European Higher Education Institutions: Building a Typology of Research

European Higher Education Institutions: Building a Typology of Research

Ben Jongbloed, Benedetto Lepori, Carlo Salerno and Stig Slipersæter

An interim report for the project Changes in University Incomes:

Their Impact on University-Based Research and Innovation (CHINC)

Under contract with the European Commission No. 22537-2004-12 F1ED SEV NO

1

The project “Changes in University Incomes: Their Impact on University-Based Research and Innovation (CHINC)” is done under service contract with the European Commission, represented by the Institute for Prospective Technological Studies of the Joint Research Centre under contract number 22537-2004-12 F1ED SEV NO The following institutions take part in the project:

Centre for Higher Education Studies (CHES), Czech Republic

The Danish Centre for Studies in Research and Research Policy, Århus

University, Denmark

Observatoire des Sciences et des Techniques, France

Innovation Research Centre, Corvinus University of Budapest, Hungary

Scuola Superiore, Sant' Anna di studi universitari e di perfezionamento, Italy

Center for Higher Education Policy Studies (CHEPS), Universiteit Twente,

the Netherlands

NIFU STEP, Norway

Institute for Innovation and Knowledge Management INGENIO, Spain

Istituto Comunicazione e Formazione – IceF, Universita della Svizzeria italiana,

Switzerland

SPRU (Science and Technology Policy Research), University of Sussex,

United Kingdom

2

Contents Towards a Typology of European Higher Education Institutions ...................................... 4

Introduction..................................................................................................................... 4 Classification of systems................................................................................................. 5 Classification of institutions ........................................................................................... 6 Conceptualizing and constructing a practical typology ................................................ 10 Operationalization......................................................................................................... 13 Developing a classification ........................................................................................... 14

Preliminary analysis of the sample ................................................................................... 18 Methodological limitations ........................................................................................... 19 Higher education institutions in Europe – a first overview .......................................... 20 A first attempt at a typology ......................................................................................... 24

References......................................................................................................................... 30 Methodological Annex...................................................................................................... 31

3

Towards a Typology of European Higher Education Institutions

Introduction When making a survey of the changes in university income categories – in

particular the revenues for research – it is useful to have at one’s disposal a typology for

classifying the institutions in the survey. A classification, or typology, of higher

education institutions is a useful mechanism for describing institutional diversity. It

provides a framework for characterizing institutions on the basis of a set of dimensions

and in this way can assist the typology’s users when analyzing or drawing comparisons

among varied higher education providers. In no place is the need for such a classification

more necessary today than in Europe, where more than 3,000 higher education

institutions stretching across some 30 countries arguably form the most diverse collection

of both education and research programs seeking to operate under a more general, and

more common, framework.

Such diversity is one of the budding system’s strengths, yet when the degree of

diversity becomes so high, transparency inherently becomes low. To those within and

outside of higher education (e.g., students, businesses, academic researchers, institutional

researchers and policy analysts) that must constantly grapple with higher education’s

complexity, it is often unclear where to begin when comparisons or analyses are

warranted. To get an overview, this complexity needs to be reduced to manageable

proportions. In the course of doing this, the objects (in our case: higher education

institutions, universities) subjected to the classification are grouped together into a

limited set of categories that each have their distinctive characteristics.

A classification can help if one would like to know to which family, class or group

a particular higher education institution belongs. However, the first thing that one needs

to know is in which country the higher education institution is situated. In other words, it

can make a large difference if one knows that university A is located in country B.

Therefore, the objects of classifications are not restricted to providers of education and

4

research but may also be extended to entire higher education systems, or indeed science

(R&D) systems. Although the main aim of the Changes in University Incomes (CHINC)

project is to provide insight into the changes in the research portfolio of individual higher

education institutions, it makes sense to also briefly mention system-wide classifications.

Classification of systems In the field of higher education studies, some attempts were made to classify

national systems of higher education. Scholars who focused on the national level, such as

Clark (1983), Goedegebuure et al (1994), Teichler (1988), and van Vught (1989), have

analyzed characteristics of higher education systems in a comparative way, but their

methods do not lend themselves for a classification of the countries involved on the basis

of measurable indicators. At best higher education systems have been compared on the

basis of subjective dimensions that allowed for a relative positioning of countries without

these dimensions being combined into discriminating categories. Clark’s “triangle of

coordination”, for example, allowed for a positioning of countries on the dimensions

‘state versus market coordination’, ‘market versus academic oligarchy coordination’, and

‘academic oligarchy versus state coordination’, but these dimensions were not

operationalized in the sense that Clark did not present indicators that could be used to

interpret the position of a country on the dimension. Therefore it was impossible to

determine/measure whether, for example, France was on the dimension state versus

academic oligarchy closer to the state extreme of the dimension than Sweden.

More recently, and more close to the topic of the CHINC project, the OECD has

looked into the steering and funding of research institutions and, in analyzing the data,

has made an attempt to group countries according to a functional taxonomy.1 The OECD

taxonomy relates to the countries’ patterns of funding and R&D performance. Taking into

account only the public sector research, it came to the conclusion that government is the

main funder in all OECD countries observed, but in addition it made the following

typology of public research systems (between brackets our ‘CHINC-countries’):

1 OECD (2001), AD HOC WORKING GROUP "STEERING AND FUNDING OF RESEARCH INSTITUTIONS" 13-14 March 2001. See: http://www.oecd.org/dataoecd/43/10/2507864.pdf

5

1. Countries where universities are the most important research performer (Denmark, Italy, Netherlands, Norway, Spain, Switzerland)

2. Countries where there is a balance with regard to performance between universities and public research institutions (France, Germany)

3. Countries (in particular in Central Europe) where public research institutions play the major role for public sector research (Czech Republic, Hungary).

