<<

The academic structures of Boston, London and : a comparison Report prepared for CNRS

CLIENT: Centre National de la Recherche Scientifique (CNRS)

DATE: October 2016

[email protected] - www.sirisacademic.com

2

THE ACADEMIC STRUCTURES OF BOSTON, LONDON AND PARIS: A COMPARISON

October 2016

Submitted to:

Centre Nationnal de la Recherche Scientifique

By SIRIS Academic

Av. Francesc Cambó, 17 08003 Barcelona Spain Tel. +34 93 624 02 28 [email protected] www.sirisacademic.com

For more information and comments about this report, please contact us at [email protected]

3

EXECUTIVE SUMMARY ...... 6 KEY RECOMMENDATIONS ...... 8 LINKS TO THE INTERACTIVE VISUALISATIONS PRESENTED IN THIS REPORT ...... 12 Rankings overview ...... 12 Maps of science ...... 12 Visualisation of ranking’s qualitative vs. quantitative criteria ...... 12 Visualisations of ranking’s qualitative criteria vs. citation impact ...... 12 Visualisations of scientific production and impact ...... 12 INTRODUCTION ...... 13 PART 1. A FIRST GLANCE: THE SIMILARITIES...... 18

OVERVIEW ...... 18 CITY RANKINGS ...... 20 HIGHER EDUCATION ...... 21 AND DEVELOPMENT ...... 21 PART 2. SCIENTIFIC PRODUCTION, RANKINGS AND RECOGNITION ...... 23

SCIENTIFIC PRODUCTION ...... 23 Specialisation ...... 24 Maps of Science ...... 24 RANKINGS ...... 30 Overview ...... 30 Qualitative vs. quantitative criteria ...... 31 Conclusions ...... 36 EFFICIENCY-RELATED CRITERIA ...... 37 PRELIMINARY REMARKS ON THE DISTRIBUTION OF SCIENTIFIC POTENTIAL ...... 42 PART 3. COMPARING THE STRUCTURATION OF THE INSTITUTIONAL LANDSCAPES...... 43

BOSTON AREA ...... 43 Legal categories and missions ...... 43 Carnegie classification ...... 45 LONDON AREA ...... 46 Legal categories and missions ...... 46 Classification ...... 47 PARIS AREA ...... 49 Legal categories and missions ...... 49 Classification ...... 50 COMPARING BOSTON, LONDON AND PARIS IN TERMS OF INSTITUTIONAL STRUCTURATION ...... 50 PRELIMINARY CONCLUSION(S) ...... 56

DO WORLD-CLASS UNIVERSITIES INCREASE SOCIAL STRATIFICATION? ...... 56 DO WORLD-CLASS UNIVERSITIES HAVE A STRONG POSITIVE IMPACT ON THEIR ENVIRONMENT? ...... 59 THE EVOLUTION OF THE PARISIAN LANDSCAPE AND WORLD-CLASS UNIVERSITIES ...... 61 THE CHALLENGES AHEAD ...... 62 BIBLIOGRAPHY (MAIN REFERENCES) ...... 67 APPENDICES...... 69

APPENDIX A: REMARK ON THE PERIMETER OF THE THREE AREAS ...... 69 APPENDIX B: THE CARNEGIE CLASSIFICATION...... 69 APPENDIX C: VIKKY BOLIVERS CLUSTERS OF UK UNIVERSITIES BY VARIABLE ...... 71

4

APPENDIX D: OVERVIEW OF THE MAIN CLASSIFICATIONS ...... 72 APPENDIX E: WEB OF SCIENCE QUERIES USED IN THE REPORT...... 73 Queries by Area ...... 73 APPENDIX F: DATA PROBLEMS: PARIS- ACCORDING TO WEB OF SCIENCE: ...... 78 APPENDIX G: LIST OF INSTITUTIONS PER AREA ...... 79 Paris ...... 79 London ...... 86 Boston ...... 89

5

Executive Summary

The following report compares the institutional landscape of Paris, London and Boston.

The first part of the report outlines the deep similarities in terms of overall scientific productivity of the three cities, whilst highlighting one key difference: Paris has no institution as well ranked as Harvard, MIT, Imperial or UCL.

We argue that this difference is critical for three reasons: - It was used to justify the launch of the Programme Investissements d’avenir and is constantly used to defend ongoing institutional change, both by politicians and by the institutions themselves. - A global Matthews effect is enabling leading institutions to attract ever more resources and gain ever more visibility. - Increasing competition from countries such as China is predominantly affecting institutions who have nearly but not quite broken into the leading group.

We are thus at a tipping point, with a small number of elite global institutions pulling away from the rest and a large number of new institutions aspiring to join the elite1. If no French institution is able to join this elite within the next 5 years, they will never be able to.

The second part of the report compares the scientific production, rankings and recognition of the institutions of the three cities. Whatever the criteria, the order of Boston and London institutions remains the same: Harvard, MIT, UCL and Imperial are followed by the rest and the rest are generally a long way behind. On the contrary, in Paris the order of institutions varies depending on the criteria being compared: the best ranked are either the four main research universities (Paris Descartes, Paris Diderot, Paris Sud and UPMC), or grandes écoles such as École Normale Supérieure or École Polytechnique. They never reach the scores of Harvard, MIT, UCL or Imperial and are usually a long way behind. All the comparisons we tried reflect this general picture, sometimes in quite a striking manner. There is, for example, a strong correlation between the number of publications per staff and the total number of staff amongst Boston institutions, whereas in Paris, the distribution is far more random with a tendency for smaller institutions to publish more per researcher. In Boston, researchers who publish most thus tend to be employed by the institutions with the largest number of research staff (Harvard followed by MIT) whereas in Paris this is not the case. Surprisingly, the leading existing Parisian institutions tend to have better results than those of the Parisian ComUE.

Because the scientific potential is equivalent, the differences are presumably linked to the institutional landscape of the three cities. In the third part of the report we compare the

1 The Nature Rising Stars index lists the 100 institutions that have most increased their number of publications in leading journals. 8 of the top 10 and 40 of the top 100 are Chinese (11 are American, 9 UK, 8 German, 5 Indian, 4 Swiss, 3 Australian and Swedish and 2 each from Canada, (INSERM 75th, UPMC 87th), Italy, Poland and South Korea). The top 10 European rising stars includes both UCL and Imperial (alongside EPSL Lausanne and ETH Zurich) whereas no French university or grande école breaks the leading 25. See: https://www.natureindex.com/supplements/nature-index-2016-rising-stars/index#ni-articles

6 differentiation of these landscapes according to the two core missions of their institutions: teaching and research2. The hierarchy between institutions in terms of research output is clear in all three cities, however, there is a big difference between Boston (where the curve is exponential) and Paris (where UPMC leads the three other main research universities but the level of concentration is lesser than in the two other cities). The main difference occurs in terms of student distribution. The seven universities with most undergraduate students are all situated in Paris (Paris Diderot, Paris Descartes, UPMC, Paris Sud, Panthéon Sorbonne, Paris-Ouest and Paris Dauphine). They are followed by Boston University and King’s College London. Furthermore, the four leading Parisian universities in terms of publications have less graduate students than Boston University, Harvard, UCL or King’s despite Paris concentrating proportionally more graduate students (37%) than both Boston (32%) or London (29%). Institutionalising the existing ComUE (by mergers or other means), without tackling problems of research intensity or distribution of undergraduate students would increase the divergence between the Parisian landscapes and those of London and Boston. From a purely institutional perspective, the Parisian landscape of Higher Education would thus become more similar to that of countries dominated by large, heterogeneous universities such as southern Europe or southern America.

The report ends by debating two important questions: do world-class universities increase social stratification? and do they have a positive effect on their environment? We emphasise the importance of tackling the first question seriously, even if we are not able to answer it conclusively in this report. The answer to the second question, on the other hand, is clearly positive and shows that the reasons given for launching the PIA were correct, even if the institutional transformations have not yet achieved the expected results.

2 Figures 29 for Boston, figure 30 for London and figures 31 and 32 for Paris and Parisian ComUEs compare institutions by number of undergraduate students, number of graduate student and research output.

7

Key recommendations

The reforms launched in France over the last 10 years have transformed the French landscape of HE&R. However, they have not yet created the conditions for World-Class Universities to emerge. Grandes écoles, universités and even ComUE are all institutions of higher education and research, but none can aspire to become either “world-class research universities” or “leading teaching universities” without undergoing considerable further transformation. Seen from or Boston, they remain idiosyncratically French.

The ComUE have played a key role as facilitating structures. They have enabled the process of institutional transformation to start by creating a space for institutional dialogue between research organisations, grandes écoles and universités. There is however a danger that they lose their focus and become institutional goals per se. This would defeat the aim for which they were set up, namely to coordinate the activities of institutions of higher education and research in a given territory. Indeed, by comparing the performances of individual institutions with those of the ComUE of which they are part, the report shows that the institutionalisation of ComUE would actually increase the differences between the landscapes of Paris, London and Boston (notably in terms of student distribution). Furthermore, the resultant institutions would be less competitive than the leading existing institutions on a number of important criteria, such as students per staff, impact per publication or publications per staff.

Viewing the ComUE in this way makes it easier to refocus on the aims of the IDEX and the supposed “expectations” of the international jury. It highlights the fact that the problem is not to “institutionalise the ComUE” but to create institutions that are able to tackle key issues that hold back would-be world-class universities such as selection, funding and governance. The problem is one of mission and vision.

Clarifying and differentiating institutional mission

World-class universities cannot be created in isolation. They are necessarily part of a long-term vision for the university system as a whole. As Birnbaum underlines: “what we really need in countries everywhere are more world-class technical institutes, world-class community colleges, world-class colleges of agriculture, world-class teachers’ colleges, and world-class regional state universities. The United States don’t have a world-class higher education system because it has many world-class universities; instead it has world-class universities because it has a world-class higher education system”. (Birnbaum, 2007).

In Paris, there has been a strong tendency to try and ensure either that no-one gets left behind and/or that all types of cursus are integrated into a single structure. This commendable aim is bound to failure because it is less competitive than comprehensive systems in which different institutions have clearly identified missions: Lasell College focuses on education for undergraduate students, Harvard on producing world-class research.

The Parisian system hosts an incredibly diverse set of institutions from the purely legal perspective but it lacks diversity in terms of institutional missions. It desperately needs institutions to focus on becoming teaching universities, vocational institutions or regional universities. Because without this

8 ecosystem, it will not be able to foster the emergence of 4 or 5 research universities and 2 or 3 world-class universities.

The existing ComUE can help transform the landscape by encouraging internal differentiation between institutions: they should be concentrating their efforts on non-selective undergraduate programmes and the development of true vocational cursus.

Only 2 or 3 Parisian institutions should aspire to be comprehensive world-class research universities. They need to focus exclusively on their core mission: being world-class in research and education means that most members of the academic staff have the potential to be in the top 10% in their field and most undergraduate students have the potential to do a PhD. They need greater autonomy to enable them to choose their mission and select their students and academic staff. Harvard, MIT, Imperial or UCL are part of well-oiled competitive systems of HE&R in which other institutions fulfil the roles of vocational institutions or regional state universities.

Finally, a last striking fact when looking at the landscape of higher education and research of Boston and London is the diversity of great institutions. Alongside Harvard, MIT, UCL and Imperial there is room for excellent research universities such as King’s, Tufts or Boston University. There are remarkable small research institutions such as the London School of Hygiene and Tropical Medicine, Research Hospitals such as MGH, specialised institutions such as London School of Economics. And, of course, there are wonderful vocational institutions and teaching universities that do not feel the need to claim that they are doing cutting-edge research.

Creating coherent world-class institutions in Paris

The Parisian Idex have succeeded in launching remarkable initiatives, however they still need to solve two key problems. They lack the internal coherence to be immediately recognised as universities by external observers. And they lack that “little something”, the “dark matter” that marks the difference between most institutions whose position varies from ranking to ranking and the very few who are always at the top whatever the criteria selected.

Coherence and the institutional debate

Institutional coherence is sometimes considered synonymous with being recognised by ARWU, CWTS Leiden or THE. This approach leads to discussions on legal statuses, lobbying and attempts to define minimal conditions that all take up a large amount of energy and resources. None of these questions solves the underlying problem of institutional coherence.

No two World-Class universities are structured alike. They combine departments, research institutes, governmental laboratories, faculties, student colleges and a vast array of other institutional objects in a myriad of forms. Furthermore, all these institutional objects are themselves subject to an even greater diversity of function and status. This is not a problem.

However, no world-class university contains within it a structure that has prerogatives overlapping with those of the university itself. To give one much cited example, the problem of comparing Cambridge with French ComUE is not one of legal status (Colleges as listed bodies versus University as recognised body versus Schools, Departments and Faculties without legal status) but one of

9 mission: the Colleges of Cambridge provide tutorship to students and are represented on some of the boards of the central university (4 seats on the 25 member university council), the university faculties organise teaching and research and are also represented on the board of the university. A conflict can arise between colleges or between faculties over resource allocation but no conflict can exist between them and the university on their mission.

The merger of Paris-Sorbonne and UPMC will recreate a coherent comprehensive at the heart of Paris. The respective responsibilities of the university, the faculties, laboratories, departments and other internal structures need to be defined but there is no conflict about mission.

The cases of Sorbonne Paris-Cité, PSL or are more complex. Specialised institutions such as Institut d’Optique or Observatoire are not problematic: they are research institutes in a specific field that can easily find their place within the future university, whatever form it takes. On the other hand, comprehensive institutions cannot be members of a comprehensive institution, they can only exist within university systems (like UCL in University of London or UC Berkeley in University of California) or as the heart of the future university.

Density and “dark matter”

The term “dark matter” was coined by (Usher & Savino, 2007), who talk of the gravitational pull that certain institutions (Harvard, Oxford, etc.) or types of institutions exert on all rankings, regardless of the specific indicators and weightings used.

The idea of “dark matter” harks back to a much-maligned concept introduced in the first IDEX calls: the famous périmètre d’excellence, but a périmètre d’excellence whose limits correspond to those of the institution. The leading institutions of Boston and London are dense: they are leaders in all fields, their students are the best students and the percentage of highly cited publications of their researchers is extremely high. To be leaders they must ensure that the level of excellence is homogeneous. They cannot afford to retain laboratories that are not leaders in their field or deliver diplomas to students who are unlikely to be accepted by a Master programme at one of their peer institutions.

Sorbonne University will not immediately acquire this “dark matter”, however the fundamentals are in place and the core institutions are comparable. can succeed with a strong internal strategy of excellence and strong governmental support to enable it to compete on an equal footing with the world’s elite institutions (notably in terms of student attractiveness). The situation of PSL, Saclay or Sorbonne Paris-Cité is, once again, more complex because the teaching and research excellence of the institutions that integrate the project are not homogeneous. They can lead the world but they need to rethink their structure and define conditions modelled on those of the institutions to whose status they aspire.

The Path ahead

There is no reason for which the Parisian system should not host world-class universities: it already hosts some of the most highly selective institutions in the world, as well as some of the best research groups. It not only publishes a comparable quantity of articles to Boston or London but hosts a higher percentage of graduate to undergraduate students than either of these two cities (37%, 32% and 29%). Despite the dichotomy between grandes écoles and universités, despite the

10 confusion of roles between excellence in research and non-selective undergraduate education, Parisian institutions still succeed in appearing in the top 100 in the world. Furthermore, the CNRS, the most prestigious national research organisation in the world, is fully integrated at the heart of all the important research-based institutions and is already playing a key role in terms of differentiation of the Parisian landscape of HE&R.

If the main objective is to consolidate 2 to 3 world-class universities and 4 to 5 research universities, then the path ahead is simple: enable a differentiation in terms of mission, create good teaching colleges, encourage the diversity of the landscape of HE&R and let the flagship universities concentrate on their mission of attracting the best students and scholars and producing world-class research.

11

Links to the interactive visualisations presented in this report

These links are integrated into the document. You can click on visualisations marked by an orange dot to follow the link to the online interactive version.

Rankings overview http://sirislab.com/lab/cnrs/vizs/#!/view1

Maps of science http://sirislab.com/lab/cnrs/map-science/#/

Visualisation of ranking’s qualitative vs. quantitative criteria  ARWU http://sirislab.com/lab/cnrs/rankings/ARWU.html  QS http://sirislab.com/lab/cnrs/rankings/QS.html  THE http://sirislab.com/lab/cnrs/rankings/THE.html

Visualisations of ranking’s qualitative criteria vs. citation impact  ARWU http://sirislab.com/lab/cnrs/rankings/ARWU_cites.html  QS http://sirislab.com/lab/cnrs/rankings/QS_cites.html  THE http://sirislab.com/lab/cnrs/rankings/THE_cites.html

Visualisations of scientific production and impact  Citation impact vs. no. of publications http://sirislab.com/lab/cnrs/impact/citations.html  Publications/staff vs. students/staff http://sirislab.com/lab/cnrs/impact/students.html  Publications/staff vs. staff http://sirislab.com/lab/cnrs/impact/publications.html

12

Introduction

Over the last 15 years two major trends have accelerated the transformation of the landscape of HE&R (higher education and research) worldwide. On the one hand the transition towards a knowledge economy has led to massive growth of the sector and increased scrutiny of its social relevance (link with innovation, value of specific educational programmes, etc.). On the other hand, the globalisation of HE&R has increased competition between institutions, not only on a national level but internationally.

The tipping point is usually associated with the publication of the ARWU (or Shanghai) ranking in 2003, which, for the first time, provided a ranking of the best universities in the world. This reinforced the so-called Matthew effect in HE&R3 whereby the best institutions attract the best scholars and the best students and thus get even better. But, more importantly, it globalised this phenomenon, increasing competition and differentiation between institutions worldwide.

These trends consolidated the “world-class university”4 as a status symbol for all countries and led to the launch of initiatives to create or develop world-class universities in countries such as China, Korea, Germany, Japan, Russia and many more5. The transformation of the French system of HE&R over the last 10 years6 is thus part of a much more global phenomena, which takes on specific forms in each different country.

Contextualisation This report is written shortly after the 4-year review of the 8 first French IDEX projects. It does not aim to analyse the reasons of the failure of some of the projects7 nor to discuss their future but simply to widen the debate by comparing the Parisian ecosystem of HE&R with those of Boston and London.

Of course, the time needed to gain prestige, publish articles and be recognised as a single university can explain in part why the IDEX that were awarded in 2011-2012 have not yet enabled the emergence of any “world-class university”. However, this argument would only be valid if

3 Term coined by Robert K. Merton in his The Sociology of Science: Theoretical and Empirical Investigations, 1973. cf. his more recent reassessment in On Social Structure and Science, 1996. 4 “World-class university” is a fuzzy term – we will describe it in more detail in this report, however for now we take it to be synonymous with a ranking in the top 20 of the main international rankings – whatever their shortcomings. The 20th position in ARWU is the position around which the exponential take-off of the “top” institutions begins: in terms of total score, the distance between the 1st and the 20th is greater than that between the 20th and the 500th. 5 See the list of Recent Research Excellence Initiatives in Appendix F of Salmi, 2009. The challenge of establishing world-class universities. 6 In France, the last ten years have been marked by legal changes increasing the autonomy of universities, the creation of new types of institutions such PRES and ComUE and large scale calls for projects of which the most important, the IDEX, aimed to «faire émerger sur le territoire français 5 à 10 pôles pluridisciplinaires d’excellence d’enseignement supérieur et de recherche de rang mondial». See: www.agence-nationale- recherche.fr/investissementsdavenir/documents/ANR-AAP-IDEX-2010.pdf. 7 Between 2011 and 2012, a first batch of eight projects were selected. When they were evaluated in spring 2016, only 3 projects were definitely confirmed (, Aix- and Strasbourg), none of which is in the top 50 of any international ranking.

13 institutional stability had increased. This is clearly not the case8: currently Parisian institutions are experimenting a wide range of possible institutional models, including mergers of universities (Paris- Sorbonne and UPMC), “integration” of institutions (PSL and Paris-Saclay), loose federations (HéSAM), possible mergers of Grandes écoles with universities (Paris and Paris Est), mergers of Grandes écoles (École Centrale and Supélec). Many other models have been proposed, such as recreating Paris University or separating undergraduate from graduate education. This is causing instability at the level of the system with institutions leaving one alliance to join another and, in some cases, even parts of one institution attempting to leave it in order to join another.

In this situation, the CNRS has a key role to play for two fundamental reasons: 1) As a partner of all the universités, Grande écoles and ComUE in Paris and its region, it benefits from a unique overview of the institutional landscape. And as a transversal institution, it has a great impact on the structuring of the scientific landscape. It is, in many ways, the only HE&R institution that can influence the overall structure of the system of HE&R. 2) The CNRS is by far the largest contributor to the scientific potential of Paris and its region and as such it has a unique responsibility to help shape the future landscape of HE&R: there can be no world-class university in France that does not benefit from substantial support from the CNRS.

This report aims to provide useful elements of benchmark against which the current institutional transformations of Paris can be compared.

Justification of the method Individual universities do not exist in isolation. They form part of wider HE&R ecosystems and can only be understood as components of these wider HE&R ecosystems. This is true both in terms of concentration of excellence and in terms of institutional diversification.

This is why nearly all leading universities are situated in global hubs such as Boston, Singapore or the golden triangle between Cambridge, London and Oxford9. These hubs concentrate both a large number of leading universities, including nearly all the top 20 of international rankings, and a large number of other institutions fulfilling a diversity of roles and notably that of providing quality undergraduate teaching without undertaking research10. This situation explains the methodological choice of this report: comparing regional HE&R systems, by opposition with isolated institutions, or national situations.

In France, the only global hub is situated in Paris. Although Aix-Marseille, Bordeaux and Strasbourg are great universities, they are unlikely to ever acquire a status comparable to that of Harvard or UCL: the concentration of research in Paris is such that only Paris has the potential to enable universities of this standard to emerge. Indeed, the 4 IDEX projects that could have theoretically entered the top 50 of the ARWU ranking were all situated in Paris11.

