See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323955584

Chinese scholars in China and overseas: Comparative analysis on research

productivity and impact

Article  in  Current science · March 2018

DOI: 10.18520/cs/v115/i1/49-55

CITATIONS READS

0 271

3 authors, including:

Wen Lou Siluo Yang

East China Normal University Wuhan University
19 PUBLICATIONS   45 CITATIONS    48 PUBLICATIONS   201 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Evaluating synthetically the International impact of Chinese research works in Humanities and Social Sciences View project

Evaluating synthetically the International impact of Chinese research works in Humanities and Social Sciences View project

All content following this page was uploaded by Siluo Yang on 23 March 2018.

The user has requested enhancement of the downloaded file.

 Chinese scholars in China and overseas: comparative
analysis on research productivity and impact
Wen LOU

Department of Information Management, Faculty of Economics and Management, Institution

of Academic Evaluation and Development, East China Normal University. Room 323,
Building of law and business, East China Normal University, 500 Dongchuan rd, Minhang,
Shanghai, China.

8
Email: [email protected]

01
/2
/3
Hui WANG

12
Department of Information Management, Faculty of Economics and Management, East China

on
Normal University. Building of law and business, East China Normal University, 500
Dongchuan rd, Minhang, Shanghai, China.
d
he
Email: [email protected]
is
bl
pu

Siluo YANG*
n

School of information management, Wuhan University

io

299 Bayi Road, Wuhan, China.

rs
ve

Email: [email protected]
d

Siluo YANG is the corresponding author of this paper.

ite
ed
Un

1
Abstract
Chinese scholars both in China and overseas have played an important role in global
scientific research. This paper focuses on the differences in research productivity and
impact between Chinese scholars both in China and overseas. A total of 1190 Chinese
scholars in China and 1983 Chinese scholars overseas were recognized out of 6306
papers in six journals over ten years. Research performance was evaluated by
informetrics indicators, including the absolute value, proportion, and average number
of authors, publications, citations, and usage count. Statistics metrics, including

8
standard deviation and coefficient of variance, were used to test the viability of two

01
groups. These metrics conclude the following: 1) in general, Chinese scholars in China

/2
have fewer advantages than Chinese scholars overseas on all indicators, namely, the

/3
absolute and proportion values of authors, publications, citations, and usage counts; 2)

12
Chinese scholars in China have more research potentiality than other scholars due to

on
the faster speed of development; and 3) regarding differences in research areas, in
computer science, Chinese scholars in China have a greater population but much less
d
he
output than Chinese scholars overseas. In addition, from the perspective of average
is

impact, Chinese scholars in China have a slightly stronger impact in most research areas
bl

than Chinese scholars overseas. Overall, Chinese scientific research in China is more
pu

developed and better at science than social science.

n

Keywords
io

Chinese scholar; Research productivity; Research impact; Scientific communication;

rs
ve

Usage metrics
d
ite
ed
Un

2
1. Introduction
China is facing a great tide of returnees who are finishing their studies or were
working abroad and are now ready to make a living in China. According to the annual
report from Center for China and Globalization 1, two million returnees holding
international degrees have come back to China. The large population not only
represents the large number of Chinese scholars who studied overseas but also indicates
that China has increasing opportunities and that beneficial policies in China have been
attracting students and workers abroad back to China 2. On the one hand, Chinese

8
scholars overseas certainly have participated in many important scientific projects and

01
produced many significant outcomes 3. On the other hand, the Chinese research level

/2
in China is no longer far behind the global level and is perhaps even higher in some

/3
subjects 4. Chinese scientific research has captured increasing attention in the global

12
academic environment. Chinese scholars’ names have become common on papers and

on
awards both individually in international journals and conferences and together with
scholars in other countries 5.
d
he

The two abovementioned scenarios represent the research output of Chinese

is

scholars in China and Chinese scholars overseas. Chinese scholars in China and
bl

overseas used to share a similar level of education in China. We wonder if there is any
pu

difference for those who moved abroad in terms of research productivity and impact.
n

In this paper, we conduct an individual-oriented analysis on the differences between

io
rs

Chinese scholars in China and overseas. The differences are displayed by a performance
ve

evaluation of Chinese scholars through the following questions: (1) What are the
different research productivities and impacts between Chinese scholars working in
d
ite

China and overseas? (2) What are the differences between the two groups in different
ed

research areas? The potential implications will be knowing the current situation of
Un

Chinese international communication as well as future policy on scholarly

communication.
2. Related Work
2.1 Research performance evaluation metrics
Performance evaluation involves evaluation objects and indicators. Countries,
institutions, and individual scholars have all been evaluation objects. Different
indicators are utilized on different objects. One indicator is sufficient for evaluation
sometimes, but generally, a complex indicator system is more convincing 6.

3
 Bibliometrics-oriented evaluation is the most common method 7. All kinds of
scientific results can be used to measure object performance through bibliometrics.
Among them, the number of publications and citations are the most common 8. Scholars’
productivity is usually measured by the number of papers, patents, presentations, and
awards. Citation metrics have been overused to measure scholars’ impact, including
total citations, citations per paper, h-index and other deformations 9. Many perspectives
have been discussed recently since the absolute quantity on only a few measurements
10,11
cannot satisfy in an era of fast-changing data . For this reason, Altmetrics is

8
proposed 12. Then, as the leader in literature integration platforms, the Web of Science

01
13
put forward Usage Count to adjust the utility of user browsing and saving behavior

/2
instead of using citations only. Indeed, the relationship between the number of citations

/3
14,15
and downloads has been discussed for a long time , and it attracted increasing

12
attention until Altmetrics started being utilized on social media .

on
2.2 Performance evaluation of China
Since China entered the realm of international scholar communication, the
d
he
Chinese have approached China's current developmental situation and scientists from
is

other countries have also been highly interested in Chinese issues 16. Previous studies
bl

have focused on two aspects. One is the approach to the performance evaluation of
pu

