You are on page 1of 20

Journal of Marketing Analytics

https://doi.org/10.1057/s41270-020-00081-9

ORIGINAL ARTICLE

Why and how to merge Scopus and Web of Science during bibliometric
analysis: the case of sales force literature from 1912 to 2019
Saïd Echchakoui1

Revised: 14 August 2019


© Springer Nature Limited 2020

Abstract
To conduct a bibliometric analysis, several researchers retrieve publications from Scopus or/and Web of Science (WOS)
databases. When these scholars consider both Scopus and WOS databases, they mostly make two bibliometric analysis: one
from Scopus database and the other from WOS database. A few researchers merge the two databases to conduct a single
analysis, but they do not specify how they did it. This paper aims to advance the bibliometric analysis by addressing two
points. First, this research claims that making a bibliometric analysis that takes information from Scopus or/and separately
from WOS cannot give a broader view of knowledge and tendencies in a field. To prove this claim, we retrieve papers from
Scopus and WOS databases to make a bibliometric analysis of sales force literature that covers from 1912 to 2019. Results
show that there are many disparities between WOS and merged database, and between this latter and WOS database regarding
bibliometric analyses, especially among primary productive authors, the most influential papers, and keyword occurrences.
Second, this research proposes a four-step procedure that merges these two databases to allow more reliable bibliometric
analyses. This procedure was explicitly shown by using the bibliometric analysis of sales force literature during 1912–2019.

Keywords Bibliometric analysis · Scopus · WOS · Merged database · Sales force literature · R package · VOSviewer

Introduction Such analyses have been used to assess collaborative econ-


omy (Ertz and Leblanc-Proulx 2019), Industry 4.0 (Muhuri
The definition of bibliometric analysis has expanded since et al. 2019), tourism and sustainability (Garrigos-Simon
its introduction. In 1969, it referred to “the application of et al. 2018), and creativity in business economics (Castillo-
mathematical and statistical methods to books and other Vergara et al. 2018) literature.
media of communication” (Pritchard 1969, p. 348), but a Bibliometric analysis clusters into three techniques (Fab-
contemporary definition from Muhuri et al. (2019) suggests regat-Aibar et al. 2019, Hall 2011). In the first, the evaluative
that it represents an analysis method of research structure technique, analysis focuses on three measures—influence
and tendencies in a body of literature. Ellegaard and Wallin (e.g., citation numbers), productivity (e.g., publication num-
(2015) argue that bibliometric analysis can be employed to bers), and hybrid (e.g., h-index), which includes both influ-
categorize aspects of science, such as the most contributing ence and productivity (Hall 2011). The second, the relational
authors, journals, institutions, universities, and countries. technique, identifies networks among publications, journals,
Bibliometric analysis is, thus, paramount to studying lit- or authors (Fabregat-Aibar et al. 2019). The third, called the
erature tendencies and the state of knowledge across fields. review technique, is the classical approach, during which
researchers conduct systematic literature reviews based on
Electronic supplementary material The online version of this the literature or meta-analyses (Fabregat-Aibar et al. 2019).
article (https​://doi.org/10.1057/s4127​0-020-00081​-9) contains We focus on the first and second techniques—evaluative and
supplementary material, which is available to authorized users. relational.
To conduct bibliometric analyses, researchers retrieve
* Saïd Echchakoui
said_echchakoui@uqar.ca publications primarily from the Web of Science (WOS)
and Scopus databases. Some authors assess only one or the
1
Management Department, University of Quebec other, and only a few use both. When researchers use both
at Rimouski, 1595 Boulevard Alphonse‑Desjardins, Lévis, databases during bibliometric analysis, they analyze them
QC G6V 0A6, Canada

Vol.:(0123456789)
S. Echchakoui

separately because it is difficult to merge the two manually general characteristics and coverage (Norris and Oppenheim
due to disparate tag fields. For example, author citations in 2007; Archambault et al. 2009), citation-based rankings
Scopus are represented by the “Cited-By” tag field, which is and citation counts (Harzing and Alakangas 2016), journal
equivalent to the “Cited-Reference” tag in WOS. Thus, even classification methods (Wang and Waltman 2016), keyword
if a researcher retrieves publications from the two databases assignation to articles (Bartol and Mackiewicz-Talarczyk
using the same library format as the bib file, it is difficult to 2015), and Meyer’s Index, which compares the singularity
merge the two to conduct unique analyses, especially when of a database (Sánchez et al. 2017). Based on these compari-
the databases are large. A few researchers merge the two sons, we describe singularities and correlations between the
databases to conduct a single analysis, but they do not spec- two databases.
ify how they did it (Ertz and Leblanc-Proulx 2019), which is
important because if the merged file method is unsuitable, all WOS and Scopus singularities
subsequent bibliometric analyses are likely inappropriate. To
limit this problem, we conduct bibliometric analysis of sales Comparisons between WOS and Scopus have revealed
force literature by retrieving information from both WOS advantages, or singularities, of WOS. Goodman and Deis
and Scopus. We then conduct three bibliometric analyses (2007) show that WOS providing greater coverage across
using Scopus, WOS, and a merged database, the purpose of years, and Archambault et al. (2006) argue that WOS cov-
which is demonstrating disparities among results between ers western, English-language journals well. Chirici (2012)
MDB and Scopus bibliometric analysis as well as between demonstrates that across numerous disciplines, WOS offers a
MDB and WOS. varied range of information. Disadvantages of WOS regard-
This paper makes two main contributions to the litera- ing coverage in comparison to Scopus were used by some
ture based on the bibliometric analysis of sales force lit- authors to espouse advantages of WOS. Bradford’s Law sug-
erature that covers 1912 to 2019. First, we show that even gests that about 90% of significant literature in an area is
if there are correlations between Scopus and WOS, biblio- found in only a few journals (Garfield 1971). Zyoud et al.
metric analysis results are different from considering the (2017), thus, argue that WOS including the most reliable,
merged database. Specifically, this study shows that among high-impact scientific studies.
ten commonly used bibliometric analysis (e.g., productive The major advantage of Scopus is its greater publication
authors, frequency words), there are seven (70%) disparities coverage. Sánchez et al. (2017) argue that Scopus covers
between these databases (MDB and Scopus) and 80% differ- wine tourism literature better than WOS doing, reporting
ences between MDB and WOS databases. Second, although that regarding unique documents, WOS contains 35% of
many bibliometric analyses appear in the literature that use journals and 40% of articles, and Scopus includes 59% of
Scopus and WOS, none demonstrate how to conduct appro- journals and 63% of articles. Mingers and Lipitakis (2010)
priate analyses using the two databases. So, in this research, similarly suggest that WOS does not cover publications in
we propose a procedure containing four steps to perform administration and business management literature well, and
a merged database that combines Scopus and WOS and Goodman and Deis (2007) argue that Scopus offers superior
removes duplicate publications. journal selection. Another advantage of Scopus is that in
addition to articles, the database contains other documents
such as conference proceedings and books (Fingerman
Comparison overview between WOS 2006). De Moya-Anegón et al. (2007) argue that Scopus
and Scopus covers publications from developing countries better than
WOS does, suggesting that Scopus also includes more non-
Previously called Web of Knowledge, WOS is a group of English-language publications.
databases managed by Thomson Reuters Institute of Sci-
entific Information (ISI) that includes a collection of data- WOS and Scopus correlation
bases that cover such topics as science, social sciences, and
humanities (i.e., Core Collection), computing (i.e., Inspec), Although there are many singularities for each database,
and biology (i.e., Biosis). WOS was traditionally the exclu- there is also a correlation between them. Archambault
sive and largest accessible database for bibliometric analysis, et al. (2009) illustrate that although the two databases dif-
but since its introduction by Elsevier and its ease of acces- fer in terms of scope and policy coverage, they correlate
sibility in universities worldwide, Scopus became a major highly. Gavel and Iselid (2008) argue that WOS contained
competitor of WOS for conducting such analyses. The ques- 54% of Scopus’ journal titles, and Scopus covers 84% of
tion of which to use during analysis has, thus, progressively WOS’ titles. Sánchez et al. (2017) found a high correlation
become an important topic in the literature. Researchers use ­(R2 = 0.78) between WOS and Scopus regarding the num-
various indicators to compare the two databases, including ber of articles. They also report that researchers have found
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Fig. 1  Annual publication in sales force literature in Scopus, WOS, and merged databases