8 This instability can be measured in the frequency of AEF reports since spring 2016 announcing major institutional shakeups, it concerns all Parisian institutions. 9 Smaglik, 2005. “Golden Opportunities”, Nature 436 10 There are, for example, 54 institutions of HE&R in Boston alone. 11 (PSL (27th), Paris-Saclay (28th), Sorbonne University (35th) and Sorbonne Paris-Cité -56th) according to the simulations in the ARWU ranking undertaken by Docampo, Egret and Cram, 2014. French COMUEs and the Shanghai Ranking. Technical report. Three of them were given an extra probationary period (Paris-Saclay, Paris

14

We chose to compare Paris with Boston and London because they are all three well-positioned on the “knowledge economy map” with:  roughly similar population within the overall metropolitan area,  similar levels R&D investment,  comparable numbers of students and scientific publications,

Despite the similarity between Boston, London and Paris in terms of total scientific potential, there is one major difference: Paris has no institution as well ranked as Harvard, MIT, Imperial or UCL.

Figure 1 Number of universities ranked per city in the three main rankings ●http://sirislab.com/lab/cnrs/vizs/#!/view1

Sciences et Lettres, Sorbonne Universités) and one was definitely discontinued (Sorbonne Paris-Cité) by the International Jury.

15

The absence of world-class universities was the critical issue used to justify the launch of the programme investissements d’avenir12 and is constantly used to defend ongoing institutional change, both by politicians and by the institutions themselves.

Our starting hypothesis is that this absence is due, at least in part, to a Important remark different institutional structuration of This report relies, as far as possible, on available data. the HE&R sector, based on different Sometimes, data is not available, or the quality is uncertain. choices regarding basic decisions such Two issues need to be highlighted in this respect: as the selectivity of the institutions, the 1. A city can be defined in many ways: in an administrative sense as the municipality or the region, as a scope of their missions, their autonomy metropolitan area, as an , etc. The definition of or again student fees. each of these entities changes from country to country as does data availability even for data as simple as population. We will therefore look at how the Similarly, the list of institutions to include is not necessarily institutional structure of the Parisian clear: to cite two examples, Compiègne is not in any of the system, with its universities, Grandes previously defined areas but the Université de Technologie écoles and research organisations, de Compiègne is part of the ComUE Sorbonne Universités. supports or impedes the emergence of Brown University is in Providence, Rhode Island, but often world-class universities and if larger considered part of the Combined Boston Statistical Area. groupings like the ComUEs are better 2. The bibliometric data are not fully satisfactory. Some issues deal with disciplinary field representation, such suited to this aim. as the limits of data in Social Sciences and , others with the capacity at correctly identifying the scientific Methodological issues production of a given institution. For example, many Attempting to identify the structural publications in fields such as astrophysics or medicine are reasons for which Paris does not host mistakenly attributed by Web of Science to Université Paris- an institution equivalent to Harvard or Sorbonne, which is specialised in humanities. Imperial is obviously very ambitious, That said, we are confident that problems with the data do and bound to be controversial. The not affect the key interpretations made in the report. weight of context and limits our capacity to simply compare the performances of different systems. Furthermore, our starting hypothesis relies on assumptions and interpretations, which are difficult to put to empirical test. Therefore, the present report adopts a modest stance: it aims at identifying available data that supports a, necessarily partial, reading of the assets and weaknesses of the Parisian system of HE&R.

In order to do that, a series of methodological choices have been made: 1. Proceed through a benchmarking approach, which focuses on a sub-national scale. We made the choice of comparing big metropolitan areas, because they are the ones where the tension between the missions of ensuring world-class research and a good basic undergraduate university education are clearest. 2. Distinguish the gathering of data about “inputs” (number of students, researchers, money, etc.) and “outputs” (number of graduates, publications, rankings, etc.) of the system, from

12 As A. Juppé and M. Rocard wrote in their report Investir pour l’avenir (p.27) : « Pour critiquables qu’ils soient, les classements et indicateurs internationaux font état de prestations médiocres: le classement de Shanghai ne place que trois universités françaises dans les cent premières (dont la première à la 40e place seulement en 2009), tandis que le classement du Times Higher Education Supplement considère que seuls quatre établissements français figurent parmi les deux cents meilleurs mondiaux. ».

16

the interpretation of the role of the institutional structure, to ensure that the reader can clearly identify when we move from facts to interpretation and discussion.

In terms of data, we relied on bibliometric data from the Web of Science, demographic data from public sources in the US, UK and France, HE&R-specific statistical data, data extracted from HE&R rankings and specific benchmarks at the level of individual institutions.

Structure of the report The first part of the report presents the three metropolitan systems taken as a whole, in order to establish a basic idea of their affinities. It relies on data from ministries, national statistic agencies, international bodies, as well as on bibliometric data from the Web of Science. It shows how the three metropolitan areas compare in terms of scientific production, as well as in terms of higher education.

The second part presents an analysis of the scientific production and the performance in rankings of Paris, London and Boston. We assess, first, if the three areas are comparable in terms of disciplinary coverage or if we can identify reasons for the head start that London and Boston have in the rankings. We then analyse in detail the performance of the three regions in terms of rankings with the aim of understanding if Paris is disadvantaged by its institutional structure.

The third part of the report then focuses on the institutional organisation of each of the 3 areas. This more qualitative and interpretative part pinpoints the extent to which we can correlate the institutional structure with the systemic performance.

Finally, the conclusion discusses the Interactive visualisations relevance of a “world-class university” A series of interactive representations have been approach by showing the impact of this developed for this report. They are available online by type of institution on the system as a whole clicking on the images marked by an orange dot. ● The and by asking if concentrating on excellence full list of links is also available at the front of the report. necessarily results in less social fairness.13

We emphasise that, in order to understand and transform an individual institution, it is necessary to take a comprehensive view of the missions of the HE&R system as a whole, from the production of world-class research and selective academic education, to guaranteeing the best possible level of education for the population as a whole.

13 We were not able to develop this section in as much detail as we would have liked both because of time constraints and the difficulty in accessing relevant data.

17

Part 1. A first glance: the similarities

As explained above, the decision to compare Paris with London and Boston was motivated by the following considerations: 1) Paris, London and Boston are at the centre of metropolitan areas of comparable population and economic activity. Therefore, it is reasonable to assume that their respective HE&R-systems need to provide comparable services. 2) The academic landscape of these cities is surprisingly diverse. Paris and Boston have a large variety of institutions: Grandes Écoles, universités and national research organisations in Paris, and, in Boston, community and state colleges, vocational schools and public as well as private universities. At first glance, London seems much less diverse in this respect because, since the disappearance of the polytechnics in 1992, universities are virtually the only kind of HE&R institution. However, when looking more precisely at their functioning, the British universities are actually highly differentiated across a spectre ranging from the Russell group to the former polytechnics (post-1992 universities). 3) Universities which aspire to a “world-class” status typically benchmark themselves with universities such as Harvard or UCL14 and the same is true for whole HE&R systems, which often use the UK and US systems as references.

Overview

In this part of the report we review briefly the demography, economic activity and HE&R sector of the three cities, to justify the relevance of comparing their HE&R systems. The following table summarises the main indicators for the three regions and shows their similarity in terms of population and GDP per capita (the difference in terms of R&D funding is due to different perimeters being measured as explained later):

Paris London Boston Geographical perimeter City population 2 254 262 8,673,713 667,137 Paris, Greater London and Boston

Greater city population 12,005,077 9,787,426 4,732,161 Ile de France, Greater London Built-up Area, Greater Boston Large metropolitan area 12 405 426 14,031,830 8,041,303 Aire urbaine, London Metro Area, population Boston-Worcester-Providence Combined Area Greater city GDP 54,069 52,047 60,902 Ile de France, Greater London, /capita (€) Greater Boston Total R&D (M€) 18,664 4,429 18,774 Ile de France, Greater London, Massachusetts

Figure 2: Key city data (source: Eurostat, INSEE, UK Office for National Statistics, United States Office of Management and Budget, National Science Foundation. Latest year available).

The Brookings institution curates a “Global metro monitor”15, with some diverging perimeters and figures to the ones we use as a basis, but which is also useful in comparing the three regions at hand:

14 As this report will show, the differences between London and Boston are such that the notion of an Anglo- Saxon model makes little sense. 15 www.brookings.edu/research/reports2/2015/01/22-global-metro-monitor

18

Paris London Boston

Figure 3: Brookings’ global metro monitor for Paris, London and Boston16

16 The full report is available here: www.brookings.edu/~/media/Research/Files/Reports/2015/01/22-global-metro-monitor/bmpp_GMM_final.pdf?la=en

19

There is thus a close correspondence between the basic demographic and economic data especially of Paris and London. The GDP per capita, the employment and the distribution of industrial activity is also close, although the GDP growth of Paris is currently lagging behind. Boston presents only minor differences such as being less affected by the 2008 crisis in terms of GDP growth. The considerably higher GDP per capita is linked in part to the fact that the area definition chosen by Brookings considers a smaller, hence more urbanised portion of the Boston Combined Statistical Area. In terms of economic activity, the three metropoles are very similar with all 3 being strong in the Business/Finance, Local/Non-market and Trade and Tourism fields (unlike other cities surveyed).

What is relevant in this case is that the overall similarity makes it reasonable to suppose that the need for education and research and the capacity to provide for such needs, should be comparable.

City rankings

The following table gathers a selection of global city rankings to facilitate the comparison between the three cities. Rank

Paris London Boston

Global Cities Index - A.T. Kearney’s 3 2 23

Global Cities Outlook (future potential) - A.T. Kearney’s 19 3 2

2025 City Competitiveness rankings table - The Economist 7 2 19

Total Rank - Global Power City Index 2015 - Mori Foundation 3 1 23

Economy Score - Global Power City Index 2015 - Mori Foundation 218 324 191 Research and development Score - Global Power City Index 2015 - Mori Foundation 124 124 122

Figure 4: City rankings17

Unsurprisingly, London and Paris are consistently considered top-class global cities, with London outranking Paris on most accounts. More surprisingly, Boston is generally excellently placed considering its smaller size (92nd metropolitan area in terms of population vs. 24th and 16th), especially when considering future and innovation potential.

The Mori Memorial Foundation’s Global Power City Index, which is a well curated comprehensive global city analysis, reinforces the previous analysis: London leads as a global economic hub, followed by Paris and then, closely, Boston.

Particularly interesting in the table is the R&D score attributed by the Mori foundation which are almost identical. This compounded score takes into account the following criteria: - number of researchers; - world’s top 200 universities;

17 Source: A.T. Kearny’s: www.atkearney.com/research-studies/global-cities-index/2015 ; The Economist: www.citigroup.com/citi/citiforcities/pdfs/hotspots2025.pdf; The Mori Memorial Foundation: www.mori-m- foundation.or.jp/english/ius2/gpci2/

20

- academic performance in mathematics and science; - readiness for accepting foreign researchers; - research and development expenditure; - number of registered industrial property rights; - number of winners of highly-reputed prizes; - interaction opportunities between researchers18.

The following sections provide further information and statistics on the HE&R sector of the three metropolitan areas.

Higher Education

Paris, London and Boston have similar number of students although in this case the leadership of Paris in quantitative terms is clear.

Paris London Boston Total tertiary students 463,363 378,930 372,768 As a % of tertiary students in the country 19.82% 15.88% 1.85%

Total undergraduate students 292,051 260,496 263,433 Total graduate students 171,312 122,025 109,335

Proportion of undergraduate of students 63.03% 68.75% 70.67% Proportion of graduate of students 36.97% 32.20% 29.33%

Figure 5: Overview of key Higher Education Numbers (based on raw data by the Ministère de l’Education, INSEE, HEFCE, Carnegie Foundation, National Center for Education Statistics. Latest data available)

Paris hosts the largest number of tertiary students, and also concentrates the most students in its country. It is particularly interesting to note that the national concentration of students is intensified at the graduate level in Paris. The city thus hosts, proportionally more graduate students than London or Boston (37%, 32% and 29%).

Research and development

Another key indicator to consider is R&D expenditure. The following table summarizes the available data for the Paris, London and Boston areas19.

18 See http://www.mori-m-foundation.or.jp/pdf/GPCI2015_en.pdf, p. 8. 19 It should be noted that these figures present where (in which type of institution) the expenditure takes place, not which is the source of funding.

21

London + Paris East/South Boston (Ile de France) Greater London East England (Massachusetts) Total (M€) 18,664 4,429 17,908 18,774 Private (M€) 12,767 1,550 11,282 13,609 Academic (M€) 3,159 2,310 4,481 2,526 Government (M€) 2,452 394 1,641 no data

Private non-profit (M€) 286 174 504 no data

Figure 6: R&D Expenditure. Source: Eurostat 2013, NSF 2012&2014

The absolute data in these tables is difficult to compare because the numbers correspond to different areas: In Paris, an important part of the R&D activity is performed in science and business campuses outside the city, which are largely accounted for as the region considered is Île-de-France. In the case of Boston, the reference perimeter is the state of Massachusetts, which includes areas that are clearly outside the area of Boston but those not include the neighbouring parts of New Hampshire and Rhode Island that belong to the larger metropolitan area. The figure for London is low, largely because it is calculated for the most restrictive perimeter, Greater London, which is purely urban and includes just above half of its metropolitan population. We have therefore included the neighbouring regions of "South East England" (6,220M€) and "East England" (7,260M€). The sum of R&D expenditure in these three regions is 17,908M€, similar to Paris and Boston. Nevertheless, these regions contain the world-class science and technology ecosystems of Cambridge and Oxford, which are often considered separately20.

In Paris, the activity of the governmental sector (following OECD’s Frascati Manual, this item includes all national research organisations) is very important21, almost as high as the expenditure of the universities. This is not the case in London, where most expenditure of the academic sector happens at the universities. For Boston, we don’t have detailed data, but we can say, that 2,639M€ are unaccounted for after subtraction of the academic and private sector’s expenditure. The role of the government sector (state and federal) is presumably thus also fairly important, but a clear assessment would require further investigation.

In overall terms, the three regions are therefore comparable, and the potential of Paris public research institutions does not seem hindered by the overall R&D landscape.

20 As we will see later in this report the leading universities of London, Cambridge and Oxford have recently started highlighted the “golden triangle” as a leading HE&R hub. 21 We follow the definition of the sector proposed by the OECD, which includes all activity by national organisms (CNRS, CEA, INRIA, INSERM, etc). Frascati Manual 2002, Proposed Standard Practice for Surveys on research and Experimental development, www.oecd-ilibrary.org/docserver/download/9202081e.pdf? expires=1468574722 &id=id&accname=guest&checksum=59FDF3FCCA08BFB82FA80A7A64D55CAB

22

Part 2. Scientific production, rankings and recognition

This first analysis of the data has highlighted the similarities between Boston, London and Paris in terms of HE&R. It means that the absence of a top 20 university in Paris is presumably not linked to the overall potential of the city or the number of tertiary students it hosts. We will now look in more detail at how scientific production, impact and recognition are distributed in the three cities.

Scientific production

To assess the scientific production of the three metropolitan areas considered in this report, we queried the Web of Science (WoS) bibliometric databases for the years 2010-2016. Since there is no direct way to query for this, we propose below two indirect methods. The first method consists in establishing a list of HE&R institutions for the area in question and then looking for the publications attributed to those institutions. For this we used the WoS interface InCites. The second method consists in querying the database on the address line, looking for all the relevant municipalities. Neither of these methods is fully satisfactory: the first method probably underestimates the production, in particular because it is likely to miss out on publications of national research organisations such as CEA, while the second possibly overestimates because it includes publications from all organisations, including those not primarily involved in HE&R. Further problems include those of wrong affiliations (see appendix F) and double counting of publications.

Citation Impact Publications (normalised by field) Institutions London 256,974 1.62 33 Paris 159,957 1.47 39 Boston 294,694 1.94 17

Figure 7: Method 1: scientific production of academic institutions (source InCites)

Publications London 297,677 Paris 265,445 Boston 320,865

Figure 8: Method 2: scientific production by address (source WoS)

There is clearly a distance between Paris on the one hand, London and Boston on the other. In particular, by looking at the first table, which takes institutions as the starting point, Paris appears to have a lower productivity distributed over a larger number of institutions22. The citation impact (normalised by field) is also lower than that of London and especially Boston23.

22 Only institutions that publish are counted. The “total” number of institutions of HE&R is much higher, in particular in Boston. We will return to this problem of research concentration in part 3 of the report. 23 The reasons for this are complex and need studying in detail. They are probably due in part to differences in the map of science of the three cities (citation patterns tend to favour Health Sciences and as we will see, these are more developed in Boston and London), but also to in built biases in WoS and the Matthew effect, which

23

Specialisation A closer look at the production shows that its distribution in terms of scientific fields is remarkably different: Paris is particularly strong in terms of formal sciences (mathematics and physics), but comparatively weak in health sciences and social sciences.

Figure 9: Production by scientific areas. Source: WoS

Maps of Science This overall picture can be refined by looking at variations in publication practices on a field by field basis. To do this, we produced a Map of Science, where each dot corresponds to a WoS category. The size of the dot is proportional to the number of publications and thus makes it possible to identify areas of relative strength24.

On these graphs: - disciplinary categories are represented by a color code (see legend); - the size of dots is proportional to the number of publications in each field; - the spatial distribution reflects the frequency of citations between categories (closer dots mean denser networks).

The representation applies a threshold principle: if a field is not represented or is below the significant threshold, it does not appear on the graph.

combine to favour English language publications by Anglo-Saxon Journals from highly recognised institutions. Defining the degree of bias and possible counter measures that could be taken is a question for another report. 24 The spatial distribution is generated mechanically by an algorithm elaborated by Rafols, Porter and Leydesdorff. 2010 “Science overlay maps : a new tool for research policy and library management”, Journal of the American Society for Information Science & Technology 61 (9). The interactive version presented here was built as part of a collaboration between SIRIS Lab and I. Rafols (Ingenio, Polytechnic Institute of Valencia).

24

These maps are best explored interactively. They are accessible ● here (http://sirislab.com/lab/cnrs/map-science/#/) on the SIRISLab site. They present the different research profiles of each of the cities, according to the 251 different WoS categories. The table simply regroups the top 20 for each city.

The maps are followed by two tables. The first simply lists the top 20 WoS categories for each city with the number of publications and the percentage of publications that each category represents. The second regroups these categories into the 10 large disciplinary fields identified by the OST (Observatoire des Sciences et Techniques).

Figure 10: Map of Science: Paris (source: WoS, visualisation by SIRISLab)

25

Figure 11: Map of Science: London (source: WoS, visualisation by SIRISLab)

Figure 12: Map of Science: Boston (source: WoS, visualisation by SIRISLab)

26

Web of Science Categories Paris London Boston records % of total records % of total records % of total ONCOLOGY 13300 5,010 13220 4,441 21823 6,801 MEDICINE GENERAL INTERNAL 13166 4,423 10569 3,294

ASTRONOMY ASTROPHYSICS 11738 4,422 9809 3,057

NEUROSCIENCES 7462 2,811 12950 4,350 13263 4,134 BIOCHEMISTRY MOLECULAR BIOLOGY 8570 3,229 7723 2,594 13706 4,272 MULTIDISCIPLINARY SCIENCES 7689 2,897 9518 3,197 13696 4,268 ENGINEERING ELECTRICAL ELECTRONIC 11125 4,191 5913 1,986 9124 2,844 CARDIAC CARDIOVASCULAR SYSTEMS 9086 3,052 12619 3,933

CLINICAL NEUROLOGY 6107 2,301 11554 3,881 11706 3,648 PSYCHIATRY 11315 3,801 8163 2,544

SURGERY 5464 2,058 10381 3,487 10909 3,400 CELL BIOLOGY 5561 2,095 5873 1,973 12067 3,761 IMMUNOLOGY 6597 2,485 9316 3,130 9028 2,814 MATERIALS SCIENCE MULTIDISCIPLINARY 8287 3,122 PHYSICS APPLIED 8215 3,095 PUBLIC ENVIRONMENTAL OCCUPATIONAL HEALTH 9100 3,057 10097 3,147

INFECTIOUS DISEASES 8500 2,855

CHEMISTRY PHYSICAL 7293 2,747 HEMATOLOGY 6773 2,552 7399 2,486 8528 2,658 CHEMISTRY MULTIDISCIPLINARY 8366 2,607

RADIOLOGY NUCLEAR MEDICINE MEDICAL IMAGING 8235 2,566

PHARMACOLOGY PHARMACY 6398 2,410 7563 2,541 7931 2,472 ECONOMICS 6756 2,270 7347 2,290

MEDICINE RESEARCH EXPERIMENTAL 7280 2,269

PHYSICS MULTIDISCIPLINARY 5912 2,227

PHYSICS PARTICLES FIELDS 5722 2,156

OPTICS 5707 2,150

GENETICS HEREDITY 5480 2,064 6392 2,147 MATHEMATICS 5468 2,060

GASTROENTEROLOGY HEPATOLOGY 5942 1,996

RESPIRATORY SYSTEM 5787 1,944

Figure 13: Top 20 WoS categories in number of publications per city (source: WoS)

27

SCIENCES SOCIALES

SCIENCES POUR L'INGENIEUR

SCIENCES HUMAINES

SCIENCES DE L'UNIVERS

RECHERCHE MEDICALE

PHYSIQUE

NON ATTRIBUE

MATHEMATIQUES

CHIMIE

BIOLOGIE FONDAMENTALE

BIOLOGIE APPLIQUEE - ECOLOGIE

0 10 20 30 40 50 60 70 80

Boston London Paris

Figure 14: Respective weight of publications in the OST’s large disciplinary fields categories (source: WoS, Observatoire des Sciences et Techniques)

Preliminary interpretations25 The strength of Boston and London in health sciences and medicine is remarkable, with a significant difference compared to Paris in terms of publication numbers. This favours them in bibliometric terms where medical research has a disproportional weight26.

