61st IFLA General Conference - Conference Proceedings - August 20-25, 1995
Applications in teaching bibliometrics
Sara von Ungern-Sternberg, PhD, Senior lecturer, Abo Akademi University, Department of Library and Information Science, FINLAND
In recent discussions of library and information science (LIS) educators, reducing heterophily, the lack of similarity between two groups or individuals, among LIS researchers and practitioners has b
een emphasized. Two researchers with different backgrounds, one in bibliometrics and the other in case study, have observed and discussed applications of bibliometrics and case study as used in teach
ing research methods. The main thrust is in providing tools for teaching these research methods so that the gap between research and practical application could be narrowed. Dr. Leena Siitonen discus
ses in her paper applications in teaching case study research methods.
Bibliometric methods are seldom used by librarians in practical work. Still these methods grow more important when planning information provision in research libraries. New subject fields develop and
the number of interdisciplinary publications have during the last decades grown exponentially. It is, though, difficult to organise information in new fields, when the classification systems, used f
or instance by journal services, have a discipline based structure. The need to organize this information and help the user to identify relevant documents grows more important, and at the same time t
he huge amount of available documents give great possibilities to apply bibliometrics easily and in the frame of practical work. Bibliometrics provide a tool for getting the core for developing a loc
al collection in a new field. Teaching bibliometric methods could be developed by 1. seminars, where the students learn the methods and also learn to interprete their results by comparing with other
studies, and 2. by use of online systems which give good bases for different bibliometric methods.
- Key words:
Social science research
Librarians do not usually include bibliometrics in their practical work, and still it is very useful for the planning of information provision. There are several reasons for the non-use of bibliometr
ics. Bibliometric studies are time-consuming and sometimes difficult to perform; another problem is that the results of bibliometric studies give a simplified picture of a complex reality and must ta
ke into account many variables to be useful in practice.
The volume of bibliometric studies reported in the literature is big and difficult to put together and compare. There is a need for refining techniques so that the results of one study can be compare
d more satisfactorily with those of another (Broadus, 1987). Bibliometric methods, however, give opportunities to describe the content, structure and development of research, and bibliometrics is bec
oming more important as a basis for the collection development in research libraries. The reason is that scientists today to a large extent organise themselves in networks and the discipline based mo
del of organising knowledge is becoming obsolete. New subject fields emerge and the number of interdisciplinary publications grows exponentially. The classification systems, used for instance by jour
nal services, have a discipline-based structure and do not include new, interdisciplinary subject fields. If they include them, it is often on a very basic level.
Librarians, who develop library collections, databases or information services, will have difficulties in knowing which material should be included, especially in cases where there is little concensu
s even among the scientists about the definition of the field. In these cases bibliometrics can provide a tool for identifying the core for a local collection in such fields.
2. APPLICATION OF BIBLIOMETRICS
The huge amount of documents available through networks gives great possibilities to apply bibliometrics within the framework of the librarians daily work. Different bibliographic databases register
daily thousands of references to literature in almost any subject field. Bibliometric studies can be based on index terms, words in the titles of documents, authors, sources or geographical or time d
istributions. The citation indexes (for example Science Citation Index and Social Science Citation Index) provide a base for analyses based on citations.
Educational institutions have a responsibility to teach students these methods, as well as other research methods in library and information science. A way of teaching bibliometrics is proposed in th
is paper. It is based on a real library problem, which is to develop local collections in an interdisciplinary field. A course like this can be held in cooperation with the university library. The bi
bliometric studies could be applied on one of the subject fields covered by the library collections and publications by local scientists in the field the base for the studies.
