LIS research in Africa : how much is it worth ? A citation analysis of the literature , 1986-2006

This paper examines library and information science (LIS) literature as produced and published by researchers in Africa in order to establish the productivity and impact of LIS research in the region. Using publication counts, and more specifically, citation analysis, the paper demonstrates that the research output and impact of LIS literature on the continent is relatively low when compared to other disciplines in Africa, such as social sciences. Correspondingly, the research forms a small percentage of both the national and world total LIS research output. A comparison of countries indicates that South Africa presently leads in terms of both research output and citations, and Nigeria and South Africa account for over 70% of the total number of Africa's publications and citations. Other findings are discussed, in addition to recommendations for further research, and ways in which to improve the visibility of LIS research in Africa.

volume; the type of articles; the distribution of references by volume; the range and mean number of references per article; the authorship patterns of articles; the ranked list of the most prolific contributors of articles; the ranked list of authors by geographical affiliation; and the ranked list of authors by institutional affiliation.Other studies that have focused on the analysis of specific LIS journals/periodicals in order to measure productivity in LIS research include He & Spink (2002), Lipetz (1999), Raptis (1992), and Siddiqui (1997).Aina (1999) deviated from the analysis of specific journals/periodicals in terms of quantity and quality, and focused on individual researchers' productivity.He identified 34 top-ranking researchers in LIS in Africa, and examined a total of 294 papers that the researchers had published between 1990 and 1995.
Bibliometric analyses of LIS literature specific to particular geographic regions are also common.For example, Ocholla (2000) used the South African Bibliographic and Information Network (SABINET) to analyze LIS research in South Africaproduced between 1993 and 2000 -in order to determine the research capacity and potential in the country.He analyzed data using various indicators, such as research themes/subjects, institutions, quantity of the reports, language of medium, gender and population groups.In an article entitled "Library literature in Ghana, 1950-1994" Kadiri (200 I) used two bibliographies to investigate, among other things, the number of published materials in librarianship in Ghana, formats of publication, research collaboration, and sources of librarianship literature.
At the international level, Cano (1999) evaluated LIS research in Spain over a span of I 7 years, during which she identified and analyzed a total number of 354 articles.Uzun's (2002) bibliometric study of LIS research in developing countries and Eastern European countries is one of the few studies to have covered a broader spectrum of countries/geographic regions.Uzun (2002:21) examined "a set of 21 core journals in the field of library and information science (LIS) from 1980-1999" and sought articles with either principal or co-authors from developing countries (DCs) and the formerly socialist Eastern European countries (EECs) in order to identify the productivity of research articles by librarians and information scientists from the aforementioned areas in international journals.

Purpose of the study
This study examines LIS records produced by researchers in Africa between 1986 and 2006 in order to measure the quantity and quality of LIS research in Africa in terms of the number of publications and citations as well as its impact when compared with research in selected social science disciplines.In view of the above, the study sought to determine: I.The level of coverage of LIS records in the Institute for Scientific Information (lSI) and Library and Information Science and Technology Abstracts (L1STA)databases

