Apsa-Cp Newsletter of the Organized Section in Comparative Politics of the American Political Science Association

apsa-cp Newsletter of the Organized Section in Comparative Politics of the American Political Science Association

Volume 12, Issue 2 Summer 2001

Table of Contents Letter from the President Letter from the President 1 News & Notes 3 Bridging the Quantitative/Non--Quantitative Divide Continuing Debate 5 From Philosophers' Stone to Unified Theory Michael Wallerstein Rein Taagepera Northwestern University Finding the Good Theory: [email protected] Explaining Comparative Politics Herbert H. Werlin The ability to read and converse in a language other than Eng- Symposium lish is a fundamental component of most comparativists’ expertise. Academic Collaboration in Comparative Politics Linguistic diversity does not interfere with our ability to communi- Introduction cate with each other, however, since we all share a common lan- Daniel Treisman 9 guage that we use in our papers and seminars. A scholar may have Some Reflections on My spent years mastering Russian but, when he or she gives a talk, we Collaboration with expect the talk to be in English. So why are comparativists who Stephan Haggard have spent years mastering formal theory or statistical methods ex- Robert R. Kaufman 10 empt from the general requirement to present their work in the The Growth of Collaborative Research: common language? The APSR will not accept articles written in Comparative Politics and Russian. Why, many ask, does the APSR publish articles written in International Relations the language of mathematics, a language that is equally foreign to Edward D. Mansfield 11 those without specialized training? On Collaborative The answer that mathematically trained scholars give is that too Research and Writing in much precision is lost in translating mathematics into English. Economics While there are phrases in any foreign language that are impossible William J. Baumol 14 to translate precisely into English, mathematics is different. The Collaboration and Credit in the Biological Sciences content of social science, as opposed to the content of literature, Jessica Treisman 16 should not depend on the (non-mathematical) language that is used Adversarial to present the results. But, it is argued; essential content is lost in Collaboration translating social science from mathematics into English. Daniel Kahneman 18 Not surprisingly, the view of the technically trained that every- APSA Annual Meeting 20 one should learn enough formal theory and statistics to follow Book Review 21 Copyright 2001 American Political Science Association. Published with the financial assistance of the College of Letters and Science, UCLA. apsa-cp Summer 2001 1 quantitative work in the field is not well received by those without technical training. The problem is that life is finite and the number of things that could be studied is infinite. It is easy to say that everyone in political science should learn formal theory and statistics. It is difficult to say how students should find the time to master social choice, game theory, statistics, the general literature in comparative Section Officers politics, the specific literature pertaining to one’s area of specialization and a foreign language, all in two or three years of course President work. There are always tradeoffs. More time spent learning formal Michael Wallerstein theory and statistics means less time spent learning the history and Northwestern University culture of other societies. [email protected] It would help reduce the conflicts in comparative politics if par- tisans on both sides of the quantitative/non-quantitative divide Vice-President, President-Elect could accept the validity of the arguments of their opponents. Con- Evelyne Huber sider, first, the case of empirical work. All empirical work consists University of North Carolina of two steps. The first step is to gather the evidence or data. The [email protected] second step is to apply the logic of scientific inference to identify Secretary-Treasurer the conclusions that can be drawn. A caricature of the non- Atul Kohli quantitative social scientist is one who invests all of his resources in Princeton University an effort to understand one particular case in all of its complexity, [email protected] only to confront the logical impossibility of concluding anything about cause and effect from a single case. An opposing caricature 2001 APSA Program Coordinator of the quantitative social scientist is one who spends all her time Barbara Geddes performing intricate statistical tests using data that contain little in- University of California, Los Angeles formation about anything we care about. [email protected] Caricatures aside, both sides have a point. On the one hand, quantitative researchers are correct to argue that all of the logical At-Large Committee Members Nancy Bermeo problems of inference that spur the development of statistical the- Princeton University ory are equally present in non-quantitative work. In addition to the [email protected] obvious problem of small sample size, the problems of identifica- tion and selection bias are, if anything, worse in non-quantitative Kathryn Firmin-Sellers work than in quantitative work since non-quantitative scholars are Indiana University often not trained to recognize such errors in reasoning. In addition, [email protected] learning to think in probabilistic terms is a great benefit in studying social phenomena, given the severe limits of what can be known Robert Jackman University of California, Davis with certainty. On the other hand, non-quantitative researchers are [email protected] correct to say that the payoff from collecting better information often exceeds what can be gained from finding new ways to torture Susan Pharr existing data. Moreover, by paying close attention to what people Harvard University say in an open-ended setting, non-quantitative researchers may ob- [email protected] tain insights into the causal mechanisms at work that exceed what can be learned from the study of quantitative measures of people’s behavior (or people’s responses to surveys). The conflict between quantitative and non-quantitative empirical scholars is muted in comparison to the conflict between formal (i.e. mathematical) and verbal theory. Statistical work is widely accepted in comparative politics in the way that formal theory is not (yet). One reason for the greater hostility towards formal theory in comparative politics is that it is newer and, therefore, less familiar. Another reason, I think, is that the introduction of formal theory (Continued on page 23)

2 apsa-cp Summer 2001

News & Notes

Announcement of Nominations for New Officers for 2001--02 The nominations committee of the Comparative Politics Sec- Newsletter Staff tion has nominated Peter Hall, Harvard University, for the position of Vice-President and President-Elect of the Comparative Politics Section. Peter Hall would replace Evelyne Huber, who is leaving Editor the position of vice-president to replace Michael Wallerstein as Daniel Treisman President of the section. In addition, the nominations committee University of California, Los Angeles [email protected] has nominated Melanie Manion, University of Wisconsin at Madi- son, and Ian Lustick, University of Pennsylvania, to serve as at-large Assistant Editor members of the Executive Committee. Melanie Manion and Ian Elizabeth Stein Lustick would replace Kathryn Firmin-Sellers and Susan Pharr, University of California, Los Angeles whose terms are coming to an end. All positions are for two years. [email protected] The nominations committee consisted of Francis Hagopian (chair), Jean Oi, Jonas Pontusson, Kaare Strom, and Crawford Young. Regional Editors -at-Large Any five members of the Comparative Politics Section may nominate alternatives, either at the business meeting of the Com- Middle East parative Politics Section at the Annual Meetings of APSA or by pe- Leonard Binder tition sent to the president prior to the annual meetings. If addi- University of California, Los Angeles

tional nominations are received for any of the open positions, the Latin America position will be filled by secret ballot at the business meeting. Barbara Geddes University of California, Los Angeles Call for Bids to Edit Newsletter Formal Analysis and Methodology The newsletter needs a new editor. All who are interested in John Londregan becoming the next editor of the newsletter of the Comparative Poli- University of California, Los Angeles tics Section of the APSA are encouraged to submit a bid. At the last meeting of the executive committee, the following guidelines were Africa adopted. Daniel Posner The editor of the newsletter will henceforth be a four-year term. University of California, Los Angeles The next four-year term will begin in the fall of 2002. The deadline for submitting a bid is Dec. 31, 2001. Bids should be sent to, Eve- Europe lyne Huber, (University of North Carolina, Department of Political Ron Rogowski University of California, Los Angeles Science, Chapel Hill, NC 27599-3265, email: [email protected]). A three-person committee, to be appointed by the president of the Section, will select the winning bid. The selection committee's decision will be announced by April 1, 2002. While it is desirable for the Newsletter to locate in different universities, the incumbent editor or editorial team may submit a bid to continue to edit the newsletter for a second term. The selection committee will use the following criteria to evaluate the bids: a. Bidding institutions should have a comparative politics faculty sufficiently large to support an editor, an associate editor, and have a pool of possible replacements. Responsibilities of the editorial team include identifying and developing themes, contacting potential contributors, selecting and editing submissions, and overall apsa-cp Summer 2001 3 oversight of the production Weingast for their paper cluded in such a directory, you and mailing process. The edi- “Federalism and Democratiza- should send your name, full tor and associate editor must tion in Mexico.” contact information including be able to commit an esti- For the Luebbert Award for email address, plus a fairly de- mated working time of 2-3 Best Article in Comparative tailed account of your area of weeks per issue, spread out Politics Published in 1999/2000, expertise to Sue Davis. You can over a longer period of time. the committee selected one win- contact Sue Davis by telephone b. Bidding institutions should ner and three runners-up. The at (202) 483-2512 or by email at have a pool from which to winner is Nicholas Sambinas for [email protected]. v his article “Partition as a Solu- choose an assistant editor. Estimated time spent by the tion to Ethnic War: An Empiri- assistant editor is four weeks cal Critique of the Theoretical per issue. The assistant edi- Literature” inWorld Politics 52 tor is expected to handle lay- (4). The three runners-up are outs, convert email submis- Torsten Persson and Guido sions, arrange for printing and Tabellini, “The Size and Scope production, and manage a of Government: Comparative web site. Compensation for Politics with Rational Politi- this position comes from the cians” in European Economic bidding institution. Review 43; David Rueda and Jonas Pontusson, “Wage Ine- c. The bidding institution quality and Varieties of Capital- APSA in should provide office space, ism” in World Politics, 52 (3); computer equipment, copy- and Michael Wallerstein, San Francisco ing and phone support. Re- “Wage-Setting Institutions and leased time for faculty will Pay Inequality in Advanced APSA 97th Annual Meeting also be taken into account Industrial Societies” in the but is not a requirement. American Journal of Political August 30 - September 7 Proposals should include a Science, 43 (3). prospective budget and a statement of administrative Comparative Politics at Hosted by support. APSA Convention San Francisco Hilton & Towers d. Themes, directions, special The Comparative Politics Parc 55 topics and other ideas of the Section will meet at 5:30 p.m. Nikko on Friday, August 31, 2001 at bidding editors will be taken into account. the APSA convention to be held in San Francisco, Califor- e. Section dues will pay for pro- nia. The particular room for Meeting attendees: duction and mailing ex- the gathering will be an- Pre-register by August 10th, penses. The bidding univer- nounced at the event. 2001 to reserve great low rates. sity should be able to cover other expenses listed in (c). Call for Experts To qualify for low rates, Comparative Politics Section Would you like to be a pun- you must be a current member of APSA. Awards dit? The APSA often receives To join APSA or renew your membership, press queries for experts in visit: www.apsanet.org The winner of the 2001 particular political fields. In Sage Award for Best Paper in response, the APSA is consid- Comparative Politics presented ering the establishment of a at the 2000 APSA Meetings directory of names covering all goes to Alberto Diaz-Cayeros, subfields of the discipline. If Beatriz Magaloni and Barry you are interested in being in-

