Version classiqueVersion mobile
OpenEdition Books

The role of German universities in a system of joint knowledge generation and innovation

 | 
Mirja Meyborg

4. Data of Analysis and Subject of Study

Texte intégral

1As this PhD thesis is about German universities and their increasing importance regarding knowledge generation, innovation and collaboration, scientific publications and patent applications are used to measure their invention and innovation potential. Chapter two has already shown that both indicators are proper R&D results indicators which measure the successful output of R&D efforts. Thus, scientific publications are now used to picture the invention activity of all German universities, whereas patent applications are further used in order to show their innovation potential. The following two subchapters now present a detailed introduction of both innovation indicators, before shortly introducing the German universities that are considered for analyses.

4.1. Scientific Publications as Innovation Indicator

2Bibliography is traditionally the academic study of books as physical objects. Carter and Barker (2010) describe bibliography as a twofold scholarly discipline. First, they mention the organized listing of books and other works such as journal articles. Here, information on the author, the title, the publisher and place of publication as well as the date of publication amongst other can be found. Second, they bring up the systematic and detailed description of books as physical objects. In this context, it is about format and pagination statements. Generally, bibliometrics is a tool which helps to quantitatively analyze scientific literature (Oxford Dictionaries 2012). Of course, there exist different publication data-bases which can be used to analyze publication activity of certain authors and their linkages to others. The following two subchapters now firstly introduce the two most relevant data-bases, namely Scopus Elsevier and Web of Science, before presenting how the publication data from Scopus got prepared for all following analyses.

4.1.1. Scopus Elsevier versus Web of Science

  • 15 Since Elsevier, one of the main international publishers of scientific journals, is the owner of S (...)

3The Scopus data-base, developed by Elsevier15 in 2004, is an abstract and indexing data-base with full-text links. Its name comes from the bird hammerkop, in latin Scopus umbretta, which reportedly has excellent navigation skills. It took two years until the data-base was ready to extensively use, and during this time, 21 research institutions and more than 300 researchers and libarians helped to improve Scopus through verbal and behavioural feedbacks (Burnham 2006). Scopus covers almost 13.000 journals in the field of physical, health, life, and social sciences, whereof about 500 are open source (Falagas et al. 2008). The coverage period is from 1966 until today and the data-base gets updated daily. Besides, 60% of all titles are from countries other than the United States (Burnham 2006). The following figure now shows the basic search mode of Scopus.

Figure 4: Basic Search Mode of Scopus (Scopus 2012).

  • 16 The following subchapter will present the search mode used for this PhD thesis.

4As you can see from the figure, the basic search mode, for example, makes use of fill-in and drop-down boxes to search various fields. The search can be further limited to date, document type and subject area. Besides, author, affiliation, and advanced search modes exist which can be used if required (Burnham 2006). The results are generally displayed as a listing of 20 to 200 items per page which can be further saved to a list, exported, printed, or e-mailed. The results can also be refined by source title, author name, year of publication, document type and subject area. In this context, there are further fields that can be included within the output, such as citation information, bibliographical information, abstract and keywords, funding details, references and other information (Falagas et al. 2008).16

  • 17 “Web of Science was created by Thomas Scientific to make citation indices (that E. Garfield assess (...)

5In contrast, Thomson Scientific which is one of the five operating divisions of The Thomson Corporation developed the famous bibliometric database Web of Science in 2004.17 Web of Science covers almost 9.000 journals in the field of science, technology, social science, arts and humanities (Falagas et al. 2008). The coverage period is from 1900 until today and the data-base is updated weekly. The geographic coverage includes 80 countries (Burnham 2006). Of course, also Web of Science possesses different possibilities regarding its search modes. It has a quick search, an author finder, a cited reference search, and an advanced search. The following figure now shows the quick search mode of Web of Science.

Figure 5: Quick Search Mode of Web of Science (Web of Knowledge 2012).

6As it is for Scopus, this figure also shows for Web of Science that the quick search mode makes use of fill-in and drop-down boxes to search various fields. Besides, the search can be further limited to date. Limitations to document type and language can be made within the advanced search (Web of Knowledge 2012). However, the results are displayed as a listing of 10 to 50 items per page. The full title, author names, and source are provided. All related records can be further sorted by publication date, processing date, times cited, relevance, source title, conference title, and author name. Of cource, the results can be further analysed by author, country, or document type and it is possible to either e-mail all selected records or to print them (Falagas 2008).