This typology is worth keeping in mind when interpreting the results of our analysis

of funding trends in European universities. Differences in particular research revenue

streams across universities may be explained by referring to the between-country

variance as well as the within country variance.

The above information on the funding structures for public R&D were the subject

of the OECD report Governance of Public Research – Towards Better Practices (OECD,

2003). This report, also mentioned in the EUA study on The Funding of University-based

Research and Innovation in Europe (Conraths & Smidt-Södergard, 2004) makes mention

of three archetypes for science systems:

1. The Centralized Archetype: Top down priority setting from ministry/central

government, funding streams directly to public research institutions and universities, no independent funding agencies. This model partly applies to many systems in the new member states and to a certain extent in France.

2. The Dual System Archetype: Federal and regional structure and ministries, both bottom-up and top-down practice, a mix of direct funding for public research institutions and universities AND competitive grant programmes offered by independent funding agencies (Research Councils or Associations). This model is applicable to Germany, UK, Norway, The Netherlands, Italy, and Switzerland.

3. The Decentralized Archetype: Multiple ministerial research responsibilities, a primarily bottom-up agenda setting, mostly project funding via competitive grants from independent agencies. This model seems to be the least represented system in Europe.

Classification of institutions When it comes to making a typology of individual higher education institutions the

best known example of a successfully employed higher education typology is the

Carnegie Classification System. In place for more than 30 years now, its original purpose

was to support higher education and research by identifying categories of American

6

colleges and universities that are functionally similar. Over the years researchers and

policymakers gradually adopted it for other purposes; so much so that today one is hard

pressed to find academic research or policy documents related to higher education that do

not, at some point, make some reference to the Carnegie Classification. Even the general

public in the United States has more than a passing familiarity with it, as it is also used in

the influential rankings of American colleges and universities produced annually by the

popular periodical US News and World Report.

Carnegie classification What makes the Carnegie Classification so relevant and useful to the current study

is its simplicity and dynamism: two characteristics we see as crucial for the development

and implementation of a European higher education research typology. Though the soon-

to-be-released incarnation will be the most expansive typology to-date, it relishes its

widespread reputation for being “a tool for rendering the diversity of US higher education

into analytically manageable, reasonably homogeneous groupings of roughly comparable

institutions” (McCormick, 2004). Moreover, such simplicity has successfully adapted

itself over time. The functions of American colleges and universities have changed

dramatically since the early-1970s as have the issues and priorities relevant to key

stakeholders. In response, Carnegie has undergone four methodological revisions since its

introduction. The first classification was developed in 1973 by Clark Kerr. New versions

appeared in 1976, 1987, 1994 and 2000. Currently it is being revised again – this time

more drastically – for the 2005 classification due out soon (McCormick, 2004). Like the

previous systems, the new classification will include all 3 941 US universities and

colleges that are degree granting and accredited.

The 1994 Classification (i.e. the fourth edition), for instance, distinguishes: research

universities I, research universities II, doctoral universities I, doctoral universities II,

Master’s (comprehensive) colleges and universities I, Master’s (comprehensive) colleges

and universities II, baccalaureate (liberal arts) colleges I, baccalaureate colleges II,

associate of arts colleges, and specialised institutions. The qualifications “I” and “II”

were merely indicators of size: size of federal financial support, number of PhDs granted,

student enrolment, and level of selectivity. A main difference between the 2000 edition

7

and the 1994 edition is that the four doctoral institution categories have been collapsed

into two categories. Another difference between the 2000 Classification and the previous

one was the elimination of federally sponsored research as the proxy to indicate an

institution’s research emphasis.2 There are other conceptual and technical differences

between classifications over time, but the backbone remains similar: institutions are

classified on the basis of their research and teaching objectives, the degrees offered, their

size and their comprehensiveness. The backbone clearly refers to the institutions’

missions.

The Carnegie classification over the years has provided, as McCormick states, “a

single approach to characterizing the diversity in US higher education”. However, today

there are many questions for which different groupings of ‘similar institutions’ may be

more suitable. A classification should always be matched to a specific analytical purpose.

To help address the variety in questions that require a classification of some sort, the

Carnegie Foundation’s 2005 edition will replace the present single classification scheme

with a set of distinct, parallel schemes that will include different sets of institutional

characteristics. Apart from the traditional classification that looks at the institutions’

research activity (albeit based on a multidimensional set of research measures compared

to the previous classifications), other classification criteria will focus on:

(1) undergraduate programs, (2) graduate programs, (3) overall student profile, (4) size

and setting (urban, suburban, rural), (5) undergraduate education, and (6) service and

community engagement. In other words, different criteria related to institutional

characteristics will be used to assess similarities and differences between institutions in

such a way that institutions will be grouped into mutually exclusive categories.