4 apsa-cp Summer 2001 Continuing Debate

From Philosophers' Stone to Unified Theory

Rein Taagepera University of California, Irvine and Tartu University, Estonia [email protected]

Hopes of a unified “theory of everything” emerge mainly at dawn and high noon of a discipline. In its late morning, scientists are too busy packaging cumulative and interlocking knowledge into more limited theories of something. It started with the alchemists' notion of a philosophers' stone. After the advent of not only Newton and Maxwell but also Einstein, Bohr and Schroedinger with their theories of something, physicists began to look for a unified theory, semi-jokingly dubbed the “theory of everything” (TOE). In contrast, biologists have little time left for a “theory of life,” busy as they are with exciting and cumulative studies of something, such as the human genome. Political scientists are tempted to latch on to the label of “TOE” when their something does not look cumulative, given that the competing term, “philosophers' stone,” has lost its glamour. My high school philosophy of science course in Marrakech taught me that “theory” has an almost op- posite meaning in everyday speech and in philosophy of science (see Figure). What I later encountered in physics agreed with philosophy of science. But what does “theory” mean in political science? In Michael Wallerstein's lead article in apsa-cp (“Does Comparative Politics Need a TOE?” apsa-cp 12:1:1, winter 2001) “theory” seems to be synonymous with “paradigm” or “conceptual framework.” Com- parisons with relativity and quantum mechanics overlook the respective stages of development. For under- standable reasons (complexity and fluidity of the subject of study, plus limited ability to experiment) the achievements of political science, “rational choice paradigm” included, are more reminiscent of physics before Newton than after him. At this stage, dreams of a unified theory sound like “alpolitics” (cf. ‘alchemy”). To be more than a philosophers' stone, a unified theory presumes a basis in firmly established partial theories—and political science is as yet far from that stage. A hallmark of science at the level of several partial theories is cumulativeness. Successive paradigms build on each other rather than replacing the previous one. Relativistic mechanics adds specifications to the Newtonian rather than refuting it. Wave and particle approaches to light combine in a synthesis that defies everyday common sense. The last major notion fully dumped in physics was ether, two centuries ago. In contrast, contemporary political science offers a succession of fashions. Cumulativeness is obtained chiefly at the data collection level. (Does it correspond to the Tycho Brahe stage in astronomy?) On the overall paradigm level the main difference with alchemy is that the identity of the hoped-for philosophers' stone changes every quarter-century. Analogies with more developed sciences may offer a useful road map when the location of political science on this map is fixed realistically. Borrowing the term “unified theory” from the physicists' current quest and pretending that political science, too, already has several proven and stable theories in need of unification is mania grandiosa. Let us first establish reasonably firm laws (in the scientific sense), and with a sufficiently interlocking set of such laws, some theories of something eventually will emerge. As for conceptual frameworks, they are inevitable. Any empirical inquiry in any field is guided (or mis- guided) by such a framework, explicit or implicit. Empirical findings in turn affect the conceptual framework, but the latter may have considerable inertia. The Ptolemaic framework yielded to the Keplerian with considerable reluctance. This also means that major breakthroughs may come from researchers at the out- skirts of the accepted paradigms. From this viewpoint the frequent shifts in dominant frameworks that apsa-cp Summer 2001 5 Wallerstein observes in political science may speed up the process. If one is out of fashion, one may only have to wait for 25 or twice 25 years to have one's findings seriously considered. Actually, no framework has dominated completely and hopefully will not do so until fashionable promise is complemented with much firmer empirical results than presently is the case in political “science.” At the present stage it would be stifling to restrict the conceptual frameworks to a single one at a time. And don't call them theories. v

The meaning of “theory” in philosophy of science:

Wild goose chase No Reformulation of hypothesis Hunch Wording as Testing (observation testable Empirical -based) hypothesis Yes regularity --> Why? Semi-model Range of validity?

Philosophical clarification No Logical More Reformulation of concepts quantitative detailed of model model testing Scientific Yes law Many interlocking laws THEORY

The meaning of “theory” in popular parlance: “It's just a theory” = it's just a hunch (even less firm than a well-worded hypothesis).

Science: Theory is Popular: “Theory” is the ground Political science: the ultimate roof under a non-existing house ?

TheoryTheory

6 apsa-cp Summer 2001 Findinginding thethe GoodGood Theory:Theory: ment? (3) why is it that more ize or delegate power in vari- Explaining Comparative developed countries are simul- ous ways without losing con- Politics taneously more centralized trol and (b) to expand their and more decentralized than influence, reliably and pre- Herbert H. Werlin less developed countries? (4) dictably affecting the behav- Independent Consultant why is corruption devastating for ior of wider circles of citi- [email protected] poor countries, but not rich zens, participants and subor- countries? (5) why is economic dinates. In his presidential letter, globalization good for the devel- 3. Political elasticity depends “Does Comparative Politics opment of some countries, but partly on the selection of ap- Need a TOE (Theory of Every- not for others? and (6) to what propriate political hardware thing)?” in the Winter 2001 extent and how can political cul- but mostly on political soft- APSA-CP Newsletter, Michael ture be changed? ware (which has to do with Wallerstein makes a number of In my 1998 book, The Mys- the “subjective” quality of points that need to be exam- teries of Development: Studies relationships between leaders ined: (1) we are all theorists inas- Using Political Elasticity Theory and followers). much as we have explanations (University Press of America), I 4. Political software can be of what is happening; (2) we put forward political elasticity made more effective in vari- have no need for a unified the- theory as my way of answering ous commonsensical ways, ory; and (3) rational choice is these questions and, as such, requiring establishing accept- the “only theory in comparative linking comparative politics to able goals, hiring qualified politics today that is sufficiently comparative administration and personnel, encouraging train- powerful and general to be a se- development studies. Since this ing, delegating responsibility, rious contender for the unified book was summarized in the stimulating motivation and theory....” These assertions September 2000 issue of PS: competition, paying attention seem to me to require consid- Political Science and Politics, to morale, expanding two- eration of two questions: (1) including some comparative way flows of communication, what is politics? and (2) what is case studies (The Netherlands promoting legitimacy, main- theory? and Egypt - decentralization; taining supervision, cultivat- The classical meaning of Singapore and Jamaica - democ- ing contractors, protecting politics has to do with the rela- racy; Japan and Sierra Leone - independent spheres of au- tionship of leadership to fol- corruption), I will here only in- thority and developing con- lowership for the purpose of vite readers to examine the rele- flict-resolution procedures. governance. Insofar as rational vance of the following proposi- choice theory (as I understand tions for their own theorizing: 5. The enhancement of political it) refers only to the “selection 1. The more governments or software requires a balancing of policies,” rather than the those in authority can inte- of partisanship and states- “implementation of policies,” grate and alternate soft forms manship. These two mean- it cannot help us with the most of political power (linking ings of politics (i.e., the stru g- interesting questions in com- incentives to persuasion) with gle for competitive advantage parative politics, including: (1) hard forms of power and the struggle for consen- why is it that highly authoritar- (including disincentives and sus) can be considered as ian (even totalitarian govern- coercion), the more effective subdivisions (primary and ments, such as in North Korea they will be. secondary politics) of my ear- or Cuba) are usually incapable lier mentioned overarching 2. As leaders integrate and alter- of providing a high standard of definition of politics: the rela- nate soft and hard forms of living? (2) why is it that liberal tionship of leadership to fol- power, their political power democracy (as against classical lowership for the purpose of takes on “rubber band” and democracy) may undermine, governance. “balloon” characteristics, al- rather than promote, develop- lowing them (a) to decentral- apsa-cp Summer 2001 7 The theory presented here is elasticity, I try to look for multi- or explicate such political words a conceptual, rather than a for- dimensional evidence. as decentralization, corruption mal theory, with nothing more Yet, the pitfalls of formal and democracy. The justifica- than an explanatory capacity. theory seem to me far more tion for PE theory is a simple There is no effort to quantify dangerous. To ask for a social one — if it helps you, the reader, political words, separate facts science theory that permits po- in your own theorizing, that is from values or independent litical terminology to be quanti- good enough! I would appreci- from dependent variables, de- fied and conditions to be rigor- ate your communication in this velop testable hypotheses, or ously or definitely predicted is regard. v make predictions that can be as futile as asking for a perpetual verified. As such, my theory motion machine. To pretend cannot altogether escape the tau- otherwise is to be either fraudu- tological trap, particularly the lent or foolish (I am uncertain circular argument. While I may which), as do too many contem- appear to be overanxious to porary social scientists. Instead, equate success with political what I have tried to do is qualify