high average correlations between the two databases in areas four steps—identification of studies, screening, eligibility,
such as the social sciences (23%), health sciences (32%), and and inclusion.
physical sciences (30%). During identification, we used several terms to retrieve
Comparisons between WOS and Scopus regarding con- publications from both WOS and Scopus, including “sales-
ducting bibliometric analyses do not suggest superiority of person” OR “salesman” OR “salespeople” OR “sales force”
one over the other (Sánchez et al. 2017). We, thus, assume OR “salesforce”. Such identification yielded 17,365 and
that both databases are important, arguing that to conduct 11,431 publications in Scopus and WOS, respectively. Two
appropriate bibliometric analyses, it is best to use both substages were used during screening. During the first stage,
because one completes the other. Escalona et al. (2010) we removed duplicate publications from each database, and
argue that WOS and Scopus are complementary, and Mon- during the second, we used screening criteria on both data-
geon and Paul-Hus (2016) suggest that both should be used. bases. We retained only English-language publications and
Since they are complements and a correlation of publication excluded document types “book series” OR “books” OR
coverage is evident between the two databases, the question “trade publications” OR “undefined”. We, thus, retained
of how to merge them into one database is paramount. 13,273 studies in Scopus and 6884 in WOS. During the
eligibility step, we excluded publications unrelated to sales
force literature. For example, some articles associated with
Method to merge WOS and Scopus salesforce software and others generally were unrelated to
for bibliometric analysis sales force literature. We also removed publications with-
out International Standard Serial Numbers (ISSN) because
The first step to merging the two databases is preparing them we examined only publications from recognized sources.
for a topic of study (e.g., sales force literature). During the inclusion step, we exported each database to the
Endnote Library (Fig. 1). We created two composite libraries
WOS and Scopus database creation in Endnote. The first, labeled WOS, contained 5469 pub-
lications, and the second, labeled Scopus, included 7540
To construct both WOS and Scopus databases for sales publications. We did not remove publications between
force bibliometric analysis, we used the Preferred Report- WOS and Scopus databases; we conducted this operation
ing Items for Systematic Reviews and Meta-Analyses later because we needed to conduct bibliometric analyses
(PRISMA) method introduced by Moher et al. (2009), a on each database before merging them. In a subsequent sec-
method used commonly to perform bibliometric analyses. tion, we demonstrate how to remove duplicate publications
To select publications from a database, PRISMA includes using Excel VBA.
S. Echchakoui

Table 1  Main information of sales force publications in Scopus, WOS, and merged databases
Step Objective Procedures Action

Step 1 Convert WOS and Scopus databases to Convert each database separately from Select “BibTex Export” library from Endnote
WOS and Scopus to bibliography files. Endnote to “.bib” files style as shown in Annex 1.
Change the Endnote’s BibTex bibliography Use Template in Annex 2.
template.
Step 2 Convert both WOS.bib and Scopus.bib to Use Rstudio (or R) by loading the biblio- From R software do:
“bibtex” files. metrix package. > install.packages(“bibliometrix”)
>library(bibliometrix)
Follow the tutorial method in the following
link made by Aria and Cuccurullo (2017):
https​://cran.r-proje​ct.org/web/packa​ges/bibli​
ometr​ix/vigne​ttes/bibli​ometr​ix-vigne​tte.
html
Or, for non-coders, use the biblioshiny From R software do:
library. >install.packages(“bibliometrix”)
>library(bibliometrix)
>biblioshiny()
Step 3 Format the WOS and the Scopus bibtex Format the WOS.bib and the Scopus.bib 1. Used a Word VBA macro in Annex 3 to
files to have the same tag fields. files. change some field tag.
Format the Scopus.bib files. 2. Use VBA excel code in Annex 3 to remove
duplicate references between Scopus and
WOS bibliography.
Step 4 Merge the two databases in R or Rstudio Convert the WOS.bib and the Scopus. From R software do:
and remove duplicates. bib files to WOS.xlsx and Scopus.xlsx, > write_csv (M1, “…/path/WOS.csv”)
respectively, in Rstudio. > write_csv (M2, “…/path/Scopus.csv”)
M1: is the data frame including WOS’s
bibtex bibliography.
M2: is the data frame including Scopus’s
bibtex bibliography.
Path: is the file path I which you need to
store the WOS.csv and Scopus.csv.
Copy one file to the other manually. In Excel, paste for example WOS.xlsx data to
Scopus.xlsx data.
Remove duplications. Use VBA code to remove duplication in
Excel (see Annex 3).

Merging WOS and Scopus for bibliometric analysis shown in Annex 2. The purpose of this step is to export both
WOS and Scopus databases with the same tag fields.
To merge the two databases retained in the inclusion step of
the PRISMA method, we propose the following four-step Second step: convert both WOS.bib and Scopus.bib
procedure: to “bibtex” files

To convert both WOS.bib and Scopus.bib to “bibtex” files,


First step: convert WOS and Scopus databases we used Rstudio (or R) by loading the bibliometrix pack-
to bibliography files age, as Aria and Cuccurullo (2017) suggest. We could then
conduct bibliometric analyses for each bibtex file in Rstu-
To adequately convert imported WOS and Scopus databases dio, following Aria and Cuccurullo’s (2017) procedures and
references in Endnote to bibliography files, we make the codes in R. For non-coders, Aria and Cuccurullo (2017)
two following actions (see Table 1). First, we convert each created the biblioshiny library, which makes bibliometric
database separately from Endnote to “.bib” files (e.g., WOS. analysis based on bibtex or Excel files easy. We transformed
bib and Scopus.bib). The “.bib” is a file format employed the bibtex files for both the WOS and Scopus databases for
to process references generally in connection to LaTeX two reasons. While conducting bibliometric analyses using
documents. To do this operation, we selected “BibTex bibtex files, errors are common, and their causes are unob-
Export” library from Endnote style (see Annex 1). Second, servable. Excel files help with merging the two databases,
we changed the Endnote’s BibTex bibliography template as but before converting the bibtex files into Excel files, we
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Table 2  Main information Scopus WOS Merged


of sales force publications in
Scopus, WOS, and merged Documents 7540 5469 10,108
databases
Sources (journal article, conference proceeding) 1870 1072 2214
Keywords plus (ID) 26,236 13,913 29,432
Author keywords (DE) 25,049 13,850 28,445
Period 1912–2019 1991–2019 1912–2019
Average citations by documents 30.74 27.64 28.81
Authors 12,364 9117 15,855
Author appearances 19,896 14,939 26,965
Authors of single-authored documents 1092 591 1315
Authors of multi-authored documents 11,272 8526 14,540
Single-authored documents 1361 763 1727
Documents by author 0.61 0.6 0.638
Authors by document 1.64 1.67 1.57
Co-authors by documents 2.64 2.73 2.67
Collaboration index 1.82 1.81 1.73