3. COLLECTION DEVELOPMENT PROCESS
Collection development includes planning, implementation and evaluation of collections (Baughman, 1977):
Planning is to map information needs, to develop aims and make decisions about priorities. Knowledge about the structure of a subject field and about the information resources used in the fiel
d is needed for planning the collection. Bibliometric methods such as citation analysis, bibliographic coupling, coword analysis and cocitation analysis can be used to map the knowledge structure and
the use of literature. Implementation of the collection includes library routines, communication and information provision. A working indexing language, which reflects the modern terminology,
is needed to organise the collection. Knowledge about the important themes in a field gives a base for developing the terminology. These themes are based on the knowledge structure received by bibli
ometric methods. Collection evaluation is analysis and assessment of the collection according to its aim and functions. Different bibliometric methods such as citation analysis, analysis of th
e scattering of articles to journals and analysis of the obsolescence of literature are used for this purpose.
4. BIBLIOMETRICS AS RESEARCH METHOD
Bibliometrics is the application of mathematical and statistical methods for measuring quantitative and qualitative changes in collections of books and other media. By using quantitative analysis it
is, for instance, possible to measure the scattering of articles to different journals or to measure the growth and obsolescence of literature in different subject fields. These analyses show that a
small part of the journals in a field stand for a great part of the relevant articles in the field (Bradfords law) and that only a small part of the authors in a field are highly productive (Lotka's
law). Bibliometric studies can be used to study the regional pattern of research, the extent of cooperation between research groups and national research profiles. The methods are objective and repea
table, to be of practical value, however, the results must be applied to a complex reality (Ginman, 1987). The most used bibliometric methods in addition to the above mentioned are cocitation analysi
s, bibliographic coupling and coword analysis.
4.1 Cocitation analysis
Citations are often used in bibliometric analyses, and they are also the base for cocitation analysis and bibliographic coupling. In cocitation analysis the data compiled are counts of the number of
times two documents are jointly cited in later publications. The fact of having been cited together in the same new paper establishes a quantifiable link between the earlier papers, the strength of t
he link depending upon the number of times that a pair of documents are cited together. Cocitation analysis can also be based on authors or journals as units of analysis. Journals can thus be used fo
r studying the organisation of a subject literature through cocitation analysis. Cocitation of the published articles link the journals in which they were published and the journal title then represe
nts the subjects of all articles included (McCain, 1991).
4.2 Bibliographic coupling
In bibliographic coupling the hypothesis is that two articles which both cite the same previously published article have something in common. Analysis of the bibliographic coupling results in cluster
s of citing documents, when the cocitation analysis groups cited documents. In bibliometric terminology the citing articles create a research front, when a cluster of cited documents is called an int
ellectual base (Persson, 1994).
4.3 Coword analysis
Coword analysis is based upon the analysis of the co-occurence of the keywords used to index articles and other documents. This method emphasises the existence and evolution of networks of problems (
so called problematic networks) (Courtial, 1984). The method is useful for mapping the content of research in a field.
5. DESIGNING A COURSE IN BIBLIOMETRICS
A course in bibliometrics in the library and information science program can be divided in two parts.
- Seminars, where the methods and the results of the studies are discussed.
- Empirical work in the frame of a project, which could be for instance collection development in an interdisciplinary field.
An example of a course in bibliometrics is given in the following:
The two parts of the course can run parallell. The first part of the course is seminars in groups of at most eight students, where the participants prepare abstracts on literature given by the teache
r about the bibliometric methods and discuss them. The empirical part of the course is built on knowledge of online searching in databases, text processing software and software for cocitation analys
is and bibliographic coupling. The material used for the analyses is a collection of about 100 references downloaded from the citation indexes Science Citation Index (the database Scisearch) or Socia
l Science Citation Index (SSCI) by known local authors or local organisations in the field with time limitations for restricting the number of references. The format chosen for output is one with tag
ged fields. Using text processing software (e.g. WordPerfect) the citations of the articles are then analysed in different ways:
- The cited sources are sorted and analysed according to type of publication: journal article, book or other. This part of the work includes problems such as identifying journals and different spel
lings of their names. This has to be done manually, but gives a good training in identifying different types of material and emphasizes the importance of standardization of name abbreviations. The pe
rcentage of different types is calculated and gives information about the most important types of literature in the field.