  • 18 The KIT is one of five German institutions which enjoys free access for Scopus.

7Coming to a comparison of both data-bases, Falagas et al. (2008) found that overall Scopus has better features than Web of Science. They, for example, did a specific search on the word “brusellosis” and came to the conclusion that Scopus listed about 20% more articles referencing their search term in any given period compared to Web of Science. Thus, Scopus includes apparently a more expanded spectrum of journal articles than did Web of Science. On the oder hand, Web of Science provides better graphics within its citation analysis. Besides, Burnham (2006) concludes that it is easier to navigate Scopus, even for the novice user than it is for Web of Science. The ability to proper use the dropdown boxes allows for a comfortable searching within the intuitive search system. In contrast, Web of Science possesses the better coverage with its data-base going back to 1900 since 2005. However, it is obvious that both data-bases bear its own strenghts and weaknesses, so that they could be generally used interchangeable for all conducted analyses. T h e entire data preparation was extensive and not trivial at all; the decisive factors for Scopus were its much easier navigation and, of course, its free access.18

4.1.2. Data Preparation with Scopus Elsevier

8Again, as this PhD thesis is, amongst other things, about the role of German universities as knowledge generators, the publication data are taken to picture this development. Thus, all publications are filtered out from Scopus, where at least one German university has been involved with. Within the basic search mode of Scopus, the official name of all German universities has been filled in separately in the search field. In order to get the best possible matching in this regard, affiliation name could be further chosen within the drop down-boxes. This search strategy has been conducted for four years, namely 2000, 2003, 2006 and 2009. There were no further restrictions, as all document types as well as all subject areas have been considered. After choosing all document titles for one university and one time period, this data-set could be exported via comma-separated values (csv) file with the following chosen output criteria:

  • Citation information, such as authors, document title and year, and
  • bibliographical information, such as affiliations.
  • 19 See appendix for the exact and detailed assignment of each group of authors.

9As the export data of Scopus has been not as expedient as expected, several additional Excel VBA macros had to be written. Thus, after exporting all data available for the 76 German universities from 1999 until 2010, the csv-files got further prepared until each author of a publication was assigned to its own publication-ID, country-code and year. They got further classified in being a German university (subject of study), any other university, a research institute or an enterprise.19 The last step in this regard was the calculation of the weighting for each single author. In doing so, it was assured that the share of involvement was calculated exactly, i.e. if there have been four authors from the KIT, two from any other foreign university, and two from Daimler, the weighting for the KIT has been 0.5, and 0.25 for the foreign university, as well as for Daimler. This approach is especially important for institutional or geographical measures in order to avoid double or even multiple counts. Finally, in order to be able to conduct a detailed spatial analysis, each author got also assigned to its individual geographic coordinate. The detailed explanation of this procedure will be introduced in the context of the spatial analysis. The data set looks as follows:

Table 2: Example of Data Preparation by Means of Publications (own illustration).

10As can be seen from the table, each publication has got its own ID number, the name of each author, the country code, the year when the publication has been published, the number of authors that has been involved with the publication, the weight, the actors code, the title as well as its longitude and latitude. Having all this information, it is possible to conduct several detailed analyses concerning the aforementioned institutional and spatial perceptions.

4.2. Patent Data as Innovation Indicator

11While the role of all German universities as knowledge generator is pictured through scientific publications, patent applications are used to illustrate the development path of all German universities as innovators.

12Thus, in this regard, for all conducted analyses patent data are the basis. The following two subchapters now firstly introduce the phenomenon of patent statistics, before illustrating how the patent data got finally prepared.

4.2.1. Patent Statistics

13A patent is a temporary monopoly, issued by an authorized governmental agency. It grants the right to exclude anyone else from the commercial production or use of a specific new device, apparatus, or process. This right is given to the inventor of this innovative device or process after an examination that pays attention to the novelty of the claimed item and its potential utility (Griliches 1990). Thereby, the guidelines for examination in the European Patent Organisation (EPO) claim that the invention needs to be a technological one, susceptible of industrial application, novel, and finally needs to involve an inventive step in order to be qualified for patenting. If the public institution finally grants a temporary monopoly to the inventor, he is enabled to fully dispose over the returns of the invention. Of course, the inventor can assign the right to use the patent to somebody else, usually to its employer, a corporation, or sell it or license for use by somebody else. Thus, patents provide a legal framework for the protection of an invention. T h e public itself sees the stock of public knowledge increased (Griliches 1990).

