Demand for scientific articles and citations

An example from the Institut de l'information scientifique et technique (France)

 

Jean-Michel Salaün, Gresi[1], salaun@enssib.fr

Thierry Lafouge, Recodoc[2], lafouge@enssib.fr

Cherifa Boukacem, Gresi, cboukace@enssib.fr

 

 

 

Abstract

 

The patterns that appear in exchanges between researchers, scientific journal publications and the demand for scientific articles often intersect, but the logic behind each type of activity is not necessarily the same. Analyses of requests for scientific articles from document suppliers may help to interpret current developments in electronic publishing.

This study of article requests to the Institut de l'information scientifique et technique (INIST) shows that, in France, document supply customers fall into three main categories : business, academic libraries and public research organisations, in descending order. Demand focuses mainly on medicine, pharmacology, biology and chemistry, and the distribution of articles is entirely in accordance with the laws of bibliometrics.

A further comparative analysis shows a high reciprocal correlation (except in the physical sciences) between the 50 journals most frequently requested from INIST, and the 50 most frequently cited journals according to ISI (Institute for Scientific Information). The titles which did not appear in either one list or the other show that the most frequently cited physics journals are not necessarily requested from the document supplier, and that, conversely, some frequently requested journals are not often cited. It may therefore be assumed that a trade in electronic articles is likely to develop quite rapidly in disciplines which are common to both lists, although this would focus on reputed titles only, but that a different pattern of electronic document exchange would emerge for scientific literature in other disciplines.


Demand for scientific articles and citations

An example from the Institut de l'information scientifique et technique (France)

 

_______________

 

Jean-Michel Salaün, Gresi[3], salaun@enssib.fr

Thierry Lafouge, Recodoc[4], lafouge@enssib.fr

Cherifa Boukacem, Gresi, cboukace@enssib.fr

 

Introduction

Published communications between researchers[5] appear to follow a linear pattern which may be summarised as follows : author-article-journal-collection-article-reader. The link between the publishers of scientific journals and the libraries providing access to their collections reflects the commercial aspects of the process. However, this linear pattern actually masks a number of parallel processes and more complex circuits.

 

Document exchanges simultaneously involve several overlapping issues, processes and patterns of logic, although these intersect only at specific points in the chain :

-       the dynamics of scientific research are expressed in practice through research programmes, which need funding, through laboratory activities, which have to be managed, and through the careers of individual researchers. As W. D. Garvey (GAR79) showed some twenty years ago, communication between researchers is essential in maintaining these dynamic processes, and this is occurring today through the use of electronic networks by researchers. (CRA96). Articles published in journals still play an essential part, not so much as a means for disseminating knowledge (since scientific communication often uses faster, informal channels that vary with each discipline) as to achieve the recognition of knowledge. Being published means, first of all, recognition from one's peers, and secondly, becoming a reference for colleagues. The impact of an article, a researcher, a laboratory or a journal is often assessed in terms of the number of citations it receives in scientific publications, however arguable this criterion may be.

-       The situation differs from standard publication processes in that scientific authors are not paid for the writing they actually do for journals, while readers do not pay for access to the documents (since journal subscriptions are paid for by their institutions). The financial exchanges supporting research activities - which sometimes involve very high stakes - occur in a separate context. This characteristic feature has been a basis for arguing (HAR97) that publishing is not only an activity which runs parallel to communication between researchers, but that it may even have a parasitic role which could easily be short-circuited today through electronic networks.

-       The publishing process involves obtaining and selecting manuscripts, editing by an editorial board, assembling the articles in journals published on a regular basis, the purchase of journals through subscriptions taken out by libraries or document centres, and their incorporation into collections, which includes inserting their references into bibliographic tools to make them available to reader communities. In this particular chain of activities, libraries act as customers, while publishers (often through a subscription agency) act as suppliers. Any financial transactions are conducted on a standard commercial basis.

 

The steep -- and highly controversial -- rise in journal subscription prices is a symptom of the emerging crisis in journal trading. This has given rise to abundant comment, debates between professionals, and to a range of negotiations and experiments. The first large-scale survey on the subject was conducted by the Mellon Foundation for American Research Libraries (MEL92). After noting the increasing proportions of subscription budgets among American research libraries, the foundation suggested that electronic technology might offer a solution. The proposals from scientists mentioned earlier therefore agree with those made by American librarians. Publishers, librarians and the various intermediaries are being drawn into a whirlwind of experiments and negotiations in which the big question for all the protagonists is the role they will be playing in the future marketplace for electronic publications in science.

- However, the last twenty years or more have seen the development of an alternative exchange system centred on individual articles, as the smallest units of published scientific literature from the academic world. Any specific article may now be copied and supplied to individual readers on request, with the development of bibliographic databases providing the means, and inter-library loans the opportunity. This process was initiated by librarians, as a way of filling gaps in their collections by pooling their resources. Depending on the countries, document supply activities may involve specialised public bodies and private profit-making or not-for-profit organisations as well as libraries. Brown (BRO96, p.100-101) describes the development of the sector as a succession of stages, with inter-library loans as the first stage, followed by document supply through specialised organisations as an intermediate stage in a process which, according to the author, should culminate with electronic publishing.

Thus, whether the focus of interest is communication between researchers, journal publishing or demand for articles, the consensus of opinion is that electronic networks are destined to be the main vehicle for the circulation of scientific literature in the future. While the first two aspects (communication between researchers and scientific publishing) have been widely studied, and still are, the third has attracted much less interest so far. Discussion has mostly focused on library management issues (loans versus acquisitions, faster service, pricing and reader satisfaction), or on the practical implementation of new electronic services. Articles on these aspects have generally been of a descriptive, prescriptive or practical nature. None have appeared in France.

And yet, among all the changes associated with the shift towards electronic documentation, the article as a unit of exchange between different scientific disciplines could take on even greater importance. In this case, organisations which already exist as direct article suppliers would have information on use patterns which could be of considerable value in identifying trends for the future.