Methodology
The study targeted a total of 53 African countries (see Table I).Of the 53 countries, 22 are English speaking.Worth noting too is the multi-lingual nature of some countries where there are several official languages (e.g.South Africa which has II official languages, Cameroon Besides English, French is widely spoken in Africa, especially in Sub-Saharan Africa.The language is an official language in 23 countries.Portuguese-speaking countries total five, while there are II Arabic-speaking countries (see http:// www.nationsonline.org/oneworld/africanJanguages.htm ).
Two sources of data, namely lSI's Web of Knowledge and EBSCO Online were used to extract LIS publications.Specifically, two databases from the Web of Knowledge (i.e.SCI and SSCI) and one electronic database from EBSCO online (i.e.L1STA),were used to collect data.The two lSI databases cover close to 35 different document types from a variety of subject domains (mainly science and social sciences publications) while L1STA indexes more than 600 periodicals plus books, research reports and proceedings.Subject coverage in L1STAincludes librarianship, classification, cataloging, bibliometrics, online information retrieval, information management and more.
Four approaches were used to obtain LIS records, as follows: Sci 2007, 73(2) I.The total world productivity of LIS records in lSI was determined by searching for two keywords, namely, TS="Librar*" OR TS="lnformation" as 'Topics' of discussion.The search was then refined using the "Subject Categories" option as provided by lSI in order to identify the 'Information Science & Library Science' subject category records.It is worth noting that the use of only two keywords may have resulted in fewer LIS records than has been published.
2. Another search was conducted within the lSI databases (Le. the Science Citation Index [SCI] and the Social Sciences Citation Index [SSCI]) in order to obtain only LIS records as produced by researchers in Africa by: a. Identifying a country's total number of publications through an advanced search using '~D='country name"', where AD stands for 'Address' field.
b. Analyzing the identified records using lSI's '~nalyze" feature.The records were analyzed according to 'Subject categories' and whichever country that did not yield any LIS record was excluded from the final analysis.
3. In this way, LIS records were identified, downloaded and saved as .txtcomputer files and thereafter analyzed in order to measure LIS research productivity by country; proportion of LIS research to each respective country's total productivity; the cited ness and uncitedness of LIS research; the most cited works; and the trends of LIS research impactcalculated as the average number of citations per record in each year -in each country.
4. In the case of L1STA,only the names of countries were used to search for and download LIS records specific to African countries using the a uniform search strategy, i.e.AF = 'country name' where AF is the Author's Institutional Affiliation (or Address) field tag.In this way, all records produced by authors affiliated to institutions which are located in African countries were captured.
The above approaches employed an advanced search in the case of the Science Citation Index and the Social Sciences Citation Index.The same procedures were followed to identify publications published by researchers in Africa in selected social science disciplines (Le.Anthropology, Economics, Education, Geography, History, Political Science, Language and Linguistic theory, and Sociology) for comparative purposes.The Library and Information Science and Technology Abstracts (L1STA) database was used in order to compare the coverage of LIS records in the two bibliographic databases (Le.L1STAand lSI databases).A list of countries that produced LIS records was compiled from the lSI databases and then used to download L1STArecords for comparison purposes only.An attempt was made to use the Library and Information Science Abstracts (LISA) data, but because of the unavailability, on the database, of all contributing authors' addresses, it was excluded from the study.
Table I Target population (African countries) .Cape Verde In the absence of a controlled subject vocabulary to describe LIS research and given that the discipline consists of several terms/phrases which are common in other disciplines (e.g.link analysis, content analysis, co-word analysis, data mining, information technology, knowledge management, etc.), the approach that was adopted in this study provided as accurate data as possible.An attempt was made to search for LIS records in the SCI and SSCI using specific terms/phrases, such as those provided by the Association for Library and Information Science Education (2003) in its classification of LIS research areas, but this proved difficult and yielded inaccurate data.This limitation perhaps explains why various researchers evaluate LIS research using LIS-specific journals and databases.
Data extracted from the SCI and SSCI databases was analyzed using the STIKISI computer-aided software to obtain citation and publication frequencies in each year of publication in each country.The average cites per record, which was used to measure a country's publications impact and LIS publications impact in Africa, was calculated as the ratio of the total number of citations to the total number of publications.The rank of the LIS subject category relative to each country's total number of subject categories was used to measure the performance of LIS in the respective countries.
Relative performance, whose formula is herein introduced by this author, was thus calculated as follows: Relative Performance (rp) = Rank (position) of LIS Subject Category Total no. of Subject categories where 0 ~rp

~I
The relative performance of LIS was deemed high if the ratio was closer to zero (0).For instance, if a LIS subject category ranked second out of a total of 8 subject categories in country A, the rp was calculated as 2/8 = 0.25.If, in country B, the LIS subject category ranked 2nd out of a total of 50 subject categories, the rp would be 2/50=0.04.Country B's LIS performance would therefore be rated as higher than that of country A.
Finally, it should be borne in mind that the inclusion of other document types such as book reviews, editorial material, meeting abstracts, notes, letters, reviews, and reprints (see Appendix A) in the final analysis resulted in higher productivity and impact than it would have been had we analyzed only journal articles which are assumed to reflect scholarly publishing of research findings.Nevertheless, worth noting is the fact that some of these document types not only ignite debate that may lead to research but also, in the process, attract citations, an aspect that was the subject of discussion in this paper.It is also acknowledged that visibility of an author, institution or country is often measured by, among other indicators, the number of citations that docu'ments published by each of the entities receive.It is in this respect that we included all the document types in the analysis.

Results
This section provides the findings of the study under the following subheadings: the world output of LIS publications; LIS research output in Africa; trend of LIS research in Africa; LIS research's contribution to both national and world output; number of citations by country; impact of LIS research; rank and relative performance of LIS; most cited LIS records; cited ness and/or uncitedness of LIS records; and the productivity and impact of LIS and selected social sciences.

