Mel Hinich, professor of political science and economics at the University of Texas at Austin and Mike Hogg Professor of Local Government, died in a tragic fall the evening of September 6, 2010.
Melvin J. Hinich achieved an international reputation in four academic disciplines: economics, engineering, political science, and statistics. He published pathbreaking contributions in seven books and more than four hundred journal articles, ranging over 40 years, with an endless energy for work and a childlike curiosity about the science of almost everything. Hinich's scholarship blended technical virtuosity, theoretical depth, interdisciplinary sweep, and a keen eye for the main chance in terms of substantive importance. But Mel was not simply a bright but easily distracted scholar with many interests. Rather, he was a scientist, a scholar who found most problems interesting, and he was capable of making connections across fields because so many problems share a deep logical and mathematical structure.
Mel received a BS and an MS in mathematics from the Carnegie Institute of Technology in 1959 and 1960, respectively. He earned his Ph.D. in statistics from Stanford University in 1963, working under Dr. Herman Chernoff. His thesis project involved the problem of estimating the properties of a recurring but unknown waveform in Gaussian data, and it proved useful enough to have the central theoretical result serve as the basis for a variety of civilian and military applications in signal processing.
Hinich began his academic career with the Graduate School of Industrial Administration at the Carnegie Institute of Technology in September 1963. He then circulated between academic teaching, research posts, and stints in settings where he could apply his knowledge to industrial and governmental uses. He held positions at the Bell Laboratories, the Columbia University–Hudson Laboratories, the Center for Naval Analyses, and the Naval Coastal Systems Center. His academic appointments have been at Carnegie Mellon University in the Graduate School of Industrial Administration, the School of Urban and Public Affairs, and the department of statistics. He moved to Virginia Polytechnic Institute as a professor of economics in 1974 and then to the University of Texas as a professor of government and economics in 1983. He also served as a research professor in the Applied Research Laboratories at UT–Austin, a position he loved because he was in close contact with scholars from different disciplines.
Mel's impact is easily illustrated both quantitatively and qualitatively. His work has been cited more than five thousand times in the professional literature, with 20 different papers and books in economics, political science, signal processing, and statistics all having received more than one hundred references. No one paper and, in fact, no one disciplinary contribution are the “main” result. Mel's contributions were broad and deep.
Qualitatively, Mel colleagues recognized his work by naming him to numerous honors and memberships in honorary societies. He was named Fellow of the Institute of Mathematical Statistics in 1973, appointed the Sherman-Fairchild Distinguished Scholar at the California Institute of Technology in 1975, named a Fellow of the Public Choice Society in 1988, elected president of that society in 1992, and named a Fellow of the American Statistical Association in 2002. Most recently, in 2008, Mel was named one of the 21 original Fellows of the Society for Political Methodology.
Scholarly Works
Mel's scientific contributions were so sweeping that we can only summarize them briefly. In political science, Mel made seminal contributions to the use of spatial analysis in analyzing politics and public policy. In collaboration with economist Otto A. Davis, Mel formalized a general model of the spatial theory of voting and elections. Although earlier researchers qualitatively identified the perspectives of that model, it was Mel's conviction that if a model or idea was to yield scientifically testable propositions, it required formalization. Today, thanks to that effort, spatial voting choice theory occupies a role in political science that is as central as consumer choice is in economics.
Mel, however, didn't formalize for the sake of displaying his considerable mathematical talents. What made him unusual was his insistence on staying close to the data in modeling exercises. He had scant patience with top-down modelers who had little appreciation for the data that Mel thought was required to give flesh and blood to the bones of an abstract model. No model was right, wrong, or beautiful to Mel; the only criterion was whether or not it was useful. And utility to Mel meant one thing: did it lead to further insights and more interesting work?
As the methods field splintered into “theory modelers” and “statistical modelers,” Mel resolutely bridged the gap and continued to use formal models as a way of understanding data and the underlying processes of politics. As a consequence, his work exhibited careful attention to the methods of statistical analysis and deftness in applying these skills to the analysis of empirical data. In a field often admired for its abstract models and arcane mathematics but condemned for its failure to connect to the real world, Mel had a special talent for linking theories to data. He was not only a pioneer of the spatial choice revolution, but he remains one of the most productive and innovative members of the field.
