Quality of Estonian science estimated through bibliometric indicators ( 1997 – 2007 )

The main goal of this paper is to provide an analysis of bibliometric indicators of the quality of Estonian science in comparison to its neighbours Latvia and Lithuania during the 11-year period from 1997 to 2007. Since 1990, Estonian and Lithuanian scientists more than tripled the number of articles they published in journals indexed by the Thomson Reuters Web of Science. The number of articles from Latvia has decreased relative to the general increase of published articles in the world. According to the Essential Science Indicators database, papers published by Estonian scientists had the highest impact (7.87) compared to all other former Communist bloc countries including Hungary (7.83), Latvia (5.92), Lithuania (4.95), and Russia (3.98). While Latvia failed to increase the productivity and Lithuania to improve the quality of their scientific publications, Estonia succeeded in reducing the gap both in the productivity and impact of its publications compared to the world leading countries. The observation of changes during the last three years allows identifying Agricultural Sciences, Molecular Biology & Genetics, and Social Sciences as three fastest growing fields in Estonia.


INTRODUCTION
The idea of bibliometric analysis has transformed from an intriguing possibility (de Solla Price, 1965) into a regular tool for evaluation of the scientific quality of countries and institutions (European Commission, 2005;Moed, 2005;Must, 2006).Estonia, like its two neighbours, Latvia and Lithuania, is an interesting historical case.Before regaining independence in 1991, they belonged to one of the most inefficient scientific systems in the world.At that time 5.5% of all indexed scientific publications in the world were produced by scientists working in the former Soviet Union.At the same time, only 1.7% of all citations were on works authored by Soviet scientists.This rate was about 10 times less than, for example, citations of Dutch scientists.In terms of bibliometric performance one Dutchman alone was equal to the impact of a small research institute in Moscow (Allik, 1998).Although the quality of science has certainly improved in Russia, according to a recent press release (24 February 2008) the percentage of Russia's share of all papers published in the world is about 2.8% while less than 1% of all citations are on works authored by Russian scientists (http://sciencewatch.com/dr/sci/08/feb24-08_1/).Therefore it is not only interesting but also significant to see what has happened to Estonian science after almost two decades of autonomous development.
Bibliometric indicators have shown that although Estonian science has considerably improved during the years of independence, it is still less intensive than science in the most scientifically advanced countries such as Switzerland, Sweden, or Finland (Allik, 1998(Allik, , 2003)).However, nearly two decades is a sufficient period for looking back how political decisions and administrative reforms have influenced the development of scientific research systems in Estonia in comparison to two other Baltic countries, Latvia and Lithuania, with similar historical experience (Kristapsons et al., 2003).This analysis is particularly relevant because the weakness of a set of policies crucial for longer-term development, such as innovation and technology policies, has led in the last decade to a deterioration rather than strengthening of the competitive advantages of Eastern European economies including Estonia (Tiits et al., 2008).
The main goal of this paper is to provide an analysis of bibliometric indicators of the quality of science in Estonia in comparison to its neighbours, Latvia, Lithuania, and Finland, during the 11-year period from 1997 to 2007.

METHODS
The analysis is based on the Internet version of the Thomson Reuters Web of Science (WoS) databases covering the period from 1997 to 2007.The WoS includes about 11 000 of the most influential research journals in the world.It contains three citation indices: Science Citation Index (1900-present), Social Sciences Citation Index (1956-present), and Arts & Humanities Citation Index (1975-present).The WoS also provides different analytical tools for the analysis of countries/ territories and research institutions.
The Essential Science Indicators (ESI ) is a resource that provides analytical tools for ranking scientists, institutions, countries, and journals.It is based on journal article publication counts and citation data from the WoS and analyses a ten-year and plus n-months rolling period.ESI covers 10 million articles in 22 specified fields of research (excluding humanities), and is updated every two months.ESI provides both total citation counts and cites per paper scores for different countries or territories.ESI is limited to the journal articles indexed in the WoS only.No books, book chapters, or articles published in journals not indexed by the WoS are taken into account in ESI, either in terms of publication or citation counts.