advise making format changes to the bibtex files. These Results of bibliometric analysis
changes helped us format our bibtex bibliography to avoid
errors in keywords frequencies and author influence. To conduct sales force bibliometric analysis on the WOS,
Scopus, and merged databases, we considered eight analysis
Third step: Format the WOS and the Scopus bibtex files types—data information, research growth, most productive
to have the same tag fields authors, most relevant source, most influential papers and
authors, most productive country, greatest keyword fre-
To format the WOS and the Scopus bibtex files to have the quency and occurrences, and co-country collaborations.
same tag fields, we used a Word VBA macro to format the
WOS.bib and the Scopus.bib files. We employed regular Information about the data
expressions (e.g., “regex”) to position author keywords (e.g.,
“Keywords”) on one line, separated by a colon. We also Table 2 shows information on publications identified in the
extracted the tag fields “Cited-By” from the notes bibtex tag Scopus, WOS, and merged databases. The number of pub-
fields and renamed some tag fields (e.g., “Times-Cited” to lications retrieved from Scopus was 37% larger than those
“Times-Cited” and “Cited-By” to “Times-Cited”) to allow from WOS. The merged database was also unequal regarding
recognition by the Bibliometrix package (Annex 3). the sum of the Scopus and WOS databases because 22% of
the records were duplicate publications. Table 2 shows that
Fourth step: merge the two databases in R or Rstudio papers in WOS began from 1991, whereas Scopus started
and remove duplicates from 1912. The number of author keywords and Keyword
plus was nearly double in Scopus, and author appearances
The final step is to merge the two databases in R or Rstu- were lower in WOS than in Scopus (12 364 versus 9 117).
dio and remove the duplicate bibliography. To control this However, the number of authors by document (1.67 versus
process, we convert the WOS.bib and the Scopus.bib files 1.64) and of co-authors by document (2.73 versus 2.64) were
to WOS.xlsx and Scopus.xlsx, respectively, in Rstudio. We slightly greater in WOS. The collaboration index was nearly
used the following command: write_csv (M, “…/path/WOS. equal between the two databases (1.81 versus 1.829), but
csv”). Then, copy one file to the other manually (e.g., WOS. was greater than in the merged database (1.73).
xlsx to Scopus.xlsx). To remove duplicates, sort the database
based on the Title (“TI”) tag file. Use VBA code to remove Research growth
duplicates (Annex 3). We deleted 2901 duplicates, so the
merged database included 10,108 publications. The merged Figure 1 illustrates the tendency of publications in the Sco-
Excel file can then be used to analyze the merged database pus, WOS, and merged databases for the period from 1912
using the biblioshiny package in Rstudio. If we wanted to to 2018. For the merged databases (here after “MDB”), the
analyze the file in VOSviewer, we would have had to convert growth tendency declined between 2009 and 2011. How-
it to a csv file in Excel. ever, after 2011, the number of publications grew globally
S. Echchakoui

Table 3  Most productive authors in Scopus, WOS, and merged databases


Scopus WOS Merged
Authors # Articles h-index Authors # Articles h-index Authors # Articles h-index

Laporte G 77 35 Laporte G 100 40 Laporte G 125 45


Dubinsky AJ 53 22 Gendreau M 56 32 Dubinsky AJ 63 22
Wang Y 38 11 Ahearne M 28 22 Gendreau M 59 34
Ahearne M 31 25 Cordeau JF 24 17 Wang Y 39 9
Jaramillo F 31 22 Salazar Gonzalez JJ 22 12 Boles JS 35 23
Boles Js 29 21 Dubinsky AJ 21 11 Sharma A 35 20
Cravens DW 29 20 Letchford AN 21 11 Ahearne M 34 24
Chonko LB 28 19 Wang Y 21 6 Jaramillo F 33 22
Marshall GW 26 17 Woeginger GJ 21 10 Cravens DW 32 20
Ingram Tn 24 16 Fischetti M 20 16 Chonko LB 31 19
Sharma A 24 20 Sharma A 20 14 Marshall GW 31 17
Wang X 24 8 Toth P 20 14 Ingram TN 30 19
Woeginger GJ 24 13 Deineko VG 19 8 Toth P 30 18
Evans Kr 23 17 Gouveia L 19 10 Woeginger GJ 29 12
Gendreau M 23 16 Gutin G 19 10 Cordeau JF 27 20
Li J 23 9 Rapp A 19 13 Evans KR 27 17
Sager JK 23 18 Boles Js 18 12 Fischetti M 26 18
Jones E 22 18 Glover F 18 13 Moncrief WC 26 15
Moncrief WC 22 15 Agnihotri R 17 8 Pullins EB 25 13
Rapp A 22 18 Homburg C 17 12 Rapp A 25 19

#: number

until 2018. With respect to MDB, the growth of the Scopus force literature between 1912 and 2019. These authors are,
publications was the same until 2014. After this date, the respectively, Laporte G (125 publications, h-index = 45),
number of publications’ tendency of the Scopus decline in Dubinsky AJ (65 publications, h-index = 22), and Gend-
2015 and grew in 2016, and finally decrease until 2018. For reau M (59 publications, h-index = 34). Regarding MDB,
the WOS database, the tendency of publication declines Laporte G is also the most productive author in both
weakly in 2011 with respect to MDB. It also increased less Scopus and WOS databases (Scopus: 77 publications,
than MDB between 2009 and 2011. Finally, the WOS data- h-index = 35; WOS: 100 publications, h-index = 40).
base increased at a faster rate than MDB after 2012. Con- However, while Dubinsky AJ has also the second posi-
sequently, following these results, we can argue that in the tion in Scopus (53 publications, h-index = 22), he holds
case of the sales force literature, the tendency of publications the sixth position in WOS (21 publications, h-index = 11).
in Scopus and WOS databases have some similarities and On the other hand, Gendreau M who is the third contribu-
disparities with respect to MDB. We do not include 2019 tor author in MDB, takes second place in WOS with 56
because that year has not concluded. With respect to the publications and an h-index of 32, but only the fifteen
sales force literature, based on Fig. 1, we can claim that the position in Scopus (23 publications, h-index = 19). These
scholar’s interest in this field of research has globally grown findings demonstrate that results from the merged file are
in time. not influenced only by Scopus even though it is larger, but
also by WOS.
Most productive authors Figure 2 shows that the most consistently productive
author in MDB between the 1912 and 2019 in the sales force
Table 3 shows the 20 most productive authors in the Sco- literature is Laporte G, followed equally by Dubinsky AJ and
pus, WOS, and MDB, a classification based on the number Cravens DW (see Fig. 2a). With respect to MDB, Laporte
of publications and the h-index. The first criteria ranking G has the same place as Dubinsky AJ and Cravens DW in
is a number of publications, but if two or more authors the Scopus database (see Fig. 2b). About WOS database
have the same number of publications, the second crite- (see Fig. 2c), Laporte G also has the first place, Wang Y the
rion is the h-index. About MDB, Table 3 illustrates that second, and the third position is shared between two authors
three authors are the most productive scholars in the sales Dubinsky AJ and Gendreau M.
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Fig. 2  Top authors production over time in Scopus, WOS, and merged databases. a Merged database. b Scopus database. c WOS database.

According to our comparison between MDB and Sco- G), but the differences concern the second and the third
pus as well as between MDB and WOS, we argue that consistently productive authors.
there is a similarity and also differences between MDB and Based on our finding in MDB, we can claim that the most
Scopus, as well among MDM and WOS. The similarity is productive scholars in the sales force literature between 1912
mainly on the first consistently productive author (Laporte and 2019 are , respectively, Laporte G, Dubinsky AJ, and
S. Echchakoui

Table 4  Most relevant source in Scopus, WOS, and merged databases


Scopus WOS Merged
Sources # Articles Sources # Articles Sources # Articles

Journal of Personal Selling and 564 European Journal of Operational 351 Journal of Personal Selling and 588
Sales Management ­Researcha Sales Management
European Journal of Operational 278 Industrial Marketing Management 208 European Journal of Operational 437
­Researcha Research
Industrial Marketing Management 200 Computers and Operations 199 Industrial Marketing Management 274
Research
Computers and Operations 156 Journal of Business Research 115 Computers and Operations 257
Research Research
Journal of Business Research 127 Discrete Applied ­Mathematicsa 95 Journal of Business Research 159
Journal of Business and Industrial 114 Journal of Business and Industrial 95 Journal of Business and Industrial 145
Marketing Marketing Marketing
Journal of the Academy of Market- 114 Journal of the Operational 93 Journal of the Academy of Market- 137
ing Science Research Society ing Science
Operations Research Letters 101 Journal of the Academy of Market- 88 Journal of the Operational 137
ing Science Research Society
Discrete Applied Mathematics 92 Networks 85 Discrete Applied Mathematics 131
Journal of the Operational 90 Journal of Marketing 80 Operations Research Letters 124
Research Society
Networks 81 Transportation Science 76 Mathematical Programming 121
Mathematical Programming 72 Mathematical Programming 73 Networks 118
Journal of Applied Psychology 65 Journal of Personal Selling and 70 Journal of Business Ethics 90
Sales Management
International Journal of Production 62 Journal of Business Ethics 69 Transportation Science 88
Research
Expert Systems with Applications 60 Operations Research Letters 68 International Journal of Production 84
Research
Journal of Business Ethics 54 Expert Systems with Applications 54 Journal of Marketing 82
Applied Soft Computing Journal 50 Annals of Operations Research 53 Annals of Operations Research 79
Annals of Operations Research 49 Computers and Industrial Engi- 48 Expert Systems with Applications 77
neering
Information Processing Letters 48 Journal of Heuristics 47 Operations Research 74
Operations Research 48 Applied Soft Computing Journal 46 Journal of Applied Psychology 70

