Factors Influencing Cities’ Publishing Efficiency

Open access

Abstract

Purpose

Recently, a vast number of scientific publications have been produced in cities in emerging countries. It has long been observed that the publication output of Beijing has exceeded that of any other city in the world, including such leading centres of science as Boston, New York, London, Paris, and Tokyo. Researchers have suggested that, instead of focusing on cities’ total publication output, the quality of the output in terms of the number of highly cited papers should be examined. However, in the period from 2014 to 2016, Beijing produced as many highly cited papers as Boston, London, or New York. In this paper, another method is proposed to measure cities’ publishing performance by focusing on cities’ publishing efficiency (i.e., the ratio of highly cited articles to all articles produced in that city).

Design/methodology/approach

First, 554 cities are ranked based on their publishing efficiency, then some general factors influencing cities’ publishing efficiency are revealed. The general factors examined in this paper are as follows: the linguistic environment of cities, cities’ economic development level, the location of excellent organisations, cities’ international collaboration patterns, and their scientific field profile. Furthermore, the paper examines the fundamental differences between the general factors influencing the publishing efficiency of the top 100 most efficient cities and the bottom 100 least efficient cities.

Findings

Based on the research results, the conclusion can be drawn that a city’s publishing efficiency will be high if meets the following general conditions: it is in a country in the Anglosphere–Core; it is in a high-income country; it is home to top-ranked universities and/or world-renowned research institutions; researchers affiliated with that city most intensely collaborate with researchers affiliated with cities in the United States, Germany, England, France, Canada, Australia, and Italy; and the most productive scientific disciplines of highly cited articles are published in high-impact multidisciplinary journals, disciplines in health sciences (especially general internal medicine and oncology), and disciplines in natural sciences (especially physics, astronomy, and astrophysics).

Research limitations

It is always problematic to demarcate the boundaries of cities (e.g., New York City vs. Greater New York), and regarding this issue there is no consensus among researchers. The Web of Science presents the name of cities in the addresses reported by the authors of publications. In this paper cities correspond to the spatial units between the country/state level and the institution level as indicated in the Web of Science. Furthermore, it is necessary to highlight that the Web of Science is biased towards English-language journals and journals published in the field of biomedicine. These facts may influence the outcome of the research.

Practical implications

Publishing efficiency, as an indicator, shows how successful a city is at the production of science. Naturally, cities have limited opportunities to compete for components of the science establishment (e.g., universities, hospitals). However, cities can compete to attract innovation-oriented companies, high tech firms, and R&D facilities of multinational companies by for example establishing science parks. The positive effect of this process on the city’s performance in science can be observed in the example of Beijing, which publishing efficiency has been increased rapidly.

Originality/value

Previous scientometric studies have examined cities’ publication output in terms of the number of papers, or the number of highly cited papers, which are largely size dependent indicators; however this paper attempts to present a more quality-based approach.

Abstract

Purpose

Recently, a vast number of scientific publications have been produced in cities in emerging countries. It has long been observed that the publication output of Beijing has exceeded that of any other city in the world, including such leading centres of science as Boston, New York, London, Paris, and Tokyo. Researchers have suggested that, instead of focusing on cities’ total publication output, the quality of the output in terms of the number of highly cited papers should be examined. However, in the period from 2014 to 2016, Beijing produced as many highly cited papers as Boston, London, or New York. In this paper, another method is proposed to measure cities’ publishing performance by focusing on cities’ publishing efficiency (i.e., the ratio of highly cited articles to all articles produced in that city).

Design/methodology/approach

First, 554 cities are ranked based on their publishing efficiency, then some general factors influencing cities’ publishing efficiency are revealed. The general factors examined in this paper are as follows: the linguistic environment of cities, cities’ economic development level, the location of excellent organisations, cities’ international collaboration patterns, and their scientific field profile. Furthermore, the paper examines the fundamental differences between the general factors influencing the publishing efficiency of the top 100 most efficient cities and the bottom 100 least efficient cities.

Findings

Based on the research results, the conclusion can be drawn that a city’s publishing efficiency will be high if meets the following general conditions: it is in a country in the Anglosphere–Core; it is in a high-income country; it is home to top-ranked universities and/or world-renowned research institutions; researchers affiliated with that city most intensely collaborate with researchers affiliated with cities in the United States, Germany, England, France, Canada, Australia, and Italy; and the most productive scientific disciplines of highly cited articles are published in high-impact multidisciplinary journals, disciplines in health sciences (especially general internal medicine and oncology), and disciplines in natural sciences (especially physics, astronomy, and astrophysics).

Research limitations

It is always problematic to demarcate the boundaries of cities (e.g., New York City vs. Greater New York), and regarding this issue there is no consensus among researchers. The Web of Science presents the name of cities in the addresses reported by the authors of publications. In this paper cities correspond to the spatial units between the country/state level and the institution level as indicated in the Web of Science. Furthermore, it is necessary to highlight that the Web of Science is biased towards English-language journals and journals published in the field of biomedicine. These facts may influence the outcome of the research.

Practical implications

Publishing efficiency, as an indicator, shows how successful a city is at the production of science. Naturally, cities have limited opportunities to compete for components of the science establishment (e.g., universities, hospitals). However, cities can compete to attract innovation-oriented companies, high tech firms, and R&D facilities of multinational companies by for example establishing science parks. The positive effect of this process on the city’s performance in science can be observed in the example of Beijing, which publishing efficiency has been increased rapidly.

Originality/value

Previous scientometric studies have examined cities’ publication output in terms of the number of papers, or the number of highly cited papers, which are largely size dependent indicators; however this paper attempts to present a more quality-based approach.

1 Introduction

Both the total publication output of China (Andersson et al. 2014; Grossetti et al. 2014; Morrison 2014; Zhou et al. 2009a) and its publication output in specific research areas (Kumar and Garg 2005; Lu and Wolfram 2010; Zou and Laubichler 2017; Zhou et al. 2009b) have significantly increased in the past decades. The growth rate of China’s publication output is quite extreme; however, India (Gupta et al. 2011), Iran (Moin et al. 2005), Brazil (de Almeida and Guimarães 2013; Leta et al. 2006), South Korea (Kim et al. 2012), and Taiwan (Miyairi and Chang 2012) have also recently witnessed significant growth in their total publication output. At the same time, the global share of the publication output of the most developed countries (e.g., the United States, Canada, the Western European countries, Japan, and Australia) has been slowly decreasing. Naturally, the United States still has the highest publication output in the world (Leydesdorff and Wagner 2009; Nature Index 2016), but it can easily be predicted that, due to China’s robust growth in the production of science, the global hegemony of the United States will soon cease.

Some cities in the world have long been considered as an outstanding locus of the production of science (Matthiessen and Schwarz 1999; Van Noorden 2010), and for some decades, an increasing number of cities have been involved in that process (Grossetti et al. 2014; Maisonobe et al. 2017). However, cities’ contribution to the global publication output has been changing over time. Before the rise of Chinese cities, most global output was primarily produced by Northern American cities (e.g., New York, Boston, and Los Angeles), Western European cities (e.g., London, Paris, and Rome), and Japanese cities (e.g., Tokyo, Kyoto, and Osaka). Currently, Beijing is producing the highest publication output in the world (Csomós 2018; Van Noorden 2010). Furthermore, some cities in emerging countries have been positioning themselves as major actors in the production of science. For example, the publication output of Seoul (South Korea), Tehran (Iran), and São Paulo (Brazil) has also increased significantly.

The question is whether the total publication output clearly represents the scientific performance of a city. Can we find another method to measure the scientific performance of a city, a method that is not based on total (or any kind of) output? Does the geographical pattern of the global production of science change if we focus on quality rather than quantity regarding cities’ publication output?

According to Van Noorden (2010), there are some alternatives to express the quality of a city’s scientific performance, for example, measuring the ‘average number of citations that a research paper from a city attracts’ or measuring the total number of Nature and Science articles published by researchers affiliated with that city. Recent studies recommend that, to measure the quality of cities’ publication output, the focus should be on the citation impact of the articles published in those cities. According to Bornmann and Leydesdorff (2011), Bornmann and Waltman (2011), Bornmann et al. (2011), Bornmann and Leydesdorff (2012), and Leydesdorff et al. (2014) as centres of excellence, cities can be assessed by counting the number of excellent papers (i.e., the top 1% most highly cited papers) produced in a city. These studies suggest that, based on the quality of the publication output, cities located in the most developed countries (i.e., the United States, Canada, the Western European countries, Japan, and Australia) are still in top positions.

It is, however, assumed that the higher a city’s total publication output is, the more likely it is that the output of highly cited papers will also be high (e.g., currently Beijing produces the greatest number of highly cited papers in the world). This context suggests that, instead of focusing on cities’ total publication output or the output of highly cited papers, we should focus on cities’ publishing efficiency (i.e., the ratio of highly cited papers to all papers).

Why is it important to measure a city’s publishing efficiency? It can be assumed that the higher the ratio of the number of highly cited papers to all articles produced in a city is, the more likely it is that researchers affiliated with that city conduct research resulting in new scientific breakthroughs (Van Noorden’s study also suggests this nexus). Thus, publishing efficiency shows how successful a city is at the production of science. In 2015, 2.28 percent of the world’s GDP was spent on research and development (R&D) but of course this value varied country to country. In some countries, a higher proportion of the GDP was spent on R&D (e.g., Israel, Japan, and Sweden spent more than three percent of their GDP on R&D), while most countries’ R&D expenditures remain under the world average (e.g., the United Kingdom spent less than two percent of their GDP on R&D). Publishing efficiency is a measure that informs governments on how effectively the R&D expenditures have been used (for example, the mean publishing efficiency of UK cities is almost twice as much as that of Japanese cities, while Japan has a much higher R&D expenditure). Furthermore, because publishing efficiency is measured on the city level, it allows governments to introduce more effective regional development policies.

There are many factors influencing cities’ publishing efficiency, some of which are city specific and some of which are general. Most of the city-specific factors are related to human factors (for example, how prolific a researcher or a team of researchers is), which, due to their nature, vary city to city. However, based on the general factors, typical geographical patterns can be revealed. In this paper, I aim to measure cities’ publishing efficiency worldwide and present the most significant general factors that might influence their publishing efficiency.

The structure of the paper is as follows. In Section 2, I present the data collection process and the methodology. Section 3 is divided into two subsections. In the first subsection, I rank cities based on their publishing efficiency, and in the second subsection, the most significant general factors are presented. Finally, in Section 4, I discuss the results and draw the conclusions.

2 Data and methodology

In the analysis, only cities that had at least 3,000 journal articles published in the period from 2014 to 2016 (i.e., at least 1,000 articles per year) are included. This criterion was met by 554 cities. Data of scientific publications were provided by the Clarivate Analytics’ Web of Science database. Two constraints were implemented to improve the objectivity of the study: 1) Only journal articles were selected for the analysis, and 2) journals should be included in the Science Citation Index Expanded (SCI-EXPANDED), the Social Sciences Citation Index (SSCI), and the Arts & Humanities Citation Index (A&HCI) databases.

The reason for the first constraint is that journal articles are generally considered the most prestigious of scientific publications since they are ‘the basic means of communicating new scientific knowledge’ (Braun et al. 1989: 325). Therefore, I excluded all other types of publications indicated by the Web of Science (e.g., meeting abstracts, book reviews, editorial materials, reviews, proceedings paper, etc.). It should be noted, however, that certain scientific fields conference proceedings (e.g., in computer sciences) and books (e.g., in social sciences and humanities) are also important publishing channels for researchers. However, two-thirds of the documents indexed in the Web of Science are journal articles, therefore the results are only slightly biased towards health sciences and natural sciences.

The reason for the second constraint is that, in 2015, Clarivate Analytics launched a new database in the Web of Science, the Emerging Sources Citation Index (ESCI), which includes journals of regional importance from emerging scientific fields but that are not yet listed in the Journal Citation Report (i.e., they do not have an impact factor).

The publishing efficiency of a given city (x) in the period from 2014 to 2016 (y) is obtained by dividing the number of the highly cited articles by the number of all articles produced by authors affiliated with that city (the value is multiplied by 100 to show a percentage). The formula is as follows:

PublishingEfficiencyx,y=HCAx,yAx,y100,

where HCA is the number of highly cited articles indicated by the Web of Science and A denotes all articles indexed by the Web of Science.

