Available via license: CC BY 4.0
Content may be subject to copyright.
1
Publication modalities ‘article in press’ and ‘open access’ in relation to
journal average citation
Sara M. González-Betancor *
Universidad de Las Palmas de Gran Canaria, Departamento de Métodos Cuantitativos
en Economía y Gestión, Campus de Tafira, 35017 Las Palmas de Gran Canaria, Spain.
E-mail: sara.gonzalez@ulpgc.es
ORCID: http://orcid.org/0000-0002-2209-1922
Pablo Dorta-González
Universidad de Las Palmas de Gran Canaria, TiDES Research Institute, Campus de
Tafira, 35017 Las Palmas de Gran Canaria, Spain. E-mail: pablo.dorta@ulpgc.es
ORCID: http://orcid.org/0000-0003-0494-2903
* Corresponding author
2
Publication modalities ‘article in press’ and ‘open access’ in relation to
journal average citation
Abstract:
There has been a generalization in the use of two publication practices by scientific
journals during the past decade: 1. ‘article in press’ or early view, which allows access
to the accepted paper before its formal publication in an issue; 2. ‘open access’, which
allows readers to obtain it freely and free of charge. This paper studies the influence of
both publication modalities on the average impact of the journal and its evolution over
time. It tries to identify the separate effect of access on citation into two major parts:
early view and selection effect, managing to provide some evidence of the positive
effect of both.
Scopus is used as the database and CiteScore as the measure of journal impact. The
prevalence of both publication modalities is quantified. Differences in the average
impact factor of group of journals, according to their publication modalities, are tested.
The evolution over time of the citation influence, from 2011 to 2016, is also analysed.
Finally, a linear regression to explain the correlation of these publication practices with
the CiteScore in 2016, in a ceteris paribus context, is estimated.
Our main findings show evidence of a positive correlation between average journal
impact and advancing the publication of accepted articles, moreover this correlation
increases over time. The open access modality, in a ceteris paribus context, also
correlates positively with average journal impact.
Keywords: early view; in-press articles; online first; open access; citation advantage;
CiteScore; impact factor
1. Introduction
The communication of research findings has benefited greatly from the emergence of
the Internet, and especially from two publication practices during the past decade. The
first practice is the publication of documents under the Article in Press (AIP) modality,
3
also known as online first. The AIP practice allows access to the documents from the
moment the version is accepted by the editor, and before its formal publication in an
issue. The second practice is publication under the Open Access (OA) modality,
allowing readers freely available access to the documents. With these publication
modalities publishers and authors aim to increase the visibility, usage, and citation
impact of the document. However, to date no strong evidence has been found as to
whether such modalities do in fact have the desired effect.
Open Access was defined in 2002 by Budapest Open Access Initiative as free and
unrestricted access on the public Internet to literature that scholars provide without
expectation of direct payment (Prosser, 2003). Open access is not limited to just two
modalities, though the gold and green are the main roads. Gold OA refers to articles in
fully accessible OA journals, and green OA refers to publishing in a traditional journal
in addition to self-archiving the preprint or postprint paper in a repository (Harnad et al.,
2004). With respect to gold OA, the Directory of Open Access Journals (DOAJ) is
currently the largest index presenting quality controls. According to the DOAJ list, in
April 2019 there were 12,980 OA journals, of which 9,507 were totally free and 3,415
had article processing charges (APC). There was no available information about the
possible existence of APC for 58 journals.
The OA research has produced a significant body of literature. For recent and detailed
reviews, see Tennant et al. (2016) and McKiernan et al. (2016). And for a definition of
OA and its subtypes, to assess the prevalence of OA and to examine the relative impact
of OA citations, see Piwowar et al. (2018). Here, instead, we briefly review the
literature on the OA citation advantage.
Some researchers, starting with Lawrence (2001), have found that OA articles tend to
have more citations than pay-for-access articles. This OA citation advantage has been
observed in a variety of academic fields including computer science (Lawrence, 2001),
philosophy, political science, electrical & electronic engineering, and mathematics
(Antelman, 2004), physics (Harnad et al., 2004), biology and chemistry (Eysenbach,
2006), civil engineering (Koler-Povh, Južnič & Turk, 2014), as well as biology and
medicine (Li et al., 2018).
However, there is no general agreement in the literature about the OA citation
advantage (Davis et al., 2008; Norris, Oppenheim & Rowland, 2008; Joint, 2009;
Gargouri et al., 2010; Wang et al., 2015; Dorta-González, González-Betancor & Dorta-
4
González, 2017; Dorta-González & Santana-Jiménez, 2018). Furthermore, some authors
are critical of the causal link between OA and higher citations, stating that the benefits
of OA are uncertain and vary among different fields (Craig et al., 2007; Davis &
Walters, 2011).
Kurtz et al. (2005), and subsequently other authors (Craig et al., 2007; Moed, 2007;
Davis et al., 2008), set out three postulates supporting the existence of a correlation
between OA and increased citations: (i) OA articles are easier to obtain; and therefore
easier to read and cite (Open Access postulate); (ii) OA articles tend to be available
online prior to their publication and therefore begin accumulating citations earlier than
pay-for-access articles (Early View postulate); and (iii) more prominent authors are
more likely to provide OA to their articles, and authors are more likely to provide OA to
their highest quality articles (Selection Bias postulate). Moreover, these authors
conclude that early view and selection bias effects are the main factors behind this
correlation.