RESULTS AND ANALYSIS Productivity from 1990 to 2007
In 1990, just before the collapse of the Soviet Union, Estonia, Latvia, and Lithuania had very similar starting positions.Scientists in each of these three Baltic countries published approximately 300 papers per year in journals indexed by the WoS.Seventeen years later, in 2007, Estonians, Lithuanians, and Latvians published 1295, 1067, and 426 articles, respectively.Figure 1 demonstrates the growth of publication in the three Baltic countries from 1990 to 2007. 1 Estonian and Lithuanian contributions to the world science more than 1 Figure 1 counts all publications (also abstracts) included in the WoS based not on their nominal publication date but on the actual time they were entered into the database.All publications were included since it has been argued that conference presentations are more important vehiches of scientific communication in some areas than others. 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 tripled even after taking into account the about 60% increase in the total number of publications in the 17-year period.The relative contribution of Latvia, however, even decreased from 0.036% in 1990 to 0.029% in 2007.Thus, Estonia and particularly Lithuania increased their publication activity considerably during the last two decades while Latvia stagnated in the number of publications.Nevertheless, in both absolute and relative terms the intensity of scientific publication in the Baltic countries is still low.Even after taking into account population size Estonian scientists managed to reach 34% level compared to Finnish productivity normalized per one million of population.
For comparison, Latvia and Lithuania produced 10% and 13% of the number of papers that Finnish scientists produced per capita of their country.--------Notes: * = Only countries that published 2000 or more papers during the 11-year period.The first 22 countries/territories are above the world average (9.5).5% of the comparable Finnish citation intensity.This is only slightly more citations per capita (4%) than papers authored by Russian scientists.

The impact of Estonian science (1997-2007)
Estonia occupies the 61st place in the ranking of papers and the 52nd place in the ranking of citations.This indicates that Estonian scientists are not especially prolific but if they publish their papers they have a relatively high impact.One of the possible reasons for the relatively high impact is that until 2008 only one local Estonian journal, Oil Shale with the impact factor IF 2006 = 0.371, was indexed by the WoS.
2 It seems to be a general rule that local national journals have lower impact than the mainstream international publications.For example, Lithuania has several journals indexed by the WoS (Medicina-Lithuania, Veterinarija ir Zootechnika, Transformations in Business & Economics, Lithuanian Journal of Physics, Baltic Forestry, etc.) but its 11-year impact is 4.95 (65th), which is even lower than 5.92 of Latvia (46th).Another example is Croatia with 13 journals indexed by the WoS (Andreis & Jokic, 2008) and the impact factor of 4.61 (68th).On average, the impact of Estonian science increased by 24.4% during the last three years.The most remarkable increase was achieved in Agricultural Sciences, which was unable to surpass the threshold of essential science in 2004 but did it in 2007.The largest increase (+ 68.8%) was in Molecular Biology & Genetics followed by Social Sciences general (+ 38.8%), Geosciences (+ 35.8%), and Environment/Ecology (+ 32.5%).The only field whose impact declined was Space Science (-1.8%).Pharmacology & Toxicology managed to stay close to the world average level and its increase was marginal (0.3%).

Comparison with Latvia and Lithuania
In the year 2002, the impact of Estonia, Latvia, and Lithuania was 5.03, 3.52, and 3.97, respectively (Allik, 2003).According to the release of the ESI, from 1 January 1997 to 31 December 2007, the impact factors for Estonia, Latvia, and Lithuania increased to 7.87, 5.92, and 4.95, respectively.The increase during the last five years was 56.5%, 68.2%, and 24.7%, respectively, for Estonia, Latvia, and Lithuania.Thus, although the total number of Latvian papers did not increase their quality improved considerably.At the same time Lithuania remarkably improved in the number of publications but their quality improved at a much more modest rate.According to bibliometric criteria the weakest field in Estonian science is Economics & Business, the only one which failed to exceed the threshold of essential science.Relatively modest impact is in Geosciences, Immunology, and Computer Science.

Productivity of research institutions
Table 4 lists 20 most productive research institutions with regard to the number of papers authored by Estonian scientists in the period between 1997 and 2007.Of these top 20 institutions, 11 are not located in Estonia but in some other countries (five in Sweden, four in Finland, and one both in Germany and Russia).The University of Helsinki is the fourth most productive research institution contributing to Estonian science.These 20 institutions produced virtually all Estonian papers because the total score is above 100% (a considerable number of papers have authors from several listed institutions).