The Web of Science presents the name of cities in the addresses reported by the authors of publications. Naturally, one can suggest that it is difficult, if not impossible, to compare scientometric data of cities with very different sizes and populations. For example, Beijing, the Chinese capital, with almost 22 million inhabitants and an area of 16,000 km2 is obviously not on the same tier as Guilford, Surrey, a midsized English town with nearly 137,000 inhabitants. The total publication output of Beijing, produced in the period from 2014 to 2016, exceeded that of Guilford by 56 times, and the difference was the same in terms of the number of highly cited articles. However, the publishing efficiency of Beijing and that of Guilford is equal (1.317) since publishing efficiency is calculated as the quotient of the number of highly cited articles and the number of all articles. That is, publishing efficiency is a relative value, and the method it is calculated by makes the size of the city in terms of area or population irrelevant. It should be added that the size of a city might influence its publishing efficiency: a small city or town being home to a top-ranked university can produce higher efficiency than a metropolis being the location of many universities and research institutions. Acknowledging this limitation, however, the general factors influencing cities’ publishing efficiencies can be well defined.

It should be noted that the period of data collection spanned from 26/09/2017 to 26/10/2017. The Web of Science has been continuously indexing articles in its database from previous years, especially from 2016. For this reason, in 2017, the number of articles published in 2016 (and before) has been increasing, just like the number of the highly cited articles. A repeated data collection would experience minor differences in the obtained data; however, neither the publishing efficiency nor the rank of cities would considerably change.

3 Results

3.1 Relationship between cities’ publication output and publishing efficiency

In the past decades, the publication output of China has radically increased, and the growth rate has exceeded that of any other countries in the world. Furthermore, the publication outputs of some emerging countries, such as South Korea, Taiwan, India, and Iran, have also significantly increased (Csomós 2018; Grossetti et al. 2014; Maisonobe et al. 2017). Naturally, the annual outputs of the United States, Canada, the Western European countries, and Japan have also increased, but they have witnessed a much smaller growth rate than the emerging countries. Therefore, the share of the most developed countries in the production of science has been decreasing for decades (Leydesdorff and Wagner 2009).

Some cities in the world have long been considered as an outstanding locus of the production of science (Matthiessen and Schwarz 1999; Van Noorden 2010), and the growth rate of these cities’ publication output is much higher than that of the countries in which they are located. By the beginning of the 2010s, the annual publication output of Beijing surpassed that of any other city in the world (Table 1). In the period from 2014 to 2016, it produced a greater number of scientific publications than Japan. Furthermore, Seoul, Tehran, and São Paulo have also experienced a significant increase in their publication output, and due to this development, their position in the ranking has approached that of Tokyo, Paris, New York, and Boston.

Table 1

Top 50 cities producing the greatest number of articles between 2014 and 2016.

RankCountryCityTotal number of articles (2014-2016)
1ChinaBeijing201260
2ChinaShanghai98227
3EnglandLondon92453
4South KoreaSeoul86447
5JapanTokyo77440
6FranceParis75033
7ChinaNanjing70320
8USANew York, NY68577
9USABoston, MA63789
10ChinaGuangzhou51922
11ChinaWuhan50343
12RussiaMoscow47871
13SpainMadrid47061
14IranTehran46173
15ChinaXi’an44052
16SpainBarcelona40393
17BrazilSão Paulo39916
18USACambridge, MA39121
19ChinaHong Kong39032
20ChinaHangzhou39029
21USALos Angeles, CA38740
22CanadaToronto, ON38497
23AustraliaSydney, NSW37676
24USAChicago, IL37560
25SingaporeSingapore37523
26USABaltimore, MD36528
27GermanyBerlin36509
28USAPhiladelphia, PA36117
29ChinaChengdu36032
30USAHouston, TX33869
31USAAtlanta, GA32564
32CanadaMontreal, PQ31820
33ChinaTianjin31764
34EnglandOxford31605
35GermanyMunich30886
36USASeattle, WA30779
37NetherlandsAmsterdam30498
38USAWashington, DC29986
39SwitzerlandZürich29242
40AustraliaMelbourne, VIC29198
41SwedenStockholm28599
42EnglandCambridge27907
43ChinaChangsha27442
44USAAnn Arbor, MI27322
45JapanOsaka26594
46ChinaJinan26557
47ChinaHarbin26386
48DenmarkCopenhagen25538
49ItalyRome25378
50ChinaHefei24911

However, many researchers wonder about the quality of publications produced in Brazilian, Chinese, Indian, Iranian, and even South Korean cities, which is also reflected in their low citation impact (Andersson et al. 2014; Maisonobe et al. 2017; Xie et al. 2014; Van Noorden 2010; Zhou et al. 2009a). In the period from 2014 to 2016, the greatest number of highly cited articles were produced in Beijing (see Table 2), which is not surprising, if we consider the extremely high total publication

Table 2

Top 50 cities producing the greatest number of highly cities articles between 2014 and 2016.

RankCountryCityNumber of highly cited articles (2014-2016)
1ChinaBeijing2650
2USABoston, MA2387
3EnglandLondon2337
4USANew York, NY2237
5USACambridge, MA1827
6FranceParis1601
7ChinaShanghai1208
8USASeattle, WA1191
9USALos Angeles, CA1142
10EnglandOxford1083
11USAStanford, CA1058
12USAPhiladelphia, PA1050
13USABaltimore, MD1047
14CanadaToronto, ON1024
15USAChicago, IL991
16USAAtlanta, GA971
17USAHouston, TX964
18SpainBarcelona920
19USABerkeley, CA911
20USASan Francisco, CA910
21EnglandCambridge885
22SingaporeSingapore871
23ChinaNanjing866
24USABethesda, MD825
25NetherlandsAmsterdam806
26SpainMadrid772
27USAAnn Arbor, MI765
28GermanyMunich764
29JapanTokyo734
30SwitzerlandZürich730
31DenmarkCopenhagen729
32AustraliaSydney, NSW723
33ChinaHong Kong720
34SwedenStockholm708
35South KoreaSeoul698
36GermanyBerlin678
37USAWashington, DC674
38USADurham, NC660
39USANew Haven, CT659
40CanadaMontreal, PQ656
41AustraliaMelbourne, VIC643
42ChinaWuhan637
43GermanyHeidelberg627
44CanadaVancouver, BC615
45USAPittsburgh, PA588
46USAPrinceton, NJ587
47ChinaGuangzhou552
48SwitzerlandGeneva534
49Saudi ArabiaJeddah532
50USASt. Louis, MO511

output of Beijing in terms of the number of articles. Table 2 shows that the difference between the output of Beijing and Boston in terms of the number of highly cited articles is very small, while the total output of Beijing is three times greater than that of Boston (see Table 1). Comparing the rankings in Tables 1 and 2, the positions of some top-ranked cities in terms of total output (e.g., Tokyo and Seoul) have dropped in the ranking of cities producing the greatest number of highly cited articles. In addition, such emerging cities, such as Tehran and São Paulo, which both produced a substantial number of articles between 2014 and 2016, have disappeared from the ranking of the top 50 cities with the greatest number of highly cited articles.

A different geographical pattern will emerge if we focus on measuring cities’ publishing efficiency (Figure 1). The mean publishing efficiency of the 554 cities included in the analysis is 1.818, which means that an average of 1.818% of all articles published in these cities in the period from 2014 to 2016 received enough citations to belong to the top 1% of highly cited articles. However, there are significant geographic differences behind the mean value. Figure 1 shows that the publishing efficiency of most Chinese, Japanese, and South Korean cities (many of which have high publication output in terms of the number of articles) is quite low, while the publishing efficiency of most Northern American and Western European cities is considerably higher. This information is not novel since, directly or indirectly, it has also been described by Van Noorden (2010), Bornmann and Waltman (2011), and Leydesdorff et al. (2014).

Figure 1

Download Figure

Figure 1

Geographic visualisation of cities’ publishing efficiency.

Citation: Journal of Data and Information Science 3, 3; 10.2478/jdis-2018-0014

However, a more fundamental question is whether there are factors influencing cities’ publishing efficiency. Are there any general factors producing high publishing efficiency? Can we find general factors characterising cities having low publishing efficiency? Why is the publishing efficiency of Villejuif (France), Menlo Park, California (United States), or Jeddah (Saudi Arabia) high, and what are the reasons behind the low publishing efficiency of Tehran (Iran), Shenyang (China), and Niigata (Japan)? To answer these questions, we should explore and compare the general factors characterising the most efficient and least efficient cities.

The mean publishing efficiency of the top 100 most efficient cities is 3.179, while that of the bottom 100 least efficient cities is 0.621. In the top 100 cities, in the period from 2014 to 2016, a mean of 13,830 articles per city was produced, of which a mean of 444.71 articles per city received enough citations to belong to the top 1% of highly cited articles. In the same period, in the bottom 100 cities, a mean of 8,885 articles per city was produced, of which only a mean of 60.44 articles per city received enough citations to belong to the top 1% of highly cited articles. That is, the total output in terms of the number of articles of the top 100 most efficient cities is only 1.5 times greater than that of the bottom 100 least efficient cities. In contrast to the results above, there is a difference of more than 7.4 times between the number of highly cited articles produced in the top 100 cities and those produced in the bottom 100 cities. When exploring the general factors influencing cities’ publishing efficiency, I will focus on presenting the differences between the top 100 most efficient cities and the bottom 100 least efficient cities.

The general factors examined in this paper are as follows: the linguistic environment of cities, cities’ economic development level (both derived from country-level data), the location of excellent organisations, cities’ international collaboration patterns, and their scientific field profile. The full list of the top 100 most efficient cities is available in Appendix 1, and the list of the bottom 100 least efficient cities is in Appendix 2.

3.2 Exploring factors influencing cities’ publishing efficiency

Before exploring and evaluating the general factors influencing cities’ publishing efficiency, it is necessary to present the geographical location of cities included in the analysis. The geographical location of a given city does not directly influence its publishing efficiency but allows us to draw indirect conclusions.

Most cities producing high publication output in terms of the number of articles (i.e., at least 3,000 articles in the period from 2014 to 2016) are in three geographical regions in the world: Europe, Asia, and Northern America (Table 3). The aggregate proportion of cities from other regions i.e., Africa, Latin America, and Australia/New Zealand) does not reach 9%. Not just the output but also the mean publishing efficiency of cities differs from each other depending on where they are located. Northern American cities produce the highest publishing efficiency, which is almost one-third greater than that of the European cities ranked second. However, if we divide Europe, the most complex region (there are 29 European countries in the analysis), into sub-regions, we obtain a more realistic picture. The mean publishing efficiencies of the Northern European and the Western European cities are much higher than that of the Southern European and Eastern European cities, and while the publishing efficiencies of the former groups approach the efficiencies of the Northern American cities, those of the Eastern European cities are rather close to the efficiencies of the Latin American cities. In Asia, significant differences emerge as well. The mean publishing efficiencies of cities in Southern Asia and Eastern Asia are under the mean efficiencies of Western Asian cities. Furthermore, cities located in the former two Asian sub-regions produce the lowest mean publishing efficiencies in the world.

Table 3

Number of cities and their mean publishing efficiencies by region and sub-region*.

Regions/Sub-regionsNumber of citiesPercentage in the datasetCities’ mean efficiency publishing
Africa111.991.306
Asia13123.651.009
Eastern Asia8815.880.950
Southern Asia244.330.876
Western Asia142.531.521
Europe23041.521.948
Eastern Europe254.510.989
Northern Europe6010.832.260
Southern Europe549.751.673
Western Europe9116.432.168
Latin America213.790.952
Northern America14526.172.497
Canada183.251.970
USA12722.922.572
Australia/New Zealand162.891.918
World5541.818

*Regions and sub-regions are defined by the United Nations Statistics Division in its geoscheme.

Figure 2 shows the geographical location of the top 100 most efficient cities. Most of the top 100 cities are in two major regions: Northern America (primarily in the United States) and Europe (primarily in Northern Europe and Western Europe). In this group, only three cities are outside the above regions: two of them are in Southern Africa (more precisely in South Africa), and two of them can be found in Western Asia (Saudi Arabia and Israel).

Figure 2

Download Figure

Figure 2

Geographical location of the top 100 most efficient cities.