Gaule & Maystre (2011) and Niyazov et al. (2016) found evidence of Selection Bias in
OA, but still estimated a statistically significant citation advantage even after
controlling for that bias. However, Björk and Solomon (2012) argued that the
distribution model is not related to journal impact. This result was confirmed by
Solomon, Laakso & Björk (2013), who concluded that articles are cited at a similar rate
regardless of the distribution model.
The Early View postulate is also related to the publication of AIPs and self-archiving by
authors. Nowadays, many journals post accepted articles online before they are formally
published in an issue (in-press articles). The overall publication delay (the time between
submission and publication) negatively influences citations (Luwel & Moed, 1998; Yu,
Wang & Yu, 2005; Tort, Targino & Amaral, 2012; Shi, Rousseau, Yang & Li, 2017).
Conversely, advance the publication of in-press articles increases citations (Alves-Silva
et al., 2016; Al & Soydal, 2017; Echeverría, Stuart, & Cordón-García, 2017). Thus,
many publishers provide access to in-press articles to minimize the effect of publication
delays and potentially increase citations. This is the case, among others, of Elsevier
(articles in press), Nature Publishing Group (advance online publication), Springer
(online first), Taylor & Francis (latest articles), and Wiley (early view).
Some authors archive the preprint or postprint (final draft after peer review) of their
articles to OA repositories or share them via social networks before they are available
5
on the publisher’s website. This strategy reduces the effect of publication delays and
makes it more likely that an article will be cited before it is formally published.
According to the SHERPA/RoMEO database of publishers' policies on copyright and
self-archiving, 81% of publishers formally allow some form of self-archiving (statistics
for the 2,561 publishers in the database; accessed April, 2019).
Therefore, open access and in-press access are two increasingly important phenomena
that need to be investigated, especially in terms of any possible interrelationships
between the two modalities. One of the limitations seen in the related literature, whether
supporting or refuting the citation advantage, is often the small number of analysed
articles. Another limitation concerns the concentration of a study on a particular
scientific area, especially given the well-known existence of important differences
between areas in terms of publication and citation habits (Dorta-González & Dorta-
González, 2013; Dorta-González, Dorta-González & Suárez-Vega, 2015). In view of all
the above, in this paper we conduct a large-scale analysis of the citation advantage in
both publication modalities.
Objectives and relationship between OA publishing and AIP publishing
OA publishing and AIP publishing have previously been explored in the literature
around strategies that journals can employ to increase the impact factor/citations metrics
of a journal. In an environment where publishing is slow and time for various processes
are counted in months and sometimes years, AIP publishing can be a good strategy to
compete in time with OA publishers, which are generally more efficient in time. Many
publishers now have their own megajournals (i.e., Plos One, Scientific Reports, Nature
Communications, Science Advances, SAGE Open,...). These megajournals publish
continuously so they do not have queues of documents to be published, significantly
reducing the times of publication.
The publication delay traditionally refers to the time between the acceptance of an
article and its publication and indexing in scientific databases. This delay has been
previously proposed to correlate negatively with journal impact factors (Yu, Wang &
Yu, 2005). With the development of online access, however, AIP publishing has
become commonplace. Thus, a significant fraction of publication delay now consists of
a period in which an article is available online, but has not been formally published in
print. AIP publishing increases the impact factor of a journal (Tort, Targino & Amaral,
6
2012). This is because the AIP allows journals to formally publish papers which are
'born' already with citations.
The research question about the difference in citations of open access (as published by
full OA journals) to citations in subscription journals has been covered quite extensively
(most recently by Li et al., 2018; Dorta-González & Santana-Jiménez, 2018; and Dorta-
González, González-Betancor & Dorta-González, 2017). Accordingly, the objective of
this study is focusing on the influence of early view availability to average journal
citations, and whatever synergies that can create with the open access research question.
Therefore, we address the following questions: (i) is there evidence to confirm that
advancing the publication of in-press articles improves the average impact of the
journal? (ii) is there evidence to confirm that open access improves the average journal
impact in a ceteris paribus context?
2. Methodology
The data
Part of the final dataset was downloaded directly from the Scopus website at
https://journalmetrics.scopus.com, with the rest obtained from Scopus by request.
The first file was downloaded in summer 2017 and contains information of all journals
included in Scopus for a 6-year period, starting in 2011. We restricted our analyses to
journals with CiteScore in 2016, amounting to a total of 21,529 journal titles. Each
journal is classified according to a system of 329 subject areas. The downloaded dataset
includes the following variables of interest for our research for each year from 2011 to
2016:
- Scopus ID: a unique identifier for every journal.
- CiteScore: this is a measure of average journal impact calculated by Scopus.
This indicator measures the average citations received per document published
in the journal in a window of three years (e.g. CiteScore 2016 is obtained by
dividing the number of citations in 2016 to all documents published from 2013
to 2015 by the number of documents published from 2013 to 2015).