China itself. Liu 17probed the patterns and dynamics of China’s social science research
n

via bibliometric analyses. Zhu and Willett18 conducted a bibliometric analysis of the
io

development of Chinese research in superconductivity. Second, a comparative analysis

rs
ve

has been continuously discussed between China and other countries. The topics include
19
wide ranges at the continental level , such as comparing Asia with Europe on the
d
ite

relationship between research productivity and economics considering that China is

part of Asia, as well as at the individual level 20,21, such as comparing China with India
ed

22
and Malaysia as developing countries, and at the district level , such as comparing
Un

mainland China with Hong Kong and Taiwan. Most studies have taken China in its
entirety to uncover the difference from external objects, and fewer studies have focused
on internal comparisons, such as Chinese scholars themselves but inside and outside
China. However, more studies have analyzed Chinese scholars in China and overseas
separately. These studies normally analyze Chinese scholars’ performance by
comparing international and domestic data sources, such as WOS vs. Chongqin VIP 23.
On the other hand, because the United States is a multi-national country, the analysis
on scholars’ performance within internal America is more popular 24,25
. Studies have
4
addressed productivity differences between foreign-born American scholars and
America-born American scholars as well as the performance of visiting scholars in
America 26.
In summary, the number of publications and citations are two irreplaceable
27
dimensions for research performance evaluation . In this paper, however, we make
use of Usage Count for greater insight. Second, the comparison between Chinese
scholars in China and overseas will be our only target in this study, which has been rare
in previous studies. We believe the first approach on the performance evaluation of

8
Chinese scholars in China and overseas with updated metrics will elucidate the insights

01
of the current and potential distribution of the Chinese research level to derive some

/2
ideas about Chinese scientific decision policies and current education construction

/3
projects.

12
3. Methodology

on
To compare the research performance between Chinese scholars in China and
d
overseas, we used data from six representative journals, conducted an identification
he

strategy and relied on several indicators in the following section.

is

3.1 Data sources and collection

bl

Selecting representative journals as data sources is a common method used in

pu

informetrics and scientometrics. In studies, it is common to enumerate at least one

n

journal from which to collect data 28. Journals were selected using the following criteria:
io
rs

(1) Choose three journals each from SCI journals and SSCI journals based on 2016 JCR.
ve

(2) Subjects to which journals belong must be independent, aiming to reduce the bias
of duplicate author names and multi-disciplines. For example, if “Computer Science,
d
ite

Software Engineering” is chosen, then “Computer Science, Hardware Engineering”

ed

cannot be chosen. (3) Select the top journal from the one and only associated subject.
Un

Journals in JCR are usually categorized to two or more subjects. If a journal is the top
journal in the subject but belongs to other subjects, we skip it until we find the top
journal which belongs only to this subject. For example, MIS Quarterly ranks highest
in the subject Information Science & Library Science-SSCI. However, it belongs to
another subject. Therefore, we keep looking until we find the International Journal of
Information Management. For this reason, the International Journal of Information
Management and the Journal of Business Venturing rank at 8th and 7th, respectively (see
Table 1).

5
 Articles and article proceedings published between 2005 and 2014 were
downloaded from the Web of Science platform (WoS). Reviews in American Historical
Review and Biological Reviews were also downloaded since these two journals are more
review-oriented rather than article oriented. All full-record data were collected on July
15th, 2016. We decided on the year 2014 as the end of targeting period because papers
published in 2015 and 2016 may not have been cited yet. Papers published inside China
and outside China were downloaded and processed separately.

Table 1

8
Selected journals as data sources

01
Journal Full Name Abbr. JCR Subject Rankings(

/2
S/P)

/3
ACM Transactions on Graphics ACM Computer Science, Software 1/1
Engineering-SCI

12
American Historical Review HIS History-SSCI 1/1
Biological Reviews BIO Biology-SCI 1/1

on
International Journal of Information INFOR Information Science & Library 1/8
Management Science-SSCI
d
Journal of Business Venturing BIZ Business-SSCI 1/7
he
New England Journal of Medicine MED Medicine, General & Internal-SCI 1/1
Note. Rankings are based on Impact Factor in Journal Citation Reports (JCR).
is
bl

Rankings(S) refer to the ranking of journals in the only subject of WoS. Rankings(P)
pu

refer to the ranking of journals in all subjects of WoS.

n
io

3.2 Identify Chinese scholars

rs

First, this study must demonstrate who the Chinese scholars are. Given the
ve

difficulty of identifying a scholar’s nationality, many studies have considered the

d

authors’ affiliations as their nationality for convenience. However, in this paper,

ite

affiliations will not serve as a substitute for nationality. Hence, we define individuals
ed

whose original nationality is Chinese (including Hong Kong, Taiwan, and Macau) as
Un

Chinese scholars. Thus, Chinese scholars in China are individuals whose original
nationality is Chinese and who are now working in China. Chinese scholars overseas
are individuals whose original nationality is Chinese but who are now working overseas.
This definition is for our research purposes. As we attempt to unravel the research
differences between one group who remains in China and another who used to be in
China but are now working overseas, it is better to collect data on scholars’ original
nationality instead of their current nationality. In addition, many scholars today have

6
changed nationality. Current nationality has become more complicated than original
nationality.
The question of how to know if a scholar is still working in China is simple to
answer. The affiliation information in the dataset fully represents the current working
situation. The real question is how to collection a scholar’s nationality. After examining
the scholars’ CVs, we realized that few people would establish either original or current
nationality. However, we believe educational background information could represent
the most similar information. People tend to receive their first educational stage in their

8
29
mother countries . We determined the college period as the first educational stage

01
because few people publish information about their kindergartens, primary schools, or

/2
high schools, and the college period is the most fundamental stage for having a

/3
scientific career 30.