The Spatial Theory of Electoral Competition
Hinich was not merely one of the central figures in the early development of “spatial theory,” he was the central figure, the nexus of all important early collaborations. Spatial theory is a predictive and descriptive model based on the premise that the actors being studied prefer policies “closer” to their own goals and principles. Mel's initial contribution here was the construction of mathematically explicit, empirically focused models. More important, Mel did not see that initial contribution as an endpoint of research, but only as a beginning. Mel had little interest in producing an endless series of disconnected, mathematically elegant papers that lengthened his vita but offered nothing except empirically vacuous technical reformulations that would do little to advance a clear understanding of science. His core motivation was to develop a science of politics and to give its study an empirically connected theoretically grounded base.
The Davis-Hinich collaboration on the development of the multidimensional spatial theory of the electoral process uses a conception of voter utility that is proportional to the “distance” between the candidate or party being considered, and the voter's most preferred set of policies. The Davis-Hinich model makes the concept of distance mathematically precise by defining the weighted Euclidean distance metric and allowing for both differences in salience and connected or “nonseparable” preferences on different dimensions. In later work, the nature of the dimensions themselves was extended to take into account the (empirically demonstrable) latency of ideological or values-based dimensions that are linked, both in the political rhetoric of campaigns and in voters' minds, to real political issues.
Although spatial theory has its roots in the early writings of Anthony Downs and Duncan Black, it was really Hinich, with his coauthors Otto Davis and Peter Ordeshook, who developed the formal model on which spatial theory today rests. By inserting a matrix of weights into the Downsian model, and by demonstrating the ability of the extended model to test more general claims about politics, spatial theory was given a much-needed generality and a mathematical life. In the process, Davis, Hinich, and Ordeshook transformed the model from a curiosity into a practical and powerful modeling tool.
The publication of Davis, Hinich, and Ordeshook's 1970 article, “An Expository Development of a Mathematical Model of the Electoral Process,” like the publication of The American Voter (1960), was a defining moment for contemporary political science. It was a tour de force statement of the state of the field as of 1970—and was required reading for many years afterward, not replaced until the appearance of Enelow and Hinich's Spatial Theory of Voting in 1984, which has itself become a central work in the field.
All of this research not only carries the common name of “Hinich,” but it was all also collaborative. Mel was not engaged on some academic ego trip. His views of science included the argument that the most fruitful research requires collaborative efforts wherein coauthors test each other's ideas while contributing new ones in a context wherein true scientific advances occur as the product of the give and take of intellectual argument. And Mel was a scientist! A true scientist is one who attempts to understand real-world empirical phenomena. Unsurprisingly, then, his work in spatial theory directly led to issues of how one detects and statistically estimates spatial structure in voting data. In collaboration with Lawrence Cahoon, one of his Ph.D. students in the department of statistics at Carnegie Mellon, in 1978, Mel developed a metric multidimensional scaling methodology that is based on the parametric weighted utility model for voting choice.
This methodology was further developed by Hinich over many years and is now called the Cahoon-Hinich MAP Method. The MAP program takes information about voter preferences (thermometer scores) and constructs a latent “space” where the number and location of dimensions is determined by the data themselves. The output of the estimation is analogous to the results of the Poole-Rosenthal “space” of political conflict in the Congress in the sense that it estimates the contours of the space and the location of the political actors in that space. The difference is that the MAP constructs the spatial configurations based on stated preferences, rather than reconstructing those spaces from observable actions (for Poole and Rosenthal, voting). The Cahoon-Hinich methodology is a statistical method—one that provides statistical estimates of parameters along with the means for testing their statistical significance. Once again, as a scientist, Mel saw methodologies not as a path to producing “numbers,” but as a route to testing hypotheses.
Throughout the process, Mel thought of any idea or development as nothing more or less than a place to stand to develop more insights, theories, and tests of theories. His work was a springboard for Jones' development of a spatial approach to agenda-setting in which the salience measures (the elements on the main diagonal of the A matrix) could vary over time, shifting preference orderings in unexpected ways. When Jones tentatively presented his perhaps harebrained idea at a seminar run by Mel at the University of Texas in the mid-1990s, Mel, far from being critical, was enthusiastic—appreciating the underlying idea as an explanation, even though it could not be used in an ex ante predictive model.