14Patent statistics are a crucial tool for scientists, statisticians and policy makers interested in innovation and intellectual property rights (IPR), as they measure the successful output of R&D efforts (Carpenter and Narin 1983, Griliches 1984, Schmoch et al. 1988 and Grupp 1998). As innovation indicator, patents refer to technological innovations, mirroring a part of the existing technological knowledge stock of a sector, region, or economy (Frietsch et al. 2008). Further, while many distinct advantages regarding the usage of patents as innovation indicator have been discussed over the past years, Schmoch (1990) summarised these advantages as follows:

  • Patents provide a broad documentation of nearly all fields of technology,
  • they have a complete geographic coverage,
  • the collection of exclusive technological information is possible,
  • it is possible to collect the important global innovation in German and English,
  • they have a simple retrieval due to t h e fine classification,
  • patent data have a good availability due to existing data-bases,
  • patents are rich in technological details,
  • they have a strong relatedness to certain applications,
  • they make it possible to collect bibliographic information, such as applicant 20 , inventor21 , application years, etc.,
  • they deliver up-to-date information, and
  • statistical analyses based on patent data are less cost-intensive.

15Generally, patent data make it possible to show innovation activities in time and through a geographical, technological, sectoral, and legal perspective (Tran 2011). The following subchapter now illustrates data preparation with PATSTAT.

4.2.2. Data Preparation with PATSTAT

16This PhD thesis concentrates on patents which are filed at the EPO Worldwide Patent Statistical Data-Base Version October 2010 (PATSTAT) or went through the Patent Cooperation Treaty (PCT) filing process at the World Intellectual Property Organisation (WIPO). In doing so, it is assured that it is dealt with patents with a high-expected economic value (Frietsch and Jung 2008). In order to conduct all following analyses properly, the patent data had to be prepared in a certain manner. For this purpose, all German patents have been filtered out that have been filed at the EPO from 1999 until 2008. By means of Structured Query Language (SQL), t h e following information has been gathered:

  • Application-ID,
  • year,
  • name of affiliation,
  • nuts-code,
  • country-code,
  • application sequence number
  • inventor sequence number.
  • 22 This proceeding is also done by Meyer-Krahmer and Schmoch 1998.

17After obtaining this data, each patent got further classified in either being from a German university (subject of study), any other university, a research institute, an enterprise or a private person. Here, one challenge was to filter out all German university patents, as in Germany the so called ‘Hochschullehrerprivileg’ prevailed until February 2002. This means that employees of universities (professors or scientific assistants) could freely dispose of their intellectual property rights, and thus appear as patent applicants (Meyer-Krahmer and Schmoch 1998). In order to filter out all university patent applications, it has been looked at the academic title to further check all patent applications with professors as applicants. Hence, it could have been determined whether the professor really worked at a university, and to which university he was really affiliated.22 Besides, it is further concentrated on co-applicant networks, i.e. this PhD thesis focuses on networks of applicants, not on networks of inventors. Thereby, a connection between two actors is existent, if two applicants are on the same patent. Thus, the overall activity of German applicants is illustrated and can be compared to the changing role of the German universities as innovators. The prepared data set looks finally as follows:

Table 3: Example of Data Preparation by Means of Patents (own illustration).

18Again, the table shows an excerpt of the data panel including the application-ID, the year when the patent has been filed, the name of each actor, the country code, the inventor and applicant number of each actor, the actor's code and the weight each actor has on the patent. A detailed description of the data application will be provided for all hypotheses within the following chapters, where the conception and methodology are introduced.

4.3. Subject of Study

19This PhD thesis tackles the question how the German universities behave in a system of joint knowledge generation and innovation, and how they have developed over the past ten years. Thus, in a greater context, German universities are the subject of study. But as there are around 420 academic institutions in Germany, this PhD thesis focuses on approximately one fifth of them. But how have the academic institutions been chosen?

  • 23 The German Fachhochschule is a type of German institution of higher education that emerged from th (...)

20There are basically five different types of universities in Germany; the traditional university, the pedagogical university, the theological university, the art academy and the university of applied science (the German Fachhochschule23). The following table now presents an overview of the German academic landscape.