12
The identification process is as follows (See Figure 1): (1) Manually highlight

on
potential Chinese names. According to linguistic word-formation, people's names apart
from English or other Latin languages are very different from names in Chinese Pinyin
d
he
31
. Distinguishing Chinese names from Western, Japanese, Korean, and most other
is

Southeastern Asian names is simple. We highlighted names when the last names or first
bl

names appeared to be Chinese names, including deformations. For example, “Wong”

pu

is a deformation of “Wang.” (2) Look for college information from their homepages,
n

CVs, LinkedIn profiles, ResearchGate profiles, and other resources to determine if they
io

were educated in China. We were surprised to find other helpful information, such as
rs
ve

mother language and high school information, which assured us of their original
nationalities. (3) Contact the remaining 597 uncertain authors for the information. Only
d
ite

16 authors who could not be reached had to be abandoned in the following analysis.
ed
Un

7
 8
01
/2
/3
12
Figure 1. The results and procedure for identifying Chinese authors

on
3.3 Indicators description
In the following sections, several measurements are discussed (see Table 2) for
d
he
representing productivity and impact. Paper*, P*, and R* describe the publication
is

ability of scholars. RTC*, RU1*, and RU2* describe the scientific influence of scholars.
bl

TC refers to Total Citations, and U1 and U2 are new metrics on Web of Science that
pu

represent how many times a paper has been browsed or saved. All three indicators
n

illustrate a paper’s influence. Total Papers and Author* give an absolute value on the
io

basic background. Total Papers is distinguished from Paper* since Paper* only
rs
ve

includes papers published by Chinese authors. P*s and R*(TC, U1, U2)s avoid the bias
from the scale of authors by years or by research areas.
d
ite

Table 2
ed

Indicators description and formulation in this paper

Un

Indicators Description
Total papers The total amount of selected journals in a year or a research area.
Author* The number of Chinese authors who have publications in a year or a
research area.
Paper* The number of papers only published by Chinese scholars in a year or a
research area.
P* The percentage of Papers in the total papers. P=Paper/ Total papers.
R* The ratio of Papers per author. R=Paper/Author.
TC* The number of total citation of a paper on Web of Science platform.
U1* The count of the number of times the full text of a record has been accessed
or a record has been saved in the last 180 days on Web of Science platform.

8
 U2* The count of the number of times the full text of a record has been accessed
or a record has been saved since February 1, 2013, on Web of Science
platform.
RTC* The total citation ratio of Papers per author. RTC=TC/Author.
RU1* The ratio of U1 of Papers per author. RU1=U1/Author.
RU2* The ratio of U2 of Papers per author. RU2=U2/Author.
Note. Indicators with a star (*) mean they contain sub indicators, e.g. AuthorC refers to
Chinese authors in China. AuthorO refers to Chinese authors overseas. C is short for
China, and O is short for overseas.

8
3.4 Calculation concerns

01
For the calculation of co-author papers, these papers were all published in top

/2
journals, and authors in these journals could be acknowledged as outstanding scholars,

/3
so we used the normal count approach in which each author is counted as one

12
publication. Since our study is author-oriented research, the counting calculation is
from the perspective of authors. For instance, an article entitled “Creating Consistent
on
Scene Graphs Using a Probabilistic Grammar” was written by Liu, Tianqiang;
d
he
Chaudhuri, Siddhartha; Kim, Vladimir G.; Huang, Qixing; Mitra, Niloy J.; Funkhouser,
Thomas. Then, Liu, Tianqiang and Huang, Qixing were identified as Chinese scholars.
is
bl

We assigned both Liu, Tianqiang and Huang, Qixing as having one publication, which
pu

denotes that both Liu, Tianqiang and Huang, Qixing have published an article. In this
n

case, author order was not considered. Regardless whether the order of an author is
io

third or thirtieth, she or he published a paper based on the premise that all authors are
rs

outstanding scholars in these journals. In addition, we abandoned counting methods

ve

involving weights because some papers had more than 50 authors, and in such cases,
d

adjusted count and weighted count approaches1 would not work.

ite

In addition, we determined that people with the same names working with different
ed

affiliations were different people. However, if the same names were shown in one paper,
Un

we considered that one person was working with different affiliations. In addition, if a
person worked both in China and overseas, we counted the paper once both in China
and overseas.

1Two methods were used to measure the different contribution of each author in a paper. The adjusted count
approach gives every author a mean of 1 divided by the total number of authors in a paper. The weighted count
approach gives authors different weights by order of author.

9
4. Findings
4.1 The numbers of authors in China and overseas
In total, we found 1983 Chinese authors overseas and 1190 Chinese authors in
China. In different years, we calculated 2270 Chinese authors overseas and 1654
Chinese authors in China. The reason for this inequality is that some authors published
more than one paper in ten years.
The first part of Figure 2 shows the comparison sequentially. Generally, AuthorC
and AuthorO both increase over the years. AuthorC is almost always less than AuthorO,

8
except in 2013, because in 2013, there was a sudden increase of Chinese authors in

01
China in computer science and medicine, according to Figure 3. Additionally, the

/2
increasing and decreasing tendencies of AuthorC and AuthorO are similar, especially

/3
from 2006 to 2011. However, in 2012 and 2013, a dramatic reversal happens, as

12
mentioned above.

on
The second part of Figure 2 shows the comparison in different research areas.
d
Apparently, AuthorO is larger than AuthorC except in computer science. In information
he

science, business, and medicine areas, AuthorO approximately doubles AuthorC, which
is

implies that more Chinese overseas are good in these research areas than Chinese
bl

scholars in China 32. However, scholars in China are better in computer science since
pu

Chinese scholars overseas are 147 less than in China. We also found few Chinese
n

scholars in history regardless of group. This finding is reasonable because the journal
io
rs

(American History Review) mainly focuses on American history, which leads to few
ve

other nationalities in this study area.

d
ite
ed
Un

10
 8
01
/2
/3
12
on
d
is he
bl
pu
n

Figure 2. Comparison of the numbers of Chinese authors

io
rs

In the history area, we found only one Chinese author and only one paper written
ve

by a Chinese scholar; therefore, we excluded the analysis of the history area in Figures
d

3, 7, and 9 when they referred to the detailed analysis of research areas in the following
ite

sections. In addition, we displayed the analysis of the history area in tables and figures
ed

when they contain an overview analysis. For instance, even though Figure 5 describes
Un

the detailed analysis of research areas, it includes history because it contains an

overview of Total papers.