Announcing QUALMETH - a New Discussion List The Consortium on Qualitative Research Methods is pleased to announce the creation of QUALMETH, an electronic discussion list for political scientists and other scholars to analyze, debate, develop, critique and apply qualitative methods, broadly defined. To subscribe to QUALMETH send an email to [email protected] with the subject line left blank, and with the message text: subscribe QUALMETH yourfirstname yourlastname. For more information see: http://www.asu.edu/clas/polisci/cqrm/qualmeth.html QUALMETH will post messages on a broad range of topics, including queries on particular methods, discussions of texts advocating or critiquing qualitative methods, analysis of work applying qualitative methods, and up-to-date information on opportunities for, or problems with, archival access. QUALMETH will also publish book reviews, paper abstracts, relevant job descriptions, and notices of the posting of syllabi and working papers on the CQRM electronic database. In addition to these regular activities, QUALMETH will host scheduled roundtables in the CQRM discussion series. The series posts recent or draft articles and book chapters in which scholars analyze, debate, develop, critique and apply a variety of qualitative methods, or discuss issues relevant to their application. For more information, see: http://www.asu.edu/clas/polisci/cqrm/discussion.series.html Any questions on QUALMETH or CQRM's other activities can be emailed to [email protected], or mailed to Dr. Colin Elman, Executive Director, Consortium on Qualitative Research Methods, Box 872001, Arizona State University, Tempe, AZ 85287-2001.

8 apsa-cp Summer 2001 Symposium

Academic Collaboration in Comparative Politics

Editor Daniel Treisman University of California, Los Angeles [email protected]

Introduction Political science used to be almost entirely the preserve of soloists. In 1980, only one fifth of the articles that came out in the American Political Science Review had more than one author. Last year, there were as many co-authored as single-authored APSR pieces, including a number by comparativists. In World Politics, the share of co-authored articles rose from about 8 percent in 1980 to almost 16 percent in 2000. Comparing just two years is a rough-and-ready way to gauge change that would make statisticians cringe, but the figures do seem to fit a general perception. (They also match the results of some more systematic counting reported by Ed Mansfield in his contribution to this symposium.) The apparent change afoot raises an important question: if collaboration is catching on, how will this reshape—how is it reshaping—our field? Some practical issues are becoming familiar. Who, these days, has not sat in on a promotion case for a promising scholar whose work is copious but all or almost all of it co-authored? How should one parcel out credit and guess where the creativity or hard labor lay? What is the secret of great collaboration? What should one look for in a collaborator—besides, of course, brilliance, generosity and a Puri- tan work ethic? What are the advantages and difficulties of collaboration between junior and senior scholars? Across disciplines? Across nationalities? Across the breakfast table? We asked a variety of scholars—not just in comparative politics but from a range of fields—to share their insights. Robert Kaufmann, a comparativist with a long history of collaboration, offers a few thoughts on the cooperative efforts that went into crafting his well-known work with Steph Haggard. Ed Mansfield surveys the growth of collaboration in both comparative politics and IR in the 1990s, and mines the major journals for hints about why people are doing more joint research. How does collaboration work in other scientific disciplines? William Baumol, an economist with a record of distinguished collaborations that goes back more than four decades, reflects on the practice in economics. In the natural sciences, the titles of journal articles are often dwarfed by the list of authors, which sometimes seems to include everyone associated with the lab in question down to the jani- tor. To explain the norms that have developed to attribute credit for such collaborations, we picked at random an up-and-coming biochemist from the NYU Medical School whose (co-authored) articles have appeared in Cell, Developmental Biology, and Nature. In psychology, a new way of using collaboration to advance the research process may be emerging. One senior scholar, Daniel Kahneman, when asked by a journal editor to write a response to an empirical attack upon his work, recently invited the attackers to join forces with him. Together, they designed experiments that could resolve their disagreement, and published a joint article laying out the results and their different interpretations. We publish here the protocol for “adversarial collaborations” that Kahneman developed.

apsa-cp Summer 2001 9 Some Reflections on My of seminars on the debt crisis gether to produce two other vol- Collaboration with of the early 1980s, organized umes: Fragile Coalitions, also Stephan Haggard* through the efforts of Tom edited by Nelson, and The Poli- Biersteker and Miles Kahler. tics of Economic Adjustment, Robert R. Kaufman In one of those seminars, we edited by Haggard and Kauf- Rutgers University each presented individually man. The first of these was ad- [email protected] authored papers, which were dressed mainly to a “policy audi- eventually published in a spe- ence.” The second, which also For about the first 15 years cial edition of IO and in The enlisted contributions from John of my professional life, my pub- Politics of International Debt, Waterbury and Peter Evans, at- lications consisted primarily of both edited by Kahler. I had tempted to build on earlier case single-authored works. Since the not known Steph prior to our study work by focusing more mid-1980s, on the other hand, a presentations, but remember explicitly on “cross-cutting” ana- very large portion of my work being incredibly impressed by lytical issues: the influence of has involved collaborative re- the quality of his paper. the IMF, the politics of inflation search and writing. During that In the course of these meet- and stabilization, anti-poverty time, I have been fortunate to ings, a number of participants efforts, privatization and the role team up with a variety of very began to engage in more inten- of the state. able scholars, including Barbara sive discussions about the politi- The discussions, research, Stallings, among others. The cal economy of stabilization and and learning that went on during most sustained collaboration, structural adjustment, and to co- this period provided the founda- however, has been with Stephan ordinate their ideas into a relation for my co-authored book Haggard. We published our first tively coherent research agenda. with Haggard, The Political co-authored article on the debt In addition to Steph and me, Economy of Democratic Transi- crisis in 1989, and since that the other “core” participants in- tions. The ideas elaborated in time we have worked together cluded Joan Nelson, Barbara that volume were initially on a book, two edited volumes, Stallings, Miles Kahler, and sketched out in the introduction and at least seven other articles Tom Callaghy. Eventually, un- and conclusion of the co-edited and book chapters, all related to der Joan Nelson's leadership, we book. Our objective was to ex- the political economy of adjust- acquired research funding from plore the way our earlier analy- ment and democratization. I am Ford and Rockefeller for a col- sis of the politics of market- proud of what Steph and I have laborative project. An important oriented reform intersected with accomplished together, but I first step consisted of case stud- the strategy-oriented approaches will leave it to others to judge ies and small-N comparisons of that then characterized the de- the contributions these works adjustment experiences in thir- mocratization literature. have made to scholarship. For teen poor and middle-income A number of factors helped the present purpose, it is a rea- countries; these were published to make this collaboration work. sonable gauge of success that we in Economic Crisis and Policy For example, we had comple- have been able to complete so Choice, edited by Nelson. All of mentary knowledge of different many projects and still remain the chapters were individually regions; Steph's comparative ad- good friends. authored, but the book as a vantage was East Asia, whereas I suppose the main point to whole was a joint effort that in- mine was Latin America. He emphasize in the history of this volved careful coordination of also knew more about IPE, collaboration was that it evolved initial research questions, fre- whereas I had done more on the "organically" and quite slowly out quent meetings, and extensive issue of regime change and de- of a series of intellectual encoun- discussion of the conclusions mocratization. Moreover, since ters which involved not only that could be derived from the Steph and I had both established Steph and me, but a number of individual reform stories. fairly visible records of individual other people as well. Steph and I Though the cast of charac- scholarship prior to our collabo- first met as participants in a series ters changed somewhat, this ration, there were none of the group continued to work to- issues that sometimes arise about

