Jounal of Busines Research 133 (2021) 285-296,
Contents lists available at ScienceDirect
Journal of Business Research
ELSEVIER, journal homepage: www. elsevier.comlocateibusres
®
How to conduct a bibliometric analysis: An overview and guidelines an
Naveen Donthu*”, Satish Kumar“, Debmalya Mukherjee“, Nitesh Pandey”, Weng Mare Lim“**
Deparment of Maring Ceri Ste Une, Aa 0 30902, USA
*bepstnen of Manapoent Sede, Maley National nt of Teco Jane, 902017 eta Ii
< apanent of Manag, Cleo Bars, The Unity of Alo, 259 South Be, Aloo, OF M825
© Sroure Buse So. Sytaure Omer of Tool, Jom Soe, $122 ardor Wer, Aa
*sohol Buss, Sumas Ubety of Tele, Jn Soa Tes, 95850 Ruhr, Sarak, Melt
ARTICLE INFO apsTRact
ower
ime anal
Performance ani
o-aton aa
hippie coupling
ibliometrie annyse fe a popular and rigorous method for exposing and analyzing arg wokines of scene
dat, 1 enables us co unpack the evolutionary nuances ofa specie fl wile shedding Ught on the emerging
tes in that Geld. Yer, its application in busaes eseaich i eatvely evr, and in many istanoes, underde
‘eloped, Accordingly, we endeavor to present an overview ofthe biblloneni mehodeogy, with a paicula
focus on its diferent tecnies, while offering step-by-step adetines tat can be relied pon to rigovonsy
perfom bibliometie analysis with confidence To this end, we also shed Hight on when and how biblomeaie
‘alysis should be wed vis--sis othe sila techniques such as neta analysis and sptemuai tert eviews.
[Asa whol, this paper should be a sel resource fr gaining lsights on the avalale techniques and procedures
for earying out studies using bibliometic analysis
1. Introduetion
Bibliomettic analysis has gained immense popularity in business
research in recent years (Donth, Kunuar, & Patnaik, 2020b; Donthu,
Kuna, Pataik, & Lim, 2021; Khan et, 2021), and is popularity can
be attributed 10 (1) the advancement, availability, and necessity of
bibliometric software such as Gephi, Leximancer, VOSviewer, and sc-
‘entific databases such as Scopus and Web of Science, and (2) the eros.
disciplinary pollination of the bibliometric methodology from infor.
mation seienee to busines research. More importantly, the populity of|
bibliomerrie ansiysis in business research is not a fad but rather &
reflection ofits utility for (1) handling lage volumes of scientific date,
‘and (2) producing high research impact.
‘Scholars use bibliometric analysis fr a variety of reasons, such a8 (0
uncover emerging trends in article and journal performance, collabo:
"ation parters, and research constiments, and to explore the intellectual
strete of a specie domain in the extant iteratire (Dont, Kunin,
Pandey, & Lim, 20214; Verma & Gustafsson, 2020; Donthn et al.
2020e). The data that takes center stage in bibliometric analysis tends ro
bbe massive (eg, ndreds, f not chousands) and objective in nature (e
{g itmber of eitations and publicntions, occurrences of keywords and
* Coxresponding autor
topics), though its interpretations often rely on both objective (e.,
performance analysis) and subjective (e., thematic analysis) evalua
‘ons established through stormed techniques and procedures. n other
words, bibliometric analysis is useful for deciphering and mapping the
cumulative scientific knowledge and evolutionary nuances of well
established fields by making sense of large volumes of unstmetured
data in rigorous ways. Therefor, bibliomerric studies that are well done
«an build firm foundations for advancing afield in novel and meaningful
ways—it enables and empowers scholars to (1) gain a one-stop over
ew, (2) identify knowledge gaps, (3) derive novel Ideas for invest
gation, and (4) position their intended contributions tothe field
Notwithstanding its merits, bibliometric analysis reninins relatively
new in business research, and in many instances, its deployment does
not make fll use of is potential. This occurs when bibliometric studies
rely om a limited set of bibliometric data and techniques and provide
only a piecemeal understanding of the field under study (eg, perfor
mance analysis without science mapping—e.g., Brown, Park, & Pitt
2020). Is important to note that an axthortaive guide to bibliometric
‘analysis in business research remains absent, which poses asa signif
cant challenge for business seiolars who wish to learn more about the
bibliometric methodology and its application for business research in &
mall addreses: odouthudysuedu (N. Dowthe), skumardns@amitacin (S. Kuma), dmubher@uakioa.edu (D. Mukherjee), 2018RBN0016@maitacin
(6%. Pandey, limarvengmate com (WML Lm).
psd. og/ 10.1016, uses 2021.04.070
Received 13 March 2021; Recelved in revised fum 25 Apil 2021; Accepted 28 Apll 2021
Available online 14 May 2021
(0148-2963/© 2021 Eevee All ight reserved holistic yet ensy-to-digest manner. Though muthoritaive guides on sys
tematic iterate reviews are available (e. Palmatier, Houston, &
Hullané, 2018; Snyder, 2019), they do not provide adequate breadth
‘nd depth on the bibliometric methodology:
Given the aforementioned gaps, this paper aims to offer (1) an
‘overview of the bibliometric methodology and (2) step-by-step guidelines
for conducting bibliometric analysis for business research. In particular,
this paper introduces bibliometric analysis to business scholars, wherein
its fundamentals, cechniques, and procedures, with exemplars and ra
Uonales, are provided. The conteburions of this paper are manifold.