- Open Access (OA): journals covered by Scopus are catalogued as OA if listed
in the DOAJ or the Directory of Open Access Scholarly Resources (ROAD).
7
- The ISSN as well as the e-ISSN are also given.
Scopus also offers other journal metrics in addition to CiteScore, like SJR and SNIP,
which are among the set of citation metrics offered by Scopus since 2010. However, the
recent CiteScore is a much simpler citation metric -in contrast to SJR and SNIP- that
can be easily validated. In this respect, we decided to use CiteScore, instead of a
different journal metric, because of its transparency and simplicity to users.
The second file, Scopus Source List, was updated to April 2017 and obtained by
request. It has one entry per title with information related to each publisher:
- Scopus ID (used to merge both files).
- Articles in Press (AIP): indicating whether the journal publishes accepted
articles before its official publication in an issue, and if they are considered
when computing the CiteScore.
- Coverage: indicating the years that the journal has been indexed in Scopus.
- Article language: indicating the different languages in which the title publishes
its articles.
- Branch: indicating the scientific branch to which the title belongs by grouping
the Scopus classification system into 5 main branches. Titles with more than
one subject category can be assigned to more than one branch.
- Publisher’s name.
- Publisher’s country.
New variables
In the first file we generated the variable ‘Number of subject categories’, describing the
number of subject categories to which each title is assigned (out of 329). This variable
ranges from one to thirteen and is asymmetrically distributed with a median of two.
In the second file we generated the variable ‘Number of branches’. This variable ranges
from one to four and is asymmetrically distributed with a median of one.
We also generated a variable called ‘Number of languages’, as some journals publish
their articles in different languages while others publish all their papers in just one. The
maximum number of editing languages of these journals is six.
We also recoded the variable ‘Publisher’s Name’, given the excessively large number of
categories that the original variable had (11,387). First, we generated a new variable of
8
number of journals per publisher. This variable had a wide range, showing publishers
with just one title and others with more than 2,000. Thus, we decided to recode the
original variable into a different one called ‘Publisher’, with one category for each well-
known publisher (Elsevier, Emerald, SAGE, Springer Nature, Taylor & Francis, and
Wiley-Blackwell) -all with more than 350 titles- and all other publishers -all with less
than 350 titles- grouped into a single category called ‘Others’.
We also decided to recode the variable ‘Publisher’s Country’ for the same reason. We
grouped all countries into continents excluding the four major publishing countries
which were given their own category: Germany, Netherlands, United Kingdom, and the
United States.
Finally, we decided to differentiate certain characteristics for all publishers. We
generated one variable showing the percentage of OA journals for each publisher, and
another showing the percentage of journals that publish in-press articles for each
publisher.
Merged dataset
After generating all the variables we merged both datasets. The final combined dataset
includes all variables that could explain -and be correlated with- the journal average
impact, measured in terms of its CiteScore:
- Articles in Press (AIP): dichotomous variable that indicates if the journal
publishes articles in press or not (0 = No; 1 = Yes).
- Open Access (OA): dichotomous variable that indicates if it is an OA journal or
not, i.e., registered in DOAJ and/or ROAD by April 2017 (0 = No; 1 = Yes).
- e-ISSN: dichotomous variable that indicates if the journal is accessible on-line
or not (0 = No; 1 = Yes).
- Indexed year: discrete variable that indicates the first year the journal was
indexed in the Scopus database.
- Number of languages: discrete variable for the number of editing languages in
the journal.
- English/Spanish/Chinese: dichotomous variables that indicate if the journal
articles are edited in English/Spanish/Chinese or not (0 = No; 1 = Yes).
9
- Number of subject categories/branches: discrete variable that describes the
number of subject categories/branches the journal is included under.
- Branch: categorical variable that indicates the branch in which the journal is
included (Health Sciences, Life Sciences, Physical Sciences, Social Sciences,
General). Journals assigned to more than one branch were included under the
category “Variety of branches”.
The final database also contains variables related to the journal’s publisher:
- Country: categorical variable that indicates the publisher’s geographical area
(Africa, America, Asia, Europe, Oceania, Germany, Netherlands, United
Kingdom, and the United States).
- Publisher: categorical variable that indicates the name of the publisher (Elsevier,
Emerald, SAGE, Springer Nature, Taylor & Francis, Wiley-Blackwell, Others).
- Percentage of AIP/OA journals: continuous variable that indicates the amount of
AIP/OA journals in relation to the total amount of journals of each publisher.
3. Results and Discussion
In this section we quantify both publication modalities (AIP and OA) and perform a
bivariate analysis to test hypotheses on the relationship between the CiteScore and the
AIP and OA variables, differentiating among groups by geographical area or language.
Finally, we estimate a linear regression to explain the correlation between all considered
variables and the CiteScore 2016, in a ceteris paribus context.
Prevalence of Articles in Press and Open Access
The prevalence of the AIP and gold OA publication modalities was quantified (Table
1). One out of every two journals (10,475) indexed in the Scopus database in 2016 used
one or other of these publication modalities. To be more precise, 37% (7,953) of all
journals published under the AIP practice, while 16% (3,522) of all journals used the
gold OA modality. Only 2.3% (500) of all journals used both publication practices.