11
 8
01
/2
/3
12
on
d
Figure 3. The number of authors between China and overseas in different research areas
is he

First, the trend line for every research area increases. The number of Chinese
bl

authors in total in computer science increases the most, while the number of Chinese
pu

authors in biology and business increase the least. Second, in the first seven years, the
n

increasing tendency is mild, but in 2012 and 2013, the number of Chinese authors in
io
rs

total increases more. The reason for this greater increase is that Chinese scholars had
ve

realized the importance of international scientific communication, and the Chinese

government was providing increasing support to scientists studies abroad 5. Third,
d
ite

Chinese authors in China generally still have fewer advantages in the number of
ed

research areas because every yellow line that represents Chinese authors in China is
Un

below the green line that represents Chinese authors overseas, except in computer
science. In addition, the two types of lines do not always align but are more often
inversely related. In general, when the number of Chinese authors overseas rises, the
number of Chinese authors in China goes down, such as in the years 2012 and 2013 in
medicine. This situation occurred mainly because of scholarly communication and
exchange to some extent.

12
4.2 Chinese authors in China and overseas productivity performance
Six journals published 6306 papers in ten years (shown in the first part of Figure
4). The number of publications slightly increases yearly after 2006. Chinese authors
overseas published 1319 papers in ten years. Meanwhile, Chinese authors in China only
published 400 papers. Pc and Po also show the huge gap between two groups. Po
mainly stays at the same level, which is between 20% and 25%, with two exceptions,
while Pc continues to rise from 2% to nearly 10%. Therefore, the trend of Pc will
potentially keep growing, which means that Chinese scholars in China will publish an

8
increasing number of papers in the future.

01
In the second part of Figure 4, the number of total papers in all research areas

/2
aligns with the number of total authors in Figure 2. The only exception is Pc, which is

/3
not higher than Po in computer science, which is opposite of the author number in

12
computer science.

on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

13
 Figure 4. Comparison of total papers and P

A close examination of every research area (see Figure 5) shows first that there
are no common rules for total publication. Overall, the number of total publications
increases by various amounts for each area. The number of total papers in computer
science, information science, and biology doubled in ten years, while in medicine 300
papers were published every year. In addition, the publication number in history was
unpredictable.

8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

Figure 5. Comparison of total papers and P in different research areas

Pc changed inversely with Po in most areas. Especially in computer science, the

changes in Pc were the opposite of Po. However, in information science and business
only, Pc fell commensurate with Po in the last three years. The same phenomenon
occasionally occurred in biology and medicine. As in Figure 3, we assume the reason
14
for the reverse proportion changes can be described by the following scenario. When
some outstanding Chinese scholars go abroad, the number of papers published by
Chinese in China will drop because of the loss of talent who otherwise would have
published papers under their affiliations inside China. When some scholars return to
29
China, the scenario reverses . This interesting phenomenon may have other
explanations that can be discussed further by other papers.
After examining the situation from the Total papers perspective, let us narrow the
analysis to papers only published by Chinese scholars (all tables and figures in the

8
following sections use Paper*). Apparently, the gap between inside and outside China

01
is large as well. Even though both PaperC and PaperO increased over ten years, the

/2
absolute value of PaperO is at least twice PaperC. However, PaperC increased six

/3
times, which was much higher than the growth of PaperO, namely, 60%. The situation

12
is similar to the comparison between Pc and Po. Therefore, Chinese authors in China

on
may publish more papers in the future. However, Ro stays in the range of 0.5 to 0.7,
which means that two scholars co-author a paper every year. Despite small spikes, Ro
d
he
has dropped slightly in ten years. On the other hand, Rc has slightly increased in ten
is

years but ranges from 0.2 to 0.3, which means three or four scholars co-author
bl

approximately one paper every year. The comparison shows that Chinese authors in
pu

China publish fewer papers per person, while the degree of co-authoring is higher for
n

Chinese authors in China than for Chinese authors overseas.

io

In different research areas (second part of Figure 6), Chinese scholars in China
rs
ve

still published fewer papers than Chinese scholars overseas. Even though there were
more Chinese authors in China in computer science (See Figure 2), their productivity
d
ite

in ACM was less than that of the group overseas. In addition, in medicine, double the
ed

number of Chinese authors overseas published seven times the number of papers as
Chinese authors in China. Meanwhile, the more Paper* published in one area in total,
Un

the larger the gap of papers per author between the two groups. In medicine, Ro is five
times Rc. In addition, in computer science and information science where Paper* is
less than in medicine, the multiple is double. However, the exception is in the business
area. Scholars in business published fewer papers than in information science, but the
gap between Rc and Ro is the smallest. We believe this result implies scholars in
business share the same co-author model globally.

15
 8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed

Figure 6. Comparison of Paper and R

Un

Unlike PaperC and PaperO in total, which both continue to increase

synchronously, PaperC and PaperO in each area are unpredictable. On the one hand,
the growth in each area varies. On the other hand, unlike the changes in Figure 6, the
growth of PaperC and PaperO in Figure 7 does not change at the same pace all the
time. For instance, PaperO in information science increases greatly in 2011, while
PaperC decreases. The reason for this contradiction is the small scale of publication in
the areas of information science, business, and biology. Therefore, these results can
hardly predict the future of Chinese author productivity in these three areas. However,
16
in computer science and medicine, the growth in PaperC and PaperO correspond more
with each other. Especially in ACM, PaperC rises ten times in ten years, which suggests
Chinese authors in China will publish more papers in the future.
Rc and Ro in each area change inversely, especially in ACM and MED. Rc and Ro
in information science tend to be reversed as well. Due to the impact from PaperO and
PaperC, Rc and Ro in business and biology areas change dramatically. Overall, Rc and
Ro change inversely, as we mentioned above that Pc and Po change inversely. This
finding is one more piece of evidence to prove that the number of publications is

8
affected by scholarly exchange.

01
/2
/3
12
on
d
is he
bl
pu
n

Figure 7. Comparison of Paper and R in different research areas

io
rs
ve

Since there is no significant result from the absolute value of Paper* in each area,
we calculated the statistics for Paper* and R* (See Table 3). The coefficient of variance
d
ite

(C.V) can be used to compare the volatility of each area by years since standard
ed

deviation (SD) is the absolute value in our study. All C.Vs are above 1.00, which means
Paper* and R* in each area are not stable. For instance, PaperC in business and biology
Un

areas and Rc in MED have the highest C.V, which means the productivity performance
of Chinese scholars in China is not stable in these areas. Another point is that almost
all C.Vs of PaperC are higher than PaperO, and almost all C.Vs of Rc are higher than
Ro. This finding implies that Chinese authors in China are more unstable but developing.
Because high variance creates possibilities, and Chinese authors in China to date have
published increasingly over the years, there would be a possibility of constantly
increasing publication.

