Evolution of the interdisciplinary characteristics of information and library science

Authors


Abstract

An empirical investigation of citations to 150 publications in the field of Information and Library Science (ILS) has enabled mapping of the development of the interdisciplinary scope of the ILS field. The publications were drawn randomly in six years between 1975 and 2000, with 25 articles each from the selected years. Network-based graphical presentation of number of extradisciplinary citations show that the field attracts a significant wide spectrum of disciplines in the domains of science, social science, and the humanities, and that the kinds of disciplines interested in the field vary by year. ANOVA result based on the number of extradisciplines was significant and the linear contract between the year group of 1975,1980,1985 and the year group of 1990,1995, 2000 was highly statistically significant. Interdisciplinary diversity was further examined through measures of Citations Outside Category and Brillouin's Index. Kruskal Wallis test showed significant results, however, when the two measures were considered independently, only the contrast of the year of 1990 to the prior three years was significant.

Introduction

The interdisciplinary characteristics of Information and Library Science (ILS hereafter, used in this paper as interchangeable to Library and Information Science) is a long debated issue and the discussion has been often embedded in the exploration of the disciplinary nature and intellectual structure of Library Science or Information Science itself (Bates, 1999; Saracevic, 1999). The bibliometric analysis of ILS interdisciplinarity, on the other hand, often reaches the conclusion that ILS is a relatively insular field with limited impact on other disciplines (Broadus, 1971; Peritz, 1981; Bracken & Tucker, 1989). Such a conclusion, nevertheless, ought to be accurately validated through a detailed inspection with chronological mapping. This study attempts to outline the interdisciplinary breadth of the ILS field through studying citations to journal publications within the field. Moreover, the development of the ILS interdisciplinarity is captured through examining citations to randomly selected articles in six selected years from 1975 to 2000. The diversity and change in the number of extradisciplinary citations over three decades is greatly impressive, and the evolutionary trend gives a very good indication of directions of future development of ILS.

Literature Review

Numerous papers have declared that information science is an interdisciplinary, meta-field (Brook, 1980; Bates, 1999; Saracevic, 1999), however, as suggested by Karki (1996), there has not been very much empirical evidence either to support or to refute the interdisciplinarity of information science. On the other hand, studies by Broadus (1971), Gatten (1991), and Bracken & Tucker (1989) found that library science is much less interdisciplinary than other social science disciplines. The self-citation rate for ILS was found to be in the range of 74–80% (Peritz, 1981; Bracken & Tucker, 1989). In 1999, Buttlar's data show a lowered rate of self-citation, that of 50%. Buttlar examined 61 dissertations in Library and Information Science and found that roughly half of the time the citations are to the field itself. On the other hand, disciplines such as Education and Computer Science were highly cited. The author thus concluded, “Library and information science is definitely an interdisciplinary field and has close relationships with the fields of education, computer science, health/medicine, psychology, communications, and business, among others” (Buttlar, 1999, p241).

Multiple measures of interdisciplinarity such as Citations Outside Category (COC), Brillouin's Index, and Pratt Index have been proposed and used in the literature to quantitatively assess the openness of an academic discipline (Choi, 1988; Hurd, 1992; Steele & Stier, 2000; Morillo, Bordons, & Gomez, 2001). Among them, Pratt index measures the concentration and scattering of documents (Pratt, 1977). The COC approach calculates the percentage of total citations that are outside the cited journal's subject category. When the citing journal has a subject category that is different from that of the cited journal, the citing document is perceived as coming from an area of study that is extradiscipline. Steele and Stier (1999) argue that the COC approach does not provide a complete and accurate account of citation data. Instead, they suggest that Brillouin's diversity index (Brillouin, 1956) better captures the richness and relative abundance of observations. The Brillouin's index (H) is expressed through the following equation:

equation image

Where N is the total number of observations, and ni is the number of observations in category i.

Despite rich theoretical grounding and empirical studies on interdisciplinary nature of various subject matters, such research work on ILS scholarship lacks substantial data that map the evolution of interdisciplinary breadth of the field as shown through journal citations. This paper is motivated by the idea of gathering evidence to capture the developmental path of ILS intellectual structure and to further perform statistical tests on the change in interdisciplinary configuration of the ILS field.

Research Questions

  • 1The study addresses the following three questions:
  • 2What is the interdisciplinary breadth of Information and Library Science (ILS) manifested through citations to the ILS journal publications?
  • 3What is the trend of development of ILS interdisciplinarity manifested through six years in four decades?
  • 4Are there any statistically significant differences in terms of the frequencies of extradiscipline citations among the years selected?