2 The reasons cited are: “First, federal support is at best a rough proxy for an institution’s research

activity that suffers from several weaknesses. Not all research is federally funded, and institutions differ in the proportion of all research that is funded from federal sources. Similarly, academic fields differ in their reliance on federal research funding and also in the costs associated with research. Thus a focus on federal dollars pays greater attention to fields that are heavily dependent on federal funding and also to fields where research requires substantial investments. The federal obligations data are also blind to the pass-through of funds from one institution to another, as happens in the case of large projects involving research teams at different institutions. All of these factors compromise the accuracy of federal obligations as a gauge of overall research activity (see: http://www.carnegiefoundation.org/Classification/CIHE2000/background.htm).

8

The latest editions therefore are based on different definitions than Kerr’s original

ones – they are more teaching than research based, focusing on the number and type of

degrees an institution awards, rather than on research or selectivity in admissions. The

reason is that the research-based classifications were criticized by teaching institutions

and other actors for providing a positive bias towards the research function in higher

education, and consequently for stimulating academic drift based on research. Given this

development it is obvious that for our project the first three editions are more interesting

and relevant. If one would like to come to a European version of the Carnegie

classification, the 1987 edition of the Carnegie classification would seem to be most

relevant for our purposes.

Speaking of European classification attempts, it is good to mention the recent

initiative by a group of (mostly Dutch) researchers that aims to explore the viability of

designing a typology of higher education institutions in Europe. The project was granted

funding in the framework of the Socrates program and so far has produced a progress

report that outlines the goals of the project and proposes a set of criteria for a typology

(Van Vught et al., 2005). Like the Carnegie Foundation initiative, it mentions the wish to

produce a multi-dimensional classification based on objective data. It will be interesting

to see to how this European initiative develops and, in particular, how it operationalizes

the research (versus teaching) dimension as it is the key focus of our project.

Summarizing the above we would like to state that a different perspective (i.e. the

perspective of college attendance, relevant to prospective students and their parents,

versus the perspective of academic research, relevant for research-based classifications)

will lead to a different typology. Because the world – and indeed the world of academic

research – is dynamic and very much driven by market forces, one has to be aware that

classification dimensions will be able to capture the important developments one intends

to study. Institutions will change places in a typology over the years, but typologies

themselves also change.

9

Conceptualizing and constructing a practical typology Identifying the relevant framework and subsequent criteria on which a typology can

be based is complicated on many fronts; not the least of which is securing relevant and

objective data. When undertaking a classification exercise across a wide range of

countries one also runs into the thorny issue of different (and oftentimes multiple or even

overlapping) contexts and institutional settings. The core concern at this stage of the

CHINC project is developing a typology for characterizing accredited European higher

education institutions with respect to their varied research portfolios, yet it is clear that a

concept like “research” is neither easily defined nor operationalized.

Research-based typologies can be developed from a number of different angles

depending on what needs the typology is meant to serve. What type of research do

institutions do? Is it basic, applied or possibly strategic? What kind of institution is

doing the work? To what extent is research being conducted in traditional universities,

specialized institutions (e.g., technical institutes), or professional education institutions?

Who sponsors the research? Does funding come primarily from government (perhaps

integrated with annual appropriations), research councils, industry, charitable foundations

or international sources? How much research capacity do institutions have? Does an

institution have a relatively large number of research staff? Does it train sizeable cohorts

of PhDs? How much output is produced? Does an institution only conduct a relatively

low amount of research? Is the institution’s research concentrated in a small number of

programs or is it more evenly distributed across the entire institution’s disciplinary base?

Of course none of the groupings above are mutually exclusive. More complicated and

more elaborate frameworks might incorporate some or possibly even all of the

perspectives above.

Hence, like the Carnegie Classification, developing and implementing the simplest

yet most encompassing typology is critical. The number of dimensions on which it is

based cannot be too large and the necessary data for positioning institutions ought to be

available in all countries or at least possible to collect without incurring too great a cost.

Nevertheless, given that one of the CHINC project’s overall objectives is to broadly

characterize European higher education research activities, the best typology would

incorporate as many of these varied considerations as is possible.

10

Taking each of the factors above into account, we chose to build the typology

around the following two dimensions:

1. Research versus teaching orientation. This is intended to reflect institutions’

observed missions.

2. Research intensity. This is intended to reflect the extent to which institutions engage in research relative to other activities.

From an implementation standpoint, the use of such dimensions enjoys support

from the fact that the Carnegie Classification has historically used similar measures.

Perhaps the strongest justification though is that these particular dimensions arguably

capture the broadest characterizations of institutions’ research: relative importance and

intensity in comparison to institutions’ other core tasks. Notions like type of research

done, the type of institution doing it or sources of financial support are clearly important.

However, these measures can be thought of as sub-classifications of research

intensity/distribution. In other words they are: 1) inherently narrow in their focus, and

2) possibly inapplicable to certain types of higher education providers. Dimensions that

account for mission and intensity are appealing specifically because they are inclusive.

They make it possible to position all higher education providers, not just research

institutions, including those that do very little or perhaps no research.

Because the unit-of-analysis is the institution, it is also critical that our typology

account for the relatively large number of providers that are not comprehensive in scope

but instead have narrow disciplinary foci. Since relative research intensity depends

greatly on the disciplines doing the work, the volume of research done in technical or

natural sciences institutions (where higher capital costs translate into larger project

budgets) will systematically vary in different ways than it will at institutions that are

strongly humanities-, social science- or performance arts-based. What is more, since

different disciplines often conduct different types of research (e.g., engineering faculties

generally do a balanced amount of basic and applied research and their income portfolio

is likely to include both public and private funding whereas physicists and geologists are

more prone to doing basic research and securing the bulk of their funding from

11

governments and national research councils), relying solely on the two dimensions will

invariably produce misleading and sometimes even strange results.