On the other hand, Parisian scientific production is distinguished by its focus on formal sciences, a field where the number of recorded publications is over double that of London.

This difference in global focus is reflected on a field by field basis. The top 10 fields in terms of scientific production of both London and Boston are present in the top 20 of the other city, but 4 of these fields (Medecine General Internal, Psychiatry, Public Environmental Occupational Health and Cardiac Cardiovascular diseases) are not in the top 20 fields in Paris.

25 More detailed interpretations would require comparing WoS results with those that would be obtained using Scopus, looking at the evolution of the patterns over time, and, of course, revising the quality of data curation. Two examples of potential areas of study that would be useful for policy making could be (1) modelling the emergence of world-class clusters (for example the potential of environmental studies in Paris) ; (2) studying the impact of consolidated clusters on big topic research (climate change, etc.) within that same city (here the approach could be both based on a study of funded research projects and co-publication networks). 26 A good example of the impact of this situation is the way Nature became the top journal for experimental physicists. As R. Werner explains: “Life scientists are more numerous and use more citations than physicists, so the impact factors of Science and Nature, which cover all sciences, easily beat that of any non-review physics journal” (Werner 2015, “The focus on bibliometrics makes papers less useful”, Nature 517).

28

On the contrary, 3 of the top 10 fields in Paris and 7 of the top 20 are present in the top 20 of neither Boston, nor London (Material Sciences Multidisciplinary, Physics Applied, Chemistry Physical, Physics Multidisciplinary, Physics Particles Fields, Optics and Mathematics).

The results in should not be over-interpreted, in part because there is a bias in favour of English-speaking journals, but especially because WoS does not reference a sufficient number of significant journals in this field27. This said, it would probably make sense to analyse in more detail why London’s institutions manage to publish so much more, in WoS indexed journals, in these fields than those in Paris, even where France has traditionally had a strong reputation such as Political Sciences (769 publications for Paris versus 2959 for London) or International Relations (336 publications for Paris versus 2073 for London). The weight of Economics in both Boston and London is also remarkable.

Finally, Environmental Sciences, despite their small relative size, are worth noting and include fields such as as Geosciences where Paris publishes more than Boston and London combined (4586 versus 1593 and 2589).

It is interesting to note that Paris and London have a greater overall degree of research heterogeneity than Boston28. However, it is clear that the overall scientific profile of Boston and London are more similar than either of them are to Paris.29 This corresponds, unsurprisingly, to what is known about investment in science in the different fields and also highlights the strategic choices made in the UK and the US in favor of Health Sciences shown for example in terms of budget (in the USA, the proportion of the research budget dedicated to biomedical research is significantly higher than what it is in France30) but also with the opening of the Francis Crick Institute in London31.

The analysis of the HE&R of the three cities shows that their scientific potential is comparable and that the differences are not as important as the overall similarities between them. As we will see in the following chapter, the institutional distribution of this potential is however remarkably different, in particular when we compare leading institutions.

27 The reasons for this are well explained by H. Nowotny, the former director of the ERC, in an article that concerns the social sciences as a whole but is particular relevant to the situation in France: “The social sciences and particularly the humanities, by contrast, inhabit a landscape that is still overwhelmingly rural in character. In order to avoid living too close to their neighbour scholars can easily migrate and settle in an adjacent valley. It is therefore easy to avoid competition and to cut down communication. The rural lifestyle nourishes a different kind of individualism compared to the urban life style. It can easily become complacent and it is clear that the system is paying a high price for it – that of fragmentation and slow rates of growth.” (Nowotny, 2005 “Humanities in European Research”, IWM-Post 89). The fragmentation of the field makes it very difficult for WoS to identify correctly a limited number of key journals. 28 We can show this by calculating the Shannon entropy for the 3 cities based on the distribution of their publications by categories of the Web of Science. 29 This can be shown by calculating the Kullback-Leibler divergence between the distribution of the WoS categories in the 3 cities. This divergence measures the similarity of the research pattern in the three cities, as captured by the WoS. 30 See D. Sarewitz: “The most obvious indicator of this diversity is biomedical science, which in the United States commands almost 50% of the total federal nondefense R&D budget, compared to 4% in Japan and Germany, 6% in France, and 20% in the United Kingdom.” Sarewitz, D. 2007. “Does Science Policy Matter?”, Issues in Science and Technology 23 (4). 31 Created in 2016 as a joint venture of King’s College, Imperial College, University College London and several UK health funding agencies, with the aim to give visibility to that field in London.

29

Rankings32

Overview The following table, that we have already seen above, depicts the distribution of the ranked universities in the three regions. An interactive and more complete dashboard for these visualisations is available (http://sirislab.com/lab/cnrs/vizs/#!/view1) on the SIRISLab website.

Figure 15: Number of Universities ranked in each city by ARWU, QS and THE (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/vizs/#!/view1

32 This report does not discuss the numerous methodological issues of rankings. These issues are important and are described in detail in many studies (including the SIRIS report: University rankings: an introduction). However, we consider that they do not change our key conclusions.

30

Figure 16: overview of the ranking tool (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/vizs/#!/view1

The overall picture that emerges from the ARWU ranking corresponds relatively well with that of the respective scientific production of the 3 cities with 7 institutions ranked in Boston and Paris and 8 in London.

Surprisingly, in QS and THE rankings, the total number of ranked institutions is much higher in Paris and London than in Boston (QS 15, 18 and 8; THE 11, 18 and 7). The presence of a large number of British institutions may well be due, at least in part, to the fact that the methodology on which both QS and THE are based was initially defined by a British team who had closer contacts with British universities than those in other parts of the world. However, the low number of institutions in Boston has a different explanation: the landscape of HE&R in Boston is much more hierarchical with institutions having far more precise missions. As we will see later, this means that teaching universities have no research, no policy in terms of international visibility and do not aspire to enter international rankings: they attract a different type of student, use different criteria to measure their success and compare themselves using national based rankings for their specific type of institution.

In terms of public perception, however, what counts is less the total number of ranked institutions, and more the number of institutions ranked in the top 20. Here, the difference is clear: whereas both Boston and London are present, no Parisian institution manages to appear in any of the three rankings.

In the following sections, we will present a series of visualisations that compare the different criteria used by the rankings to measure reputation, impact and effectivity. These visualisations are too complex to be adequately presented in the format of a report. The complete interactive versions are available on-line, accessible by clicking on the visualisations marked by an orange dot, ● and these interactive versions of the visualisations are the authoritative ones.

Qualitative vs. quantitative criteria The first exercise consists in comparing qualitative versus quantitative criteria in the three rankings: 1. Quantitative criteria are clear in all three rankings (mainly: bibliometric data on research production and research impact) 2. Qualitative criteria are also clear in QS (Academic and Employer Reputation). In THE we used the categories “Research” and “Teaching”, which are composite criteria but for which the main weight is given to reputational surveys. For ARWU, we used Alumni and Award that measure the number of Nobel Prizes and Field Medals to provide a proxy for long term reputation that we could represent opposite the more immediate bibliometric data (number of papers in Nature and Science, etc.)

31

The following graphs present in abscissa the total score of institutions with respect to quantitative criteria, and in ordinate their total score with respect to qualitative criteria.

Figure 17: “ Qualitative” noble prize data vs “quantitative” bibliometric data in ARWU (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/rankings/ARWU.html

There is a clear correlation between purely quantitative bibliometric data and the number of Nobel Prizes and Field Medals awarded to Academic Staff and former students. At the top, Harvard is an outlier, followed at a fair distance by MIT (and a number of other mainly American institutions). UCL and Imperial appear at a considerable distance, closely followed by UPMC.

It is interesting to note that 4 Parisian institutions have a disproportionate number of prizes compared to their current scientific production: Paris Sud, ENS and (with very low scores in N&S + Pub) Paris-Dauphine and Mines ParisTech. In this sense, they are similar to Cambridge, University of and Princeton University all 3 of which are, however, far stronger in quantitative terms. This is not particularly reassuring in terms of the future evolution of their position because Awards progressively loose value as time passes and, unlike Cambridge, University of Chicago and Princeton University, these three institutions publish far less than their main competitors.

32

Figure 18: “Qualitative” reputational survey data vs “quantitative” bibliometric data in QS (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/rankings/QS.html

33

Figure 19: “Qualitative” reputational survey data vs “quantitative” bibliometric data in QS, including all institutions in the country (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/rankings/QS.html

The pattern in QS is similar in general terms for institutions from Boston and London but this time Harvard, MIT, Imperial and UCL all cluster tightly together in the top right hand corner. However, unlike in ARWU, the 2 French institutions closest to them are ENS and Polytechnique (with UPMC a little further behind). These are situated at a similar level to LSE, Boston University and Brown University.

Once again, surprisingly, there is a Parisian cluster on the left. This time, however, the two members have no Nobel Prizes: they rely on a mistaken identity. Paris 1 Panthéon-Sorbonne and Paris 4 Paris- Sorbonne profit massively from, what we assume, is a double mistake. On the one hand, people answer “Sorbonne” when asked which institution is most prestigious in their field, on the other hand, QS attributes “Sorbonne” to Paris 1 and Paris 4.

34

Figure 20: “Qualitative” data vs “quantitative” bibliometric data in THE (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/rankings/THE.html

35

Figure 21: “Qualitative” data vs “quantitative” bibliometric data in THE, including all institutions in the country (visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/rankings/THE.html

The data from THE33 is closer to QS than to ARWU, which is logical because like QS, THE relies on reputational surveys. Once again, Harvard, MIT, UCL and Imperial cluster at the top. In this case, however, they benefit from a clear phase transition, where above a certain point, reputational increase is exponential (and benefits a large cluster of British and American institutions). The 2 top ranked French institutions, ENS and École Polytechnique are just below this point (again with UPMC and bit further behind), closely clustered with institutions such as Brown University or Boston University.

Conclusions It is particularly interesting to note that whereas the same institutions lead all rankings in Boston and London, in Paris they change: UPMC and Paris Sud appear in ARWU (with surprisingly different profiles – UPMC, thanks mainly to bibliometric indicators, Paris Sud to awards and alumni), ENS and École Polytechnique appear in QS and THE, thanks to their reputation and their results on indicators such as staff/student). In both cases, they fail to make it up to the top group of institutions.

33 The data from the Times Higher Education ranking should be interpreted with care. Ideally, we would have liked to use THE reputational surveys for “qualitative” data, however THE embeds this data within the research and teaching criteria, alongside more quantitative data. We decided to use these two criteria as proxies for qualitative data because in both cases the reputational surveys weigh far more than the other criteria. www.timeshighereducation.com/news/ranking-methodology-2016

36

As we will see throughout this report, one probable reason for this failure is that neither universités, nor Grandes écoles are able become top 20 universities for structural reasons, even with the support of national research organisations. The problem is that ComUEs will not be able to either, as we will show in the next set of visualisations. One last point needs to be made from these visualisations: whereas Boston and London appear within a cloud of American and British institutions, Parisian institutions are far more isolated: there are few other French institutions and all of them are ranked well below those from Paris. This once again underlines the importance of the Parisian ecosystem of HE&R for the country as a whole.

Efficiency-related criteria

The following visualisations explore these results in more detail by moving away from the rankings and exploring the results of Boston, London and Paris institutions in different aspects: in terms of total number of publications versus citation impact, in terms of student/staff versus paper/staff and in terms of staff/publication per staff34.

Figure 22: citation impact vs number of publications (data WoS, 2010-2016, visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/impact/citations.html

34 These visualisations also take into account small research centres that are not ranked because they do not fulfil the ranking agencies criteria as universities.

37

Leaving aside institutions with under 100 publications such as the Central School of Speech and Drama or the École Nationale des Chartes, there is a clear correlation between the total number of publications and the normalised citation impact of an institution.

In the case of Boston, the correlation is almost absolute along a line that stretches from Harvard in the top right corner (180651 publications, normalised citation rate 2,19) down to Plymouth State University at the bottom left corner (239 publications, normalised citation rate 0,65).

The distribution of clusters is interesting: - Group A: Harvard University (a cluster in se, isolated and perfectly aligned with the log dashed line) - Group A’: Massachusetts General Hospital and MIT (outliers, with higher normalised citation rates than Harvard, despite publishing less) - Group B: Boston University - Group C: Tufts, North-Eastern University and Dartmouth College - Group C’: Brandeis University (outlier, with a normalised citation rate higher than expected) - Group D: Boston College, University of New Hampshire and University of Massachusetts, Boston - Group E: University of Massachusetts, Lowell - Group F: Bridgewater State College, Salem State College and finally Plymouth State College (all with less than 500 publications)

London presents a fuzzier distribution, but one which remains interesting: - London University35 is the only institution, which reaches close to the total number of publications of Harvard with 186 000. Its normalised citation rate is however relatively poor (1,69), about the same as Dartmouth College. - UCL, Imperial and King’s with large volumes of publications (between 40 000 and 80 000) and normalised citation rates fairly similar to those of University of London (above in the case of UCL and Imperial). - Small institutions with volumes of publications between 4500 and 20 000 and relatively high citation rates: Institute of Cancer Research, London School of Hygiene and Tropical Medicine, St Georges University London, Queens Mary University London. All these institutions are either medical research centres or universities with large university hospitals. The only outlier in the group is the London School of Economics. - After a gap in terms of citation rates, the remaining London institutions36 are distributed similarly to those from Boston Groups D to F with: Brunel University 7053 publications, normalised citation rate 1,37), Birbeck University London, University of Greenwich, University of Westminster, Kingston University, Roehampton University, Middlesex University, London Metropolitan University, University of East London, London South Bank University, Buckinghampshire New University, St Mary’s University, Twickenham and University of West London (95 publications, normalised citation rate 0,71)

Paris institutions form an even fuzzier cloud than London in which we can underline the following facts:

35 To obtain the scores of London University we aggregated all publications of the member institutions such as UCL or King’s. 36 Leaving aside specialised institutions such as the Royal Academy of Music or London Business School.

38

- Paris Diderot (26357 publications, normalised citation rate 1,82) is, overall, the best performing institution, with almost identical results to Boston University (Boston group B). Paris-Sud has comparable results. - UPMC publishes more (43383 publications) but with lower normalised citation rates than Paris Diderot - After these institutions there is a gap to institutions publishing less than 8000 publications with highly disparate normalised citation rates in which we find (in order of normalised citation rates): UVSQ, Observatoire, ENS, Collège de France, University Paris-Est Créteil, ESPCI, École Polytechnique, Université d’Évry, Agro Paris Tech, Chimie Paris-Tech, EPHE, Université Paris-Sorbonne, Université Paris 13, Centrale Supélec, ENS and on down to Université Sorbonne Nouvelle (547 publications, normalised citation rate 0,34)

Clearly, the 4 main Parisian universities are the only ones that compare to leading Boston and London ones in terms of volume of publications with Paris Diderot in particular having a fairly good normalised citation rate. Beyond these universities, no institution has more than 8000 publications and, unlike in Boston and London, the small elite institutions such as ENS or École Polytechnique do not differentiate themselves in terms of normalised citation rates.

What is most interesting is to observe what happens when we place the CoMUE on the map:

Figure 23: Normalised citation impact vs number of publications – with ComUE data (data WoS, visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/impact/citations.html

39

For the ComUE, the total number of publications increases but they remain below UCL and Imperial with less than 60,000 in total, even in the case of Sorbonne Paris-Cité, Université Paris-Saclay and Sorbonne Universités.

However, the normalised citation rates all drop considerably to the level of Boston Group C, alongside institutions such as North-Eastern University and this time clearly below both UCL and Imperial. This is due to the aggregation of institutions with different normalised citation impact figures, which lowers the weighted average37.

The next graph adds another dimension by plotting the number of papers per staff against the number of students per staff38.

Figure 24: publications/staff vs student/staff (data WoS and other sources, visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/impact/students.html

37 This impact would be clearest in the CWTS Leiden rankings where the results of ComUE would, presumably, be far lower than those of the leading existing institutions in terms of PP1% and PP10%. 38 Data on number of staff is taken from institutional websites and/or QS.

40

In this case, the graph takes the shape of a waxing moon, in which the leading institutions are all concentrated in the top left hand corner.

Boston forms an almost perfect crescent with: - Group A and A’: Massachusetts General Hospital (with, unsurprisingly a very low student/staff ratio), followed by Harvard (with by far the highest rate of publications per staff) and MIT - Followed by the rest of the institutions in a similar order to that of the Groups we distinguished above, but with some Colleges gaining some places over universities thanks to the lower student/staff ratios: Dartmouth College, Tufts University and Boston University, Brandeis University, Northeastern University, Boston College, etc.

London also forms a crescent with: - 2 small medical research institutes: Institute of Cancer Research and London School of Hygiene and Tropical Medicine - Followed by the 3 leading universities: Imperial, UCL and a bit further back King’s - The rest of institutions specialised in Health Sciences (St Georges, Queen Mary, Royal Veterinary) - University of London, clearly below Imperial and UCL - The rest of London’s institutions following the pattern already identified

In the case of Paris, the pattern is more confused with: - A group of Grandes écoles in the top-left corner with performances as good if not better than Harvard, MIT, UCL and Imperial: Observatoire, Muséum, ESPCI, École Polytechnique, Institut d’Optique, AgroParisTech. - The 4 main universities: Paris Sud, UPMC, Diderot and Descartes - The rest following a similar pattern to previously.

Once again, the introduction of ComUEs in the picture is significant: they perform, in overall terms, worse, not only than the group of Grandes écoles but also than UPMC or Paris 7. PSL appears towards the bottom of the cluster of Grandes écoles with Saclay below PSL; Sorbonne Universités and Sorbonne Paris Cité appear on the right-hand side, below UPMC and Paris Diderot.

Finally, the following graphs plot the total number of publications per staff against the total number of staff per institution.

Figure 25: publications/staff vs number of staff (data WoS and other sources, visualisation: SIRISLab) ● http://sirislab.com/lab/cnrs/impact/publications.html

41

The results are particularly striking for Boston where we have a strong, and surprising, correlation between number of staff and publications per staff, with the usual suspects in the usual order (Harvard, Massachusetts General Hospital and MIT, followed by Tufts, Boston University and Dartmouth College and so on). This means that in Boston the largest institutions also have the most productive researchers.

In London, UCL, Imperial and King’s again appear at the top right with larger staff numbers but similar publications per staff to the specialised medical research centres. These are followed by the other universities which have less staff and less publications per staff. University of London performs less well than UCL, Imperial or King’s.

In Paris, on the contrary to Boston, the correlation is almost inversed with smaller institutions more likely to publish more per staff. Again, at the top left, mainly Grandes écoles, top right the 4 leading universités. Again, the ComUE perform worse in terms of publications per staff than either of the two clusters. PSL and Sorbonne Universités are close to Paris Descartes (with Sorbonne Universités larger in terms of staff); Sorbonne Paris Cités and Paris Saclay have almost identical scores with less publications per staff than either PSL or Sorbonne Universités and larger numbers of staff. This is clearly indicative of a structural disadvantage of the Paris system, presumably amplified by the lack of mobility of good researchers from poorer/smaller institutions to better/larger institutions, as happens in Boston.

Preliminary remarks on the distribution of scientific potential

The similitudes in scientific potential between Boston, London and Paris, which we highlighted in the beginning of this chapter, have thus left way to a clear differentiation once we look at the distribution of this scientific potential in institutional terms.

The results clearly prove the fallacy of using size as a proxy for increasing visibility in the Parisian context. They show that (1) the existing institutional structure does not perform as well, in global terms, as either Boston or London and (2) the ComUE actually perform less well than both the Grandes écoles or the leading universités currently do when benchmarked against London’s and Boston’s institutions on key indicators.

This is coherent with the ComUE’s stated aim of facilitating the transformation of the existing system and helping the emergence of world-class universities at their heart. It also proves that the institutionalisation of ComUE in their current form would not solve the problems of performance that have just been highlighted.

This is why, in the third part of this report we will turn to institutional structure.

42

Part 3. Comparing the structuration of the institutional landscapes

In part 2, we clustered institutions together in function of their performance according to different parameters: rankings, student numbers, staff numbers, bibliometric data, etc.

In Boston, this approach enabled us to define clear groups that perform similarly across all parameters and range from world-class research universities (Harvard and MIT) down to teaching colleges. In London, the results are comparable, even if the top group (Imperial and UCL) does not perform as highly as in Boston.

In Paris, on the other hand, there is a clear dichotomy whereby, depending on the parameters examined, the best performers are either a group of Grandes écoles or a group of 4 universités. The distance of each group from the world-class universities of Boston thus varies from parameter to parameter. When the ComUE are added to the picture they systematically perform less well than either group.

In order to understand better why this is happening, we will try and analyse in more detail the institutional structure of the system of HE of each city, by looking first at legal distinctions, then at existing classifications and finally by comparing institutions in terms of their two key missions: teaching and research. It would have been interesting to go into closer detail into the way missions are divided between institutions, however no suitable tools exist and data gathering would be highly complex39.

Boston Area

Legal categories and missions

The system of HE&R in Massachusetts, like in most other American states, can be divided into three categories: public, private not-for-profit and private for-profit40.

The Public System41 is itself divided into 3 categories: - University of Massachusetts, which is a university system, like University of California and which includes two universities situated in the Boston area: University of Massachusetts Lowell and University of Massachusetts Boston. - Nine State Universities, including in Boston area: Bridgewater State University, Massachusetts College of Arts and Design and Salem State University. - 15 Community Colleges, including in the Boston area: Berkshire Community College, Bunker Hill Community College, Great Bay Community College, Massachusetts Bay Community College, Massasoit Community College, Middlesex Community College-Bedford, Northern Essex Community College.