- In many subject fields journals are the most used type of literature and analysis of them is important. The cited journals are sorted according to frequency of use, and in this way names of the c
ore journals are received.
- The scattering of articles to different journals according to Bradford's law is analysed on the basis of the frequency list. The cumulated numbers of articles and journals are calculated and plot
ted on a linear-logarithmic scale, from which the Bradford zones can be calculated. The number of journals needed to cover about one third, two thirds or all the relevant articles in the field are id
- The relations between the journals are studied with a cocitation analysis. A software for the analysis of (small volumes) of references developed at Umea university, BIBMAP, can be used for clust
ering the cocited journals (Persson, 1994). To visualize the internal relations in the clusters they are further analysed with a software for multidimensional scaling (MDS). MDS describes similaritie
s between two concepts in a group as distances in a map so that the greater distance between the concepts, the smaller is their similarity. The closer two journals are, the higher is the frequency of
cocitations. A journal's placement in the map reflects the congruence between the content of that journal's articles and the research interest of authors publishing in the field represented by the s
et of journals studied. Journals with high mean cocitation rates may be considered the more significant titles (McCain, 1991).
- The knowledge structure of the subject field is mapped with bibliographic coupling. This is done by sorting the citations according to the authors and again manually identifying manually differen
t spellings and ways of writing the names. The resulting frequency list of cited documents according to authors is the base for analysing the bibliographic coupling. The same computer programs can be
used for this analysis as for the cocitation analysis. The resulting clusters show documents which are related. The themes of the individual clusters can be identified and named and together these t
hemes form the research front of the field.
- A coword analysis, which also shows the knowledge structure is done in the same way as in the bibliographic coupling, but instead of citations index terms are used. These terms can be taken from
the titles of the publications and substitute the citations in the SciSearch-references. Also in this case the resulting clusters create themes which together form the research front of the field.
The results of the analyses have to be interpreted in relation to other similar studies and should therefore be followed up by parallell seminars, where the students read and discuss literature descr
ibing bibliometric studies and try to understand their own results in relation to them. The two methods of mapping the research front are compared and discussed in the seminars.
A project such as this will give the students knowledge about how to gather information for collection development in a new, interdisciplinary field. The students will in the future use electronic ne
tworks to a large extent, they will be familiar with online databases and electronic publications. The barrier towards bibliometrics will vanish, because the studies will be easier to perform. An imp
ortant task for the educational institutions besides teaching the methods will be to teach how the results should be interpretated and implemented in the daily work of the librarian.
Baughman, J.C. Toward structural approach to collection development.
College & Research Libraries 38 (3): 241-248, 1977
Broadus, Robert N. Some notes on research in bibliometrics.
Journal of Education for Library and Information Science 28 (2): 152-153, 1987
Courtial, J.-P., Callon, M. Is indexing thrustworthy? Classification of articles through co-word analysis. Journal of Information Science 9: 47-56, 1984
Ginman, Mariam. Lagar & regelbundenheter inom litteraturen.[Laws and regularities in literature]. In: Bibliometri. Abo: Abo Akademi University, 1986.
Publication/Department of Library and Information Science. 1. P. 1-16. [in swedish]
McCain, Katherine W. Mapping economics through the journal literature: An experiment in journal cocitation analysis.
Journal of the American Society for Information Science 42 (4): 290-296, 1991
Persson, Olle. The intellectual base and research front of JASIS 1986-1990. Journal of the American Society for Information Science 45 (1):31-38, 1994
von Ungern-Sternberg, Sara. Verktyg för planering av tvärvetenskaplig informationsförsörjning. En tillämpning på ämnesområdet bioteknik i Finland.[A tool for
planning the information provision in an interdisciplinary field. An application on the field of biotechnology in Finland]. Abo: Abo Akademis förlag, 1994. 318 pp. [in swedish]
Sara von Ungern-Sternberg, PhD,
Abo Akademi University,
Department of Library and Information Science,
Henriksgatan 9, FIN-20500 Turku, FINLAND