World productivity of LIS records 1986-2006
Table 2 compares LIS publications as indexed in lSI's SCI and SSCI to L1STA databases.There were a total of 15934 records in lSI, while L1STAyielded a total of 823199 publications between 1986 and 2006.The Table illustrates a mixed pattern of growth, whereby some years recorded a growth rate as low as 5 (in the case of lSI) and 1290 records in L1STA.
The highest increase in the number of publications was recorded in 1991 (lSI) and 2003 (L1STA), which yielded a difference of 165 and 13962 records between 1990and 1991(lSI) and 2002and 2003 (L1STA), respectively.Similarly, the number of records peaked to 890 in 1998 in lSI and 76902 in 2005 in L1STA.

LIS research output in Africa 1986-2006
Only records that were downloaded from the lSI databases were analyzed and presented in Table 3, because the main focus of the study was to compute and compare productivity and impact of LIS publications in Africa.Only the lSI databases provided both research indicators, i.e. number of publications and citations.L1STAprovided only the number of publications.The results in Table 3 show that there were a total of 26 countries in Africa which produced at least one lSI listed LIS record.The leading country was South Africa which produced a total of 439 records, followed by Nigeria (259), while Botswana and Ghana were ranked number three with 59 records each.Kenya (37) came fifth, followed by Egypt (36), Swaziland (29) and Ethiopia (25).At the bottom of the   -,.... M 't" U'"I -.0 to-. (Xl C1' 0 -,.... M 't" U'"I -.0 "Ei ... ,.... ,.... ,.... ,.... ,.... ,.... to t--co en 0 .,-N (") '<t L() to t--co en 0 .,-N (") '<t L() to co co co co en en en en en en en en en en 0 0 0 0 0 0 0 en en en en en en en en en en en en en en 0 0 0 0 0 0 0 .,-.,-.,-.,-Further analysis of the lSI data, as shown in Table 4, indicates that South Africa's contribution accounted for a mere 0.46% of the total national output and 2.76% of the world's total LISoutput, while Nigeria's productivity accounted for 1.22% of the country's national output and 1.63% of the world's total LISoutput.Other findings were as follows, in the order of percentage of the total national output and the world's total LIS output: Botswana (3.33%, 0.37%), Ghana (1.87%,0.37%)and Kenya (0.31%,0.23%).It should be noted that the highest LISproducers in terms of the number of LIS records (Le.South Africa, Nigeria, Botswana, Ghana, and Kenya) were not the leaders when their percentage contribution to the total national output was considered.The highest national contribution was recorded by Swaziland (4.32%).As regards L1STA,each country's contribution fell below 0.1%.The highest contribution was recorded by South Africa (0.0658%) followed by Nigeria (0.0238%), Botswana (0.0 I 12%), Ghana (0.0083%), and Kenya (0.0050%) as shown in Table 4.

Most cited LIS records
The 'TC' (i.e.Total Cites) field tag was used to identify the most cited records in Africa.Table 7 provides the records which had received 6 or more citations at the time of conducting this study (i.e.April 2007).They include, in descending order: Lawan SM ( 43 7 is the dominance of South Africa as one of the top most cited countries.Out of the top 34 most cited records, 23 (67.6%) originated from South Africa while Nigeria yielded 6 (17.6%) followed by Egypt and Kenya which produced 2 (5.9%) each and Tanzania (I, 2.9%).

