Hix, simon. a global ranking of political scince departments
-
Upload
tbattibugli -
Category
Technology
-
view
285 -
download
4
Transcript of Hix, simon. a global ranking of political scince departments
A Global Ranking of Political ScienceDepartmentsSimon HixLondon School of Economics
Rankings of academic institutions are key information tools for universities, funding agencies,students and faculty. The main method for ranking departments in political science, throughpeer evaluations, is subjective, biased towards established institutions, and costly in terms oftime and money. The alternative method, based on supposedly ‘objective’ measures of outputsin scientific journals, has thus far only been applied narrowly in political science, using pub-lications in a small number of US-based journals. An alternative method is proposed in thispaper – that of ranking departments based on the quantity and impact of their publicationsin the 63 main political science journals in a given five-year period. The result is a series ofglobal and easily updatable rankings that compare well with results produced by applying asimilar method in economics.
Rankings of academic institutions are key information tools for universities,public and private funding agencies, students and faculty. For example, toinvestigate whether and why European universities lag behind their competi-tors in the US, the European Economics Association commissioned research intothe ranking of economics departments on a global scale (see, especially, Coupé,2003).
A variety of different ranking methods have been used in the natural sciencesand have started to emerge in the social sciences, especially in economics (see, for example, Scott and Mitias, 1996; Dusansky and Vernon, 1998). Allmethods have disadvantages and trade-offs. Nevertheless, the best methodstend to have three elements: (1) they rank institutions on a global scale ratherthan in a single country; (2) they use ‘objective’ measures of research outputs,such as publications in journals, rather than subjective peer evaluations; and(3) they are cheap to update, for example by allowing for mechanized annualupdates.
However, no such global, objective or easily updated method exists in politicalscience. This research aims to fill this gap by proposing and implementing anew method for ranking departments in this field. To this end, in the nextsection I review the existing methods in our discipline. In the third section Ithen propose and justify an alternative method, based on research outputs inthe main political science journals in a particular five-year period. And in thefourth section I present the results of an analysis of the content of 63 journalsbetween 1993 and 2002.
POLITICAL STUDIES REVIEW: 2004 VOL 2, 293–313
© Political Studies Association, 2004. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UKand 350 Main Street, Malden, MA 02148, USA
294 SIMON HIX
Existing Rankings of Political Science DepartmentsAs in other disciplines, two main methods have been used to rank politicalscience departments: peer assessments, and content analysis of scientific jour-nals. However, both methods, as applied thus far, have their limitations.
Peer AssessmentsThe most widely used method for ranking political science departments is peerassessments – where senior academics are asked to evaluate the quality of otherdepartments. For example, this method is used by the US National ResearchCouncil and the U.S. News and World Report to rank doctoral programs in theUS (see PS: Political Science and Politics, 1996a, b), and in the Research Assess-ment Exercise in the UK for the allocation of central government researchfunding.
The problems with this method are well known. First, peer assessments are sub-jective, by definition. No ranking method is perfectly ‘objective’. For example,the content of scientific journals is determined by the subjective judegmentsof journal editors and article reviewers. However, journal editors and articlereviewers are experts on the subjects of the papers they publish or review. Also,peer evaluation in the journal publishing process is repeated thousands oftimes, which reduces bias. In contrast, rankings based on periodic peer assess-ments rely on a small sample of academics, who cannot possibly be experts inall areas of research produced by the institutions they rank. As a result, rank-ings based on peer assessments are less ‘objective’ than rankings based on the content analysis of journals (if a sufficiently large sample of journals isused).
The resulting biases of this subjectivity have been well documented. Becausethe sample of academics has only limited information about the output of allinstitutions, they are forced to base their judgements on other factors. Thisresults in a bias towards large established departments and against new andup-and-coming departments (Katz and Eagles, 1996). The overall reputation ofthe university has an effect on the respondents’ expected performance of apolitical science department – known as the ‘halo effect’ (Lowry and Silver,1996; Jackman and Siverson, 1996).
Second, the peer assessment method is highly costly and time-consuming. Thisis because of the need either to survey a large number of senior faculty (as inthe cases of the US National Research Council and the U.S. News and WorldReport) or to prepare and read the submissions of all the universities (as in thecase of the Research Assessment Exercise). Hence, rankings based on peerassessments are invariably updated only every five years (in the case of theResearch Assessment Exercise and the U.S. News and World Report) or evenlonger (in the case of the US National Research Council).
Third, all existing peer assessment rankings are nationally specific. If similarmethods were used in different countries, a global ranking based on peerassessments could be produced. However, different methods tend to be used
POLITICAL SCIENCE DEPARTMENTS 295
in different countries. For example, in the U.S. News and World Report, depart-ments are scored out of five in a number of criteria, and then averaged (withthe top departments scoring between 4.7 and 4.9). In the Research AssessmentExercise, departments are scored in bands from 5* to 2 (with the top depart-ments scoring 5*). As a result, relative performance on a global scale is diffi-cult to establish.
Content Analysis of Scientific JournalsIn a conscious effort to improve on these peer assessment results, political scientists have begun to develop more objective methods of ranking politicalscience departments. Following the practice in other disciplines, the mostpopular method is the analysis of the content of the leading political sciencejournals (see, for example, Welch and Hibbing, 1983). The assumption behindthis method is that, in contemporary political science, the main output forresearch results is publication in a professional journal.
Publication of books is more common in political science than in economics.Hence, ideally, a ranking based on book publications as well as journal articleswould produce the best results (see Rice et al., 2002). However, analysing thecontent of books and the number of citations to particular book series is costly,since there is not a single database of book publications and book citations likethe Social Science Citation Index (SSCI) for journal publications. One solutioncould be to use peer assessments to rank book publishers (see, for example,Goodson et al., 1999). But this would go against the aim of creating a rankingusing only non-subjective assessments.
Also, a ranking based on book publications may have little added value,because there is probably a high correlation between the outputs of depart-ments in books and journals. At the individual level, some political scientistsprefer to write books while others prefer to write articles. However, top depart-ments probably produce a lot of books as well as a lot of articles, whereas less-good departments probably produce less books and articles. Hence, therankings resulting from these two measures should be similar, at least for thelarger departments.
Consequently, most researchers have analysed journal publications rather thanbook publications. But there are problems with the way this method has beenapplied thus far. First, existing studies have counted only a small number ofjournals. Miller et al. (1996) only looked at the content of the American Polit-ical Science Review (APSR); Teske (1996) looked at APSR, the Journal of Politics(JOP), and the American Journal of Political Science (AJPS) (see Garand andGraddy, 1999); McCormick and Rice (2001) counted articles in APSR, AJPS, JOP,the Western Political Quarterly (WPQ) and Polity; and Ballard and Mitchell(1998) looked at APSR, JOP, AJPS, World Politics, Comparative Politics, theBritish Journal of Political Science (BJPolS), WPQ, Polity and Political ScienceQuarterly. But even nine journals is a rather limited sample of the main jour-nals in political science. The SSCI contains 143 journals in the fields of politicalscience, international relations and public administration. With modern
296 SIMON HIX
computer technology, there is no reason why all, or at least a larger and morerepresentative sample, of these journals cannot be counted.
Second, and partly due to the limited sample of journals coded, the existingrankings based on the content of journals have tended to be biased towardsinstitutions in the US. For example, although APSR is widely respected as the top political science journal, it is nonetheless the ‘in-house’ journal of theAmerican Political Science Association. Not surprisingly, only 7 per cent of articles in APSR between 1996 and 1999 were by scholars based outside the US(Schmitter, 2002). This small number may be a fair reflection of the quality orquantity of research outside the US. However, studying the content of onejournal inevitably risks a high degree of error.
Even in Ballard and Mitchell’s (1998) study of nine journals, only one journalbased outside the US was included (BJPolS). Not surprisingly, not a single non-American department appeared in their top 50. It might be the case that nodepartment outside the US is good enough to be in the top 50. But one wouldbe more inclined to accept this conclusion if this result was based on thecontent of a larger sample of journals and more journals based outside the US.
An Alternative MethodBuilding on existing bibliometric research, the method proposed here ranksacademic institutions on the basis of the quantity and impact of articles pub-lished in the main journals in political science in a given period. To establishthis ranking, decisions were made about the following: (1) what time periodto use; (2) what to count as the ‘main’ political science journals; (3) what tocount as a publication in a journal; (4) how to measure the impact (‘quality’)of a journal; (5) how to construct a ranking from this information; and (6) howto create a ‘quasi-error’ term.