Everything Is Linked
Mel's early work centering on signal processing addressed a problem that the Navy otherwise found intractable. The Navy had to decide between two alternative antisubmarine warfare (ASW) systems for detecting and locating enemy submarines. One alternative was a system of sonar devices scattered about the ocean that provided range-only data (distance from a detected object). The second, more expensive alternative provides both direction and range data. For a given cost, the first alternative allowed the ocean to be covered with many more units. As a result, it was not so clear which system was more capable. Such a problem presented a myriad of statistical and modeling problems, not least because the acoustic properties of the ocean are confounded by innumerable nonlinearities, thereby rendering any simplistic approach worthless. For Mel, the problem was like red meat before a hungry lion.
At the same time, Mel, in his inimitable style at Carnegie, would hold conversations with anyone about anything, provided that there was intellectual content to the conversation. And it was in some of those lunchtime conversations with Otto Davis that Mel saw the connection between his work in ASW and the problem of statistically estimating the positions of candidates in a multidimensional issue space when surveys provided “range-only” data in the form of thermometer scores.
The Cahoon-Hinich methodology of spatial analysis was a large additional step in estimating position using range-only data. This work made it possible for the spatial theory of elections to be empirically studied and tested in a wide variety of contexts. Since the publication of The Spatial Theory of Voting, which documents the methodology in detail, the Cahoon-Hinich methodology has been applied to elections in many countries, including Chile, Germany, Korea, Russia, Taiwan, Turkey, Ukraine, and the United States.
Mel loved working with co-authors. In addition to Davis and Ordeshook, Hinich began a long and fruitful collaboration with James Enelow in 1978 regarding both theoretical expansions of spatial theory and using the Cahoon-Hinich MAP program to estimate spatial maps of candidates in American politics using the thermometer scores from Inter-University Consortium for Political and Social Research (ICPSR) surveys. This collaboration produced the first book on the modern spatial theory of elections: The Spatial Theory of Voting in 1984. Then they produced an edited volume on spatial theory, Advances in the Spatial Theory of Voting in 1990.
After Hinich moved to the University of Texas's department of government, he began a collaboration with Michael Munger, an assistant professor in the department whose doctorate was in economics. Their extension of the spatial theory of elections resulted in several papers and three books: Ideology and the Theory of Political Choice (University of Michigan Press, 1994), Analytical Politics (Cambridge University Press, 1997), and Empirical Studies in Comparative Politics (Kluwer Academic, 1999).
In 1977, Hinich published a paper in the Journal of Economic Theory that introduced a model of probabilistic voting. The model has both theoretical and empirical implications, but one of the most interesting is that it illustrates how democratic systems can be stable and avoid the cycling that the open agendas spatial theory predicts. This model was an outgrowth of a simpler probabilistic voting model that Hinich had developed with Peter Ordeshook and John Ledyard several years before. This paper is still widely cited by political economy scholars. Hinich and Munger used this probabilistic voting theory to develop a type of general equilibrium model in Ideology and the Theory of Political Choice. This book unites three streams of thought in Hinich's work, because it is based on the idea of a latent, constructed dimension but rests on a model of campaign contributions as the vector through which elites can influence politics. The third theoretical connection with Mel's earlier work is equilibrium concept, based on probabilistic voting. By uniting these three separate bodies of theory, the book illustrates the predictions of a more general model of political competition in which elites are an important moving part but social construction of political belief and rhetoric continues to play a key role.
Mel's work influenced many generations of scholars and students in several fields. Howard Rosenthal writes, “Most of my empirical work owes its inspiration to Mel's awakening my interest in models and the spatial model in particular.” Much of Theo Panagiotidis work is based on Mel's studies with Doug Patterson on nonlinearities in stock market returns; Theo notes, “I have followed his writing and his work appears in most of my papers.” These are the human faces of the many scholarly citations his work has received over the years.