10 apsa-cp Summer 2001 who contributed what; the work is to be integrated and co- reasons for the growth of col- "Haggard and Kaufman" brand herent; a strict division of labor laborative studies in compara- name is based entirely on the between narrow specialists will tive politics and international alphabet (although I have at not do. So, opportunities for relations, the latter of which is times considered dropping the collaborative research do not my field of expertise. I then sur- letter "K" from my last name). obviate the need for individual vey some leading journals to get Perhaps the most important scholars to acquire knowledge a preliminary sense of how thing was that we approached of a relatively wide range of much jointly authored work is problems of comparative re- skills and subjects. being published in these fields search in similar ways. Although There are, of course, many and who is conducting such both of us were guided by the- ways in which such problems work. ory, we also wanted to confront can be overcome, and I am not Collaborative research has the specificities of individual sure that my own experience always played an important role cases, a commitment that usu- provides readily generalizable in comparative politics and in- ally pulled us in the direction of lessons. It is not easy to repro- ternational relations. In interna- small-N comparisons. These duce the circuit of seminars and tional relations, for example, instincts were reinforced by the informal encounters which al- some of the best-known studies, numerous encounters I have lowed so many people from dif- such as Robert O. Keohane and just described. In part because ferent academic institutions to Joseph S. Nye’s Power and In- we spoke a “common language,” come together and build on terdependence (Little, Brown, our way of working did not in- their common interests. Almost 1977) and Glenn H. Snyder and volve a clear division of labor. assuredly, there are many for- Paul Diesing’s Conflict Among After we discussed a project or a mulas for successful collabora- Nations (Princeton University chapter, one or the other of us tions besides the one I have de- Press, 1977), were co-authored. would take his hand to a first scribed. This one, however, And the Correlates of War Pro- draft. We would then pass the evolved incrementally, begin- ject, which is led by J. David draft back and forth between us, ning with efforts that involved Singer at the University of sometimes as many as ten times, relatively limited commitments Michigan, has given rise to a until we were both either satis- and then grew from there. steady stream of collaborative fied or exhausted. There were work for almost four decades. plenty of arguments and mutual *Thanks to Steph Haggard and But despite the prominence of criticisms. But there was also a Joan Nelson for comments and certain collaborative enterprises, genuine synergy, and I strongly suggestions. co-authored studies have tradi- believe that the results were far tionally comprised a relatively better than either of us could small portion of the research have produced alone. conducted in comparative poli- On the basis of my work The Growth of Collaborative tics and international relations. with Steph as well as with other Research: Comparative Politics Even a casual glance at the lit- colleagues, I have become con- and International Relations erature suggests that this ten- vinced that collaborative re- dency has started to change: in search should, and will, play an Edward D. Mansfield recent years, there has been a increasingly important role in University of Pennsylvania and marked rise in the amount of our discipline. It is simply im- Ohio State University work involving investigative possible to expect any single in- [email protected] teams. dividual to master the increas- It seems to me that this de- ingly specialized domains of Over the past few decades, velopment can be traced to at knowledge that now characterize there has been a noticeable n-i least three factors. First, work in political science. On the other crease in the amount of co- comparative politics and interna- hand, collaborators must also be authored research being con- tional relations is becoming more able to speak mutually compre- ducted by political scientists. In technical and demands high lev- hensible languages if their joint this article, I speculate on the els of specific types of expertise. apsa-cp Summer 2001 11 Some research involves the use of how pervasive such research ranging from a low of about 15 of formal models and/or the has become and who is con- percent in Comparative Politics quantitative analysis of relatively ducting it. To obtain some data to a high of about 35 percent in large and complicated data sets. that might help address these Comparative Political Studies. Other studies stand at the inter- questions, I surveyed the articles Thus, while collaborative work section of international relations in comparative politics and in- has become more common, it is and comparative politics, political ternational relations (excluding by no means the norm in com- science and economics, or politi- review essays) published be- parative politics or international cal science and sociology. As tween 1990 and 2001 in the fol- relations. well, comparative and interna- lowing journals: (1) American Interestingly, the growing tional research often demands Journal of Political Science, (2) use of quantitative methods in expertise in particular regions of American Political Science Re- these subfields and the rise of the world. Given the time and view, (3) British Journal of Po- collaborative research seem to diverse talents needed to conduct litical Science, (4) Comparative be closely related. Regardless of such work, there are often sub- Political Studies, (5) Compara- which journal is surveyed, arti- stantial gains to be realized from tive Politics, (6) International cles written by a team of authors collaboration among two or more Organization, (7) International are more likely to use quantita- individuals with different skills. Studies Quarterly, (8) Journal of tive methods than solely au- Second, collaborative work Politics, and (9) World Politics. thored articles. This tendency is has grown easier and less expen- These periodicals differ on especially pronounced in the sive to conduct, especially for many dimensions, including the subfield journals, although it is individuals who are not located amount of collaborative re- also apparent in the general in- at the same institution. Central search they publish, the fre- terest journals. in this regard has been improve- quency with which they publish Who are the collaborators ments in computer and commu- articles on comparative politics in comparative politics and in- nications technologies, which and international relations, and ternational relations publica- have allowed collaborators to whether they are geared to sub- tions? Depending on the journal exchange ideas, data, results and field specialists or political scien- in question, between 20 percent drafts of manuscripts quickly tists in general. Each, however, and 30 percent of this research and at very low cost. Also note- is a high-quality outlet for re- involves a faculty member and a worthy is the dip in the cost of search in comparative politics graduate student. But the major- travel over the past few decades and/or international relations ity of collaborative work in these and the willingness of some col- that does not discriminate subfields is undertaken by fac- leges and universities to bear against co-authored work. ulty alone. Moreover, faculty these costs when collaborators I began by determining the members conducting such work need to meet in person. percentage of comparative poli- tend to be located at research Third, having a record of tics and international relations universities and think tanks. Per- conference papers and pub- articles that were co-authored in haps not surprisingly, faculty at lished articles has become in- each of these nine journals. In teaching colleges rarely publish creasingly important for gradu- each of the general interest jour- in the journals being surveyed. ate students entering the job nals (that is, the American Jour- When they do, however, the market. Faculty sometimes have nal of Political Science, Ameri- articles are almost never co- tried to relieve these pressures can Political Science Review, authored. by collaborating with their stu- British Journal of Political Sci- Especially pronounced is dents. Clearly, faculty-student ence, and Journal of Politics), the tendency for collaborative collaboration is nothing new, roughly 50 percent of the arti- research on comparative politics but my sense is that it is more cles were co-authored. In con- and international relations to common now than in the past. trast, the corresponding figure in involve a full professor. On av- Related to why co- the subfield journals is about 25 erage, about 60 percent of the authored research has become percent, although there is con- collaborative articles that were more common are the issues siderable variation among them, surveyed included at least one