39 No comparative tool yet exists. The classification tool U-rank (http://www.u-map.org/ – not to be confused with U-Multirank) could provide an interesting starting point but currently it only includes 117 institutions, of which only one (UVSQ) is situated in Boston, London or Paris. 40 We will leave the for-profit sector aside. One example, included in the graph below is University of Phoenix- Massachusetts. 41 As mentioned in the first part, when possible we include institutions within Greater Boston but not in Massachusetts in our analysis. These include public institutions from New Hampshire such as University of New Hampshire-Main Campus

43

Each category has a clear mission42: Community Colleges Community Colleges “are committed to Community colleges (or junior colleges) are excellence in teaching and learning and traditionally two-year colleges focusing on provide academic preparation for transfer employability. They typically award the associate to four-year institutions, career degree that counts as a HE degree in the US, but is not preparation for entry into high demand recognised as such in most other countries. occupational fields, developmental Conventions often exist with 4-year senior colleges, coursework, and lifelong learning but much less frequently with universities. There is a opportunities.” trend, however, for Community Colleges to build up to either offer bachelor programmes, or to develop into State Universities are “Committed to "Comprehensive Community Colleges", that are closely excellence in instruction and to providing associated to four-year-institutions, allowing their responsive, innovative, and educational students to move on to further education if desired. In programs of high quality, they seek to terms of the Carnegie classification, community develop each student’s critical thinking, colleges are mostly considered as associate’s colleges quantitative, technological, oral, and or, if they have their own bachelor, as written communication skills, and practical Baccalaureate/Associate’s Colleges. appreciation of the arts, sciences, and humanities as they affect good citizenship and an improved quality of life”.

The University of Massachusetts mission “is to provide an affordable education of high quality and conduct programs of research and public service that advance our knowledge and improve the lives of the people of the Commonwealth”.

Unlike in other American states like California or A “university” in Massachusetts Wisconsin, the main public universities in “A university must meet the requirements of a four- Massachusetts are not considered amongst the year or senior college, and must provide graduate best in the world. programs in four or more distinct professional fields This said, THE does rank University of of study. A university must clearly identify graduate Massachusetts 141st (and it was in the top 100 studies as a distinct element within its organization, up until 2015)43 ; whereas ARWU ranks both and must provide the additional , facilities, and University of Massachusetts-Amherst and resources necessary to support sound graduate University of Massachusetts-Medical School in programs.” the top 150 in the world. Massachusetts Board of Higher Education: Finally, it is interesting to note that MIT is a http://www.mass.edu/forinstitutions/academic/documents/610C land-grant university and thus receives federal MR.pdf funding despite being private.

The private not for profit sector includes a large number of highly diverse institutions, including not only leading universities such as Harvard or MIT, but specialised research institutes, liberal art colleges and specialised schools in fields such as Business, etc.

42 See: http://www.mass.edu/system/aboutsystem.asp#state and the Performance Measurement Report of Massachusetts Board of Higher Education (http://www.mass.edu/bhe/lib/documents/2010Performance MeasurementReport.pdf) 43 Only ENS and École Polytechnique are thus ranked above University of Massachusetts in THE! On a side note, the fact that THE ranks University of Massachusetts as a whole but ranks University of Wisconsin and University of California on a campus per campus basis is inconsistent.

44

Their missions are defined by each institution independently and vary from Harvard’s wonderful declaration that “Harvard University does not have a formal mission statement” to the more traditional: “The mission of the Massachusetts Institute of Technology is to advance knowledge and educate students in science, technology, and other areas of scholarship that will best serve the nation and the world in the 21st century. We are also driven to bring knowledge to bear on the world’s great challenges”, and down to very pragmatic ones for the large number of institutions that do not claim to engage in research: “Lasell College engages students in the practice of their fields of study through collaborative learning that fosters lifelong intellectual exploration and social responsibility”.

Carnegie classification Another way to approach the institutional structure in Boston is through the Carnegie Classification of Institutions of Higher Education, a framework for classifying colleges and universities in the United States. It includes all accredited, degree-granting colleges and universities that are included in the National Center for Education Statistics Integrated Postsecondary Education Data System (IPEDS). It was last reworked in 2005. Institutions are distinguished mainly by the number of programmes leading to different degrees: Within the doctoral universities, the number of doctoral programmes and their ratio to Master programmes is the main distinction. Master’s colleges are basically Master-centred universities, and Bachelor- and Associate-colleges are institutions that propose mainly these degrees. For Boston, this gives us the following landscape44:

No. Associate’s Bachelor Master’s Doctoral Doctoral universities 13 183 28,318 20,711 6,009 - R1 7 - R2 4 - R3 2 Master’s Colleges/Unis 17 90 12,098 7,387 256 Baccalaureate Colleges 6 62 2,205 125 0 Special Focus 4-year 33 948 3,975 2,475 1,381 Special Focus 2-year 4 215 1 0 0 Associate’s Colleges 9 6,366 0 0 0 Total 82 7,864 46,579 30,698 7,646

Figure 26: Number of institutions and degrees awarded in Boston per type of institution according to the Carnegie Classification (source: Carnegie classification)

The Carnegie classification corresponds quite closely to the division defined for the public sector and extends it to the rest of institutions. It is interesting to compare it to the French context:  The Community Colleges are all classified as Associate Colleges (with 9 subcategories). They correspond in the French context to IUTs or BTS but function as independent institutions with a state system45.  There are a large number of 4 year undergraduate schools, either generalist such as Wellesley College and classified as Baccalaureate Colleges or specialist in Law, Engineering, Business or Medicine, classified as Special Focus 4-year institutions. They do not have a French equivalent but are most similar to Grandes écoles with the classe préparatoire phase

44 This is a simplified version: in total, there are 33 categories of HE&R institutions. 45 The CPGE could also be included in this category, rather than in the following one.

45

included, both in terms of length (4 years rather than 5 in the case of the Grandes écoles), and in terms of fields of specialisation.

Finally, the universities are classified in 4 categories depending on their research focus: - Highest intensity research (7 institutions: Harvard, MIT, Brandeis, Northeastern, Boston and Tufts Universities and Boston College in Massachusetts) - Higher intensity research (4 institutions: University of Massachusetts-Boston and Lowell, Dartmouth College and Worcester Polytechnic Institute) - Moderate intensity research (3 institutions: Suffold, Lesley and Clark University) - Master’s colleges / universities (17 institutions including Salem State, Bridgewater universities, Emerson and Cambridge College, etc.) They would correspond to French universités and leading comprehensive Grande écoles such as ENS or École Polytechnique. However, unlike in France, they are perceived as belonging to different categories and have clearly differentiated missions.

The classification is not fixed: it is reviewed every 5 years and institutions regularly change categories, depending both on their performance and on their strategic choices46.

London area

Legal categories and missions Universities in the UK Unlike in Boston, in London the only HE British institutions can only use the title institutions stricto sensu are Royal Charter “university” if they have been granted a charter universities, all authorised to deliver PhD by the Privy Council, under the terms of the diplomas. These universities are considered Further and Higher Education Act 1992 public institutions if they are funded by the Higher Education Funding Council for England Further Education Colleges (HEFCE). These Colleges are very versatile and their role in There are currently very few private universities the educational system should not be in the UK. In London, they include the not-for- underestimated. They are extremely profit Regent’s University London and the for- heterogeneous, teaching pupils from 16 onward profit BPP University (owned by the Apollo up to lifelong learning in special fields. They are Group) as well as the University of Law47. roughly comparable to American Community Colleges or 2-year Special-Focus-Colleges, but are Alongside Universities, the London landscape generally not considered part of the HE system. includes two other major categories: Only a portion of 13,000 FEC-students in London is participating in HE strictly speaking. They have specialised Colleges and Further Education not been included in our visualisations. They Colleges. figure, however, in the overview chart on diplomation in the three systems, and their omission puts into perspective the overall lower number of HE institutions in London. 46 For example, between 2005 and 2010, both Northeasternwww.londoncolleges.com University and Boston College were reclassified as Highest Intensity rather than Higher Intensity, whereas Dartmouth College was reclassified as Higher Intensity rather than Highest Intensity. 47 The situation is changing with the decision of the government to encourage private competition in HE&R: https://www.timeshighereducation.com/news/treasury-plan-seeks-put-competitive-pressures-universities

46

Specialised Colleges are highly diverse. They include world-renowned institutions that deliver all levels of diploma including the PhD, such as London Business School, Saint Georges - University of London or the London School of Hygiene and Tropical Medicine; Art Schools such as the Courtauld Institute of Art or the Guildhall School of Music and Drama.

Unlike in Boston, no attempt has been made to differentiate institutions in terms of mission. Thus, University of London, as a university system of some 20 members, resembles in many ways the Massachusetts public system of Higher Education. However, the mission of University of London remains incredibly vague48: “The objects of the University, carried out through the Colleges primarily, and also through the Central Academic Bodies and Central Activities, are, for the public benefit, to promote education of a university standard and the advancement of knowledge and learning by teaching and research; and to encourage the achievement and maintenance of the highest academic standards. In pursuit of these objects, the University will serve and support the interests of the Colleges.”

Classification Up until 1992, institutions of HE&R were typically divided between polytechnics and universities. The granting of university status to polytechnics increased their visibility (notably for international students) but also immediately led to attempts to create new categories such as the Russell Group, set up by leading research universities as a lobby platform.

The main classifications include those based on historical aspects or on organisational aspects as well as more or less coherent alliances:  historical categories: “old civics”, “redbricks”, “greenfields”, “technological”, “new”, etc.49  organisational categories: Unitary universities, Examining Board universities, Federal universities, Collegiate universities  alliances: Russell Group, University Alliance, Million+, GuildHE, etc.

Vikki Boliver recently criticised the use of these categories in functional terms and proposed a data- driven approach, more in line with the Carnegie Institute50. She considers a range of criteria including: research activity, teaching quality, economic resources, academic selectivity, to the socioeconomic student mix.51 The classification is done using a progressive cluster-analysis approach instead of a variable-based approach such as that used by the Carnegie classification.52

The most remarkable difference is between clusters 2 and 3: the former comprising more than 70% of all Old universities (pre-1992 universities) and the latter of more than 70% of all New ones (the former polytechnics). Cluster 2 universities have more than 3 and a half times more research income, more than six times more income from endowment and investments, and spend more than 50% more on academic services by student. They also have 50% more postgraduate students and

48 Information available in the statutes: http://www.london.ac.uk/fileadmin/documents/about/governance/ ordinances/SandO2015/Statutes_wef_8_Oct_2015.pdf 49 This is a classical view going back to Tight, 1988. “Institutional typologies”. Higher Education Review, 20 and Tight, 1996. “Institutional typologies re-examined”. Higher Education Review, 29. 50 Boliver, 2015 “Are there distinctive clusters of higher and lower status universities in the UK?”, Oxford Review of Education, 41.5. 51 Details on the criteria and the distribution among the clusters is included in annex. 52 Carnegie has some characteristics of a cluster-based classification, but since the 1970s, its reforms have relied more and more on variable tweaking. Currently, it should be considered a variable-based approach.

47 much better results on the Research Assessment Exercise. They are distinctively more selective and have a larger percentage of students from higher social classes. The only exception is teaching: the difference between cluster 2 and 3 in quality here is negligible. Oxford and Cambridge alone stand out from cluster 2 and in consequence form the first cluster. The differences with other universities are clear in all criteria, again, except teaching. Oxford and Cambridge do not have significantly better student evaluations or completion rates than cluster 2 and even cluster 3 universities. Cluster 4 institutions have significantly lower scores on all criteria. V. Boliver does not name her clusters, but it seems sensible to consider clusters 1 and 2 as research universities and clusters 3 and 4 as comprising essentially teaching institutions.

The clustering is thus driven by research criteria, whereas teaching is remarkably homogeneous across the clusters with much lesser variation in results in this area than in any of the others. This said, V. Boliver does not include criteria related to outcome of the students such as expectations of employability, salary, or position. The inclusion of such criteria would most likely differentiate this field.

The following table lists the type of institutions by cluster for the London area53.

UG PG Total Cluster Cluster 2: 7 institutions of 39 national, 18% 71,665 53,930 125,595

Goldsmiths College 5,235 2,930 8,165 2 Imperial College of Science, Technology & Medicine 9,015 7,595 16,610 2 King's College London 17,610 11,120 28,730 2 London School of Economics and Political Science 4,415 6,185 10,600 2 Queen Mary University of London 11,390 4,570 15,960 2 Royal Holloway / Bedford New College 7,170 2,745 9,915 2 University College London 16,830 18,785 35,615 2 Cluster 3: 10 institutions of 67 national, 15% 124,780 42,020 166,800

Brunel University London 9,695 4,025 13,720 3 City University 9,695 8,585 18,280 3 Kingston University 16,975 4,935 21,910 3 London South Bank University 13,020 4,715 17,735 3 Middlesex University 13,675 3,805 17,480 3 Roehampton University 5,985 1,665 7,650 3 University of Greenwich 16,105 5,190 21,295 3 University of West London 8,980 1,510 10,490 3 University of Westminster 16,145 4,320 20,465 3 University of the Arts, London 14,505 3,270 17,775 3 Cluster 4: 2 institutions of 19 national, 11% 22,710 7,230 29,940

London Metropolitan University 11,030 3,055 14,085 4 University of East London 11,680 4,175 15,855 4 No cluster: 21 institutions 37,750 18,610 56,360

Birkbeck College 9,680 4,255 13,935 Buckinghamshire New University 7,970 1,065 9,035

53 There are 21 institutions that were not considered, as they are not proper universities, but specialised institutions. They are simply not accounted for in the classification and should not be considered as lower than the rest.

48

Conservatoire for Dance and Drama 1,070 190 1,260 Courtauld Institute of Art 165 305 470 Guildhall School of Music and Drama 555 355 910 Heythrop College 350 355 705 London Business School 0 1,790 1,790 London School of Hygiene and Tropical Medicine 0 1,245 1,245 Ravensbourne 1,985 70 2,055 Rose Bruford College 730 60 790 Royal Academy of Music 360 410 770 Royal College of Art 0 1,445 1,445 Royal College of Music 425 380 805 St George's Hospital Medical School 4,575 930 5,505 St Mary's University, Twickenham 3,905 1,430 5,335 Institute of Cancer Research 0 300 300 Royal Central School of Speech and Drama 635 355 990 Royal Veterinary College 1,635 510 2,145 School of Oriental and African Studies 3,015 2,890 5,905 Trinity Laban Conservatoire of Music and Dance 695 270 965 Total: 40 institutions 256,905 121,790 378,695

Figure 27: Classification of London’s institutions according to V. Boliver 2015

Paris area

Legal categories and missions This report will not overview the legal categories of French institutions, however it is interesting to note that the structuration of the public system of HE&R is profoundly different in Paris from both London (where universities exist alongside specialised colleges and research institutes) and Boston (where the structuration is largely dependent on institutional mission). The key categories are as follows:

49

UG PG Total Private Public Total Type of institution students students students inst. inst. instit. Autres établissements 3,867 3,867 5 6 11 EPSCP-dependent 474 474 3 2 5 Écoles d'ingénieurs and alike 6,810 6,810 9 4 13 École normale supérieure 3,829 3,829 2 2 Grand établissement 4,966 25,854 30,820 20 20 Université 280,379 117,319 397,698 16 16 Total 285,345 158,153 443,498 17 50 67

Figure 28: Number of students by category of institution (source: Open Data Portal Enseignement Supérieur et Recherche, OS for institutions not depending on this ministry)54

Classification Unlike in Boston or London, to the best of our knowledge, no attempt has been made to classify French institutions in other categories than those defined above. The main associations regroup institutions with the same legal status, even if they sometimes accept other institutions or create subgroups per mission. For example, the CPU represents French Universités, but welcomes other types of institutions, to which it links on its search page55: Université, Grand établissement, École centrale, Institut national des sciences appliquées, Regroupements universitaires, École normale supérieure, Institut national polytechnique, École française à l’étranger.

A proposal to regroup Paris Descartes, Paris Diderot, Paris Sud, UPMC, École normale supérieure and École Polytechnique in a single category could seem logical to external observers. In France, it would surprise many because institutions are classified primarily according to their legal categories rather than their actual mission.

Comparing Boston, London and Paris in terms of institutional structuration

In order to better understand the structural differences between the three cities, the following figures compare the distribution of research publications indexed in WoS with that of undergraduate and graduate students in each institution of each city. In the case of Paris, a fourth figure aggregates student and publication numbers by ComUE.

54 Students in classes préparatoires have not been included. 55 http://www.cpu.fr/page-annuaire/

50

Harvard stands far out with 180651 registered Boston publications 50000 90000

45000 80000

40000 70000 35000 60000 30000 50000 25000 40000 20000 30000 15000 10000 20000 5000 10000

0 0

Curry College Curry

Lasell College Lasell

Fisher College Fisher

Quincy College Quincy

Boston College Boston

Babson College Babson

Tufts University Tufts

Hebrew College Hebrew

Laboure College Laboure

Endicott College Endicott

Emerson College Emerson

Lesley University Lesley

Newbury College Newbury

Simmons College Simmons

Bay State College State Bay

Wellesley College Wellesley

Suffolk University Suffolk

Boston University Boston

Wheelock College Wheelock

MCPHS University MCPHS

Bentley University Bentley

Mount Ida College Ida Mount

Harvard University Harvard

Cambridge College Cambridge

Merrimack College Merrimack

Dartmouth College Dartmouth

Brandeis University Brandeis

Anna Maria College Maria Anna

Pine Manor College Manor Pine

Marian Court College Court Marian

William James College James William

Boston Baptist College Baptist Boston University State Salem

Salve Regina University Regina Salve

Northeastern University Northeastern

Urban College of Boston of College Urban

Northpoint Bible College Bible Northpoint

Berklee College of Music of College Berklee

Episcopal Divinity School Divinity Episcopal

Wheaton College-Norton Wheaton

The Boston Conservatory Boston The

Eastern Nazarene College Nazarene Eastern

Montserrat College of Art of College Montserrat

Plymouth State University State Plymouth

Emmanuel College-Boston Emmanuel

New England School of Law of School England New

Boston Architectural College Architectural Boston

Bridgewater State University State Bridgewater

Massachusetts School of Law of School Massachusetts

Berkshire Community College Community Berkshire

Great Bay Community College Community Bay Great

Massasoit Community College Community Massasoit

Bunker Hill Community College Community Hill Bunker

North Shore Community College Community Shore North

The New England Institute of Art of Institute England New The

Hult International Business School Business International Hult

ITT Technical Institute-Wilmington Technical ITT

Wentworth Institute of Technology of Institute Wentworth

New England College of Optometry of College England New

University of Massachusetts-Lowell of University

Northern Essex Community College Community Essex Northern

University of Massachusetts-Boston of University

MGH Institute of Health Professions Health of Institute MGH

Andover Newton Theological School Theological Newton Andover

New England School of Acupuncture of School England New

University of Phoenix-Massachusetts of University

Longy School of Music of Bard College Bard of Music of School Longy

Franklin W Olin College of Engineering of College Olin W Franklin

Massachusetts Institute of Technology of Institute Massachusetts

Massachusetts Bay Community College Community Bay Massachusetts

Middlesex Community College-Bedford Community Middlesex

Northeastern University Global Network Global University Northeastern

The New England Conservatory of Music of Conservatory England New The

Massachusetts College of Art and Design and Art of College Massachusetts

Benjamin Franklin Institute of Technology of Institute Franklin Benjamin

School of the Museum of Fine Arts-Boston Fine of Museum the of School

University of New Hampshire-Main Campus Hampshire-Main New of University

New England College of Business and Finance and Business of College England New

Boston Graduate School of Psychoanalysis Inc Psychoanalysis of School Graduate Boston

Lawrence Memorial Hospital School of Nursing of School Hospital Memorial Lawrence

Le Cordon Bleu College of Culinary Arts-Cambridge Culinary of College Bleu Cordon Le Hellenic College-Holy Cross Greek Orthodox School of… School Orthodox Greek Cross College-Holy Hellenic

Total undergraduate Total postgraduate WoS count

Figure 29: Main institutions of HE&R in Boston (source: Carnegie classification, WoS)

51

London 50000 90000

45000 80000

40000 70000 35000 60000 30000 50000 25000 40000 20000 30000 15000

10000 20000

5000 10000

0 0

Ravensbourne

BirkbeckCollege

HeythropCollege

The City The University

Goldsmiths College Goldsmiths

Kingston University Kingston

Royal College of College Art Royal

Middlesex University Middlesex

Rose Bruford College Bruford Rose

King's College London College King's

Royal College of College Music Royal

Roehampton University Roehampton

London Business School LondonBusiness

Royal Academy of Music of Academy Royal

Courtauld Institute of Institute Courtauld Art

Brunel University London Brunel

University College London UniversityCollege

The University of Greenwich University of The

The Royal College Veterinary The

The University of East London University of The

University of the Arts, London Universityof

London South Bank University LondonSouth Bank

The University of Westminster University of The

The University of West University London of The

London Metropolitan University LondonMetropolitan

The Institute of Institute Research The Cancer

Buckinghamshire New University BuckinghamshireNew

St Mary's University, Twickenham StUniversity, Mary's

Queen Mary University of London University Mary of Queen

St George's Hospital Medical School StGeorge's

Conservatoire for Danceand for Drama Conservatoire

Guildhall School of Music and of Drama School Music Guildhall

Royal Holloway and Bedford New and Holloway College Bedford Royal

The School of Oriental andStudies African Orientalof School The

Imperial College of Science, Technology and… Technology Science, of College Imperial

Trinity Laban Conservatoire of Music and of Dance Conservatoire TrinityLaban

The Royal Central School of Speech ofDrama† Speech Royal and School The Central

London School of Hygiene and Tropical Hygiene Medicine Londonof School London School of Economics and Science Economics Political Londonof School