Citation: Journal of Data and Information Science 3, 3; 10.2478/jdis-2018-0014

The bottom 100 least efficient cities are primarily in three major regions in the world: Asia (primarily in Eastern Asia and Southern Asia), Europe (primarily in Eastern Europe), and Latin America. There is no Northern American city among the least efficient cities, and only two cities from Northern Europe and Western Europe belong to this group. Compared to the number of cities from other regions in the world, the number of African cities (6 out of 100 cities) is insignificant in this group; however, 55% of all African cities in the dataset of 554 cities belong to the bottom 100 least efficient cities.

The geographical location of the top 100 most efficient cities and the bottom 100 least efficient cities is indicative information but allows us to deduce some of the general factors influencing cities’ publishing efficiency. One of the most crucial factors is related to linguistic features, more precisely to the dominance of the English language.

3.2.1 The linguistic environment of cities as a factor influencing cities’ publishing efficiency

It is a generally accepted fact that English has acquired an almost exclusive status as the international language of scientific communication (i.e., the neutral “lingua franca”), leaving little space for other languages in science (Björkman 2011; López-Navarro et al. 2015; Tardy 2004; van Weijen 2012). Although the most important indexing and abstracting databases (i.e., the Web of Science and Scopus) have been including an increasing number of non-English language journals, English language journals are still significantly overrepresented (Li et al. 2014; Mongeon and Paul-Hus 2016). According to Paasi (2005), ‘Anglo-American journals dominate the publishing space in science’, and the international journal publication space is ‘particularly limited to the English-speaking countries’. Furthermore, as Braun and Dióspatonyi (2005), Braun et al. (2007), and Leydesdorff and Wagner (2009) asserted in terms of gatekeepers like editors-in-chief and editorial board member positions, the dominance of the United States is still unchallenged. Considering the above facts, the English language is assumed to be one of the main factors that influence cities’ publishing efficiency.

Figure 3

Download Figure

Figure 3

Geographical location of the bottom 100 least efficient cities.

Citation: Journal of Data and Information Science 3, 3; 10.2478/jdis-2018-0014

In this paper, I classified cities according to the Anglosphere system introduced by Bennet (2007). In this system, the United States, the United Kingdom, Canada, Australia, New Zealand, and Ireland belong to the Anglosphere−Core. Countries in the Anglosphere−Middle sphere (e.g., Nigeria and South Africa) have several official languages, including English (which is the principal language of administration and commerce), but ‘where the primary connections to the outside world are in English’. The Anglosphere−Outer sphere consists of English-using states of other civilisations, including India, Pakistan, Bangladesh, the Arab states formerly under British control (primarily in the Middle East), and the Islamic former colonies of Britain (e.g., Malaysia and African states).

A total of 230 cities (out of 554 cities included in the analysis) are in countries in the Anglosphere, from which 195 cities are in countries in the Anglosphere−Core. Countries outside the Anglosphere are home to 324 cities. The mean publishing efficiency of cities in countries in the Anglosphere is 2.271, while that of the rest of the cities is 1.497. That is, the mean publishing efficiency of cities in countries in the Anglosphere is greater than that of the rest of the cities by 50%. If we focus on the mean publishing efficiency of cities located in countries in the Anglosphere− Core, it increases to 2.439.

As for the top 100 most efficient cities, 73% of them are in countries in the Anglosphere, and 70% of them can be found in countries in the Anglosphere–Core. The mean publishing efficiency of cities belonging to the latter group is 3.235. In contrast, 85% of the bottom 100 least efficient cities are in countries outside the Anglosphere, and 99% of them are in countries outside the Anglosphere−Core. Loughborough (England), having a publishing efficiency of 0.868, is the only city in the group of the bottom 100 cities that can be found in the Anglosphere–Core.

In conclusion, the publishing efficiency of cities located in countries in the Anglosphere (especially in the Anglosphere−Core) is much higher than that of any other cities located in countries outside the Anglosphere. That is, English is not only the international language of scientific communication but also the most fundamental factor influencing cities’ publishing efficiency.

3.2.2 Economic development level of cities as a factor influencing publishing efficiency

Some researchers have observed linear or exponential correlation between scientometric indicators (e.g., the number of publications) and economic development indicators (e.g., GDP per capita or income per capita) (de Solla Price 1978; Kealey 1996; King 2004), while others assert that the correlation between these different sets of indicators is far from clear (Lee at al. 2011; Meo et al. 2013; Vinkler 2008; Vinkler 2010). It is, however, more commonly accepted that the higher the GDP per capita or the income level of a country is, the more likely it is that a greater number of publications will be produced in that country. The question is whether there is a relationship between cities’ publishing efficiency (as a scientometric indicator) and cities’ per capita income level (derived from country-level data).

The classification of countries (and cities) by income level is based upon data obtained from the World Bank Country and Lending Groups database. In this database, countries are classified into four income-level groups: low-income countries (GNI per capita of $1,005 or less in 2016), lower middle-income countries (GNI per capita between $1,006 and $3,955), upper middle-income countries (GNI per capita between $3,956 and $12,235), and high-income countries (GNI per capita of $12,236 or more).

Results show that 434 out of 554 cities included in the analysis are in high-income countries, 93 of them are in upper middle-income countries, and only 27 cities can be found in lower middle-income countries. None of the cities are in low-income countries. That is, most cities producing high publication output in terms of the number of articles (i.e., at least 3,000 articles in the period from 2014 to 2016) are in high-income countries. The mean publishing efficiency of cities from high-income countries is 2.057, that of cities located in upper middle-income countries is 0.997, and the mean publishing efficiency of cities from lower middle-income countries is only 0.881. There is a difference of more than double between the mean publishing efficiency of cities located in high-income countries and that of cities located in upper middle-income countries. The difference between the mean publishing efficiency of cities in upper middle-income countries and that of cities in lower middle-income countries seems to be insignificant.

As for the top 100 most efficient cities, 98% of them are in high-income countries, and only 2% of them can be found in upper middle-income countries. As compared to the quasi-homogeneous group of the top 100 cities, the bottom 100 least efficient cities show a more complex picture; 18% of them are in lower middle-income countries, 46% of them are in upper middle-income countries, but 36% of the bottom 100 least efficient cities are in high-income countries. Based on former studies available in the literature, this latter result might not have been expected; therefore, it requires more explanation.

As was mentioned above, most of the top 100 cities were in Northern America (primarily in the United States) and Europe (primarily in Northern European and Western European countries). Almost all countries in these regions are high-income countries. Contrary to the most efficient cities, none of the least efficient cities are in Northern America. Furthermore, only 17% of the bottom 100 cities are in European countries; except for five cities, all of them are in Eastern European countries (including Russia). Results show that 11 out of the 17 least efficient European cities are in high-income countries, and six of them are in Poland. Figure 4 illustrates that many cities producing low publishing efficiency are in Eastern Asian high-income countries. Half of these cities are in South Korea (11 cities), and another half are in Japan (12 cities); i.e., in countries that belong to the most developed countries in the world in terms of income level. One might suggest that if South Korea and Japan are high-income countries, cities located in South Korea and Japan should produce high publishing efficiency. One reason for this discrepancy may be that both Korean and Japanese languages are considered language isolates (Campbell, 2010), and it is well studied how problematic it is for Japanese people to acquire sufficient communicative skills in English (even if they are researchers) (see, for example, Butler and Iino, 2005, Iwai, 2008). In contrast, for example, one survey shows that the Dutch have the world’s best non-native English skills (see, EF EPI English Proficiency Index, https://www.ef.co.hu/epi/) Therefore, it is not surprising that the collaboration intensity of both Japan and South Korea with the United States is lower (the proportion of co-authored papers indexed in the WoS during 2014–2016 was 10.6 and 13.8 percent for these countries, respectively) than that of European high-income countries, such as Germany and the Netherlands (the proportion of co-authored papers indexed in the WoS during 2014–2016 was 15.9 and 18.0 percent for these countries, respectively). It has however been determined that the higher the intensity of the collaboration between a country and the United States is, the more likely it is that co-authored papers will receive a higher number of citations (Pan et al., 2012; Sud and Thelwall, 2016), and have a greater chance to become highly cited papers.

Figure 4

Download Figure

Figure 4

Geographical location of the bottom 100 least efficient cities in terms of countries’ income levels.

Citation: Journal of Data and Information Science 3, 3; 10.2478/jdis-2018-0014

Loughborough (England) is the only city in the bottom 100 least efficient cities that is in a high-income country belonging to the Anglosphere−Core. Beer-Sheva (Israel), a city in the group of the bottom 100 least efficient cities, is also in a high-income country, but is in the Anglosphere−Outer sphere. In fact, many of the bottom 100 cities are in countries in the Anglosphere−Outer sphere, but all of them are in lower middle-income countries, primarily in Southeast Asia (11 cities are in India, and one is in Pakistan).

East Asia is home to 46% of the bottom 100 least efficient cities. Beside Japan and South Korea, most of these cities are in China. While none of the East Asian countries included in the analysis belong to the group of the low-income or lower middle-income countries (Japan and South Korea are high-income countries, and China is an upper middle-income country), the publishing efficiency of the East Asian cities is rather low. Kawaguchi (Japan), the city producing the highest publishing efficiency in the region, is ranked only 138th. The facts above suggest that the economic development level of the cities is a key factor influencing publishing efficiency, which is reinforced by the fact that almost all cities in the group of the top 100 cities are in high-income countries, but it is not the most important factor.

The examination of factors like the dominance of the English language and cities’ economic development level will bring us closer to understanding why cities’ publishing efficiency differs from each other; however, we need deeper insight to obtain a precise picture of publishing efficiency. For example, country-level data allows us to understand why the publishing efficiency of Canadian and Chinese cities significantly differ from each other but does not help us to understand why the publishing efficiency of Kawaguchi is higher or why that of Niigata is lower than the mean publishing efficiency of Japanese cities. To examine cities’ publishing efficiency in a more precise way, we need to focus on some general as well as more city-specific factors, like the location of excellent organisations, cities’ international collaboration patterns, and the productivity of specific research areas.

For example, in Kawaguchi, most publications were produced by the Japan Science and Technology Agency, one of Japan’s excellent scientific organisations; therefore, the publishing efficiency of Kawaguchi is considerably higher than that of other Japanese cities. That is, which cities in the world are home to excellent organisations (e.g., universities and governmental and international research institutions) should be examined. The question is whether these organisations are exclusively located in cities producing high publishing efficiency or whether some of them might be found in cities with low publishing efficiency.

3.2.3 Location of excellent organisations as a factor influencing cities’ publishing efficiency

In the paper by Van Noorden (2010: 907) an important question arose: What is the reason Boston ranks top in several analyses of scientific quality? A brief answer was given by José Lobo, a statistician and economist who was affiliated with Arizona State University at Tempe: ‘Take three or four of the best universities in the world, put them in a city with a seaport, and voilà!’ Naturally, the question requires a more complex answer (as was later also explained by Van Noorden), but it calls attention to a key factor: the scientific performance of cities significantly depends on whether they are home to top-ranked universities.

Although many research institutions, hospitals, governmental organisations (e.g., ministries and departments), NGOs, and companies have a significant publication output (see, for example, Archambault and Larivière 2011; Csomós and Tóth 2016; Hicks 1995), scientific publications are primarily produced by universities all over the world. In recent years, university rankings have gained in popularity. The main goal of ranking and comparing universities in terms of scientific output (of which the publication output is a vital component) is to make the most prestigious universities visible worldwide. There are several different world university rankings available (e.g., CWTS Leiden Ranking, The Times Higher Education World University Rankings, QS World University Rankings, and Academic Ranking of World Universities – ARWU), which are all based upon different input data. However, each ranking attributes more or less significance to bibliometric indicators, such as the number of publications produced in a given university, the quality (citation impact) of scientific publications, or the number of articles published in top journals (see, for example, Docampo et al. 2015; Frenken et al. 2017; Piro and Sivertsen 2016; Shehatta and Mahmood 2016). Naturally, the methodologies of how university rankings are produced differ from each other; thus, university rankings are different in terms of top university rankings (Abramo and D’Angelo 2015; Lin et al. 2013).