Wagner 2015). A Practical Comparison of Scopus and Web of Science Core Collection (https​://ubir.buffa​lo.edu/xmlui​/handl​e/10477​/38568​)
a
Table 4 shows that WOS and Scopus include different article counts for the same journal. Two factors can explain these differences. First, pub-
lications from Scopus database were available between 1912 and 2019, whereas these from WOS were between 1991 and 2019. This point can
probably explain why some journals (e.g., Journal of Personal Selling) have a greater number of papers in Scopus database. Second, it is well
known that Scopus has some weakness as author clustering problem, institution clustering problem, and random missing articles (Wagner 2015).
So, these reasons can explain why some journal (notably: European Journal of Operational Research, Computers and Operations Research, Jour-
nal of Business Ethics) have more number in WOS than Scopus. Indeed, in addition to the missing papers in Scopus, in the PRISMA procedure
steps, we delete all papers that did not have identified authors or institutions

Gendreau M. In addition, the most consistently productive publications) and the Industrial Marketing Management
author in the sales force literature is Laporte G, followed (274 publications). With respect to MDB, the most rel-
equally by Dubinsky AJ, and Cravens DW. evant source in Scopus is the same with similar orders.
However, in the WOS database, the Journal of Personal
Most relevant sources Selling and Sales Management only holds thirteenth place
(70 publications), but the Industrial Marketing Manage-
Results in Table 4 show that the most relevant source ment has a better place (second) with 208 publications.
in MDB is the Journal of Personal Selling and Sales Likewise, European Journal of Operational Research has
Management, with 588 articles, followed, respectively, also a first place (351 publications) instead the second
by the European Journal of Operational Research (437 position in MDB.
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

The above results show that there is a similarity between study the traveling salesperson problem and optimization.
MDB and Scopus databases about the most relevant sources. We, thus, argue that sales force literature is influenced
Nevertheless, there is a similarity and difference between greatly by research that assesses the latter subject.
MDB and WOS database. The similarity is about the Euro-
pean Journal of Operational Research and the Industrial Most productive country
Marketing Management. These journals are classed between
the most third influential sources in the sales force literature Analysis of the most productive countries was based on two
in both MDB and WOS databases. However, the biggest indices—an author’s affiliated country (Table 6) and total
difference concerns the position of the Journal of Personal citations by country (Table 7). About MDB, Table 6 shows
Selling and Sales Management (first in MDB versus thirteen that the USA has the greatest number of author affiliations (2
in WOS). 636). China and Canada hold the second and third positions
Based on the merged database, we can argue that the (respectively 713 and 431). By comparison to MDB, Table 5
sales force literature was influenced by the three following reveals that these three most productive countries (the USA,
journals: (1) Journal of Personal Selling and Sales Manage- China and Canada) are also the most productive countries in
ment, (2) European Journal of Operational Research, and (3) both Scopus and WOS databases. However, there are some
Industrial Marketing Management. differences if we consider the ten most productive countries.
Indeed, for example, The United Kingdom holds the fourth
Most influential papers and authors position in MDB (271 authors) but holds sixth place in WOS
with 167 author affiliations. Likewise, Japan has the sixth
Table 5 shows the most influential papers in Scopus, WOS, place in the MDB, but the seventh position in the Scopus
and the MDB from 1912 to June 2019. We constructed the database.
list based on publications with the greatest number of cita- Table 7 shows the top ten cited countries in sales force
tions (Knoke and Yang 2008; Strozzia et al. 2017). Dorigo literature from 1912 to June 2019. TheUSA is the most cited
et al. (1996), titled “Ant system: Optimization by a colony country in MDB with 98,174 citations, followed consecu-
of cooperating agents,” published in IEEE Transactions on tively by Canada (16,676 citations) and Italy (12,315 cita-
Systems, Man, and Cybernetics has the most citation in the tions). With respect to MDB, the most cited in the sales
MDB databases at 5269. The second and the third papers force literature countries in the WOS database are the same.
with the most citations in the MDB are, respectively, Dorigo However, there are two differences between MDB and Sco-
and Gambardella (1997), “Ant colonies for the traveling pus database. Indeed, despite the most cited country in the
salesman problem,” in Biosystems with 5 209 citations, and two databases being the USA (77 438 citations in Scopus),
Hopfield and Tank (1985), ““Neural” computation of deci- the second position in Scopus is Belgium with 16,072 cita-
sions in optimization problems,” in Biological Cybernetics tions, whereas Canada, which has the second position in
(3 576 citations). These most influential papers and authors MDB, holds the fourth position in Scopus with 7020 cita-
are the same between MDB and Scopus databases. Regard- tions. In addition, the third most cited country in the Scopus
ing the WOS database, the first and the second positions are is Germany (7755 citations). However, Italy which holds the
similar to MDB, but the third place is taken by Xu (2005), third position in MDB falls to the ninth place in Scopus at
“Survey of clustering algorithms,” IEEE Transactions on 4240 citations.
Neural Networks. Therefore, there is a similarity between Drawing on the merged database (see Tables 6 and 7), we
MDB and Scopus databases about the most influential claim that the most productive countries in the sales force
papers and authors. Nevertheless, there is only a difference research between 1912 and 2019 with respect to the cor-
in the third position between MDB and WOS database. responding author’s countries were successively: the USA,
Based on the MDB, we may claim that the three most China, and Canada. However, with respect to the total cita-
influential papers with authors in the sales force research tions by country, the three most influential countries were
during 1912–2019 were, respectively, (1) “Dorigo et al. the following: USA, Canada, and Italy.
(1996). Ant system: Optimization by a colony of cooperat-
ing agents IEEE Transactions on Systems, Man, and Cyber- Most frequent words
netic” with 5269 citations, (2) “Dorigo and Gambardella
(1997). Ant colonies for the traveling salesman problem. We analyze both keyword frequency and occurrences.
Biosystems” with 5209 citations, and (3) “Hopfield and Tank Table 8 shows the 20 most relevant keywords in the three
(1985). “Neural” computation of decisions in optimization databases for sales force literature from 1912 to June 2019.
problems. Biological Cybernetics” with 3576 citations. The first most frequency word is “The traveling salesper-
Only one paper, Podsakoff et al. (2000), which assesses son problem” in MDB with 4578 occurrences. The second
sales force management, is ranked in WOS. All other papers most frequency word is “Business and economic” (2081
S. Echchakoui

