1. Introduction
In April 1986, on the occasion of her visit to the
Science Dynamics Department of the University of Amsterdam, Dr. S. Cozzens of
the National Science Foundation raised the question of whether it would be
possible to use the instruments for the measurement of relations between
journals as described in our article on 'The Development of Frames of
References'1, to explore the following questions in the area of
oceanography:
1.Is there a specific journal structure for
oceanography, and if so, can this structure be shown to change over time?
2.Is the growing international cooperation in the
exploitation of deep sea resources reflected in these journals as an increase
in joint scientific publications among research institutes in different countries?
The two questions are connected, because if the relevant
journal set changes over the years, this should have implications for the
relevant 'universes' to examine for international 'co‑ authorships'.
In this report, it will be argued that the pattern of
aggregated journal‑journal citations in 'oceanography' have changed, as
measured with the data in the Science Citation Index, particularly in
the 1970s. In the 1980s the journal structure has become almost completely
stable.
However, since 1977 the (underlying) factor structure of
inter‑journal citations in this and related fields has been rather
stable, and hence a core journal set can be identified in order to follow
international co‑authorship between France and California between 1974‑1985.
(California was chosen because of the location of the Scrips Institute in La
Jolla.)
From the results of this analysis, we will conclude that
this type of international co‑authorship happens occasionally throughout
the whole period at a frequency of about two times per year. These results are
then compared with general frequencies of international co‑authorship
between France and California.
2. The journal‑journal structure
2.1. Methods
The Journal Citation Reports of the Science
Citation Index are used to determine the evolution of the journal
structure. These contain between 2450 and 3300 journals for the relevant years
(1974‑ 19842), with an annual turn‑over of approximately
7%. 3
As core journal we selected the Journal of Physical
Oceanography, because in more recent years this journal has become central
to the field (see fig. 1 for the 1984 main journals citing this journal,
and being cited by this journal). (Initially, the Journal of Geophysical
Research was considered as a core journal for the analysis as well‑‑
at the suggestion of dr. Cozzens‑‑ but this journal was finally not
selected because it is closely linked with journals in other areas as well.)
The analysis was performed as described in the above‑
mentioned article 4, namely as follows:
1.From the entry journal the three journals which cite
that journal most heavily were listed, and then this procedure was reiterated
for the new batch of journals either until the method converged or until we
diverged more than two arrows from our entry. Subsequently, the same procedure
was followed for 'citing' as well, and by combining the two lists a
comprehensive journal list for that specific year was created.
2.For each year, a journal‑journal citation matrix
with numeric values was constructed. Correlation matrices were computed for
each matrix and its transposition, and used for factor‑analysis and Q‑factoring
respectively.
3.The sum‑value of mutual citations between two
journals was used for a multidimensional scaling representation. In these
journal‑maps the relevant structures were drawn in accordance with the
results of the factor analyses.
4.Additionally, a dendogram was produced from the
cluster analysis of the journals, using Wards' method of analysis on a Pearson
correlation matrix5. The dendograms provide information about the
level at which the journals belong together.
Qualitative analysis (ad 1.) was pursued for 1974, 1977,
1981, and 1984; Quantitative analyses (ad 2, 3, and 4) for 1974, 1977, and 1984
only.
2.2. Results
In 1974, the Journal of Physical Oceanography was
not yet included in the Journal Citation Reports (although it is
included in other sections of the Science Citation Index for that year.)
Therefore, Deep‑Sea Research, which in later years was closely
linked to the Journal of Physical Oceanography, was taken as the entry
journal.
The qualitative pictures for 1974 (fig. 2), show
clearly that although Deep‑Sea Research was among others directly
linked to the leading journal in geophysical research (the Journal of
Geophysical Research), its relations are much more scattered. The inclusion
of the Journal of Physical Oceanography in 1977 only makes it possible
to identify a definite cluster of journals on 'oceanography' from that year
onwards.
Factor‑analysis for 1977 (table 1) reveals
'geophysics' as a first factor accounting for 31.8% of the common variance. As
a second factor (27.8%), 'atmospheric sciences' seem of major importance for
this field, being linked to the 'geophysics' cluster by the Journal of Geophysical
Research‑ Oceans and Atmospheres, a special section of this journal
which in later years was not included separately in the Journal Citation
Reports. These titles suggest a predominance of attention (and mutual
citations) for what we might call 'surface‑problems'.
Only the third and the fourth factor (16.7% and 12.3%
respectively) consist of journals which have the word 'oceanography' in their
titles. The third cluster is the biological part of the field, while the fourth
is 'physical oceanography'. Finally, in 1977, a fifth cluster can be discerned
corresponding to 'fluid mechanics' (11.4% of the common variance).