Whatever the opinions on the value of indicators based on the number of citations, which were popularised in particular by E. Garfield and the Institute for scientific information (ISI), it must be recognised that they serve a useful function in contributing to institutional assessments of research and researchers' career patterns. The important point here is not to determine whether they give an objective picture of scientific truths, but whether they perform a benchmark function for the scientific community, in the same way as audience ratings (which are no more reliable as a representation of reality) are used by television.

However, in view of the trend towards more and more widespread networking of scientific literature, it would be of interest to discover whether these indicators give a true picture of actual demand for articles. If this is indeed the case, they could also serve as indicators of trends in the electronic publications market. If not, this would reflect a hiatus between the economics of article circulation and the economics of academic research. Such a discrepancy would, in turn, affect the organisation of electronic publishing. Indications of this sort would be of considerable importance at a time when both information and the benefit of hindsight are lacking in negotiations on the prices of electronic services (HUN97).

Drawing on statistical data from the Institut de l'information scientifique et technique (INIST), the main French document supplier, this article aims to provide additional input to the debate by formulating the following questions :

- Is it possible to characterise demand in scientific articles ? Who requests articles, and how frequently ? What articles are requested, in which disciplines and from which journals ?

- Can any relationship be identified between requests for scientific articles and the number of citations in the relevant journals ? Is there any correlation between the citation indicator and demand for articles ? In other words, can such indicators be considered relevant as a measure of actual demand for articles while also providing information on the economics of article circulation ?

- Can any conclusions be drawn from these indicators on developments to be anticipated in the field of electronic documentation ?

These questions need to be answered as part of a much wider research programme which aims to clarify the economics involved in current changes in the circulation of scientific literature.

 

 

1. Value and limitations of available data

 

The data analysed are the article orders sent to INIST (Institut de l'information scientifique et technique) during the month of January 1997, i.e. a total of 50000 orders. Description of INIST data :

The lines in each order form contain :

Order date

Customer code

Customer category code

Document type code (book, journal, etc.)

Code indicating whether the document is held at INIST.

Journal title

Year of publication

Code identifying the volume required

ISSN

These data did not allow us to identify individual articles, but they did allow analyses by volume.

The data are contained in four Excel files, one per week in January 1997. All statistical results given in this article were obtained from an analysis of :

                .the four-week period covered by the files

                .each separate week

This precaution was taken to avoid distortions induced by external effects and to provide a sound basis for our conclusions on the characteristics of demand at INIST, even though journal circulation was studied over one month only. For example, we checked that the most frequently requested articles (cf annexe 2) were actually requested each week. Results given as intervals are those where the figures were calculated separately for each file and placed within their minimum-maximum range.

 

To find out whether this set of data is representative of a broader context, two questions must be answered : first, is the month of January sufficiently representative of average monthly orders sent to INIST ?  Secondly, are the orders sent to INIST representative of overall demand ?

 

In the first case, our sample (see Annexe 1) suggests that the answer should be yes. The trends identified were confirmed when the data for the whole year were processed.

The second question cannot be answered categorically. However, INIST is the main document supplier in France, with about 700 000 orders annually, which is equivalent to the number of documents requested through inter-library loans throughout the country. Although this suggests that some relationship exists between the structure of demand at INIST and the structure of overall demand for journal articles, it does not mean they have the same pattern. Informal exchanges of articles are extremely frequent, and we will be suggesting hypotheses throughout this article to attempt a characterisation of requests to INIST through libraries and the Internet.

 

As a first step, we will be looking at the data set from the customer-type angle, then from the journal-title angle. Secondly, part of the data - the fifty articles most frequently requested from INIST - will be compared to ISI findings [6] on the most frequently cited journals (ISI data are described in detail in Annexe 2).  This ISI data set has often been criticised. For the purpose of this analysis, we have selected the critical study by Magri and Solari (MAG96), which demonstrates the lack of uniformity of the data and draws attention in particular to a minority group of journals which produce «extreme» results in all indicators. Since our interest here lies in the most frequently requested and cited articles, our analysis focuses on this particular group.

 

 

 

 

 

2. INIST article requests in January 1997

 

85% of INIST requests concerned articles or conference proceedings. INIST used external suppliers to satisfy 8% of all requests in January (12% over the whole year).

 

The first lesson to be derived here concerns the extent of demand, which is larger than generally assumed in France and increasing steadily. Trading in journal articles is no longer a marginal form of business. It would be interesting to look at the monetary value of this activity and review the problems raised by copyright. These questions, which are of particular concern to publishers in view of their misgivings as to the development of a competing sector, have often been discussed within the profession, for example by Line et Wood (LIN75) some twenty years ago. The situation has changed, however, and the issue today is not so much to determine whether one form of document distribution is overshadowing another, as to attempt to discern those features in the current pattern which may shed more light on the shift to electronic communications, which concerns everyone with an interest in the exchange of scientific articles.

The first data processing stage provided a picture of overall customer distribution characteristics (in terms of institutional categories and behaviour patterns), and of journal distribution (in terms of the number of articles ordered by journal or by date).

 

 

2.a Customer typology

 

This was based on the survey carried out by the INIST marketing department, bringing the number of categories down to eight. The table 1 below shows distribution by customer numbers (in other words, the number of customers in each category who ordered at least one article in January 1997). This  table shows also customer distribution by the number of requests sent (in other words, the total number of requests sent in each customer category in January 1997).

 

 


 

 

Number of orders

% of orders

Number of customers

% customers

Public or private sector companies

22565

44,4

837

33,5

Higher education establishments

12094

23,8

709

28,4

Research organisations

12366

24,2

579

23,2

Hospitals, medical centres & clinics

1414

2,8

111

4,5

Information centre

859

1,7

51

2

Other public or private sector organisations

755

 1,5

84

3,4

Individuals

486

1

100

4

Other

 320

  0,6

20

1

TOTAL

50859

 100

2495

100

Table 1: distribution by number of requests sent to the INIST and by number of  document supply customers (January 1997)

 

image

 

Diagram 1 :Distributions of customers and orders

Three categories stand out clearly : companies (i.e., research and development), higher education establishments (mainly academic libraries) and research organisations (CNRS laboratories and leading public sector laboratories such as those run by the CEA, INSERM or INRA [7]). However, their relative contributions need to be interpreted in different ways.

