return
Citation environment of Angewandte
Chemie
CHIMIA 61(3), 104-109, 2007.
Lutz
Bornmann*, Loet Leydesdorff, and Werner Marx
Abstract
Recently, aggregated journal-journal citation networks were
made accessible from the perspective of each journal
included in the Science Citation Index (see
http://www.leydesdorff.net/). The local matrices can
be used to inspect the relevant citation environment of a journal using
statistical analysis and visualization techniques from social network analysis.
The inspection gives an answer to the question what the local impact of this
and other journals in the environment is. In this study the citation
environment of Angewandte Chemie was analysed. Angewandte Chemie is
one of the prime chemistry journals in the world. Its
environment was compared with that of the Journal of the American Chemical
Society. The results of the environment analyses give a detailed insight
into the field-embeddedness of Angewandte Chemie.
The impacts of the German and international editions of this journal are
compared.
Key words
Angewandte Chemie,
citation environment, citation network, local impact, Journal Citation Reports
1
Introduction
In 1976 Eugene
Garfield – the founder of the Institute for Scientific Information (ISI,
now Thomson Scientific, Philadelphia, PA, USA) – introduced
the Journal Citation Reports (JCR) as an instrument to evaluate the
significance of scholarly journals.[1] Today, the most important
journals (currently about 7,500 journals from more than 3,300 publishers in
over 60 countries) are listed in the JCR with a series of bibliometric data and
indicators (e.g., total citations, Journal Impact Factor, Journal Immediacy
Index, Journal Cited Half-Life). Through the publication of the indicators,
essentially the Journal Impact Factor (JIF), JCR has become an authority for
evaluating scholarly journals.[2, 3] To get from the JCR
information on journals within disciplines for interdisciplinary comparison,
each journal is classified by using 172 subject categories. For chemistry, the
categories analytical, applied, inorganic/ nuclear, medicinal,
multidisciplinary, organic, and physical are used (Angewandte Chemie –
International Edition, Angew Chem Int Edit, belongs to
multidisciplinary).
The JCR are an
index of journal-journal links based on a grouping and summation of condensed
citations using the journal name as the sorting key.[1] For each journal, there are
two basic print-outs – cited and citing journals. Cited journal data
show the number of times papers (original research and
review articles) published in a certain year (in journals
covered by JCR) cited papers (original research and review articles) published in a certain journal. Citing journal data show the number of times papers published in journals (covered by JCR)
were cited in a certain journal in a certain year. Both data combined define a huge matrix of cited
and citing journals.[4] From this huge matrix, local matrices can be extracted. Recently, local matrices were made accessible from the perspective
of each journal covered by JCR (in 2004 from the perspective of 7,379 journals,
see http://www.leydesdorff.net/jcr04). The local matrices can be used to inspect the relevant citation
environment of one journal (i.e., the aggregated journal-journal citation network from
the perspective of one journal) by using statistical analysis and visualization
techniques from social network analysis.[5] The inspection may give answers to the following
questions: what is the scientific impact of this and other journals in the
environment (defined each by the number of published papers and their
citations)? What is the share of the journal’s self-citations among their local
impact? Furthermore, and most importantly, environments can be divided in
field-specific clusters by means of a journal’s citation patterns; relationships
between journals and clusters can be identified.
In this study the citation environment of Angew Chem
(we refer to the German and international Angewandte Chemie
editions with this abbreviation) was analyzed. Angew Chem is one
of the prime chemistry journals in the world, with a JIF currently at 9.596
(JCR for 2005). The environment of Angew Chem was
compared with that of the Journal of the American Chemical Society (J
Am Chem Soc). J Am Chem Soc also belongs to multidisciplinary
chemistry in the JCR with a JIF currently at 7.419 (JCR for 2005). Both Angew Chem and J Am Chem Soc are written
and read primarily for the purpose of communication of original research
findings.
