Like many political methodologists, I have been interested in the ecological inference problem for a long time and have benefited from innumerable conversations with many colleagues. I wrote several papers on the subject while an undergraduate, and my first paper in graduate school and my first paper after graduate school attempted to make inferences from aggregate data to individuals. My thanks to Leon Epstein, Bert Kritzer, and Jerry Benjamin for teaching me about the problems in the literature and the more serious ones in my papers. For most of the years since, I followed the literature while working on other projects, but my interest was heightened by a grant I received at the end of the 1990 redistricting process from a now-defunct nonpartisan foundation. This grant included a donation of the largest set of U.S. precinct-level electoral data ever assembled in one place (originally collected to aid minorities during redistricting). The U.S. National Science Foundation provided another key grant (SBR-9321212) to enable me over the next several years to clean these data, merge them with U.S. Census data, and develop ecological inference methods for their analysis. My thanks goes to officials at the NSF Programs in Political Science (Frank Scioli, John McIver), Geography (J. W. Harrington, David Hodge), and Methods, Models, and Measurement (Cheryl Eavey) for arranging the grant, and for a creative conspiracy to introduce me to geographers working on related problems. I find that new methods are best developed while analyzing real data, and these data have proved invaluable. Thanks to my colleague Brad Palmquist, who joined with me in leading the data project soon after he arrived at Harvard, and to the extraordinarily talented crew of graduate students--Greg Adams, Ken Benoit, Grant Emison, Debbie Javeline, Claudine Gay, Jeff Lewis, Eric Reinhardt, and Steve Voss--and undergraduate students--Sarah Dix, Jim Goldman, Paul Hatch, and Robert Hutter--for their research assistance and many creative contributions.
Nuffield College at Oxford University provided a visiting fellowship and a wonderful place to think about these issues during the summer of 1994; my appreciation goes to Clive Payne for hosting my visit and for many interesting discussions. The John Simon Guggenheim Memorial Foundation deserves special thanks for a fellowship that enabled me to spend an exciting year (1994-1995) devoted full time, almost without interruption, to this project. The departments of Political Science and Geography at the University of Colorado provided a forum in September 1994, so that I could discuss my initial results. I presented the final version of the statistical model at the March 1995 meetings of the Association of American Geographers, and in May I presented this model along with most of the empirical evidence offered here in the political science and statistics seminar sponsored by the Institute for Governmental Studies at the University of California, Berkeley. My sincere thanks to Luc Anselin for very insightful comments at the geography meetings, and to Andrew Gelman for going beyond the call of duty in helping me correct several proofs and work out some computational problems following my Berkeley presentation. My first exposure to tomographic research came from Andrew's dissertation work at Harvard, and my understanding of its use as a heuristic device for portraying the model and developing diagnostic procedures came during our conversations and Andrew's many insightful suggestions, some of which are also acknowledged in the text of the book. Without what I learned from Andrew during our collaborations on other projects over the last decade, this book would probably look very different.
I am indebted to the exceptionally talented scholars who attend the always lively summer meetings of the Political Methodology Group. They have provided many years of encouragement, perceptive feedback, and numerous new ideas. The meeting in July 1995 was no exception: Chris Achen was my formally designated discussant, and numerous others were very helpful at the meetings and almost continuously thereafter. Mike Alvarez, Neal Beck, and John Freeman provided written suggestions and answered numerous questions about many topics. John also gave me some especially helpful comments as my discussant at the American Political Science Association meetings in August 1995. Neal Beck's remarkable ability to read my manuscript and simultaneously to write trenchant e-mail comments and respond to relentless further inquiries about it kept me very busy and improved the final product immeasurably; I have learned a great deal over the years from our frequent electronic communications. Thanks also to Henry Brady, Nancy Burns, Gary Chamberlain, Charles Franklin, Dave Grether, Jerry Hausman, Ken McCue, Doug Rivers, Jim Stock, Wendy Tam, Søren Thomsen, Waldo Tobler and Chris Winship for helpful conversations, to Larry Bartels, Gary Cox, Dudley Duncan, Mitchell Duneier, David Epstein, Sharon O'Halloran, and Bert Kritzer for insightful written comments, to Sander Greenland for helping me learn about epidemiological research, and to Danny Goldhagen for help with the literature on the Nazi vote.
My colleagues Jim Alt, Mo Fiorina, Brad Palmquist, and Sid Verba were very helpful throughout the process, providing many comments, numerous insightful discussions, and much encouragement. Nothing beats the off-hand comments at the mailboxes in the Department of Government. Brad Palmquist's careful readings of the manuscript, many useful suggestions, and deep knowledge of ecological inference saved me from several blunders and improved the final product substantially. Alex Schuessler and Jonathan Katz provided very insightful comments on painfully early versions of the manuscript. Curt Signorino was far more than my lead research assistant; he provided important, perceptive reactions to many of my ideas in their earliest forms, helped me reason through numerous difficult statistical and mathematical issues, corrected several proofs, and helped me work through a variety of computational issues. My appreciation goes to Chuck D'Antonio, Yi Wang, and William Yi Wei for their computer wizardry.
Alison Alter, Ken Scheve (Harvard); Barry Burden, David Kimball, Chris Zorn (OSU); Jeff Lewis, Jason Wittenberg (MIT); Fang Wang (Cal Tech); and other faculty and students deserve special thanks for letting me experiment on them with alternative drafts and computer programs and for their continual inspiration and suggestions. Most of these political scientists participated with me in a ``virtual seminar'' held over the 1995-1996 academic year: they helped me improve the manuscript by identifying passages that were unclear, unhelpful, or untrue, and I tried to return the favor with immediate explanations via e-mail of anything that was holding them up. No doubt I got the better end of this bargain!
When the project was farther along still, I had the great fortune to receive comments on this work from presentations I gave at Michigan State University's Political Institutions and Public Choice Program (February 16, 1996), the University of Iowa (February 29, 1996), the Harvard-MIT Econometrics Workshop (April 4, 1996), Columbia University's Center for Social Science (April 5, 1996), the University of California, Santa Barbara (April 10, 1996), the California Institute of Technology (April 11, 1996), the University of California, Los Angeles (April 12, 1996) and the ICPSR program at the University of Michigan (17 July 1996). I am also grateful to Jim Alt, Steve Voss, and Michael Giles and Kaenan Hertz for graciously providing access to their valuable (separate) data sets on race and registration. I especially appreciate the talented staff at Princeton University Press, including Malcolm Litchfield and Peter Dougherty (editors), Jane Low (production manager), and Margaret Case (copy editor), for their professionalism and dedication.
Elizabeth has my deepest appreciation for everything from love and companionship to help with logic and calculus. The dedication is to our daughter, who learned how to laugh just as I was finishing this book. I am reasonably confident that the two events are unrelated, although I will let you know after she learns to talk!