On average, each company orders nearly twice as many articles as academic libraries. This is probably due not only to the growth of research and development activities, but also to a preference in such organisations for building up their collections through orders for individual articles rather than journal subscriptions, whereas individual orders are generally a back-up solution for libraries, as they tend to look first to titles in their own catalogues to satisfy requests.

Nevertheless, the scale of demand from libraries would seem to indicate that they are less and less able to satisfy reader demand from their internal collections, or even through inter-library loans.

Demand from research organisations is very unevenly distributed. Some have obviously delegated the supply of all their scientific literature needs to INIST, while for others, INIST is only a backup solution. The distribution of demand among laboratories is unexpected from a strictly institutional point of view.

 

In the other customer categories, demand is low enough to be considered almost marginal. Our study shows a very low level of demand from individuals, and none from public libraries and distribution networks (distributors, subscription agencies, publishers, etc.). The French research community appears to form a self-contained whole which communicates very little with the outside world.

 

2.b Customer behaviour

 

A more detailed analysis of the data shows a very wide gap between a handful of very large clients, all within the research community, and numerous one-off orders.

56% of all customers in all categories sent less than seven requests in the month. In the business and academic library categories, which account for 68% of all orders as indicated above, the two largest customers account for only 4.4% (businesses) and 3.8% (libraries) of all requests. One customer in the business category ordered 27 articles on average. This is not a significant average, because the standard deviation  works out at 781, indicating very high dispersion.

 

This high degree of dispersion obviously raises tremendous management problems for the document supply organisation since it greatly increases the administrative workload involved in document handling, invoicing for small orders, managing copyright fees, etc. Conversely, however, a small number of large customers carry a threat of financial vulnerability, since the loss of any one of these would give considerable cause for concern.

 


2.c Types of orders

 

In terms of scientific disciplines, four topics account for nearly all requests : medicine, pharmacy, biology and chemistry. This is true both for the most frequently requested journals (see below) and for those least often requested.  Physics is very poorly represented.

Other subjects barely appear, especially in the human and social sciences, mathematics and information technology. English-language titles are ubiquitous, and the top-ranking French-language title, La revue du praticien (Paris), in 75th place with 54 orders, is a popularising publication.

Our analysis confirmed that these characteristics are also valid for journals that were only requested once. This result was derived from a random sample because of the large quantities involved (cf. example in Annexe 3). One notable difference was the substantial number of foreign-language titles. A previous study showed that, in chemistry at least, each article is generally requested once only (DEL89). This is confirmed by the data from this study, in terms of volume.

 

2.d Distribution of orders

 

About 20.000 journal titles were needed to satisfy the 50.000 requests sent in January 1997. INIST's own collection contains 23.000 titles. The distribution of orders by journal title (results are given with margins) shows that 52% to 55% of titles are requested once only, and 3% to 4% of titles are requested more than 10 times.

 

image

Diagram 2 : INIST journal ranking by article requests

(January 1997)


The Zipfien-type distribution of journal circulation (HAI82) observed (Diagram 2) obeys standard bibliometric laws and is common to most libraries (LEE92). A very small number of journals is in considerable demand, while the majority are only requested occasionally, which results in high dispersion : the distribution of demand at INIST is similar to that in libraries or document systems. The regularity of the above curve is characteristic of Zipf's standard law of bibliometrics.

The distribution of requests in relation to the year of publication of articles provides another classic example of distribution in scientific and technical information : the more recent the article, the greater the demand.  However, dispersion is again very high : although 26% of orders were for articles published in 1996, 14% of articles requested were published before 1980.

 

INIST's annual statistical study found the same distribution result. The standard 80/20 rule (80% of orders concern only 20% of titles) has been confirmed regularly over the six years since the organisation's statistics system has been able to calculate it. From this point of view, INIST is a classic example of librarianship.

INIST therefore functions in the same way as a library, with high dispersal of demand in terms of both titles (collections) and years (conservation).

 

 

3. Journal requests and journal citations

 

The second stage in our analysis of the data set involved comparing the 50 journals most frequently requested from INIST with the 50 journals most frequently cited according to ISI, and - in a less detailed analysis - with their citation frequency in French academic libraries. This was similar to the Scales study carried out twenty years ago (SCA76) on requests to the British Library.  However, the two studies are not readily comparable since the context as well as the time scales and geographical criteria have changed a great deal.

This was not a neutral sample, since the focus is no longer on a population of representative journals but on a group with very specific characteristics. Any conclusions drawn from the analysis are therefore only valid for those characteristics.

 

3.a Journals most frequently requested from INIST

 

The table in annexe 2 shows the fifty journals most frequently requested from INIST in January 1997. The distribution of these journals, Diagram 3, is identical to  zoom in on the first part of the curve described earlier, diagram 2.


 

image

Diagram 3 : the fifty journals most frequently requested from INIST

(January 1997)

 

 

The curve shows that the first four journals fulfil a specific function. These are the Journal of Biological Chemistry, Lancet, Proceedings of the National Academy of Sciences (USA) and the New England Journal of Medicine. In the case of the first, third and fourth of these, one of the reasons for their ranking is the number of articles published each year. According to ISI, the Journal of Biological Chemistry is by far the most productive in terms of the annual number of articles published, along with Physical Review B (which is not in the INIST list, see below). The third and fourth titles are also among the top-ranking journals in terms of article numbers.

Demand for Lancet is a special case. This is borne out by observations for the entire year and cannot therefore be considered incidental. This point will be discussed later on in this article.

Beyond these three journals, the curve decreases steadily.

 

One interpretation of the pattern of demand at INIST might be that there are gaps in the material offered by libraries. To check this, references to the relevant journals in the French Union Catalogue were noted (cf. annexe 1) to see whether there was any link between demand at INIST and the presence of the journals in libraries. The link could have operated either way, depending on whether collections were built up according to demand or on the basis of inter-library sharing agreements.  In fact, no clear correlation emerged from the figures. Contrary to conclusions drawn twenty years ago by Lines and others for the UK, there is no obvious relationship between title availability in libraries, as indicated by the French Union Catalogue, and demand at INIST.