Methods

The basic data used in this study were collected through searching three citation indexes produced by Institute for Scientific Information. In the following sections, the description of the research methodology is progressed through research design and data analysis methods.

Research Design

The sample data of the study include citation counts to the publications in the subject category of “Information Science & Library Science.” Science Citation Index (dialog files 434 and 34), Social Science Citation Index (dialog file 7), and Arts and Humanities Citation Index (dialog file 439) were searched, with the publications limited to journal articles (limitall/art) and years of publications being 1975, 1980, 1985, 1990, 1995, and 2000. Duplicates were removed. For each year, 25 publications were randomly selected and citation frequency and subject category distribution of the citing documents were recorded. Overall, data are citations to 150 ILS documents, 25 from each of the six years. For each publication, the total number of citations was noted, as was the total number of self-citations and extradisciplinary citations. For extradisciplinary citations, the actual distribution of subject categories was recorded. An extradisciplinary citation is defined as the citation coming from a discipline outside the discipline of the cited document. In this study, any citation that is published in a journal that do not belong to “Information Science & Library Science” ISI subject category is considered to be an extradisciplinary citation.

The exact searching sequence is as follows:

  • (1)Searching ILS journal article publications. For example, in searching citations to 1975 publications, the commands are:b 434, 34, 7, 439; limitall/artS sc=“information science & library science” and py=1975rd
  • (2)Capturing selected publications. Six sets of 25 random numbers were generated. Selected publications were typed or displayed and basic bibliographic information of each document was captured through dialoglink “capture file” option.
  • (3)Searching citation information. With each selected publication, citation information was collected through searching the CR (Cited Reference) field. For example, in searching journal article citations to the article below:

TITLE: OPTIMIZATION OF LIBRARY EXPENDITURE ON BIOCHEMICAL JOURNALS

AUTHOR(S): LINE MB

JOURNAL: JOURNAL OF DOCUMENTATION, 1975, V31.N1.P36–37

LANGUAGE: ENGLISH DOCUMENT TYPE: ARTICLE

Following commands were issued:

B 434, 34, 7, 439; limitall/art

Scr=line mb, 1975, v31,?

At the time of searching (November 2003), the system retrieved 27 citations, with 25 citing documents from the ILS field itself, and 3 from others. Extradisciplines included communication, multidisciplinary sciences, and veterinary medicine. Note that one citing document had two subject categories — “communication” and “information science & library science,” and thus the total of self-citation and extradisciplinary citation exceeds 27. The data set contained many cases for which the total citation count is more than the simple addition of self-citation and extradisciplinary citation because the fact that many items belong to multiple subject categories. In cases where there are subdisciplines of a general discipline, only the top-level discipline name was recorded. For example, the label “Computer Science” represents all instances of “Computer Science, Theory & Methods,” “Computer Science, Interdisciplinary Applications,” “Computer Science, Artificial Intelligence,” among others.

Data Analysis

The independent variable for the study is the year of publication, whereas dependent variables include various computed counts, such as total number of citations, number of self-citations, number of extradisciplinary citations, and other counts such as binary counts of extradisciplines (i.e., counted 1 if an extradiscipline cited a given ILS publication regardless of frequency, 0 if no extradisciplinary citations for the publication).

Descriptive data analysis was performed to report extradisciplinary citation distributions and change in interdisciplinary citation patterns. Correlation analysis of various dependent variables was performed. One-way ANOVA, with the contrast of two year-groups embedded, was conducted to test differences in extradisciplinary citations to ILS publications. Both COC and Brillouin's Index were used to test year-wide structural change. Results of analyses are outlined in the following section.

Results

The presentation of results will progress from reports of descriptive data, to social network-based graphical charting of interdisciplinary configuration, then to inferential statistical tests on the significance of change in the years included. Statistical tests on measures of interdisciplinary diversity, namely, COC and diversity index, are also presented.

Publication and Citation Distributions

Figure 1 displays the total frequency of ILS publications in the year of 1975, 1980, 1985, 1990, 1995, and 2000. While there appears an increment of 500 more publications in 1980 as compared to 1975, the uprising peak of 1990 is more apparent, as it reaches nearly 2,600 publications.

Figure 1.

Total number of publications of sample years

As a comparison, Figure 2 illustrates the total number of citations to the 25 randomly selected ILS publications of the six years. As it is shown, 1995 is the peak time for the total number of citations, and 1990 received only 51 citations, which are almost 30 citations less than that of 1995. This could mean that in 1990 there are relatively more noncitations of the publications than those in other years.