17
Table 3
Statistics of Paper and R in different research areas
Indicators ACM BIO INFOR BIZ MED
PaperC SD 13.189 1.252 3.604 1.430 4.447
Mean 9.840 0.840 2.520 0.960 3.400
C.V 1.340 1.490 1.430 1.489 1.308
PaperO SD 14.851 1.252 4.614 1.889 16.667
Mean 11.680 0.960 3.200 1.500 13.100
C.V 1.271 1.304 1.442 1.259 1.272
Rc SD 0.058 0.477 0.346 0.422 0.185
Mean 0.044 0.429 0.240 0.360 0.123
C.V 1.313 1.114 1.440 1.171 1.511

8
Ro SD 0.052 0.471 0.140 0.165 0.072

01
Mean 0.048 0.400 0.109 0.142 0.065
C.V 1.085 1.179 1.285 1.160 1.120

/2
Note. Distributions with C.V < 1 are considered as low-variance, while those with

/3
C.V > 1 are considered as high-variance.

12
on
4.3 Chinese authors in China and overseas impact analysis
d
Total citations (TC) tend to go down over ten years both inside and outside China,
he

while U1 and U2 tend to rise. This finding is because citations have a time lag, and the
is

newer the paper is, the fewer citations it may have. However, usage count (especially
bl

U2) is a cumulative amount, which always increases.

pu

The absolute value of each indicator between inside and outside China has a ten-
n

digit difference, regardless of the total numbers or the means. C.V is useful in this case.
io
rs

The results (see Table 4) show that C.Vs from inside China are twice those from outside
ve

China. Therefore, the impact of papers published by Chinese authors in China is more
d

unpredictable, and the impact of papers published outside China is stable. Taking TC
ite

as an example, TCo has a four- or five-year cycle, while TCc continues rising at the
ed

beginning but recently has a two-year cycle. We explain this result by saying that papers
Un

published by Chinese authors in China are active.

18
Table 4
Statistics of TC/U1/U2 over time
PY TCo TCc U1o U1c U2o U2c
2005 38978 2589 297 27 2082 200
2006 33798 2410 258 24 1832 218
2007 24812 3563 194 59 1311 326
2008 26259 7745 278 76 1692 722
2009 30551 9070 539 108 2698 901
2010 30225 3254 384 114 2874 769
2011 29060 4279 514 107 3713 920
2012 26156 1725 563 82 4572 775
2013 17573 4695 614 320 5399 2469

8
2014 12073 1767 563 211 3240 1142

01
Total 269485 41097 4204 1128 29413 8442
SD 7707.464 2485.488 154.718 90.110 1318.059 653.083

/2
Mean 26948.500 4109.700 420.400 112.800 2941.300 844.200

/3
C.V 0.286 0.605 0.368 0.799 0.448 0.774

12
From the perspective of absolute value (See Table 4 and 5), we found a large gap

on
of impact between inside and outside China both by years and by disciplines. TCo, U1o,
and U2o in medicine are seven times those inside China. The smallest gap is in
d
he

computer science. However, after being divided by Paper* separately, papers published
is

outside China do not have such big advantages. Because Chinese authors overseas
bl

published more papers than in China, the absolute values of each indicator in citation
pu

and usage are higher reasonably. In contrast, papers published by Chinese authors in
n

China have more impact in most areas from the perspective of per-paper value, except
io

in biology. Per-paper TCo in biology is five times that of TCc, and per-paper U1o and
rs
ve

U2o are twice those of U1c and U2c. In other research areas, per-paper TCc, U1c, and
U2c are generally higher than those of TCo, U1o, and U2o. The opposition between the
d
ite

absolute value and the per-paper value implies papers published by Chinese authors in
ed

China have the same or even better quality than papers published by Chinese authors
overseas, even though the latter has a larger population and has published more papers.
Un

Table 5
TC/U1/U2 in total in different research areas
Indicators ACM HIS BIO INFOR BIZ MED
TCo 5710 1 812 1628 765 260569
per PaperO 13.96 1 62.46 15.96 20.68 344.21
TCc 3775 0 90 402 413 36417
per PaperC 14.63 0 12.86 12.97 29.5 404.63
U1o 218 2 114 282 260 3328
per PaperO 0.53 2 8.77 2.76 7.03 4.40
U1c 210 0 29 144 147 598

19
 per PaperC 0.81 0 4.14 4.65 10.5 6.64
U2o 1831 7 1015 2675 1745 22140
per PaperO 4.48 7 78.08 26.23 47.16 29.25
U2c 1629 0 239 1110 960 4504
per PaperC 6.31 0 34.14 35.81 68.57 50.04

The method using Paper* to divide TC*, U1*, and U2* addresses paper quality.
The method using Author* to divide indicators addresses author quality.
TC has gone through a dramatic route over ten years. Both RTCc and RTCo have
tended to decrease yearly with only slight internal growth. RTCo decreases greatly from

8
more than 200 to less than 50. RTCc basically remains under 50. Therefore, there is a

01
large overall gap in Chinese author impact between China and overseas. However, the

/2
gap degree varies in different research areas. The largest gap of RTC between the two

/3
groups is in medicine followed by biology. However, in business, RTCc is larger than

12
RTCo. We could say Chinese authors overseas have more impact than Chinese authors
in China in biology and medicine compared to other areas.
on
The usage count numbers in China are not comparable to those overseas. Both
d
he

RU1c and RU2c are always below RU1o and RU2o, and the latter are twice as much as
is

the former almost every year. Similar to other R*, RU1 and RU2 in the two groups also
bl

changed inversely over the years. Therefore, scholar exchange not only affects
pu

publishing performance but also impacts results.

n

It is interesting to see that some research areas have fewer publications in which
io
rs

RU1 and RU2 are larger than in other research areas with more publications. For
ve

instance, Chinese authors in biology and business published fewer papers than in
computer science and medicine. However, RU1 and RU2 of both groups in biology and
d
ite

business are larger than those in computer science and medicine. This finding implies
ed

papers published by Chinese authors receive more attention in biology and medicine.
Un

Additionally, unlike other research areas, the usage count numbers inside China in
business are larger than those outside China, which means Chinese authors in China
have more impact than Chinese authors overseas in the business area.