3.b Journals most frequently cited according to ISI

 

The distribution curve  (diagram 4) for the ISI list of the fifty articles most frequently cited in 1996 (cf. table in annexe 2)[8] has the same shape as the first curve :

 

image

 

Diagram 4 : the most frequently cited 50 journals according to ISI

(JCR 1996)

 

Again, the curve shows an obvious, if slightly less abrupt drop after the first five titles. These are the Journal of Biological Chemistry, Proceedings of the National Academy of Sciences (USA), Nature, Science, and the Journal of the American Chemical Society. Again, the number of articles published may explain the popularity of three of the top-ranking journals.

However, two titles, Nature and Science, appear to fulfil a different role in the circulation of ideas, and even have a specific format. In both cases, the pattern remains atypical throughout the year, and is comparable to the pattern noted above for the Lancet which fulfils a similar role, albeit in the more specialised field of medicine.

One possible explanation may be that these journals are commonly found in laboratories as researchers generally use them to scan the latest research news. However, data for the whole year show that INIST fulfils a conservation function for Nature and Science, since 60% of requests for these titles concern articles published before 1990 (which is the reverse of the overall trend). Requests for the Lancet concern more recent articles. A more detailed analysis of requests in January 1997 shows that these requests are mostly from the business

sector, probably pharmaceutical laboratories whose documentary practices carry significant weight in the data set.

The structure of the ISI sample was analysed in more general terms by M.-H. Magri and A. Solari (MAG96), using the "box-plot" method. The authors showed that the sample included an «extreme» group of journals, accounting for less than 10% of the sample, for which the citation indicators were well beyond average values. By limiting the analysis to the fifty top titles, any conclusions refer to the top segment of a group which already has exceptional characteristics. This is an important point which has to be taken into account in our interpretation.

 

3.c Comparison between the INIST and ISI lists

 

The relationships between trade in articles and citation indicators were assessed on the basis of reciprocal correlation between the two lists. Specifically, two linear correlation coefficients were calculated.

 

First, the linear correlation (r) between demand for the 50 journals most frequently requested from INIST and the number of citations in the ISI sample was calculated. The result was r=0.640, and checking r resulted in a theoretical r* value of 0.452 with 0.1% for 48 degrees of freedom, which confirms the correlation to within 0.1%.  The graph below shows the cluster obtained.

 

image

 

Diagram 5 : relationship between requests and citations for the 50 journals most frequently requested from INIST(January 1997)


 

The graph raises three comments/questions :

- 3 journals (Lancet, Science, Nature) are located at the periphery of the cluster, which confirms that they have a specific function.

-cluster density is far from uniform. The correlation was therefore calculated without the four top-ranking INIST journals (see graph for the 50 journals most frequently requested from INIST). The resulting correlation

 (0.349) is significantly lower. Checking r resulted in a theoretical r* value of 0.308 with 5% for 44 degrees of freedom, which confirms the correlation to within 5%.

- several extensively requested journals demonstrate low citation rates - see also the inverse case, Diagram 6 below.

We then calculated the linear correlation between the citation frequency of the 50 top-ranking journals in the ISI list and the frequency of requests for the same journals from INIST. This worked out as r= 0.600, confirming the correlation to within 0.1%. Correlation is therefore substantially similar in both cases. The graph below shows the cluster obtained

 

image

 

Diagram 6 : relationship between requests and citations for the 50 most frequently cited journals according to ISI. (JCR 1996)

Again, the resulting cluster is not uniform, although less obviously so than in the previous case. The position of the journals which play a significant part in the structure of the cluster stands out in a similar way. When the four top-ranking ISI journals are left out of the calculation, the resulting correlation coefficient, r=0.331, is much lower. However, the same conclusion is still valid.

Similarly for Diagram 6, where we observe two averaged cited journals (Physics letter B, Physical review D) having zero demand and some with very low request ratio. It is the latter observations that indicate the existence of hidden and closed communication circles in electronic form (not electronic publishing) ; the first observation (Diagram 5) shows that some journals are heavily requested, but their use does not seem to be mirrored by citations in futher work. This implies that although a statistical correlation exist for the large portion of journals demanded/cited, the correlation is not valid at the end of the curves. If demand=use to know the demand may supplement knowledge about cited journals, e.g. in application/patent areas, where citations given are scarce.

To confirm the correlation with another method, for a longer series, the titles which appeared among the first 100 in both lists were counted, in intervals of 10, in the journals most frequently requested from INIST :

 

 

INIST

classification

Of which x titles in the 100 journals most frequently cited according to INIST

Top 10

x= 5

Top 20

x=10

Top 30

x=16

Top 40

x=18

Top 50

x=26

Top 60

x=27

Top 70

x=35

Top 80

x=42

Top 90

x=49

Top 100

x=57

 

Table 2

Intersection of the first 100 in the both lists : Inist, Isi.

 

There is a surprising degree of regularity in the series, which differs from the pattern of British Library requests observed twenty years ago by Scales (SCA76), where the increase in the proportion of titles appearing in both lists was very gradual. Two apparently contradictory conclusions may be drawn from this regularity :

- on the one hand, the number of citations does provide a fairly reliable market indicator for articles from the group of reputed journals, since an obvious correlation emerges ;

- on the other hand, the degree of notoriety as measured by the number of citations does not seem to have had any influence on demand for half of the top-ranking group. The indicator therefore leaves a significant number of titles aside. A more thorough analysis of titles which do not appear in both lists is required.

 

3.d Journals not included in either list

 

The following tables show the journals which are included in one of the lists but rank below the top 200 according to the criteria used in the other list.


 

 

Journal title

INIST rank

INIST

requests

ISI

rank

ISI

cit.

Journal of virology

248

29

36

49304

Journal of molecular biology

258

28

37

47133

Journal of bacteriology

301

25

39

46238

Inorganic chemistry

311

25

46

41161

Physical review A

515

19

44

42320

Astrophysical journal

830

12

18

71220

Physics  letters  B

0

0

40

45827

Physical review D

0

0

49

38641

 

Table 3

Titles included in the 50 most frequently cited journals according to ISI (JCR 1996) but rarely or never requested from INIST (January 1997)

 

This table clearly highlights a group of physical sciences and astrophysics journals which seem to be very popular with scientists, but whose contents do not seem to be exchanged through conventional document supply channels. One reason for this may be the growth of pre-print servers in these disciplines, based on the service launched by P. Guinsparg, in which case it may be said that electronics have superseded hard copy documentation.