From the point of view of this analysis, university rankings contain indicative information only. I chose to use the Academic Ranking of World Universities (ARWU) published annually by the Shanghai Ranking Consultancy because the importance of the Shanghai ranking has become recognized by both governments and universities; further, according to Docampo and Cram (2014), the ‘ranking has become a major resource for exploring the characteristics and quality of academic institutions and university systems worldwide.’ I examined whether there is a relationship between the location of top-ranked universities and cities’ publishing efficiency. Top-ranked universities correspond to universities having been ranked among the top 100 universities on one of the ARWU lists of 2014, 2015, and 2016.

In the period from 2014 to 2016, the top 100 universities were in 95 cities, some of which were home to more than one top-ranked university (e.g., New York, London, Boston, Pittsburgh, Munich, Stockholm, and Zurich). The publishing efficiency of cities that were home to the top 100 universities averages 2.641, while that of the rest of the cities averages 1.648. That is, the mean publishing efficiency of cities that are home to the top 100 universities is higher than that of the rest of the cities by 60%. These results suggest that the location of top-ranked universities significantly influences cities’ publishing efficiency. In other words, it seems to be a logical assumption that top-ranked universities are primarily located in the most efficient cities. Thus, we should examine which of the top 100 most efficient cities are home to top-ranked universities.

Figure 5 shows that it is not an exclusive privilege of the most efficient cities to be home to top-ranked universities. Only 43% of the top 100 universities are in the top 100 most efficient cities. Furthermore, there are many cities worldwide (including Chinese and Japanese cities), that do not belong to the top 100 most efficient cities; yet, they are home to top-ranked universities. In the group of the bottom 100 cities, Moscow (Russia) is the only city that is home to a top-ranked university.

Figure 5

Download Figure

Figure 5

Geographical location of cities that are home to the top 100 universities as ranked by ARWU.

Citation: Journal of Data and Information Science 3, 3; 10.2478/jdis-2018-0014

The location of top-ranked universities is considered an important but not decisive factor influencing cities’ publishing efficiency. Examining the ranking of the most efficient cities, there are two cities (Villejuif, France and Menlo Park, California, USA) topping the ranking that are not home to top-ranked universities as ranked by the ARWU.

The question arises as to what kind of organisations (but not universities) are in cities like Villejuif, Menlo Park, California, Upton, New York (United States), Greenbelt, Maryland (United States), Didcot (England), etc., which produce very high publishing efficiency. The explanations are as follows.

Villejuif, the city with the highest publishing efficiency in the world, is home to the ‘Institut Gustave Roussy’, one of the world’s leading cancer-research institutions and the premier oncology centre and teaching hospital in Europe. Although Villejuif is a city (commune) having 50 thousand inhabitants, it is a suburb of Paris, about seven kilometres from its centre.

Menlo Park is home to the SLAC National Accelerator Laboratory, a linear accelerator that is owned by the US Department of Energy and operated by the Stanford University. Currently, SLAC is the world’s largest linear accelerator and is one of top research centres for accelerator physics. The city of Menlo Park, with a population of 32 thousand, is in the San Francisco Bay Area between San Francisco and San Jose (i.e., in one of the fastest growing regions in the world that is home to many innovative companies and top-ranked universities). Additionally, Didcot has 25 thousand inhabitants and is 16 km south of Oxford. Didcot is home to the Rutherford Appleton Laboratory, a world-renowned research centre for particle physics and space science.

Cities such as Villejuif, Menlo Park, and Didcot can be characterised the same way; they are smaller cities, towns, or villages located in metropolitan areas and are home to quasi-independent research institutions (e.g., national laboratories) generally operating under the umbrella of prestigious universities. Naturally, top research institutions are in large cities as well, but being surrounded by universities, their visibility in terms of publication output is much lower, even if they produce very high publishing efficiency. For example, the total publication output of Geneva (Switzerland) is produced by many organisations, including the European Organisation for Nuclear Research (CERN), the World Health Organization (WHO), and the University of Geneva. In the period from 2014 to 2016, almost 60% of Geneva’s total publication output came from the University of Geneva, which has been ranked among the top 100 universities on the ARWU list, and which publishing efficiency is as high as 3.33. However, if we compare the publishing efficiency of the University of Geneva to that of the CERN (5.37) and the WHO (6.86), it seems rather low. The same pattern appears in large cities like New York, London, Paris, Los Angeles, and Tokyo.

In conclusion, a positive relationship can be detected between the location of top-ranked universities and cities’ high publishing efficiency. However, it should be noted that publications, primarily in large cities, come from different types of organisations, many of which have lower publishing efficiency than universities. Thus, some cities that are home to top-ranked universities have not been included in the top 100 most efficient cities. Furthermore, there are several top-ranked cities that are not home to top-ranked universities (or any universities); yet, they produce a very high publishing efficiency.

3.2.4 International collaboration pattern as a factor influencing cities’ publishing efficiency

In recent years, the number of publications produced by single authors has been decreasing, while the number of co-authored publications and number of co-authors in publications have been increasing rapidly (Abramo et al. 2017; Castelvecchi 2015; Uddin et al. 2012). Therefore, cities’ international collaboration patterns have become more complex (i.e., authors affiliated with a given city have been collaborating with a growing number of co-authors affiliated with other cities in other countries). Naturally, cities’ international collaboration patterns are influenced by many factors, including differences between the productivity of scientific disciplines (Larivière et al. 2006; Paul-Hus et al. 2017; Zhou et al. 2009b), the size of the national research system (Van Raan 1998), and linguistic features (Csomós 2018; Maisonobe et al. 2016). These facts might suggest that international collaboration patterns vary city to city worldwide, making it impossible to predict cities’ publishing efficiency. However, this question remains to be answered.

In this section, I aim to examine whether cities with high publishing efficiency and cities with low publishing efficiency are characterised by specific international collaboration patterns. Data obtained from the Web of Science database allows us to reveal countries with which the co-authors are affiliated. For example, in the period from 2014 to 2016, 27,322 articles were produced in Ann Arbor, Michigan (United States), from which 765 received enough citations to belong to the top 1% highly cited articles. If we focus on the international collaboration pattern of all articles produced in Ann Arbor between 2014 and 2016, 8.76% of the articles were written with co-authors affiliated with China, 7.23% had co-authors affiliated with Canada, 7.13% had co-authors affiliated with England, 6.78% had co-authors affiliated with Germany, 5.16% had co-authors affiliated with France, and so on. That is, in the case of all articles, the top collaborator with Ann Arbor is China, and the second ranked collaborator is Canada, and so on.

However, if we focus on the international collaboration pattern of the highly cited articles, a different pattern will emerge. Most highly cited articles were written with co-authors affiliated with England (27.32%), with 25.49% from Canada, 23.53% from Germany, 20.26% from France, 17.39% from Italy, and so on. That is, in the case of highly cited articles, the top collaborator of Ann Arbor is England (replacing China as the top collaborator in all articles), and the second ranked collaborator is Canada, and so on.

I examine which countries are the top collaborators (i.e., collaborators ranked 1–5) in the case of all articles and in the case of highly cited articles produced in a given city in the period from 2014 to 2016. Furthermore, I compare the typical international collaboration patterns of the top 100 most efficient cities to that of the bottom 100 least efficient cities. My aim is to reveal whether there is a relationship between cities’ international collaboration patterns and cities’ publishing efficiencies and whether there is a difference between the typical collaboration patterns of the top 100 cities and the bottom 100 cities. When examining cities’ international collaboration patterns, I implemented a geographical constraint. The group of the top 100 cities was divided into two sub-groups (i.e., the most efficient non-US cities and the most efficient US cities), and they were examined separately.

Table 4 shows the countries occupying the top 1–5 positions as collaborators in all articles and their frequency of occurrence in those positions. The top collaborator of the most efficient non-US cities (48 out of the top 100 cities) is the United States, whose frequency of occurrence in the top 1–5 positions is 100% (in the top position in 81.25% of the cases). This means that the United States has a very intense collaboration with every single city belonging to the group of the most efficient non-US cities. Germany is ranked second by collaborating with 87.50% of the most efficient non-US cities in one of the top 1–5 positions. As compared to that of the United States, the frequency of occurrence of Germany in the top position is only 8.33%. In the case of all articles, the top 1–5 collaborators of the most efficient non-US cities are the United States, Germany, England, France, and Italy. As top collaborators, other countries (like the Netherlands, Australia, Spain, etc.) are rather marginal, primarily appearing in the top 4–5 positions.

Table 4

Top collaborators* in the case of all articles.

Top collaborators of the most efficient non-US cities occurring in the 1–5 positionsFrequency of occurrence in the 1–5 positions in percentageTop collaborators of the most efficient US cities occurring in the 1–5 positionsFrequency of occurrence in the 1–5 position in percentageTop collaborators of the least efficient cities occurring in the 1–5 positionsFrequency of occurrence in the 1–5 positions in percentage
1USA100.00Germany98.11USA98.00
2Germany87.50England98.11Germany78.00
3England75.00China94.34England69.00
4France75.00Canada84.91France39.00
5Italy52.08France66.04China31.00
6Netherlands27.08Australia15.09Australia30.00
7Australia18.75Italy15.09Japan25.00
8Spain18.75Japan5.66Canada23.00
9China16.67Netherlands5.66Italy23.00
10Scotland8.33South Korea5.66South Korea18.00

* In this context, collaborators correspond to countries with which co-authors are affiliated.

In the case of all articles produced in the most efficient US cities (52 out of the top 100 cities), the most frequently occurring countries as collaborators in the top 1–5 positions are Germany, England, China, Canada, and France (Table 4). China, the top collaborator of the most efficient US cities, has surpassed England by almost 2%. The United States has had a traditionally close scientific relationship with Western European countries (especially the United Kingdom) and Canada (Adams 2013), but on the city level, China has recently been occupying a more significant position (Csomós 2018; Tian 2016). Naturally, the top international collaborator of most Chinese cities has been the United States for a long time (He 2009; Wang et al. 2013; Zhang and Guo 1997). If we merge the groups of the most efficient non-US cities and the most efficient US cities into a single group, it turns out that all the co-authors are affiliated with 21 countries occupying one of the top 1–5 positions.

Table 4 illustrates that the international collaboration patterns of the bottom 100 least efficient cities resemble a mixture of the international collaboration patterns of the most efficient non-US cities and US cities. The United States (in the top position in 85% of the cases), Germany, England, France, and China appear in the top 1–5 positions in most cases. However, two facts should be highlighted: 1) As for the international collaboration patterns of the bottom 100 cities, the frequency of occurrence of countries following the United States is much lower than in the case of the most efficient non-US cities. The mean frequency of occurrence of the top 1–5 collaborator countries in articles produced in the most efficient non-US cities is 77.92%. This value is 88.30% in articles produced in the most efficient US cities, but it reaches only 63% in the bottom 100 least efficient cities. 2) The least efficient cities collaborate with a greater number of countries (33) occupying one of the top 1–5 positions than the most efficient cities (21). Many of these countries (e.g., Saudi Arabia, Brazil, Iran, Russia, and South Korea) produce low publishing efficiency; thus, the collaboration has a negative effect on cities’ publishing efficiency (i.e., these collaborations result in a smaller number of articles that receive enough citations to belong to the top 1% highly cited articles).

It is, however, more important to know which countries (more precisely the co-authors affiliated with that country) are the top collaborators of cities (more precisely the authors affiliated with that city) in highly cited articles. According to my hypothesis, countries as the top 1-5 collaborators of cities in highly cited articles differ from those occupying top positions in the total number of articles. The publishing efficiency of cities is heavily influenced by where the top collaborators are in the case of highly cited articles.

Table 5 shows that the collaboration pattern of the most efficient non-US cities in highly cited articles is almost the same as the collaboration pattern that emerged in the total number of articles; however, the relative weight of Germany, France, and England has increased. In the total number of articles, the mean frequency of occurrence of the top 1–5 collaborators was 77.92, while in highly cited articles, this value has increased to 79.17. In highly cited articles produced in the most efficient US cities, the frequency of occurrence of England is 100%, which means that England occupies one of the top 1–5 positions of every single city (in the top position in 57.69% of the cases). Germany has the same frequency of occurrence in highly cited articles than in the total number of articles, but the frequency of occurrence of Canada and especially that of France has significantly increased. China, the third most frequently occurring country in the total number of articles, has vanished from the group of the top collaborators in highly cited articles. This means that, although the total number of articles in US cities shows intense collaboration with China, the collaboration results in only a small number of highly cited articles. In highly cited articles, the mean frequency of occurrence of the most efficient US cities with the top 1–5 collaborators is 81.92%, which is a bit less than in the total number of articles.