Time PeriodCreating annual rankings would have the advantage of being able to trackshort-term changes in the performance of departments. However, looking atthe content of only one year of each journal would be a small sample size, andso would produce a high degree of measurement error. Conversely, a newranking every ten years would be more accurate, but would not measure moresubtle changes. As a result, counting articles on a rolling five-year basis wouldprobably be the most effective method. This allows for a larger sample in eachjournal and allows a new ranking to be produced every year – in other words,1993–1997, 1994–1998, and so on. This is also a similar time period to otherrankings, such as the U.S. News and World Report and the Research AssessmentExercise.
The Main Political Science JournalsFour steps were taken to define the ‘main’ political science journals. Step oneinvolved the full list of journals in the field in the SSCI, which contained 143
POLITICAL SCIENCE DEPARTMENTS 297
journals in political science, international relations and public administrationin 2002.
Step two involved adding some missing journals to this list. The SSCI does notinclude all major political science journals. The Institute for Scientific Informa-tion (ISI) follows a careful procedure for selecting which journals to include inthe SSCI.1 However, several prominent international political science journalsare not listed in the SSCI. For example, whereas the main journals of the British,German and Scandinavian political science associations are in the SSCI, the mainjournals of the French, Italian and Dutch associations are not. Also, severalmajor sub-field journals were not included before 2002, such as the Journal of Public Policy, European Union Politics, Nations and Nationalism, History ofPolitical Thought, the Journal of Legislative Studies, and Democratization.Adding these journals to the SSCI list makes a total of 152 journals.2
Step three involved setting and applying two simple criteria for divining the‘main’ political science journals from this list of 152. First, many journals are infact journals in other fields of social science, such as law, economics, geogra-phy, sociology, history, psychology, social policy, communications, philosophy,or management. For the sake of simplicity, a political science journal can bedefined as a journal that is (a) edited by a political scientist and (b) has a major-ity of political scientists on its editorial board (in departments or institutes ofpolitical science, politics, government, international relations, public adminis-tration or public policy).
Second, many journals in the SSCI list have a marginal impact on the disciplineof political science. For example, almost one third of the journals had less than100 citations to articles published in any issue of these journals by the articlespublished in the over 8,000 other journals in the SSCI in 2002. Removing thesenon-political-science journals and journals that have only a marginal impact left60 journals.
Step four, however, involved adding back three journals that have a low impactbut are the national political science association journals of three countries: theAustralian Journal of Political Science, Politische Vierteljahresschrift (publishedby the German political science association) and Scandinavian Political Studies.It is reasonable to include these journals despite their low impact, since the ISIhad already decided that these are important journals. In other words, nationalpolitical science association journals are included in the analysis either if theyare in the SSCI or if they are not in the SSCI list but receive more than 100 cita-tions per year.
This left 63 journals for the analysis, which are listed in Table 1. For the 54 journals in the SSCI, data on the content of these journals between 1993 and2002 was purchased from the ISI. For the nine journals not in the SSCI and for the issues of the SSCI journals that are not in the database (for example,where a journal existed for a number of years prior to being included in theSSCI), the content was coded by hand. In total, the content of 495 annualvolumes was collected electronically and the content of 117 volumes was collected by hand.
298 SIMON HIX
Table 1: Journals included in the Analysis
Volumes codedJournal by hand Volumes in SSCI Impact Score
American Political Science Review 1993–2002 8.82American Journal of Political Science 1993–2002 6.91International Organization 1993–2002 5.21Foreign Affairs 1993–2002 4.72Journal of Politics 1993–2002 4.13International Security 1993–2002 3.93Journal of Conflict Resolution 1993–2002 3.72World Politics 1993–2002 3.66Journal of European Public Policy 1994–1996 1997–2002 3.34International Studies Quarterly 1993–2002 3.28Public Choice 1993–2002 3.22Journal of Common Market Studies 1993–2002 2.94British Journal of Political Science 1993–2002 2.84Journal of Peace Research 1993–2002 2.82Journal of Law Economics and Organization 1993–2002 2.80Comparative Political Studies 1993–2002 2.79Journal of Democracy 1993–1994 1995–2002 2.75Europe-Asia Studies 1993–2002 2.64European Union Politics 2000–2002 2.59Political Research Quarterly 1993–2002 2.58West European Politics 1993–1999 2000–2002 2.58Political Studies 1993–2002 2.56PS: Political Science and Politics 1993–2002 2.53European Journal of Political Research 1993–2002 2.46Public Administration 1993–2002 2.44Party Politics 1995–2002 2.38European Journal of International Relations 1995–1996 1997–2002 2.30Comparative Politics 1993–2002 2.27Electoral Studies 1993–2002 2.26Post-Soviet Affairs 1993–2002 2.18Review of International Studies 1993–1994 1995–2002 2.18Security Studies 1993–1995 1996–2002 2.17Politics and Society 1993–2002 2.14Governance 1993–1994 1995–2002 2.09Legislative Studies Quarterly 1993–2002 2.08Political Communication 1993 1994–2002 2.08Political Behavior 1993–1996 1997–2002 2.06International Interactions 1993–2002 2.00Journal of Theoretical Politics 1993–2002 2.00American Politics Quarterly 2001–2002 1993–2000 1.99Millennium-Journal of International Studies 1993–2002 1.96Publius-The Journal of Federalism 1993–2002 1.93Political Theory 1993–2002 1.91Journal of Public Policy 1993–2002 1.85International Affairs 1993–2002 1.82Philosophy and Public Affairs 1993–2001 1.81Political Science Quarterly 1993–2002 1.75International Political Science Review 1993–2002 1.74Democratization 1994–2002 1.70Nations and Nationalism 1995–2002 1.70Australian Journal of Political Science 1993–2002 1.69Journal of Legislative Studies 1995–2003 1.69Canadian Journal of Political Science 1993–2002 1.64Political Quarterly 1993–2002 1.64East European Politics and Societies 1993–2002 1.63Scandinavian Political Studies 1993 1994–2002 1.60Polity 1993–2002 1.53Politische Vierteljahresschrift 1993–2002 1.52Revue française de science politique 1993–2002 1.49Cooperation and Conflict 1993–2002 1.45History of Political Thought 1993–2002 1.40Acta Politica 1993–2002 1.38Rivista Italiana di Scienza Politica 1993–2002 1.33
Note: All issues of journals between 1993 and 2002 were coded. So, if a year is missing in the table, either a journal had notbeen published yet, or a journal was not published in that particular year.
POLITICAL SCIENCE DEPARTMENTS 299
Counting ArticlesSeveral different types of articles are published in these journals. All main articles and research notes were included, and all editorial comments, bookreviews and short notes were excluded. I decided to treat each article orresearch note in the same journal as equivalent regardless of its length, becauseI see no justification for assuming that a shorter article is less important thana longer article in the same journal. There were just over 18,000 such publica-tions in the 63 journals between 1993 and 2002.
Each article was then counted as follows: an article by a single author with asingle institutional affiliation, or by two or more authors from a single insti-tution, scored 1.0 for the institution; an article by two authors from two dif-ferent institutions, or by a single author with two institutional affiliations,counted as 0.5 for each institution; an article by three authors or three institutions counted as 0.333 for each institution; and so on. This method is notideal, as it undervalues collaborative research. However, the alternative isworse: in other words, counting multi-authored articles as having more valuethan single authored articles. Observations where an institutional affiliationcould not be derived from the editorial information were excluded. This left atotal of approximately 24,000 single observations for analysis.
Measuring ImpactSome articles are more significant than others. I assume that an article is as sig-nificant as the overall impact of the journal in which it is published. Two dif-ferent articles in the same journal may have vastly different impacts on thefield. Conversely, some articles may be cited because of the fame of the author.Hence, if one assumes that a journal applies a common standard for acceptanceof a paper for publication, it is reasonable to assume that all articles in a par-ticular journal are of approximately equal quality.
A common measure of the relative ‘impact’ of a journal is the average numberof citations to a journal in a given period. For example, the ISI calculates an‘impact factor’, which is the total number of citations by all other articles inthe ISI database to all articles in a journal in the previous two years, divided bythe number of articles in the journal in the previous two years.