10
Table 1. Prevalence of publication modalities AIP and OA in number of journals 2016
Articles in Press
No Yes Total
Open Access
No 10,554 7,453 18,007
Yes 3,022 500 3,522
Total 13,576 7,953 21,529
Source: Scopus
The CiteScore difference of journals according to whether or not they publish Articles
in Press is statistically significant, showing a great advantage -more than double- in
favour of journals that do publish AIP (Table 2). On the contrary, the CiteScore
difference of journals according to whether or not they are OA journals shows a slight
statistically significant advantage in favour of journals that do not use the gold OA
modality. Both differences are statistically significant, although the former is
considerably more pronounced than the latter. Thus, it seems that open access journals
(gold OA) have lower CiteScores than non-OA journals (Table 2, second column), and
publishing AIP means having higher CiteScores than not publishing them (Table 2, first
column).
Table 2. Average CiteScore 2016 by publication modalities
Articles in Press Open Access
No 0.872 1.298
Yes 1.965 1.164
Source: Scopus
However, considering both types of publishing at the same time, the relation between
gold OA and CiteScore reverses. Without taking into account the AIP variable, non-OA
journals show to have a higher journal impact (Table 2). But, by splitting the sample of
journals in two groups -those that do and those that do not publish AIP-, OA journals
show to have higher impact factor in both groups (Figure 1).
11
Figure 1. Distribution of CiteScore 2016 by type of publication modalities (AIP and OA) with
its mean. Number of journals at the top of each distribution. Outliers are not shown, to keep the
Y-axis on a scale that facilitates comparison between distributions.
This could seem to be a paradox, yet it is just another example of the so called
Simpson’s paradox or Yule-Simpson effect (Yule, 1903; Simpson, 1951), which states
that a trend appears in several different groups of data but disappears or reverses when
these groups are combined. Although it could seem to be counter-intuitive, the paradox
in this case is easily explained by the total frequency of each type of journals in each
group. In the OA group of journals, very few of them publish AIP (500 out of 3,522),
which are the journals with higher impact (Figure 1). And the opposite happens in the
group of non-OA journals, where the proportion of journals that publish AIP is much
higher (7,453 out of 18,007), rising the journal impact of the whole group of non-OA
journals.
As can be seen in Figure 1, within the non-OA group, journals that use the in-press
modality have an average impact which is more than double that of journals which do
not. Likewise, within the OA group, the average impact of journals that use the in-press
modality is almost twice as high as that of journals that do not. Thus it can be concluded
that there is evidence that advance publication of accepted articles is the most important
factor in terms of citation advantage, even though this type of journals have a greater
variability.
7453
500
3022
10554
12
Bivariate analysis
For a more in-depth analysis of the relation of the publication modalities on the journal
average citation, we tested for statistically significant differences in the CiteScore over
the last six years depending on the two publication practices analysed.
Table 3. Two-sample t-tests for equality of means (OA vs. non-OA journal)
Open Access N Mean S.D. p-value 95% CI for the
diff.
CiteScore 2011 No 16,077 1.239 1.989 0.000 0.185 0.315
Yes (12.2%) 2,237 0.989 1.378
CiteScore 2012 No 16,940 1.243 2.030 0.000 0.208 0.322
Yes (13.6%) 2,675 0.978 1.264
CiteScore 2013 No 17,657 1.277 2.083 0.000 0.178 0.290
Yes (14.4%) 2,979 1.043 1.300
CiteScore 2014 No 18,176 1.231 2.006 0.000 0.118 0.229
Yes (14.8%) 3,161 1.058 1.367
CiteScore 2015 No 18,674 1.259 1.964 0.000 0.073 0.185
Yes (15.3%) 3,369 1.130 1.444
CiteScore 2016 No 18,874 1.283 1.999 0.000 0.066 0.175
Yes (15.8%) 3,535 1.163 1.399
The number of OA journals in Scopus experienced a slight but regular increase over the
6-year period, representing 12.2% of the whole dataset in 2011 and 15.8% in 2016
(Table 3). The mean CiteScore of OA journals is slightly lower than the mean CiteScore
of non-OA journals. This difference is statistically significant in every year, but
decreases with time, ranging between 0.185 and 0.315 in 2011 and decreasing to a range
between 0.066 and 0.175 in 2016. Thus, although the average impact of OA journals is
slightly lower than that of non-OA (Dorta-González et al., 2017), empirical evidence
indicates that this difference is gradually becoming smaller and tending to disappear.
The proportion of journals that publish AIPs (e.g. 35.9% in 2016, Table 4) is higher
than the proportion of OA journals (e.g. 15.8% in 2016, Table 3) in the whole dataset.
Table 4 shows that there is also a statistically significant difference in CiteScore in
favour of those journals that publish articles in press compared to those that do not. This
difference (AIP vs non-AIP) is much larger than that of Table 3 (OA vs. NOA) and has
been increasing over the last six years, ranging from -1.000 to -0.888 in 2011 and from -
1.145 to -1.036 in 2016. This reveals evidence in favour of the publication of AIPs.