Figure 2.

Total number of citations to 25 publications in the same years

Table 1 contains raw counts of citations in categories of total citations, self-citations, and extradisciplinary citations. It is worth noting that the three years prior to 1990 are all below 20 whereas in and after 1990 the counts are all above 20. The year of 1990 saw the highest ratio of extradisciplines versus self-citations.

Table 1. Raw Frequency of Citation Distributions
YearTotalSelfExtra
1975594516
1980494210
1985554614
1990513334
1995805330
2000775923

Number of Extradisciplines

If counting the number of disciplines in extradisciplinary citations of a given year, the year of 1990 again stands out to be the one that received the highest number of extradisciplinary citations per publication (1.16). While all other years received less than 1 extradisciplinary citation, the three years prior to 1990 received less than 0.5 extradiscipline per publications. Figure 3 illustrates the distribution.

Figure 3.

Average number of extradisciplinary citations per publication

The correlation analysis of absolute frequencies of total citations, self-citations, and extradisciplinary citations revealed that there are significant correlations between all three dependent variables. While Pearson's correlation between self-citations and total citations is the strongest (r = 0.920), the correlation between the counts of total and that of extradiscipline is also more than a random chance (r = 0.613).

Development of Interdisciplinary Configuration

The multiplicity of citation patterns in each year makes it difficult to include all extradisciplines in one table. Table 2 includes only the citing disciplines that are common among three or more years. Note that the close affiliation of ILS and computer science is evidenced by the fact that for every year selected, there are citations from computer science to ILS publications. Communication also cites ILS quite frequently, citing five out of six years. The next level of disciplines includes education and management science. Business, engineering, multidisciplinary sciences, and psychology are the disciplines that cite ILS in three different years. Note that the values here in each cell are absolute citation frequencies.

Table 2. Some Most Common Extradisciplines
Discipline197519801985199019952000
Computer Science2322183
Communication201135
Education014360
Management100433
Business400120
Engineering100011
Multidisciplinary Sciences110001
Psychology011001

Figure 4 shows the proportion of extradisciplinary citations over total citation counts. Only four disciplines -computer science, communication, management, and education are included. It is clear that computer science saw a dramatic climax in 1990. Apparently around the time when the World Wide Web started, computer science became greatly involved in the scholarly communication exchange with library and information science researchers. The involvement could take the form of collaboration with ILS researchers or borrowing/citing from the ILS literature (as per Klein, 1990).

Figure 4.

Proportion of total citations that are from computer science, communication, management, and education

Figure 5 illustrates the proportion of citations from the four disciplines over the total number of extradisciplinary citations. While computer science again has its highest point in the year of 1990, education reaches its peak in 1985. Both communication and management science have their highest point in the year of 2000.

Figure 5.

Proportion of total extradisciplinary citations that are from computer science, communication, management, and education

The representation of the interdisciplinary citation traffic within each year is actualized through a piece of software for social network analysis (Borgatti, Everett, & Freeman, 1999). Using netdraw, both the direction and the frequency of citations are illustrated in the graphics, and Figure 6 is an example. An arrow pointing to ILS indicates one discipline citing ILS. The thickness of the line is proportional to the frequency of the citation. Figure 6 reflects the disciplinary distribution of citations to ILS publications in 1990. Note that the thickness of the line is a reflection of the absolute citation frequency. Citations to 1990 ILS publications consist of papers from 14 disciplines, including self-citations. There are a total of 30 self-citations, 20 citations from computer science, six citations from mathematics, four citations from management, and two citations from education as well as from social science.

Figure 6.

Citation pattern to ILS 1990 publications

Overall Interdisciplinary Citation Patterns. Figure 7 displays the full range of all the extradisciplines citing ILS publications in six years. The total number of extradisciplines are 34, ranging from scientific disciplines such as physics, chemistry, engineering, mathematics, medical sciences, to social sciences such as sociology, business, psychology, political science, and to arts and humanities fields such as literary review, geography, and history, to list but a few.

The nodes in the shape of square are the disciplines that cite ILS for more than 10 times, and those in the shape of diamond are the ones that cite ILS more than four times. In terms of the total frequency, the top four disciplines that often cite ILS are (1) computer science, which cited ILS 39 times, (2) education, 14 times, (3) communication, 12 times, and (4) management, 11 times. Disciplines that cite ILS more than four times include business (7 times), mathematics (7 times), and medical science, political science, and urban studies (all 5 times respectively). Note the frequency ratio used above is the absolute total frequency, which is the sum of citation counts to all sampling publications in six years. This measure is indifferent with regard to whether the citation count is for individual publications or across all publications. In the next section, the measure of citation frequency is converted to the binary count of citations, using publication as the unit of analysis regardless the citing frequency of individual publications.