Using a similar method, we could calculate the average annual citations to allarticles in a journal in the ten-year period. However, because it takes time foran article to be noticed, recently published articles are less cited than articlespublished several years ago. Hence, simply counting the average annual cita-tions would create a bias against recently established journals that have nothad long enough to build up their stock of citations.
However, if we assume that the evolution in the number of citations followsthe same functional form, a fixed-effect regression model of annual citationscan be estimated. This would produce a constant for each journal that is ameasure of its relative importance. But the common trend in citations for aparticular journal is non-linear: there tends to be a plateau in the number of
300 SIMON HIX
citations for several years followed by a decline in the number of citations inthe most recent years. Hence, the appropriate common functional form is anegative quadratic equation:
where j (journal) = 1, ... , 63; y (year) = 1, ... , 10; and JOURNAL is a vector of63 binomial variables, one for each journal.
Estimating this model using ordinary least-squares regression produces the fol-lowing results (t-statistics in parentheses): b2 = 17.944 (2.65), b3 = 0.709 (0.590),and 63 constants, ranging from a high of 882.49 citations per year for APSR toa low of 133.49 for the Rivista Italiana di Scienza Politica (RISP).3 An ‘impactscore’ for each journal was than produced from the constants by dividing eachjournal’s constant by 100 (see Table 1). In other words, a paper in APSR is aboutas important as seven papers in RISP.
The journal ‘impact scores’ calculated by this method are highly correlated(0.757) with the SSCI ‘impact scores’ in 2002 for the 54 journals in both the SSCIand my list.4 The high correlation between my index and the SSCI impact indexis not surprising, as both methods are based on the number of citations to arti-cles in journals in a given period. However, there are two advantages of myimpact scores over the SSCI scores. First, my method allows for impact scores tobe calculated for journals that are not included in the SSCI. Second, by assum-ing a common trend in the number of citations over time, my method correctsfor an inherent bias against new journals in the SSCI method.
Finally, it should be noted that because journals that were not main-stream political science journals were removed from the SSCI list, the rankingdoes not include outputs published elsewhere in the social sciences. This mayproduce a bias against departments that try to contribute to general socialscience rather than the narrow discipline of political science. Nevertheless, themethod used to calculate an impact score for each journal reintroduces ameasure of the breadth of a contribution, as the impact score for a journal iscalculated from all citations to articles in the journal from any journal in theSSCI.
Construction of the RankingSome people may be interested in the total output of a department, whereasothers may be interested in the average quality of these outputs or the averageproductivity of a department. For example, the central administration of a uni-versity may wish to know the relative per capita productivity of a department,whereas a prospective graduate student may seek a large department with alot of research-active staff.
So, five separate rankings were created from the data:
• Rank 1 (Quantity) – the total number of articles in the journals by scholarsfrom a particular institution in a five-year period.
ANNUAL CITES JOURNAL YEAR YEARjy y jy jy jy_ = - - +b b b e1 2 32
POLITICAL SCIENCE DEPARTMENTS 301
• Rank 2 (Impact) – the total number of articles in the journals by scholarsfrom a particular institution in a five-year period multiplied by the ‘impactscore’ of the journal in which the article was published.
• Rank 3 (Quantity/Faculty Size) – the total number of articles in the journalsby scholars from a particular institution in a five-year period (as used toproduce Rank 1) divided by the faculty size of the political science depart-ment of that institution.
• Rank 4 (Impact/Faculty Size) – the total number of articles in the journals byscholars from a particular institution in a five-year period multiplied by the‘impact score’ of the journal in which the article was published (as used toproduce Rank 2) divided by the faculty size of the political science depart-ment of that institution.
• Overall Rank – the average position of the institution in the other fourranks.
The overall ranking is consequently an unweighted sum of the other four rank-ings (compare with Coupé, 2003). Invariably, people will have different opin-ions about the relative importance of Ranks 1, 2, 3 and 4. Hence, the positionsof the institutions in each of the four individual ranks are also reported so thatan interested person can calculate a different overall rank using a different setof weighting of the other ranks.
The information on the size of a department was gathered from two sources.First, for the British universities, the data is the number of full-time staff submitted in the Politics and International Relations section of the 2001Research Assessment Exercise. Second, for all other universities (including those British universities who did not make a submission for this section in2001), we counted the number of full-time staff with a rank of full, associateor assistant professor (or equivalent) listed on a department’s website inNovember to December 2003.5 In other words, this includes only the numberof staff in a political science department plus related institutes, or the numberof political scientists in a department or faculty of social science. For example,according to the Harvard University website, the number of permanent facultyin the Department of Government plus the number of permanent faculty inthe Kennedy School of Government who describe themselves as ‘political scientists’ is 87.
Several things are worth noting here. First, this method of counting the size ofa department assumes that the number of political scientists in a particularinstitution remains constant, which clearly is not the case. Second, this methodonly counts academics in political science departments, whereas the methodfor counting research output counts anyone publishing in one of the journalsfrom a particular institution, regardless of where they are based in an institu-tion. For example, if someone from a business school, an economics depart-ment or a law department publishes in one of the journals, this person iscounted as a political scientist, but is not included as a member of the politi-cal science faculty in their institution. However, although there may be peopleoutside a political science department who do political science research, the
302 SIMON HIX
size of the political science department is probably a reasonable proxy for thesize of the overall political science community in an institution.
A Quasi-ErrorFinally, a ‘quasi-error’ in the overall rank of each institution was calculated.There are two sources of measurement error in this analysis. First, in countingthe number of articles published by an institution, an article may have beenmissed. For example, in the computer data, an article may have been misla-belled as a minor note rather than a proper article, the institutional affiliationof an author may have been entered incorrectly (although each entry in thedata was carefully checked), or an author who was listed as having no institu-tional affiliation may have been based in a particular academic institution.Second, it is extremely difficult to accurately measure the faculty size of adepartment. For example, different academic systems have different ways ofdescribing their faculty (for example, many German universities only list theirfull professors). Also, information on the departments’ websites is invariablyout of date or inaccurate.
Using these two sources of error, a ‘quasi-error’ was worked out by calculatingwhere an institution would have been placed in the overall ranking if the institution had produced one more/less article in a journal with a mean impactscore (2.52) and if the department was 5 per cent smaller/larger than it hadbeen measured.
For example, in 1998–2002, the London School of Economics, with a faculty sizeof 76, produced 143.31 articles with an impact of 338.87. This placed it 2nd inRank 1 (Quantity), 4th in Rank 2 (Impact), 31st in Rank 3 (Quantity/Faculty Size),57th in Rank 4 (Impact/Faculty Size), and 15th overall. If it had produced onemore article in a mean-impact score journal and had 5 per cent less staff, itsposition would not have changed in Ranks 1 and 2, but would have risen to24th in Rank 3, 51st in Rank 4, and 12th overall. Conversely, if it had one lessarticle and 5 per cent more staff, it would have been 18th overall. So, the quasi-error at 15th was 12–18 (or plus/minus three places).
ResultsTable 2 lists the ‘Global Top 200’ political science institutions on the basis oftheir output in the main political science journals in the five years between1998 and 2002.6 Anyone with a cursory knowledge of the discipline would recognize most of the names on the list.
One way of assessing the validity of the method is to compare the results tothose using a similar method in economics (Coupé, 2003). In the political sciencerankings for 1998–2002, there was one department outside the US in the top10, five in the top 20, fourteen in the top 50, thirty-six in the top 100, and 103in the top 200. In the comparable ranking in economics, there were no depart-ments outside the US in the top 10, one in the top 20, ten in the top 50, thirty-four in the top 100, and eighty-eight in the top 200.
POLITICAL SCIENCE DEPARTMENTS 303
One obvious criticism is that these rankings are biased towards English-speaking countries, since nine of the top 10, nineteen of the top 20, forty-eightof the top 50, ninety-one of the top 100, and 163 of the top 200 are from theUS, the UK, Australia, Canada or Ireland. However, the equivalent rankings in economics are equally as dominated by Anglo-Saxon institutions: with all of the top 10, all of the top 20, forty-seven of the top 50, eighty-seven of thetop 90, and 155 of the top 200 coming from these same five English-speakingcountries. In other words, the dominance of institutions from these countriesmay simply be a reflection of the dominant position of English as the globallanguage in the social sciences.