13
Moreover, the citation advantage to those journals that use this publication modality
increases with time.
Table 4. Two-sample t-tests for equality of means (AIP vs. non-AIP)
Article in Press N Mean p-value 95% CI for the diff.
CiteScore 2011 No 11,136 0.838 0.000 -1.000 -0.888
Yes (39.2%) 7,178 1.782
CiteScore 2012 No 12,235 0.834 0.000 -1.048 -0.936
Yes (37.6%) 7,380 1.826
CiteScore 2013 No 13,076 0.855 0.000 -1.118 -1.002
Yes (36.6%) 7,560 1.915
CiteScore 2014 No 13,594 0.829 0.000 -1.093 -0.982
Yes (36.3%) 7,743 1.867
CiteScore 2015 No 14,135 0.863 0.000 -1.102 -0.996
Yes (35.9%) 7,908 1.912
CiteScore 2016 No 14,362 0.873 0.000 -1.145 -1.036
Yes (35.9%) 8,047 1.963
We also graphically analysed whether there were differences in the CiteScore
depending on geographical area and the publication (or not) of AIPs (
Figure 2
). In this
case, we decided to employ another central position measure, the median CiteScore
instead of the mean, in case of any doubts about the accuracy of the latter.
14
Figure 2. Distribution of CiteScore 2016 by geographical area and publication of AIP. America
refers to the rest of the continent excluding the United States, and Europe refers to the rest of
the continent excluding Germany, the Netherlands, and the United Kingdom. Outliers are not
shown, to keep the Y-axis on a scale that facilitates comparison between distributions.
In almost all the geographical areas the pattern remains the same, with the CiteScore of
journals that publish AIPs higher than the CiteScore of those that do not. With the
exception of Africa, there is a marked difference for all geographical areas.
Table 4 shows that publishing AIPs is related to having a higher CiteScore, though this
could be because the journals that publish AIPs are more likely to publish in English as
the universal language of science. Thus, we decided to test if the difference in
CiteScore, apart from publishing AIPs, could be related to the language of publication.
We therefore tested for differences in the CiteScore of journals that publish articles in
English compared to those that do not publish in that language, and according to
whether the journal publishes AIPs or not.
Table 5. Two-sample t-tests for equality of means of CiteScore 2016 (Articles in English vs.
Articles not in English)
Articles in Press English N Mean p-value 95% CI for the diff.
No No 2,072 0.248 0.000 -0.801 -0.729
Yes 11,490 1.013
Yes No 285 0.421 0.000 -1.676 -1.529
Yes 7,695 2.024
The number of journals that publish articles in English is clearly higher in both groups
(Table 5), accounting for 85% of the group of journals that do not publish AIPs and
96% of the group that do. Table 5 shows a clear statistically significant difference in
CiteScore in favour of journals that publish articles in this language in both groups. This
difference is around 0.7 to 0.8 for the first group (AIP = No), and around 1.5 to 1.7 for
the second group (AIP = Yes). This is strong evidence of something that is well known:
research that is communicated in English has more impact.
15
Linear regression
All conclusions that are derived from a bivariate analysis have to be confirmed through
a more robust estimation technique. Thus, in this section we estimate a linear regression
for the CiteScore 2016, the most updated value that we have, explained by the variables
of interest, and all other available variables that could also influence on the CiteScore
and that therefore have to be included as control variables. The results of this estimation
are shown in Table 6.
At the beginning of the table we show the variables of interest, some of which are
related to the journals and others to their publishers. After this group of variables, we
show all control variables, with some also related to the journals and some to their
publishers.