Our solution is to distinguish institutional types in the typology by using a fixed

number of symbols that capture the scope and intensity of an institution’s disciplinary

base (see Figure 1 below). Of immediate concern is how disaggregated the disciplinary

base needs to be. With sufficiently detailed information and a well-defined purpose, it is

possible to construct dozens of disciplinary subcategories. In our case, limitations on the

project’s parameters and, more importantly, data availability make such an extensive

disaggregation impractical. Based on our own experience and a brief analysis of the

existing data we propose to classify disciplines into one of five broad areas:3

1. Humanities and social science 2. Performance arts (including theater, music and design) 3. Natural sciences 4. Technology and engineering 5. Medicine (and other health-related fields)

From our perspective, the categories above provide the most disaggregated

discipline clusters that are feasible for this study and provide an acceptable degree of

group homogeneity. Not only is the type of research conducted by disciplines falling

within any of the groups similar (e.g., output types like books, journal articles,

performances or patenting) but the groupings also reflect the differing cost structures and

technologies endemic to research in that area. When combined, these categories can also

be used to characterize the many institutions that fall between those with capacity in one

or all of the above areas.

Combining the dimensions and the institution groupings produces a graphical

representation like that found in Figure 1 below. In this example, stars might possibly

represent comprehensive research universities, circles for health-related professional

education institutions and squares for performance arts institutes.

3 Jongbloed, Salerno and Kaiser (2003) developed per-student cost estimates for higher education institutions in the Netherlands. Because cost structures vary by discipline, the authors were faced with a situation similar to that encountered here: suitably disaggregating disciplines in the face of fairly aggregated data.

12

Figure 1. – Typology Dimensions and Institution Groupings Research

oriented

low research intensity

high research intensity

Teaching

oriented

Operationalization To situate higher education institutions within a framework like that in Figure 1, it

is necessary to define proxy indicators for the two axes. Here again one runs into a

number of logical problems. There are many ways to operationalize concepts like

research, research intensity and research orientation, yet all measurements are inherently

deficient. Quantitative indicators are laced with qualitative aspects and in many cases,

because the input or output units are ill-defined, there is little in the way of uniformly

collected data. Many researchers and policymakers today rightly believe that research

measures cannot be adequately captured in a single measure (Jongbloed, et al.,

13

forthcoming) yet the problem they face, and that which we do here as well, is determining

what components most appropriately belong in the relevant basket of indicators. Should

they be input-oriented? Output-oriented? Both? And if both, should input and output

measures be equally weighted? It is a well regarded maxim in higher education research

that definition and choice of research measurement is the quintessential methodological

shortcoming to empirical studies.

With these factors in mind, we believe that two particular quantitative measures

would be most useful: 1) numbers of degrees conferred and 2) research income earned

by institutions. The first is readily observable, though some differences clearly exist

between countries when it comes to the type of degrees conferred below the PhD level.

What is more, the PhD is, for all intents and purposes, universally recognized for what it

bestows in individuals. The other indicator, research income, is the core topic of the

CHINC project and is used in our framework for evaluating research intensity. In our

view, other useful indicators like research or academic staff, publication counts or time

spent on research are of dubious value if only for a lack of reliable and comparable data.

To sum, while it is possible to select a number of different proxy measures for the

two axes, because we seek to be as inclusive and broad as is possible, our choice of

measures was designed to accommodate data availability problems. Though this study

only looks at a fixed number of European countries, the expectation is to apply this

typology in a much larger pan-European framework. With different countries using

different reporting standards and collecting data for their own particular purposes, our

experience leads us to believe that, while not perfect, the measures represent a suitable

“second-best” solution.

Developing a classification The next step is to create a classification scheme from the typology. We do this by

taking the typology above and analyzing it with respect to the following four criteria (an

example of the overlay between the two can be found in Figure 2 below):

1. Highest type of degree offered. When this is a PhD, the institution belongs to

Group A, which naturally includes most, if not all, universities. Group B denotes

14

the Master’s (or similar level) colleges and specialized institutions like business or law schools. Group C consists of those institutions offering baccalaureate degrees or lower-level (but still higher education) diplomas and certificates.

2. Breadth of program offerings. Does an institution only offer degree programs? Does it only carry out research in one of the five categories listed above or does it cover multiple disciplinary fields?

3. Doctorates awarded relative to total education activities. Higher percentages would imply stronger research intensity.

4. Level of research income relative to an institution’s total income. Research income includes core research funding from public agencies, competitive funding from public and private sources and contract income earned from research-related activities. This last component must still be problematized in light of the available data (see brief discussion immediately below).

Figure 2. – Typology Dimensions and Classification Quadrants Research

oriented

The first two criteria relate to the institution’s mission and the latter to its research

intensity. In general the first two points are also straightforward to address. The main

Teachingoriented

high research intensity

low research intensity

A2 A1

B BB2 B1

C2 C1

15

problem is drawing a somewhat arbitrary line between comprehensive and specialized

institutions, but this is partially resolved by our decision to treat institution types as

different plot points in the typology dimensions.