So, if one assumes that the global spread of good and bad departments isbroadly similar in the disciplines of political science and economics, then themethod outlined and applied here is as good as the most similar ranking ineconomics.
Table 3 shows the rank-order correlations between the five rankings, using theresults for the top 200 in the 1998–2002 period. As would be expected giventhe calculation method, there are high correlations between Ranks 1 and 3 andbetween Ranks 2 and 4. However, the correlations suggest that each rankingmethod measures something different.
Finally, Table 4 shows the ‘rolling’ rankings for the six five-year periods between1993 and 2002. One of the striking things here is the stability of the top three,with Stanford, Harvard and Columbia swapping places at the top of the list,and none of these institutions dropping below third. Only two other institu-tions remained in the top 10 throughout this period (Indiana and the Univer-sity of California, San Diego), and thirty-six institutions remained in the top 50throughout the period. The biggest climbers, who climbed more than thirtyplaces between 1993–1997 and 1998–2002, were the State University of NewYork, Binghamton (from 117th to 19th), Aberystwyth (136th to 39th), PennState (101st to 33rd), Geneva (104th to 43rd), Trinity College Dublin (96th to40th), University College London (83rd to 46th), Illinois-Urbana Champaign(80th to 44th) and Georgetown (50th to 16th). Nevertheless, almost fifty per cent (twenty-four) of the institutions in the top 50 in 1998–2002 rose orfell less than ten places from their positions in 1993–1997.
ConclusionA reasonably objective, easily updated and global ranking of departments inpolitical science can be produced by borrowing a method used in other disci-plines – that of measuring the research output of institutions in the main jour-nals in the field in a given period, and controlling for the number of full-timestaff in a department. This method produces a series of rankings that seemintuitively correct and compare well with the equivalent rankings in econom-ics, in terms of the regional and national balance of institutions in the top 10,20, 50, 100 and 200.
One possible problem with these rankings is the apparent English-languagebias in the results, which undermines the aspiration to be truly ‘global’.
304 SIMON HIX
Tab
le 2
:Th
e G
lob
al T
op
200
Po
litic
al S
cien
ce D
epar
tmen
ts,
1998
–200
2
Qu
anti
ty (
1)Im
pac
t (2
)Q
uan
tity
/Siz
e (3
)Im
pac
t/Si
ze (
4)A
vera
ge
Qu
asi-
Erro
rO
vera
llFa
cult
yN
o.
of
Art
icle
s*A
rtic
les/
Imp
act/
of
Ran
ksR
ank
Un
iver
sity
Co
un
try
Size
Art
icle
sR
ank
Imp
act
Ran
kFa
c. S
ize
Ran
kFa
c. S
ize
Ran
k1
to 4
Bes
tW
ors
t
1C
olu
mb
iaU
SA45
120.
694
420.
102
2.68
27
9.33
62
3.75
11
2H
arva
rdU
SA87
204.
371
700.
931
2.34
915
8.05
76
5.75
23
3St
anfo
rdU
SA38
90.9
46
342.
883
2.39
314
9.02
34
6.75
23
4O
hio
Sta
teU
SA43
84.9
510
327.
875
1.97
624
7.62
59
12.0
04
55
EUI
Ital
y17
62.0
817
157.
9730
3.65
22
9.29
23
13.0
05
5
6U
C,
San
Die
go
USA
3774
.22
1225
4.53
112.
006
226.
879
1615
.25
69
7U
C,
Irvi
ne
USA
3271
.34
1321
4.71
172.
229
186.
710
1916
.75
68
8In
dia
na
USA
3367
.49
1622
4.68
162.
045
216.
808
1717
.50
712
9Pr
ince
ton
USA
4987
.50
731
9.40
71.
786
396.
518
2218
.75
612
10Y
ale
USA
5291
.90
532
3.59
61.
767
416.
223
2519
.25
714
11U
C,
Ber
kele
yU
SA45
86.6
28
268.
3410
1.92
529
5.96
331
19.5
06
1212
Mic
hig
an S
tate
USA
2957
.00
2420
4.69
191.
966
257.
058
1320
.25
815
13C
hic
ago
USA
3970
.98
1423
8.39
131.
820
356.
113
2822
.50
918
14U
C,
Los
An
gel
esU
SA51
85.5
69
317.
138
1.67
848
6.21
826
22.7
510
1815
LSE
UK
7614
3.31
233
8.87
41.
886
314.
459
5723
.50
1218
16=
Geo
rget
ow
nU
SA43
77.9
111
233.
4214
1.81
236
5.42
840
25.2
513
1816
=Es
sex
UK
2355
.83
2514
0.44
342.
427
136.
106
2925
.25
1318
18M
ITU
SA30
53.9
128
201.
4220
1.79
738
6.71
418
26.0
013
2119
=A
NU
Au
stra
lia25
61.5
318
127.
5143
2.46
110
5.10
045
29.0
019
2119
=SU
NY,
Bim
gh
amto
nU
SA16
42.0
050
120.
8648
2.62
58
7.55
410
29.0
019
2119
=O
xfo
rdU
K70
122.
083
302.
999
1.74
444
4.32
860
29.0
016
21
22B
irm
ing
ham
UK
2049
.08
3311
5.82
522.
454
115.
791
3532
.75
1927
23C
amb
rid
ge
UK
1849
.25
3210
6.41
622.
736
55.
912
3333
.00
2227
24Fl
ori
da
Stat
eU
SA25
43.5
345
167.
6926
1.74
145
6.70
820
34.0
019
2925
=Sh
effi
eld
UK
2250
.41
3111
3.01
532.
291
165.
137
4335
.75
2429
25=
Was
hin
gto
nU
SA23
39.9
154
161.
1029
1.73
546
7.00
414
35.7
522
30
27M
ich
igan
USA
4867
.69
1525
0.72
121.
410
775.
223
4136
.25
2232
28=
Joh
ns
Ho
pki
ns
USA
2143
.03
4712
2.48
472.
049
205.
832
3437
.00
2432
28=
Texa
s A
&M
USA
4158
.75
2022
5.92
151.
433
745.
510
3937
.00
2233
30Em
ory
USA
3248
.83
3518
1.00
231.
526
615.
656
3739
.00
2436
31C
olo
rad
oU
SA22
42.4
149
123.
3945
1.92
828
5.60
938
40.0
025
3432
Am
eric
anU
SA20
36.9
561
126.
9644
1.84
834
6.34
823
40.5
025
3833
Pen
nsy
lvan
iaU
SA25
40.0
653
150.
6232
1.60
253
6.02
530
42.0
030
3934
Bri
sto
lU
K14
38.3
159
83.0
478
2.73
65
5.93
132
43.5
031
3835
UN
C C
hap
el H
illU
SA37
53.4
529
177.
5124
1.44
572
4.79
850
43.7
530
41
POLITICAL SCIENCE DEPARTMENTS 305
36G
eorg
e W
ash
ing
ton
USA
4459
.18
1920
5.19
181.
345
874.
663
5244
.00
3143
37=
Car
dif
fU
K10
30.3
375
74.5
290
3.03
34
7.45
211
45.0
033
3937
=U
W M
adis
on
USA
3957
.94
2216
7.32
271.
486
694.
290
6245
.00
3142
39A
ber
ystw
yth
UK
2548
.50
3810
6.76
601.
940
264.
270
6547
.25
3344
40=
TCD
Irel
and
927
.50
8871
.42
933.
056
37.
936
747
.75
3943
40=
Van
der
bilt
USA
1528
.83
8110
3.23
651.
922
306.
882
1547
.75
3446
42C
orn
ell
USA
3148
.08
3913
3.22
401.
551
594.
297
6149
.75
3648
43G
enev
aSw
itze
rlan
d12
30.1
776
74.5
889
2.51
49
6.21
527
50.2
539
4844
Illin
ois
, U
rban
a-C
ham
pai
gn
USA
4052
.28
3018
2.68
221.
307
974.
567
5551
.00
3948
45R
ice
USA
1528
.16
8799
.00
701.
877
326.
600
2152
.50
3951
46U
CL
UK
525
.67
9651
.75
117
5.13
41
10.3
501
53.7
546
5047
=SU
NY,
Sto
ny
Bro
ok
USA
1524
.33
102
109.
5255
1.62
251
7.30
112
55.0
044
5147
=U
C,
Dav
isU
SA25
36.3
063
127.
