As a new Ph.D. student almost a quarter of a century ago, I was introduced to the study of identification by Franklin Fisher, who devoted a large part of the required course in econometrics to the problem of identification of linear simultaneous equations models. Some of my research from the early 1970s through the middle 1980s concerned the identification of discrete response models. But the perspective on identification presented in this book took shape more recently.
In the spring of 1987, Irving Piliavin stimulated me to examine the basic probabilistic structure of the selection problem associated with survey nonre-sponse. Later, after assuming the directorship of the Institute for Research on Poverty, I found myself grappling with the selection and mixing problems that arise in the evaluation of social programs and with the reflection problem faced in the study of neighborhood effects. My research on these subjects yielded a stream of new findings, and I gradually came to see a set of themes that warranted development into a book. Throughout this period of discovery I had the immense good fortune of being able to discuss ideas with, and receive encouragement from, my colleague Arthur Goldberger.
In 1992 I was invited by the American Sociological Association to present some of my work on identification at its annual meeting and to write a synthesis article for Sociological Methodology (Manski, 1993a). I am grateful to Clifford Clogg, Robert Mare, Peter Marsden, and William Mason for their roles in creating this opportunity for me to organize the disparate strands of my research into a more coherent whole.
With the article completed, writing a book seemed less daunting. The perfect environment for composing a first draft was provided by the Center for Advanced Studies in the Behavioral Sciences, where I was able to work with close to total concentration during the academic year 1992—93. I appreciate the financial support that made this possible, obtained from National Science Foundation grant SES90-22192 and from the University of Wisconsin Graduate School.
Arthur Goldberger, Jeffrey Dominitz, and Elizabeth Uhr have offered many constructive substantive comments and editorial suggestions on a draft of the manuscript. It has been a pleasure to work with Michael Aronson and Elizabeth Gretz of Harvard University Press.
June 1994