Figure 3 provides a multidimensional scaling
representation of this journal‑journal citation matrix. The stress is
acceptable (dhat= 0.11). Figures 4 and 5 show the results of the cluster
analysis over the 'being cited' patterns and the 'citing' patterns
respectively, clearly indicating the changing position of the interdisciplinary
character of GP7 (JGP‑ Oceans and Atmospheres) as the main
difference between these two perspectives. For reasons explained in detail in
'The Development of Frames of References', we will focus on the perspective of
regarding the volumes as cases and citations to them as variables6.
In 1984, the number of relevant journals, according to
the methods described above, has declined from 16 to 9. So has the number of
factors, if we take as the cut‑off criterion that a factor should explain
more of the common variance than an average variable. Three instead of five
factors emerge; hence the reduction in complexity is about the same, resp. 3/9
versus 5/16).
The factor matrix for 1984 is given in table 2.
Despite the fact that the Journal of Physical
Oceanography (OC1) was taken as the entry, 'physical oceanography' is still
a minor (third) factor in this matrix, explaining only 16.9% of the common
variance.
Of greater significance are the other factors of
'oceanography' (with the biological journal Limnology and Oceanography
correlating negatively with the Journal of Physical Oceanography)
and 'geophysical research'. The dendogram and the MDS‑ map are given in
figures 6 and 77.
2.3. Conclusions
Although differences appear in the journal‑journal
structures over the years, mainly in the degree of relatedness to other areas
such as atmospheric sciences, there is also a great resemblance between the
1977 and 1984 structures.
Between 1974 and 1977 there are important differences,
but these may in part be a consequence of the inclusion of the Journal of
Physical Oceanography, which in itself, however, is a major fact.
Moreover, the impact‑factors for the journals
involved (table 3) are also rather stable. Therefore, it seems reasonable in
this case to work with a constant journal set if one wants to measure another
variable such as, in this case, international co‑authorships.
Section 3 describes this constant journal set.
The important factors in the area as measured by journal‑journal
citations are:
1. geophysics
2. oceanography‑ biologically oriented
3. oceanography‑ physically oriented
4. atmospheric sciences
5.Science and Nature, which also play a
role in some of the intervening years. .pa
3. The constant journal set
3.1. Methods
The constant journal set was constituted on the basis of
the journals involved in the qualitative analyses of 1974, 1978, 1981 and 1984,
on the one hand, and the results of the various factor‑analyses on the
other. Science and Nature were added because they do play a role
in some years, and they provide a clear point of reference.
We think that the following list can be used for a
longitudinal analysis of other variables in this field:
1. Journal of Physical Oceanography OC1
2. Journal of Marine Research OC2
3. Deep‑Sea Research OC3
4. Limnology and Oceanography OC4
5. Marine Biology OC5
6. Journal of Geophysical Research GP1
7. Geophysical Journal of the Roy.Astr.Soci. GP2
8. Planetary and Space Sciences GP3
9. Geophysical Research Letters GP4
10. Journal of Atmos. Terr. Physics GP8
11. Journal of Atmospheric Sciences AS1
12. Monthly Weather Review AS2
13. Nature
14. Science
As we have noted, the Journal of Physical
Oceanography was not yet included in 1974, and the same holds true for Geophysical
Research Letters.
Matrices were composed for these 14 (12) journals, and
all the analyses mentioned in section 2.1. were carried out.
3.2. Results
The journal‑journal citation matrices between
these years are highly comparable. Table 4 shows the Pearson correlations
between these matrices for different years. The 1981‑ and the 1984‑matrix
correlate above .99.
Despite the much lower correlation between 1977 and 1984
(.69), the factor structure for these two years is almost identical, as is
illustrated by tables 5 and 6. The only difference is that 'physical
oceanography' in 1984 explains 12.2% of the common variance of this matrix
(factor 4), as against 10.0% in 1977 (factor 5). This increase has been at the
expense of 'atmospheric sciences', which explained about 3% less of the common
variance in 1984 than in 1977.
However, these differences are extremely small.
Figure 8 displays the superposition of the
multidimensional scaling solutions for 1977 and 1984, once more demonstrating
their great stability. Finally, the dendograms for 1974, 1977, 1981, and 1984
are also provided (respectively figures 9, 10, 11, and 12).
3.3. Conclusion
A fixed journal set for the area can be defined for the
years 1977‑1984. Some reservations have to be made for the validity of
this choice for the years 1974‑1977 because it cannot be checked in the JCR.
However, the high stability of the factor matrix in later years provides some
support for the validity of the inference that the journal set may be stable
over the whole period.