Our analysis qualifies this interpretation which, if based on transposition, may lead to confusion between cause and effect. The success of the Guinsparg initiative was also due to article exchange practices which already existed between researchers in this discipline, and also to existing journal subscriptions taken out by laboratory libraries. In fact, the evidence suggests that the small number of official document requests is not a recent phenomenon, and also that there is a longstanding specific relationship with these publications within the physics community :

- thus, even Line and Wood back in 1975 found only one physics title (Journal of Applied  Physics) among the journals most frequently photocopied at the British Library (LIN75).

- de la Vega (VEG99), observing communication practices between researchers in theoretical physics, showed that this was a very close-knit community with a high degree of loyalty to its own journals, which were always available in hard copy in laboratory libraries. In addition, American, European and French physics societies, which also publish the main physics journals, play a fundamental part in the exchange of knowledge and the organisation of the community. These special features have made the physics community a kind of laboratory for studies on developments in electronic publishing. Regarding the topic we are concerned with here, a point of particular interest is that the incorporation of electronic journals into pre-print servers is raising entirely new questions for the various interested parties, as described by one authority on the subject (TIM98) :

Archiving has always been the responsibility of libraries, which unfortunately are also the most important customers of a scientific publisher. The basic conundrum is that if publishers make an electronic archives available, they will undercut one of the most important activities of their major customers, that of archiving the historical record. An online archive makes it unnecessary to go to the library to access old copies of the journal; therefore, the library loses its incentive to subscribe to the journal. Furthermore, an archive searchable by title, author, full-text, etc. moves the task of locating the information from the library catalogue to the desk of the user. It also assigns the responsibility for supporting that service to the publisher, not the librarian. Thus, no matter how you look at it, an electronic archive, offered by the publisher, is a threat to the publisher's major customer -- not a situation that encourages rapid innovation.

 

 

 

Journal title

ISI rank

ISI Cit.

INIST rank

INIST requests

Journal of applied polymer science

201

11980

33

74

Journal of materials science

209

11663

19

83

Chemical and pharmaceutical bulletin

277

9206

18

89

Chemical engineering science

301

8383

44

70

Polymer  engineering and science

472

5486

26

77

Chemosphere

519

4914

48

69

Pharmaceutical research

526

4848

40

72

International journal of pharmaceutics

592

4373

11

98

 

Table 4

Titles included in the 50 journals most frequently requested from INIST (January 1997),

but not often cited according to ISI ( JCR 1996)

 

 

This table is not so readily interpreted.  Particular care is needed with regard to the bias in the ISI sample. An overall rank in 200th position can be an excellent result in a specialised discipline. However, the journals indicated here appear to focus more on research and development than on academic research : in other words, a key criterion for assessing articles from these journals might be whether an application has been made for a patent on the results, rather than the number of citations. If confirmed, this interpretation would demonstrate the value of coupling the two criteria to reflect the characteristics of article demand.

 

Conclusion

 

Apart from the reservation made in the previous paragraph, there does seem to be a relationship between citations and demand for articles in medicine, pharmacology, biology and chemistry, at least for the most frequently requested articles.  Over one year, these disciplines alone account for 63% of all articles requests to INIST.

One possible interpretation of this empirical observation may lie in the widespread and longstanding use of bibliographic tools and data bases which were pioneered by these disciplines, through the National Library of Medicine and Medline on the one hand and the American Chemical Society and Chemical Abstracts on the other. However, these tools owe their existence to the specific way in which scientific information circulates in any case in these particular disciplines, and also to the fact that articles in these subjects are in considerable demand, especially from industry. Industrial competition is very strong in these disciplines which is not the case elsewhere, and this is a determining factor in the way articles are circulated.

This suggests that there is indeed a potential market for electronic articles from the most reputed journals in these disciplines.

Three journals, Nature, Science, and The Lancet, fulfil a specific function. The use of these journals and their articles deserves a specific analysis in terms of the circulation of scientific information. Their role as magazines could provide them with a unique function in communications networks : indeed, Science already has an ambitious marketing strategy in this area. In view of the structure of demand, marketing the collection as a whole may be well prove to be profitable.

 

Patterns for the other titles and disciplines using electronic networks may prove to be very different. Although we do not claim to have covered the subject exhaustively, nor even to provide a fully researched basis for the following arguments, they may help to interpret our findings :

- journals in the physical sciences and astrophysics are frequently cited, but are much more rarely requested from INIST, if at all, although they are included in INIST's holdings. Our interpretation for this was given earlier, and it may be assumed that in these disciplines, exchanges of material through electronic networks will continue to evolve in a distinct way which will be largely determined by the organisation of the scientific community itself.

-       the absence of mathematics journals in both the ISI and INIST lists may be due to the lesser value ascribed to citations, to the smaller size of the community, to the fact that independent documentation circuits have existed for many years in France (Math-doc unit) and abroad, and to the lesser degree of interest from the business community. As for the physics community, it is likely that electronic document circulation will continue to be managed to a large extent by the community itself.

-       In computer and engineering sciences, where citations are considered differently and where we have shown not only the extent but also the dispersion of demand, the future pattern of electronic publishing seems much less clear-cut. Neither the scientific community, nor the publishers, nor the libraries appear to be capable of establishing an adequate configuration for electronic publishing.

 

No such deductions can be drawn from our data on journals in the social and human sciences. This group is much larger than the others in terms of the number of researchers and publications, but does not use INIST as a channel for the distribution of information. The citations criterion does not seem particularly relevant either. This suggests that in these disciplines, the journal medium - considered as a single uniform support for articles - together with the importance of language itself - implies a different type of relationship. Our opinion is that in these disciplines, any interpretation should be based on a different analysis altogether.

 

                                                                                              April 1998, revised juillet 99.