Collaboration with other countries
Table 5 provides a list of the top 20 countries/territories contributing to papers authored by Estonian scientists.As expected, the largest number of papers were written in collaboration with colleagues from Sweden, Finland, Germany, USA, and England (almost 50% of all papers).The proportion of papers written in collaboration with Russian scientists has decreased and is now only 4.2%.There is no doubt that the re-orientation towards scientifically stronger partners appears to be one of the success factors of Estonian science.At the same time it may be a weakness as well because of the  2008), there were 58 papers co-authored by Estonian scientists that were within the top 1% of the most cited papers in their particular fields.From these highly cited papers relatively few have exclusively Estonian authorship (Aarik et al., 1999;Niinemets, 1999;Raidal, 2004;Kaljurand et al., 2005).The analysis revealed how political decisions taken by governments of Estonia, Latvia, and Lithuania during the last 17 years of independence have differently shaped the scientific productivity and quality.Latvia failed to increase the productivity of its scientists although the quality of a relatively small number of papers published in international peerreviewed journals has increased considerably during the last five years.In five fields it exceeds the world average level.Although Latvia succeeded in maintaining the high quality of its publications, several political decisions have put Latvian science very close to a critical mass that is necessary to keep up the research activity (Allik, 2003).
Lithuania, in turn, demonstrated a considerable increase in the total number of publications in peerreviewed international journals, including their own journals indexed in the WoS, but without parallel increase in their overall quality.It is not because Lithuania failed to re-orient the network of its scientific collaboration.Among papers published between 1997 and 2007, 8.6%, 8.5%, 7.7%, and 5.7% of all articles were written in collaboration with scientists from Germany, USA, Sweden, and France, respectively.The role of collaboration with Russia (3.6%) was even lower than that in Estonia (4.2%) and Latvia (7.9%).There is also no evidence that Lithuania supports fields with smaller impacts because Physics and Chemistry are the two most productive research areas.It is also stressed that the number of WoS publications has been one of the main criteria for financing decisions in Lithuanian science (Kristapsons et al., 2003).One possible explanation is that the increase of Lithuanian publications in journals indexed by the WoS was achieved mainly by the increase of publications in low-impact journals including their own local ones.In any case, this could serve as a warning for science administrators that the number of WoS publications alone cannot serve as the criterion for decisions, including financing ones.
Estonia succeeded in exceeding the threshold of essential science in all 22 categories except Economics & Business.For a country with a research community of about 1500 actively publishing authors this is a remarkable achievement.However, even Iceland, four times smaller than Estonia, is represented in 20 fields of the 22.In several fields, particularly Agricultural Sciences, Molecular Biology & Genetics, Social Sciences, Geosciences, Environment/Ecology, and Clinical Medicine, the growth of the impact during the last three years has been noteworthy.Some of these fast developing fields are already on a high international level, such as Environment/Ecology and Molecular Biology & Genetics, but some of them, like Geosciences and Social Sciences, are still behind the world average.
The above-presented analysis provides also some evidence that one potential factor behind the relative success of Estonian science could be partnership with scientifically more advanced countries, particularly with Sweden, Finland, Germany, and the United States.A considerable proportion of publications is prepared and published in co-authorship with colleagues from countries that are ahead of Estonia both in terms of the intensity and impact of research.This is a kind of 'hidden money' that is difficult to take into account in the statistics on R&D (Allik, 2003).In the list of the 20 most productive research institutions contributing to Estonia's international publications a half are located outside the border.However, a more detailed analysis is required to reveal how the international network of collaboration stimulates or inhibits Estonian science.

Fig. 1 .
Fig. 1.Total number of publications in the Web of Science database authored by Estonian, Latvian, and Lithuanian scientists.

Table 1 .
Ranking of countries/territories according to the number of citations per paper in the Essential Science Indicators during the 11-year period 1 January 1997-31 December 2007

Table 1 .
Continued Table 2 compares twoESI 11-year periods, 1994ESI 11-year periods,  -2004ESI 11-year periods,    and 1997ESI 11-year periods,  -2007, for Estonian science.In the ESI the whole area (except Humanities) is divided into 22 categories.For the period 1997-2007, Estonia exceeded the threshold for countries/territories (50%) in all areas except Economics & Business.All fields are ranked according to the increase of their impact (citation per paper) from 2004 to 2007 (the last column).

Table 2 .
Table 3 compares strengths and weaknesses in 22 fields of science in Estonia, Latvia, and Lithuania.It is Estonian Essential Science Indicators (ESI) for two periods, 1997-2007 and 1994-2004,and increase in citations per paper(CPP) 2004(CPP)  -2007

Table 3 .
Essential science in Estonia, Latvia, and Lithuania, 1 January 1997-31 December 2007.Percentages above the world average are highlighted in bold These few papers were cited + 82.4% more frequently than papers of this category on average, suggesting that this was due to lucky coincidences.Another small number anomaly is 10 Latvian papers in Psychiatry/Psychology, which were cited + 220% above the world average.most successful subfield in Estonian science is certainly Material Science, which exceeds the world average citation rate by + 43.5%.Traditionally strong is the research in Environment/ Ecology and Plant & Animal Science categories (+ 10.5% and 0%, respectively).In addition, Pharmacology & Toxicology, Chemistry, and Molecular Biology & Genetics are the fields that are very close to the world average level.