Discussion and conclusions
Worldwide, and as shown in Table 2, productivity of LIS publications appears constant, implying a steady output.However, there were several instances during which productivity reduced by large margins, for example between 1999and 2000(lSI), and between 1996and 1997 (L1STA).Equally important were the instances during which research output leapt by almost similar margins to the drops during reductions, e.g. between 1990 and1991 (lSI)  show that LIS authors are not consistent in terms of their research activities, Le. productivity or publication.Whether this could perhaps be attributed to financial or time constraints could not be substantiated from the data.It would be interesting to determine the cause of the pattern witnessed in this study, and subsequently recommend areas of improvement.
A trend in keeping with the above was witnessed in Africa, whereby South Africa emerged the most productive country.While various researchers (e.g.Alabi and Saracevic in Jacobs, 2000) have noted a tremendous growth in scientific activities in most third world countries, LIS research has generally remained low in Africa, as indicated by each country's LIS research output from the perspective of national and world research output.Some of the factors that have been said to influence or affect productivity in Africa include lack of funds and basic facilities, the intellectual and physical isolation of researchers, insufficient personnel to run programs, fragmentation of effort in research, lack of vision and direction by the governments of Africa, and the poor self-image of the region in basic research (Mweene, n.d.).
A comparison of productivity in each African country indicates that South Africa and Nigeria produced over 70% of the total LIS research output in Africa (Le.992).The remaining countries (which are the majority, Le. 24) produced a mere 30% of Africa's total LIS research output.Previous studies (e.g.(Narvaez-Berthelemont, Russell, Arvanitis, Waast, & Gaillard, 200 I) produced similar results.Generally, South Africa's dominance in terms of research output in Africa, especially in the post apartheid era, has also been acknowledged in other studies (e.g.Jacobs, 2002).The high research output by the two countries could be attributed to several factors, chief among them being the countries' research policies.Whereas Nigeria's research policy is not clear to the author, South Africa regards research highly, and has therefore put in place a number of mechanisms that enhance research output.For instance, the government, through the Ministry of Education, provides subsidies for every research paper/article that is published in an accredited journal by researchers in institutions of higher learning in the country.The generated funds are split between the author's university of affiliation, the department that he/she works for, and the individual author.This incentive may have contributed to the high pattern of productivity witnessed in this and other studies.
Additionally, the number of LIS schools and researchers in each country may influence productivity.South Africa boasts the largest number of university LIS schools/departments (Le.15) followed by Nigeria (7), Kenya (3), Sudan (2), and Tanzania (2) while Botswana, Cameroon, Democratic Republic of Congo, Ethiopia, Ghana, Malawi, Namibia, Senegal, Sierra Leone, Uganda, Zambia and Zimbabwe have one LIS school/department each (Minishi-Majanja & Ocholla, 2004).There was no data offering the number of LIS schools in Arabic countries.Notably, despite the fact that there are two LIS schools in Sudan, there were no LIS records produced by that country as indexed and reflected in lSI's databases.
A search of ~F=Sudan' in L1STA,however, produced three records.Other factors that may influence productivity are the number of scholarly journals that are published in a country.According to the Urlich Periodical Directory (accessed on 12 th June 2007) Nigeria and South Africa publish a total of 204 and 483 academic/scholarly publications, respectively.In comparison, Botswana, Ghana and Kenya produce 9, 30 and 71 academic/scholarly publications.It was not possible, however, to ascertain the LIS-specific journals that are published in these countries.It was noted that none of the L1Sspecific journals which are published in the African countries investigated are indexed in lSI databases.
Further research is therefore recommended to study the influence of the number of LIS schools/departments and researchers (including such personal characteristics of researchers as age, academic qualifications, etc) on productivity.Another notable observation was the dismal performance of Northern Africa's countries.This may be attributed to the language factor.The lSI prefers indexing records that are published in English, and whenever a paper is prepared in any other language, the institute requires that an English language version should be provided.This limits its coverage of records that are published in any other languages, including those written in Arabic or other African languages.
Concerning the number of citations, South Africa led with 498 citations, followed by Nigeria (232).Again, the two countries' combined productivity exceeded the sum total of the rest of the countries, numbering 24.There were four countries (i.e.Algeria, Angola, Sierra Leone and Tunisia) that did not receive citations.It should be noted, however, that these countries were among the least productive in terms of the total number of records.This does not, however, mean that the more the number of records a country produces, the higher the number of citations it is likely to receive.It simply means that high research productivity may result in a higher number of citations in a country since a country with such high productivity may have more chances of receiving more citations than a country with low research productivity.
In terms of the impact of LIS research, which was measured by the average citations per record, it was noted that less productive countries dominated the top positions of countries with the highest average citations per record.These included Tanzania, whose 12 records generated 2.7 citations per record.Others in this category include Zimbabwe, Senegal and Malawi and Zambia.A comparison was also made between the number of cited and uncited records in order to determine the extent of the influence and visibility of LIS research conducted in Africa.It was observed that most records, especially in the most productive countries, remained uncited.Whether the low average citations per record and uncitedness of LIS records implies poor quality, non-visibility, or the low impact of African LIS research is difficult to tell, because cited ness and/or citation impact factors do not always indicate quality (Garfield, 1993;Seglen, 1997).Nevertheless, further research should be conducted to ascertain the factors that contribute towards the uncitedness of LIS research records and thereby recommend solutions.
The ranking and relative performance of LIS in each of the 26 countries as shown in Fig 2 demonstrated that the subject domain ranked poorly in most countries, while its relative performance was found equally wanting.The best performance was reported in Botswana where LIS was ranked number 10 out of a total of 175 subject categories in which research was conducted, thus producing the performance ratio of 0.06.This may indicate that LIS is not a highly prioritized area of research in a given country especially in situations where governments influence research output, particularly in situations where the governments commission and fund research.Other factors that may explain the poor ranking of LIS may include the following: • LIS researchers may be fewer than researchers from other disciplines that are performing well.
• LIS research is largely conducted at institutions of higher learning, unlike research in pure sciences which is conducted at both industry level and within institutions of higher learning • LIS research is largely basic research while research in pure sciences is mainly applied or action research hence the need for the industry and other stakeholders to invest in the latter.• LIS research is mainly published in local (non-international or non-lSI) journals, thus affecting international visibility.When compared to other social sciences, LIS research productivity and impact (average citations per record) performed fairly well.In fact, there was little difference in the number of publications, citations and citations per record between LIS and other selected social sciences as shown in Table 9.While it ranked 4th out of lain terms of the number of publications, it was placed 6 th and 8 th positions in terms of the number of citations received and the average number of citations per record respectively.This implies that LIS performed better in terms of productivity, averagely in terms of the number of citations and poorly in terms of impact.Initiatives such as current awareness services, whereby authors alert