 

 


 

 

 

 

Acknoledgments

 

We would like to thank Marc Guichard and his associates for their support and for providing us with access to these data and. http://www.inist.fr.

 

We would like to thank Frédéric Dubois, Head of INIST's statistical processing department, for his careful editing of this article. The comments and data which he provided have been incorporated into our analysis.

 

 

REFERENCES :

1.     (BRO96)

D. J. BROWN, Electronic publishing and libraries, Planning for the impact and growth to 2003, British Library Research, Bowker Saur, 1996.

 

2.     (CRA96)

S. Y. CRAWFORD , J. M. HURD, A. C. WELLER, From Print to Electronic, The Transformation of Scientific Communication. Asis, Information Today, Medford NJ Usa, 1996.

 

3. (DEL89)

A. DELARBRE, T. LAFOUGE, Des statistiques à la bibliométrie, Revue française de bibliométrie, (1989) No 4, 38-49.

 

4. (GAR91)

E. GARFIELD, In truth, the"flood" of scientific literature is only a myth, The Scientist,  5 (1991) No 17.

http://165.123.33.33/yr1991/sept/garfield_p11_910902.html

 

5. (GAR79)

W.D. GARVEY, Communication : the essence of science - Faciliting information exchange  among Librarians, Scientists, Engineers, and students, Pergamon Press 1979.

 

6. (HAI82)

S.D HAITUM, Stationary scientometric distributions, Scientometrics, 4 (1982) No 1, 5-25.

S.D HAITUM, Stationary scientometric distributions, Scientometrics, 4 (1982) No 2, 89-104.

S.D HAITUM, Stationary scientometric distributions, Scientometrics, 4 (1982) No 3, 181-194.

 

7. (HAR97)

S. HARNAD, How to Fast-Forward Serials to the Inevitable and the Optimal for Scholars and Scientists, Serials Librarian , 30 (1997) 73-81.

http://www.cogsci.soton.ac.uk/~harnad/Papers/Harnad/harnad97.learned.serials.html

Reprinted in :

C. Christiansen & C. Leatham, Eds. Pioneering New Serials Frontiers: From Petroglyphs to CyberSerials,  NY: Haworth Press, and in Traduction française :

Comment accélérer l'inéluctable évolution des revues érudites vers la solution optimale pour les chercheurs et la recherche

http://www.enssib.fr/eco-doc/harnadinteg.html

 

8. (HUN97)

K. Hunter, The Effect of Price: Early Observations, Conférence de la fondation Andrew MellonScholarly Communication and Technology, 24-25 avril 1997, Emory University, USA.

http://www.arl.org/scomm/scat/hunter.html

 

9. (LEE92)

LEEMANS, M.J, MAES, M.,  ROUSSEAU, R., & RUTS C., The negative binomial distribution for circulation data in flemish public libraries, Scientometrics, 25 (1992) No 1, 47-57.

 

 

10. (LIN75)

M. B. LINE, D. N. WOOD, The effect of large-scale photocopying service on journal sales, Journal of Documentation, 31 (1975) No 4, 234 -245

 

11. (MAG96)

M.-H. MAGRI, A. SOLARI, Description of the JCR journal population based on the number of citations received, number of source items, impact factor, immediacy index and cited half-life, Scientometrics, 35 (1996) No 1 (1996), 93-117.

 

12. (MEL92)

Fondation A. W. MELLON, University Libraries & Scholarly Communication, library administration, 23 (1996) No 3/4.

Published by Association for research libraries, November 1992.

http://www.lib.virginia.edu/mellon/mellon.html

 

13. (SAL97)

J.-M. SALAÜN, Le modèle bibliothéconomique, Economie et bibliothèques, Editions du Cercle de la librairie, Paris, 1997.

 

14. (SCA76)

P.A. SCALES, Citation analyses as indicators of the use of serials : a comparison of ranked title lists produced by citation counting and from use data, Journal of documentation, 32 (1976) No 1, 17-25.

 

15. (TIM98)

T. Timothy, Archives in a New Paradigm of Scientific Publishing, Physical Review Online Archives (PROLA),

D-Lib Magazine, May 1998.

http://mirrored.ukoln.ac.uk/lis-journals/dlib/dlib/dlib/may98/05thomas.html

 

16. (VEG99)

J. de la VEGA, La communication scientifique à l'épreuve de l'internet, Editions de l'Enssib, 1999.

 

WEB Sites:

 

Agence bibliographique de l'enseignement supérieur (French Union Catalogue Management)

http://www.abes.fr/

 

Institut de l'information scientifique et technique

http://www.inist.fr/

 

Institute for scientific information

http://www.isinet.com/

 

Economie du document - circulation des articles scientifiques

http://www.enssib.fr/eco-doc


 

 

 

Annexe 1

 

DESCRIPTION OF CCN DATA

The French Union catalogue of Serial Publications (CCNPS - Catalogue Collectif National des Publications en Série) :

The CCN-PS (Catalogue Collectif National des Publication en série) is the national serial publication union catalogue in France.

Existing since 1983, it is an essential tool for article suplying since periodical articles constitute the majority of documents requested on inter library loan, nationaly.

The data base was set up in 1983 and combines several files to provide a list of more than one million journal collections, or over 280 000 different titles in nearly 3000 libraries.

The availability of the CCN in Cd-Rom is named Myriade, a useful tool for libarries to enable the location of both French and foreign-language serial publications.