Total undergraduate Total postgraduate WoS count

Figure 30: Main institutions of HE&R in London (source: HEFCE, WoS) 52

Paris 50000 90000

45000 80000

40000 70000 35000 60000 30000 50000 25000 40000 20000 30000 15000

10000 20000

5000 10000

0 0

ENSTA

ECAM-EPMI

Institute Curie Institute

Agro ParisTech Agro

CentraleSupélec

Telecom SudParis Telecom

Collège de France de Collège

Telecom Paristech Telecom

Université Paris-Sud Université

École polytechnique École

Observatoire de Paris de Observatoire

Université Paris Diderot Paris Université

Université Paris Descartes Paris Université

Université Paris-Dauphine Université

Université Paris-Sorbonne Université

EPF - École d'ingénieur-e-s École - EPF

Université Panthéon-Assas Université

Institut catholique de Paris de catholique Institut

École nationale des Chartes des nationale École

ESPE de l'académie de Paris de l'académie de ESPE

École de biologie industrielle biologie de École

Université de - de Université

Institut de physique globe du physique de Institut

ESPE de l'académie de Créteil de l'académie de ESPE

Université Paris 13 - Paris Nord Paris 13 - Paris Université

Université d'Évry-Val d'Essonne d'Évry-Val Université

Institut protestant de théologie de protestant Institut

Université Pierre et Marie et Pierre Université

ESPE de l'académie de Versailles de l'académie de ESPE

École française d'Extrême-Orient française École

École pratique des hautes études hautes des pratique École

Institut national d'histoire de l'art de d'histoire national Institut

École normale supérieure de Paris de supérieure normale École

Institut d'Optique Graduate School Graduate d'Optique Institut

École nationale vétérinaire d'Alfort vétérinaire nationale École

Institut d'études politiques de Paris de politiques d'études Institut

École d'ingénieur de la ville de Paris de ville la de d'ingénieur École

Université Paris-Est -la-Vallée Paris-Est Université

École nationale supérieure maritime supérieure nationale École

École normale supérieure de Cachan de supérieure normale École

Muséum national d'histoire naturelle d'histoire national Muséum

Université Sorbonne Nouvelle - Paris 3 - Paris Nouvelle Sorbonne Université

École nationale des ponts et chaussées et ponts des nationale École

Université Paris 1 - Panthéon Sorbonne Panthéon 1 - Paris Université

École d'ingénieurs du monde numérique monde du d'ingénieurs École

Institut supérieur de mécanique de Paris de mécanique de supérieur Institut

Université Paris-Est Créteil Val-de-Marne Créteil Paris-Est Université

Institut supérieur d'électronique de Paris de d'électronique supérieur Institut

École nationale supérieure Louis Lumière Louis supérieure nationale École

Conservatoire national des arts et métiers arts des national Conservatoire

Université Paris 8 - - Saint-Denis Vincennes 8 - Paris Université

Université Paris Ouest Nanterre La Défense Nanterre Ouest Paris Université

École nationale supérieure d'arts et métiers d'arts supérieure nationale École

École des hautes études en sciences sociales sciences en études hautes des École

École Polytechnique de l'Université Paris-Sud l'Université de Polytechnique École

École nationale supérieure de chimie de Paris de chimie de supérieure nationale École

école des Hautes Etudes Commerciales de Paris de Commerciales Etudes Hautes des école

Facultés Libres de Philosophie et de etPsychologie de Philosophie de Libres Facultés

Institut d'administration des entreprises de Paris de entreprises des d'administration Institut

Université de Versailles Saint-Quentin-en- Versailles de Université

École supérieure des techniques aéronautiques et de… aéronautiques techniques des supérieure École

Institut national des langues et civilisations orientales et civilisations langues des national Institut

École nationale supérieure de l'électronique et de ses… de et l'électronique de supérieure nationale École

École d'ingénieur d'agro-développement international d'agro-développement d'ingénieur École

Groupe des écoles nationales d'économie et statistique d'économie nationales écoles des Groupe

Institut agronomique, vétérinaire et forestier de France de et forestier vétérinaire agronomique, Institut

Institut de management et de communication interculturels etcommunication de management de Institut

École nationale supérieure d'informatique pour l'industrie et… l'industrie pour d'informatique supérieure nationale École

École supérieure des sciences économiques et commerciales économiques sciences des supérieure École

Ecole Superieure de Physique et de Chimie Industrielles de la… de Industrielles Chimie de et Physique de Superieure Ecole

Institut national du sport, de l'expertise et de la performance la de et l'expertise de sport, du national Institut

Institut national supérieur de formation et de recherche pour… recherche et de formation de supérieur national Institut

École supérieure d'ingénieurs des travaux de la construction de… construction la de travaux des d'ingénieurs supérieure École

École spéciale des travaux publics, du bâtiment et de l'industrie de et bâtiment publics, du travaux des spéciale École

École internationale des sciences du traitement de l'information de traitement du sciences des internationale École École d'ingénieur généraliste en informatique et technologies du… technologies et informatique en généraliste d'ingénieur École

UG PG Web of Science Documents

Figure 31: Main institutions of HE&R in Paris (source: Open Data Portal Enseignement Supérieur et Recherche, QS for institutions not depending on this ministry, WoS). 53

USPC stands far out with 78.335 undergrad students and Saclay with 93.966 publications

50000 v 90000 45000 80000 40000 70000 35000 60000 30000 50000 25000 40000 20000 30000 15000 10000 20000 5000 10000 0 0

UG students PG students Web of Science Documents

Figure 32: Main ComUE in Paris (source: Open Data Portal Enseignement Supérieur et Recherche, OS for institutions not depending on this ministry, WoS).

The comparison of the 4 tables from the preceding pages clearly highlights the overall similarity between the structure of HE&R in the 3 cities in terms of research production with a strong hierarchy between institutions.  This hierarchy is clearest in Boston with only 16 institutions indexed in WoS, 11 with more than 3000 publications, 4 with more than 30 000.  In London, 31 institutions have publications indexed in WoS (almost double), but there are only 12 with more than 3000 publications and 3 with more than 30 000.  In Paris, 35 have indexed publications (similar to London), there are 15 with more than 3000 publications and 1 with more than 30 000.  If the ComUE became universities then the number of Parisian institutions would be cut to 8, but surprisingly only 3 would have more than 30 000 publications and all would remain below both Imperial and UCL (without mentioning Harvard) in terms of total number of publications.

What is most interesting, however, is to compare the institutions in terms of student numbers:  The seven largest universities in number of undergraduate students are all situated in Paris: Paris Diderot, Paris Descartes, UPMC, Paris Sud, Panthéon Sorbonne, Paris-Ouest Nanterre and Paris 9 all have more undergraduate students than the university with most undergraduates in Boston (Boston University) or in London (King’s).

54

 Furthermore, the four leading French universities in terms of publications also have less graduate students than Boston University, Harvard, UCL or King’s (despite Paris concentrating proportionally more graduate students than both Boston or London)  Needless to say, the ComUE cause the total number of undergraduate students to skyrocket: Sorbonne Paris Cité would have over 8 times more undergrad students than Harvard for just over twice the number of graduate students and less than one third the number of publications… The only exception would be Paris Sciences et Lettres, which would resemble Imperial in terms of student numbers and undergraduate / graduate proportions but would still have less than half the total number of publications

55

Preliminary Conclusion(s)

The comparisons we have made so far between Boston, London and Paris seem to assume that the “Harvard model” is an ideal. This was probably unavoidable, seeing the data considered (rankings, publications in WoS, number of students, etc.) and the cities we selected. We think that it is justified because the reforms underway explicitly make reference to rankings and aspiring world-class universities usually benchmark themselves with Anglo-Saxon institutions.

This said, it is important to open this narrative to a more critical perspective by questioning the stated aim of institutions to become “world-class” and examining the supposed impact of “world- class universities”.

Do World-Class Universities increase social stratification?

The capacity to educate students efficiently is a cornerstone for the assessment of the quality of an educational system: it should enable the greatest number of people possible to reach their potential. Measuring and comparing the relative capacity of the Boston, London and Paris systems of HE&R is however extremely difficult, especially seeing the wide variety of possible criteria that could be taken into account, such as:

 educational attainment;  the role and impact of a sub-tertiary vocational training;  graduation rates and time to graduation;  profile of the HE systems in terms of output disciplines;  social mobility;  employment and earnings.

It is undeniable that the UK and the US perform badly in terms of social mobility in Higher Education compared to other OECD countries56. Furthermore, although we have found no clear evidence that institutional structure has an impact on social mobility, the most socially mobile countries (Scandinavia and Canada) seem to have less stratified HE&R systems57.

This explains why specialists such as E. Hazelkorn argue against a “Harvard-here” model that “aims to replicate the experience of Harvard University or the Ivy League by encouraging greater vertical or hierarchical (reputational) differentiation between HEIs [Higher Education Institutions], with greater distinction between research (elite) universities and teaching (mass) HEIs” and in favour of a socio- democratic model that “seeks to balance excellence and equity by supporting the development of a world-class system of higher education across a country by strengthening horizontal (mission or functional) differentiation across a diverse portfolio of high-performing HEIs, some of which may be globally or regionally focused”.

56 See the excellent OECD-report “Education at a Glance” from 2015 http://www.oecd.org/education/education-at-a-glance-19991487.htm 57 It would be necessary to study the Scandinavian and Canadian models in more detail before reaching any conclusions.

56

This would be a strong argument against the current reforms, if social mobility was higher in France. However, this is not the case.

In terms of percentage of population with a Higher Education diplomas, the US and the UK system outrank the French system by 10 percentage point distance and progress in France over the last 10 years has been only half that which the British system has witnessed and under both the OECD and the EU21 averages. 50 45 40 35 30 25 20 15 10 5 0 2000 2005 2010 2014 France 22 25 29 32 UK 26 30 38 42 US 36 39 42 44 OECD 22 26 30 34 EU21 20 24 28 32

Figure 33: Trends in educational attainment: tertiary. (Source OECD, Education at a Glance, Table A1.4a)58

Social mobility is affected by a plethora of elements beyond higher education (family, elementary and higher education, economic structure and job market, taxation and redistributive policies, public welfare services, urban planning and the housing market, discrimination, etc.), but higher education is understood to play an important role to it, particularly at the higher end. And, in terms of social mobility, the three countries are close:

Figure 34: Strength of the link between individual and parental earning in OECD countries. Reproduced from A Family Affair: Intergenerational Social Mobility Across OECD-Contries (OECD 2010)

58 These numbers are aggregated.

57

Figure 35: Social Mobility and the Gini Index (source: Miles Corak, 2013)

When looking at the French higher education landscape, in terms of social mobility, a number of striking facts emerge: 1. Between 2006 and 2010, the share of university students coming from lower socioeconomic levels has decreased from 35% to 31%, while those of the higher classes have increased from 32 to 36%. In the Grandes écoles, the students from the working class represent only 10% of the total.59

2. The existence of a clear cleavage, or two-tier system, that separates the Grandes écoles from the regular universities. Grandes écoles are very selective, and are training a smaller proportion of the French population than in the past. Universities are not selective, and are training a larger number and share of students. The democratisation of French higher education since WWII has largely been tackled by the universities, leaving the Grandes écoles sytem relatively unaffected; this disparate effort has led to an increase in stratification. This stratification could have an impact on the social outcomes of the system, since graduates from the Grandes écoles (particularly the best ones, normally located in Paris) are very disproportionally occupying the best positions in both the public and private sector. 60

3. The rates of failure in French universities are massive: about 50%61 of first year students do not make it to the second year.62 So, although higher education is accessible in France (it remains free, and regular universities don’t select students) this accessibility doesn’t translate into academic success. Furthermore, failure rates are higher amongst university

59 http://www.challenges.fr/economie/20120907.CHA0555/grandes-ecoles-universites-le-vrai-prix-des-etudes- superieures-en-france.html, based on OECD data. 60 Brezis and Hellier, 2013 “Social mobility at the top: Why are elites self-reproducing?”, ECINEQ WP 312. 61 Réussite et échec en premier cycle, http://cache.media.enseignementsup- recherche.gouv.fr/file/2013/44/7/NI_MESR_13_10_283447.pdf 62 http://www.letudiant.fr/etudes/fac/valerie-pecresse-lance-son-plan-anti-echec-a-la-fac/universite-pourquoi- tant-dechecs-19057.html

58

students with lower socioeconomic profiles63. These failure rates have massive social costs, and due to this double differentiation, they disproportionally affect the worse-off social strata, hindering social mobility.

4. In parallel, students which at the end of the higher education would like to pursue specialised or shorter-cycle studies in the STS (sections de techniciens supérieurs) and IUT (Instituts Universitaires de Technologie) get rejected from these institutions64, which are also selective, and enrol in non-selective, longer, comprehensive universities. This phenomenon of misallocation, which leads to misalignment between the objectives, expectations and capacities of the student, and those offered by French universities, hinder academic success, and, most probably, social mobility.

5. Tertiary education studies remain almost tuition-free in France, but, according to the OECD, this doesn’t enable full access to higher education, since scholarships and other aids dependant on social status are low in comparison with other countries with higher social mobility (ex. Norway, Australia or New Zealand). In fact, the best-funded non-tuition public transfer to French university students is the rent subsidy (aide au logement), which is independent of the socioeconomic background of the recipient.65

6. The precedent elements are compounded by the fact that expenditure per student is significantly higher in top Grandes écoles than in regular universities, around 3.5 times higher66. Analysing the breakdown or the rationality of this higher budget per student is beyond the scope of this report, but it could have a negative impact on the educational development of students in the less well-financed institutions, which are also less well-off in socioeconomic terms. According to research by Brezis and Hellier67 , this could have a counter redistributive effect of public spending, hindering social mobility and preserving economic and cultural elites: “Our main result is that, the higher the difference in expenditure per student between the elite and standard universities, the lower the upward social mobility of the middle class, and the more self-reproducing the elite group”.

All these elements are simple observations, they do not prove anything and certainly should not be taken as the basis for any theory on the impact of either the existing French system or world-class universities on social mobility. However, they do highlight the importance of taking social mobility into account when creating world-class universities.

Do World-Class Universities have a strong positive impact on their environment?

If the evidence of world-class universities impact on social mobility within a system of HE&R remains unclear, their impact in terms of visibility is clear: “world-class universities are today more than just

63 La réussite des étudiants selon les difficultés financières et la perception d’une allocation d’études, http://cache.media.enseignementsup-recherche.gouv.fr/file/2014/91/3/NI_ESR_14_05_340913.pdf 64 Ibid 3. 65 http://www.challenges.fr/economie/20120907.CHA0555/grandes-ecoles-universites-le-vrai-prix-des-etudes- superieures-en-france.html 66 Observatoire Boivigny. 67 ibid

59 cultural and educational institutions - they are points of pride and comparison among nations that view their own status in relation to other nations.”68

This said, changing a system of HE&R should not be about national pride or status but about improving the overall performance of the system. And the value of investing large amounts of time and money into creating world-class universities rather than attempting to improve the performance of the system as a whole needs to be discussed.

World-class universities have an important positive effect on systems of HE&R in which they are situated due to two phenomena, which are themselves linked: natural distribution curve and Matthew effect. Institutions tend to be distributed according to a power-law at both local and more importantly, global levels. This is illustrated by multiple examples such as following figure, which shows the distribution of institutions by total score in the ARWU ranking.

Figure 36: The distribution of institutions by total score and overall rank in ARWU (source Liu and Cheng 2005)69

The superstar or Matthew effect70 postulates that the best scholars get more recognition than they work warrants and that publications in well referenced journals get more citations than they warrant71. As a result, a very reduced number of elite institutions garner the vast majority of benefits.

For example, M. H. Medoff showed that the Matthew Effect applied to the Economics Departments of 19 US universities but nearly all the advantage accrued to two departments: that of Chicago and Harvard72. These two universities are, not surprisingly, the top 2 universities in ARWU’s ranking in the field of Economics and Business.

Similarly, the impact of country of publication has been studied in a series of papers by M. Bonitz and his collaborators that reflect “the national ability for competition in global science”73. These showed that, for publications from 1990-1994, France was already suffering from a relative national

68 Justin Lin (World Bank VP and Chief Economist) in Salmi, 2009 op. cit. 69 Liu, N. C., & Cheng, Y. (2005). The Academic Ranking of World Universities. Higher Education in Europe, 30(2), 127–136. 70 Merton 1968. The Matthew effect in science. Science, 159 (3810). 71 Larivière and Gingras, 2010. The impact factor’s Matthew effect: a natural experiment in bibliometrics, Journal of the American Society for Information Science and Technology, 61/2. 72 Medoff, 2006. “Evidence of a Harvard and Chicago Matthew effect”. Journal of Economic Methodology, 13(4). 73 For example, Bonitz, Bruckner, and Scharnhorst, 1997. “Characteristics and impact of the Matthew effect for countries”, Scientometrics, 40(3).

60 loss in terms of citations with exceptions in the fields of Mathematics, Engineering and Physics, in particular in comparison with countries such as the UK and US.

As a result, although the HE&R systems of Boston, London and Paris remain similar, the ongoing globalisation of HE&R is likely to increase differentiation. Because national system of HE&R no longer exist in isolation, the existence of a powerful attractor in the form of a world-class university is key not only to attract new talent but to retain existing talent. These universities act as lighthouses, attracting investment and talent not only to their institution but to the region as a whole and the neighbouring institutions of HE&R. And the fact that Paris does not host any of the top 20 world-class institutions is, potentially, a major disadvantage, especially with the growing competition from emerging systems of HE&R.

The evolution of the Parisian landscape and world-class universities

Most attempts to characterise world-class universities agree on a list of basic features, which include74: - excellent students and faculty; - abundant resources; - favourable governance structures.

This list explains why, in the 2000s, external observers were pessimistic about the potential of the French system to enable the emergence of world-class universities. J. Salmi thus wrote: “In the case of France, for example, mergers would augment the critical mass of researchers and bring about a higher place in the [ARWU] ranking that favours research output, but they would not address the fundamental limitations of French universities, including inflexible admission policies, a weak financial basis, rigid governance arrangements, and outdated management practices.”75

Five years after the launch of the IDEX calls, the institutional reforms underway in Paris have deeply transformed the landscape.

Both management practices and governance arrangements have improved; even though they remain far from international standards76. More importantly, thanks in particular to the strategy of the CNRS, National Research Organisation and Universities have successfully integrated their activities thus ensuring that “research is integrated at all levels”, which is one of the key characteristics of world-class universities77.

74 Amongst the extensive literature on this topic, see for example: Alden and Lin, 2004 “Benchmarking the Characteristics of a World-Class University: Developing an International Strategy at University Level.” Leadership Foundation for Higher Education; Salmi, 2009 The Challenge of Establishing World-Class Universities; Heyneman and Lee 2012 “World Class Universities: the Sector Requirements”, in: Institutionalization of World Class Universities in Global Competition. 75 Salmi, 2009, op. cit. 76 See Aghion et al. 2007, Higher Aspirations: An Agenda for Reforming European Universities.” Bruegel Blueprint Series 5 and Aghion 2010 L'excellence universitaire: leçons des expériences internationales for the impact of governance and notably the strong correlation between university autonomy and research performance. This is a topic that we do not raise in this report. 77 For example: Salmi, 2009, op. cit. “the strict separation between the research institutes affiliated with the Centre National de la Recherche Scientifique (the National Center for Scientific Research, or CNRS) and the research departments of the universities results in the dispersion of human and financial resources”.

61

Finally, the IDEX endowment has helped diversify funding and encouraged universities to implement new fundraising approaches even if most universities remain severally constrained and, when compared to the leading Anglo-Saxon institutions, clearly underfunded.

However, the reforms have not managed to tackle the core issue of differentiation within the French system of HE&R. In particular, none of the main Parisian ComUEs comes close to fulfilling “the first and perhaps foremost determinant of excellence [which] is the presence of a critical mass of top students and outstanding faculty” or, more precisely78 of:  Having an international reputation for its teaching;  Attracting the most able students and producing the best graduates79;  Being able to recruit students from an international market;  Attracting a high proportion of postgraduate students;  Attracting a high proportion of students from overseas.

In this sense, the situation remains similar to 2004, when Orivi wrote the first French report on the ARWU ranking. He underlined at the time the dichotomy between grande écoles and universités: “Si l’on voulait donner à l’enseignement supérieur français quelque chance de devenir compétitif au niveau international, on voit bien que seule une fusion des avantages de certaines grandes écoles et de certaines universités pourrait y parvenir”. Although, the reforms of French HE&R had not yet been launched he perfectly foresaw the problems that institutions such as Paris-Saclay currently face: “Mais même si on obtenait ces financements additionnels, serait-on assuré du succès? Non, car rien ne prouve que les meilleurs étudiants cesseront de fréquenter les classes préparatoires aux grandes écoles, puis les grandes écoles elles-mêmes. Il faudrait d’abord qu’ils soient convaincus que l’excellence sera au rendezvous, et il faudrait aussi que les avantages qu’ils trouvent au système actuel disparaissent. Or ils ne disparaîtront pas spontanément, et si les meilleurs étudiants ne font pas le premier pas pour peupler les futures universités d’excellence, celles-ci risquent d’attendre longtemps avant d’y parvenir.”

The challenges ahead

The reforms launched in France over the last 10 years have transformed the French landscape of HE&R80. However, they have not yet created the conditions for World-Class Universities to emerge. Grandes écoles, universités and even ComUE are all institutions of higher education and research, but none can aspire to become either “world-class research universities” or “leading teaching universities” without undergoing considerable further transformation. Seen from Shanghai or Boston, they remain idiosyncratically French.