It is not yet clear whether the research intensity measure can be constructed from

composite indicators (depending on data availability). For the moment then, our

preference is to base it solely on the numbers of PhD degrees (criteria 3) and if feasible,

create a composite measure later based on criteria 3 and 4.

Overlapping Figures 1 and 2 gives rise to Figure 3. For illustrative purposes, one

might treat the circles as comprehensive universities and the stars as professional

education institutions. One would expect the research universities scattered in the top

right section of Figure 3 as well as some technical universities and institutions with

medical schools. Institutions that are qualified to confer doctorates but in reality do so

very seldom (like some of the Arts schools or music academies) would be placed towards

the extreme northwest. Institutions like the German Fachhochschulen or Dutch

hogescholen would show up in the right-middle. Those institutions falling into the bottom

two categories would include the diploma granting institutions discussed earlier. Note

that the low volume of research conducted by largely teaching-oriented institutions

implies there should be only a few observations in the lower right classification and

where there are they should lie close to the vertical axis.

16

Figure 3. – Hypothetical Mapping of European Institutions

Teachingoriented

high research intensity

low research intensity

Research oriented

A2 A1

BB2 BB1

C2 C1

Logistically, there is still a problem with defining when an institution falls on either

side of the vertical axis. In other words, what is the borderline between categories A1 and

A2, categories B1 and B2, and C1 and C2? In our view, this will first require an

examination of the available data, particularly that coming from research-intensive

institutions.

17

Preliminary analysis of the sample

Data collection for the CHINC project began in February 2005. The primary

purpose of this initial data gathering exercise was twofold: first to get basic information

about the higher education institutions in the countries covered by the study and second

to test the applicability of the typology expounded upon in the first part of this

intermediary report. Importantly, our sample covers more than just universities. We also

include institutions usually classified in the second tier of many countries higher

education systems, like the Fachhochschulen in Germany, Hogescholen in the

Netherlands and the Høgskoler in Norway. Though such institutions may have strong

education orientations, some possess considerably diverse research portfolios or have

ambitions to increase their research intensity. Because this study focuses specifically on

European higher education institutions’ research portfolios, the inclusion of such

institutions is both logical and, in our view, necessary for fully exploring the underlying

dynamics of academic research in Europe.

Project partners collected data for their representative country. These individuals

are not only capable of collecting such information, indeed they were selected because of

their regional expertise, knowledge of higher education indicators and access to key

networks within their countries. Such expertise is invaluable as no national system is

completely transparent. The knowledge of the national correspondents has already proven

to be remarkably helpful in the process of securing and vetting the data collected to this

point as well as for understanding the limitations the arise in the process.

The sources behind the data collected thus far are publicly available information

coming from within the respective countries. To the extent it was possible, information

was collected for 2002. It should be noted however that in a small number of cases, we

rely on data from other years.4

4 Country experts were asked to annotate all data and explain deviations from all reference points.

18

Methodological limitations While the data collected to now include straightforward and commonly measured

information, definition issues impose a number of limitations on the reliability and

validity of the data. The variety and differentiation of what constitutes an undergraduate

program provides one particularly illustrative example. Many countries have rapidly

adopted the Bologna model of 3 + 2 years while still retaining their former degree

structures. Moreover, in some study fields programs are inherently structured differently

from the “typical” European model, such as the 3- and 4-year undergraduate program

cycles in the United Kingdom. This would likely have the effect of under-estimating the

numbers of undergraduate students at institutions in countries like the UK and France

relative to that which we estimated for institutions in our other sample countries.

Our initial approach was to classify programs into one of four broad categories: 1)

humanities and social sciences, 2) natural sciences, 3) technology, and 4) medicine. After

examining the data collected to this point we revised our program classification by

making performance arts institutions their own category (rather than include them with

the humanities) and to add engineering programs to the technology classification.

An additional problem relates to our coverage of institutions’ scientific fields and

specializations. There are considerable differences between institutions that offer one

economics course and those that offer a range of social science courses, hence it is highly

subjective whether having one social science course is enough to classify a university as

covering the humanities and social sciences. Likewise, how many courses or scientists

working in areas like “technology and human resources management”, “business

administration” or “social aspects of technology” can a technical university have before it

stops being a specialized technical university and a becomes a university within

technology and business administration, or even a university covering technology and

social sciences? From our perspective such problems will be handled on a case-by-case

basis. Here we simply wish to point out that such borderline cases exist and run the risk

of being situated in what others might consider to be the wrong classification.

Defining whether an institution is private or public is tricky because there are

numerous ways in which distinctions between the two sectors are made. We define public

19

institutions as those which receive recurrent public core funding. There can, however, be

institutions which have such funding in most years but not all and consequently there can

be some doubt of the classification. For the more detailed sample which will be collected

later in the project, there will be more detailed data on core funding which will make it

easier to attribute institutions on this variable.

Finding the exact year of foundation is most difficult in cases where an institution is

the result of one or more mergers over the years. As the history of such institutions can be

difficult to track, we have in some cases decided to use the year of foundation for the

institution in its present state as the year of foundation.

Higher education institutions in Europe – a first overview The sample discussed here covers higher education institutions in 11 European

countries, and includes approximately 1000 institutions. Given that the total number of

European5 higher education institutions has been estimated as somewhere in the

neighborhood of 3000, our sample, while not randomly distributed includes

approximately one-third of the population.