2
Data and descriptive statistics
The data for the analysis of the Angew
Chem environment originated from the 2004 journal-journal citation database
of the JCR. In addition to the JCR data, we extracted the number of citations
that Angew Chem received from other journals from the Expanded
Science Citation Index at Web of Science (Thomson Scientific). Angew Chem has been published since 1962
in the form of both a German and an international edition, which has led to a
number of problems with respect to citation analyses.[6, 7] Since Angew Chem
papers are contained in different volumes and appear on different page numbers
in the two editions, these papers are counted as if they were two completely
different publications.[6] References to papers in Angew
Chem relate to either one or the other edition or even both editions
(double-citations), as practiced, e.g., for references in papers in Angew
Chem to other Angew Chem papers.[6] As double-citations lead to
an overestimation of the citations in JCR by the percentage of citations from
both issues of the same paper,[8] we used Angew Chem
data from the Expanded Science Citation Index
that are corrected for double-citations in this study.
The local
citation environment for Angew Chem was determined by including all
journals which cited the journal to the extent of 1% of its citation rate: 76,904
(total citations to Angew Chem Int Edit) – 7,512 (self-citations to Angew
Chem Int Edit) / 100 = 693 citations (on this, see the JCR for 2004). Among the 869 journals that cited Angew Chem Int Edit in
2004, only 22 cited it more than the 1% threshold of 693 times. Thus, the
distribution of citations to the journal is highly skewed.[9] From the 22 journals
that cited Angew Chem Int Edit more than 693 times all 852,008
references in 21,958 records were organized in a relational database. Among
these 852,008 citations, 41,495 (5%) are to “Angew Chem” (both editions). One hundred
and eleven of these records are incomplete (e.g., “In Press Angew Chem”);
32,416 (78%) of the remaining 41,495 records cited “Angew Chem Int Edit” and
the other 9,079 (22%) “Angew Chem” without the indication of the international
edition. We assume that this represents the German edition. In 8,054 cases, two
consecutive records in the same document cited both the German and the
international editions. When two citations follow upon each other, one to the
German and the other to the international edition and with the same author name
and publication year we identified these citations as double-citations. Among the
8,054 cases, we found 7,343 double-citations and 702 citations solely to the
German edition. On the basis of this calculation, the impact of Angew Chem
is overrepresented in this journal set (due to double-citations) to the extent
of 21.5%: [7,343 / (41,495 – 7,343)] * 100. In a previous study[6], an overestimation of the
(global) JIF with 15% was found, but the effect of ‘double-citations’ is larger
in the Angew Chem citation environment because journals published by
German publishers are overrepresented.
Table 1 shows the publisher and the subject categories of the
22 journals that determine the local citation environment of the Angew Chem
as well as the total numbers of papers published in 2004 by each of the 22
journals. Seven journals are published by the American Chemical Society
(Washington, DC, USA), five journals by Wiley - VCH Verlag (Weinheim, Germany),
and three journals by the Royal Society of Chemistry (London, UK) and Pergamon
(Amsterdam, The Netherlands), respectively. Most of the journals have been
categorized by Thomson Scientific and Ulrich’s International Periodical
Directory[10] in the fields of
multidisciplinary (five journals), inorganic/ nuclear (seven journals), and
organic (12 journals) chemistry (many journals have been categorized in more
than one field in both databases). Table 1 further shows
that the number of papers published by the 22 journals averaged out about 1,015
with the highest number for J Am Chem Soc (3,167) and the lowest for Chem
Rev (183).
Table 2 shows the number of papers published in 2004
by the 22 journals that cited papers published in Angew Chem
(international edition, German edition, sum of both editions (double-citations
included), and sum of both editions corrected for double-citations) in all
years. The separated numbers (row ‘Total citations’) indicate
that the international edition contributed about 70% (32,416) and the German
edition about 30% (9,079) to the total number of citations.[6] However, there is a large
amount of double-citations, especially produced by Angew Chem Int Edit (n=2,945)
and Chem-Eur J (n=1,823). The table (column ‘sum of both editions
corrected for double-citations’) shows that Angew Chem was cited 3,991 times (12%)
by papers published in the same journal. That means about one tenth of
citations from journals which contributed more than 1% to the Angew Chem
impact are those of authors who published in Angew Chem. This within-journal
citation rate of 12% is lower than, e.g., the rate of J Am Chem Soc (it
has a share of about 20% within-journal citations among those journals that
cited it more than 1% in the same year), but it is definitely higher than the
within-journal citation rate of, e.g., Chem Commun (with much less than
10%). Although within-journal
citations are often not self-citations by authors, they can be considered as an
indicator of the inwardness of a community supporting a journal.[11] As Table 2
shows, 14% of citations to Angew Chem resulted from papers published in J
Am Chem Soc and 7% from ones published in Chem-Eur J and J Org
Chem, respectively. The remaining 60% of citations were provided by the
other 18 journals listed in Table 2 (on average 1,138
citations per journal).