12 apsa-cp Summer 2001 author of this rank. In rather professors, and full professors, strong norm in most political striking contrast, only about respectively, although it is note- science departments that junior one-quarter of the solely au- worthy that there has been a faculty need to demonstrate the thored articles were written by marked jump in multiple- ability to publish alone in order full professors. On the other authored articles involving a full to be promoted. That such a hand, roughly 40 percent of the professor in certain journals large portion of single-authored single-authored articles were (namely, Comparative Politics work in the journals I surveyed written by assistant professors and International Organization). are single-authored pieces by and faculty of this rank tend to Clearly, these figures should assistant professors suggests that engage in less collaborative re- be interpreted with considerable junior faculty are well aware of search than their more senior caution. Particularly important is the potential pitfalls associated counterparts. Further, unlike that they pertain to only a frac- with becoming too involved in both full professors and assistant tion of the research on com- collaborative research at the be- professors, associate professors parative politics and interna- ginning of their careers. are about as likely to write sin- tional relations, most notably Freed from the need to im- gle-authored articles as they are because no account is taken of press promotion committees, to write co-authored articles. research published as or in however, full professors have Finally, I made a prelimi- books. much greater latitude to choose nary attempt to address whether Nonetheless, one of the whether or not to do collabora- collaboration in comparative most striking patterns emerging tive research. In addition, senior politics and international rela- from this analysis is that faculty faculty have greater discretion to tions has become more or less are more likely to engage in col- pursue projects that are interest- pervasive over the past decade laborative research as they being but out of the mainstream. by comparing publication pat- come more senior. This ten- Rather than make the terns from 1990-1995 to pat- dency is not difficult to under- (potentially substantial) invest- terns from 1996-2001. I found stand. After all, assistant profes- ment needed to acquire the ex- that there has been a modest sors start their careers with dis- pertise necessary to complete increase in the amount of col- sertation research that, except in such a project by herself, it may laborative research in these very unusual circumstances, is make more sense for a faculty fields during the recent past and conducted alone. Moreover, member to team up with some- that this development stems pri- they have strong incentives to one already having that skill. Fi- marily from the growth of jointly publish that research in high- nally, full professors are espe- authored work published in sub- quality outlets, like the ones sur- cially likely to mentor graduate field journals. In general interest veyed here. Equally, it can be students and we saw that a non- journals, the portion of articles quite hazardous for assistant trivial portion of collaborative on comparative politics and n-i professors to become overly in- articles during the past decade ternational relations written by volved in co-authored work. have been jointly written by fac- teams of scholars has changed One obvious problem in evalu- ulty members and students. very little since 1990. There is ating such work is determining During my career, I have also some evidence of an in- each author’s contribution to taken part in a fair amount of crease in the amount of jointly the final product. An assistant collaborative research, having authored studies involving a fac- professor with a publication re- worked with my students, my ulty member and a student, al- cord composed solely of co- dissertation advisor, my depart- though this varies quite a bit by authored research who is being mental colleagues, and faculty at journal. But the amount of col- considered for tenure might other institutions. These experi- laborative research published by find it difficult to convince a ences have been remarkably en- faculty at both universities and committee reviewing her work riching. I have learned a great teaching colleges has been fairly that she played enough of a role deal from my co-authors and stable. So too has the portion of in this research to warrant pro- have enjoyed the process of col- co-authored articles written by motion. Further, my sense is laborating with them. Moreover, assistant professors, associate that there continues to be a many of the resulting papers sim- apsa-cp Summer 2001 13 ply could not have been written from working together. They span several such sub-fields, as by any one of us individually. can stimulate one another’s ef- frequently occurs, urgently calls That, it seems to me, is the best forts; their discussions can lead for collaboration. reason to do collaborative re- to new ideas and approaches; Disputes among collaborators search: it allows scholars to pur- and if they find writing to be an sue a wider range of topics with onerous task, each author will Yet collaboration is not an greater expertise and insight than appreciate the opportunity to easy thing. Working methods would be possible working share the burden. can differ. One participant may be neurotic about meeting dead- alone. Why the growth in collaborative lines, another may be chroni- work? cally inclined to postponement But it is my hypothesis that until the deadline looms, and such collaboration between in- some may disregard deadlines On Collaborative Research vestigators whose talents are sub- and Writing in Economics* altogether. Peace among col- stitutable rather than comple- laborators is considerably facili- mentary is not the reason for the William J. Baumol tated if they select themselves as marked growth in frequency of a group with homogeneous atti- New York University and such arrangements. The incen- Princeton University tudes on such matters and on tive of burden sharing and mu- working procedures more gen- [email protected] tual stimulation has long been erally. I understand that disputes present. Rather, I believe, an among collaborating authors are Substituteitute andand complementarycomplementary explanation that is more persua- collaborators quite frequent and can some- sive is the marked increase in times grow bitter, perhaps on The vital ingredient of my specialization and the difficulty the oft-noted principle that there many fruitful collaborations over of the technical analytic require- is so little at stake. I remember the years has been my areas of ments that have characterized how such a dispute arose many ignorance. Recognition that many academic disciplines in years ago, when a close friend of much of what I was currently in- recent years. Today the general mine, a noted biologist, was wri t- vestigating was beyond my range economist is virtually extinct. ing an elementary textbook with of knowledge and analytic skills We divide ourselves not only a colleague who, I understand, many times has led to partner- into macro and micro practitio- was also eminent. The book ship with someone who was able ners, but within each of these contained a photograph of a sea to provide what would otherwise there is a profusion of speciali- turtle making its way across a have been missing. To the extent zations: growth theorists, finance beach as part of the egg-laying that I, too, could provide a con- specialists, labor economists, process. The one author had tribution that would have been pure theorists, applied econo- written, “The turtle comes difficult for the other person, the metricians, and so forth. ashore to lay its eggs.” This result was superadditive — the Whereas forty years ago an eco- aroused the ire of his partner whole was greater than the sum nomics department normally who insisted that it be revised to of the parts. More than that–this offered one or two seminars that read, “The turtle comes ashore pattern helps to explain why my were attended by most of the and lays its eggs.” The teleologi- collaborators and I have, literally, faculty and graduate students, cal battle was protracted and never quarreled. We each knew nowadays these seminars have emotional. I remember one dis- that in relation to what the other proliferated substantially, and cussion that literally lasted the was writing we had a good deal to each is usually attended by a entire night, even enlisting the be modest about. handful of participants, the spe- participation of a famous phi- Of course, the division of cialists in the sub-field. More- losopher. My point is not that the territory in this way is not the over, a superficial acquaintance the distinction was unimportant. only possible way of proceeding. with such a field and its vast lit- To specialists it may well be vital. Clearly, two analysts with similar erature simply will not do. But I suspect that among the abilities may nevertheless profit Hence, research and writing that many freshmen and sophomores

14 apsa-cp Summer 2001 who read the book, there were But there are costs of which no person’s performance, intuitive few if any who would have no- portion can legitimately be at- evaluation based on discussions ticed, let alone being brought up tributed to any one of the regu- with the person, and other bits short, by what implicitly was be- lated firm’s products. An obvi- and pieces of evidence that can ing claimed. ous example is railroad rate be presumed to supplement the Disentangling contributions: regulation, where a substantial limited inferences that can be evidence for promotion and proportion of the cost is made derived from the evidence of tenurtenure up of track, bridges, tunnels and collaborative publications alone. so forth, all of which are used in Conclusion The growth in collaborative common in transporting steel, research and writing has evi- coal, lumber, wheat and the Collaboration is not all dently created severe impedi- many other products the rail- benefit. The process of working ments for evaluation of the work road carries. Predictably, hoping together can, for example, tend of the participants by their de- to get a low rate for their clients, to discourage heterodox ap- partments. It is clearly not easy the attorneys who represent the proaches and conclusions, or to evaluate the contribution and shippers of feathers argue that unconventional modes of ex- abilities of an assistant professor weight is the proper basis for the pression. Still, it does tend to who has taken part in the prepa- apportionment of cost, while add to the range of knowledge ration of 20 articles, all of them those who represent lead ship- taken into account in the work, having at least one coauthor. pers argue for the use of bulk as and gives each participant the The attempt to disentangle the appropriate criterion. The opportunity to benefit from the the evidence is surely under- result is extended litigation, stimulation of the ideas, sugges- taken in every university, but its heavy expenditure on attorneys tions and criticisms of the oth- success is plainly rather limited. and expert witnesses, and irra- ers. There is plainly room for Indeed, I am skeptical about its tional decisions, but no illumina- both collaborative work and prospects. Experience in other tion. work that is individually pro- fields leads me to take a dim These are not the only per- duced. By and large I feel that view of attempts to divide up the ils that beset the search for sys- the trend toward more frequent indivisible. Unless an article is tematic methods to assign the co-authorship is a beneficial de- divided into sections — with unassignable. One may, for ex- velopment; but, that is a conclu- each piece clearly emanating ample, be tempted to use a sion for which no firm and ob- from a different author — how “common element” criterion jective foundation can be can one hope to disentangle not associated with J.S. Mill — the claimed. only the words contributed by view that if one author is a col- each but also, more important laborator common to a number * I am grateful to the C.V. Starr and much more elusive, the of particularly successful articles, Center for Applied Economics relative values of those words? she, as the one common con- at NYU for support of this Attempts to find a systematic tributor, can be presumed to work. method for approaching the bear the primary responsibility matter can easily contribute fur- for their success. But that, as ther confusion by appearing to Jacob Viner was fond of point- provide some light, where they ing out, is like observing that merely camouflage the difficulty since bourbon and soda, scotch and thereby only add to dark- and soda, and brandy and soda ness. In governmental price are equally inebriating, it is the regulation the attribution issue is soda that must be responsible. a familiar problem and its dan- My conclusion is that the gers have been thoroughly docu- problem has no systematic an-