 

Annexe 2

The fifty journals most frequently cited and requested

 

The fifty journals most frequently cited according to the ISI sample

(JCR 1996)

 

Journal title

ISI Citations

INIST requests

INIST ranking

Myriade

1.   Journal of biological chemistry

293024

184

1

157

2.   Proceedings of the national academy of science - USA

276659

158

3

214

3.   Nature

270077

107

5

391

4.   Science

221696

79

24

344

5.   Journal of the american chemical society

174540

99

10

152

6.   Cell

149477

60

61

143

7.   Physical review letter

132797

50

86

120

8.   Physical review B -Condensed matter

130789

35

182

69

9.   Journal of chemical physics

120281

81

21

117

10.New england journal of medicine

118106

115

4

176

11.Lancet

100526

179

2

171

12.Biochemistry

91747

58

67

28

13.Journal of immunology

89503

69

46

111

14.American journal of physiology

89048

82

20

98

15.Cancer research

82343

91

15

91

16.Journal of physical chemistry - US

73237

79

23

107

17.Journal of clinical investigation

72294

95

13

111

18.Astrophysical journal

71220

12

830

54

19.Biochimica et biophysica acta

69557

104

6

81

20.Journal of cell biology

67402

69

45

133

21.Blood

67238

44

112

102

22.Circulation

64644

97

12

81

23.Brain research

63958

70

43

49

24.Embo journal

62920

38

156

97

25.Journal of geophysical  research

61912

51

83

89

26.Journal of experimental medicine

60034

57

70

98

27.Molecular and cellular biology

58647

40

134

68

28.Biochemical and biophysical research communications

56993

103

8

116

29.Journal of organic chemistry

56632

70

42

98

30.Journal of applied physics

55355

79

22

133

31.Applied physics letters

55205

42

123

90

32.Nucleic acids research

54784

45

104

86

33.JAMA - Journal of  the american medical association

53060

77

28

111

34.Tetrahedron letters

52693

50

88

83

35.Biochemical journal

50291

75

32

80

36.Journal of virology

49304

29

248

82

37.Journal of molecular biology

47133

28

258

111

38.Cancer

47014

57

71

103

39.Journal of bacteriology

46238

25

301

126

40.Physics letters B

45827

0

0

69

41.Febs letters

45436

93

14

101

42.British medical journal

44982

78

25

97

43.Analytical biochemistry

42553

74

35

120

44.Physical review A

42320

19

515

45

45.Endocrinology

42263

49

91

18

46.Inorganic chemistry

41161

25

311

72

47.Journal of neuroscience

40357

48

97

25

48.Gastroenterology

39531

40

141

78

49.Physical review D

38641

0

0

89

50.Analytical chemistry

37777

74

36

191

 


The fifty journals most frequently requested from INIST

(January 1997)

 

Journal title

INIST requests

ISI Citations

ISI ranking

Myriade

1.   Journal of biological chemistry

184

293024

1

157

2.   Lancet

179

100526

11

171

3.   Proceedings of the national academy of science - USA

158

276659

2

214

4.   New england journal of medicine

115

118106

10

176

5.   Nature

107

270777

3

391

6.   Biochimica et biophysica acta

104

69557

19

81

7.   Chest

104

19456

107

68

8.   Biochemical and biophysical research communications

103

56993

28

116

9.   Journal of chromatography

100

30985

62

99

10.Journal of the american chemical society

99

174540

5

152

11.International journal of phrmaceutics

98

4373

592

19

12.Circulation

97

64644

22

81

13.Journal of clinical investigation

95

72249

17

111

14.Febs letters

93

45436

41

101

15.Cancer research

91

82343

15

91

16.American journal of epidemiology

90

15778

145

46

17.Annals of the New York academy of science

89

22362

90

63

18.Chemical and pharmaceutical bulletin

89

9206

277

34

19.Journal of materials science

83

11663

209

54

20.American journal of physiology

82

89048

14

98

21.Journal of chemical physics

81

120281

9

117

22.Journal of applied physics

79

55355

30

133

23.Journal of physical chemistry

79

73237

16

107

24.Science

79

221696

4

344

25.British medical journal

78

44982

42

97

26.Polymer engineering and science

77

5486

472

27

27.The journal of urology

77

26622

76

59

28.JAMA - Journal of the american medical association

77

53060

33

111

29.Journal of clinical microbiology

75

22248

92

39

30.Neurology

75

30132

65

54

31.Archives of internal medicine

75

17256

132

69

32.Biochemical journal

75

5291

35

80

33.Journal of applied polymer science

74

11980

201

49

34.American journal of cardiology

74

27656

71

58

35.Analytical biochemistry

74

42553

43

120

36.Analytical chemistry

74

37777

50

191

37.Biochemical pharmacology

74

20338

103

49

38.Archives of biochemistry and biophysics

73

21420

97

86

39.Journal of colloid and interface science

72

16676

136

67

40.Pharmaceutical research

72

4848

526

15

41.Environmental science and technology

72

14199

161

3

42.Journal of organic chemistry

70

56632

29

98

43.Brain research

70

63958

23

49

44.Chemical engineering science

70

8383

301

46

45.Journal of cell biology

69

67402

20

133

46.Journal of  immunology

69

89503

13

111

47.Annals of internal medicine

69

36850

52

91

48.Chemosphere

69

4914

519

11

49.Neuroscience letters

68

22249

91

30

50.Applied an environmental microbiology

67

25831

79

85

 


ISI journal citations (JCR 1996) not included in the INIST list (January 1997)

(ranking from 1 to 50)

 

 

Journal title

INIST rank

INIST requests

ISI rank

ISI citations

Myriade

Cell

61

60

6

149477

143

Biochemistry

67

58

12

91747

28

Journal of experimental medicine

70

57

26

60034

98

Cancer

71

57

38

47014

103

Journal of geophysics research

83

51

25

61912

89

Physical review letters

86

50

7

132797

120

Tetrahedron letters

88

50

34

52693

83

Endocrinology

91

49

45

42263

18

The journal of neuroscience

97

48

47

40357

25

Nucleic acids research

104

45

32

54784

86

Blood

112

44

21

67238

102

Applied physics letter

123

42

31

55205

90

Molecular and cellular biology

134

40

27

58647

68

Gastroenterology

141

40

48

39531

78

Embo journal

156

38

24

62920

97

Physical review B

182

35

8

130789

69

Journal of virology

248

29

36

49304

82

Journal of molecular biology

258

28

37

47133

111

Journal of bacteriology

301

25

39

46238

126

Inorganic chemistry

311

25

46

41161

72

Physical review A

515

19

44

42320

45

Astrophysical journal

830

12

18

71220

54

Physics  letters  B

0

0

40

45827

69

Physical review D

0

0

49

38641

89

 


 

Journals requested from INIST (January 1997)and not included in the ISI list (JCR 1996)

(Ranking from 1 to 50)

 

 