Table 6. Ordinary Least Squares estimations for CiteScore 2016
Variables Coef. t Sig. Beta Beta CI 95%
Constant 20.503 14.37 ***
Variables of interest
Related to journals
Open Access (Ref. No) 0.265 4.80 *** 0.054 (0.037 / 0.071)
Articles in Press (Ref. No) 0.537 11.62 *** 0.145 (0.125 / 0.165)
Related to publishers
Percentage of OA journals 0.001 1.74 * 0.020 (-0.011 / 0.051)
Percentage of AIP journals -0.001 -1.54 -0.023 (-0.054 / 0.008)
Control variables
Related to journals
e-ISSN (Ref. Only print) 0.079 2.51 ** 0.021 (0.006 / 0.036)
Indexed year -0.011 -14.96 *** -0.100 (-0.119 / -0.081)
Number of languages -0.153 -4.82 *** -0.035 (-0.049 / -0.021)
English (Ref. No) 0.476 10.16 *** 0.084 (0.068 / 0.100)
Spanish (Ref. No) 0.002 0.02 0.000 (-0.014 / 0.014)
Chinese (Ref. No) 0.247 2.53 ** 0.018 (0.005 / 0.031)
Number of subject categories 0.034 3.18 *** 0.023 (0.009 / 0.037)
Number of branches 0.789 18.40 *** 0.203 (0.184 / 0.222)
Branch (Ref. Variety of branches)
Health Sciences 0.899 17.40 *** 0.219 (0.193 / 0.245)
Life Sciences 1.091 14.71 *** 0.133 (0.118 / 0.148)
Physical Sciences 1.226 21.10 *** 0.290 (0.263 / 0.317)
Social Sciences 0.492 8.52 *** 0.126 (0.101 / 0.151)
16
General 1.340 6.45 *** 0.042 (0.030 / 0.054)
Related to publishers
Percentage of e-ISSN journals 0.002 3.93 *** 0.036 (0.003 / 0.069)
Country (Ref. United States)
Africa -0.776 -7.30 *** -0.049 (-0.061 / -0.037)
America -0.713 -10.24 *** -0.082 (-0.096 / -0.068)
Asia -0.857 -18.76 *** -0.154 (-0.169 / -0.139)
Europe -0.700 -15.88 *** -0.140 (-0.157 / -0.123)
Oceania -0.345 -3.45 *** -0.023 (-0.036 / -0.010)
Germany -0.570 -10.61 *** -0.081 (-0.096 / -0.066)
Netherlands -0.314 -6.06 *** -0.050 (-0.065 / -0.035)
United Kingdom -0.028 -0.79 -0.007 (-0.022 / 0.008)
Publisher (Ref. Others)
Elsevier 0.741 12.62 *** 0.124 (0.105 / 0.143)
Emerald 0.155 1.46 0.010 (-0.004 / 0.024)
SAGE 0.424 5.82 *** 0.039 (0.027 / 0.051)
Springer Nature 0.163 2.76 *** 0.025 (0.008 / 0.042)
Taylor & Francis -0.546 -9.80 *** -0.094 (-0.111 / -0.077)
Wiley-Blackwell 0.233 3.77 *** 0.032 (0.016 / 0.048)
Number of obs. 19,976
F(32, 19943) 170.62
Prob > F 0.000
R-squared 0.215
Root MSE 1.596
Note: *** Significant at 1%; ** Significant at 5%; * Significant at 10%
Starting with the variables of interest, note that the two variables related to the journals
are statistically significant at the 1% significance level and have a positive sign. This
means that those journals that are OA or publish AIPs have higher CiteScore values,
ceteris paribus. This confirms the result obtained by Li et al. (2018) also for the
CiteScore but with a different methodology for the OA citation advantage. Of these two
variables, the one with the highest influence is the one related to publishing AIPs (with
a standardized coefficient between 0.12 and 0.15). That is, the most important factor is
not being an OA journal but rather making the journal articles themselves immediately
accessible.
Of the two explanatory variables related to the publishers, only one is statistically
significant at the 10% significance level, specifically the variable related to the
17
percentage of OA journals of the publisher. The positive sign of the estimation shows
that the higher the percentage of OA journals the higher the CiteScore.
The control variables included show some interesting features related to their
correlation with the CiteScore. Nearly all are statistically significant. Among the
variables related to journals, we can see -in a ceteris paribus context- that recent
incorporation in the Scopus index correlates negatively with CiteScore, as does
publishing articles in an increasing number of languages. On the other hand, publishing
in English -or even in Chinese- and being indexed in an increasing number of Scopus
subject categories or branches, correlates positively with CiteScore. The highest
CiteScore corresponds to the Physical Sciences branch (with a standardized coefficient
between 0.26 and 0.32) followed by the Health Sciences, Life Sciences, Social Sciences
and the General branches.
The control variables related to publishers refer to their country of publication and to
some specific publishing houses. First, it is shown that the average CiteScore in all
regions is lower than that of the United States, except for the United Kingdom which
shows no statistically significant difference compared to the average CiteScore of
journals published in the United States. It is also shown that Elsevier, SAGE, Wiley-
Blackwell, and Springer Nature have a positive and statistically significant coefficient,
which means that their average publication’s CiteScore is higher than that of other
publishers, ceteris paribus. In contrast, Taylor & Francis publications have a negative
and statistically significant coefficient, which means that its average publication’s
CiteScore is lower than that of other publishers, ceteris paribus.
4. Conclusion
The communication of research findings has benefited greatly from the emergence of
the Internet and especially from some publication practices that have their origins in the
widespread use of the web. Nowadays, many journals post accepted articles online
before they are formally published in an issue (in-press articles). Moreover, a growing
number of journals are making their articles available free of charge (gold open access).
Both publication practices aim to increase visibility, usage, and citation impact.
The main objective of this paper was to search for evidence to confirm or refute the
hypothesis that advancing the publication of in-press articles, or being an OA journal,
18
improves the average impact of a journal. With this aim, the Scopus dataset was
statistically analysed and the following conclusions drawn:
First, the data reveal evidence that shows the highest important relation to
citation advantage is advance publication of accepted articles.
Second, although the average impact of OA journals is slightly lower than that
of non-OA journals, the data indicate that this difference is gradually
diminishing and tending to disappear.
Third, the citation advantage of journals that publish articles in press is
increasing with time.
Finally, using a set of control variables related to language, categories, branches,
geographical location and publisher, it was found that those journals that are
either OA or publish articles in press have higher CiteScore values in a ceteris
paribus context. Moreover, publishing articles in press relate stronger to the
impact factor of the journal than being OA.