Table 5  Most influential papers in Scopus, WOS, and merged databases


Authors Title Source # Citations

Merged
Dorigo et al. (1996) Ant system: Optimization by a colony of coop- IEEE Transactions on Systems, Man, and 5269
erating agents Cybernetic
Dorigo and Gambardella (1997) Ant colonies for the traveling salesman prob- Biosystems 5209
lem
Hopfield and Tank (1985) “Neural” computation of decisions in optimiza- Biological Cybernetics 3576
tion problems
Geem et al. (2001) A new heuristic optimization algorithm: Har- Simulation 3096
mony search
Dorigo and Gambardella (1997) Ant colonies for the traveling salesman prob- Biosystems 2542
lem
Xu (2005) Survey of clustering algorithms IEEE Transactions on Neural Networks 2355
Černý (1985) Thermodynamical approach to the traveling Journal of Optimization Theory and Applica- 1962
salesman problem: An efficient simulation tions
algorithm
Podsakoff et al. (2000) Organizational citizenship behaviors: A critical Journal of Management 1837
review of the theoretical and empirical litera-
ture and suggestions for future research
Mladenović and Hansen (1997) Variable neighborhood search Computers and Operations Research 1528
Dorigo et al. (1999) Ant algorithms for discrete optimization Artificial Life 1396
Scopus
Dorigo et al. (1996) Ant system: Optimization by a colony of coop- IEEE Transactions on Systems, Man, and 5269
erating agents Cybernetic
Dorigo and Gambardella (1997) Ant colonies for the traveling salesman prob- Biosystems 5209
lem
Hopfield and
Tank (1985) “Neural” computation of decisions in optimiza- Biological Cybernetics 3576
tion problems
Geem et al. (2001) A new heuristic optimization algorithm: Har- Simulation 3096
mony search
Sttzle and Hoss (2000) Max–Min ant system Future Generation Computer Systems 2013
Mladenović and Hansen (1997) Variable neighborhood search Computers and Operations Research 1965
Černý (1985) Thermodynamical approach to the traveling Journal of Optimization Theory and Applica- 1962
salesman problem: An efficient simulation tions
algorithm
Dorigo et al. (1999) Ant algorithms for discrete optimization Artificial Life 1286
Reinelt (1991) The traveling salesman computational solu- Lecture Notes in Computer Science 1226
tions for TSP applications
Kirkpatrick (1984) Optimization by simulated annealing—quanti- Journal of Statistical Physics 1168
tative studies
WOS
Dorigo et al. (1996) Ant system: Optimization by a colony of coop- IEEE Transactions on Systems, Man, and 5269
erating agents Cybernetic
Dorigo and Gambardella (1997) Ant colonies for the traveling salesman prob- Biosystems 2542
lem
Xu (2005) Survey of clustering algorithms IEEE Transactions on Neural Networks 2355
Podsakoff et al. (2000) Organizational citizenship behaviors: A critical Journal of Management 1837
review of the theoretical and empirical litera-
ture and suggestions for future research
Mladenović and Hansen (1997) Variable neighborhood search Computers and Operations Research 1528
Dorigo et al. (1999) Ant algorithms for discrete optimization Artificial Life 1396
Sttzle and Hoss (2000) Max–Min ant system Future Generation Computer Systems 1345
Blum and Roli (2003) Metaheuristics in combinatorial optimization: Acm Computing Surveys 1280
Overview and conceptual comparison
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Table 5  (continued)
Authors Title Source # Citations
Aurenhammer (1991) Voronoi diagrams: A survey of a fundamental Computing Surveys 1131
geometric data structure

Table 6  Corresponding author’s Scopus WOS Merged


countries in Scopus, WOS, and
merged databases Country # Articles Country # Articles Country # Articles

USA 2 175 USA 1 128 USA 2 636


China 493 China 407 China 713
Canada 284 Canada 274 Canada 431
United Kingdom 271 Germany 242 United Kingdom 357
Germany 235 Japan 172 Germany 356
France 210 United Kingdom 167 Japan 292
Japan 196 France 166 France 287
Australia 176 Italy 157 Taiwan 221
India 160 Taiwan 137 Italy 210
Taiwan 135 Spain 116 India 207

Table 7  Total Citations by Scopus WOS Merged


country in Scopus, WOS, and
merged databases Country # Citations Country # Citations Country # Citations

USA 77 438 USA 4 4530 USA 98 174


Belgium 16 072 Canada 13 137 Canada 16 676
Germany 7 755 Italy 9 842 Italy 12 315
Canada 7 020 Belgium 5 714 Belgium 12 047
United Kingdom 6 962 China 5 292 Germany 9 879
France 6 221 Germany 5 049 United Kingdom 9 285
Australia 4 909 France 4 562 France 7 911
Korea 4 788 United Kingdom 4 470 Netherlands 5 941
Italy 4 240 Netherlands 3 739 Korea 5 403
Japan 3 312 Spain 2 715 Australia 4 949

occurrences) and the is and “Algorithm” (1732 occurrences). four clusters are distinguishable. The first cluster (green;
The first most frequency word in Scopus and WOS is also traveling salesperson problem) represents research into solv-
“The traveling salesperson problem” words with 3420 fre- ing how to choose the shortest route for a salesperson’s visits
quencies and 3035 frequencies for Scopus and WOS, respec- across cities. In this cluster are nodes such as scheduling,
tively. However, the second and the third frequency words mathematical model, graph theory, and optimization. The
differ between MDB and Scopus databases. In this latter, second cluster (blue; operations research and management
the second most frequency word is “Algorithm” with 1090 nodes) focuses on transportation and delivery, and contains
occurrences, and the third most frequency word is “optimi- nodes such as time, delivery problems, and vehicle routine
zation” (1007 occurrences). With respect to WOS database, problems. Business and economics is the largest node in the
the second most frequency words “Business and economic” third cluster (red), which contains sales force management
is the same as MDB. Nevertheless, the third most frequency themes such as salesperson performance, control systems,
word is “Computer science”1616 occurrences instead of and buyer–seller relationships. The fourth cluster includes
Algorithm in the case of MDB. Nevertheless, Computer sci- three small nodes—human, male, and female—and focuses
ence is in the third position with 1616 occurrences instead on differences between males and females in sales forces.
of Algorithm in the case of MDB. Figure 3b shows that author keywords in Scopus sepa-
Figure 3a shows the classification of sales force literature rate into four clusters. Each cluster comprises nodes (cir-
based on author keywords extracted from MDB, in which cles) with similar colors, and includes links or relationships
S. Echchakoui

Table 8  Most frequency words in Scopus, WOS, and merged databases


Scopus WOS Merged
Keyword Occurrences Keyword Occurrences Keyword Occurrences

Traveling salesperson problem 3420 Traveling salesperson problem 3035 Traveling salesperson problem 4578
Algorithm 1090 Business and economic 2081 Business and economic 2081
Optimization 1007 Computer science 1616 Algorithm 1732
Heuristic algorithm 777 Operation research and manage- 1607 Operation research and manage- 1665
ment ment
Combinatorial optimization 664 Algorithm 1201 Computer science 1620
Genetic algorithm 469 Engineering 937 Optimization 1226
Computational complexity 366 Mathematic 784 Mathematic model 1093
Human 307 Salesperson 741 Engineering 938
Scheduling 300 Optimization 659 Salesperson 914
Graph theory 299 Performance 577 Heuristic method 821
Integer programming 298 Model 530 Genetic algorithm 642
Salesperson 296 Vehicle routing problem 496 Performance 638
Vehicle routing problem 280 Heuristic algorithm 468 Vehicle routing problem 611
Mathematical model 274 Behavior 407 Model 581
Approximation algorithm 242 Genetic algorithm 398 Combinatorial optimization 521
Simulated annealing 227 System 279 Scheduling problem 382
Computer simulation 221 Search 276 Behavior 374
Ant colony optimization 208 Combinatorial optimization 272 Approximation algorithm 320
Operation research 201 Problem 268 Transportation 304
Artificial intelligence 200 Salesperson performance 260 Sales 293

(lines) between nodes. The first cluster (yellow) includes occurrence of this keyword in the merged database is 4578.
the most frequent keyword, traveling salesperson problem, This keyword is also the main node of the big cluster that
as the dominant node. Many other keywords constitute this shape the research in the sales force (see Fig. 3a).
cluster, such as integer programming, combinatorial experi-
ment, and vehicle routing problem. Heuristic algorithm is Co‑country collaboration
the largest node in the second cluster (green), which includes
other nodes such as evolutionary algorithm, genetic algo- The co-country collaboration keyword was assessed using
rithm, and learning algorithm. The third cluster (red) con- VOSviewer. Three sub-figures are shown in Fig. 4, each of
tains many small nodes such as human, neural network, con- which represents the three databases. The figures show that
trol study, and computer simulation. The last cluster (blue) the merged database had the greatest number of clusters (9),
includes the algorithm’s keyword as the largest node, and followed by Scopus (7) and WOS (4). The United States
includes nodes such as approximation algorithm, mathemati- is the largest node in the three databases, especially in the
cal model, and constraint theory. MDB and Scopus database. Some countries are related in
For WOS, Fig. 3c shows that sales force literature can MDB, but not in other databases. For example, in MDB Ger-
be divided in two clusters based on author keywords. The many and Japan belong to the same cluster (red), in which
first cluster (green) contains the traveling salesperson prob- Canada has the largest node. However, in Scopus, Germany
lem as the dominant node, and includes other nodes such and Japan do not collaborate on sales force research because
as computer science, operations research and management, each belongs to a different cluster (e.g., pink for Germany
and vehicle routing problem. Business and economics is and red for Japan). In addition, in WOS, Germany belongs
the largest node in the second cluster (red), which includes to the same cluster as the United States (red) but does not
many other nodes such as model, salesperson performance, relate to Japan (green).
strategy, and job satisfaction. These nodes primarily concern Figure 4a shows that the merged database has nine clus-
sales force management. ters. The first (green) includes Canada, Netherlands, Ger-
Based on the above analysis, we can argue that “Trave- many, and Japan, the second (blue) The United States as
ling salesperson problem’ is the most influential keyword the largest node, in Australia, Greece, and Egypt, the third
in the sales force literature between 1912 and 2019. The (red) China, India, Sweden, and others, the fourth (yellow)
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Fig. 3  Keyword Co-occurrences in Scopus, WOS, and merged databases. a Merged database. b Scopus database. c WOS database.
S. Echchakoui