4. International co‑authorship
In this section we
address the question of whether the supposed internationalization of the field
can be measured by the number of articles with addresses from different
countries. The analysis has been limited to California, a US‑state with
major research facilities in the discipline, and to France as an exemplary
advanced country with a share in economic activities.
4.1. Methods
The analysis was done straightforwardly on‑line on
DIALOG. The intersection of Californian AND French addresses for each year was
combined with an AND for the relevant journals. This order in the search was
chosen to enable us to compare the number of co‑ authored articles with
general trends in co‑ authorship among researchers with Californian and
French addresses.
For reasons of cost‑effectiveness, only the
following eight journals representing the core of the field were used:
1. Journal of Physical Oceanography (OC1)
2. Journal of Marine Research (OC2)
3. Deep‑Sea Research (OC3)
4. Limnology and Oceanography (OC4)
5. Journal of Geophysical Research (GP1)
6. Geophysical J. of the Roy. Astr. Soci. (GP2)
7. Planetary and Space Sciences (GP3)
8. Geophysical Research Letters (GP4)
Hence Marine Biology (OC5), the Journal of
Atmospheric and Terrestrial Physics (GP8), the two 'atmospheric sciences'
journals, were left out, as were Science and Nature.
In this case the Journal of Physical Oceanography
posed no problems, because it had been included in the 1974 Science Citation
Index although not in the Journal Citation Reports of that year. The
Geophysical Research Letters are only included from 1975 onwards.
It should be noted that these years relate not to
"publication years" but to the dates‑of‑entry as
indicated by ISI for the DIALOG‑installation of the SCI.
4.2. Results
Co‑authorship on
institutional addresses in France and California does exist, but represents
less than 1% of both French and Californian publications. Table 7 presents the
total numbers and the respective percentages. However, the percentage is going
up: for France, from 0.42% in 1974 to 0.93% in 1985, and for California from
0.31% to 0.84% in 1985.
The fifth column of this table lists the number of co‑authorships
with these two addresses for the eight journals involved. It is clear that
these figures are extremely low, and there is no substantial increase over
time. Moreover, all co‑ authored articles are in 'geophysical' journals,
as can be seen from the latter column.
International co‑authorship in this field seems
rather to be something which happens from time to time, particularly in
geophysics8. We found no such co‑authorships in the
oceanography‑cluster at all, and hence also no increase.
The full bibliographic details of the articles
(including the OATS‑nrs for ordering) are given in Appendix B.
4.3. Conclusion
It is not possible to measure quantitatively the
'internationalization of oceanography' in terms of co‑authorships from
researchers with institutional addresses in different locations, in this case
France and California (US).
However, an increasing French participation can be
observed over the years in the universe of journal sets employed here (table
8). French participation nearly doubled between 1974 and 1985, against a more
modest growth from almost 4 to almost 5% in the total French share of world
publications.
Californian participation has always been high in this
field (as compared to an overall share of world publications of about 5.2%),
but tends to diminish somewhat in more recent years. (Of course, this effect
could have other causes as well.)
5. Summary
With regard to the original questions raised in section
1 of this report, namely
1. whether a specific journal structure for oceanography
can be identified, and if so, whether this structure can be shown to change
over time? and
2.whether growing international cooperation in the field
is reflected in a rise in scientific publications with addresses both in France
and in California?
we can now conclude as follows:
1.A journal structure emerges more stable during the mid
1970s, partly because important journals in this field are taken up in the Science
Citation Index, partly because the already present underlying citation
structures become stronger. This process comes to an end by 1981.
However, already for 1974 a dendogram with the full
structure could be produced as well (figure 8).
2.Because of the presence of a journal structure over
the whole period, in this case it is possible to identify a fixed journal set,
and hence to control for this variable while varying another. Such a fixed
journal set has been determined, and has been found to be extremely stable over
time in its structure, although some variance in other aspects can be found in
the earlier years (1974‑ 1981).
3.Eight journals were
chosen from this set to investigate in terms of internationalization as
operationalized by sharing both a French and a Californian address in one
record of the Science Citation Index.
This form of
operationalization proved unworkable in this case.
However, internationalization can be measured if operationalized
in a different way. French participation in the relevant journal set almost
doubled in the course of a decade, against some decline of Californian
participation (which is traditionally relatively high!) between 1982 and 1985.
Notes and References
1.L. Leydesdorff, 'The Development of Frames of
References', Scientometrics, 9 (1986) 103‑ 125.
2. At the time of this analysis (April 1986) the JCR for
1985 has not yet been issued.