5 Here we consider the 25 member countries of the European Union and 8 additional countries.

20

Table 1 Higher education institutions in selected European countries according to the awarding of a PhD or not. Number of institutions and number of undergraduate students. 2002.

Awarding PhD Non PhD Total Country N Students N Students N StudentsCzech Republic 25 217 502 39 10 917 64 228 419Denmark 1 15 40 55 France 86 1 053 121 19 25 104 105 1 078 225Germany 125 1 377 166 209 542 885 334 1 920 051Hungary 21 238 587 45 135 543 66 374 130Italy 70 1 707 527 7 6 839 77 1 714 366Netherlands 18 180 382 54 325 771 72 506 153Norway 16 98 515 28 63 683 44 162 198Spain 58 1 324 753 8 22 121 66 1 346 874Switzerland 12 74 386 7 28 338 19 102 724United Kingdom 87 1 182 420 3 48 670 90 1 234 090Total 533 7 454 359 459 1 209 871 992 8 664 230

1 At the time of writing, the figures for Denmark were not yet ready.

In Table 1 we have split the institutions according to awarding a PhD or not. The

table shows a relative equal distribution of number of institutions in the two groups.

There are 533 institutions (55 percent) which have the right to award a PhD within at

least one scientific field. 459 institutions do not award a PhD.

21

Figure 4 Average size of sample institutions by number of undergraduate student. 2002.

0

5 000

10 000

15 000

20 000

25 000

30 000

Czech

Rep

ublic

Denmark

Franc

e

German

y

Hungary Ita

ly

Netherlan

ds

Norway

Spain

Switzer

land

UK

Avera

ge

Awarding PhDNon PhD

In all countries except for the UK, the average size of the PhD-awarding institutions

in our sample is larger than the non-PhD-awarding institutions (approximately 10 000 for

the former and around 5 000 for the latter). The exception is in Italy and Spain where the

average size of PhD-granting institutions are notably large: enrolments of 24 400 and

22 800 students respectively. Our sample also includes 16 universities with more than 50

000 students. With the exception of the Open University in UK (which is a special case)

all are in Italy (9) and Spain (6). This highlights two important considerations that will

need to be factored into the later analysis. First, the diverse educational offerings of such

“mega-universities” likely include some programs that are offered by second-tier

institutions in other countries. Second, these institutions may contain pockets of strong

research but because of their size be regarded as not very research oriented on the whole.

Note also that the UK is the only country in our sample where the non-PhD-

awarding institutions are larger than the PhD-awarding ones.

22

There are also large differences between the countries in terms of the number of

institutions in both categories. Figure 5 shows the per-country distribution of institutions

by highest degree awarded.

Figure 5 Distribution of higher education institutions per country according to

the awarding of a PhD or not. 2002.

0 %

20 %

40 %

60 %

80 %

100 %

Czech

Repu

blic

Denmark

France

German

y

Hungary Ita

ly

Netherl

ands

Norway

Spain

Switzerl

and UK

Averag

e

Non PhDAwarding PhD

A clear pattern can be seen in Figure 5 between those countries operating dual (CZ,

DK, D, HU, NL and CH) and unitary (F, I, ESP and UK) systems. Systems in the former

tend to have larger numbers of second-tier (e.g. professional education) providers than

the latter. These institutions possess at least three characteristics that are important to the

typology laid out earlier in Figure 2. One, on average, they tend to enroll fewer students.

Two, they are often more specialized in their program offerings. Three, on a volume basis

they conduct much less research than traditional universities. Taken together these factors

suggest that second-tier institutions should, as a rule, be located in the C2 area of our

typology. However, this will not be the case for all such institutions. In some countries

(e.g. Switzerland or Finland) such providers tend to have specific research missions or

foci, and will thus have higher research intensity. Moreover, because we have elected to

23

use per-capita measures in order to control for the wide range of operational scale a

Dutch hogescholen, for example, with a low volume of research output can be plausibly

regarded as “research intensive” if they also have notably low student enrolments and/or

a relatively high faculty to student ratio. To the extent that such differences are in

evidence will become clearer as more data is collected. At this point we simply bring the

issue forward to highlight our awareness of the methodological difficulties that exist and

to provide further evidence for the idea that research intensity cannot be adequately

captured through the use of one or even two indicators.

Given that most institutions in unitary systems award PhDs, the natural inclination

is to expect them, a priori, to be research intensive. However, as can be seen in Figure 4,

many also teach large numbers of students (see “mega universities” issue addressed

earlier). If class sizes are small, student-faculty interaction is extensive or campus

infrastructure is geared more towards education rather than research activities, then it

would not be surprising to find many PhD-granting institutions within unitary systems to

be relatively less research-intensive than their peers in other countries or possibly even

other tiers in other systems. Again, the confirmation of our suspicions by the data

collected to this point suggests that careful attention will need to be given to developing

as large a basket of institutional research indicators as is feasibly possible.