Fig. 4  Co-countries collaboration in Scopus, WOS, and merged databases. a Merged database. b Scopus database.c WOS database.
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Taiwan, Hong Kong, and Indonesia, the fifth (light blue) Scopus). For example, the most three co-country collabora-
South Africa, and New Zealand, the sixth (purple) Bel- tion clusters in MDB are: (1) cluster 1 including Canada,
gium, Norway, and Chile, the seventh (pink) South Korea, Netherlands, Germany, and Japan, (2) cluster 2 containing
the eighth (brown) Finland, and the ninth (orange) Spain. United States, Australia, Greece, and Egypt, and (3) cluster
Figure 4b shows that Scopus contains seven clusters. The 3 incorporating China, India, Sweden, and others. In the
first (green) includes the United States, the Netherlands, and case of Scopus database, these three clusters are: (1) cluster
Finland, the second one (blue) France, United Kingdom, and 1 including United States, the Netherlands, and Finland, (2)
Sweden, the third (red) China, Japan, Turkey, and Russia, cluster 2 containing France, United Kingdom, and Sweden,
which replaces the Russian Federation, the fourth (yellow) and (3) cluster 3 covering China, Japan, Turkey, and Russia.
Spain and Norway, the fifth (pink) Hong Kong, Germany, Table 9 also shows that there are several disparities in the
Switzerland, and India, the sixth (orange) Taiwan and China, bibliometric analysis results between MDB and WOS data-
and the seventh (brown) only Greece. bases. Explicitly, there is an eight in ten disparities (80%)
Figure 4c shows that WOS contains four clusters. The between these databases (MDB and WOS). For example,
first (red) includes the United States, Germany, Russia, and for the most three keyword co-occurrence clusters, analysis
France, the second (blue) Canada and others such as The shows three clusters in MDB (1. Traveling salesperson prob-
United Kingdom and South Korea, the third (green) Japan, lem, 2. Operations research and management, 3. Business
Switzerland, Singapore, and others, and the fourth (yellow) and economics). However, in the case of WOS database, our
Spain and Greece. analysis reveals only two clusters (1. Traveling salesperson
Globally, we can argue that the USA is the most influ- problem, 2. Business and economics).
ential country for research in the sales force literature. The The third objective was to conduct a bibliometric analy-
USA’s researchers collaborate with some scholars belong- sis of sales force literature, which served to illustrate the
ing to other countries like Greece, Australia and Egypt (see first and second objectives and provide an overview of the
Fig. 4a). state and research trajectories of knowledge in sales force
literature. Regarding this last point, our results reveal (see
Table 9) that the research in sales force continue to grow;
General discussion the most three productive authors are Laporte G, Dubinsky
AJ, and Gendreau M; the most three consistently productive
The first objective of this study is to show how to make authors are Laporte G, Dubinsky AJ, and Cravens DW; the
MDB by combining and removing duplicate publications most three relevant sources are Journal of Personal Selling
from Scopus and WOS databases. To show the importance and Sales Management, European Journal of Operational
of MDB to better make a bibliometric analysis, we con- Research, and Industrial Marketing Management; the most
sidered the sales force literature between 1912 and 2019. three influential papers and authors are Dorigo et al. (1996),
Specifically, we make a bibliometric analysis comparison Dorigo and Gambardella (1997), and Hopfield and Tank
between MDB and Scopus, as well as among MDB and (1985); the most three corresponding author’s countries
WOS. Table 9 summarizes similarities and differences of are the United States, China, and Canada; the most three
these comparisons based on ten analysis types of bibliomet- cited countries are the United States, Canada, and Italy; the
ric analysis in the case of the sales force literature between most three frequent words are Traveling salesperson prob-
1912 and 2019. These analyses types are: (1) research lem, Business and economic, and Algorithm; the most three
growth, (2) most three productive authors, (3) most three Keyword co-occurrence clusters are Traveling salesperson
consistently productive authors, (4) most three relevant problem, Operations research and management, and Busi-
sources, (5) most three influential papers and authors, (6) ness and economics; the most three co-country collaboration
most three corresponding author’s countries, (7) most three clusters are (1) Canada, Netherlands, Germany, and Japan,
cited countries, (8) most three frequent words, (9) most three (2) the United States, Australia, Greece, and Egypt, and (3)
keyword co-occurrences clusters, and (10) most three co- China, India, Sweden, and others.
countries collaboration clusters. We focused only on the Based in our results and objectives, we discuss in the fol-
most three elements for two reasons. First, these elements lowing two elements: (1) Contribution to the bibliometric
are the most important ranked positions. Second, our objec- analysis, and (2) contributions to the sales force field.
tive is only to show there is some differences between MDB
and both Scopus and WOS databases. Contribution to the bibliometric analysis
Table 9 illustrates that there are many differences in
bibliometric analysis results between MDB and Scopus Bibliometric analyses are mainly performed by retrieving
databases. Specifically, among ten analysis, there is seven publications from WOS database or from Scopus databases.
(7/10 = 70%) disparities between these databases (MDB and Only, few scholars use both of them without indicating how
S. Echchakoui

Table 9  Synthesis of most similarities and differences between MDB and Scopus and between MDB and WOS
Analysis types MDB versus Scopus MDB versus WOS
Similarity Difference Similarity Difference

Research growth Tendency between 1912 Tendency between 2014 Tendency between 1912 Tendency between 2013 and
and 2014 and 2018. and 2013. 2018.
Most three productive First: Laporte G Third: First: Laporte G Second:
authors Second: Dubinsky AJ. MDB: Gendreau M. MDB: Dubinsky AJ.
Scopus: Wang Y. WOS: Gendreau M.
Third:
MDB: Gendreau.
WOS: Ahearne M.
Most three consistently Laporte G, Dubinsky AJ Laporte G has the: First: Laporte G Second:
productive authors and Cravens DW are First position in MDB. MDB: Dubinsky AJ and
the most consistently Same position as Dubinsky Cravens DW in MDB.
productive authors. AJ and Cravens DW. WOS: Wang Y.
Dubinsky AJ and Cravens Third author:
DW are the most have WOS: Gendreau M and
same position. Dubinsky AJ.
Most three relevant sources First: Journal of Personal First:
Selling and Sales Man- MDB: Journal of Personal
agement. Selling and Sales Manage-
Second: European Journal ment.
of Operational Research. WOS: European Journal of
Third: Industrial Market- Operational Research.
ing Management Second:
MDB: European Journal of
Operational Research.
WOS: Industrial Marketing
Management.
Third:
MDB: Industrial Marketing
Management.
WOS: Computers and
Operations Research.
Most three influential First: Dorigo et al. (1996). First: Dorigo et al. (1996). Third:
papers and authors Second: Dorigo and Gam- Second: Dorigo and Gam- MDB: Hopfield and Tank
bardella (1997). bardella (1997) (1985).
Third: Hopfield and Tank WOS: Xu (2005).
(1985).
Most three corresponding First: United States. First: United States.
author’s countries Second: China. Second: China.
Third: Canada. Third: Canada
Most three cited countries First: United States. Second: First: United States.
MDB: Canada. Second: Canada.
Scopus: Belgium. Third: Italy
Third:
MDB: Italy.
Scopus: Germany
Most three frequent words First: Traveling salesper- Second: First: Traveling salesper- Third:
son problem MDB: Business and son problem. MDB: Algorithm.
economic Second: Business and WOS: Computer science.
Scopus: Algorithm. economic.
Third:
MDB: Algorithm.
Scopus: Optimization.
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Table 9  (continued)
Analysis types MDB versus Scopus MDB versus WOS
Similarity Difference Similarity Difference