Hinich supervised a total of 16 Ph.D. students at Texas, several of whom have contributed to the theoretical and empirical development of spatial theory. His most recent and last student was Chih-Cheng (Almond) Meng, who finished his dissertation in the spring of 2010. Mel influenced many more UT students thorough his courses and service on their committees. Jan Box-Steffensmeier, a government department student in the 1990s, commented, “I knew right away that Mel was a scholar with incredible reach when my spouse, who was a grad student in the electrical engineering department, was also reading work by Mel Hinich.”
Work outside Political Science
In addition to his work in political science on spatial theory and estimation, Mel has a very serious presence in the statistical study of signal processing and time series and nonlinear dynamics. We will simply relay an anecdote. When Jones was head of political science at Texas A&M, he got a call from Manny Parzen, the distinguished statistician. Manny said that the statistics department was inviting Mel over to give a presentation on his time-series work and wanted to know if political science wanted to invite him for a talk. Jones said sure, that he had meant to do that beforehand. “Bryan, does Mel have a reputation in political science?” Manny asked. He was quite surprised to hear that the answer was yes.
Mel's early development of important, straightforward ideas about stationarity and regime shift in data lie at the core of many key results in time series, even today. He invented bispectral and trispectral tests for nonlinearities in stationary stochastic processes. His article, “Testing for Gaussianity and Linearity of a Stationary Time Series,” published in the Journal of Time Series Analysis in 1982, remains Mel's single most-cited article (only his “Spatial Theory of Voting” with Enelow has been cited more). He and Professor Douglas Patterson used his bispectrum method to discover nonlinearities in daily stock returns, which meant that the efficient market thesis based on Gaussian returns could not be correct. Their paper in the Journal of Business and Economic Statistics has been cited over four hundred times. Later, they used the Hinich bicorrelation and cross bicorrelation tests to show that nonlinearities in asset returns are episodic in nature. As Doug Patterson explains, “On most days, stock rates of return appeared to be drawings from a Gaussian distribution. On other days, the data was highly nonlinear.”
“Be Well, Mel”
Mel's commitment to science and his appetite for collaborators who could learn from him and teach are the central themes of his scholarly life. Moreover, when he was wrong, he not only admitted it, but he was quite capable of bringing on the individual pointing out the error as a co-author. In a 1965 seminar at Rochester in which Davis and Hinich presented their early work, Ordeshook offered an alternate proof to one of the theorems in the paper. Amazingly, after Davis had presented the main ideas, someone in the audience asked about that very theorem, whereupon Mel stepped to the blackboard and filled it in thoroughly. Upon the conclusion of the seminar, the graduate student approached “Professor Hinich” and asked if he could show him his alternative proof to see if there was a flaw in the argument. After writing it on the blackboard, Mel nodded his head and said, “Yes, that's better than mine.” Three years later, when the student entered the job market, Ordeshook learned he had a “wired” offer from Carnegie Mellon. Turns out, Mel remembered that Rochester seminar, remembered that student and his assessment that “we can do science together,” and ensured that Carnegie hired Peter Ordeshook when his Ph.D. work was done.
Personally, Mel was warm and generous and loved a good conversation on almost anything. A comment by one of his friends, Patricia Murrieta, is so typical of Mel. She commented,
I remember many hours of long and interesting conversations with him, even when we could have opposite ideas about politics and migration…. I really enjoyed listening about his trips, his life, his family, politics, education and many other things he would always be willing to discuss and share with us; including what was going on with our friends in Chile. Some of these wonderful conversations took place with a good bottle of tequila at our house in Mexico.
Boaz Golany summarized this complexity perhaps best of all:
Ironically, Mel was a professor of “government.” Instead, he should have been a professor of “peoples.” For the most part, he hated governments and was highly critical of the incapable (often corrupt) leaders who led them. But he really loved people—regardless of their race, gender, religion and any other dividing lines that exist in our societies.
But Mel could as well be harsh and acerbic, especially toward those whom he considered academic inferiors (and this invariably included almost any academic administrator he had ever encountered). As Harrison Wagner noted, Mel “had the unusual ability to make me feel smaller and less significant, without making me feel any the worse about myself … [but he also] was inclined to make dire forecasts of the future of the world, or the U.S., or the University of Texas, or the UT government department.” There was a bright, expansive Mel, a man in whom every remark seemed insightful and worthy of thought and reaction. And there was the dark Mel, who was capable of viewing the world through lenses that left one depressed. Not that Mel was wrong; he was almost always right. But it was the extensiveness of the negativity that could shock. The trick was to move from the latter to the former, engaging his intellectual curiosity and boundless humanity.