A first attempt at a typology Research orientation, as indicated by the right to award PhD degrees, is one of the

dimensions of our typology. As shown in table 1, slightly over half of the institutions in

our sample have an orientation toward basic research. The second dimension in our

typology is research intensity. This reflects the actual extent to which institutions engage

in research relative to other activities (in particular teaching). This second dimension

differs from the first dimension in the sense that uses a relative measure instead of an

absolute one. As stated in the first part of this report, the relative level of the institutions’

research income and the doctorates awarded relative to total educational activities are

selected as indicators of research intensity. Given the state of our data collection so far,

we only have information the second indicator. In the remainder of this section we will

24

employ this indicator to look at the research intensity of the PhD awarding institutions

(mainly universities) in our sample.

On the basis of the PhD/student ratio one can observe that, on average, large

institutions are not very research intensive. For the subset of PhD-awarding institutions,

Table 2 gives a first impression of how research intensity is distributed among the

countries and institutions.

Table 2 Research intensity expressed by PhD degrees awarded per 100 undergraduate students. 2002.

Country PhD degrees awarded

No. of under-graduate students

PhD degrees per 100 students

Czech Republic 1 228 228 419 0,54 Denmark 961 France 9 568 1 078 225 0,89 Germany 23 675 1 920 051 1,23 Hungary 7 430 374 130 1,99 Italy 4 341 1 714 366 0,25 Netherlands 2 538 506 153 0,50 Norway 739 162 198 0,46 Spain 6 245 1 346 874 0,46 Switzerland 2 743 102 724 2,67 UK 10 934 1 231 090 0,89 Total 70 402 8 664 230 0,80

The table shows considerable variation for this indicator; from 0.25 PhDs awarded

per 100 undergraduate students enrolled in Italy to 2.67 in Switzerland. The average ratio

is 0.80 (Denmark excluded).

The difference between the countries can probably to some extent be explained by

factors like the absorption of PhDs in the industrial and academic workforce, traditions

for high level academic training in the populace and the availability of educational

programs and research opportunities at this level. In other words, there are structural and

institutional differences between the countries higher education and research systems.

The differences as expressed by the PhD/student indicator are quite substantial, showing

a magnitude of almost 1:11 between the highest and lowest ranking nations.

25

When it comes to individual institutions, one can once again observe a large

variation in research intensity across as well as within countries. Again, structural and

institutional differences can be put forward. However, the variation across individual

universities within countries is due partly to the country’s national policy for resourcing

education and research and the individual institution’s teaching and research strategy and

its research portfolio. Some institutions – like some of the big Italian or Spanish

universities – will have a large vocational education component. Others have bigger

research budgets at their disposal, enabling them to engage in research training. The aim

of our (CHINC) project is to look more deeply into the issue of research budgets to

analyze which revenue sources over time have grown or experienced decline and to what

extent national or individual university policy is causing this.

The number of PhDs awarded in relation to the institution’s undergraduate

portfolio, therefore is a reflection of the institution’s mission, which partly is a result of

national policies and priorities and partly of the individual institution’s strategy and

opportunities. Therefore, at this point in the project we have chosen not to present the

outcomes in terms of the values for the PhD/student ratio for individual institutions, as

this would pose too many questions still; questions that can only adequately be addressed

once more data are available. A list – or even a ranking – of institutions based on the

value of this indicator will list only some of the well-known European research

universities in the research-intensive part of the classification (Zürich, Oxbridge, LSE,

München, Paris VI), whereas others (e.g. Heidelberg, Oslo, La Sapienza, Delft, Oslo)

only show up in the less research-intensive (i.e. bottom) part of the list. Yet other

institutions – many of them specialised in fields like arts and engineering, or even

performing arts – turn out to be the research-intensive institutions.

However, an exercise focusing on the PhD/student ratios is instructive for several

reasons. First it triggers many questions about the data for individual institutions, about

definitions and possible anomalies. Second, it immediately shows the need to properly

take into account the national context. Third, it points to the institutional mission and

profile – some institutions are primarily teaching oriented, others are specialized in

particular fields, and yet others are heavily oriented on research in engineering or medical

disciplines with a large potential to generate revenues. Fourth, and most importantly, our

26

first attempt to quantitatively address research intensity shows the need for a basket of

indicators to assess the research portfolio of institutions. A one-dimensional picture on

the basis of a readily available indicator clearly is insufficient.

What has also become clear after closer inspection of the data for the PhD/student

indicator is that the profile of the individual institution has a large impact on the actual

value for this ratio. Some institutions are relatively small, specializing on particular fields

– either in undergraduate education and/or postgraduate education and/or in particular

disciplinary fields. For instance, our exercise shows that the institutions having a

relatively high value of the PhD/student ratio on average have less than 10 000

undergraduates, were established before the year 1900, and quite a few of them are

technical universities.

For the 50% of PhD-awarding institutions in our sample that have the highest

scores in terms of PhD/student ratio, figure 6 shows the number of PhDs awarded per

100 students enrolled on the vertical axis (as a proxy for research intensity) and their

coverage of scientific fields on the horizontal axis. The figure illustrates that institutions

that are research intensive in terms of the PhD/student ratio also have a tendency to be

specialized in terms of scientific fields covered. All institutions with a ratio over 6 are

covering 2 fields or less. For ratios under 6 the coverage is more evenly distributed.

27

Figure 6 Research intensity by scientific fields covered

0 1 2 3

SciFields4

2,0

4,0

6,0

8,0

10,0

PhD

_stu

d

PhD_stud: Number of PhD-degrees awarded per undergraduate students enrolled. SciFields: Number of scientific fields covered by the institutions.