Journal title

ISI rank

ISI citations

INIST rank

INIST requests

Myriade

Annals of internal medicine

52

36850

47

69

91

Journal of chromatography

62

30985

9

100

99

Neurology

65

30132

30

75

54

American journal of cardiology

71

27656

34

74

58

Journal of urology

76

26622

27

77

59

Applied an environmental microbiology

79

25831

50

67

85

Neuroscience letters

91

22249

49

68

30

Annals of the New York academy of science

90

22362

17

89

63

Journal of clinical microbiology

92

22248

29

75

39

Archives of biochemistry and biophysics

97

21420

38

73

86

Biochemical pharmacology

103

20338

37

74

49

Chest

107

19456

7

104

68

Archives of internal medicine

132

17256

31

75

69

Jornal of colloid and interface science

136

16676

39

72

67

American journal of epidemiology

145

15778

16

90

46

Environmental science and technology

161

14199

41

72

3

Journal of applied polymer science

201

11980

33

74

49

Journal of materials science

209

11663

19

83

54

Chemical and pharmaceutical bulletin

277

9206

18

89

34

Chemical engineering science

301

8383

44

70

46

Polymer  engineering and science

472

5486

26

77

27

Chemosphere

519

4914

48

69

11

Pharmaceutical research

526

4848

40

72

15

International journal of pharmaceutics

592

4373

11

98

19

 


Annexe 3

Random sampling of journals requested once only from INIST

(January 1997)

_________________________

 

Random samples were taken four times in succession. No outstanding differences were observed. One of the samples is given below as an example.

 

-10TH NAMCR : CONFERENCE, 1982 MAY : HAMILTON, CANADA

-APPLICATIONS AND THEORY OF PETRI NETS - 6TH EUROPEAN CONFERENCE JUNE 1985

-CHEMICKE ZVESTI

-ENGINEERING MANAGEMENT JOURNAL

-INDIAN DRUGS

-JOURNAL OF FLUORINE CHEMISTRY

-MAGNESIUM IN VEHICLE DESIGN

-PALEOBIOLOGY

-RECENTI PROGRESSI IN MEDICINA

-THE ITALIAN JOURNAL OF GASTROENTEROLOGY

-ADVANCES IN APPLIED MICROBIOLOGY

-BIOCHEMISTRY : (EASTON)

-COASTAL ENGINEERING : (AMSTERDAM)

-EUROPEAN JOURNAL OF OBSTETRICS AND GYNECOLOGY

-INFLAMMATION RESEARCH

-ADVANCES IN PERITONEAL DIALYSIS

-ASPECTS OF SPONGE BIOLOGY, 1975 MAY, ALBANY NY

-CANCER BIOTHERAPY

-CRITICAL CARE CLINICS

-EUROPEAN WORKSHOP IN PERIODONTOLOGY FEB 1993 ITTINGEN

-HUMAN PSYCHOPHARMACOLOGY

-IZVESTIA SIBIRSKOGO OTDELENIA AKADEMII NAUK SSSR. SERIA HIMICESKIH NAUK

-JOURNAL OF MORPHOLOGY : (1931)

-LOZARSTVO I VINARSTVO

-NIHON KYOBU GEKA GAKKAI ZASSHI

-POLYMER-PLASTICS TECHNOLOGY AND ENGINEERING : (SOFTCOVER ED)

-REVUE DES MALADIES RESPIRATOIRES

-TECHNIQUES AND APPLICATIONS OF PATH INTEGRATION

-SOVIET JOURNAL OF OPTICAL TECHNOLOGY

-SOVIET JOURNAL OF SUPERHARD MATERIALS

-SOVIET PHYSICS JOURNAL

-SOVIET PHYSICS, JETP

-SOVIET PHYSICS, LEBEDEV INSTITUTE REPORTS

-SOVIET PHYSICS. TECHNICAL PHYSICS

-SOVIET PHYSICS. USPEKHI

-SPECIAL PUBLICATION - ROYAL SOCIETY OF CHEMISTRY

-SPECIALITY CHEMICALS : (REDHILL)

-SPECTROCHIMICA ACTA. PART A : MOLECULAR SPECTROSCOPY

-SPECTROCHIMICA ACTA. PART B : ATOMIC SPECTROSCOPY

-SPEECH MONOGRAPHS

-SPRINGER SERIES IN CHEMICAL PHYSICS

-STAHL UND EISEN : (1881)

-STATISTICESKIE PROBLEMY UPRAVLENIA

-STATISTIQUE ET ANALYSE DES DONNEES

-STEEL - STRENGTHENING MECHANISMS - PROCEEDINGS F A SYMPOSIUM, 1969 MAY, ZU

-STEEL RESEARCH

-STEROIDS

-STOCHASTIC PROCESSES AND THEIR APPLICATIONS

-ANNALS OF HUMAN BIOLOGY

-ANNALS OF MATHEMATICAL STATISTICS

-ANNALS OF MATHEMATICS

-ANNALS OF SURGERY

 



[1]Cresi, Enssib, 17,21 Bvd du 11 novembre 1918, 69623 Villeurbanne Cedex France.

[2]Recodoc, Université Claude Bernard Lyon 1 bat 721, 43 Bvd du 11 novembre 1918, 69622 Villeurbanne cedex France.

[3]Gresi, Enssib, 17,21 Bvd du 11 novembre 1918, 69623 Villeurbanne Cedex France.

[4]Recodoc, Université Claude Bernard Lyon 1 bat 721, 43 Bvd du 11 novembre 1918, 69622 Villeurbanne cedex France.

[5]This study was made as part of a research programme of which full details are available on the Web site. See : économie du document_-_circulation des articles scientifiques.

http//www.enssib.fr/eco-doc

[6]http://www.isinet.com

[7] CNRS - National Centre for Scientific Research; CEA - Atomic Energy Commission; INSERM - National Institute of Health and Medical Research; INRA - National Institute of Agronomic Research

[8] For our purposes, the relevant indicator is the number of journal citations in the previous year. The well-known "impact factor", a ratio calculated for citations of the most recent articles, is not  meaningful for the pruposes of this study, because the analysis of INIST article requests was made for the entire collection and not just recent issues.