3
Results
For our set with
journals that cited Angew Chem to the extent of 1% of its citation rate
a 22 x 22 citation matrix was composed with two dimensions: one in the ‘cited’
(columns) and another in the ‘citing’ (rows). Both dimensions represent the
relevant Angew Chem citation environment (corrected for double-citations)
covering the (local) impact of the most important journals. Each journal is
represented in the matrix with the number of citations that it received from
each of the other 22 journals (‘cited’ dimension) and with the number of
citations that each of the other 22 journals received from this journal
(‘citing’ dimension). The matrix was (1) imported into SPSS – a statistical
analysis software[12] – for ‘decomposition’ of the
environment with a reduction scheme, and (2) read into Pajek – a program for
large network analysis[13] – for visualization of the
environment (the visualization is
based on using the algorithm of Tomihisa Kamada and
Satoru Kawai).[14]
‘Decomposition’ of the environment
With exploratory factor analysis
we tried to find a reduction scheme that indicates how the citation patterns
(‘being cited’) of the 22 journals in the Angew Chem citation
environment cluster or hang together. By the use of principal components in the
factor analysis our set of correlated variables (the journals’ citation
patterns) was transformed into a set of uncorrelated variables (components or
factors). The goal of this analysis was that a small number of components would
account for most of the variance (>75%) in the patterns.[15] One result of the calculated factor analyses with four components (eigenvalues
larger than 1) explained 80% of the variance and was taken as the final result
(see Table 3).
The
categorization of the journals among the components by using factor loadings
greater than .4 (these loadings are marked in Table 3)
showed that the journals can be categorized in terms of disciplinary
affiliations (see also Table 1). A first group
of journals (e.g., Tetrahedron Lett and Synlett) were cited in the
environment according to a pattern of organic chemistry (factor 1 explains 40%
of the variance in the matrix); a second and third pattern are specific for
multidisciplinary chemistry (e.g., J Am Chem Soc and
Angew Chem; also for applied chemistry: Adv Synth Catal) and
inorganic/ nuclear chemistry (e.g., Inorg Chem
and Z Anorg Allg Chem; factor 2 and factor 3 explain 25% and 9% of the
variance, respectively). A fourth group can be designated to organometallic
chemistry (e.g., Z Organomet Chem and Organometallics; factor 4
explains 6% of the variance).
As Dalton T and Eur J
Inorg Chem embrace both aspects of the chemistry of inorganic and
organometallic compounds, their citation patterns are not only specific for
inorganic but also for organometallic chemistry. Because of low and negative
factor loadings in Table 3, J
Phys Chem B and Org Biomol Chem could not clearly be designated to
one of the four factors. All in all, the results of the factor analysis show
that the Angew Chem citation environment is determined by citation
patterns that are characteristic for organic (factor 1), multidisciplinary
(factor 2), and inorganic/ nuclear (organometallic,
factor 3 and 4) chemistry journals.
Visualization
of the citation environment
Figure 1 shows the citation environment of Angew Chem
by using nodes for indicating a specific journal and solid lines for indicating
links between journals. The links are defined by a similarity measure: the
cosine coefficient.[16] This coefficient is
similar to the well-known Pearson’s correlation coefficient and specifies each
correlation between the citation patterns (‘being cited’) of
the 22 journals.[11] Links between journals
representing cosine values below .2
are suppressed in Figure 1 in
order to display only the most important correlations.