6942
1.45
271
5.10
844
55.0
042
5149
=A
rizo
na
USA
2235
.25
6710
8.51
571.
602
534.
932
4856
.25
4552
49=
Vir
gin
iaU
SA35
44.3
744
164.
1928
1.26
810
24.
691
5156
.25
4451
51D
uke
USA
4455
.73
2617
0.44
251.
267
103
3.87
475
57.2
545
5152
Osl
oN
orw
ay33
44.7
542
122.
8446
1.35
686
3.72
277
62.7
552
6753
Cla
rem
on
t G
rad
uat
eU
SA8
19.4
512
863
.30
105
2.43
112
7.91
38
63.2
552
5754
Pitt
sbu
rgh
USA
2941
.33
5111
2.75
541.
425
763.
888
7363
.50
5262
55Le
iden
Net
her
lan
ds
2238
.81
5782
.21
801.
764
423.
737
7663
.75
5263
56Io
wa
USA
2837
.33
6011
9.75
491.
333
904.
277
6465
.75
5263
57N
ew M
exic
oU
SA16
23.6
710
410
0.74
681.
479
706.
296
2466
.50
5268
58N
ew Y
ork
USA
3847
.58
4013
5.96
381.
252
107
3.57
888
68.2
552
6859
Min
nes
ota
USA
5557
.79
2319
5.42
211.
051
138
3.55
392
68.5
052
6860
Geo
rge
Mas
on
USA
3542
.50
4813
0.13
411.
214
109
3.71
878
69.0
056
70
61H
ebre
w,
Jeru
sale
mIs
rael
2943
.33
4610
0.61
691.
494
673.
469
9669
.50
5270
62A
rizo
na
Stat
eU
SA24
32.7
569
102.
8566
1.36
584
4.28
563
70.5
056
7363
Hu
llU
K21
35.7
565
75.7
687
1.70
247
3.60
886
71.2
555
7364
Mar
ylan
dU
SA45
48.9
734
156.
6731
1.08
813
03.
482
9572
.50
5573
65G
eorg
iaU
SA19
29.0
080
79.9
083
1.52
661
4.20
567
72.7
556
74
66=
SUN
Y, B
uff
alo
USA
1526
.33
9465
.79
971.
755
434.
386
5873
.00
5874
66=
Ho
ust
on
USA
2732
.08
7111
7.25
501.
188
112
4.34
359
73.0
058
7468
No
rth
wes
tern
USA
3132
.56
7014
4.52
331.
050
139
4.66
253
73.7
558
7369
Pen
nsy
lvan
iaU
SA29
38.9
256
103.
4464
1.34
288
3.56
790
74.5
057
7570
Cal
Tec
hU
SA9
14.7
516
179
.38
841.
639
498.
820
574
.75
5875
71=
Sou
ther
n C
alif
orn
iaU
SA18
25.3
399
82.1
781
1.40
778
4.56
556
78.5
062
8071
=W
arw
ick
UK
2738
.50
5887
.31
741.
426
753.
234
107
78.5
063
8073
Tel
Avi
vIs
rael
2131
.33
7275
.39
881.
492
683.
590
8778
.75
6280
74M
ann
hei
mG
erm
any
1828
.33
8565
.71
981.
574
563.
651
8480
.75
6883
75St
rath
clyd
eU
K18
28.6
682
64.3
310
01.
592
553.
574
8981
.50
6983
306 SIMON HIX
Tab
le 2
:Th
e G
lob
al T
op
200
Po
litic
al S
cien
ce D
epar
tmen
ts,
1998
–200
2: C
on
tin
ued
Qu
anti
ty (
1)Im
pac
t (2
)Q
uan
tity
/Siz
e (3
)Im
pac
t/Si
ze (
4)A
vera
ge
Qu
asi-
Erro
rO
vera
llFa
cult
yN
o.
of
Art
icle
s*A
rtic
les/
Imp
act/
of
Ran
ksR
ank
Un
iver
sity
Co
un
try
Size
Art
icle
sR
ank
Imp
act
Ran
kFa
c. S
ize
Ran
kFa
c. S
ize
Ran
k1
to 4
Bes
tW
ors
t
76M
isso
uri
USA
2127
.17
9086
.09
761.
294
994.
100
7083
.75
6881
77W
ash
ing
ton
USA
4546
.98
4113
7.04
361.
044
140
3.04
511
984
.00
7183
78A
arh
us
Den
mar
k46
54.1
727
116.
3751
1.17
811
42.
530
147
84.7
574
8379
=N
ort
h T
exas
USA
2930
.95
7410
7.74
581.
067
133
3.71
579
86.0
074
8679
=Su
ssex
UK
2735
.67
6686
.52
751.
321
933.
204
110
86.0
071
84
81U
W,
Milw
auke
eU
SA20
25.3
399
82.2
279
1.26
710
34.
111
6987
.50
7185
82A
ber
dee
nU
K17
26.4
293
58.7
211
01.
554
583.
454
9789
.50
7488
83N
ewca
stle
-Up
on
-Tyn
eU
K14
22.8
310
851
.42
119
1.63
150
3.67
382
89.7
576
9284
UC
San
ta B
arb
ara
USA
2428
.37
8484
.83
771.
182
113
3.53
593
91.7
576
8885
Gla
sgo
wU
K14
21.2
511
451
.62
118
1.51
863
3.68
781
94.0
081
100
86Le
ices
ter
UK
1321
.00
119
47.5
212
71.
615
523.
655
8395
.25
8110
287
Man
ches
ter
UK
3136
.08
6487
.43
731.
164
118
2.82
013
196
.50
8495
88R
och
este
rU
SA29
28.6
583
101.
3467
0.98
815
03.
494
9498
.50
8296
89Lo
uis
ian
a St
ate
USA
2325
.50
9777
.40
851.
109
122
3.36
599
100.
7585
103
90=
Bir
kbec
k, L
on
do
nU
K13
20.2
512
346
.20
134
1.55
857
3.55
491
101.
2584
106
90=
Ru
tger
sU
SA53
44.5
343
136.
3037
0.84
018
02.
572
145
101.
2586
100
92Sy
racu
seU
SA40
39.0
355
106.
6261
0.97
615
22.
666
139
101.
7586
102
93To
ron
toC
anad
a66
58.0
021
140.
2535
0.87
916
92.
125
183
102.
0086
102
94K
ansa
sU
SA25
27.1
591
80.2
382
1.08
613
13.
209
109
103.
2585
104
95U
C,
Riv
ersi
de
USA
1114
.62
163
56.0
611
41.
329
925.
096
4610
3.75
8510
5
96B
rad
ford
UK
1723
.17
106
52.8
111
61.
363
853.
106
116
105.
7587
107
97H
um
bo
ldt
Ger
man
y17
22.2
511
156
.11
113
1.30
996
3.30
110
410
6.00
8610
698
Wes
tern
Au
stra
liaA
ust
ralia
815
.50
152
31.8
817
41.
938
273.
985
7210
6.25
8710
999
=Ed
inb
urg
hU
K16
22.5
011
049
.86
124
1.40
679
3.11
611
510
7.00
8610
799
=Le
eds
UK
2831
.00
7374
.40
911.
107
124
2.65
714
010
7.00
8910
7
101
Du
rham
UK
814
.83
158
32.6
816
71.
854
334.
085
7110
7.25
8710
910
2A
lab
ama
USA
1317
.17
142
50.4
012
31.
321
933.
877
7410
8.00
8710
710
3Q
MU
LU
K14
21.0
011
943
.38
145
1.50
065
3.09
911
711
1.50
8810
910
4D
artm
ou
thU
SA21
21.8
311
270
.28
941.
040
141
3.34
710
111
2.00
9210
910
5So
uth
Car
olin
aU
SA36
30.1
776
103.
6163
0.83
818
22.
878
129
112.
5096
107
POLITICAL SCIENCE DEPARTMENTS 307
106
Mo
ntr
eal
Can
ada
2224
.90
101
63.4
610
31.
132
121
2.88
512
811
3.25
9510
910
7Tu
fts
USA
1819
.83
127
60.4
310
81.
102
125
3.35
710
011
5.00
9611
410
8M
ax P
lan
ckG
erm
any
3229
.83
7876
.63
860.
932
161
2.39
515
411
9.75
103
121
109
Texa
s, A
ust
inU
SA59
41.3
351
133.