The ComUE have played a key role as facilitating structures. They have enabled the process of institutional transformation to start by creating a space for institutional dialogue between research

78 Salmi, 2009, op. cit. and for a detailed list Alden and Lin, 2004, op. cit. 79 EPFL is sometimes highlighted as a non-selective university at the undergraduate level. However, only students with the maturité gymnasiale are automatically enrolled, whereas requirements for foreign students are high (Baccalaureat S and an average of 16 for French students and / or an exam). Most important of all, in Switzerland there are no selective grandes écoles that compete for the best students. 80 Amongst other things, they have increased institutional autonomy, enabled the successful merger of universities that had been broken apart in 1969, transformed the role of national research organisms, increased the proportion of competitive funding for research and created a national evaluation agency.

62 organisations, Grandes écoles and universités. There is however a danger that they lose their focus and become institutional goals per se. This would defeat the aim for which they were set up, namely to coordinate the activities of institutions of higher education and research in a given territory. Indeed, by comparing the performances of individual institutions with those of the ComUE of which they are part, the report shows that the institutionalisation of ComUE would actually increase the differences between the landscapes of Paris, London and Boston (notably in terms of student distribution). Furthermore, the resultant institutions would be less competitive than the leading existing institutions on a number of important criteria, such as students per staff, impact per publication or publications per staff.

In this context, the following paragraph of J. Salmi’s report on The Challenges of Establishing World- Class Universities81, is striking: “It is difficult to maintain high selectivity in institutions with rapidly growing student enrolment and fairly open admission policies. The huge size of the leading universities of American countries such as Mexico or Argentina is certainly a major factor in explaining why these universities have failed to enter the top league, despite having a few excellent departments and research centers that are undoubtedly world-class. At the other extreme, University maintained its overall enrolment at less than 20,000”.

Viewing the ComUE in this way makes it easier to refocus on the aims of the IDEX and the supposed “expectations” of the international jury. It highlights the fact that the problem is not to “institutionalise the ComUE” but to create institutions that are able to tackle key issues that hold back would-be world-class universities such as selection, funding and governance. The problem is one of mission and vision.

In cases such as Bordeaux, , Aix-Marseille, Strasbourg or Sorbonne University, mergers make sense because they reunite what were de-facto faculties within a comprehensive university. The merger enables the (re)creation of a comprehensive university on a model well recognised all over the world. For an institution such as Sorbonne University, it is the first step on a path that could lead to being a top 20 world-class university. However, the merger alone will not enable Sorbonne University to compete with UCL, Imperial, Harvard or MIT. It is the start of a deep process of transformation that will require strong internal and external support.

In more complex situations such as Paris Saclay, PSL, Sorbonne Paris-Cité or the leading Grandes écoles and/or universités risk losing their reputation in favour of an institution that cannot aspire to the same level of recognition in its current form. In these cases, a merger of all institutions would be counterproductive. In order to move forward, these ComUE need to accept that they will never be world-class in their totality because becoming so would mean leaving large numbers of staff and student by the wayside. They need to identify and institutionalise a perimeter of excellence.

81 Salmi, 2009, op cit.

63

Clarifying and differentiating institutional mission

World-class universities cannot be created in isolation. They are necessarily part of a long-term vision for the university system as a whole. As Birnbaum underlines: “what we really need in countries everywhere are more world-class technical institutes, world-class community colleges, world-class colleges of agriculture, world-class teachers’ colleges, and world-class regional state universities. The United States don’t have a world-class higher education system because it has many world-class universities; instead it has world-class universities because it has a world-class higher education system”. (Birnbaum, 2007). In Paris, there has been a strong tendency to try and ensure either that no-one gets left behind and/or that all types of cursus are integrated into a single structure. This commendable aim is bound to failure because it is less competitive than comprehensive systems in which different institutions have clearly identified missions: Lasell College focuses on education for undergraduate students, Harvard on producing world-class research. The Parisian system hosts an incredibly diverse set of institutions from the purely legal perspective but it lacks diversity in terms of institutional missions. It desperately needs institutions to focus on becoming teaching universities, vocational institutions or regional universities. Because without this ecosystem, it will not be able to foster the emergence of 4 or 5 research universities and 2 or 3 world-class universities. The existing ComUE can help transform the landscape by encouraging internal differentiation between institutions: they should be concentrating their efforts on non-selective undergraduate programmes and the development of true vocational cursus. Only 2 or 3 Parisian institutions should aspire to be comprehensive world-class research universities. They need to focus exclusively on their core mission: being world-class in research and education means that most members of the academic staff have the potential to be in the top 10% in their field and most undergraduate students have the potential to do a PhD. They need greater autonomy to enable them to choose their mission and select their students and academic staff. Harvard, MIT, Imperial or UCL are part of well-oiled competitive systems of HE&R in which other institutions fulfil the roles of vocational institutions or regional state universities. Finally, a last striking fact when looking at the landscape of higher education and research of Boston and London is the diversity of great institutions. Alongside Harvard, MIT, UCL and Imperial there is room for excellent research universities such as King’s, Tufts or Boston University. There are remarkable small research institutions such as the London School of Hygiene and Tropical Medicine, Research Hospitals such as MGH, specialised institutions such as London School of Economics. And, of course, there are wonderful vocational institutions and teaching universities that do not feel the need to claim that they are doing cutting-edge research.

Creating coherent world-class institutions in Paris

The Parisian Idex have succeeded in launching remarkable initiatives, however they still need to solve two key problems. They lack the internal coherence to be immediately recognised as universities by external observers. And they lack that “little something”, the “dark matter” that marks the difference between most institutions whose position varies from ranking to ranking and the very few who are always at the top whatever the criteria selected.

64

Coherence and the institutional debate

Institutional coherence is sometimes considered synonymous with being recognised by ARWU, CWTS Leiden or THE. This approach leads to discussions on legal statuses, lobbying and attempts to define minimal conditions that all take up a large amount of energy and resources. None of these questions solves the underlying problem of institutional coherence. No two World-Class universities are structured alike. They combine departments, research institutes, governmental laboratories, faculties, student colleges and a vast array of other institutional objects in a myriad of forms. Furthermore, all these institutional objects are themselves subject to an even greater diversity of function and status. This is not a problem. However, no world-class university contains within it a structure that has prerogatives overlapping with those of the university itself. To give one much cited example, the problem of comparing Cambridge with French ComUE is not one of legal status (Colleges as listed bodies versus University as recognised body versus Schools, Departments and Faculties without legal status) but one of mission: the Colleges of Cambridge provide tutorship to students and are represented on some of the boards of the central university (4 seats on the 25 member university council), the university faculties organise teaching and research and are also represented on the board of the university. A conflict can arise between colleges or between faculties over resource allocation but no conflict can exist between them and the university on their mission.

The merger of Paris-Sorbonne and UPMC will recreate a coherent comprehensive research university at the heart of Paris. The respective responsibilities of the university, the faculties, laboratories, departments and other internal structures need to be defined but there is no conflict about mission. The cases of Sorbonne Paris-Cité, PSL or Saclay are more complex. Specialised institutions such as Institut d’Optique or Observatoire are not problematic: they are research institutes in a specific field that can easily find their place within the future university, whatever form it takes. On the other hand, comprehensive institutions cannot be members of a comprehensive institution, they can only exist within university systems (like UCL in University of London or UC Berkeley in University of California) or as the heart of the future university.

Density and “dark matter”

The term “dark matter” was coined by (Usher & Savino, 2007), who talk of the gravitational pull that certain institutions (Harvard, Oxford, etc.) or types of institutions exert on all rankings, regardless of the specific indicators and weightings used. The idea of “dark matter” harks back to a much maligned concept introduced in the first IDEX calls: the famous périmètre d’excellence, but a périmètre d’excellence whose limits correspond to those of the institution. The leading institutions of Boston and London are dense: they are leaders in all fields, their students are the best students and the percentage of highly cited publications of their researchers is extremely high. To be leaders they have to ensure that the level of excellence is homogeneous. They cannot afford to retain laboratories that are not leaders in their field or deliver diplomas to students who are unlikely to be accepted by a Master programme at one of their peer institutions.

Sorbonne University will not immediately acquire this “dark matter”, however the fundamentals are in place and the core institutions are comparable. Sorbonne University can succeed with a strong internal strategy of excellence and strong governmental support to enable it to compete on an equal footing with the world’s elite institutions (notably in terms of student attractiveness).

65

The situation of PSL, Saclay or Sorbonne Paris-Cité is, once again, more complex because the teaching and research excellence of the institutions that integrate the project are not homogeneous. They can lead the world but they need to rethink their structure and define conditions modelled on those of the institutions to whose status they aspire.

The Path ahead

There is no reason for which the Parisian system should not host world-class universities: it already hosts some of the most highly selective institutions in the world, as well as some of the best research groups. It not only publishes a comparable quantity of articles to Boston or London but hosts a higher percentage of graduate to undergraduate students than either of these two cities (37%, 32% and 29%). Despite the dichotomy between grandes écoles and universités, despite the confusion of roles between excellence in research and non-selective undergraduate education, Parisian institutions still succeed in appearing in the top 100 in the world. Furthermore, the CNRS, the leading national research organisation in the world, is fully integrated at the heart of all the important research-based institutions and is already playing a key role in terms of differentiation of the Parisian landscape of HE&R.

If the main objective is to consolidate 2 to 3 world-class universities and 4 to 5 research universities, then the path ahead is simple: enable a differentiation in terms of mission, create good teaching colleges, encourage the diversity of the landscape of HE&R and let the flagship universities concentrate on their mission of attracting the best students and scholars and producing world-class research.

66

Bibliography (main references)

Alden, J., and G. Lin. 2004. “Benchmarking the Characteristics of a World-Class University: Developing an International Strategy at University Level.” Leadership Foundation for Higher Education, London.

Aghion, P., M. Dewatripont, C. Hoxby, A. Mas-Colell, and A. Sapir. 2007. “Higher Aspirations: An Agenda for Reforming European Universities.” Bruegel Blueprint Series 5, Brussels. Retrieved July 11th, 2016, from http://bruegel.org/wp-content/uploads/imported/publications/BPJULY2008 University.pdf

Aghion, P. 2010. L'excellence universitaire : leçons des expériences internationales - Rapport d'étape, MESR. Retrieved July 11th, 2016, from http://www.ladocumentationfrancaise.fr/var/storage/ rapports-publics/104000043.pdf

Altbach, Philip G. 2004. “The Costs and Benefits of World-Class Universities.” Academe 90 (1): 20.

Boliver, V. 2015. “Are there distinctive clusters of higher and lower status universities in the UK?”, Oxford Review of Education, 41.5.

Bonitz, M., Bruckner, E. & Scharnhorst, A. 1997. Characteristics and impact of the Matthew effect for countries, Scientometrics, 40(3): 407-422.

Brezis and Hellier. 2013. Social mobility at the top: Why are elites self-reproducing?, ECINEQ Working paper 312. Retrieved July 11th, 2016, from http://www.ecineq.org/milano/WP/ECINEQ2013-312.pdf

Docampo, D., Egret, D. and Cram, L. 2014. French COMUEs and the Shanghai Ranking. DEC Technical Report. Retrieved July 11th, 2016, from https:// www.researchgate.net/publication/266557914 _French_COMUEs_and_the_Shanghai_Ranking

Hazelkorn: E. 2013. World-Class Universities or World-Class Systems: Rankings and Higher Education Policy Choices', in: E. Hazelkorn, P. Wells, M. Marope (eds), Rankings and accountability in higher education: uses and misuses, 71-94, Paris: UNESCO

Heyneman, S.P. and Lee, J. 2012. “World Class Universities: the Sector Requirements”, in Shin J.C. and Kehm, B. (eds.) Institutionalization of World Class Universities in Global Competition New York: Springer Publishers, 45-58.

Juppé, A. and Rocard, M. 2009. Investir pour l’avenir, Priorités stratégique d’investissement et investissement national. Retrieved July 11th, 2016, from www.gouvernement.fr/sites/default /files/contenu/piece-jointe/2014/08/rapport_juppe_rocard.pdf

Larivière, V. and Gingras, Y. 2010. The impact factor’s Matthew effect: a natural experiment in bibliometrics, Journal of the American Society for Information Science and Technology, 61/2, 424- 427, 2010.

Liu, N. C., and Cheng, Y. 2005. The Academic Ranking of World Universities. Higher Education in Europe, 30(2), 127–136.

67

Medoff, M.H. 2006. Evidence of a Harvard and Chicago Matthew effect. Journal of Economic Methodology, 13(4): 485-506.

Merton, R. K. 1968. The Matthew effect in science. Science, 159 (3810): 56-63.

Merton, R. K. 1973. The Sociology of Science: Theoretical and Empirical Investigations, Univ. of Chicago Press, Chicago/London

Merton, R. K. 1996. On Social Structure and Science, Univ. of Chicago Press, Chicago/London.

Nowotny, H. 2005. “Humanities in European Research”, IWM-Post, 89, 28-31

Orivel, F. 2004. “Pourquoi les universités françaises sont-elles si mal classées dans les palmarès internationaux?” Notes de l’IREDU, Dijon. Retrieved July 11th, 2016, from: http://iredu.u- bourgogne.fr/images/stories/Documents/Publications_iredu/Notes_Iredu/note044.pdf

Rafols, I, Porter, A. L. and Leydesdorff, L. 2010. “Science overlay maps : a new tool for research policy and library management”, Journal of the American Society for Information Science & Technology 61 (9), 1871-1887. (with annexes at www.leydesdorff.net/overlaytoolkit/)

Salmi, J. 2009. The challenge of establishing world-class universities. Washington, D.C.: The World Bank. Retrieved July 11th, 2016, from: http://siteresources.worldbank.org/EDUCATION/Resources /278200-1099079877269/547664-1099079956815/547670-1237305262556/WCU.pdf

Sarewitz, D. 2007. “Does Science Policy Matter?”, Issues in Science and Technology 23 (4).

Smaglik, P. 2005. “Golden Opportunities”, Nature 436, 144-147.

Tight, M. 1988. “Institutional typologies”. Higher Education Review 20, 27–51.

Tight, M. 1996. “Institutional typologies re-examined”. Higher Education Review 29, 57–77.

Werner, R. 2015. “The focus on bibliometrics makes papers less useful”, Nature 517, 245.

68

Appendices

Appendix A: Remark on the perimeter of the three areas

Defining the limits of Boston, London or Paris for the sake of the analysis and comparison between datasets can be a challenge. Each time that the data were available, we chose to focus on the metropolitan area which focuses on the functional integration. When data were not available at this level, then we focused either on city-based data, regional data or national data.

For Paris, Île-de-France and the Paris Metropolitan Area, which mostly converge have been used. For Boston, we used as much as possible the Boston Combined Statistical Area, that extends beyond the frontiers of Massachusetts to the south82, and for London, the London Metropolitan Area or Southeastern Metropolitan Area. We also considered other area definitions where data was available, that we could not retrieve for the precise region considered (Massachusetts state sources, etc.), if the context rendered a comparative imprecision acceptable.

This difficulty also extended to the set of institutions of HE&R to include. For example, the Université de Technologie de Compiègne is situated in the Hauts-de-France Region, but is part of the ComUE Sorbonne Universités and is more closely linked to the greater Parisian system of HE&R than to that of Lille.

Due to time constraints, we were not able to systemise the datasets to include these institutions that are located outside political boundaries, however we do not think that this has a major impact on the conclusions reached since the main and largest institutions are within the perimeters studied.

Appendix B: The Carnegie Classification

The analysis relied in part on the Carnegie Classification, designed to categorise accredited degree- granting HE-institutions in the USA.

Its basic classification distinguishes 7 basic types of institutions detailed in a number of subcategories. There are five further classification levels that precise the basic one, but for our purposes of comparison, we can limit our considerations to the basic level and retain the following aspects of this classification as pertinent. Our presentation is therefore not entirely faithful to the Carnegie classification, but already adapted.

1. Doctoral Universities83  R1 (115): Doctoral Universities – Highest research activity  R2 (107): Doctoral Universities – Higher research activity  R3 (113): Doctoral Universities – Moderate research activity

82 The metropolitan area of Boston is statistically much less well documented than its French and English counterparts. For our demographic analysis, we will therefore rely on a smaller unity of statistical description for Boston comprising just short of 5M inhabitants. In some cases, it is necessary to use state statistics for the whole of Massachusetts. 83 Includes institutions that awarded at least 20 research/scholarship doctoral degrees excluding professional practice doctoral-level degrees, such as the JD, MD, PharmD, DPT, etc. Excludes Special Focus Institutions.

69

2. Master's Colleges and Universities84  M1 (393): Master's Colleges and Universities – Larger programs  M2 (207): Master's Colleges and Universities – Medium programs  M3 (141): Master's Colleges and Universities – Smaller programs

3. Baccalaureate Colleges85  B1 (295): Baccalaureate Colleges: Arts & Sciences Focus  B2 (324): Baccalaureate Colleges: Diverse Fields

4. Special Focus Institutions86  ASF: Two-Year (Health Professions, Technical Professions, Arts & Design, Other Fields)  BSF: Four-Year (Faith-Related Institutions (Medical Schools & Centres, Other Health Professions Schools, Engineering Schools, Other Technology-Related Schools, Business & Management Schools, Arts, Music & Design Schools, Law Schools, Other Special Focus Institutions)

5. Baccalaureate/Associate's Colleges87  B/A1 (259): Baccalaureate/Associate's Colleges: Mixed Baccalaureate/Associate's Colleges  B/A2 (342): Baccalaureate/Associate's Colleges: Associate's Dominant

6. Associate's Colleges (A)88

84 Includes institutions that awarded at least 50 master's degrees and fewer than 20 doctoral degrees. Excludes Special Focus Institutions. 85 Includes institutions where baccalaureate or higher degrees represent at least 50 percent of all degrees but where fewer than 50 master's degrees or 20 doctoral degrees were awarded. Excludes Special Focus Institutions. 86 1-3 or 4-year degrees, where a high concentration of degrees is in a single field or set of related fields. 87 Includes four-year colleges (by virtue of having at least one baccalaureate degree program) that conferred more than 50 percent of degrees at the associate's level. Excludes Special Focus Institutions. 88 Institutions at which the highest level degree awarded is an associate's degree. Excludes Special Focus Institutions.

70

Appendix C: Vikky Bolivers clusters of UK universities by variable

Figure 37 source: Vikki Boliver, “Are there distinctive clusters of higher and lower status universities in the UK?”

71

Appendix D: Overview of the main classifications

1 year 2 years 3 years 4 years 5 years 6 years 7 years 8 years The Carnegie Classification (a classification in functional terms) Doctoral University More than 20 doctorate programmes Master's college and University Focus on Master, sometimes some PhD programmes Baccalaureate Colleges Main activity focused on Baccalaureate, but sometimes a few Master programmes Special Focus Colleges 2- and 4-year Special Focus Colleges exist Baccalaureate/Associate Colleges Awards associate and Bachelor degrees Associate Colleges Awards almost exclusively Associate degrees

The French System Université Classe préparatoire CP do not award proper HE degrees. They provide ECTS convertible into a Bachelor Grande École GEs award mostly specialised diplomas such as engineering diploma, some also now award masters

The English System University All universities have Further Education College89 a significant number of doctoral students

The American System (in legal terms) University

Liberal Arts College Senior College (some CCs) Community College In Massachusetts, "University" does not necessarily include doctorate programmes

89 These colleges offer a range of degrees and trainings, not only, but mostly, vocational. They serve around half a Million students in London, but they are addressed at a public from 16 to 60. They have a great role in lifelong learning, but in the end, only 13.000 students study in actual HE programmes. Therefore, the Colleges are usually omitted from the English HE system in a stricter sense. www.londoncolleges.com/en/Colleges_in_London/Facts_London_Colleges.cfm 72

Appendix E: Web of Science queries used in the report.

Web of Science queries for the three regions

As stated, for method 1 (procedure by list of institutions), we created an InCites dashboard with all the institutions registered in Île-de-France. The following queries on the contrary, concern method 2 (procedure by municipalities).

All the queries were executed on the 23rd of June 2016.