mented. Price regulators cus- swer. We are forced to stumble tomarily, and with good reason, toward an evaluation on the ba- seek to take costs into account. sis of casual observation of the apsa-cp Summer 2001 15 can cause problems in deciding have a Ph.D., are continuing Collaboration and Credit in both credit and responsibility. their training in another area of thethe BiologicalBiological SciencesSciences Other disciplines use alternative research, and also must establish conventions; for example, many a foundation of accomplishments Jessica Treisman high-energy physics papers list on which to build their own labs. New York University their numerous authors alpha- They are generally more inde- School of Medicine betically.1 pendent than graduate students, [email protected] in part because of their greater Division of labor within a lab experience, and in part because Experimental research in A biology lab is normally they need to find a direction to the natural sciences is almost headed by a faculty member or follow in the future that will not exclusively a collaborative effort. principal investigator, and may directly compete with the lab in The most common type of col- also contain one or more re- which they trained. laboration is between one or search technicians, graduate stu- Although this collaborative more junior members of a labo- dents and postdoctoral fellows. structure allows faster progress ratory and the head of that lab; Each of these individuals has dif- than individuals could achieve, at however, collaborations among ferent goals and responsibilities. times it can be frustrating for all two or more labs are becoming In general, the responsibilities of concerned. Junior members of increasingly frequent. The first the lab head are to supervise the the lab may feel they are not type of collaboration is a result other lab members, provide n-i given enough freedom to pursue of the extended training period tellectual input to their research, their own ideas, perhaps because required to gain experience in assist in writing papers, write of commitments made to fund- complex experimental tech- grants to raise money for salaries ing agencies. The lab head, who niques, as well as the significant and supplies, publicize the re- has been trained as an experi- amount of time that must be de- search of the lab through semi- mentalist, may find it difficult to voted to fund-raising to support nars and conferences, and per- adjust to management and watch research that requires expensive haps teach undergraduates or less experienced people bungle equipment and materials. The graduate or medical students. his or her ideas. A few lab heads second is due to the realization Research technicians are often do manage to continue their own that many problems require a recent graduates seeking lab ex- experimental work, but in most multidisciplinary approach be- perience before applying to institutions promotion depends yond the expertise that can be graduate or medical school, al- on growth in lab size and grant assembled in one lab, and also though they can also be older support, making this difficult. to the increasing convergence of individuals who have chosen to Recognizing contributions in interests between scientists make their career in developing print working on apparently different a particular technical expertise. questions that turn out to be re- They may have independent The division of labor between lated. Although such collabora- projects, but also are usually ex- lab members results in multi- tions often result in publications pected to assist others in the lab author publications in which the that are deeper and more intel- with the less intellectually stimu- order of authors reflects the dif- lectually satisfying than what a lating aspects of their work, such ferences between their contribu- single group could have as ordering supplies or making tions. Conventions for ordering achieved, they complicate the solutions. Graduate students are authors in the credits seem to process of awarding credit to the expected to spend an average of have evolved in a decentralized individuals involved. In the bio- five years on an integrated re- way, but they are widely under- logical sciences, a system of search project that will intro- stood by scholars in the field. rules and conventions has been duce them to both the practical Typically, the lab head is the last developed to deal with this, and and intellectual aspects of sci- author, and this position is taken in general contributions are ence, as well as leading to sig- to mean that the person had a evaluated appropriately; how- nificant findings and publica- guiding intellectual role in the ever, the exceptions that arise tions. Postdocs, who already project rather than making a

16 apsa-cp Summer 2001 “hands-on” contribution to the equally to the paper. Although the second author and the lab research. The first author is the this is the most common way to head the second to last author. person, usually a student or appease an unhappy second au- If a third lab becomes involved, postdoc, who did the bulk of the thor, credit will not in fact be its members will take up author- experimental work. The differ- equally perceived by those who ial positions correspondingly ent meanings of the first and last simply see the paper listed on a closer to the middle of the list. positions mean that conflicts be- c.v. without accompanying aster- A more difficult situation can tween these two authors are isks. In general, conflicts can arise if two groups discover that rare. However, conflicts fre- best be avoided by deciding the they have been working on the quently arise between the first order of authors as early in the same topic. One option is to author and other contributing project as possible. This allows publish back-to-back papers re- “middle” authors. A position as those who know they will be peating much of the same data, second, third, fourth, etc. author middle authors to put in only but another possibility is to pool means that the person has made the amount of effort justified by the results in a single paper to a contribution to the work; how- such a position. This approach avoid duplication of effort. ever, this contribution may vary may run into problems if one However, if the contributions of greatly, from simply donating a person has to leave the lab be- both labs have been fairly equal, reagent to carrying out a large fore the project has been com- it can be difficult to decide both proportion of the experimental pleted. This person may have the first-author and last-author work. Because of this variability, done a large amount of work positions. There is no good the credit assigned to a second and have assumed that he or she mechanism yet to acknowledge or third author is significantly would be the first author of the equal contributions of the two less than that assigned to the resulting publication. However, lab heads. first author. Postdocs applying completing the project may take Awarding credit for faculty positions are judged significantly more work than an- predominantly on their first- ticipated, due to technical diffi- Variations in the relative con- author papers, and there is also culties or reviewers’ criticisms, tributions of authors arising for pressure on students to publish giving the person who completes the reasons listed above can lead as the first author. Thus indi- it an equally legitimate claim to to difficulties in evaluating candi- viduals in a lab normally prefer first authorship. The latter dates for hiring or promotion. to publish their more time- claimant then has the advantage In general, postdocs applying consuming projects as first- of proximity to the lab head. for a faculty position are judged predominantly on the first- author papers. Counteracting Multi--lablab collaborationscollaborations this is the need to have a com- author papers on their c.v. This plete story, often more than one Collaborations between labs probably means that their con- person can accomplish in a rea- raise a different set of issues. tributions to papers on which sonable period of time, in order Such collaborations can arise in there are several other authors to publish in the top journals. a number of different ways. The are over weighted, while their This may result in the lab head members of one lab may realize contributions to papers on prevailing on other lab members that a set of techniques beyond which they were the second au- to contribute their work to a pa- their expertise would be neces- thor are under weighted. How- per on which they will be the sary for a full analysis of the ever, collaboration as a second second or third author. problem on which they are author can still be beneficial to No completely satisfactory working, and turn to another lab such candidates, as it enables strategy to recognize the contri- for assistance. In such cases it is them to discuss the collaborative butions of such authors has often understood that the sec- work in their seminar. Faculty been found. Many journals al- ond lab will make an important members applying for promotion low an asterisk to appear after but minority contribution to the are judged predominantly on several of the authors’ names, finished work, which is normally their last-author papers, although followed by a note to say that recognized by making the stu- additional middle-author papers those individuals contributed dent or postdoc from that lab can be helpful as evidence of the apsa-cp Summer 2001 17 ability to collaborate. Thus an Rather than publish a series of empirical analysis, they might over-enthusiasm for collabora- critiques and rejoinders, the two agree on different ways to rean a- tion may be penalized if it re- set out to design, conduct and lyze the same data. Kahneman duces the time available for write up joint experiments that proposes the following protocol more independent research. promised to resolve their dis- for those inspired to try an adver- Paradoxically, single-author pa- agreement. The two brought in sarial collaboration. We are pers, written by rare faculty a third psychologist, Barbara grateful to Psychological Science members who continue their Mellers, to act as “arbiter,” re- and its editor, Sam Glucksberg, own experimental research, may solving disputes about proce- for permission to reprint it. be valued less than multiple- dures. To cut a long story short, — DT author papers, which demon- the collaborators did not end up strate that the last author has agreeing in all respects. But the trained additional scientists in differences narrowed on the ba- Suggestions for Adversarial his or her lab. sis of the new results. Collaboration Although the mechanisms As the authors write in the re- for evaluating an individual’s sulting article: “Collaborating Daniel Kahneman contributions to collaborative with an adversary is not easy… Princeton University papers are imperfect, fore- Despite our mishaps, we hope [email protected] knowledge of these mecha- the approach catches on. In an nisms allows some adjustment ideal world, scholars would feel 1. When tempted to write a cri- of effort. The rewards of col- obliged to accept an offer of ad- tique or to run an experimen- laboration are the satisfaction versarial collaboration. Editors tal refutation of a recent publi- of contributing to a study that would require adversaries to col- cation, consider the possibility is more complete and of laborate prior to, or instead of, of proposing joint research greater interest than the part of writing independent exchanges. under an agreed protocol. We it that one could have accom- Scientific meetings would allot call the scholars engaged in plished alone, as well as the time for scholars engaged in ad- such an effort participants. If intellectual stimulation and in- versarial collaboration to pre- theoretical differences are sights gained from discussion sent their joint findings. In short, deep and/or if there are large with one’s collaborators. adversarial collaboration would differences in experimental become the norm, not the ex- routines between the laborato- 1. Nature (1997) 387, 831. ception.” A second type of ad- ries, consider the possibility of versarial collaboration, pio- asking a trusted colleague to neered by Kahneman with dif- coordinate the effort, referee ferent collaborators,2 is the joint disagreements and collect the Adversarial Collaboration commentary on a critique: data. We call that person an Must collaborations occur rather than just responding in arbiter. between friends? Not, accord- print, the scholar whose conclu- 2. Agree on the details of an ing to one leading psycholo- sions have been challenged joins initial study, designed to sub- gist. In a recent article in the with his critic to co-author a ject the opposing claims to an journal Psychological Science, commentary sharply defining informative empirical test. Daniel Kahneman and several their areas of agreement and dif- The participants should seek colleagues introduced the conference. to identify results that would cept of “adversarial collabora- Experiments are rare in politi- change their mind, at least to tion.”1 Ralph Hertwig of the cal science. Nevertheless, there some extent, and should ex- Max Planck Institute for Hu- may be ways in which such a plicitly anticipate their inter- man Development in Berlin practice could be adapted. pretations of outcomes that had challenged a finding of Scholars might agree on what would be inconsistent with Kahneman about common bi- would constitute a “critical case,” their theoretical expectations. ases in probabilistic reasoning. and then look for examples. In These predictions should be