In agreement with the results of the factor analysis, the figure (in the
centre) shows that Angew Chem is strongly embedded in a core group of five
multidisciplinary chemistry journals (journals with red nodes). Organic and inorganic/ nuclear (together with organometallic) journals are connected to this core group at specific sites:
organic chemistry journals at the top of Figure 1
(journals with green nodes) and inorganic/ nuclear chemistry at the bottom
(journals with blue nodes). Also in agreement with the factor analysis
findings, J Phys Chem B appears in the Angew Chem environment as
a more or less isolated point. This journal presumably represents an
environment according to a citation pattern of chemical physics and physical
chemistry (surface chemistry, electrochemistry, and in particular the strongly
increasing nanoscience). Almost 40% of the J Phys Chem B papers
published in 2004 and citing Angew Chem (both editions) deal with topics
related to nanoscience. At the bottom of Figure 1, Z Anorg Allg Chem clearly
belongs to the inorganic journal group, but the node also appears somewhat
isolated. We assume that Z Anorg Allg Chem is still noticed in the scientific
community as one of the traditional German chemistry journals (and is not perceived
as an international chemistry journal), although the number of papers published
in English is currently larger than the number of papers published in German.
The shape of the
nodes in Figure 1 is used for indicating the percentage
contribution of each journal to the citation environment both including and
excluding within-journal citations: the greater the vertical size of one node,
the greater the share of received citations among the total number of citations
in the environment. The share of received citations is not only dependent on
the impact of the papers published by a certain journal but also by the total
number of papers in the journal published in 2004; both citations and number of
articles strongly correlate (Spearman’s rank correlation = 0.76). Whereas the
vertical size of the nodes in Figure 1 indicates the
share of citations including journal self-citations, self-citations are not
considered for the horizontal size. The more ‚stretched’ a node in the longitudinal
axis appears, the higher is the share of self-citations. By inspecting the shape of the nodes in the
figure one is able to see how much a journal is dependent on an inner-circle of
authors citing one another. Note that within-journal citations can be both
self-citations of authors and citations among authors publishing in the same
journal.[11]
As the results
in Figure 1 show, most of citations in the environment
account for J Am Chem Soc with a share of 24% (19% without within-journal
citations); about 8% account for Angew Chem (6% without within-journal
citations). The high citation share for J Am Chem Soc compared to Angew
Chem may especially be due to the fact that J Am Chem Soc published about
2.6 times more papers in 2004 than Angew Chem (see Table
1). Figure 1 further shows that J Org Chem
and Tetrahedron Lett, repectively, received a share of about 10% (including
within-journal citations). The shape of the nodes for these dominant journals
in the environment indicates that the number of self-citations does not weight
heavily: the citation patterns are
not dominated by within-journal citation rates. The
lowest number of citations in the Angew Chem citation environment
account for Org Biomol Chem (0.2% with as well as without within-journal
citations). If we sum the shares of the journals’ citations according to the
field categorizations in Table 3, the following
percentage values arise: 47% for multidisciplinary chemistry (citation rate
received by 7,411 papers published in seven journals), 38% for organic
chemistry (citation rate received by 8,801 papers published in nine journals),
and 12% for inorganic/ nuclear and organometallic chemistry (citation rate
received by 3,024 papers published in four journals).
For the purpose
of comparison with the Angew Chem citation environment, the environment
of J Am Chem Soc is shown in Figure 2. Among the 1,566 journals which cited J Am Chem Soc in 2004 at least one
time, 21 cited it above the citation threshold of 1%. Whereas sixteen of the 21
journals appear also in the citation environment of the Angew Chem (see Figure 1), six journals drop out. Except for one journal (Tetrahedron-Asymmetr),
these journals are published by German publishers (Adv Synth Catal, Eur
J Inorg Chem, Synlett, Synthesis-Stuttgart, and Z Anorg Allg Chem;
see Table 1). Five journals in the citation environment of J Am Chem Soc
are not able to exceed the 1% threshold value in the Angew Chem
environment: Journal of Physical Chemistry A (J Phys Chem A), Journal
of Chemical Physics (J Chem Phys), Macromolecules, Biochemistry
and Langmuir.