Another collateral conclusion of the research supports the well-known axiom related to
the language of scientific publication, showing that research that is communicated in
English has more impact.
Finally, after controlling for a set of control variables, the highest average impact factor
in 2016 was found to be for Elsevier publications.
Our study does have certain limitations. One first limitation is the impact of self-
archiving on the NOA journals. This is because we work at journal level. Paywalled
journals often allow authors to deposit preprint or postprint versions of the paper in
repositories. According to the SHERPA/RoMEO database of publishers' policies on
copyright and self-archiving, 81% of publishers formally allow some form of self-
archiving. In this way, paywalled journals benefit from their subscriber network and at
the same time from the efforts of many authors who may facilitate access via the green
route. In order to control for the self-archiving, we suggest further analyses at article
level in future.
A second limitation of our study is the data source. It is based on a database with
selective coverage. Scopus tends to feature top international journals rather than lower
impact or more nationally oriented journals. This is the case in all research areas but is
especially true in the Arts & Humanities and Social Sciences categories. In order to
19
control more nationally oriented journals, we suggest further analyses with other data
source as Google Scholar in future.
A third limitation is the impact of publishing cost on the OA journals. We do not take
into account the influence of article processing charge (APC) costs. The APCs of top-
ranked journals are evidently higher than those of lower ranked ones. For this reason,
many authors cannot publish in their preferred gold OA journals, especially the top-
ranked ones. In order to control for these costs, we suggest further analyses with data
sources that include this information in future.
Our results have managerial implications to stakeholders of journals such as publishers,
editors, and authors. The early view is beneficial -both for the publisher and the authors-
, as by means of an early distribution they may obtain also an increase in the number of
citations. In the case of authors, they are increasingly demanding to make their results
accessible in the shortest possible time. In the case of publishers, nowadays there are no
actual reasons to delay the publication date of an accepted article, just to make it
coincide with a set of articles in a particular journal number. This may be an
anachronism of the physical distribution era that has no sense in the digital era anymore.
Therefore, AIP publishing is a strategy that journals can employ to increase the impact
factor/citations metrics of a journal, as the early view allows journals to formally
publish papers which afterwards will 'born' already with citations. In an environment
where publishing times are slow and publication processes are counted in months -and
sometimes years-, AIP publishing turns to be a good strategy to be more efficient and
competitive.
As a final recommendation, publishers should make articles accessible on a continuous
basis, once the version accepted by the editor is available, based on the results of this
study, which provides empirical evidence of a citation advantage in this strategy and
confirms previous results (Tort, Targino & Amaral, 2012; Yu, Wang & Yu, 2005). In
fact, many journals ‘born’ as open access (e.g., Plos One, Scientific Reports, Nature
Communications, Science Advances, SAGE Open) employ this system of continuous
publication outside the rigid structure in fixed date issues. These journals are published
continuously, so they do not have queues of documents to be published, which
significantly reduces the distribution times.
20
5. References
Al, U., & I. Soydal (2017) Publication lag and early view effects in information science
journals. Aslib Journal of Information Management, 69(2), pp. 118–130.
Alves-Silva, E., A.C.F. Porto, C. Firmino, H.V. Silva, I. Becker, L. Resende, et al.
(2016) Are the impact factor and other variables related to publishing time in ecology
journals? Scientometrics, 108(3), pp. 1445–1453.
Antelman, K. (2004) Do open-access articles have a greater research impact? College &
Research Libraries, 65(5), pp. 372–382.
Björk, B.C., & D. Solomon (2012) Open access versus subscription journals: A
comparison of scientific impact. BMC Medicine, 10(1), pp. 73–73.
Craig, I.D., A.M. Plume, M.E. McVeigh, J. Pringle, & M. Amin (2007) Do open access
articles have greater citation impact? A critical review of the literature. Journal of
Informetrics, 1(3), pp. 239–248.
Davis, P.M., B.V. Lewenstein, D.H. Simon, J.G. Booth, & M.J. Connolly (2008) Open
access publishing, article downloads, and citations: Randomised controlled trial. British
Medical Journal, 337(7665), pp. 343–345.
Davis, P.M., & W.H. Walters (2011) The impact of free access to the scientific
literature: a review of recent research. Journal of the Medical Library Association,
99(3), pp. 208–208.
Dorta-González, P., & M.I. Dorta-González (2013) Comparing journals from different
fields of science and social science through a JCR subject categories normalized impact
factor. Scientometrics, 95(2), pp. 645–672.
Dorta-González, P., M.I. Dorta-González, & R. Suárez-Vega (2015) An approach to the
author citation potential: Measures of scientific performance which are invariant across
scientific fields. Scientometrics, 102(2), pp. 1467–1496.
Dorta-González, P., S.M. González-Betancor, & M.I. Dorta-González (2017)
Reconsidering the gold open access citation advantage postulate in a multidisciplinary
context: an analysis of the subject categories in the Web of Science database 2009-
2014. Scientometrics, 112(2), pp. 877–901.
21
Dorta-González, P., & Y. Santana-Jiménez (2018). Prevalence and citation advantage of
gold open access in the subject areas of the Scopus database. Research Evaluation,
27(1), 1–15.