3839
0.70
122
22.
261
168
120.
0010
311
711
0G
ron
ing
enN
eth
erla
nd
s15
18.2
513
546
.27
133
1.21
710
83.
085
118
123.
5010
312
6
111
Sou
tham
pto
nU
K19
21.0
811
550
.93
120
1.10
912
22.
681
138
123.
7510
712
711
2=G
eorg
ia S
tate
USA
2223
.17
106
57.1
711
11.
053
137
2.59
914
312
4.25
108
126
112=
Ken
tuck
yU
SA23
21.0
011
969
.08
950.
913
162
3.00
312
112
4.25
107
126
114
Live
rpo
ol
UK
1618
.75
130
47.3
112
91.
172
115
2.95
712
512
4.75
107
129
115
Bri
tish
Co
lum
bia
Can
ada
2927
.33
8966
.94
960.
942
159
2.30
816
212
6.50
108
131
116
Am
ster
dam
Net
her
lan
ds
6248
.66
3710
9.31
560.
785
193
1.76
322
212
7.00
108
126
117
Man
ches
ter
Met
rop
olit
anU
K8
12.3
318
329
.16
186
1.54
160
3.64
585
128.
5010
813
611
8N
otr
e D
ame
USA
4834
.83
6810
7.03
590.
726
214
2.23
017
412
8.75
108
129
119
Den
ver
USA
610
.83
199
25.1
821
21.
805
374.
197
6812
9.00
107
133
120
SUM
Y, A
lban
yU
SA25
23.8
310
361
.18
106
0.95
315
72.
447
152
129.
5011
013
3
121
Ken
tU
K16
18.5
013
445
.13
136
1.15
611
92.
821
130
129.
7510
813
312
2Ex
eter
UK
1619
.25
129
42.9
414
61.
203
110
2.68
413
713
0.50
108
133
123
Hel
sin
kiFi
nla
nd
1720
.25
123
43.5
414
41.
191
111
2.56
114
613
1.00
108
133
124=
Bro
wn
USA
2018
.67
132
58.8
910
90.
934
160
2.94
512
613
1.75
110
133
124=
Wes
t V
irg
inia
USA
1718
.67
132
47.0
613
11.
098
129
2.76
813
513
1.75
110
133
126
No
ttin
gh
amU
K21
21.0
811
550
.78
121
1.00
414
52.
418
153
133.
5011
013
412
7Li
verp
oo
l Jo
hn
Mo
ore
sU
K4
8.50
236
20.3
524
12.
125
195.
088
4713
5.75
108
149
128=
CU
NY
USA
5036
.33
6295
.90
710.
727
212
1.91
820
113
6.50
117
134
128=
Lafa
yett
eU
SA9
13.0
017
729
.04
188
1.44
473
3.22
710
813
6.50
108
143
130
Mu
rdo
chA
ust
ralia
712
.50
182
22.8
522
31.
786
393.
264
105
137.
2511
014
3
131
East
An
glia
UK
1416
.00
148
38.8
115
01.
143
120
2.77
213
413
8.00
116
139
132
Mel
bo
urn
eA
ust
ralia
2021
.33
113
44.7
613
81.
067
133
2.23
817
313
9.25
115
142
133
Ko
nst
anz
Ger
man
y29
25.5
097
60.9
210
70.
879
169
2.10
118
613
9.75
119
138
134
QU
BU
K22
20.8
112
249
.31
126
0.94
615
82.
241
172
144.
5012
214
313
5=U
CD
Irel
and
1313
.50
173
38.5
015
21.
038
142
2.96
212
314
7.50
126
147
135=
St A
nd
rew
sU
K11
14.5
016
528
.97
189
1.31
895
2.63
414
114
7.50
122
151
137
Twen
teN
eth
erla
nd
s12
15.1
715
529
.81
184
1.26
410
62.
484
149
148.
5012
415
613
8Te
xas
Tech
no
log
ical
USA
2118
.75
130
47.2
413
00.
893
167
2.25
016
914
9.00
131
151
139
Tru
man
Sta
teU
SA8
11.0
019
725
.07
213
1.37
581
3.13
411
315
1.00
126
158
140
Bre
men
Ger
man
y15
16.5
014
633
.67
163
1.10
012
72.
245
170
151.
5013
115
8
308 SIMON HIX
Tab
le 2
:Th
e G
lob
al T
op
200
Po
litic
al S
cien
ce D
epar
tmen
ts,
1998
–200
2: C
on
tin
ued
Qu
anti
ty (
1)Im
pac
t (2
)Q
uan
tity
/Siz
e (3
)Im
pac
t/Si
ze (
4)A
vera
ge
Qu
asi-
Erro
rO
vera
llFa
cult
yN
o.
of
Art
icle
s*A
rtic
les/
Imp
act/
of
Ran
ksR
ank
Un
iver
sity
Co
un
try
Size
Art
icle
sR
ank
Imp
act
Ran
kFa
c. S
ize
Ran
kFa
c. S
ize
Ran
k1
to 4
Bes
tW
ors
t
141
Will
iam
an
d M
ary
USA
1312
.67
180
38.7
615
10.
975
154
2.98
212
215
1.75
130
152
142
Kee
leU
K31
26.0
095
55.8
711
50.
839
181
1.80
221
715
2.00
134
149
143
Sim
on
Fra
ser
Can
ada
2018
.00
137
44.0
013
90.
900
165
2.20
017
615
4.25
134
158
144
Bo
sto
nU
SA33
23.3
310
564
.04
101
0.70
722
11.
941
199
156.
5013
515
814
5N
UST
No
rway
1815
.83
150
41.5
714
70.
879
169
2.30
916
115
6.75
134
159
146
Mis
siss
ipp
iU
SA12
10.1
720
440
.14
148
0.84
817
73.
345
102
157.
7513
416
414
7N
ijmeg
enN
eth
erla
nd
s16
15.9
114
934
.62
161
0.99
414
92.
164
177
159.
0013
416
114
8U
pp
sala
Swed
en38
27.0
892
65.3
199
0.71
321
71.
719
231
159.
7513
515
814
9N
ebra
ska
USA
1512
.25
184
43.5
814
30.
817
187
2.90
512
716
0.25
135
160
150
Loyo
laU
SA14
14.2
016
631
.76
175
1.01
414
32.
269
166
162.
5013
516
8
151
Gri
ffith
Au
stra
lia16
17.0
014
331
.62
177
1.06
313
51.
976
196
162.
7513
516
815
2Io
wa
Stat
eU
SA27
16.8
314
463
.75
102
0.62
325
22.
361
157
163.
7514
216
815
3=Fl
ori
da
USA
4729
.83
7873
.26
920.
635
246
1.55
924
416
5.00
144
167
153=
UW
EU
K13
13.1
717
630
.42
181
1.01
314
42.
340
159
165.
0013
717
115
5C
op
enh
agen
Den
mar
k38
28.2
586
57.1
611
20.
743
207
1.50
425
716
5.50
143
167
156
Okl
aho
ma
USA
3522
.83
108
63.4
010
40.
652
238
1.81
121
316
5.75
143
167
157
Ber
nSw
itze
rlan
d17
14.8
315
836
.23
157
0.87
217
32.
131
181
167.
2514
117
115
8V
ien
na
Au
stri
a23
17.4
214
045
.44
135
0.75
720
01.
976
196
167.
7514
417
115
9Ill
ino
is,
Ch
icag
oU
SA22
14.6
716
250
.50
122
0.66
723
32.
295
164
170.
2514
317
416
0N
ew S
ou
th W
ales
Au
stra
lia17
15.3
315
332
.11
173
0.90
216
41.
889
204
173.
5014
717
7
161
Sou
ther
n I
llin
ois
USA
1712
.58
181
39.9
014
90.
740
208
2.34
715
817
4.00
150
176
162
No
ttin
gh
am T
ren
tU
K17
14.8
315
832
.75
166
0.87
217
31.
926
200
174.
2514
717
816
3Sy
dn
eyA
ust
ralia
2017
.33
141
35.1
715
90.
867
175
1.75
922
517
5.00
148
174
164
Ree
dU
SA4
5.50
340
20.7
723
81.
375
815.
193
4217
5.25
138
197
165
Port
lan
d S
tate
USA
36.
0031
714
.64
319
2.00
023
4.88
049
177.