20
 8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve

Figure 8. Comparison of TC/U1/U2 per Chinese authors

d
ite

As for individual research areas, RTC tends to decrease, while RU1 and RU2 have
ed

ups and downs, which is similar to the overall distribution. Moreover, similar to before,
Un

all indicators for overseas are larger than those for China. Except in business, three
indicators for China are larger than those for overseas in some years. Another finding
is that the change in three indicators, including growth and decline, is generally
synchronous. This result may demonstrate informetrics studies of a potential research
point that usage count may be relevant to citation metrics.

21
 Figure 9. Comparison of TC/U1/U2 per Chinese authors in different research areas

8
01
/2
All RTCo apparently decrease, but some RTCc increase slightly, such as the one

/3
in business. We also see there are two peaks in the RTCc line in business. We believe

12
these peaks are caused by the global economic depressions in 2007 and 2011 so that
many Chinese scholars returned to China instead of studying abroad 33.

on
All of RU1 shows a very different situation in different research areas. When all
d
of RU1c has more spikes, all of RU1o remains more stable and vice versa. U1 refers to
he

the match with user information needs in a short time, so the uneven lines have a
is
bl

reasonable explanation. The increasingly stable lines in RU2 prove that U2 represents
pu

a long-term result. However, RU2 of Chinese authors in China in information science

and business are less stable than in other research areas. We reason it is because
n
io

information science and business are new for Chinese authors in China as they explore
rs

the global academic environment.

ve

5. Conclusion and discussion

d

To answer the first research question, our results show the big differences between
ite

Chinese scholars in China and overseas on both research productivity and impact. 1)
ed

One of the differences between the two groups is based on static data. Chinese scholars
Un

in China have fewer advantages than Chinese scholars overseas on all indicators,
namely, the absolute and proportional values of authors, publications, citations, and
usage counts. In fact, those four aspects of indicators represent research power, output
34
capability , formal impact, and informal impact individually. Therefore, Chinese
scholars in China lack these abilities comparatively. We chose the highest IF journals
as data sources to display the top level of research. Only good research results would
be published in top journals, even though IF has been argued about for a long time 35.
Chinese papers in total share a small proportion of the world in this study. Only when
22
Chinese scholars can publish papers in top journals can they reach the top research level.
2) The second difference between the two groups is the difference in tendency. Chinese
scholars in China have more potential development than overseas. Regardless of which
research area or which indicator is used, the metrics of Chinese scholars in China have
a faster speed of development.
As for the second research question, the differences vary between two groups from
different research areas as follows. 1) Regardless of the shortages of Chinese scholars
in China generally, they are good at some indicators in some research areas. For

8
example, in computer science, the number of Chinese scholars in China is larger than

01
Chinese scholars overseas. In other words, the research power of Chinese scholars in

/2
China in computer science is stronger than the other group. 2) For the output capability

/3
represented by the number of publications, the gaps between two groups vary from

12
different research areas. Larger gaps exist in computer science and medical science. 3)

on
From the perspective of impact, on average, papers written by Chinese scholars in
China have a slightly stronger impact in computer science, information science, library
d
he
science, business, and medicine than papers written by Chinese scholars overseas.
is

Meanwhile, in biology, Chinese scholars overseas have more impact both generally and
bl

on average. Using only one journal as a data source representing one research area is
pu

controversial. However, despite the small scale of publication, one representative for a
n

research area is a common way to conduct informetrics analysis. In addition, the top
io

journals are acknowledged to be good representatives for their respective research areas
rs

36
ve

. In addition, even in six journals, more than 50,000 authors in total are recognized.
Due to the limitation of manual identification, selecting additional journals requires
d
ite

more identification work.

ed

We used a primary approach to determine the differences between Chinese

scholars in China and overseas. We should work on more journals and other metrics in
Un

the future to improve this study. Two further considerations are demonstrated below.
On the one hand, Chinese scientific research is certainly progressing, but there is
much room for improvement in the future. Therefore, studying or working abroad for
Chinese scholars is very necessary. In fact, our data for Chinese scholars in China
included scholars in Hong Kong and Taiwan, which enlarged the data on Chinese
scholars in China. If the data only included papers from mainland China, the gap would
be even bigger, as we could expect.

23
 On the other hand, studying or working outside China requires different policies
in different circumstances. We have seen that fewer scholars work in social science
than in science. Under the Double World-class project, universities with a strong
background in science could improve science even more if they sent out scholars to
further their world-class disciplines. Universities with a comprehensive discipline
background could improve social science if they expanded social science studies
outside China, so they could build up the world-class universities. For instance,
American history has been scarcely discussed by Chinese scholars. We did not remove

8
it from our data because it may remind Chinese scholars of a research blank.

01
Acknowledgements

/2
We thank the support from Professor Rongying Zhao, the key project of National

/3
Social Science Foundation of China (grant number 17ATQ009), the youth project of

12
National Social Science Foundation of China (grant number 17CTQ025), and the

on
National Social Science Foundation of China (grant number 16BTQ055).
d
he

References
is

1. Center for China and Globalization, Investigative report on Chinese returnees

bl
pu

employment in 2016. 2016.

2. Hao, J., Wen, W., and Welch, A., When sojourners return: Employment
n
io

opportunities and challenges facing high-skilled Chinese returnees. Asian

rs

Pacific Migr. J., 2016, 25, 22–40.

ve

3. Zou, Y. and Laubichler, M. D., Measuring the contributions of Chinese

d

scholars to the research field of systems biology from 2005 to 2013.

ite

Scientometrics, 2017, 110, 1615–1631.

ed

4. Zong, F. and Wang, L., Evaluation of university scientific research ability

Un

based on the output of sci-tech papers: a D-AHP approach. PLoS One, 2017,
12, e0171437.
5. Yang, S. and Wang, F., Visualizing information science: Author direct citation
analysis in China and around the world. J. Informetr., 2015, 9, 208–225.
6. Radicchi, F. and Castellano, C., Analysis of bibliometric indicators for
individual scholars in a large data set. Scientometrics, 2013, 97, 627–637.