18 apsa-cp Summer 2001 recorded by the arbiter to results, and a statement of 8. The arbiter should have the prevent future disagree- agreed conclusions. If signifi- casting vote in selecting a ments about remembered cant disagreements remain, venue for publication, and interpretations. the participants should write editors should be informed 3. If there are disagreements individual discussions. The that requests for major revi- about unpublished data, a length of these discussions sions are likely to create im- replication that is agreed by should be determined in ad- possible problems for the both participants should be vance and monitored by the participants in the exercise.

included in the initial study. arbiter. An author who has more to say than the arbiter 1 Barbara Mellers, Ralph Hertwig 4. Accept in advance that the allows should indicate this and Daniel Kahneman, “Do Fre- initial study will be inconclu- fact in a footnote and pro- quency Representations Eliminate sive. Allow each side to provide readers with a way to Conjuction Effects? An Exercise in Adversarial Collaboration,” Psy- pose an additional experi- obtain the added material. ment to exploit the fount of chological Science, Vol. 12, July 2001, pp. 269-275. hindsight wisdom that com- 6. The data should be under the control of the arbiter, monly becomes available 2 who should be free to pub- Dan Ariely, Daniel Kahneman when disliked results are ob- and George Loewenstein, “Joint tained. Additional studies lish with only one of the original participants if the Commentary on ‘The Importance should be planned jointly, of Duration in Ratings of, and with the arbiter resolving dis- other refuses to cooperate. Choices Between, Sequences of agreements as they occur. Naturally, the circumstances Outcomes,’” Journal of Experi- of such an event should be mental Psychology, December 5. Participants should agree in part of the report. 2000, 129, pp. 524-529. v advance to produce an article with all participants as 7. All experimentation and writ- authors. The arbiter can ing should be done quickly, take responsibility for sev- within deadlines agreed in ad- eral parts of the paper: an vance. Delay is likely to breed introduction to the debate, discord. the report of experimental

Renew Today!

How to Subscribe: Subscriptions to the APSA-CP Newsletter are a benefit to members of the Organized Section in Comparative Politics of the American Political Science Association. To join the Section, check the appropriate box when joining the APSA or renewing your Association membership. Section dues are currently $7 annually, with a $2 surcharge for foreign addresses. The printing and mailing of the Newsletter are paid for out of members’ dues. To join the APSA, contact: American Political Science Association 1527 Hampshire Ave., NW Washington, DC 20036 USA Telephone: (202) 483-2512 Facsimile: (202) 483-2657 Email: [email protected]

apsa-cp Summer 2001 19 APSA Annual Meeting

11-24 Taxation and Public Sector Spending Comparative Politics in San Francisco and Capacity Thu 8:45 am The preliminary schedule for this year’s APSA 11-25 The Politics of Economic Reform Thu 10:45 am Convention in San Francisco includes the 11-26 Regionalism, Political Institutions and following panels: Voting Behavior Sun 8:45 am 11-27 Establishing Justice and the Rule of Law 11-1 Institutional Arrangements and Their Policy in New Democracies Fri 1:30 pm Implications in Emerging Democracies Fri 1:30 pm 11-28 How Institutions Shape Politics Sat 8:45 am 11-2 Evaluating Success and Failure in Post- 11-29 Central Bank Independence, Labor Insti- Communist Economic Reform Thu 1:30 pm tuitions, Inflation and Unemployment Fri 10:45 am 11-3 Political Foundations of Economic 11-30 Party Systems in New Democracies Sun 8:45 am Development Sun 10:45 am 11-31 The Effect of Federalism on Economic 11-4 Processes of Diffusion in Political and Institutions and Performance Sat 3:30 pm Economic Liberalization Thu 3:30 pm 11-32 Democracy and Growth Thu 1:30 pm 11-5 The Politics and Economics of Post- 11-33 The Effect of the Economy on Public Communist Reform: A Theoretical and Opinion and Voting in New Democracies Fri 1:30 pm Methodological Discussion Fri 8:45 am 11-34 The Politics of Decentralization Thu 3:30 pm 11-6 The Institutional Determinants of 11-35 The Politics of Welfare Spending and Governmental Performance Sat 8:45 am Income Inequality Fri 10:45 am 11-7 The Comparative National Elections 11-36 Contested Transitions to Democracy Thu 8:45 am Project Fri 10:45 am 11-37 The Politics of Environmentalism Sat 10:45 am 11-8 The International Diffusion of Policy Innovation Thu 8:45 am 11-38 Women, Politics and Family Welfare in Comparative Perspective Sat 1:30 pm 11-9 Rethinking Democracy and Dictatorship in Interwar Europe Sun 8:45 am 11-39 New Research on the European Union Fri 3:30 pm 11-10 What We Know and How We Know It in 11-40 Comparative Electoral Systems Fri 1:30 pm Comparative Politics Fri 8:45 am 11-41 Veto Players and Political Analysis Thu 10:45 am 11-11 Rules and Social Norms: Formal and 11-42 Comparative Analyses of Human Rights Informal Institutional Dynamics in the Violations and Protective Mechanisms Sun 10:45 am Politics of Late Developers Sun 10:45 am 11-43 Putting the Politics back into Privatization: 11-12 Government Capacities and Policy- Cross-National Studies of Transitions in Making in Latin America Thu 10:45 am Socialist, Post-Socialist and Developing 11-13 Southeast Asian Industrialism on Trial Fri 8:45 am Countries Thu 3:30 pm 11-14 Market Reform, Institution Building and 11-44 State-Market Relations in Post-Communist Democratization Sun 10:45 am Settings Sun 10:45 am 11-15 Comparative Politics, Rational Choice and 11-45 Economic Change and Political Response: the Hermeneutics of Strategic Action Fri 3:30 pm The Politics of Market Liberalization Fri 1:30 pm 11-16 Politics in Japan: Regime Shift or 11-46 Globalization, Domestic Politics and Political Incremental Change? Sat 10:45 am Causes: Tracing the Causal Pathways Fri 8:45 am 11-17 When Politicians Switch Parties or Parties 11-47 Globalization, Democratization and the Accumulation of Human Capital Fri 3:30 pm Switch Positions: A Comparative Thu 3:30 pm 11-48 Informal Institutions and the State: A 11-18 New Political Economy of Latin Cross-Regional Perspective Sat 3:30 pm America Sat 1:30 pm 11-49 Taking Democracy Seriously: The Role of 11-19 The Role of Mass Protest during Transitions Political Parties in Southeast Asia Fri 1:30 pm Sat 3:30 pm 11-20 Strategic Behavior during the Mexican 11-50 Regime Transition and the Radical Right Transition Sat 8:45 am in Central and Eastern Europe Thu 1:30 pm 11-21 Popular Attitudes toward Economic 11-51 Scaling in Comparative Research Fri 8:45 am Reform Sat 10:45 am 11-52 Race, Ethnicity and the Power of Cross- 11-22 Models of Bureaucratic Accountability National Comparison Sun 10:45 am and Reform Thu 1:30 pm Division chair: Barbara Geddes, University of 11-23 New Approaches to Understanding Ethnicity Sat 1:30 pm California, Los Angeles