Indexes=SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, ESCI

Timespan=2010-2016

Queries by Area

Île de France CI=(PARIS OR ACHÈRES-LA-FORÊT OR ANNET-SUR-MARNE OR ARBONNE-LA-FORÊT OR ARMENTIÈRES-EN-BRIE OR OR AVON OR BAGNEAUX-SUR-LOING OR BAILLY-ROMAINVILLIERS OR OR BLANDY-LES-TOURS OR BOIS LE ROI OR OR BOISSISE-LA-BERTRAN OR BOISSISE-LE-ROI OR BOISSY-LE-CH TEL OR BRAY-SUR-SEINE OR BRIE-COMTE-ROBERT OR BUSSY SAINT-GEORGES OR OR OR CHAMPEAUX OR CHAMPS-SUR-MARNE OR OR CHATENOY OR CHAUMES-EN-BRIE OR CHELLES OR CHESSY OR CLAYE-SOUILLY OR COLLÉGIEN OR COMBS-LA-VILLE OR OR CONCHES-SUR-GONDOIRE OR CONGIS-SUR-THÉROUANNE OR OR COUILLY PONT AUX DAMES OR OR COULOMMIERS OR OR OR CRÉCY-LA-CHAPELLE OR CRÉGY-LES- OR CROISSY-BEAUBOURG OR OR CUISY OR DAMMARIE-LÈS-LYS OR DAMMARTIN-EN-GOËLE OR DAMMARTIN-SUR- OR DONNEMARIE-DONTILLY OR ECHOUBOULAINS OR ÉGREVILLE OR ÉMERAINVILLE OR ÉPINAY-SUR-SEINE OR OR ETREPILLY OR EVERLY OR OR FÉROLLES-ATTILY OR FERRIÈRES-EN-BRIE OR OR FONTENAY-TRÉSIGNY OR OR GRETZ- ARMAINVILLIERS OR GUÉRARD OR OR OR OR JOUY-LE-CH TEL OR LA CHAPELLE-GAUTHIER OR LA CHAPELLE-RABLAIS OR LA FERTÉ-GAUCHER OR LA FERTÉ-SOUS-JOUARRE OR LA HOUSSAYE OR LA ROCHETTE OR LAGNY-SUR-MARNE OR OR LE MÉE-SUR-SEINE OR LE PLESSIS- L'EVÊQUE OR LE PLESSIS-TRÉVISE OR LE VAUDOUÉ EN GATINAIS OR LÉSIGNY OR LIZY-SUR-OURCQ OR OR OR MAGNY-LE-HONGRE OR OR MAISON-ROUGE EN BRIE OR MAREUIL-LÈS-MEAUX OR MARLES EN BRIE OR MARNE-LA-VALLÉE OR MAROLLES EN BRIE OR MARY-SUR-MARNE OR MEAUX OR MITRY-MORY OR MOISSY-CRAMAYEL OR MONTEREAU-FAULT- OR MONTÉVRAIN OR MONTGÉ-EN-GOËLE OR OR MORET-SUR-LOING OR MOUSSY-LE-NEUF OR MOUSSY-LE-VIEUX OR NANDY OR OR NANTEUIL-LÈS-MEAUX OR OR OR NOYEN-SUR-SEINE OR OR OZOIR-LA-FERRIÈRE OR OZOUER-LE-VOULGIS OR PÉCY OR POLIGNY OR OR PONTAULT- COMBAULT OR PONTCARRÉ OR PRINGY OR OR QUINCY-VOISINS OR

73

REBAIS OR ROISSY-EN-BRIE OR SAINT-CYR-SUR-MORIN OR SAINT-FARGEAU- PONTHIERRY OR SAINT-GERMAIN LAVAL OR SAINT-GERMAIN-SUR-MORIN OR SAINT-HILLIERS OR SAINT-JEAN-LES-DEUX-JUMEAUX OR SAINT-MAMMÈS OR SAINT-MARTIN-DU-BOSCHET OR SAINT-PATHUS OR SAINT-PIERRE-LÈS-NEMOURS OR SAINT-SAUVEUR-SUR-ÉCOLE OR SAINT-SOUPPLETS OR SAINT-THIBAULT-DES- VIGNES OR SAMOIS-SUR-SEINE OR OR SAVIGNY-LE-TEMPLE OR SÉNART OR SERRIS OR SERVON OR SIGNY-SIGNETS OR SOISY-BOUY et MONTRAMÉ OR OR SOUPPES-SUR-LOING OR OR OR THORIGNY-SUR-MARNE OR TORCY OR TOURNAN-EN-BRIE OR OR USSY- SUR-MARNE OR VAIRES-SUR-MARNE OR VANVILLE OR VAUX-LE-PÉNIL OR OR VERNOU LA CELLE OR VERT-SAINT-DENIS OR OR VILLEPINTE OR VILLEVAUDÉ OR VILLIERS-SAINT-GEORGES OR VILLIERS-SUR- MORIN OR VILLIERS-SUR-SEINE OR OR VULAINES-SUR-SEINE OR YÈBLES OR OR ACHÈRES OR ANDELU OR ANDRESY OR OR AULNAY- SUR-MAULDRE OR AUTEUIL-LE-ROI OR BAILLY OR OR BAZOCHES-SUR- GUYONNE OR OR BEYNES OR BOIS D'ARCY OR BONNIÈRES-SUR- SEINE OR BOUAFLE OR OR BRÉVAL OR BRUEIL-EN-VEXIN OR BUC OR OR CARRIÈRES-SOUS- OR CARRIÈRES-SUR-SEINE OR CERNAY- LA-VILLE OR OR OR CH TEAUFORT OR CHANTELOUP-LES- VIGNES OR OR OR OR OR CONFLANS SAINTE-HONORINE OR CRESPIÈRES OR CROISSY-SUR-SEINE OR DAMPIERRE-EN- YVELINES OR OR OR ELANCOURT OR ÉPÔNE OR EVECQUEMONT OR OR OR FLINS-SUR-SEINE OR FONTENAY-LE-FLEURY OR GAILLON-SUR-MONTCIENT OR OR OR GARANCIÈRES OR OR GOMMECOURT-CLACHALOZE OR OR OR OR OR OR JOUARS-PONTCHARTRAIN OR JOUY-EN-JOSAS OR OR BOISSIÈRE-ÉCOLE OR CELLE-SAINT-CLOUD OR VERRIÈRE OR LANVILLE-EN-VEXIN OR CHESNAY OR MESNIL-LE-ROI OR MESNIL-SAINT-DENIS OR PECQ OR PERRAY-EN-YVELINES OR PORT-MARLY OR TREMBLAY-SUR-MAULDRE OR VÉSINET OR ALLUETS LE ROI OR CLAYES-SOUS-BOIS OR LES-ESSARTS-LE-ROI OR LOGES-EN-JOSAS OR VÉSINET OR MUREAUX OR LIMAY OR OR MAGNY-LES-HAMEAUX OR MAISONS- LAFFITTE OR MARCQ-EN-YVELINES OR MAREIL-LE-GUYON OR MAREIL-MARLY OR MAREIL-SUR-MAULDRE OR MARLY-LE-ROI OR MAULE OR OR MAUREPAS OR MEULAN-EN-YVELINES OR MÉZIÈRES-SUR-SEINE OR MEZY-SUR- SEINE OR MONTAINVILLE OR MONTALET-LE-BOIS OR OR MONTFORT- L'AMAURY OR MONTIGNY-LE-BRETONNEUX OR NEAUPHLE-LE-CH TEAU OR NÉZEL OR NOISY-LE-ROI OR OINVILLE-SUR-MONTCIENT OR OR ORGEVAL OR OR PLAISIR OR POISSY OR OR ROCQUENCOURT OR ROSNY- SUR-SEINE OR SAINT-CYR-L'ÉCOLE OR SAINT-GERMAIN-DE-LA-GRANGE OR SAINT-GERMAIN-EN-LAYE OR SAINT-LÉGER-EN-YVELINES OR SAINT-NOM-LA- BRETÈCHE OR SAINT-QUENTIN-EN-YVELINES OR SAINT-RÉMY-L-HONORÉ OR SAINT-RÉMY-LÈZ-CHEVREUSE OR OR SAULX-MARCHAIS OR OR OR TACOIGNIÈRES OR TESSANCOURT-SUR-AUBETTE OR THIVERVAL-GRIGNON OR TOUSSUS-LE-NOBLE OR -EN-YVELINES OR TRIEL-SUR-SEINE OR VAUX-SUR-SEINE OR VÉLIZY-VILLACOUBLAY OR VERNEUIL-SUR-SEINE OR VERNOUILLET OR VERSAILLES OR VILLENNES-SUR- SEINE OR OR VILLIERS-LE-MAHIEU OR VILLIERS-SAINT-FRÉDÉRIC

74

OR OR VOISINS-LE-BRETONNEUX OR ANGERVILLE OR ARPAJON OR AVRAINVILLE OR ATHIS-MONS OR AUVERS-SAINT-GEORGES OR BALLAINVILLIERS OR BALLANCOURT OR BIÈVRES OR BONDOUFLE OR BOUSSY-SAINT-ANTOINE OR BOUTIGNY-SUR-ESSONNE OR BOUVILLE OR BRÉTIGNY-SUR-ORGE OR BREUILLET OR BREUX-JOUY OR BRIIS-SOUS-FORGES OR OR BRUYÈRES-LE-Ch TEL OR BURES-SUR-YVETTE OR CERNY OR OR CHEPTAINVILLE OR CHEVANNES OR CHILLY-MAZARIN OR CORBEIL OR COURCOURONNES OR CROSNE OR DOURDAN OR OR ÉPINAY-SOUS-SÉNART OR ÉPINAY-SUR-ORGE OR ÉTIOLLES OR ETRÉCHY OR ÉVRY OR FLEURY-MÉROGIS OR FONTENAY-LES-BRIIS OR FORGES-LES-BAINS OR GIF-SUR-YVETTE OR GOMETZ-LA-VILLE OR GOMETZ- LE-CH TEL OR IGNY OR ITTEVILLE OR JANVRY OR JUVISY-SUR-ORGE OR LA FERTÉ-ALAIS OR LA VILLE-DU-BOIS OR LE COUDRAY-MONTCEAUX OR OR LEUVILLE-SUR-ORGE OR LIMOURS OR LINAS OR LISSES OR OR LONGPONT-SUR-ORGE OR MAISSE OR OR MAROLLES-EN-HUREPOIX OR MASSY OR MAUCHAMPS OR MENNECY OR MILLY-LA-FORÊT OR MOIGNY-SUR-ÉCOLE OR OR MONTLHÉRY OR MORANGIS OR MORIGNY-CHAMPIGNY OR MORSANG-SUR-ORGE OR NAINVILLE-LES-ROCHES OR NOZAY OR OLLAINVILLE OR ORMOY OR OR OR PARAY-VIEILLE-POSTE OR OR QUINCY-SOUS-SÉNART OR RIS-ORANGIS OR ROINVILLE-SOUS-DOURDAN OR SAINT-AUBIN OR SAINT-CHÉRON OR SAINT-GERMAIN-LÈS-ARPAJON OR SAINT- GERMAIN-LÈS-CORBEIL OR SAINT-JEAN-DE-BEAUREGARD OR SAINT-MICHEL-SUR- ORGE OR SAINT-PIERRE-DU-PERRAY OR SAINTE-GENEVIÈVE-DES-BOIS OR SAULT LES CHARTREUX OR SAVIGNY-SUR-ORGE OR SOISY-SUR-SEINE OR TIGERY OR VAUGRIGNEUSE OR VAUHALLAN OR VERRIÈRES-LE-BUISSON OR VERT-LE-GRAND OR VIGNEUX-SUR-SEINE OR VILLABÉ OR VILLEBON-SUR-YVETTE OR VILLEMOISSON-SUR-ORGE OR VILLIERS-LE-B CLE OR VILLIERS-SUR-ORGE OR VIRY-CHATILLON OR WISSOUX OR OR ANTONY OR ASNIÈRES-SUR-SEINE OR BAGNEUX OR BOIS- OR BOULOGNE-BILLANCOURT OR BOURG-LA- REINE OR CH TENAY-MALABRY OR CH TILON OR OR OR CLICHY-LA-GARENNE OR COLOMBES OR OR OR OR ISSY-LES-MOULINEAUX OR LA GARENNE-COLOMBES OR LE PLESSIS-ROBINSON OR LEVALLOIS OR OR MARNES LA COQUETTE OR OR OR NANTERRE OR NEUILLY-SUR-SEINE OR OR RUEIL-MALMAISON OR SAINT-CLOUD OR SCEAUX OR SÈVRES OR OR OR OR VILLE-d'AVRAY OR VILLENEUVE-LA-GARENNE OR WISSOUS OR OR AULNAY-SOUS-BOIS OR OR OR OR CLICHY-SOUS-BOIS OR COUBRON OR OR ÉPINAY-SUR- SEINE OR OR GOURNAY-SUR-MARNE OR L'ÎLE-SAINT-DENIS OR OR LE BLANC-MESNIL OR OR OR LES PAVILLONS-SOUS-BOIS OR LIVRY-GARGAN OR OR MONTREUIL OR NEUILLY-PLAISANCE OR NOISY-LE-GRAND OR NOISY-LE-SEC OR OR PIERREFITTE-SUR-SEINE OR PRÉ SAINT-GERVAIS OR OR ROSNY- SOUS-BOIS OR SAINT-DENIS OR SAINT-OUEN OR OR STAINS OR TREMBLAY-EN-FRANCE OR OR VILLEPINTE OR OR ABLON SUR SEINE OR OR OR BOISSY-SAINT-LÉGER OR BONNEUIL-SUR-MARNE OR BRY-SUR-MARNE OR CACHAN OR CHAMPIGNY-SUR-MARNE OR CHARENTON-LE-PONT OR CHENNEVIÈRES-SUR-MARNE OR CHEVILLY-LARUE OR CRÉTEIL OR CRÉTEIL OR CROSNE OR FONTENAY-SOUS-BOIS OR FRESNES OR

75

GENTILLY OR IVRY-SUR-SEINE OR JOINVILLE-LE-PONT OR KREMLIN-BICÊTRE OR L'HAY-LES-ROSES OR LA QUEUE-EN-BRIE OR LE-PERREUX-SUR-MARNE OR LIMEIL-BRÉVANNES OR MAISON-ALFORT OR NOGENT-SUR-MARNE OR OR ORMESSON-SUR-MARNE OR PÉRIGNY-SUR-YERRES OR QUINCY-VOISINS OR OR SAINT-MANDÉ OR SAINT-MAUR-DES-FOSSÉS OR SAINT-MAURICE OR OR SUCY-EN-BRIE OR OR OR OR VILLENEUVE-LE-ROI OR VILLENEUVE-SAINT-GEORGES OR VILLIERS-SUR-MARNE OR VINCENNES OR VITRY-SUR-SEINE OR OR AINCOURT OR AMBLEVILLE OR ANDILLY OR OR ARNOUVILLE-LES- OR ARRONVILLE OR ASNIÈRES-SUR- OR AUVERS-SUR-OISE OR BAILLET EN FRANCE OR BEAUCHAMP OR BEAUMONT-SUR-OISE OR BÉTHEMONT-LA-FORÊT OR BELLOY-EN-FRANCE OR BERVILLE OR BESSANCOURT OR OR CERGY OR CHAMPAGNE-SUR-OISE OR CHAUVRY OR CORMEILLES-EN-PARISIS OR CORMEILLES-EN-VEXIN OR DEUIL-LA-BARRE OR DOMONT OR OR ECOUEN OR ENGHIEN-LES-BAINS OR ENNERY OR ÉPIAIS-RHUS OR ÉRAGNY-SUR- OISE OR OR ÉZANVILLE OR FOSSES OR FRANCONVILLE OR GARGES-LÈS- GONESSE OR GOUSSAINVILLE OR GROSLAY OR HERBLAY OR JOUY-LE-MOUTIER OR L'ILE-ADAM OR FRETTE-SUR-SEINE OR ROCHE-GUYON OR PLESSIS-BOUCHARD OR THILLAY OR LUZARCHES OR MAGNY-EN-VEXIN OR MARLY-LA-VILLE OR MENOUVILLE OR MERIEL OR MÉRY-SUR-OISE OR MONTLIGNON OR MONTIGNY-LÈS- CORMEILLES OR MONTMAGNY OR MONTMORENCY OR MOUSSY OR MOURS OR NERVILLE-LA-FORÊT OR NEUVILLE-SUR-OISE OR OR PARMAIN OR PARMAIN-JOUY-LE-COMTE OR PIERRELAYE OR PISCOP OR PONTOISE OR PONTOISE OR PRESLES OR SAINT-BRICE-SOUS-FORÊT OR SAINT-GRATIEN OR SAINT-LEU-LA-FORÊT OR SAINT MARTIN DU TERTRE OR SAINT-OUEN-L'AUMÔNE OR SAINT-PRIX OR OR OR SOISY-SOUS-MONTMORENCY OR OR VALLANGOUJARD OR VAURÉAL OR VEMARS OR VILLIERS-ADAM OR VILLIERS-LE-BEL)

London CI= (London OR Camden OR "City of London" OR Westminster OR Hammersmith OR Fulham OR Kensington OR Chelsea OR Wandsworth OR Hackney OR Newham OR "Tower Hamlets" OR Haringey OR Islington OR Lewisham OR Southwark OR Lambeth OR Bexley OR Greenwich OR "Barking & Dagenham" OR Havering OR Redbridge OR Waltham Forest OR Enfield OR Bromley OR Croydon OR Merton OR "Kingston upon Thames" OR Sutton OR Barnet OR Brent OR Ealing OR Harrow OR Hillingdon OR Hounslow OR "Richmond upon Thames") AND CU=England

Boston CI= (Abington OR "Amesbury Town" OR Andover OR Arlington OR Ayer OR Bellingham OR Belmont OR Beverly OR Boston OR Boxford OR "Braintree Town" OR Bridgewater OR Brockton OR Brookline OR Burlington OR Cambridge OR Chelsea OR Cochituate OR Danvers OR Dedham OR Derry OR Dover OR Dover OR Durham OR Duxbury OR "East Pepperell" OR Epping OR Essex OR Everett OR Exeter OR Farmington OR Foxborough OR Framingham OR Franklin Town OR Gloucester OR "Green Harbor-Cedar Crest" OR Groton OR Hampton OR "Hampton Beach" OR Hanson OR Haverhill OR HinghamOR Holbrook OR Hopkinton OR Hudson OR Hull OR Ipswich OR

76

Kingston OR Lawrence OR Lexington OR Littleton Common OR Londonderry OR Lowell OR Lynn OR Lynnfield OR Malden OR Marblehead OR Marion Center OR Marlborough OR Marshfield OR Marshfield Hills OR "Mattapoisett Center" OR Maynard OR Medfield OR Medford OR Melrose OR Methuen Town OR Middleborough Center OR "Millis-Clicquot" OR Milton OR Milton OR Milton Mills OR Nahant OR Needham OR Newburyport OR Newfields OR Newmarket OR Newton OR "North Lakeville" OR North Pembroke OR "North Plymouth" OR "North Scituate" OR Norwood OR "Ocean Bluff-Brant" Rock OR Onset OR Peabody OR Pepperell OR Pinehurst OR Plymouth OR Portsmouth OR Quincy OR Randolph OR Raymond OR Reading OR Cambridge OR Newton OR Revere OR Rochester OR Rockport OR Rowley OR Salem Salisbury OR Saugus OR Scituate OR "Seabrook Beach" OR Sharon OR Shirley OR Somersworth OR Somerville OR "South Duxbury" OR Southfield OR Stoneham OR Swampscott OR The Pinehills OR Topsfield OR Townsend OR Wakefield OR Walpole OR Waltham OR "Wareham Center" OR "Watertown Town" OR Wellesley OR "West Concord" OR "West Wareham" OR Weweantic OR "Weymouth Town" OR "White Island Shores" OR Wilmington OR Winchester OR Winthrop Town OR Woburn) AND CU=USA AND PS=MA

Web of Science queries by institutions Done using InCites.

77

Appendix F: data problems: Paris-Sorbonne according to Web of Science:

78

Appendix G: List of institutions per area

Paris

Institutions Type Legal ComUE UG PG Total Web of Category Times % Citation Status Science Normalize Cited Docs Impact Documen d Citation Cited ts Impact CEA Public Université Paris- 0 0 0 35737 1.65 434494 0.747 12.16 Saclay 3 CentraleSupélec Grand Public Université Paris- 0 4390 4390 4052 1.19 20584 59.25 5.08 établissement Saclay Collège de France Grand Public Université de 0 0 0 3832 1.65 46916 73.25 12.24 établissement recherche Paris sciences et lettres - PSL Research University Conservatoire national des arts et Grand Public Hautes Études- 0 0 0 0 0 0 0 0 métiers établissement Sorbonne-Arts et Métiers ECAM-EPMI École d'ingénieurs Private Université Paris- 0 0 0 0 0 0 0 0 Seine École de biologie industrielle École d'ingénieurs Private Université Paris- 0 0 0 0 0 0 0 0 Seine école des Hautes Etudes Autre Private Université Paris- 0 0 0 490 1.76 3089 67.55 6.3 Commerciales de Paris établissement Saclay École des hautes études en sciences Grand Public Université de 0 2618 2618 2073 0.99 5317 36.71 2.56 sociales établissement recherche Paris sciences et lettres - PSL Research University École d'ingénieur d'agro- École d'ingénieurs Private Université Paris- 0 0 0 0 0 0 0 0 développement international Seine École d'ingénieur de la ville de Paris Etablissement Public 0 0 0 0 0 0 0 0 relevant d'un

79

autre département ministériel rattaché à un EPSCP École d'ingénieur généraliste en École d'ingénieurs Private 0 0 0 0 0 0 0 0 informatique et technologies du numérique École d'ingénieurs du monde École d'ingénieurs Private 0 0 0 0 0 0 0 numérique École française d'Extrême-Orient École française à Public Université de 0 0 0 45 1.62 43 22.22 0.96 l'étranger recherche Paris sciences et lettres - PSL Research University École internationale des sciences du Établissements Private Institut 0 0 0 0 traitement de l'information d'enseignement polytechnique supérieur privés rattachés à un EPSCP École nationale des Chartes Grand Public Université de 0 114 114 19 0.11 5 10.53 0.26 établissement recherche Paris sciences et lettres - PSL Research University École nationale des ponts et Grand Public Université Paris- 0 1503 1.07 7768 66.4 5.17 chaussées établissement Est relevant d'un autre département ministériel École nationale supérieure d'arts et Grand Public Hautes Études- 0 1636 1636 0 0 0 0 métiers établissement Sorbonne-Arts et Métiers École nationale supérieure de Écoles nationales Public Université de 0 342 342 2097 1.36 24635 78.97 11.75 chimie de Paris supérieures recherche Paris d'ingénieurs sciences et lettres - PSL Research University

80

École nationale supérieure de Écoles habilitées à Public Institut 0 837 837 0 l'électronique et de ses applications délivrer un polytechnique de Cergy diplôme Grand Paris d'ingénieur École nationale supérieure Écoles nationales Public Université Paris- 0 567 567 6 0.78 21 83.33 3.5 d'informatique pour l'industrie et supérieures Saclay l'entreprise d'ingénieurs École nationale supérieure Louis Autre Public Université Paris 0 0 0 0 0 0 0 Lumière établissement Lumières École nationale supérieure maritime Grand Public 0 0 0 0 0 0 0 établissement relevant d'un autre département ministériel École nationale vétérinaire d'Alfort Etablissement Public Université Paris- 0 0 0 858 1.18 6238 69.7 7.27 relevant d'un Est autre département ministériel rattaché à un EPSCP École normale supérieure de Cachan École normale Public Université Paris- 0 2106 2106 2231 1.17 14288 66.38 6.4 supérieure Saclay École normale supérieure de Paris École normale Public Université de 0 1723 1723 6578 1.67 69357 72.77 10.54 supérieure recherche Paris sciences et lettres - PSL Research University École polytechnique École d'ingénieurs Public Université Paris- 0 2944 2944 7844 1.55 83139 74.71 10.6 Saclay École Polytechnique de l'Université École d'ingénieurs Public Université Paris- 0 0 Paris-Sud Saclay École pratique des hautes études Grand Public Université de 0 1208 1208 2178 1.33 19482 68.87 8.94 établissement recherche Paris sciences et lettres - PSL Research University École spéciale des travaux publics, Établissements Private 0 0 0 0 0 0 0