3.H. R. Coward, J. J. Franklin, L. Simon, ABRC
Science Policy Study: Co‑Citation Bibliometric Models, Center for
Research Planning, Philadelphia, Penn., 1984, p. 4. For my critique of the use
of a fixed journal set, see: L. Leydesdorff, 'Increases in British and Dutch
Scientific Performance', Nature (forthcoming).
4. Leydesdorff, 1986.
5.L. Leydesdorff, 'Different Methods for the Mapping of
Science' (in preparation).
6. Leydesdorff, 1986, p. 106.
7.When we transpose the matrix, the pattern of 'being cited'
(as opposed to the 'citing') pattern reveals a remarkable‑‑
negative!‑‑ relation between Limnology and Oceanography and Nature,
which may also be a result of other effects, such as being heavily cited. See
also Leydesdorff, 1986, p. 106.
8.About geophysics see, e.g.: C. S. Gillmor and C. J.
Terman, 'Communication Modes of Geophysics: The Case of Ionospheric Physics', EOS,
54 (1973) 900‑08; C. S. Gillmor, 'Citation Characteristics of the JATP
Literature', Journal of Atmospheric and Terrestrial Physics, 37 (1975)
1401‑04.
APPENDIX A
OC1 Journal of Physical Oceanography
OC2 Journal of Marine Research
OC3 Deep‑Sea Research
OC4 Limnology and Oceanography
OC5 Marine Biology
GP1 Journal of Geophysical Research
GP2 Geophysical J. of the Royal Astron. Society
GP5 Geological Society of America Bulletin
GP6 Bulletin of the Seismological Society of America
GP7 Journal of Geophysical Research‑ Oceans and Atmospheres
FM1 Journal of Fluid Mechanics
FM2 Physics of Fluids
FM3 Proceeding of the Roy. Soci. of London Series A
FM4 AIAA Journal
AS1 Journal of Atmospheric Sciences
VAR1 Journal of the Acoustical Society of America
Legend_to_figures_3,_4_and_5
1. Journal of Physical Oceanography OC1
2. Journal of Marine Research OC2
3. Deep‑Sea Research
OC3
4. Limnology and Oceanography OC4
5. Marine Biology OC5
6. Journal of Geophysical Research GP1
7. Geophysical J. of the Royal Astron. Society GP2
8. Geological Society of America Bulletin GP5
9. Bulletin of the Seismological Society of America GP6
10. Journal of Geophysical Research‑ Oceans and Atmospheres
GP7
11. Journal of Fluid Mechanics FM1
12. Physics of Fluids FM2
13. Proceeding of the Roy. Soci. of London Series A FM3
14. AIAA Journal FM4
15. Journal of Atmospheric Sciences AS1
16. Journal of the Acoustical Society of America VAR1
Legend_to_figure_1
OC1 Journal of Physical Oceanography
OC2 Journal of Marine Research
OC3 Deep‑Sea Research
OC4 Limnology and Oceanography
GP1 Journal of Geophysical Research
GP2 Geophysical J. of the Royal Astron. Society
GP3 Planetary and Space Sciences
GP4 Geophysical Research Letters
NATURE
Legend_to_table_2
OC1 Journal of Physical Oceanography
OC2 Journal of Marine Research
OC3 Deep‑Sea Research
OC4 Limnology and Oceanography
GP1 Journal of Geophysical Research
GP2 Geophysical J. of the Royal Astron. Society
GP3 Planetary and Space Sciences
GP4 Geophysical Research Letters
NATURE
Legend_to_figures_6_and_7
1. Journal of Physical Oceanography OC1
2. Journal of Marine Research OC2
3. Deep‑Sea Research OC3
4. Limnology and Oceanography OC4
5. Journal of Geophysical Research GP1
6. Geophysical J. of the Royal Astron. Society GP2
7. Planetary and Space Sciences GP3
8. Geophysical Research Letters GP4
9. Nature NATURE
Table_3
IMPACT FACTORS FOR RELEVANT JOURNALS IN DIFFERENT YEARS
1974
1977 1984
OC1 Journal of Physical Oceanography ‑‑
1.750 1.625 OC2 Journal of Marine Research 0.574
1.945 3.237 OC3 Deep‑Sea Research
1.739 1.663 1.568 OC4 Limnology and Oceanography
1.487 1.937 2.628
OC5 Marine Biology 0.858
1.292 1.811
GP1 Journal of
Geophysical Research 2.536 3.552 2.393 GP2
Geophysical J. of the Royal Astron. Society 1.691 2.003 1.668