2.
The total world productivity of LIS records between 1986 and 2006 3. The total number of records produced by Africa over the same period 4. The trend of LIS research in Africa from 1986 to 2006 5. LIS records' share of both the national and world output 6.The total number of citations received by LIS records by African country 7.The average number of citations per LIS record in each country 8.The most cited LIS records 9.The performance of LIS research when compared to research in other social sciences & Info Sci 2007, 73(2)

Figure 2
Figure2Rank and relative performance of LIS in each country ), Miller J & Doyle BA (38), Turoff M, Hiltz SR, Bahgat ANF & Rana AR (34), and Money A, Tromp D & Wegner T (26) while the records authored by Vaughan PW, Rogers EM, Singhal A & Swalehe RMM and Cosijn E & Ingwersen P received 25 citations apiece.Also evident from Table and between 2002 and 2003 (L1STA).It is difficult to speculate what factors underlie such patterns of productivity.Nevertheless, the patterns do

Table 2
World productivity of LIS research ranked according to year ofpublication, 1986-2006

Table 3
Yearly distribution of the LIS records by African country from lSIdata, 1986-2006

Table 4
Record contribution as a percentali!e of the national (151 only) and world productivity LIS research's contribution to both the national and world output As mentioned above, of the 53 African countries only 26 authored at least one LISrecord each between 1986 and 2006.

Table 5
Total number of citations by year of publicationforAfricancountries, 1986Africancountries,  -2006

Table 6
Rank and relative performance of LIS in each countryThis section presents the rank distribution and relative performance of LIS in each African country by comparison with other subject areas of research in the respective countries.The data is reported in Figure2.The rank distribution and relative performance ratio ofthe subject domain in each country was as follows:Algeria (146/188, rp=0.78),Angola(62/91, rp=0.68).Benin (103/188, rp=0.55),Botswana(10/175,rp=0.06),Cameroon(165/191,rp=0.86).Egypt (156/222, Average cites per record by African country and year of publication Table6provides the average citations per record in each year by country.The Table reveals that Tanzania recorded the highest average number of citations per record (Le.2.7), followed by Egypt (2.6), Zimbabwe (1.7), Senegal (1.6), Ethiopia ( I.5), and Malawi (I .4).Others that recorded one or more citations per record were, in descending order: Zambia (1.3), Kenya (1.2),South Africa (1.1), and Swaziland (1.1) while Gabon, Ivory Coast, and Morocco each yielded 1.0 citation per record.Again, Algeria, Angola, Sierra Leone and Tunisia produced zero citations per record.Table6also indicates that, other than South Africa, no country received citations continuously throughout the period of study.4.7 Zambia (54/174, rp=0.31) and Zimbabwe (171/206, rp=0.83).Fig 2 provides this distribution in graphic form.
4.9 Cited ness and/or uncitedness of LIS records Table 8 provides the total number of cited and uncited records in each African country.There were a total of 308 uncited records in South Africa, accounting for 70.16% of the total LIS records produced by LIS researchers in the country.The SAJnl Libs & Info Sci 2007, 73(2) data in

Table 8
Citedness and/or uncitedness of LISrecords Comparing the performance of research and researchers in different disciplines is common in bibliometrics studies.LIS research was compared with research in selected social sciences disciplines in order to find out the subject domain's performance in each country relative to other subject domains.This comparison is shown in Table9.The data shows that LIS research output was ranked 4 th between 1986 and 1990 with 192 publications and 307 citations and 1.60 citations per record.In considering the number of publications, the first position was held by Economics -in the order of publications, I 4.10 Productivity and impact of research in LIS and selected social sciences SAJnl Libs & Info Sci 2007, 73(2)

Table 9
Productivity and impact of research in LISand other social sciences