24
7. Ding, Y., Chowdhury, G. G., and Foo, S., Bibliometric cartography of
information retrieval research by using co-word analysis. Inf. Process. Manag.,
2001, 37, 817–842.
8. Petersen, A. M., Wang, F., and Stanley, H. E., Methods for measuring the
citations and productivity of scientists across time and discipline. Phys. Rev. E
- Stat. Nonlinear, Soft Matter Phys., 2010, 81, 36114.
9. Radicchi, F. and Castellano, C., A reverse engineering approach to the
suppression of citation biases reveals universal properties of citation

8
distributions. PLoS One, 2012, 7, e33833.

01
10. Barnes, C., The Use of Altmetrics as a Tool for Measuring Research Impact.

/2
Aust. Acad. Res. Libr., 2015, 46, 1–14.

/3
11. Bornmann, L., Do altmetrics point to the broader impact of research? An

12
overview of benefits and disadvantages of altmetrics. J. Informetr., 2014, 8,

on
895–903.
12. Priem, J., Taraborelli, D., Groth, P., and Neylon, C., Alt-metrics: a manifesto.
d
he
2010, pp. 1–5.
is

13. Wang, X., Fang, Z., and Sun, X., Usage patterns of scholarly articles on Web
bl

of Science: a study on Web of Science usage count. Scientometrics, 2016, 109,

pu

917–926.
n

14. Bollen, J., Van De Sompel, H., Smith, J. A., and Luce, R., Toward alternative
io

metrics of journal impact: A comparison of download and citation data. Inf.

rs
ve

Process. Manag., 2005, 41, 1419–1440.

15. Moed, H. F., Statistical relationships between downloads and citations at the
d
ite

level of individual documents within a single journal. J. Am. Soc. Inf. Sci.
ed

Technol., 2005, 56, 1088–1097.

16. Zhou, P. and Leydesdorff, L., The emergence of China as a leading nation in
Un

science. Res. Policy, 2006, 35, 83–104.

17. Liu, W., Hu, G., Tang, L., and Wang, Y., China’s global growth in social
science research: Uncovering evidence from bibliometric analyses of SSCI
publications (1978-2013). J. Informetr., 2015, 9, 555–569.
18. Zhu, Q. L. and Willett, P., Bibliometric analysis of Chinese superconductivity
research, 1986-2007. Aslib Proc. , 2011, 63, 101–119.
19. King, D. A., The scientific impact of nations. Nature, 2004, 430, 311–316.

25
20. Wong, C. Y. and Goh, K. L., The sustainability of functionality development of
science and technology: Papers and patents of emerging economies. J.
Informetr., 2012, 6, 55–65.
21. Larsen, P. O., Maye, I., and Ins, M. von, Scientific Output and Impact: Relative
Positions of China, Europe, India, Japan and the USA. Collnet J. Sci. Inf.
Manag., 2008, 2, 1–10.
22. Li, F., Yi, Y., Guo, X., and Qi, W., Performance evaluation of research
universities in Mainland China, Hong Kong and Taiwan: Based on a two-

8
dimensional approach. Scientometrics, 2012, 90, 531–542.

01
23. Hennemann, S., Wang, T., and Liefner, I., Measuring regional science

/2
networks in China: A comparison of international and domestic bibliographic

/3
data sources. Scientometrics, 2011, 88, 535–554.

12
24. Walsh, J. P., The impact of foreign-born scientists and engineers on American

on
nanoscience research. Sci. Public Policy, 2015, 42, 107–120.
25. Lee, S., Foreign-born scientists in the United States-do they perform differently
d
he
than native-born scientists? Georgia Institute of Technology, 2004.
is

26. Chao, X., Chinese visiting scholars’ academic socialization in US institutions

bl

of higher education: a qualitative study. Asia Pacific J. Educ., 2015, 35, 1–18.
pu

27. Bar-Ilan, J., Informetrics at the beginning of the 21st century-A review. J.
n

Informetr., 2008, 2, 1–52.

io

28. Appio, F. P., Cesaroni, F., and Di Minin, A., Visualizing the structure and
rs
ve

bridges of the intellectual property management and strategy literature: a

document co-citation analysis. Scientometrics, 2014, 101, 623–661.
d
ite

29. Leung, M. W. H., Unraveling the skilled mobility for sustainable development
ed

mantra: An analysis of China-EU academic mobility. Sustain., 2013, 5, 2644–

2663.
Un

30. Xu, H., Hou, Z.-J., Tracey, T. J. G., and Zhang, X., Variations of career
decision ambiguity tolerance between China and the United States and between
high school and college. J. Vocat. Behav., 2016, 93, 120–128.
31. Zhu, H., A Cultural Comparison of English and Chinese Family Names. Asian
Soc. Sci., 2010, 6, 226–229.
32. Fu, H.-Z. and Ho, Y.-S., Independent research of China in Science Citation
Index Expanded during 1980–2011. J. Informetr., 2013, 7, 210–222.

26
33. Grossman, R. S. and Meissner, C. M., International aspects of the Great
Depression and the crisis of 2007: Similarities, differences, and lessons. Oxford
Rev. Econ. Policy, 2010, 26, 318–338.
34. Veugelers, R., Towards a multipolar science world: Trends and impact.
Scientometrics, 2010, 82, 439–456.
35. Alberts, B., Impact Factor Distortions. Science (80-. )., 2013, 340, 787.
36. Walters, W. H., Do subjective journal ratings represent whole journals or
typical articles? unweighted or weighted citation impact? J. Informetr., 2017,

8
11, 730–744.

