our choices about what is best and what works. Who today would argue that it is morally wrong to close a manufacturing plant and ruin a town when a statistical study can can show that the plant is not economically viable and would not return to profitability? Porter explains why we so "trust in numbers" and how we reached this point. The book is also a very entertaining read!

The historical overview and case studies of how numbers and statistics become part of today's management practices.

Essential for official statistics undergraduate