Table 5

Top collaborators* in the case of the highly cited articles.

Top collaborators of the most efficient non-US cities occurring in the 1–5 positionsFrequency of occurrence in the 1–5 positions in percentageTop collaborators of the most efficient US cities occurring in the 1–5 positionsFrequency of occurrence in the 1–5 position in percentageTop collaborators of the least efficient cities occurring in the 1–5 positionsFrequency of occurrence in the 1–5 positions in percentage
1USA100.00England100.00USA98.00
2Germany89.58Germany98.08Germany72.00
3France79.17France88.46England70.00
4England77.08Canada86.54France43.00
5Italy50.00Australia36.54Australia34.00
6Netherlands22.92Italy34.62China29.00
7Spain18.75China21.15Italy29.00
8Switzerland16.67Spain9.62Spain26.00
9Australia14.58Netherlands9.62Canada25.00
10Canada12.50Switzerland7.69Japan13.00

* In this context, collaborators correspond to countries with which co-authors are affiliated.

Not surprisingly, in highly cited articles, the bottom 100 least efficient cities have a very intense collaboration with the United States. In 98 cities, the United States occupies one of the top 1–5 positions and is in the top position in 79% of the cases. The frequency of occurrence of countries following the United States is much lower than in the most efficient cities. The mean frequency of occurrence of the top 1–5 collaborators in highly cited papers produced in the least efficient cities is only 63.4%. In the top 1–5 positions, the bottom 100 least efficient cities collaborate with a total of 30 countries, while this value in the top 100 most efficient (non-US and US) cities is 16.

In the case of the highly cited articles, there are fundamental differences between the international collaboration patterns of the most efficient cities and the least efficient cities. Although both groups of cities have roughly the same top collaborators, the least efficient cities collaborate with a much greater number of countries than the most efficient cities. It seems that this difference significantly influences the publishing efficiency of cities.

In conclusion, if co-authors are primarily from countries of the United States, Germany, England, France, Canada, and Italy, which are leading countries in science, articles will have a greater chance to receive enough citations to belong to the top 1% highly cited articles.

3.2.5 The scientific field profiles of cities as a factor influencing publishing efficiency

Beside the factors detailed above, cities’ publishing efficiency is significantly influenced by the productivity of scientific disciplines. The most productive disciplines vary city to city, and the productivity of different disciplines in terms of highly cited articles differs as well (Bornmann et al. 2011). In each city, the most productive disciplines will be revealed both in the case of all articles and in the case of highly cited articles.

For example, in the period from 2014 to 2016, authors from Ann Arbor, Michigan produced articles in 151 disciplines: 8.16% of the 27,322 articles were published in the discipline of physics, 7.41% in engineering, 6.43% in ‘science, technology, and other topics’ (as it is indicated in the WoS), 4.99% in chemistry, 4.91% in psychology, and so on. The greatest number of highly cited articles was produced in quite different disciplines; 15.11% of the 765 highly cited articles were written in ‘science, technology, and other topics’, 11.27% in general internal medicine, 9.35% in physics, 9.22% in oncology, 5.89% in astronomy and astrophysics, and so on.

To obtain a better understanding of why the publishing efficiency of the most efficient cities and that of the least efficient cities differ significantly, we need to reveal the characteristics of the most productive discipline in those cities. Table 6 shows that, in the case of the top 100 cities, the most productive discipline occurring in the top 1–5 positions is ‘science, technology, and other topics’. In the Web of Science, articles published in multidisciplinary journals (e.g., Nature, Science, Proceedings of the National Academy of Sciences of the United States of America, and PlosONE) are classified into the discipline of ‘science, technology, and other topics’. It is well-known that articles published in high-impact multidisciplinary journals become highly cited at a very great proportion. For example, 45.67% of all articles published between 2014 and 2016 in Nature and 40.44% of all articles published in the same period in Science have received enough citations to belong to the top 1% highly cited articles.

Table 6

Most productive scientific disciplines in all articles.

The most productive scientific disciplines occurring in the top 1–5 positions in the most efficient citiesFrequency of occurrence in the 1–5 positions in percentageThe most productive scientific disciplines occurring in the top 1–5 positions in the least efficient citiesFrequency of occurrence in the 1–5 positions in percentage
1Science, Technology, and Other Topics84.00Chemistry99.00
2Physics63.00Engineering85.00
3Neurosciences and Neurology47.00Physics84.00
4Chemistry45.00Materials Science80.00
5Engineering41.00Science, Technology, and Other54.00
Topics
6Astronomy and Astrophysics38.00Mathematics15.00
7Oncology29.00Environmental Sciences and12.00
Ecology
8Environmental Sciences and Ecology21.00Oncology12.00
9Psychology20.00Pharmacology and Pharmacy11.00
10Materials Science16.00Agriculture7.00

In general, articles published in the top 100 most efficient cities can be classified into two major scientific fields: natural sciences (e.g., physics, chemistry, and engineering) and health sciences (e.g., neurosciences and neurology, oncology, and psychology). Contrary to the top 100 cities, most articles produced in the bottom 100 least efficient cities can be classified into disciplines that are natural sciences, while the field of health sciences is almost absent. In the case of the least efficient cities, oncology is the most frequently occurring health science discipline with a frequency of occurrence of only 12% (i.e., it occurs in the top 1–5 positions in only 12% of the least efficient cities). In contrast to health sciences, natural sciences (e.g., chemistry, engineering, physics, and material science) produce a very high frequency of occurrence (Table 6). Chemistry is in the top 1–5 positions in almost every bottom 100 city, and it occupies the top position in 54% of the cases. This means that, in more than half of the least efficient cities, chemistry is the most productive research area.

In the case of the highly cited articles published in the top 100 most efficient cities, the discipline of ‘science, technology, and other topics’ is even more dominant; it is in the top 1–5 positions in 91% of all cities but occurs in the top position in only 20% of the cases. In 35% of the top 100 cities, general internal medicine occupies the top position but ranked second based on the aggregate frequency of occurrence (Table 7). In highly cited articles produced in the most efficient cities, both the number and frequency of occurrence of health disciplines are greater than in all articles. When examining all articles produced in the most efficient cities, general internal medicine is in the top 1–5 positions in only 5% of cases, but in the highly cited articles, this value increases to 69%. Furthermore, the frequency of occurrence of oncology and the discipline of the cardiovascular system and cardiology increased by more than 50%.

Table 7

Most productive scientific disciplines in highly cited articles.

The most productive scientific disciplines occurring in the top 1–5 positions in the most efficient citiesFrequency of occurrence in the 1–5 positions in percentageThe most productive scientific disciplines occurring in the top 1–5 positions in the least efficient citiesFrequency of occurrence in the 1–5 positions in percentage
1Science, Technology, and Other Topics91Science, Technology, and Other74
Topics
2Physics69Chemistry66
3General Internal Medicine69Physics56
4Astronomy and Astrophysics54Engineering54
5Oncology42General Internal Medicine37
6Chemistry28Materials Science37
7Cardiovascular System and Cardiology21Astronomy and Astrophysics25
8Biochemistry and Molecular Biology18Environmental Sciences and Ecology20
9Environmental Sciences and Ecology15Oncology20
10Neurosciences and Neurology14Mathematics17

In highly cited articles produced in the bottom 100 least efficient cities, most of the dominant disciplines are in natural sciences. In the least efficient cities, the discipline of ‘science, technology, and other topics’ occupies the top position, but its frequency of occurrence is less than in the most efficient cities.

When we examined the international collaboration patterns in both the cases of all articles and in highly cited articles produced in the top 100 cities and produced in the bottom 100 cities, respectively, we found that they differ in the frequency of occurrence of the top collaborators. However, the countries with which they collaborate (i.e., the location of co-authors) were primarily the same. As for the scientific disciplines, there are significant differences between the top 100 cities and the bottom 100 cities in not only the frequency of occurrence of the most productive disciplines but also in the disciplines themselves. In the most efficient cities, highly cited articles are produced in disciplines that are in natural sciences and health sciences to almost the same degree, while, in the least efficient cities, health disciplines are rather marginal. Furthermore, the frequency of occurrence of the discipline of ‘science, technology, and other topics’ is much higher in articles produced in the most efficient cities than in articles produced in the least efficient cities. This fact suggests that, in the most efficient cities, a greater number of articles are published in high-impact multidisciplinary journals than in the least efficient cities.

4 Discussion and conclusion

In this paper, I examined whether there were general factors influencing cities’ publishing efficiency (i.e., the ratio of highly cited articles to all articles produced in that city). I have found the following five fundamental factors: the dominance of the English language, cities’ economic development level, the location of excellent organisations, cities’ international collaboration patterns, and the productivity of scientific disciplines.

The dominance of the English language seems to be one of the most (if not the most) significant factors influencing cities’ publishing efficiency. About three-quarters of the most efficient cities are in countries in the Anglosphere–Core, and the rest of are in Northern and Western European countries. Contrary to the most efficient cites, 99% of the least efficient cities are in countries outside the Anglosphere–Core.

The economic development level of cities (derived from country-level data) as a factor influencing the publishing efficiency seems less significant than the linguistic environment. Results show that 98% of the most efficient cities are in high-income countries. It might suggest that there is a relationship between cities’ high-income level and cities’ high publishing efficiency, but it turned out that one-third of the least efficient cities were also located in high-income countries. The reason for this is that countries that are home to cities with low efficiency but high-income level do not belong to the Anglosphere, reinforcing the fact that the dominance of the English language (i.e., the linguistic environment of cities) as a factor has a greater significance in influencing cities’ publishing efficiency than the cities’ economic development level has.

It is well-known fact that scientific publications are primarily produced by universities. We can assume that the most efficient cities should be home to the most prestigious universities in the world, while top-ranked universities are not expected to be in the least efficient cities. Results show that this hypothesis is basically correct, at least when we focus on the location of top-ranked universities in the least efficient cities. However, the picture is more complex in the case of the most efficient cities, because half of those cities are not home to top-ranked universities. Moreover, many top-ranked universities are in cities that are not the most efficient cities. The reason for this is that there are many towns and small or mid–sized cities that are home to world-renowned national or international research institutions producing even higher publishing efficiency than top-ranked universities. These settlements are all characterised by the fact that they are within metropolitan areas, while the research institutions they host operate under the umbrella of prestigious research universities.

In the case of the highly cited articles, an overlap can be detected between the international collaboration patterns of the most efficient cities and the least efficient cities. In both cases, the top collaborators are the United States (primarily in the top position), Germany, England, France, Canada, and Australia/Italy. If we merely focus on who the top collaborators of cities are, we cannot predict whether its publishing efficiency will be high. However, the magnitude of the collaboration intensity between cities (more precisely the authors affiliated with those cities) and the leading countries in science (more precisely the co-authors located in those countries) even more significantly influences cities’ publishing efficiency. The higher the collaboration intensity is, the more likely it is that cities will produce high publishing efficiency.

In the most efficient cities, highly cited articles are produced in disciplines of natural sciences and health sciences to the same degree. In the least efficient cities, almost all highly cited articles are produced in the field of natural sciences (primarily in chemistry), while hardly any articles are published in health sciences. In the case of both groups of cities, ‘science, technology, and other topics’ is the most frequently occurring discipline in highly cited articles; however, its frequency of occurrence in articles produced in the most efficient cities is much higher than in the least efficient cities.

Based on the above research results, we can draw the conclusion that a city’s publishing efficiency will be high if meets the following conditions:

  1. It is in a country in the Anglosphere–Core;

  2. It is in a high-income country;

  3. It is home to top-ranked universities and/or world-renowned research institutions;

  4. Researchers affiliated with that city most intensely collaborate with researchers affiliated with cities in the United States, Germany, England, France, Canada, and Australia/Italy; and

  5. The most productive scientific disciplines of highly cited articles are ‘science, technology, and other topics’ (i.e., most articles are published in high-impact multidisciplinary journals), disciplines in health sciences (especially general internal medicine and oncology), and disciplines in natural sciences (especially physics, astronomy, and astrophysics).

Approximately 60% of the top 100 most efficient cities meet the above criteria, but if we expand the geographical dimension beyond the Anglosphere, 86% of the top 100 cities will meet the criteria.