Figure 7.

Extradisciplines that cite ILS in absolute counts

Binary Counts of Interdisciplinary Citations Figure 8 illustrates the interdisciplinary breadth of ILS, using binary counts at the publication level. Note that ILS_Self is also included in the chart. ILS self citation binary counts are 89, which means about 60% of publications received citations from within. Similar to the absolute frequency counts, other disciplines that cite more than six ILS publications, represented by the square node shape, include computer science (binary counts = 21), communication (binary counts = 11), education (binary counts = 9), and management (binary counts = 7). Disciplines that cite more than two ILS publications, namely, business, engineering, economics, medical science, psychology, and multidisciplinary sciences, are represented in the diamond shape. Note that at this level a slightly different set of citing disciplines emerged. Instead of political science, urban studies, and mathematics, the disciplines that cite multiple ILS documents are psychology, multidisciplinary science, engineering, and economics. Other disciplines are the same as those based on the absolute frequency count.

Figure 8.

Extradisciplines that cite ILS in binary counts

Table 3 shows the proportion of ILS publications being cited by the four heavily interested disciplines. Note that in 1990, 40% of the ILS publications were cited by scholars from computer science.

Table 3. Proportion of ILS Publications being cited
 197519801985199019952000
Computer Science0.040.080.080.40.120.12
Communication0.0800.040.040.080.2
Management0.04000.080.040.12
Education00.040.080.120.120

The structure of citing disciplines is represented by Figures 9 and 10. The former illustrates those disciplines that are common among a minimum of two independent years. The latter represents unique disciplines by each year. There are a total of 17 disciplines that are interrelated among the years. Disciplines that are on the left side of Figure 9 are the ones that cite ILS publications of three or more than three years.

Figure 9.

Interdisciplinary breadth based on common citing disciplines

Figure 10.

Interdisciplinary breadth based on unique disciplines

As shown in Figure 10, the year of 1980 did not have any disciplines that are unique. The range of unique extradisciplines in the remaining five years shows that the variety of disciplines interested in ILS has been growing. The early years saw interests from a variety of domains but mostly soft science (social sciences and the humanities), the post 1990 years received citations from mostly hard science, with 1995 publications attracting telecommunication interests and 2000 publications gaining interests from biology related areas. The trend of change may be more comprehensive and accurately modeled in Figures 11 and 12, where the interdisciplinary citation structure is charted along two year groups: 1975, 1980, and 1985 (Figure 11) and 1990, 1995, and 2000 (Figure 12).

Figure 11.

Interdisciplinary breadth of ILS in years of 1975, 1980, and 1985

Figure 12.

Interdisciplinary breadth of ILS in years of 1990, 1995, and 2000

A visual comparison of the two figures indicates that the cross-disciplinary citation structures are much more complex in the second year group than that of the first year group. This may be an indication that the ILS discipline became mature in or after 1990, and thus obtained attention from a variety of extradisciplines. This observation is further tested through inferential statistics, as reported in the next section.

Year Contrasts of Interdisciplinary Breadth

The ANOVA test on the number of extradisciplinary citations of the six years yielded significant result (F=2.384, df=5, p<0.05). The linear contrasts of year between before and after 1990 are highly significant. Table 4 Table 5 contain SPSS results, showing that while contrasts 1 (1975, 1980, and 1985 vs. 1990, 1995, and 2000) and 3 (1980 vs. 1990) are significant, contrast 2 (1975 vs. 1995) is not.

Table 4. Contrast Coefficients
ContrastYEAR
 197519801985199019952000
1111−1−1−1
21000−10
3010−100
Table 5. Contrast Tests
 contrastStd errortdfSig.(2-tailed)
Assume equal1.555−3.09144.002
 2.321−1.87144.063
Variances3.321−2.62144.010
Does not assume equal1.555−3.09109.002
 2.323−1.8638.5.071
Variances3.348−2.4234.3.021

ILS Interdisciplinary Diversity

When data were converted into two interdisciplinary citation measures — COC and Brillouin's diversity index, the year contrasts present a slightly different picture. Both COC and diversity index contain information of the richness of extradiscipline citations, with individual publications accounted for. Figure 13 displays the comparison of two sets of values, with the top line as average COC measure and bottom as average diversity index measure. The COC measure shows a distinctively high rising pike of the citation diversity in the year of 1990.