The distribution of scientific fields covered by the institutions has only been

analyzed for the institutions having the highest PhD/student ratio. A further analysis of

particular subsets of institutions in our sample could provide additional insights in across-

country and within-country differences between, for instance, the technological

institutions.

Our sample also allows analysis at the regional level as all institutions are classified

according to the Eurostat NUTS standard (level 2). This opens up the possibility of

analyzing regional distributions along the same axis as at national level.

28

The analysis of the first, relatively simple set of data illustrates in our opinion the

need for building a typology of European higher education institutions based on data for

the institutional level. What our analysis up to now has shown is that research intensity is

a dimension that needs to be captured on the basis of more than just one indicator. Our

one-dimensional PhD/students indicator does not allow us properly assess research

intensity. From this exercise and other experiences we know Europe to show a very large

diversity of national settings. Knowledge of the context is thus of utmost importance

when making comparisons between countries or higher education institutions and

inferring conclusions. Indicators will always have to be interpreted in their national

context.

A multi-dimensional indicator clearly is needed, as research intensity is a

function/reflection of other factors like the composition of an institution’s revenues, its

number of personnel and academic specializations, as well its strategies.

The next stage in our project, therefore, is to collect information that allows us to

adequately capture research orientation, intensity, and research portfolio management on

the national and – in particular – institutional level.

29

References McCormick, A.C. (2004), The 2005 revision of the Carnegie Classification System. Presentation to the Washington Higher Education Secretariat. Washington D.C. June 8, 2004. available online at: http://www.carnegiefoundation.org/Classification/future.htm Clark, B.R. (1983). The Higher Education System: Academic Organization in Cross-National Perspective. Berkeley: University of California Press. Jongbloed, B. W. A., Salerno, C. & Kaiser, F. Kosten per student. Methodologie, schattingen en een internationale vergelijking. Beleidsgerichte studies Hoger onderw. & Wetensch. onderzoek, 100. Den Haag: Ministerie van Onderwijs, Cultuur en Wetenschap, 138 pp. isbn. 905910191X. (Translation: Per student costs: Methodology, estimates and an international comparison) Jongbloed, B.W.A., Salerno, C.S., Huisman, J., & Vossensteyn, J.J. (2005). Researchprestatiemeting in zes landen. CHEPS: University of Twente. (Translation: Research performance measurement in six countries) L. Goedegebuure, F. Kaiser, P. Maassen, L. Meek, F. van Vught, and E. de Weert (eds.), Higher Education Policy: An International Comparative Perspective, Pergamon Press: Oxford. McCormick, A.C. (2004), The 2005 revision of the Carnegie Classification System. Presentation to the Washington Higher Education Secretariat. Washington D.C. June 8, 2004. available online at: http://www.carnegiefoundation.org/Classification/future.htm Teichler, U. (1988), Changing patterns of higher education systems. The experience of three decades. London: Jessica Kingsley. Van Vught et al. (2005), Towards a typology of higher education institutions in Europe. Progress report March 2005.

30

Methodological Annex Definition of variables for data collection on all higher education institutions All variables refer to the year 2002. Basically all higher education institutions should be included, including universities, technical universities, professional universities etc. Institutions delivering only curricula of less than three years and institutions giving only continuing education should be excluded. General description Definitions / instructions Code Pre filled. Do not change the numbers. Name of the institution Full name of the institution (university, college etc) in the national

language. Institution acronym Fill inn if institution has an acronym or abbreviation. If not, leave

empty. English name of the institution

Insert the official English name of the institution. If no such name exists, leave open.

Year of foundation In case the institution is a product of earlier mergers between institutions etc, use the year of foundation for the oldest part of the institution. Use the "Comments" column for further explanations when necessary.

Type of institution Insert type of institution according to the national system of higher education institutions. Use national language. Example France: If institution belongs to the Grande Ecoles category, insert "Grande Ecole". Example Germany: If institution belongs to the Hochschulen category, insert "Hochschule".

Public / Private Insert "1" if the institution belongs to public sector, i.e. institutions receiving annual core funding form the state. Insert "2" if the institution belongs to private sector, i.e. all institutions not included in public sector.

Region Insert code for the region where the institution is situated. See spreadsheet "Regions definitions" for codes to be used. Codes used are Eurostat NUTS, 2003 revision, level 2.

Number of under-graduate students

Total number of undergraduate students enrolled (headcount). Exclude vocational and postgraduate education. Bologna masters should be included.

31

Scientific fields covered The broad scientific fields covered by the institution. Use the following codes: "1": Humanities and social sciences "2": Natural sciences "3": Technology "4": Medicine If several codes, separate by comma.

Awarding PhD Insert "1" if the institution is awarding at least one PhD degree or another degree of equivalent level. Otherwise, insert "2".

Number of PhD degrees awarded

The total number of PhD degrees awarded by the institution.

Specialisation classification

If the institution is in some way specialized, indicate kind of specialisation, examples:

• Technical university • Business school • Arts • Theology • Agriculture.

If necessary add new categories Sample Insert “1” if you suggested this institution should be included in

the projects’ final sample of 100+ institutions. Criteria to be used for the choice: availability of data; interest for the case. Indicate more institutions than required as the total sample has to be representative. A shortlist will be made.

Comments Open for comments on the variables.

32