Echeverría, M., D. Stuart, & J. Cordón-García, J.-A. (2017) The influence of online
posting dates on the bibliometric indicators of scientific articles. Revista Española de
Documentación Científica, 40(3), e183.
Eysenbach, G. (2006) Citation advantage of open access articles. PLoS Biology, 4(5),
e157.
Gargouri, Y., C. Hajjem, V. Larivière, Y. Gingras, L. Carr, T. Brody, et al. (2010) Self-
selected or mandated, open access increases citation impact for higher quality research.
PLoS ONE, 5(10), e13636.
Gaule, P., & N. Maystre (2011) Getting cited: does open access help? Research Policy,
40(10), pp. 1332–1338.
Harnad, S., & T. Brody (2004) Comparing the impact of open access (OA) vs. non-OA
articles in the same journals. D-Lib Magazine, 10(6).
Harnad, S., T. Brody, F. Vallières, L. Carr, S. Hitchcock, Y. Gingras, et al. (2004) The
access/impact problem and the green and gold roads to open access. Serials Review,
30(4), pp. 310–314.
Joint, N. (2009) The Antaeus column: does the “open access” advantage exist? A
librarian’s perspective. Library Review, 58(7), pp. 477–481.
Koler-Povh, T., P. Južnič, & G. Turk (2014) Impact of open access on citation of
scholarly publications in the field of civil engineering. Scientometrics, 98(1033), pp.
1033–1045.
Kurtz, M.J., G. Eichhorn, A. Accomazzi, C. Grant, M. Demleitner, E. Henneken, et al.
(2005) The effect of use and access on citations. Information Processing &
Management, 41(6), pp. 1395–1402.
Lawrence, S. (2001) Free online availability substantially increases a paper’s impact.
Nature, 411(6837), pp. 521–521.
Li, Y., C. Wu, E. Yan, & K. Li (2018) Will open access increase journal CiteScores?
An empirical investigation over multiple disciplines. PLoS ONE, 13(8): e0201885.
22
Luwel, M., & H.F. Moed (1998) Publication delays in the science field and their
relationship to the ageing of scientific literature. Scientometrics, 41(1–2), pp. 29–40.
McKiernan, E., P. Bourne, C. Brown, S. Buck, A. Kenall, J. Lin, D. McDougall, B.A.
Nosek, K. Ram, C.K. Soderberg, J.R. Spies, A. Updegrove, K.H. Woo, & T. Yarkoni
(2016) How open science helps researchers succeed. eLife, 5:e16800.
Moed, H.F. (2007) The effect of open access on citation impact: An analysis of ArXiv's
condensed matter section. Journal of the American Society for Information Science and
Technology, 58(13), pp. 2047–2054.
Niyazov, Y., C. Vogel, R. Price, B. Lund, D. Judd, A. Akil, et al. (2016) Open access
meets discoverability: Citations to articles posted to Academia.edu. PLoS ONE, 11(2),
e0148257.
Norris, M., C. Oppenheim, & F. Rowland (2008) The citation advantage of open-access
articles. Journal of the American Society for Information Science and Technology,
59(12), pp. 1963–1972.
Piwowar, H., J. Priem, V. Larivière, J.P. Alperin, L. Matthias, B. Norlander, A. Farley,
J. West, & S. Haustein (2018) The state of OA: a large-scale analysis of the prevalence
and impact of Open Access articles. PeerJ, 6:e4375.
Prosser, D. (2003) Institutional repositories and open access: The future of scholarly
communication. Information Services and Use, 23(2), pp. 167–170.
Shi, D., R. Rousseau, L. Yang, & J. Li (2017) A journal’s impact factor is influenced by
changes in publication delays of citing journals. Journal of the Association for
Information Science and Technology, 68(3), pp. 780–789.
Simpson, E.H. (1951) The Interpretation of Interaction in Contingency Tables. Journal
of the Royal Statistical Society: Series B (Methodological), 13(2), pp. 238–241.
Solomon, D.J., M. Laakso, & B.C. Björk (2013) A longitudinal comparison of citation
rates and growth among open access journals. Journal of Informetrics, 7(3), pp. 642–
650.
Tennant, J.P., F. Waldner, D.C. Jacques, P. Masuzzo, L.B. Collister, C.H. Hartgerink
(2016) The academic, economic and societal impacts of Open Access: an evidence-
based review. F1000 Research, 5:632.
23
Tort, A.B.L., Z.H. Targino, & O.B. Amaral (2012) Rising publication delays inflate
journal impact factors. PLoS ONE, 7(12), e53374.
Wang, X., C. Liu, W. Mao, & Z. Fang (2015) The open access advantage considering
citation, article usage and social media attention. Scientometrics, 103(2), pp. 555–564.
Yu, G., X.-H. Wang, & D.-R. Yu (2005) The influence of publication delays on impact
factors. Scientometrics, 64(2), pp. 235–246.
Yule, G.U. (1903) Notes on the Theory of Association of Attributes in Statistics.
Biometrika, 2(2), pp. 121–134.