0013
919
7
166
Stir
ling
UK
79.
0022
319
.57
253
1.28
610
02.
796
133
177.
2513
918
816
7Ta
sman
iaA
ust
ralia
1414
.00
170
27.7
619
81.
000
147
1.98
319
517
7.50
146
179
168
Rea
din
gU
K16
13.3
717
532
.64
168
0.83
618
32.
040
189
178.
7515
018
0
POLITICAL SCIENCE DEPARTMENTS 309
169
Lan
cast
erU
K18
15.2
315
432
.26
172
0.84
617
81.
792
218
180.
5015
318
017
0=Sc
ien
ce-P
oFr
ance
8548
.83
3591
.03
720.
574
277
1.07
134
218
1.50
160
175
170=
Bat
hU
K19
14.5
816
435
.18
158
0.76
719
81.
852
206
181.
5015
718
3
172
INSE
AD
Fran
ce4
5.50
340
18.5
226
71.
375
814.
630
5418
5.50
143
206
173=
Bo
wd
oin
USA
158.
5023
649
.60
125
0.56
728
13.
307
103
186.
2515
019
517
3=TU
Dar
mst
adt
Ger
man
y5
6.50
299
18.5
126
81.
300
983.
702
8018
6.25
143
207
175
GIIS
Swit
zerl
and
57.
5826
715
.64
308
1.51
664
3.12
811
418
8.25
143
207
176
Go
then
bu
rgSw
eden
1814
.17
167
31.7
117
60.
787
191
1.76
222
318
9.25
159
195
177
Wes
tmin
iste
rU
K8
9.33
216
18.4
627
11.
166
117
2.30
816
219
1.50
155
207
178
De
Mo
ntf
ort
UK
1311
.00
197
25.9
721
00.
846
178
1.99
819
319
4.50
161
200
179
Leh
igh
USA
76.
8329
122
.35
227
0.97
615
23.
193
112
195.
5015
320
818
0Q
uee
nsl
and
Au
stra
lia22
16.5
014
632
.35
171
0.75
020
31.
470
267
196.
7516
919
9
181
Lun
dSw
eden
3421
.00
119
44.8
613
70.
618
253
1.31
928
719
9.00
172
204
182
Leu
ven
(K
UL)
Bel
giu
m24
15.6
715
136
.30
156
0.65
323
71.
513
254
199.
5017
220
518
3U
CLA
NU
K6
7.00
283
17.7
527
91.
167
116
2.95
812
420
0.50
153
217
184
Tub
ing
enG
erm
any
68.
0024
814
.88
317
1.33
390
2.48
015
020
1.25
159
226
185
Vic
tori
aC
anad
a15
11.0
019
727
.86
196
0.73
321
01.
857
205
202.
0017
220
8
186
EU V
iad
rin
aG
erm
any
119.
0022
323
.40
221
0.81
818
62.
127
182
203.
0017
021
418
7B
ryn
Maw
rU
SA5
6.33
306
16.2
029
61.
266
105
3.24
010
620
3.25
157
228
188
Flo
ren
ceIt
aly
3021
.00
119
36.3
715
50.
700
225
1.21
231
720
4.00
172
202
189
CEU
Hu
ng
ary
3118
.17
136
43.9
114
00.
586
269
1.41
627
320
4.50
173
205
190
Erla
ng
en N
urn
ber
gG
erm
any
24.
5038
911
.52
381
2.25
017
5.76
036
205.
7515
923
1
191
Staf
ford
shir
eU
K8
7.83
257
18.9
726
10.
979
151
2.37
115
520
6.00
165
228
192
Fern
of
Hag
enG
erm
any
108.
0024
822
.79
224
0.80
018
92.
279
165
206.
5017
222
019
3C
on
nec
ticu
tU
SA33
17.5
013
947
.33
128
0.53
029
11.
434
269
206.
7517
720
819
4Yo
rkU
K20
14.0
017
030
.57
180
0.70
022
51.
529
253
207.
0017
221
119
5B
erg
enN
orw
ay16
11.9
218
626
.11
207
0.74
520
61.
632
235
208.
5017
521
5
196=
McM
aste
rC
anad
a19
12.8
317
929
.49
185
0.67
523
11.
552
245
210.
0017
721
719
6=So
uth
ern
Met
ho
dis
tU
SA15
10.5
020
227
.21
202
0.70
022
51.
814
211
210.
0017
521
519
8=C
arle
ton
Can
ada
3519
.90
126
43.7
014
20.
569
280
1.24
930
821
4.00
179
211
198=
Juan
Mar
chSp
ain
45.
5034
013
.77
337
1.37
581
3.44
398
214.
0016
525
120
0U
lste
rU
K9
8.17
244
18.5
026
90.
908
163
2.05
618
821
6.00
172
240
310 SIMON HIX
Table 3: Correlations between the Ranks of the Top 200 PoliticalScience Institutions, 1998–2003
Rank 1 Rank 2 Rank 3 Rank 4
Rank 1 (Quantity) –Rank 2 (Impact) 0.962 –Rank 3 (Quantity/Faculty Size) 0.429 0.405 –Rank 4 (Impact/Faculty Size) 0.507 0.583 0.896 –Overall Rank 0.862 0.879 0.759 0.832
Method: Spearman’s rank-order correlation.
However, English is the international language for the publication and citationof research in political science, as in other social sciences and the natural sci-ences. Because of the ease of reading, publishing in and teaching from theseinternational journals, scholars in English-speaking universities are inevitablymore closely integrated into the global discipline than scholars outside theEnglish-speaking world. As a result, a ranking of departments using researchpublished in the ‘top’ international journals in a field is inevitably not a fairrepresentation of the quality of departments outside the English-speakingworld.
One possible solution would be to include more non-English-language journalsin the analysis. However, given the low number of citations to research published in non-English-language journals, it is hard to make a case for includ-ing some non-English journals while omitting others, or even for including non-English-language journals with low citations while excluding some journalswith higher citations.
A second problem is that book publications are more common and importantin political science than in economics. As discussed, if one assumes that a gooddepartment would produce a lot of articles as well as books, then only mea-suring journal publications may not make a difference to the ranking of insti-tutions at the departmental level. Nevertheless, this hypothesis can only bechecked if a similar ranking could be established using book publications, andthe results of the two rankings are compared and perhaps integrated.