20 apsa-cp Summer 2001 Book Review

From Voting to Violence: who would accept democratic elites are more likely to use na- Democratization and peace theory and helps the tionalist appeals than in more Nationalist Conflict reader understand why in theory developed settings because these By Jack Snyder democratization may cause na- citizens are more susceptible to New York: W.W. Norton and tionalist conflict, in this reader’s such nationalist appeals than Company. 2000. 382p. $29.95 view Snyder’s test lacks the em- those in more developed politi- pirical rigor that would allow the cal settings. Further, the scare Robert A. Dowd reader to join the author in dis- tactics based on ethnic or na- University of California, Los Angeles missing the possibility of having tional identity used by political [email protected] both democratization and peace elites are more likely to result in in relatively poor countries. violence than where there is a Jack Snyder claims that all Snyder does a fine job of higher level of economic devel- good things do not go together clearly presenting his argument, opment and the citizenry is gen- and that while democratization which he calls the elite persua- erally better educated. may be a good thing in the long sion argument, and explaining Snyder does a very good job run for stability and economic its underlying logic (pp. 31-32). of comparing and contrasting growth, in the short run it pro- Snyder notes that nationalist several country cases in order to duces nationalist conflict and conflict is not necessarily violent show that there is a correlation therefore makes war within and and argues that whether nation- between the pace of democrati- between states more likely. alist conflict is rather subdued zation, largely defined as the ra- While both peace and democra- or results in violence depends pidity with which suffrage was tization are good and desirable largely on the level of economic expanded, and the type and in- goals, Snyder suggests that there development in a country, the tensity of nationalist conflict that is reason to expect that the con- extent to which the careers of manifested itself. For example, ditions that prevail in many political elites who rule under an where suffrage, political rights countries, especially poverty, authoritarian regime are threat- and civil liberties expanded make having a great deal of both ened by democratization, and gradually, as in Britain, Snyder peace and democratization very the strength of political and so- convincingly argues that nation- unlikely if not impossible. Sny- cial institutions (pp. 38-39). In alism was less belligerent, less der argues that too much de- short, Snyder argues, à la Sey- violent and less destructive than mocracy, meaning universal or mour Martin Lipset (1959) that in France and Germany, where widespread suffrage, too fast it is very unlikely that poor suffrage, political rights and civil leads to the most violent and countries that long have been liberties expanded more sud- destructive types of nationalist under authoritarian rule will de- denly (pp. 76-79). Although conflict. The author traces mocratize without violent na- Snyder’s portrayal of events in causal processes within and be- tionalist conflict. As pressure to Rwanda and Burundi is gener- tween historical cases of democ- democratize mounts, suffrage is ally less convincing, he does ratization, such as Britain extended widely or universally, show that there is a correlation France, Germany, the Balkans, and elections are convoked, po- between the pressure for power Malaysia and Sri Lanka, and litical elites who have benefited sharing in those countries and contemporary cases of democra- from authoritarian rule become the intensity of ethnic nationalist tization, such as India, Russia, threatened and are more likely conflict in Rwanda and Burundi. Yugoslavia, Rwanda, and Bu- to use nationalist appeals and However, as the old adage rundi to test the plausibility of scare tactics to drum up elec- goes, correlation is not causa- the argument. While Snyder toral support. In a poorly devel- tion, and Snyder does not pro- raises many important points oped country, where educa- vide the kind of evidence to that serve as a caution to those tional attainment is low, political confirm that the relationship apsa-cp Summer 2001 21 between the pace of democrati- other words, democratizing argument against rival explana- zation and nationalist conflict is countries do not have a monop- tions for the rise of nationalist in fact a causal one or that it is oly on nationalist conflict. conflict. Snyder uses his very thor- causal in the direction that he While this need not destroy ough case studies to show that the proposes. It was as if Snyder Snyder’s argument, Snyder elite persuasion argument is plau- noted that where there is nation- needs to explain more ade- sible, but the reader is left won- alist conflict there is democrati- quately the cases that do not dering: plausible compared to zation and therefore concluded seem to fit well with the elite what? Some serious treatment of that democratization must cause persuasion argument. rival arguments about the rise of nationalist conflict without ade- It also should be noted that nationalist conflict would have quately recognizing that there it is possible that, rather than helped. also has been nationalist conflict democratization causing nation- In spite of the shortcom- and aggressive international be- alist conflict as Snyder proposes, ings, Snyder’s work stimulates havior where there has been lit- democratization may in fact be thought and should serve as a tle in the way of democratiza- the result of protracted and de- caution to those who uncritically tion, such as in Western Europe structive nationalist conflict over accept the democratic peace before Britain and France de- control of the state. This at first theory. The book is a must read mocratized and, more recently may seem like an odd argument. for those interested in critically in the Sudan since its independ- However, it is important to note assessing the democratic peace ence and Iraq since the coming that the world’s oldest and most theory and those interested in to power of Saddam Hussein. In consolidated democracies were explaining the variation in na- plural societies and experienced tionalist conflict within and be- nationalist conflict within their tween states. It is reasonable to The Newsletter invites borders and wars with nations think that nationalist conflict is doctoral students to submit book reviews for this beyond their borders long be- likely to accompany democrati- section. If the book re- fore democratization. Minorities zation. When the vast majority viewed is recent, of suffi- almost always resisted majorities of the people who long have ciently general interest to or vice versa, depending on been disenfranchised suddenly comparativists, and the which was in power, and occu- become enfranchised, we can review thoughtful and of pied nations almost always expect there to be a great deal of publishable quality, then sought to throw off occupying tumult and instability. However, we will try to find room for it in the Newsletter. If you nations. It is possible that de- it is important to ask oneself are interested, please con- mocratization was set in motion whether the tumult and instabil- tact the Editor or Assistant because political elites of vari- ity attributed to expanded suf- Editor for further informa- ous nations and/or ethnic frage and democratization by tion and style guidelines. groups gradually recognized that Snyder would have been even the costs of the nationalist vio- greater without the expansion of lence were outweighing the suffrage and democratization. v benefits and that in order to Book Reviews promote peace and the wealth Robertt A.A. DowdDowd is an advanced Needed! that comes with it there would graduate student in the Depart- need to be some kind of power ment of Political Science at Note to authors and pub- sharing arrangement among UCLA and currently a Visiting lishers: The Newsletter will them. Such an arrangement, in Scholar at the Helen Kellogg In- not find reviewers for un- order to last, would need some stitute for International Studies, solicited manuscripts. But kind of popular approval. Sny- University of Notre Dame. if you wish to help fill our der would have done well at bookshelves and landfills, least to address this possibility. keep them coming! Snyder’s argument would be strengthened if he had more thoroughly tested the elite persuasion

22 apsa-cp Summer 2001 (Continued from page 2) to focus first on the country they non-quantitative work learn has changed the questions that know best, which is usually the enough formal theory and sta- are asked. Comparativists who United States. But in part, the tistics to understand the general might welcome new ways to limited use of formal theory in principles that underlie the think about the questions they comparative politics reflects the work of their more quantitative have spent their careers studying nature of the questions com- colleagues. It also implies that have less reason to adopt new parativists have asked. The big producers of quantitative work conceptual tools that only ap- questions in comparative poli- read enough of the non- pear to be helpful in studying tics, such as the origins and sur- quantitative work to understand topics they find uninteresting. vival of democratic and authori- the questions that motivate their Again, both sides have a tarian regimes or the role of the less quantitative colleagues. In point. There is no substitute for state in promoting (or retarding) my vision of a healthy disci- the precision of mathematics in economic growth or income pline, some specialize in quanti- determining the conclusions that equality, are complex and diffi- tative work, some specialize in follow logically from a set of as- cult to specify with the precision non-quantitative work and sumptions. The work of the for- that mathematical models re- some do both, but all learn mal theorist is to eliminate am- quire. One reasonable response enough of each other’s lan- biguity and to simplify until any- is to limit one’s research to nar- guage to be able to participate one can check the validity of the rower topics that can be studied in a common conversation. v deductive argument. If doing so in a rigorous fashion. Another, leads to new research questions, equally reasonable response is so much the better. The intro- to study the big questions, doing duction of new questions is an the best one can, on the important way in which science grounds that the questions are advances. It is a gain for com- too important for human wel- parative politics that there is fare to ignore. much more research today on The recognition that quanti- the comparative organization of tative and non-quantitative work legislatures or on national differ- have different advantages and ences in how votes are aggre- disadvantages has led some to gated than there was ten years argue that the best approach is Visit the ago. to do both. Such advice, how- APSA-CP Newsletter On the other side of the ever, ignores the potential bene- online at ledger is the stubborn fact that fits of a division of labor. While http://www.shelley.polisci. few applications of formal the- I believe that the field will ad- ucla.edu/apsacp. ory have yielded new insights vance most rapidly with a com- sufficiently powerful to have a bination of quantitative and significant impact on the study non-quantitative work, it doesn’t Back issues are being of the questions that concern follow that each individual in added. most non-mathematical com- the field should try to be a mas- There is a one-year delay parativists. Of course, new appli- ter of both trades. What is most before issues cations of formal theory are con- important for the future of the appear on the web site. tinually being developed and no field is for each of us to make one can predict where the next the effort to read articles and Subscribe! advance will occur. But the fact attend seminars by scholars remains that applications of for- working on the other side of mal theory are far less pervasive the divide. The potential bene- in comparative politics than in fits of the division of labor can- American politics. In part, this is not be realized if producers re- due to the natural inclination of sist trading with one another. mathematically trained scholars This implies that producers of apsa-cp Summer 2001 23 Use the Newsletter in the classroom! Visit the APSA-CP Newsletter online at The APSA has authorized university teachers http://www.shelley.polisci.ucla.edu/apsacp. to reproduce articles from the Newsletter for use in the classroom at no charge. Back issues are8 being added. Take advantage of this policy, and introduce There is a one-year delay before issues your graduate students to the latest research, appear on the web site. Subscribe! issues and debates in comparative politics.

How to Subscribe

Subscriptions to the APSA-CP Newsletter are a benefit to members of the Organized Section in Comparative Politics of the American Political Science Association. To join the Section, check the appropriate box when joining the APSA or renewing your Association membership. Section dues are currently $7 annually, with a $2 surcharge for foreign addresses. The printing and mailing of the Newsletter are paid for out of members’ dues. To join the APSA, contact:

American Political Science Association Telephone: (202) 483-2512 1527 New Hampshire Ave., NW Facsimile: (202) 483-2657 Washington, DC 20036 Email: [email protected] USA Changes of address for the Newsletter take place automatically when members change their address with the APSA. Please do not send change of address information to the Newsletter.

APSA-CP Newsletter Non-Profit Organization Professor Daniel Treisman, Editor Department of Political Science U.S. Postage University of California, Los Angeles 405 Hilgard Avenue PAID Los Angeles, California 90095-1472 USA UCLA

UCLA Mail Recharge Code PR39

24 apsa-cp Summer 2001