Figure 13.

Interdisciplinary diversity of ILS as measured by average COC and average diversity index

Kruskal Wallis test based on both COC and diversity index values was significant (χ2=15.165, df=5, p<.05). However, since the calculation of both COC and diversity index is based on the same information, they should be used independently. When COC measure and diversity index were separated, the Kruskal Wallis was no longer significant. On the other hand, if the post-1990 years are excluded, ANOVA tests produced significant results, indicating a remarkable change of diversity taking place in the year of 1990. In ANOVA of COC test (F=2.77, df=3, p<.05), the linear contrasts of the year 1990 to each of the prior years are significant. On the other hand, even though the ANOVA result is also significant for diversity index (F=2.74, df=3, p<.05), only the linear contrast between the year of 1990 and the year of 1975 is significant.

Discussion and Conclusion

The study produced a rich set of results relevant to the evolutionary mapping of the interdisciplinary breadth of the ILS field. With a total of 34 extradisciplines citing ILS publications sampled in six years of four decades, the data appear to support the assertion that ILS has reached its scholarship maturity and that it is highly interdisciplinary field that attracts learned interests from a variety of disciplines from the domains of science, social science, and the humanities.

To address three research questions separately, it is clear that ILS has an abundance of intellectual substances that are of interest to multiple extradisciplines, including computer science, communication, education, and management science. If a comparison of this study with Buttlar's data is made, a relatively full picture may occur. Buttlar (1999) looked at ILS dissertations from 1994 to 1997, and found that 11.45% of the citations are to education, 5.72% to computer science, and 3.79% to medicine and to sociology. Relatively small proportions of ILS citations were to communication (1.96%) or to business (1.84% to economics). Figure 14 is the netdraw chart of Buttlar's original data. Only the disciplines that received more than 1% of the total citations are included. Buttlar found that publications from more than 29 extradisciplines were cited by the 61 ILS dissertations. The disciplines that received less than 1 % of the ILS citations include engineering, mathematics, chemistry/physics, law, and political science, among others. In Figure 14, the disciplines that are in the square shape are the ones that were cited by ILS dissertations for more than 3.5% of the time.

Figure 14.

Citations from ILS publications (source: Buttlar, 1999)

The results of this study, if combined with Buttlar's data, reveal citation exchange between extradisciplines and ILS. The fields that have mutual scholarly communication with ILS include computer science, education, communication, business, psychology, political science, chemistry, law, and biology, and many others. Table 6 contains comparative data between citation proportions based on this study (Citing ILS %) and those from Buttlar's paper (Cited by ILS %). The 18 fields of studies that hold mutual citation dialogue with ILS range from science, social science, and the humanities. This serves as a convincing piece of evidence for the interdisciplinary nature of the field. Readers are warned that since the two sets of data were drawn from different samples and research settings, the values may only be used as a rough estimation of the full range of activities of the ILS interdisciplinary structure.

Table 6. Citing and Cited Disciplines by ILS Publications
DisciplineCiting ILS % (this study)Cited by ILS % (Buttlar, 1999)
Computer Science10.515.72
Education3.7711.45
Communication3.231.96
Business1.891.90
Mathematics1.890.11
Health/Medicine1.353.79
Political Science1.350.24
History1.081.13
Geography1.080.14
Psychology0.812.58
Economics0.811.84
Engineering0.810.35
Chemistry0.540.62
Law0.540.23
Sociology0.273.79
Arts0.271.55
Literature0.271.31
Biology0.270.15

In answering the second and third questions regarding the year contrast in interdisciplinary development of ILS, it is clear that the year of 1990 is a high point and turning period for the field. On one hand, statistical tests have confirmed that 1990 was the year that holds the richest level of diverse interests from extradisciplines to the field. On the other hand, the period seems to echo the movement of the Internet and Web technology, since it marked a significantly high concentrated attraction from computer science. Such a rich and high level of attention is an indication of prominent correlation between computer science and ILS. The impact of computer science to the ILS field continues to be strong since 1990 till this day. There are increasing interests from communication, with the year 2000 holding 20% of the citations from communication. Management science also shows the highest level of interest in 2000, whereas education has reduced to no interest at all in 2000 from relatively high interests in prior years. Overall speaking in 2000, more citations were coming from biosciences, it appears that the ILS is moving towards more technology oriented and being applied to highly specialized subject areas.

Future studies on subject categories of references of each of the 150 publications, matched with the citation data covered in this paper, would help to establish a much well-rounded assessment of the ILS interdisciplinary structure. Such a project is currently under way.

Ancillary