Most of the bottom 100 least efficient cities are in countries outside the Anglosphere. If we do not consider the determinant significance of the linguistic factor, the patterns of the Japanese, South Korean, and European cities resemble the patterns of the most efficient cities. All of them are in high-income countries and have more or less similar international collaboration patterns as that of the most efficient cities. Moreover, most of the highly cited articles are produced in similar disciplines (although disciplines in natural sciences are overrepresented). Naturally, there are several top-ranked and prestigious universities and research institutions in Japanese and South Korean cities (especially in Tokyo, Kyoto, Nagoya, Osaka, and Seoul); yet, they produce low publishing efficiency.

The question is: What can the city administration do to increase the city’s performance in science (e.g., to increase the city’s publishing efficiency)? Naturally, cities have limited opportunities to compete for components of the science establishment. Universities, hospitals and most governmental research institutions are generally tied to their original loci. However, cities can compete to attract innovation-oriented companies, high tech firms, and R&D facilities of multinational companies by for example establishing science parks. The positive effect of this process on the city’s performance in science can be observed in the example of Beijing (Andersson et al. 2014; Liefner et al. 2006; Zhou 2005). Furthermore, cities can compete to acquire cutting-edge international research facilities. For example, in 2009, founding member states of the European Spallation Source (ESS) (the most powerful linear proton accelerator in the world) decided to support for placing ESS in Lund, selecting it from the competition of three European cities. The ESS will attract thousands of researchers from all over the world to Lund.

Some of the further research directions based upon the results of the study are as follows: What kind of local factors influence cities’ publishing efficiency? If publishing efficiency is an indicator of cities’ performance in science, what can city administrations do to improve it? If cities have very different sizes and populations (even publication output) worldwide, what kind of territorial demarcation can be introduced to balance these differences?

Acknowledgement

The publication is supported by the EFOP-3.6.1-16-2016-00022 project. The project is co-financed by the European Union and the European Social Fund.

References

  • Abramo, G., & D’Angelo, C. A. (2015). Evaluating university research: Same performance indicator, different rankings. Journal of Informetrics, 9(3), 514−525.

  • Abramo, G., D’Angelo, A. C., & Murgia, G. (2017). The relationship among research productivity, research collaboration, and their determinants. Journal of Informetrics, 11(4), 1016−1030.

  • Adams, J. (2013). Collaborations: The fourth age of research. Nature, 497(7451), 557−560.

  • Andersson, D. E., Gunessee, S., Matthiessen, C. W., & Find, S. (2014). The geography of Chinese science. Environment and Planning A, 46(12), 2950−2971.

  • Archambault, É., & Larivière, V. (2011). Scientific publications and patenting by companies: A study of the whole population of Canadian firms over 25 years. Science and Public Policy, 38(4), 269–278.

  • Bennett, J. C. (2007). The Anglosphere Challenge: Why the English-Speaking Nations Will Lead the Way in the Twenty-First Century. Lanham, Maryland, USA: Rowman & Littlefield Publishers.

  • Björkman, B. (2011). Pragmatic strategies in English as an academic lingua franca: Ways of achieving communicative effectiveness? Journal of Pragmatics, 43(4), 950−964.

  • Bornmann, L., & Leydesdorff, L. (2011). Which cities produce more excellent papers than can be expected? A new mapping approach, using Google Maps, based on statistical significance testing. Journal of the American Society for Information Science and Technology, 62(10), 1954−1962.

  • Bornmann, L., & Leydesdorff, L. (2012). Which are the best performing regions in information science in terms of highly cited papers? Some improvements of our previous mapping approaches. Journal of Informetrics, 6(2), 336−345.

  • Bornmann, L., Leydesdorff, L., Walch-Solimena, C., & Ettl, C. (2011). Mapping excellence in the geography of science: An approach based on Scopus data. Journal of Informetrics, 5(4), 537−546.

  • Bornmann, L., & Waltman, L. (2011). The detection of “hot regions” in the geography of science-A visualization approach by using density maps. Journal of Informetrics, 5(4), 547−553.

  • Braun, T., Glänzel, W., & Schubert, A. (1989). Some data on the distribution of journal publication types in the science citation index database. Scientometrics, 15(5–6), 325−330.

  • Braun, T., & Dióspatonyi, I. (2005). World flash on basic research: The counting of core journal gatekeepers as science indicators really counts. The scientific scope of action and strength of nations. Scientometrics, 62(3), 297−319.

  • Braun, T., Zsindely, S., Dióspatonyi, I., & Zádor, E. (2007). Gatekeeping patterns in nano-titled journals. Scientometrics, 70(3), 651−667.

  • Butler, Y.G., & Iino, M. (2005). Current Japanese reforms in English language education: The 2003 “action plan”. Language Policy, 4(1), 25−45.

  • Campbell, L. (2010). Language Isolates and Their History, or, What’s Weird, Anyway? In Proceedings of the 36th Annual Meeting of the Berkeley Linguistics Society (pp. 16−31). Berkeley: Berkeley Linguistics Society.

  • Castelvecchi, D. (2015). Physics paper sets record with more than 5,000 authors. Nature News, 15/05/2015.

  • Csomós, G., & Tóth, G. (2016). Exploring the position of cities in global corporate research and development: A bibliometric analysis by two different geographical approaches. Journal of Informetrics, 10(2), 516−532.

  • Csomós, G. (2018). A spatial scientometric analysis of the publication output of cities worldwide. Journal of Informetrics, 12(2), 547−566.

  • de Almeida, E. C. E., & Guimarães, J. A. (2013). Brazil’s growing production of scientific articles-how are we doing with review articles and other qualitative indicators? Scientometrics, 97(2), 287−315.

  • de Solla Price, D. (1978). Toward a model for science indicators. In Y. Elkana, J. Lederberg, R. K. Merton, A. Thackray, H. Zuckerman (Eds.), Toward a Metric of Science: The Advent of Science Indicators (pp. 69–95). New York: John Wiley & Sons.

  • Docampo, D., & Cram, L. (2014). On the internal dynamics of the Shanghai ranking. Scientometrics, 98(2), 1347−1366.

  • Docampo, D., Egret, D., & Cram, L. (2015). The effect of university mergers on the Shanghai ranking. Scientometrics, 104(1), 175−191.

  • Frenken, K., Heimeriks, G. J., & Hoekman, J. (2017). What drives university research performance? An analysis using the CWTS Leiden Ranking data. Journal of Informetrics, 11(3), 859−872.

  • Grossetti, M., Eckert, D., Gingras, Y., Jégou, L., Larivière, V., & Milard, B. (2014). Cities and the geographical deconcentration of scientific activity: A multilevel analysis of publications (1987–2007). Urban Studies, 51(10), 2219−2234.

  • Gupta, B. M., Kshitij, A., & Verma, C. (2011). Mapping of Indian computer science research output, 1999-2008. Scientometrics, 86(2), 261-283.

  • He, T. (2009). International scientific collaboration of China with the G7 countries. Scientometrics, 80(3), 571−582.

  • Hicks, D. (1995). Published papers, tacit competencies and corporate management of the public/private character of knowledge. Industrial and Corporate Change, 4(2), 401–424.

  • Iwai, Y. (2008). The perceptions of Japanese students toward academic English reading: Implications for effective ESL reading strategies. Multicultural Education, 15(4), 45–50.

  • Kato, M., & Ando, A. (2017). National ties of international scientific collaboration and researcher mobility found in Nature and Science. Scientometrics, 110(2), 673−694.

  • Kealey, T. (1996). The Economic Laws of Scientific Research. New York: St. Martin’s Press.

  • Kim, H., Yoon, J. W., & Crowcroft, J. (2012). Network analysis of temporal trends in scholarly research productivity. Journal of Informetrics, 6(1), 97−110.

  • King, D. A. (2004). The scientific impact of nations. What different countries get for their research spending. Nature, 430, 311–316.

  • Kumar, S., & Garg, K. C. (2005). Scientometrics of computer science research in India and China. Scientometrics, 64(2), 121-132.

  • Larivière, V., Gingras, Y., & Archambault, É. (2006). Canadian collaboration networks: A comparative analysis of the natural sciences, social sciences and the humanities. Scientometrics, 68(3), 519−533.

  • Lee, L. C., Lin, P. H., Chuang, Y. W., & Lee, Y. Y. (2011). Research output and economic productivity: A Granger causality test. Scientometrics, 89(2), 465−478.

  • Leta, J., Glänzel, W., & Thijs, B. (2006). Science in Brazil. Part 2: Sectoral and institutional research profiles. Scientometrics, 67(1), 87-105.

  • Leydesdorff, L., & Wagner, C. (2009). Is the United States losing ground in science? A global perspective on the world science system. Scientometrics, 78(1), 23-36.

  • Leydesdorff, L., Wagner, C. S., & Bornmann, L. (2014). The European Union, China, and the United States in the top-1% and top-10% layers of most-frequently cited publications: Competition and collaborations. Journal of Informetrics, 8(3), 606−617.

  • Li, J., Qiao, L., Li, W., & Jin, Y. (2014). Chinese-language articles are not biased in citations: Evidences from Chinese-English bilingual journals in Scopus and Web of Science. Journal of Informetrics, 8(4), 912–916.

  • Liefner, I., Hennemann, S., & Lu, X. (2006). Cooperation in the innovation process in developing countries: Empirical evidence from Zhongguancun, Beijing. Environment and Planning A, 38(1), 111–130.

  • Lin, C. S., Huang, M. H., & Chen, D. Z. (2013). The influences of counting methods on university rankings based on paper count and citation count. Journal of Informetrics, 7(3), 611−621.

  • López-navarro, I., Moreno, A. I., Quintanilla, M. Á., & Rey-Rocha, J. (2015). Why do I publish research articles in English instead of my own language? Differences in Spanish researchers’ motivations across scientific domains. Scientometrics, 103(3), 939-976.

  • Lu, K., & Wolfram, D. (2010). Geographic characteristics of the growth of informetrics literature 1987-2008. Journal of Informetrics, 4(4), 591−601.

  • Maisonobe, M., Eckert, D., Grossetti, M., Jégou, L., & Milard, B. (2016). The world network of scientific collaborations between cities: domestic or international dynamics? Journal of Informetrics, 10(4), 1025−1036.

  • Maisonobe, M., Grossetti, M., Milard, B., Jégou, L., & Eckert, D. (2017). The global geography of scientific visibility: a deconcentration process (1999–2011). Scientometrics, 113(1), 479−493.

  • Matthiessen, C.W., & Schwarz, A.W. (1999). Scientific centres in Europe: An analysis of research strength and patterns of specialisation based on bibliometric indicators. Urban Studies, 36(3), 453−477.

  • Meo, S. A., Al Masri, A. A., Usmani, A. M., Memon, A. N., & Zaidi, S. Z. (2013). Impact of GDP, Spending on R&D, Number of Universities and Scientific Journals on Research Publications among Asian Countries. PLoS ONE, 8(6), e66449

  • Miyairi, N., & Chang, H. W. (2012). Bibliometric characteristics of highly cited papers from Taiwan, 2000-2009. Scientometrics, 92(1), 197−205.

  • Moin, M., Mahmoudi, M., & Rezaei, N. (2005). Scientific output of Iran at the threshold of the 21st century. Scientometrics, 62(2), 239−248.

  • Mongeon, P., & Paul-Hus, A. (2016). The journal coverage of Web of Science and Scopus: a comparative analysis. Scientometrics, 106(1), 213−228.

  • Morrison, J. (2014). China becomes world’s third-largest producer of research articles. Nature News, 06/02/2014.

  • Nature Index (2016). US tops global research performance. Nature Index, 20/04/2016. https://www.natureindex.com/news-blog/us-tops-global-research-performance

  • Paasi, A. (2005). Globalisation, academic capitalism, and the uneven geographies of international journal publishing spaces. Environment and Planning A, 37(5), 769–789.

  • Pan, K. R., Kaski, K., & Fortunato, S. (2012). World citation and collaboration networks: uncovering the role of geography in science. Scientific Reports, 2, 902

  • Paul-Hus, A., Mongeon, P., Sainte-marie, M., & Larivière, V. (2017). The sum of it all: Revealing collaboration patterns by combining authorship and acknowledgements. Journal of Informetrics, 11(1), 80−87.

  • Piro, F. N., & Sivertsen, G. (2016). How can differences in international university rankings be explained? Scientometrics, 109(3), 2263−2278.