01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

27
Table 1
Selected journals as data sources
Journal Full Name Abbr. JCR Subject Rankings(
S/P)
ACM Transactions on Graphics ACM Computer Science, Software 1/1
Engineering-SCI
American Historical Review HIS History-SSCI 1/1
Biological Reviews BIO Biology-SCI 1/1
International Journal of Information INFOR Information Science & Library 1/8
Management Science-SSCI
Journal of Business Venturing BIZ Business-SSCI 1/7
New England Journal of Medicine MED Medicine, General & Internal-SCI 1/1

8
Note. Rankings are based on Impact Factor in Journal Citation Reports (JCR).

01
Rankings(S) refer to the ranking of journals in the only subject of WoS. Rankings(P)

/2
refer to the ranking of journals in all subjects of WoS.

/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

28
Table 2
Indicators description and formulation in this paper
Indicators Description
Total papers The total amount of selected journals in a year or a research area.
Author* The number of Chinese authors who have publications in a year or a
research area.
Paper* The number of papers only published by Chinese scholars in a year or a
research area.
P* The percentage of Papers in the total papers. P=Paper/ Total papers.
R* The ratio of Papers per author. R=Paper/Author.
TC* The number of total citation of a paper on Web of Science platform.
U1* The count of the number of times the full text of a record has been accessed

8
or a record has been saved in the last 180 days on Web of Science platform.

01
U2* The count of the number of times the full text of a record has been accessed

/2
or a record has been saved since February 1, 2013, on Web of Science
platform.

/3
RTC* The total citation ratio of Papers per author. RTC=TC/Author.

12
RU1* The ratio of U1 of Papers per author. RU1=U1/Author.
RU2* The ratio of U2 of Papers per author. RU2=U2/Author.
Note. Indicators with a star (*) mean they contain sub indicators, e.g. AuthorC refers to

on
Chinese authors in China. AuthorO refers to Chinese authors overseas. C is short for
d
China, and O is short for overseas.
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

29
Table 3
Statistics of Paper and R in different research areas
Indicators ACM BIO INFOR BIZ MED
PaperC SD 13.189 1.252 3.604 1.430 4.447
Mean 9.840 0.840 2.520 0.960 3.400
C.V 1.340 1.490 1.430 1.489 1.308
PaperO SD 14.851 1.252 4.614 1.889 16.667
Mean 11.680 0.960 3.200 1.500 13.100
C.V 1.271 1.304 1.442 1.259 1.272
Rc SD 0.058 0.477 0.346 0.422 0.185
Mean 0.044 0.429 0.240 0.360 0.123
C.V 1.313 1.114 1.440 1.171 1.511

8
Ro SD 0.052 0.471 0.140 0.165 0.072

01
Mean 0.048 0.400 0.109 0.142 0.065
C.V 1.085 1.179 1.285 1.160 1.120

/2
Note. Distributions with C.V < 1 are considered as low-variance, while those with

/3
C.V > 1 are considered as high-variance.

12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

30
Table 4
Statistics of TC/U1/U2 over time
PY TCo TCc U1o U1c U2o U2c
2005 38978 2589 297 27 2082 200
2006 33798 2410 258 24 1832 218
2007 24812 3563 194 59 1311 326
2008 26259 7745 278 76 1692 722
2009 30551 9070 539 108 2698 901
2010 30225 3254 384 114 2874 769
2011 29060 4279 514 107 3713 920
2012 26156 1725 563 82 4572 775
2013 17573 4695 614 320 5399 2469

8
2014 12073 1767 563 211 3240 1142

01
Total 269485 41097 4204 1128 29413 8442
SD 7707.464 2485.488 154.718 90.110 1318.059 653.083

/2
Mean 26948.500 4109.700 420.400 112.800 2941.300 844.200

/3
C.V 0.286 0.605 0.368 0.799 0.448 0.774

12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

31
Table 5
TC/U1/U2 in total in different research areas
Indicators ACM HIS BIO INFOR BIZ MED
TCo 5710 1 812 1628 765 260569
per PaperO 13.96 1 62.46 15.96 20.68 344.21
TCc 3775 0 90 402 413 36417
per PaperC 14.63 0 12.86 12.97 29.5 404.63
U1o 218 2 114 282 260 3328
per PaperO 0.53 2 8.77 2.76 7.03 4.40
U1c 210 0 29 144 147 598
per PaperC 0.81 0 4.14 4.65 10.5 6.64
U2o 1831 7 1015 2675 1745 22140

8
per PaperO 4.48 7 78.08 26.23 47.16 29.25

01
U2c 1629 0 239 1110 960 4504
per PaperC 6.31 0 34.14 35.81 68.57 50.04

/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

32
Figure 1. The results and procedure for identifying Chinese authors
Figure 2. Comparison of the numbers of Chinese authors
Figure 3. The number of authors between China and overseas in different research areas
Figure 4. Comparison of total papers and P
Figure 5. Comparison of total papers and P in different research areas
Figure 6. Comparison of Paper and R
Figure 7. Comparison of Paper and R in different research areas
Figure 8. Comparison of TC/U1/U2 per Chinese authors
Figure 9. Comparison of TC/U1/U2 per Chinese authors in different research areas

8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

33
 8
01
/2
/3
12
Figure 1. The results and procedure for identifying Chinese authors

on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

34
 8
01
/2
/3
12
on
d
is he
bl
pu
n

Figure 2. Comparison of the numbers of Chinese authors

io
rs
ve
d
ite
ed
Un

35
 8
01
/2
/3
12
on
d
Figure 3. The number of authors between China and overseas in different research areas
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

36
 8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs

Figure 4. Comparison of total papers and P

ve
d
ite
ed
Un

37
 8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs

Figure 5. Comparison of total papers and P in different research areas

ve
d
ite
ed
Un

38
 8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed

Figure 6. Comparison of Paper and R

Un

39
 8
01
/2
Figure 7. Comparison of Paper and R in different research areas

/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

40
 8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve

Figure 8. Comparison of TC/U1/U2 per Chinese authors

d
ite
ed
Un

41
 Figure 9. Comparison of TC/U1/U2 per Chinese authors in different research areas

8
01
/2
/3
12
on
d
is he
bl
pu
n
io
rs
ve
d
ite
ed
Un

42

View publication stats