Most three Keyword co- First: Traveling salesper- Second: First: Traveling salesper- Second:
occurrences clusters son problem. MDB: Operations research son problem. MDB: Operations research
and management. and management.
Scopus: Heuristic algo- Scopus: Business and eco-
rithm. nomics.
Third: Third:
MDB: Business and eco- MDB: Business and eco-
nomics. nomics.
Scopus: many small nodes Scopus: no class
as human and neural
network.
Most three co-countries First: First:
collaboration clusters MDB: Canada, Nether- MDB: Canada, Netherlands,
lands, Germany, and Germany, and Japan.
Japan. Scopus: United States, Ger-
Scopus: United States, many, Russia, and France.
Netherlands, and Fin- Second:
land. MDB: United States, Aus-
Second: tralia, Greece, and Egypt.
MDB: United States, Scopus: Canada and others
Australia, Greece, and such as The United King-
Egypt. dom and South Korea.
Scopus: France, United Third:
Kingdom, and Sweden. MDB: China, India, Swe-
Third: den, and others.
MDB: China, India, Swe- Scopus: Japan, Switzerland,
den, and others. Singapore, and others.
Scopus: China, Japan,
Turkey, and Russia.

MDB merged database

to merge these two databases. To prevail Scopus or WOS case of sales force literature, our results showed this correla-
database, several scholars indicated the advantage of each tion is 22%, which is close to the 23% in the social sciences
database. For example, many authors (e.g., Goodman and that Sánchez et al. (2017) report.
Deis 2007; Chirici 2012; Zyoud et al. 2017) claimed that By showing that both Scopus and WOS database results
WOS database is more reliable and offer a varied range are different from MDB in the case of the sales force lit-
of information. In this study, we retrieved initially 17 365 erature, this research makes a notable contribution to the
publications from Scopus and 5 469 publications from bibliometric analysis literature. Specifically, we claim that
WOS. After using PRISMA method, we retain only 7 540 the comparison researches between Scopus and WOS data-
(43.4% = 7 540/17 365) suitable publications from Scopus bases is not productive because none of them are similar to
and 5 469 (47.8%) from WOS database. This result illus- the MDB. Indeed, by considering only the three most ranked
trates that WOS database has slightly more precise searches positions for ten analysis types, we found 70% dissimilarities
that Scopus database in the case of the sales force literature. between MDB and Scopus as well as 80% dissimilarities
Other authors (e.g., Mingers and Lipitakis 2010; Sánchez between MDB and WOS database. So, we encourage schol-
et al. 2017) argued that Scopus has a large thematic cov- ars not only to make comparison between Scopus and WOS
erage and number of articles. In this regard, our research databases, but to enhance how to effectively make MDB.
confirms this claim by showing that the number of publica- Second, this study shows a detailed procedure of how to
tions retrieved from Scopus on sales force literature is 37% perform MDB. This procedure is based on four steps: (1)
greater than the number of publications extracted from WOS convert WOS and Scopus databases to bibliography files, (2)
(Table 2). On the other hand, some scholars stated a correla- convert both WOS.bib and Scopus.bib to “bibtex” files, (3)
tion between the two databases. For example, Sánchez et al. format the WOS and the Scopus bibtex files to have the same
(2017) reported correlations of 23% in the social sciences, tag fields, and (4) merge the two databases in R or Rstudio
32% in health sciences, and 30% in physical sciences. In the and remove duplicates. To effectively make the first and the
S. Echchakoui

second steps, we showed how to perform these operations “How does it1 affect the manufacturer’s traveling men? Not
in Endnote (see Annex 1). We also propose VBA macro to only must they and their work be considered in the plan,
perform the third step (format both WOS.bib and Scopus. but the plan, but the plan must first be sold to them. This is
bib files). Finally, we give a VBA macro to delete duplicate accomplished by salesmen’s convention” (p. 175).
publications from WOS and Scopus databases. So, by speci- Second, drawing on our results, notably on Keyword Co-
fying these steps and providing information to perm them, occurrences in MDB (see Fig. 3a), we can argue that the
this extends bibliometric analysis research by helping schol- sales force literature can be classified on three important
ars to easily make bibliometric analysis based on the MDB. streams: (1) Travel sales salesman solution (ex., artificial
intelligence, neural network),( 2) Operational research and
Contributions to the sales force field management (ex., delivery problem, transportation), and
(3) business and economic (ex., sales control, motivation,
According to Bush and Grant (1994), it is important to performance). This latter stream is known in the sales force
assess the literature regularly as salesforce to detect topics literature as the sales force management. Figure 3a shows
and trends that reflect the state of knowledge. However, few that the two first streams are close, but these streams are very
bibliometric researches on sales force literature have been distant to the sales force management. Our finding indicated
performed. These studies (e.g., Bush and Grant 1991, 1994; that the research on the Travel sales salesman solution is
Leigh et al. 2001; Plouffe et al. 2008; Richards et al. 2010; more developed (ex., more frequency words, more publica-
Schrock et al. 2016; Williams and Plouffe 2007) consid- tions’ number) than the others. So, it can probably be advan-
ered selected articles mainly from sales force management tageous for the sales force management scholars to consider
literature. For example, Williams and Plouffe (2007) con- the other streams in their research, in order to better develop
sidered 15 journals (ex., journals like Journal of Marketing, the sales force literature.
Journal of Marketing research, Journal of Personal selling
and sales management, and Industrial Marketing Industrial
Marketing Management) covering 1012 articles published Limitations and future research
between 1983 and 2002. In the period of 2003–2007, Rich-
ards et al. (2010) studied 19 journals including 388 sales The first limit of this research is the timespan of coverage
articles focusing on sales management. Schrock et al. (2016) between Scopus database (1912–2019) and WOS database
covered the period 1980–2015 and considered 721 articles (1991–2019). The second limit is our consideration of only
published only in the Journal of Personal Selling and Sales two databases to make bibliometric analysis for sales force
Management. So, to extend these researchers, this study literature. Future research should assess others, such as Pro-
makes at least two contributions. First, we cover select arti- Quest and Google Scholar. We also assess only two types
cles from not only sales management literature, but we used of documents, articles, and conference proceedings. Future
large keywords (ex., sales force, salespeople) to retrieve all research should extend document types to include books,
publications interested in the sales force issues. Our results research notes, and theses. The third limit is that for many
show that effectively a large body of research and journals years, especially the older dates, the papers in the Journal of
(ex., European Journal of Operational Research, Comput- Personal Selling and Sales Management haven’t keywords.
ers and Operations Research) cover Traveling salesperson Research should also compare the three databases assessed
problem topic. Despite the importance of this topic and the in this study on topics other than those found in sales force
method (ex., optimization, genetic algorithms) used to solve literature to assess the validity of current findings.
this problem, this is not considered in the previous biblio-
metric analysis. This gap is surprising because a traveling
salesman is considered by some historical scholars in sales Compliance with ethical standards
force research as a class of salespeople. For example, Wit-
kowski (2012) highlighted that Sheldon (1911) considered Conflict of interest On behalf of all authors, the corresponding author
states that there is no conflict of interest.
traveling salesmen as a category of sales force. In addition,
concerning the roles of the sales force in the promotion’
plan, Beard, (2015) cited the role of the traveling men, high-
lighted by Calkins (1915). In his description of this role, References
Calkins (1915) used “the manufacturer’s traveling men” as
Archambault, E., D. Campbell, Y. Gingras, and V. Larivière. 2009.
a type of salesman. Specifically, in his response to how the Comparing bibliometric statistics obtained from the Web of
promotion’s plan affects salespersons, Calkins (1915) stated

1
It: the promotion’s plan.
Why and how to merge Scopus and Web of Science during bibliometric analysis: the case of sales…