Adrian Van Deeman provided a wonderful anecdote. Mel visited him in The Netherlands frequently. Adrian commented, “He always was very warm and gentle at my home, in particular towards my children.” But once he took Mel to a party full of academic political scientists:
Mel started to talk around with the people, but he did not like it. His voice became louder, and I observed that he started to offend people. I never forgot what Mel said when we were back in the car: “Well Ad, that was wine, cheese and arrogance.” That was my dear friend Mel: warm, gentle, open, wholehearted, curious, eager to know, easy going with people, but also rough, offending, and disgusted by arrogance.
Mel apparently had trouble with water. Bob Molyneux, a student of Mel's at Virginia Tech and a long-time friend and colleague, wrote, “I picked him up at the airport for my dissertation defense—he had graciously agreed to serve on my committee—and he came down the escalator, dived for a water fountain and came up like a cresting whale with water all over his face—just so Mel, isn't it?” Gary Freeman supplied this anecdote.
Mel invited me over for a drink. Midway through our first single malt Mel got up to move the hose in his backyard. Mel was gone for what seemed an inordinately long time. Eventually I heard the back door open. Mel came in drenched from head to toe. Without a word, he sat down and reached for his glass as if nothing had happened. Mel had wrestled that garden hose and the hose had won.
His friends know that with Mel, alcohol and great thoughts intertwined like the double helix. Howard Rosenthal commented,
The 60s and 70s were leisurely times when scholars could think creatively as opposed to publishing large numbers of little papers. [At Carnegie Mellon] once every week or two, Mel, Peter Ordeshook, Jim Laing, I, and a few others would congregate one evening in someone's home to talk social science and drink a couple cases of Iron City, one case for Mel and the other for the rest of us.
Since the mid-1990s, Jones would sometimes drop by Mel's house in Austin with a six-pack of beer or would drive with Mel to a watering hole, which always led to a two- or three-hour discussion of all sorts of fascinating topics. With Mel, that was standard operating procedure. As all his friends and colleagues know, neither the quality of the drinks nor the quality of the food mattered—but the price mattered! It was the conversation that was critical. The week before he died and right after Mel had returned from Italy, Australia, and Chile, where he was working with collaborators, the two dropped by Jorge's for a drink or two and two hours of conversation. He seemed frail, leaning on Jones' arm for support as they entered Jorge's, but was immediately animated and excited with whatever ideas the two pursued.
On the day before he died, Mel talked with Bob Molyneux about their next visit and called up Munger to talk about a new chapter for the revised edition of Analytical Politics (Cambridge University Press). That revised edition should be finished soon and published at the end of 2011, a posthumous testimony to Melvin Hinich's continuing influence and focus on the development of a science of politics.
On the day he died, Mel sent Jones an e-mail with the classic Hinich sign-off: “Are you back in Austin? I am at home today. I teach tomorrow. Be well, Mel.”
Mel never quite felt fully appreciated; perhaps because his work spanned so many fields that none of us came anywhere close to grasping it, or perhaps more likely, because he was such a poor self-promoter. He would rather fume a little and then turn to the next interesting idea. The over seventy (as of October 10) of Mel's many friends, colleagues, and students from all over the world who have shared their recollections on a website that the government department at the University of Texas has established show that indeed he was appreciated as a scholar, colleague, mentor, and human being. Please share your insights into this fascinating, complex person at http://sites.la.utexas.edu/melhinich/.
The department of government has also established the Melvin Hinich Fund to support graduate student research. Members of the UT government department have donated more than $12,500 to this fund and have been joined by several other of Mel's friends and colleagues. Please join us in this effort. Contact Stuart Tendler (smtendler@austin.utexas.edu) for information or go to https://utdirect.utexas.edu/nlogon/vip/ogp.WBX?menu=LAGV. Be sure to note that the donation is for Mel Hinich and the department of government.
Mel is survived by his wife, Sonje; their daughter, Amy Leksana; and two granddaughters, Catlin and Rachel Leksana. Our condolences to them.