If we compare
the visualizations for Angew Chem and J Am Chem Soc in Figure 1 and Figure 2 it is clearly visible that both citation environments are
dominated by a strong core group of nearly the same journals with similar
citation patterns. Journals which are each dropped out in the other citation
environment are positioned in both figures primarily in the periphery around
the core group. In the Angew Chem environment, the journals are to be
found at different points in the periphery. However, in the J Am Chem Soc
environment, these journals form a separate group beside the strong core group at
the top in Figure 2. This means that these journals are
characterized by similar citation patterns which are (more or less) different
from those of the core group (except for Biochemistry). According to the
subject categories provided by JCR and Ulrich’s International Periodical
Directory,[10] J Chem Phys, J
Phys Chem A, and J Phys Chem B belong to the fields of physical
chemistry and chemical physics. Langmuir is also strongly related to
physical chemistry (e.g., surface and interface chemistry/ physics) and Macromolecules
includes papers on surface properties of polymers.
4
Discussion
The JCR publish
annual updated publication and citation data for measuring the performance of
journals.[9] In this study the relevant
citation environment of Angew Chem was analyzed by using these data
(corrected for double-citations). The most important journals and their impact
as well as field-specific journal-clusters were identified in the environment
by means of statistical analysis and visualization techniques. The results of the
environment analyses give a detailed insight into the field-embeddedness of Angew
Chem. As data source JCR data for the year 2004 were analyzed. We assume
that the results presented here for 2004 have time stability and can be extrapolated
on ‘flanking’ years because a series of studies have shown remarkable time
stabilities in journal citation patterns.[1, 17]
The
‘decomposition’ of a citation environment remains sensitive to the choice of
the seed journal (the ‘point of entrance’). Journal citation patterns span a
multi-dimensional space in which clouds can be distinguished, but the
delineation of these clouds at the edges remains fuzzy and varies with the
perspective chosen by the analyst.[18] Therefore, the analyses of
(inter-)disciplinary links between and among certain citation environments will
be topics of great interest in future journal citation network analyses.
References
[1] E. Garfield, Nature 1976, 264, 609.
[2] M.-H. Magri, A. Solari, Scientometrics 1996,
35, 93.
[3] M. Amin, M. A. Mabe, Med. Buenos Aire 2003, 63,
347.
[4] L. Leydesdorff, J. Doc. 2004, 60,
371.
[5] E. Otte, R. Rousseau, J. Inform. Sci. 2002, 28,
441.
[6] W. Marx, Angew. Chem. Int. Edit.
2001, 40,
139.
[7] R. A. Buchanan, Coll. Res. Libr. 2006, 67,
292.
[8] T. Braun, W. Glänzel, Chem. Intelligencer 1995,
31.
[9] E. Garfield, Science 1972, 178, 471.
[10] Ulrich's Periodicals Directory, 'The global source for
periodicals', R.R. Bowker, New Providence, NJ, USA, 2006.
[11] L. Leydesdorff, J. Am. Soc. Inf. Sci. Technol. accepted
for publication.
[12] SPSS for Windows, 'Rel. 14.0.1. 2006', SPSS Inc., Chicago,
USA, 2006.
[13] V. Batagelj, A. Mrvar, in 'Pajek - analysis and
visualization of large networks' (Eds. M. Juenger, P. Mutzel), Springer (series
Mathematics and Visualization), Berlin, Germany, 2003, p. 77.
[14] T. Kamada, S. Kawai, Inform. Process. Lett. 1989, 31, 7.
[15] J. Stevens, 'Applied multivariate statistics for the social
sciences', Lawrence Erlbaum Associates, London, UK, 1996.
[16] G. Salton, M. J. McGill, 'Introduction to modern
information retrieval', McGraw-Hill, London, UK, 1983.
[17] E. Garfield, Scientist 1991, 5, 11.
[18] R. Goldstone, L. Leydesdorff, Cognit. Sci. 2006,
30, 983.