81 du bâtiment et de l'industrie d'enseignement supérieur privés rattachés à un EPSCP Ecole Superieure de Physique et de École d'ingénieurs Public Université de 0 0 0 0 0 0 0 Chimie Industrielles de la Ville de recherche Paris Paris sciences et lettres - PSL Research University École supérieure des sciences Autre Private Université Paris- 0 3196 3196 481 1.05 1862 48.65 3.87 économiques et commerciales établissement Seine École supérieure des techniques École d'ingénieurs Private 0 1800 1800 0 0 0 0 0 aéronautiques et de construction automobile École supérieure d'ingénieurs des École d'ingénieurs Private 0 320 320 0 0 0 0 0 travaux de la construction de Cachan ENSTA École d'ingénieurs Public Université Paris- 0 0 0 0 0 0 0 0 Saclay EPF - École d'ingénieur-e-s École d'ingénieurs Private 0 0 0 0 0 0 0 0 ESPE de l'académie de Créteil École supérieure Public 0 3714 3714 0 0 0 0 0 du professorat et de l'éducation ESPE de l'académie de Paris École supérieure Public 0 2931 2931 0 0 0 0 0 du professorat et de l'éducation ESPE de l'académie de Versailles École supérieure Public 0 4229 4229 0 0 0 0 0 du professorat et de l'éducation Facultés Libres de Philosophie et de Autre Private 0 0 0 0 0 0 0 0 Psychologie établissement Groupe des écoles nationales Grand Public Université Paris- 0 0 0 0 0 0 0 0 d'économie et statistique établissement Saclay relevant d'un autre département ministériel Institut agronomique, vétérinaire et Autre Public 0 0 0 0 0 0 0 0 forestier de France établissement

82

Institut catholique de Paris Établissement Private 0 0 0 0 0 0 0 0 privé d'enseignement universitaire Institut d'administration des Autre Public Hautes Études- 0 0 0 0 0 0 0 0 entreprises de Paris établissement Sorbonne-Arts et Métiers Institut de management et de Autre Private 0 0 0 0 0 0 0 0 communication interculturels établissement Institut de physique du globe Grand Public Université 0 82 82 0 0 0 0 0 établissement Sorbonne Paris Cité Institut des sciences et industries du Grand Public Université Paris- 0 2000 2000 3596 1.37 30615 77.31 8.51 vivant et de l'environnement établissement Saclay relevant d'un autre département ministériel Institut d'études politiques de Paris Grand Public Université 0 9054 9054 945 1.72 2169 37.99 2.3 établissement Sorbonne Paris Cité Institut d'Optique Graduate School Établissements Private Université Paris- 0 474 474 844 1.88 8930 70.62 10.58 d'enseignement Saclay supérieur privés rattachés à un EPSCP Institut national des langues et Grand Public Université 4966 2789 7755 0 0 0 0 0 civilisations orientales établissement Sorbonne Paris Cité Institut national d'histoire de l'art Grand Public Sorbonne 0 0 0 0 0 0 0 0 établissement Universités Institut national du sport, de Grand Public 0 0 0 0 0 0 0 0 l'expertise et de la performance établissement relevant d'un autre département ministériel Institut national supérieur de Autre Public Université Paris 0 0 0 0 0 0 0 0 formation et de recherche pour établissement Lumières

83 l'éducation des jeunes handicapés et les enseignements adaptés Institut protestant de théologie Établissement Private 0 0 0 0 0 0 0 0 privé d'enseignement universitaire Institut supérieur de mécanique de Instituts et écoles Public Institut 0 671 671 0 0 0 0 0 Paris extérieurs aux polytechnique universités Grand Paris Institut supérieur d'électronique de École d'ingénieurs Private 0 0 0 0 0 0 0 0 Paris Institute Curie Public Université de 0 0 0 5380 1.72 62951 61.77 11.7 recherche Paris sciences et lettres - PSL Research University Muséum national d'histoire Grand Public Sorbonne 0 361 361 7837 1.38 63529 74.43 8.11 naturelle établissement Universités Observatoire de Paris Grand Public Université de 0 102 102 5928 1.77 89815 78.58 15.15 établissement recherche Paris sciences et lettres - PSL Research University Telecom Paristech Grand Public Université Paris- 0 1500 1500 1874 1.7 13690 53.15 7.31 établissement Saclay relevant d'un autre département ministériel Telecom SudParis École d'ingénieurs Public Université Paris- 731 0.78 1262 42.68 Saclay % Université de Cergy-Pontoise Université Public Université Paris- 10666 2942 13608 2016 0.97 10962 62.7 5.44 Seine Université de Versailles Saint- Université Public Université Paris- 12256 4938 17194 6978 1.88 78017 72.6 11.18 Quentin-en-Yvelines Saclay Université d'Évry-Val d'Essonne Université Public Université Paris- 7226 2141 9367 1692 1.42 17484 68.38 10.33 Saclay Université Panthéon-Assas Université Public Sorbonne 12743 5690 18433 0 0 0 0 Universités

84

Université Paris 1 - Panthéon Université Public Hautes Études- 28057 14777 42834 1981 0.86 4956 42.1 2.5 Sorbonne Sorbonne-Arts et Métiers Université Paris 13 - Paris Nord Université Public Université 15891 5049 20940 5061 1.2 34159 59.95 6.75 Sorbonne Paris Cité Université Paris 8 - Vincennes - Université Public Université Paris 15576 7061 22637 1304 0.78 2214 32.59 1.7 Saint-Denis Lumières Université Paris Descartes Université Public Université 28192 6835 35027 23864 1.58 235823 65.92 9.88 Sorbonne Paris Cité Université Paris Diderot Université Public Université 18073 6357 24430 26357 1.82 330005 71.28 12.52 Sorbonne Paris Cité Université Paris Ouest Nanterre La Université Public Université Paris 24874 9145 34019 1497 0.74 3642 40.68 2.43 Défense Lumières Université Paris-Dauphine Université Public Université de 7957 2175 10132 2487 1.08 7432 52.19 2.99 recherche Paris sciences et lettres - PSL Research University Université Paris-Est Créteil Val-de- Université Public Université Paris- 21203 5584 26787 4003 1.61 28637 65.93 7.15 Marne Est Université Paris-Est Marne-la-Vallée Université Public Université Paris- 7593 3055 10648 1119 1.12 3916 60.86 3.5 Est Université Paris-Sorbonne Université Public Sorbonne 13120 6982 20102 7381 1.27 16045 43.92 2.17 Universités Université Paris-Sud Université Public Université Paris- 22254 8191 30445 27891 1.75 335850 72.85 12.04 Saclay Université Pierre et Marie Curie Université Public Sorbonne 23485 9614 33099 43383 1.57 480307 72.58 11.07 Universités Université Sorbonne Nouvelle - Université Public Université 11213 5909 17122 280 0.79 391 40 1.4 Paris 3 Sorbonne Paris Cité Total 285345 158153 443498 16457790 1.49 1,542,241 66.99 9.37 %

90 Total number of document is not the sum of all the docs per institutions, since some publications are done in collaborations between several institutions.

85

Source: Open Data Portal Enseignement Supérieur et Recherche File Principaux établissements d'enseignement supérieur and file Effectifs d'étudiants inscrits dans les établissements publics sous tutelle du ministère en charge de l'Enseignement supérieur et de la Recherche. Plus information from offical websites for number of students. Students 2014-2015. InCites 2010-2016, Access 2016-07-04

London

Institution Type Legal UG PG Total Web of Science Category Times % Docs Citation status Documents Normalized Cited Cited Impact Citation Impact Birkbeck College University Public - 9680 4255 13935 3428 1.42 24791 58.37 7.23 Charity Brunel University London University Public - 9695 4025 13720 7053 1.37 46552 58 6.6 Charity Buckinghamshire New University University Public - 7970 1065 9035 235 0.64 685 53.19 2.91 Charity Conservatoire for Dance and Drama University Public - 1070 190 1260 0 0 0 0 0 Charity Courtauld Institute of Art University Public - 165 305 470 0 0 0 0 0 Charity Goldsmiths College University Public - 5235 2930 8165 987 1.39 2551 38.5 2.58 Charity Guildhall School of Music and Drama University Public - 555 355 910 0 0 0 0 Charity Heythrop College University Public - 350 355 705 513 0.44 103 6.04 0.2 Charity Imperial College of Science, Technology University Public - 9015 7595 16610 59534 1.74 634101 64.63 10.65 and Medicine Charity King's College London University Public - 17610 11120 28730 42874 1.65 378736 56.96 8.83 Charity Kingston University University Public - 16975 4935 21910 2087 1.14 8266 52.04 3.96 Charity London Business School University Public - 0 1790 1790 733 1.95 5657 67.8 7.72 Charity

86

London Metropolitan University University Public - 11030 3055 14085 1635 0.87 5504 46.3 3.37 Charity London School of Economics and University Public - 4415 6185 10600 8687 1.74 36164 49.49 4.16 Political Science Charity London School of Hygiene and Tropical University Public - 0 1245 1245 12373 2.29 153006 71.37 12.37 Medicine Charity London South Bank University University Public - 13020 4715 17735 1325 0.76 3651 49.06 2.76 Charity Middlesex University University Public - 13675 3805 17480 2232 0.94 4940 40.55 2.21 Charity Queen Mary University of London University Public - 11390 4570 15960 19042 1.89 186570 59.8 9.8 Charity Ravensbourne University Public - 1985 70 2055 0 0 0 Charity Roehampton University University Public - 5985 1665 7650 1630 1.08 5125 41.66 3.14 Charity Rose Bruford College University Public - 730 60 790 0 0 0 Charity Royal Academy of Music University Public - 360 410 770 44 1.23 18 6.82 0.41 Charity Royal College of Art University Public - 0 1445 1445 228 1.58 175 19.3 0.77 Charity Royal College of Music University Public - 425 380 805 87 0.43 64 18.39 0.74 Charity Royal Holloway and Bedford New University Public - 7170 2745 9915 4534 1.71 53260 66.65 11.75 College Charity St George's Hospital Medical School University Public - 4575 930 5505 6954 1.8 73679 57.52 10.6 Charity St Mary's University, Twickenham University Public - 3905 1430 5335 157 0.74 259 40.76 1.65 Charity The City University University Public - 9695 8585 18280 4794 1.13 17902 53.44 3.73 Charity The Institute of Cancer Research University Public - 0 300 300 6696 2.81 106379 61.62 15.89 Charity The Royal Central School of Speech and University Public - 635 355 990 27 2.61 24 33.33 0.89 Drama† Charity The Royal Veterinary College University Public - 1635 510 2145 2994 1.44 18759 64.96 6.27 Charity

87

The School of Oriental and African University Public - 3015 2890 5905 2195 1.42 3314 27.93 1.51 Studies Charity The University of East London University Public - 11680 4175 15855 1411 0.88 3567 42.95 2.53 Charity The University of Greenwich University Public - 16105 5190 21295 2501 1.25 11085 57.26 4.43 Charity The University of West London University Public - 8980 1510 10490 95 0.71 438 61.05 4.61 Charity The University of Westminster University Public - 16145 4320 20465 2119 1.19 7267 45.73 3.43 Charity Trinity Laban Conservatoire of Music University Public - 695 270 965 0 0 0 0 0 and Dance Charity University College London† University Public - 16830 18785 35615 82401 1.75 763416 60.73 9.26 Charity University of the Arts, London University Public - 14505 3270 17775 412 1.03 297 22.33 0.72 Charity Regent's University London University Private 3591 0 3591 0 0 0 0 non-profit Total 260496 121790 382286 254,387 1.64 2,189,922 59.14% 8,61%

Source: Higher Education Funding Council for England, Students 2014-2015.

InCites 2010-2016, Access 2016-07-04

88

Boston

Institution Type Legal status UG PG Total Web of Category Times % Docs Citation students Science Normalized Cited Cited Impact Docume Citation nts Impact Andover Newton Theological School Special Focus Four- Private not-for- 0 260 260 0 0 0 0 0 Year profit Anna Maria College Master's Colleges & Private not-for- 1108 322 1430 0 0 0 0 0 Universities profit Babson College Special Focus Four- Private not-for- 2107 942 3049 0 0 0 0 0 Year profit Bay State College Special Focus Four- Private for- 1108 0 1108 0 0 0 0 0 Year profit Benjamin Franklin Institute of Special Focus Four- Private not-for- 507 0 507 0 0 0 0 0 Technology Year profit Bentley University Master's Colleges & Private not-for- 4264 1301 5565 0 0 0 0 0 Universities profit Berklee College of Music Special Focus Four- Private not-for- 4743 165 4908 0 0 0 0 0 Year profit Berkshire Community College Associate's Colleges Public 2230 0 2230 0 0 0 0 0 Boston Architectural College Special Focus Four- Private not-for- 470 406 876 0 0 0 0 0 Year profit Boston Baptist College Special Focus Four- Private not-for- 94 0 94 0 0 0 0 0 Year profit Boston College Doctoral Private not-for- 9856 4461 14317 5621 1.34 44090 52.23 7.84 Universities profit Boston Graduate School of Special Focus Four- Private not-for- 0 136 136 0 0 0 0 0 Psychoanalysis Inc Year profit Boston University Doctoral Private not-for- 18017 14095 32112 31800 1.85 362803 63.34 11.41 Universities profit Brandeis University Doctoral Private not-for- 3729 2216 5945 4239 1.93 49886 59.45 11.77 Universities profit Bridgewater State University Master's Colleges & Public 9628 1559 11187 435 1.07 940 31.72 2.16 Universities

89

Bunker Hill Community College Associate's Colleges Public 14253 0 14253 0 0 0 0 0 Cambridge College Master's Colleges & Private not-for- 1140 1701 2841 0 0 0 0 0 Universities profit Curry College Master's Colleges & Private not-for- 2900 241 3141 0 0 0 0 0 Universities profit Dartmouth College Doctoral Private not-for- 4289 2009 6298 14140 1.7 123328 60.8 8.72 Universities profit Eastern Nazarene College Master's Colleges & Private not-for- 1064 147 1211 0 0 0 0 0 Universities profit Emerson College Master's Colleges & Private not-for- 3787 758 4545 0 0 0 0 0 Universities profit Emmanuel College-Boston Baccalaureate Private not-for- 2082 229 2311 0 0 0 0 0 Colleges profit Endicott College Master's Colleges & Private not-for- 2877 1552 4429 0 0 0 0 0 Universities profit Episcopal Divinity School Special Focus Four- Private not-for- 0 62 62 0 0 0 0 0 Year profit Fisher College Special Focus Four- Private not-for- 1875 0 1875 0 0 0 0 0 Year profit Franklin W Olin College of Engineering Special Focus Four- Private not-for- 350 0 350 166 0.9 950 45.78 5.72 Year profit Great Bay Community College Associate's Colleges Public 2079 0 2079 0 0 0 0 0 Harvard University Doctoral Private not-for- 10338 18453 28791 180651 2.19 2310074 63.52 12.79 Universities profit Hebrew College Special Focus Four- Private not-for- 2 144 146 0 0 0 0 0 Year profit Hellenic College-Holy Cross Greek Special Focus Four- Private not-for- 81 102 183 0 0 0 0 0 Orthodox School of Theology Year profit Hult International Business School Special Focus Four- Private not-for- 0 814 814 0 0 0 0 0 Year profit ITT Technical Institute-Wilmington Special Focus Four- Private for- 347 0 347 0 0 0 0 0 Year profit Laboure College Special Focus Four- Private not-for- 757 0 757 0 0 0 0 0 Year profit Lasell College Master's Colleges & Private not-for- 1737 333 2070 0 0 0 0 0 Universities profit Lawrence Memorial Hospital School of Special Focus Two- Private not-for- 283 0 283 0 0 0 0 0 Nursing Year profit

90

Le Cordon Bleu College of Culinary Arts- Special Focus Two- Private for- 474 0 474 0 0 0 0 0 Cambridge Year profit Lesley University Doctoral Private not-for- 1929 2930 4859 0 0 0 0 0 Universities profit Longy School of Music of Bard College Special Focus Four- Private not-for- 59 189 248 0 0 0 0 0 Year profit Marian Court College Special Focus Two- Private not-for- 277 0 277 0 0 0 0 0 Year profit Massachusetts Bay Community College Associate's Colleges Public 5369 0 5369 0 0 0 0 0 Massachusetts College of Art and Special Focus Four- Public 1967 128 2095 17 1.14 17 29.41 1 Design Year Massachusetts Institute of Technology Doctoral Private not-for- 4512 6807 11319 47559 2.32 826259 70.56 17.37 Universities profit Massachusetts School of Law Special Focus Four- Private not-for- 0 408 408 0 0 0 0 0 Year profit Massasoit Community College Associate's Colleges Public 7905 0 7905 0 0 0 0 0 MCPHS University Special Focus Four- Private not-for- 4027 2908 6935 0 0 0 0 0 Year profit Merrimack College Master's Colleges & Private not-for- 3033 304 3337 0 0 0 0 0 Universities profit MGH Institute of Health Professions Special Focus Four- Private not-for- 195 971 1166 42316 2.43 580186 62.87 13.71 Year profit Middlesex Community College-Bedford Associate's Colleges Public 9205 0 9205 0 0 0 0 0 Montserrat College of Art Special Focus Four- Private not-for- 402 0 402 0 0 0 0 0 Year profit Mount Ida College Baccalaureate Private not-for- 1293 27 1320 0 0 0 0 0 Colleges profit New England College of Business and Special Focus Four- Private for- 690 151 841 0 0 0 0 0 Finance Year profit New England College of Optometry Special Focus Four- Private not-for- 0 519 519 0 0 0 0 0 Year profit New England School of Acupuncture Special Focus Four- Private not-for- 0 160 160 0 0 0 0 0 Year profit New England School of Law Special Focus Four- Private not-for- 0 871 871 0 0 0 0 0 Year profit Newbury College Baccalaureate Private not-for- 874 0 874 0 0 0 0 0 Colleges profit

91

North Shore Community College Associate's Colleges Public 7412 0 7412 0 0 0 0 0 Northeastern University Doctoral Private not-for- 13510 6288 19798 10382 1.64 85969 60.69 8.28 Universities profit Northeastern University Global Master's Colleges & Private not-for- 3689 6413 10102 0 0 0 0 0 Network Universities profit Northern Essex Community College Associate's Colleges Public 6963 0 6963 0 0 0 0 0 Northpoint Bible College Special Focus Four- Private not-for- 303 9 312 0 0 0 0 0 Year profit Pine Manor College Baccalaureate Private not-for- 403 39 442 0 0 0 0 0 Colleges profit Plymouth State University Master's Colleges & Public 3756 1099 4855 239 0.65 750 42.68 3.1 Universities Quincy College Associate's Colleges Public 4705 0 4705 0 0 0 0 0 Salem State University Master's Colleges & Public 7600 1667 9267 237 0.95 463 35.02 1.99 Universities Salve Regina University Master's Colleges & Private not-for- 2121 618 2739 0 0 0 0 0 Universities profit School of the Museum of Fine Arts- Special Focus Four- Private not-for- 385 163 548 0 0 0 0 0 Boston Year profit Simmons College Master's Colleges & Private not-for- 1622 3180 4802 0 0 0 0 0 Universities profit Suffolk University Doctoral Private not-for- 5390 2825 8215 0 0 0 0 0 Universities profit The Boston Conservatory Special Focus Four- Private not-for- 505 206 711 0 0 0 0 0 Year profit The New England Conservatory of Music Special Focus Four- Private not-for- 413 369 782 0 0 0 0 0 Year profit The New England Institute of Art Special Focus Four- Private for- 555 0 555 0 0 0 0 0 Year profit Tufts University Doctoral Private not-for- 5177 5730 10907 16893 1.59 173635 63.28 10.28 Universities profit University of Massachusetts-Boston Doctoral Public 12700 4056 16756 3696 1.24 18038 52.11 4.88 Universities University of Massachusetts-Lowell Doctoral Public 12983 4196 17179 3059 0.97 13954 51.55 4.56 Universities University of New Hampshire-Main Doctoral Public 12831 2286 15117 5254 1.34 39221 64.98 7.46 Campus Universities

92

University of Phoenix-Massachusetts Special Focus Four- Private for- 179 19 198 0 0 0 0 0 Year profit Urban College of Boston Special Focus Two- Private not-for- 810 0 810 0 0 0 0 0 Year profit Wellesley College Baccalaureate Private not-for- 2323 0 2323 0 0 0 0 0 Colleges profit Wentworth Institute of Technology Master's Colleges & Private not-for- 4329 229 4558 0 0 0 0 0 Universities profit Wheaton College-Norton Baccalaureate Private not-for- 1587 0 1587 0 0 0 0 0 Colleges profit Wheelock College Master's Colleges & Private not-for- 869 462 1331 0 0 0 0 0 Universities profit William James College Special Focus Four- Private not-for- 0 700 700 0 0 0 0 0 Year profit Total 209645 91657 301302 303,557 1.95 3,530,31 62.75% 11.63 5

Source Carnegie Classification 2015 InCites 2010-2016, Access 2016-07-04

93

SIRIS Academic SL Av. Francesc Cambó, 17 08003 Barcelona Spain Tel. +34 93 624 02 28 [email protected] www.sirisacademic.com

94