GP3 Planetary and Space Sciences 1.645
1.381 1.388
GP4 Geophysical Research Letters ‑‑
2.395 2.091
GP5 Geological Society of America Bulletin 1.674
1.192 2.554 GP6 Bulletin of the Seismological Society of America 1.150
0.787 1.744
GP7 Journal of Geophysical Research‑ Oceans and Atmospheres ‑‑
0.263 ‑‑
GP8 Journal of Atmospheric and Terrestrial Physics 1.322
1.161 0.886
AS1 Journal of
Atmospheric Sciences 2.051 2.049 2.192
AS2 Monthly Weather Review 0.873
1.020 1.741
Nature 3.636 4.957
10.248
Science 5.412
5.745 8.209
Legend_to_tables_5_and_6,_and_to_figures_8,_9,_10,_11,_and_12
Journal of Physical Oceanography OC1 1 (not in 1974)
Journal of Marine Research OC2 2
Deep‑Sea Research OC3 3
Limnology and Oceanography OC4 4
Marine Biology OC5 5
Journal of Geophysical Research GP1 6 (not in 1974)
Geophysical Journal of the Roy.Astr.Soci. GP2 7
Planetary and Space Sciences GP3 8
Geophysical Research Letters GP4 9
Journal of Atmos. Terr. Physics GP8 10
Journal of Atmospheric Sciences AS1 11
Monthly Weather Review AS2 12
Nature 13
Science 14
Table_7
Co‑authored articles with French and Californian adresses
total % Fr. % Ca. in sample journals
(1) (2) (3) (4)
1974 68 0.42% 0.31% 1 J. Geophys. Res.
1975 80 0.37 0.35 1 Geophysical J. of the
Roy. Astr. Soci.
1976 112 0.43 0.38 4 J. Geophys. Res.
(2x); Geophys. J. Roy. Astr. Soci.
(1x);
Geophys. Res. Letters (1x)
1977 131 0.50 0.46 2 J. Geophys. Res.;
Planet. & Space Sci.
1978 150 0.61 0.50 3 J. Geophys. Res. (2x);
Geophys. J. Roy. Astr. Soci.
1979 149 0.61 0.53 1 Geophys. Res. Letters
1980 180 0.59 0.57 2 J. Geophys. Res. (2x)
1981 206 0.73 0.64 3 J. Geophys. Res. (2x);
Geophys. J. Roy. Astr. Soci.
1982 236 0.77 0.71 2 J. Geophys. Res. (2x)
1983 249 0.83 0.75 1 J. Geophys. Res.
1984 294 0.96 0.86 4 J. Geophys. Res (2x);
Geophys. Res. Letters (2x)
1985 318 0.93 0.84 0
(24)
1) total number of publications with a French and Californian
address in that year (all of the SCI).
2) 1) as a percentage of the total number of publications with a French
address
3) 1) as a percentage of the total number of publications with a
Californian address
4) number of co‑authored publications in the following journals:
Table_8
Publications with a French or Californian address
in the eight1 central journals
total France California
in central
journals N % N %
1974 1464 28 1.92% 233 15.91%
1975 1306 27 2.06% 223 17.07%
1976 2013 54 2.68% 394 19.57%
1977 1707 43 2.51% 289 16.93%
1978 1078 29 2.69% 311 28.85%
1979 1388 40 2.88% 242 17.44%
1980 1473 45 3.05% 271 18.40%
1981 1718 63 3.67% 324 18.86%
1982 1466 43 2.93% 286 19.51%
1983 1576 63 4.00% 259 16.43%
1984 1694 67 3.96% 222 13.10%
1985 1144 37 3.23% 155 13.55%
1.
1. Journal of Physical Oceanography (OC1)
2. Journal of Marine Research (OC2)
3. Deep‑Sea Research (OC3)
4. Limnology and Oceanography (OC4)
5. Journal of Geophysical Research (GP1)
6. Geophysical J. of the Roy. Astr. Soci. (GP2)
7. Planetary and Space Sciences (GP3)
8. Geophysical Research Letters (GP4)
Legend_to_figure_2
OC2 Journal of Marine Research
OC3 Deep‑Sea Research
OC4 Limnology and Oceanography
OC5 Marine Biology
MB1 Journal of Experimental Marine Biology and Ecology
MB2 Journal of the Marine Bilogical Association of the U.K.
JWPC Journal of the Water Pollution Control Federation
PHYC Journal of Phycology
FISH Journal of the Fisheries Research Board of Canada
GP1 Journal of Geophysical Research
GP3 Planetary and Space Sciences
GP8 Journal of Atmospheric and Terrestrial Physics
GP9 Reviews of Geophysics and Space Physics
SCI Science
PNAS Proceedings of the National Academy of Sciences USA
NATU Nature
JBC Journal of Biological Chemistry