Science and Scopus. Journal of the American Society for Infor- Mingers, J., and E. Lipitakis. 2010. Counting the citations; a compari-
mation Science and Technology 60 (7): 1320–1326. son of Web of Science and Google Scholar in the field of business
Archambault, E., É. Vignola-Gagne, G. Côté, V. Larivère, and Y. Gin- and management. Scientometrics 85 (2): 613–625.
gras. 2006. Benchmarking scientific outputs in the social sciences Moher, D., A. Liberati, J. Tetzlaff, and D.G. Altman. 2009. Preferred
and humanities: The limits of existing databases. Scientometrics reporting items for systematic reviews and meta-analyses: The
68 (3): 329–342. PRISMA statement. Annals of Internal Medicine 151: 1–8.
Aria, M., and C. Cuccurullo. 2017. Bibliometrix: An R-tool for com- Mongeon, P., and A. Paul-Hus. 2016. The journal coverage of Web of
prehensive science mapping analysis. Journal of Informetrics 11 Science and Scopus: a comparative analysis. Scientometrics 106
(4): 959–975. (1): 213–228.
Bartol, T., and M. Mackiewicz-Talarczyk. 2015. Bibliometric analysis Muhuri, P.K., A.K. Shukla, and A. Abraham. 2019. Industry 4.0: A
of publishing trends in fiber crops in Google Scholar, Scopus, bibliometric analysis and detailed overview. Engineering Applica-
and Web of Science. Journal of Natural Fibers 12 (6): 531–541. tions of Artificial Intelligence 78: 218–235.
Beard, F. 2015. Forgotten classics: The business of advertising, by Norris, M., and C. Oppenheim. 2007. Comparing alternatives to the
Earnest Elmo Calkins (1915). Journal of Historical Research in Web of Science for coverage of the social sciences’ literature.
Marketing 7 (4): 573–583. Journal of Informetrics 1 (1): 161–169.
Bush, A.J., and S.E. Grant. 1991. An analysis of leading contributors Plouffe, C.R., B.C. Williams, and T. Wachner. 2008. Navigating dif-
to the sales force research literature, 1980 through 1990. Journal ficult waters: publishing trends and scholarship in sales research.
of Personal Selling & Sales Management 11 (3): 47–56. Journal of Personal Selling & Sales Management 28 (1): 79–92.
Bush, A.J., and S.E. Grant. 1994. Analyzing the content of market- Pritchard, A. 1969. Statistical Bibliography or Bibliometrics? Journal
ing journals to assess trends in sales force research: 1980–1992. of Documentation 254: 348–349.
Journal of Personal Selling & Sales Management 14 (3): 57–68. Reinelt, G. 1991. Tsplib: A traveling salesman problem library. ORSA,
Calkins, E.E. 1915. The business of advertising. New York, NY: D. Journal on Computing 3 (4): 376–384.
Appleton and Company. Richards, K.A., W.C. Moncrief, and G.W. Marshall. 2010. Tracking
Castillo-Vergara, M., A. Alvarez-Marin, and D. Placencio-Hidalgo. and updating academic research in selling and sales management:
2018. A bibliometric analysis of creativity in the field of business A decade later. Journal of Personal Selling and Sales Manage-
economics. J. Bus. Res. 85: 1–9. ment 30 (3): 253–272.
Chirici, G. 2012. Assessing the scientific productivity of Italian forest Schrock, W.A., W. Zhao, D.E. Hughes, and K.A. Richards. 2016.
researchers using the Web of Science, SCOPUS and SCIMAGO JPSSM since the beginning: intellectual cornerstones, knowledge
databases. IForest-Biogeosciences and Forestry 5 (3): 101–107. structure, and thematic developments. Journal of Personal Selling
Ellegaard, O., and J.A. Wallin. 2015. The bibliometric analysis of & Sales Management 36 (4): 321–343.
scholarly production: How great is the impact? Scientometrics. Sheldon, A.F. 1911. The art of selling. Libertyville, IL: The Sheldon
https​://doi.org/10.1007/s1119​2-015-1645-z. University Press.
Ertz, M., and A. Leblanc-Proulx. 2019. Review of a proposed meth- Strozzia, F., C. Colicchia, A. Creazza, and C. Noè. 2017. Litera-
odology for bibliometric and visualization analyses for organiza- ture review on the ‘Smart Factory’ concept using bibliometric
tions: application to the collaboration economy. Journal of Mar- tools. International Journal of Production Research 55 (22):
keting Analytics 7: 84–93. 6572–6591.
Escalona, M.I., P. Lagar, and A. Pulgarín. 2010. Web of Science vs. Sánchez, A.D., M. Del Río Rama, and J.Á. García. 2017. Bibliometric
Scopus: un estudio cuantitativo en Ingeniería Química. Anales de analysis of publications on wine tourism in the databases Scopus
Documentación 13: 159–175. and WoS. European Research on Management and Business Eco-
Fabregat-Aibar, L., M. Barberà-Mariné, G. Terceño, and A. Laia Pié. nomics 231: 8–15.
2019. Bibliometric and visualization analysis of socially respon- Wagner, A.B. 2015. A Practical comparison of Scopus and Web
sible funds. Sustainability 11: 2526. of Science core collection https://ubir.buffalo.edu/xmlui/
Fingerman, S. 2006. Web of Science and Scopus: Current features handle/10477/38568
and capabilities. Issues in Science and Technology Librarianship Wang, Q., and L. Waltman. 2016. Large-scale analysis of the accuracy
48: 4. of the journal classification systems of Web of Science and Sco-
Garfield, E. 1971. The mystery of the transposed journal lists: Wherein pus. Journal of Informetrics 10 (2): 347–364.
Bradford’s law of scattering is generalized according to Garfield’s Williams, B.C., and C.R. Plouffe. 2007. Assessing the evolution of
law of concentration. Current Contents 1: 222–223. sales knowledge: A 20-year content analysis. Industrial Marketing
Garrigos-Simon, F., Y. Narangajavana-Kaosiri, and I. Lengua-Lengua. Management 36 (4): 408–419.
2018. Tourism and sustainability: A bibliometric and visualization Witkowski, T.H. 2012. Marketing education and acculturation in the
analysis. Sustainability 10: 1976. early twentieth century. Journal of Historical Research in Market-
Gavel, Y., and L. Iselid. 2008. Web of Science and Scopus: A journal ing 4 (1): 97–128.
title overlap study. Online Information Review 32 (1): 8–21. Zyoud, S.H., W.S. Waring, S.W. Al-Jabi, and W.M. Sweileh. 2017.
Goodman, D., and L. Deis. 2007. Update on scopus and web of science. Global cocaine intoxication research trends during 1975–2015: A
The Charleston Advisor 7 (3): 15–18. bibliometric analysis of Web of Science publications. Substance
Hall, C.M. 2011. Publish and perish? Bibliometric analysis, journal Abuse Treatment, Prevention, and Policy 12 (6): 1–16.
ranking and the assessment of research quality in tourism. Tour.
Manag 32: 16–27. Publisher’s Note Springer Nature remains neutral with regard to
Harzing, A.-W., and S. Alakangas. 2016. Google Scholar, Scopus and jurisdictional claims in published maps and institutional affiliations.
the Web of Science: A longitudinal and cross-disciplinary com-
parison. Scientometrics 106 (2): 787–804.
Knoke, D., and S. Yang. 2008. Social network analysis, 2nd ed. Thou-
sand Oaks, CA: SAGE Publications Inc. Said Echchakoui is a Professor of Marketing at the University of
Leigh, T.W., E.B. Pullins, and L.B. Comer. 2001. The top ten sales Quebec at Rimouski. He holds DBA degree from the University of
articles of the 20th century. Journal of Personal Selling & Sales Sherbrooke, Quebec, Canada. He is affiliated at the Association of
Management 21 (3): 217–227. Professional Engineers & Geoscientists of New Brunswick, Canada.
S. Echchakoui

His research interests focus on sales force, relationship marketing, mod- European research on management and business economics, Journal of
eling, strategy, and loyalty. His work has been published in journals Global Marketing and Global Business and Organizational Excellence,
such as the European Journal of Marketing, Journal of Retailing and and in proceedings such as The Annual Conference of the Journal of
Consumer Services, Recherche et Applications en Marketing, Jour- the Academy of Marketing Science, and The National Conference in
nal of Modelling in Management, Journal of Marketing Analytics, Sales Management.

You might also like