Table 4: Overview of Germany's Academia (DESTATIS 2012).

21The table shows all possible types of the German academic institutions from the winter terms 2009/2010, 2010/2011 and 2011/2012. In general, a slight increase of the number of the German academic institutions can be observed over the past few years. However, within the last winter term, Germany possessed 421 different academic institutions altogether, whereof 108 belonged to traditional universities, six to pedagogical universities, 16 to theological universities, 52 to art academies and 239 to universities of applied sciences (DESTATIS 2012).

  • 24 Except of the Technical University of Clausthal (4,300), the University of Flensburg (4,700) and t (...)
  • 25 It is of course possible to be in two of the three groups or even in each of them, as some few uni (...)

22As already aforementioned, this PhD thesis does not cover all 421 German universities due to a couple of reasons. First, as it is always about knowledge generation and innovation as well as about scientific collaboration, it would have not been very reasonable to examine the pedagogical, theological and art universities, especially with regard to their innovation behaviour. Second, data preparation, especially with regard to the publication data, was comprehensive and time-consuming so that it was impossible to prepare for each university all data in such a detailed manner. Hence, the German universities of applied sciences are not taken at all. Finally, off the 108 traditional German universities each university has been selected which is publicly funded, is able to reward doctorates, and possesses more than around 5,000 students24 . Finally, there are now 76 German universities which represent the subject of study. As the analyses are also based on different types of German universities, they are also assigned to three different groups. Hence, each German university can be assigned to the groups of elite, technical and/or medical universities or in none of them.25 The detailed illustration and classification of the German universities is shown within the appendix and will be further used for several statistical tests in order to identify possible differences regarding their knowledge generation, innovation and collaboration behaviour.

Notes

15 Since Elsevier, one of the main international publishers of scientific journals, is the owner of Scopus, it established the independent and international Scopus Content Selection and Advisory Board in order to maintain an open and transparent content coverage policy. The members are from all scientific disciplines and geographical areas, whose interest is to access any relevant information regardless of the publishers (Hoogendoorn 2008).

16 The following subchapter will present the search mode used for this PhD thesis.

17 “Web of Science was created by Thomas Scientific to make citation indices (that E. Garfield assessed since the early 1960s) accessible via the internet” (Falagas 2008).

18 The KIT is one of five German institutions which enjoys free access for Scopus.

19 See appendix for the exact and detailed assignment of each group of authors.

20 The applicant is the holder and owner of the patent. They can be any kind of legal persons, such as a private person, an enterprise or an institute. In most cases, it is represented by an enterprise or any organization (Tran 2011).

21 In contrast to the applicant, the inventor is the real human creator of an invention (Tran 2011).

22 This proceeding is also done by Meyer-Krahmer and Schmoch 1998.

23 The German Fachhochschule is a type of German institution of higher education that emerged from the traditional engineering schools and similar professional schools of other disciplines. It differs from the traditional university mainly through its more application or practical orientation and less research. They do not award doctoral degrees themselves. This and the rule to call professors with a professional career of at least three years outside the university system remain their major difference from traditional universities (BMBF 2004).

24 Except of the Technical University of Clausthal (4,300), the University of Flensburg (4,700) and the University of Lübeck (3,400).

25 It is of course possible to be in two of the three groups or even in each of them, as some few universities are both, medical and technical oriented and got further selected as elite university.

Table des illustrations

Légende Figure 4: Basic Search Mode of Scopus (Scopus 2012).
URL http://books.openedition.org/ksp/docannexe/image/234/img-1.jpg
Fichier image/jpeg, 132k
Légende Figure 5: Quick Search Mode of Web of Science (Web of Knowledge 2012).
URL http://books.openedition.org/ksp/docannexe/image/234/img-2.jpg
Fichier image/jpeg, 112k
Légende Table 2: Example of Data Preparation by Means of Publications (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/234/img-3.jpg
Fichier image/jpeg, 227k
Légende Table 3: Example of Data Preparation by Means of Patents (own illustration).
URL http://books.openedition.org/ksp/docannexe/image/234/img-4.jpg
Fichier image/jpeg, 170k
Légende Table 4: Overview of Germany's Academia (DESTATIS 2012).
URL http://books.openedition.org/ksp/docannexe/image/234/img-5.jpg
Fichier image/jpeg, 104k

Lire

Open access