  • Shehatta, I., & Mahmood, K. (2016). Correlation among top 100 universities in the major six global rankings: policy implications. Scientometrics, 109(2), 1231−1254.

  • Sud, P., & Thelwall, M. (2016). Not all international collaboration is beneficial: The Mendeley readership and citation impact of biochemical research collaboration. Journal of the Association for Information Science and Technology, 67(8), 1849−1857.

  • Tardy, C. (2004). The role of English in scientific communication: Lingua franca or Tyrannosaurus rex? Journal of English for Academic Purposes, 3(3), 247−269.

  • Tian, P. (2016). China’s diaspora key to science collaborations. Nature Index, 23/06/2016. https://www.natureindex.com/news-blog/chinas-diaspora-key-to-science-collaborations

  • Uddin, S., Hossain, L., Abbasi, A., & Rasmussen, K. (2012). Trend and efficiency analysis of co-authorship network. Scientometrics, 90(2), 687–699.

  • Van Noorden, R. (2010). Cities: Building the best cities for science. Nature, 467(7318), 906−908.

  • Van Raan, A. F. J. (1998). The influence of international collaboration on the impact of research results: Some simple mathematical considerations concerning the role of self-citations. Scientometrics, 42(3), 423-428.

  • Van Weijen, D. (2012). The Language of (Future) Scientific Communication. Research Trends, 31, 11/2012. https://www.researchtrends.com/issue-31-november-2012/the-language-of-future-scientific-communication/

  • Vinkler, P. (2008). Correlation between the structure of scientific research, scientometric indicators and GDP in EU and non-EU countries. Scientometrics, 74(2), 237−254.

  • Vinkler, P. (2010). The Evaluation of Research by Scientometric Indicators. Oxford: Chandos Publishing.

  • Wang, X., Xu, S., Wang, Z., Peng, L., & Wang, C. (2013). International scientific collaboration of China: Collaborating countries, institutions and individuals. Scientometrics, 95(3), 885−894.

  • Xie, Y., Zhang, C., & Lai, Q. (2014). China’s rise as a major contributor to science and technology. Proceedings of the National Academy of Sciences of the United States of America, 111(26), 9437−9442.

  • Zhang, H., & Guo, H. (1997). Scientific research collaboration in China. Scientometrics, 38(2), 309−319.

  • Zhou, P., Thijs, B., & Glänzel, W. (2009a). Regional analysis on Chinese scientific output. Scientometrics, 81(3), 839−857.

  • Zhou, P., Thijs, B., & Glänzel, W. (2009b). Is China also becoming a giant in social sciences? Scientometrics, 79(3), 593−621.

  • Zhou, Y. (2005). The making of an innovative region from a centrally planned economy: Institutional evolution in Zhongguancun Science Park in Beijing. Environment and Planning A, 37(6), 1113−1134.

  • Zou, Y., & Laubichler, M.D. (2017). Measuring the contributions of Chinese scholars to the research field of systems biology from 2005 to 2013. Scientometrics, 110(3), 1615−1631.

Appendix 1. The top 100 most efficient cities

RankCountryCityEfficiency
1FranceVillejuif6.174
2USAMenlo Park, CA5.676
3USAPrinceton, NJ4.978
4USACambridge, MA4.670
5USAStanford, CA4.658
6Saudi ArabiaJeddah4.541
7USASanta Cruz, CA4.430
8USAPasadena, CA4.400
9USASan Francisco, CA3.993
10USABerkeley, CA3.932
11USAUpton, NY3.920
12USABethesda, MD3.912
13USASeattle, WA3.870
14USARochester, MN3.830
15USASanta Barbara, CA3.778
16USABoston, MA3.742
17USAGreenbelt, MD3.679
18USARockville, MD3.667
19USARichland, WA3.618
20SwitzerlandGeneva3.566
21USANew Haven, CT3.565
22UKOxford3.427
23USADurham, NC3.400
24USAEvanston, IL3.388
25UKDidcot3.366
26USABoulder, CO3.288
27USADallas, TX3.272
28USANew York, NY3.262
29ItalyPerugia3.219
30USARiverside, CA3.201
31GermanyHeidelberg3.177
32UKCambridge3.171
33UKBrighton3.162
34USANashville, TN3.122
35FranceCréteil3.111
36IsraelRehovot3.096
37USAPortland, OR3.091
38USAPalo Alto, CA3.080
39SwitzerlandBasel3.050
40ItalyTrieste3.036
41USASt. Louis, MO3.029
42NetherlandsRotterdam3.028
43CanadaVancouver, BC3.024
44UKNorwich3.006
45USAAurora, CO3.004
46USAAtlanta, GA2.982
47UKLancaster2.976
48NetherlandsNijmegen2.963
49USASan Antonio, TX2.961
50FranceGif-sur-Yvette2.955
51USALos Angeles, CA2.948
52USAChapel Hill, NC2.937
53CanadaVictoria, BC2.924
54UKDundee2.916
55USAPhiladelphia, PA2.907
56UKLeicester2.906
57UKEdinburgh2.898
58USAResearch Triangle Park, NC2.897
59South AfricaCape Town2.896
60NetherlandsWageningen2.886
61GermanyGarching bei München2.877
62USABaltimore, MD2.866
63SwitzerlandLausanne2.866
64DenmarkCopenhagen2.855
65USARochester, NY2.852
66USAHouston, TX2.846
67EstoniaTartu2.842
68USAProvidence, RI2.840
69USADenver, CO2.837
70USABirmingham, AL2.826
71South AfricaDurban2.818
72FranceClermont-Ferrand2.802
73USAAnn Arbor, MI2.800
74ItalyFerrara2.797
75USACleveland, OH2.788
76CanadaHamilton, ON2.768
77UKSouthampton2.758
78UKCardiff2.738
79UKExeter2.738
80USASan Diego, CA2.734
81USAHanover, NH2.715
82GermanyMainz2.714
83USAGaithersburg, MD2.691
84USAWorcester, MA2.687
85SwitzerlandVilligen2.686
86UKBirmingham2.685
87DenmarkLyngby2.684
88GermanyBonn2.678
89CanadaToronto, ON2.660
90UKNewcastle2.658
91SwitzerlandBern2.657
92USAAmherst, MA2.652
93USAEugene, OR2.650
94NetherlandsAmsterdam2.643
95USAChicago, IL2.638
96GermanyEssen2.627
97BelgiumBrussels2.614
98ItalyPavia2.611
99USAWinston-Salem, NC2.594
100USATallahassee, FL2.591

Appendix 2.The bottom 100 least efficient cities
RankCountryCityEfficiency
1TunisiaSfax0.132
2RussiaYekaterinburg0.161
3South KoreaCheonan0.260
4IranShiraz0.268
5RomaniaIași0.273
6IndiaKharagpur0.283
7ChinaMianyang0.325
8PolandLublin0.333
9BrazilSão Carlos0.333
10ChinaWenzhou0.348
11IndiaVaranasi0.399
12ChinaShijiazhuang0.416
13South KoreaCheongju0.424
14JapanGifu0.436
15IranTabriz0.444
16ChileConcepción0.452
17BrazilCuritiba0.454
18JapanKumamoto0.456
19MalaysiaSerdang0.462
20TunisiaTunis0.484
21EgyptGiza0.487
22ChinaNantong0.494
23IsraelBeer-Sheva0.501
24JapanIbaraki0.503
25IndiaKanpur0.513
26ChinaBaoding0.516
27TurkeyKonya0.535
28South KoreaBusan0.537
29IranTehran0.550
30ChinaShenyang0.551
31EgyptAlexandria0.552
32JapanNiigata0.556
33FranceVilleneuve-d’Ascq0.562
34SpainAlicante0.563
35South KoreaGwangju0.563
36South KoreaJeonju0.566
37BrazilFortaleza0.567
38PolandPoznań0.568
39BrazilViçosa0.576
40TurkeyIzmir0.587
41IndiaLucknow0.587
42PortugalAveiro0.587
43ChinaZhengzhou0.588
44ChinaGuilin0.594
45ChinaYantai0.595
46South KoreaDaejeon0.596
47BrazilBelo Horizonte0.601
48IndiaKolkata0.606
49ChinaÜrümqi0.613
50IndiaChennai0.617
51JapanShizuoka0.626
52ChinaNanning0.632
53IndiaHyderabad0.632
54JapanSaitama0.634
55JapanKawasaki0.649
56BrazilRecife0.654
57ItalyMessina0.661
58EgyptCairo0.670
59TurkeyIstanbul0.673
60ChinaChangzhou0.682
61South KoreaYongin0.682
62ChinaKunming0.689
63PakistanLahore0.690
64JapanSapporo0.695
65ArgentinaCórdoba0.720
66JapanKanazawa0.732
67PolandGdańsk0.736
68PolandWrocław0.736
69ChinaQingdao0.737
70UkraineKiev0.749
71ChinaJinan0.749
72ChinaXinxiang0.754
73IndiaNew Delhi0.755
74PolandŁódź0.756
75ChinaNingbo0.758
76IndiaBangalore0.783
77South KoreaJinju0.783
78TurkeyAnkara0.791
79JapanChiba0.796
80JapanSagamihara0.798
81South AfricaPretoria0.801
82RussiaNovosibirsk0.804
83South KoreaGoyang0.804
84South KoreaDaegu0.806
85South KoreaSeoul0.807
86ChinaNanchang0.809
87ChinaTaiyuan0.810
88ChinaGuiyang0.813
89IndiaRoorkee0.829
90RussiaMoscow0.836
91ChinaWuxi0.840
92BrazilPorto Alegre0.844
93BrazilFlorianópolis0.855
94RussiaSaint Petersburg0.856
95IndiaMumbai0.865
96JapanSendai0.865
97UKLoughborough0.868
98ChinaXuzhou0.869
99BrazilCampinas0.884
100PolandKraków0.891

Footnotes

The definition of highly cited papers in the Web of Science is as follows: highly cited papers received enough citations as of July/August 2017 to place them in the top 1% of their academic fields based on a highly cited threshold for the field and publication year.

In the analysis, I focused on the document type ‘articles’ only; therefore, I narrowed the category of highly cited ‘papers’ to the category of highly cited ‘articles’.

Northern Europe includes the countries of the United Kingdom as defined by the United Nations Statistics Division in its geoscheme.

World Bank Country and Lending Groups https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups

It should be noted that ARWU is just one of the alternatives to rank universities. Naturally, other organisations produce different rankings with different universities in top positions. For example, out of the top 10 universities, only Harvard University and Stanford University appear in both the CWTS Leiden Ranking of 2017 and the ARWU list of 2017. Contrary to this example, the groups of the top 10 universities in the QS World University Rankings of 2017 and the ARWU list differ from each other by only three universities. In addition, there are many top-ranked universities that are not included in the group of top 100 universities on the ARWU list but are in cities with high publishing efficiency. For example, Rotterdam, the forty-second most efficient city in the world, is home to the Erasmus University Rotterdam, which ranked 101-150 (i.e., outside but close to the top 100 universities).

The most productive scientific discipline corresponds to the one to which the largest number of articles belong in the Web of Science.

Abramo, G., & D’Angelo, C. A. (2015). Evaluating university research: Same performance indicator, different rankings. Journal of Informetrics, 9(3), 514−525.

Abramo, G., D’Angelo, A. C., & Murgia, G. (2017). The relationship among research productivity, research collaboration, and their determinants. Journal of Informetrics, 11(4), 1016−1030.

Adams, J. (2013). Collaborations: The fourth age of research. Nature, 497(7451), 557−560.

Andersson, D. E., Gunessee, S., Matthiessen, C. W., & Find, S. (2014). The geography of Chinese science. Environment and Planning A, 46(12), 2950−2971.

Archambault, É., & Larivière, V. (2011). Scientific publications and patenting by companies: A study of the whole population of Canadian firms over 25 years. Science and Public Policy, 38(4), 269–278.

Bennett, J. C. (2007). The Anglosphere Challenge: Why the English-Speaking Nations Will Lead the Way in the Twenty-First Century. Lanham, Maryland, USA: Rowman & Littlefield Publishers.

Björkman, B. (2011). Pragmatic strategies in English as an academic lingua franca: Ways of achieving communicative effectiveness? Journal of Pragmatics, 43(4), 950−964.

