Throughout the 1970's, big-name sociologists with impeccable methodological and statistical credentials sought to persuade the discipline's journeymen that they should learn econometrics. The two most influential proponents of this view were the social statistics luminaries Hubert Blalock and Otis Dudley Duncan. Blalock was more optimistic than Duncan with regard to the ultimate payoff, but Duncan was more arrogantly dismissive of those who failed to heed his admonition.
In response, sociology and related social science journals became much more densely quantitative. Many social scientists, as a result, felt as if they had been reduced to obslescence. After all, econometrics and the other new quantitative tools, especially path analysis, which had come to dominate the discipline were difficult topics under the best of circumstances, and most social scientists lacked the mathematical training to tackle the best known econometrics texts, such as those by Jack Johnston, Jan Kmenta, and Arthur Goldberger. Many social scientists had been introduced to the econometrics mainstay, regression analysis, but not in this highly technical form.
Fortunately, the decade of the '70's also saw publication of Damodar Gujarati's introductory econometrics text, as well as the first edition of Peter Kennedy's Guide to Econometrics. Gujarati's book presented much the same material as his more insistently mathematical colleagues, but in a much more accessible form. His book could actually be used for self-instructional purposes, enabling less methodologically astute social scientists to finally figure out what was going on.
Kennedy's book was a forest-for-the-trees antidote to the mathematically dense and detailed texts, a book that enabled social scientists and other readers to identify topics that were of central importance and those that were ancillary details.
As with Gujarati, Kennedy wrote in accessible language and provided motivated readers with an overview of econometrics, enabling them to see what all the fuss was about. By including general notes and technical notes at the end of each chapter, Kennedy assured that his book was of value not not only to those of us who were less mathematically favored, but to those for whom use of econometrics was an everyday activity, one they had pretty well mastered.
In additon, while the first edition of Kennedy's book ran 175 pages, the most recent (sixth) edition is a full 575 pages. This reflects the fact that, while the book continues to provide an accessible overview of econometrics, it is also a comprehensive catalogue of regression analysis correctives. Kennedy explicitly acknowledges that his objective is to compile an accessible repository of the rapidly growing list of tests and procedures available to make regression analysis more generally applicable and informative. Anyone interested in the history of econometrics over the past thirty years would do well to begin with Kenndy's book.
Even for those of us for whom this stuff does not come easily, Kennedy's text is an invaluable reference. For the newcomer, it remains a fine overview of econometrics and a useful adjunct to any basic text. When the seventh edition comes out, it will be interesting to see what Kennedy makes of the near-obsessive concern with instrumental variable methods of causal analysis as presented, for example, in Angrist and Pischke's Mostly Harmless Econometrics.
As an addendum, Peter Kennedy is no longer with us, so unless his publisher recruits a co-author, I assume there will be no seventh edition. Finding a co-author as dedicated, knowledgeable, and who writes as well as Kennedy would be a tough job.