Despite these shortcomings, two major advantages of the method proposedhere are that it would be (i) simple to mechanize and (ii) easy to add otherjournals or books to the dataset. If ‘the discipline’, perhaps via a committee ofthe International Political Science Association, could agree a set of English andnon-English-language journals and book publishers that are the main vehiclesfor research output in the global discipline, it would not be too difficult tomodify this method and establish a mechanized system for entry and updatingof the dataset and for calculating new rankings every year. Ideally, each insti-
POLITICAL SCIENCE DEPARTMENTS 311
Table 4: The Rolling Global Top Fifty, 1997–2002
1993–1997 1994–1998 1995–1999 1996–2000 1997–2001 1998–2002
1 Stanford Stanford Stanford Stanford Columbia Columbia2 Harvard Harvard Harvard Harvard Stanford Harvard3 Columbia Columbia = Columbia Columbia = Harvard Stanford4 Indiana Indiana Essex EUI EUI Ohio State5 = UC Berkeley UC Berkeley EUI UC Berkeley UC Berkeley EUI6 ANU EUI Indiana UCSD Ohio State UCSD7 Essex Houston UC Berkeley Essex UCSD UC Irvine8 Houston Essex UCSD = Indiana Indiana Indiana9 Iowa UCSD Ohio State Ohio State Princeton Princeton
10 UCSD Princeton Yale Yale Yale Yale11 EUI Yale UCLA Princeton Michigan State UC Berkeley12 Princeton UCLA Princeton Michigan State = Chicago Michigan State13 Arizona Ohio State Oxford Birmingham MIT Chicago14 Warwick Warwick Michigan State UC Irvine = UCLA UCLA15 Chicago ANU Vanderbilt UCLA UC Irvine LSE16 Georgia Birmingham American Chicago Essex Essex17 Yale Iowa = UC Davis Vanderbilt Birmingham = Georgetown18 = Oxford Chicago UW Madison = Washington U Vanderbilt MIT19 UW Madison UC Davis Texas A&M UW Madison Johns Hopkins Oxford20 Johns Hopkins = Michigan = Houston Oxford Cambridge = SUNY
Binghamton21 Pittsburgh = Oxford Johns Hopkins UC Davis SUNY = ANU
Binghamton22 = UCLA Arizona Washington U Cambridge Oxford Birmingham23 Ohio State Washington U Chicago Johns Hopkins Georgetown Cambridge24 UC Davis Georgia Birmingham Texas A&M American Florida State25 Birmingham UCol Boulder UC Irvine Bristol = LSE Sheffield26 MIT = UC Irvine ANU Sheffield Florida State = Washington U27 Michigan UW Madison Warwick = Georgetown UW Madison Michigan28 Washington U Michigan State UCol Boulder MIT Texas A&M Johns Hopkins29 UCol Boulder Vanderbilt Georgia ANU Penn State = Texas A&M30 Michigan State Johns Hopkins Michigan Florida State = Emory Emory31 American American Bristol American = ANU UCol Boulder32 Florida State MIT Cambridge = Warwick UC Davis American33 Texas A&M Cambridge Georgetown SUNY Michigan Penn State
Binghamton34 Pennsylvania Texas A&M Arizona GWU Washington U Bristol35 SUNY Stony Glasgow = GWU Houston Sheffield UNC Chapel
Brook Hill36 UC Irvine Leiden Iowa Georgia Bristol GWU37 Strathclyde SUNY Stony LSE LSE GWU Cardiff
Brook38 Cambridge Pennsylvania Sheffield Cal Tech Rice = UW Madison39 Leiden LSE Emory SUNY Stony SUNY Stony Aberystwyth
Brook Brook40 Glasgow Florida State MIT Michigan Aberystwyth Trinity (Dublin)41 = LSE GWU Leiden Rice Trinity (Dublin) = Vanderbilt42 Cal Tech Cal Tech SUNY Stony Penn State = Arizona Cornell
Brook43 UW Milwaukee Pittsburgh Florida State = Emory Georgia Geneva44 Arizona State = South UNC Chapel Iowa Cardiff Illinois
Carolina Hill45 South Carolina Strathclyde Rice Hebrew = Cornell Rice46 Rice Emory New Mexico Pennsylvania UNC Chapel UCL (London)
Hill47 GWU = Georgetown Pennsylvania Arizona Claremont SUNY Stony
Brook48 Maryland UC Riverside Hull Maryland Geneva = UC Davis49 Vanderbilt Sheffield Maryland UCol Boulder Houston Arizona50 Georgetown = Hull Glasgow Claremont = UCol Boulder = Virginia
Note: = means that an institution is in the same position as the institution listed immediately before it.
312 SIMON HIX
tution that wanted to be included in the rankings could be asked to provideaccurate and up-to-date information about the size of their faculty.
(Accepted: 7 May 2004)
About the AuthorSimon Hix, Department of Government, London School of Economics, Houghton Street,London, WC2A 2AE, UK; email: [email protected]; website: http://personal.lse.ac.uk/hix
NotesResearch for this paper was partially funded by The Nuffield Foundation (Grant No: SGS/00889/G).I would like to thank Gabriele Birnberg, Ryan Barrett and Bjorn Hoyland for data collection; NancyBeyers at the Institute for Scientific Information for help with purchase of the Social Science Cita-tion Index data; and Daniele Archibugi, Rodney Barker, Kenneth Benoit, Jeff Checkel, Tom Coupé,Philip Cowley, Pepper Culpepper, Nicole Deitelhoff, Vedran Dzihic, Matthew Gabel, Fabrizio Gilardi,Nils Petter Gleditsch, Robert Goodin, Justin Greenwood, Metin Heper, Karl Magnus Johansson, PeterKurrild-Klitgaard, Iain McLean, Martin Lodge, Peter Mair, Paul Mitchell, Andrew Moravcsik, CasMudde, Michael Munger, Yannis Papadopoulos, Thomas Pluemper, Ben Reilly, Christian Reus-Smit,Gerald Schneider, David Shambaugh, Gunnar Sivertsen, Ulf Sverdrup, Alec Sweet, Jon Tonge, ErikVoeten, Albert Weale and the three Political Studies Review referees for comments on the researchand previous versions of this paper.
1 See ‘The ISI® Database: The Journal Selection Process’: <http://www.isinet.com/isi/hot/essays/selectionofmaterialforcoverage/199701.html>.
2 I considered adding journals of other national political science associations (such as the journalsof the Belgian, Swiss, Austrian, Irish and Japanese associations) and a number of other politicalscience journals (such as Aussenwirtschaft). However, none of these journals met the thresholdof at least 100 citations per year.
3 The adjusted R2 for the model is 0.781.
4 Part of the difference between these scores and the SSCI scores is explained by the fact that myindex is an average impact across several years, whereas the scores I have compared them againstare only for the impact of a journal in 2002.
5 More detailed information about how this was calculated for each university can be obtainedfrom the author.
6 Tables showing the top 400 in each five year period between 1993 and 2002 can be found onmy website: <http://personal.lse.ac.uk/hix>.
ReferencesBallard, M. J. and Mitchell, N. J. (1998) ‘The Good, the Better, and the Best in Political Science’, PS:
Political Science and Politics, 31 (4), 826–8.
Coupé, T. (2003) ‘Revealed Preferences: Worldwide Rankings of Economists and Economics Departments, 1969–2000’, Journal of the European Economic Association, 1 (4),<http://student.ulb.ac.be/~tcoupe/ranking.html>.
Dusansky, R. and Vernon, C. J. (1998) ‘Rankings of U.S. Economics Departments’, Journal of EconomicPerspectives, 12 (1), 157–70.
Garand, J. C. and Graddy, K. L. (1999) ‘Ranking Political Science Departments: Do PublicationsMatter?’, PS: Political Science and Politics, 32 (1), 113–16.
Goodson, L. P., Dillman, B. and Hira, A. (1999) ‘Ranking the Presses: Political Scientists’ Evaluationsof Publisher Quality’, PS: Political Science and Politics, 32 (2), 257–62.
Jackman, R. W. and Siverson, R. M. (1996) ‘Rating the Rating: An Analysis of the National ResearchCouncil’s Appraisal of Political Science Ph.D. Programs’, PS: Political Science and Politics, 29 (2),155–60.
POLITICAL SCIENCE DEPARTMENTS 313
Katz, R. and Eagles, M. (1996) ‘Ranking Political Science Departments: A View from the Lower Half’,PS: Political Science and Politics, 29 (2), 149–54.
Lowry, R. C. and Silver, B. D. (1996) ‘A Rising Tide Lifts All Boats: Political Science Department Reputation and Reputation of the University’, PS: Political Science and Politics, 29 (2), 161–7.
McCormick, J. M. and Rice, T. W. (2001) ‘Graduate Training and Research Productivity in the 1990s:A Look at Who Publishes’, PS: Political Science and Politics, 34 (3), 675–80.
Miller, A. H., Tien, C. and Peebler, A. A. (1996) ‘Department Rankings: An Alternative Approach’, PS:Political Science and Politics, 29 (4), 704–17.
PS: Political Science and Politics (1996a) ‘National Research Council Relative Rankings for Research-Doctorate Programs in Political Science’, PS: Political Science and Politics, 29 (2), 144–7.
PS: Political Science and Politics (1996b) ‘U.S. News and World Report Ranking of Graduate PoliticalScience Departments’, PS: Political Science and Politics, 29 (2), 148.
Rice, T. W., McCormick, J. M. and Bergmann, B. D. (2002) ‘Graduate Training, Current Affiliation andPublishing Books in Political Science’, PS: Political Science and Politics, 35 (4), 751–5.
Schmitter, P. (2002) ‘Seven (Disputable) Theses Concerning the Future of “Transatlanticised” or“Globalised” Political Science’, European Political Science, 1 (2), 23–40.
Scott, L. C. and Mitias, P. M. (1996) ‘Trends in Rankings of Economics Departments in the U.S.: AnUpdate’, Economic Inquiry, 34, 378–400.
Teske, P. (1996) ‘Rankings of Political Science Departments Based on Publications in the APSR, JOP,and AJPS, 1986–1995’, State University of New York, Stony Brook, manuscript.
Welch, S. and Hibbing, J. R. (1983) ‘What Do the New Ratings of Political Science DepartmentsMeasure?’, PS: Political Science and Politics, 16 (3), 532–40.