Bornmann, L., & Leydesdorff, L. (2011). Which cities produce more excellent papers than can be expected? A new mapping approach, using Google Maps, based on statistical significance testing. Journal of the American Society for Information Science and Technology, 62(10), 1954−1962.

Bornmann, L., & Leydesdorff, L. (2012). Which are the best performing regions in information science in terms of highly cited papers? Some improvements of our previous mapping approaches. Journal of Informetrics, 6(2), 336−345.

Bornmann, L., Leydesdorff, L., Walch-Solimena, C., & Ettl, C. (2011). Mapping excellence in the geography of science: An approach based on Scopus data. Journal of Informetrics, 5(4), 537−546.

Bornmann, L., & Waltman, L. (2011). The detection of “hot regions” in the geography of science-A visualization approach by using density maps. Journal of Informetrics, 5(4), 547−553.

Braun, T., Glänzel, W., & Schubert, A. (1989). Some data on the distribution of journal publication types in the science citation index database. Scientometrics, 15(5–6), 325−330.

Braun, T., & Dióspatonyi, I. (2005). World flash on basic research: The counting of core journal gatekeepers as science indicators really counts. The scientific scope of action and strength of nations. Scientometrics, 62(3), 297−319.

Braun, T., Zsindely, S., Dióspatonyi, I., & Zádor, E. (2007). Gatekeeping patterns in nano-titled journals. Scientometrics, 70(3), 651−667.

Butler, Y.G., & Iino, M. (2005). Current Japanese reforms in English language education: The 2003 “action plan”. Language Policy, 4(1), 25−45.

Campbell, L. (2010). Language Isolates and Their History, or, What’s Weird, Anyway? In Proceedings of the 36th Annual Meeting of the Berkeley Linguistics Society (pp. 16−31). Berkeley: Berkeley Linguistics Society.

Castelvecchi, D. (2015). Physics paper sets record with more than 5,000 authors. Nature News, 15/05/2015.

Csomós, G., & Tóth, G. (2016). Exploring the position of cities in global corporate research and development: A bibliometric analysis by two different geographical approaches. Journal of Informetrics, 10(2), 516−532.

Csomós, G. (2018). A spatial scientometric analysis of the publication output of cities worldwide. Journal of Informetrics, 12(2), 547−566.

de Almeida, E. C. E., & Guimarães, J. A. (2013). Brazil’s growing production of scientific articles-how are we doing with review articles and other qualitative indicators? Scientometrics, 97(2), 287−315.

de Solla Price, D. (1978). Toward a model for science indicators. In Y. Elkana, J. Lederberg, R. K. Merton, A. Thackray, H. Zuckerman (Eds.), Toward a Metric of Science: The Advent of Science Indicators (pp. 69–95). New York: John Wiley & Sons.

Docampo, D., & Cram, L. (2014). On the internal dynamics of the Shanghai ranking. Scientometrics, 98(2), 1347−1366.

Docampo, D., Egret, D., & Cram, L. (2015). The effect of university mergers on the Shanghai ranking. Scientometrics, 104(1), 175−191.

Frenken, K., Heimeriks, G. J., & Hoekman, J. (2017). What drives university research performance? An analysis using the CWTS Leiden Ranking data. Journal of Informetrics, 11(3), 859−872.

Grossetti, M., Eckert, D., Gingras, Y., Jégou, L., Larivière, V., & Milard, B. (2014). Cities and the geographical deconcentration of scientific activity: A multilevel analysis of publications (1987–2007). Urban Studies, 51(10), 2219−2234.

Gupta, B. M., Kshitij, A., & Verma, C. (2011). Mapping of Indian computer science research output, 1999-2008. Scientometrics, 86(2), 261-283.

He, T. (2009). International scientific collaboration of China with the G7 countries. Scientometrics, 80(3), 571−582.

Hicks, D. (1995). Published papers, tacit competencies and corporate management of the public/private character of knowledge. Industrial and Corporate Change, 4(2), 401–424.

Iwai, Y. (2008). The perceptions of Japanese students toward academic English reading: Implications for effective ESL reading strategies. Multicultural Education, 15(4), 45–50.

Kato, M., & Ando, A. (2017). National ties of international scientific collaboration and researcher mobility found in Nature and Science. Scientometrics, 110(2), 673−694.

Kealey, T. (1996). The Economic Laws of Scientific Research. New York: St. Martin’s Press.

Kim, H., Yoon, J. W., & Crowcroft, J. (2012). Network analysis of temporal trends in scholarly research productivity. Journal of Informetrics, 6(1), 97−110.

King, D. A. (2004). The scientific impact of nations. What different countries get for their research spending. Nature, 430, 311–316.

Kumar, S., & Garg, K. C. (2005). Scientometrics of computer science research in India and China. Scientometrics, 64(2), 121-132.

Larivière, V., Gingras, Y., & Archambault, É. (2006). Canadian collaboration networks: A comparative analysis of the natural sciences, social sciences and the humanities. Scientometrics, 68(3), 519−533.

Lee, L. C., Lin, P. H., Chuang, Y. W., & Lee, Y. Y. (2011). Research output and economic productivity: A Granger causality test. Scientometrics, 89(2), 465−478.

Leta, J., Glänzel, W., & Thijs, B. (2006). Science in Brazil. Part 2: Sectoral and institutional research profiles. Scientometrics, 67(1), 87-105.

Leydesdorff, L., & Wagner, C. (2009). Is the United States losing ground in science? A global perspective on the world science system. Scientometrics, 78(1), 23-36.

Leydesdorff, L., Wagner, C. S., & Bornmann, L. (2014). The European Union, China, and the United States in the top-1% and top-10% layers of most-frequently cited publications: Competition and collaborations. Journal of Informetrics, 8(3), 606−617.

Li, J., Qiao, L., Li, W., & Jin, Y. (2014). Chinese-language articles are not biased in citations: Evidences from Chinese-English bilingual journals in Scopus and Web of Science. Journal of Informetrics, 8(4), 912–916.

Liefner, I., Hennemann, S., & Lu, X. (2006). Cooperation in the innovation process in developing countries: Empirical evidence from Zhongguancun, Beijing. Environment and Planning A, 38(1), 111–130.

Lin, C. S., Huang, M. H., & Chen, D. Z. (2013). The influences of counting methods on university rankings based on paper count and citation count. Journal of Informetrics, 7(3), 611−621.

López-navarro, I., Moreno, A. I., Quintanilla, M. Á., & Rey-Rocha, J. (2015). Why do I publish research articles in English instead of my own language? Differences in Spanish researchers’ motivations across scientific domains. Scientometrics, 103(3), 939-976.

Lu, K., & Wolfram, D. (2010). Geographic characteristics of the growth of informetrics literature 1987-2008. Journal of Informetrics, 4(4), 591−601.

Maisonobe, M., Eckert, D., Grossetti, M., Jégou, L., & Milard, B. (2016). The world network of scientific collaborations between cities: domestic or international dynamics? Journal of Informetrics, 10(4), 1025−1036.

Maisonobe, M., Grossetti, M., Milard, B., Jégou, L., & Eckert, D. (2017). The global geography of scientific visibility: a deconcentration process (1999–2011). Scientometrics, 113(1), 479−493.

Matthiessen, C.W., & Schwarz, A.W. (1999). Scientific centres in Europe: An analysis of research strength and patterns of specialisation based on bibliometric indicators. Urban Studies, 36(3), 453−477.

Meo, S. A., Al Masri, A. A., Usmani, A. M., Memon, A. N., & Zaidi, S. Z. (2013). Impact of GDP, Spending on R&D, Number of Universities and Scientific Journals on Research Publications among Asian Countries. PLoS ONE, 8(6), e66449

Miyairi, N., & Chang, H. W. (2012). Bibliometric characteristics of highly cited papers from Taiwan, 2000-2009. Scientometrics, 92(1), 197−205.

Moin, M., Mahmoudi, M., & Rezaei, N. (2005). Scientific output of Iran at the threshold of the 21st century. Scientometrics, 62(2), 239−248.

Mongeon, P., & Paul-Hus, A. (2016). The journal coverage of Web of Science and Scopus: a comparative analysis. Scientometrics, 106(1), 213−228.

Morrison, J. (2014). China becomes world’s third-largest producer of research articles. Nature News, 06/02/2014.

Nature Index (2016). US tops global research performance. Nature Index, 20/04/2016. https://www.natureindex.com/news-blog/us-tops-global-research-performance

Paasi, A. (2005). Globalisation, academic capitalism, and the uneven geographies of international journal publishing spaces. Environment and Planning A, 37(5), 769–789.

Pan, K. R., Kaski, K., & Fortunato, S. (2012). World citation and collaboration networks: uncovering the role of geography in science. Scientific Reports, 2, 902

Paul-Hus, A., Mongeon, P., Sainte-marie, M., & Larivière, V. (2017). The sum of it all: Revealing collaboration patterns by combining authorship and acknowledgements. Journal of Informetrics, 11(1), 80−87.

Piro, F. N., & Sivertsen, G. (2016). How can differences in international university rankings be explained? Scientometrics, 109(3), 2263−2278.

Shehatta, I., & Mahmood, K. (2016). Correlation among top 100 universities in the major six global rankings: policy implications. Scientometrics, 109(2), 1231−1254.

Sud, P., & Thelwall, M. (2016). Not all international collaboration is beneficial: The Mendeley readership and citation impact of biochemical research collaboration. Journal of the Association for Information Science and Technology, 67(8), 1849−1857.

Tardy, C. (2004). The role of English in scientific communication: Lingua franca or Tyrannosaurus rex? Journal of English for Academic Purposes, 3(3), 247−269.

Tian, P. (2016). China’s diaspora key to science collaborations. Nature Index, 23/06/2016. https://www.natureindex.com/news-blog/chinas-diaspora-key-to-science-collaborations

Uddin, S., Hossain, L., Abbasi, A., & Rasmussen, K. (2012). Trend and efficiency analysis of co-authorship network. Scientometrics, 90(2), 687–699.

Van Noorden, R. (2010). Cities: Building the best cities for science. Nature, 467(7318), 906−908.

Van Raan, A. F. J. (1998). The influence of international collaboration on the impact of research results: Some simple mathematical considerations concerning the role of self-citations. Scientometrics, 42(3), 423-428.

Van Weijen, D. (2012). The Language of (Future) Scientific Communication. Research Trends, 31, 11/2012. https://www.researchtrends.com/issue-31-november-2012/the-language-of-future-scientific-communication/

Vinkler, P. (2008). Correlation between the structure of scientific research, scientometric indicators and GDP in EU and non-EU countries. Scientometrics, 74(2), 237−254.

Vinkler, P. (2010). The Evaluation of Research by Scientometric Indicators. Oxford: Chandos Publishing.

Wang, X., Xu, S., Wang, Z., Peng, L., & Wang, C. (2013). International scientific collaboration of China: Collaborating countries, institutions and individuals. Scientometrics, 95(3), 885−894.

Xie, Y., Zhang, C., & Lai, Q. (2014). China’s rise as a major contributor to science and technology. Proceedings of the National Academy of Sciences of the United States of America, 111(26), 9437−9442.

Zhang, H., & Guo, H. (1997). Scientific research collaboration in China. Scientometrics, 38(2), 309−319.

Zhou, P., Thijs, B., & Glänzel, W. (2009a). Regional analysis on Chinese scientific output. Scientometrics, 81(3), 839−857.

Zhou, P., Thijs, B., & Glänzel, W. (2009b). Is China also becoming a giant in social sciences? Scientometrics, 79(3), 593−621.

Zhou, Y. (2005). The making of an innovative region from a centrally planned economy: Institutional evolution in Zhongguancun Science Park in Beijing. Environment and Planning A, 37(6), 1113−1134.

Zou, Y., & Laubichler, M.D. (2017). Measuring the contributions of Chinese scholars to the research field of systems biology from 2005 to 2013. Scientometrics, 110(3), 1615−1631.

Journal Information

Figures

  • View in gallery

    Geographic visualisation of cities’ publishing efficiency.

  • View in gallery

    Geographical location of the top 100 most efficient cities.

  • View in gallery

    Geographical location of the bottom 100 least efficient cities.

  • View in gallery

    Geographical location of the bottom 100 least efficient cities in terms of countries’ income levels.

  • View in gallery

    Geographical location of cities that are home to the top 100 universities as ranked by ARWU.

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 39 39 25
PDF Downloads 7 7 4