Firs, the paper, which presents an overdew of bibliometric analysis and
the guidefnes on Row to conduct it, can help business scholars to learn
about te bibliometric methodofogy and co use that understanding (0
‘evaluate specific felds in che extant literanste with large bibliometric
data snd corpus. Second, the paper, which provides several siggestions
regarding the different fectniques that ean be used for biblionetric
‘analysis and when they should be used, can widen the perspective of
business scholars on the alternatives and rationales for using the
dlffeent variants of bibliometric analysis. AS a whole, this paper en
hances understanding of the bibliometric methodology for business
research with clarity and rigor, and thus, paves the way for business
scholars to use bibliometric analysis appropriately, meaningflly, and
rigorously in ther fure research
The rest ofthe paper i structured as follows, The paper begins with,
‘an overview of bibliometric analysis, followed bya toolbox of available
techniques for bibliometric analysis with accompanying guidelines on
when to use them. Next, the paper delves into the diferent research
metres and clustering algorithms used in the network analysis of bib
liomerric data, The paper then explains the process of condueting a
bibliometric analysis before concluding with its final remarks on the
limitations associated to bibliometis.
2. The bibliometric methodology
‘The bibliometric methodology encapsulates the application of
‘quantitative techniques (i. bibliomettie analysis—e.., citation ana
ysis) on bibliometric data (e.g, units of publication and citation)
2500
2000
1500
so 553
302 437
273, I |
Jeena f Buns Rater 198 (2021) 285.206
(roads, 1987; Pritchard, 1969). Early discussion on bibliometres
started inthe 1950s (Wallin, 2005), which suggests thatthe bibliometric
tethodology is not new. Ye, the proliferation of bibliometics is fishy
recent, as seen through Hts growth in the fields of “business, manage-
ment, and accounting,” “economies, econometrics, and finance,” and
"socal sciences" on Seopus using “bibiom*” asa keyword in the “article
title, abstract, and Keywords” (see Fig, 1). Specifically, publications
using bibliometrcs have grown over the years, with an average of 1021
publications in the lst deeade, whieh ean be atiibuted ro the growth of|
Scientific research itself. Yet, large bibliographic datasets have made
classic review methods cumbersome and impractical (Ramos Rodrigue
© Rule Navarro, 2004, tis noteworthy dat the emergence of scientific
Atabases such 4s Scopus and Web of Scence has made aequiring large
Volumes of bibliometric data relatively easy, an bibliometric software
sich as Gephi, Leximancer, and VOSviewer enable the analysis of sich
data in a very pragmitle way, thereby rising scholarly interest in bib
liometrie analysis in recent times, Indeed, the bibliometric methodology
has een applied in @ variety of Tels in business research, including
business strategy (Kumar, Surekha, Lim, Mangla, & Goyal, 2021),
electronic commerce (Kumar, Lim, Pandey, & Westland, 2021), finance
(Durisin & Puzone, 2009; Linnenlvecke, Chen, Ling, Smit, & Zhi, 2017;
Xu ot al, 2018), human resources (Andersen, 2019), management
CEllegaard & Wallin, 2015; Zupie & Cater, 2015), and marketing
GBackbiaws, Lgger, & Koch, 2011; Donih, Kumar, Pandey, & Soni,
20204; Dont, Kumar, & Pettnaik, 2020b; Donth, Kumar, Patina, &
im, 2021; Hu, Song, & Guo, 2019; Samice & Chabawski, 2012: Donth
ct al, 2020e), wherein the application of bibliometres ranges from
studying publication to collaboration patterns and exploring the intl:
lectual stricture of the research Feld. Her, the resent Field enn also
tanifest as journals. Indeed, the bibliometric methodalogy has been
applied to provide retrospectves of journals (e.g Journal of Business
Research), which typically occur in milestone years (Deut et al,
20200),
Acti juncture, ris importa to compare bibliometric analysis with
other frequently used review alternatives sch as mista analysis and
systematic literature reviews. In essence, meta-analysis estimates (1)
“the overall strength and direction of effets or relationships,” and (2)
1950
1701
1392
667
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
‘ig. 1. Year wise publication of biblometie popes. Note(s): This figue represents the publication trend of biblometse papers between 2005 and 2020, The data
‘was tetteve fom the Scopus database in the subject aeas of “bashes, management, and accountng,” economies, economies an Rnaee,” and soa
‘ees ing the Kesword Dion” “the across-study variance inthe distribution of effect size estimates and
the Factors that explain this variance” (Aguinis, Pierce, Bosco, Dalton, &
Dalton, 2011, p. 310}, whereas systematic literature reviews, such as
lomialn method, and theory-based reviews, encapsulate the aequis
Hon, arrangement, and assessment of the extant erature using s¥s-
tematic procedures (Palmatier et al, 2018; Tranfiel, Denyer, & Smart,
200), whieh are typically caried ou manwally (eg. content and the
‘matic analyses) by scholars (Lahiri, Mukherjee, & Peng, 2020; Lim, Yap,
‘& Makkar, 2021).
‘Similar to bibliometric analysis, meta-analysis is able o handle large
amounts of literature and provides nuanced summary ofa given fel,
though the Ueraeure considered tend co be less diverse, and the het
‘erogeneity of existing studies and the existence of publication bis can
have an adverse effect on the validity of the results obtained via meta
analysis CAguinis, Gottiredson, & Wright, 2011; Jnni, Serla, Tar,
{© Tous, 2015). In contrast, systema literature reviews using classic
‘metliods require @ narrow scope of study and thus tend to include &
lesser number of papers for review (eg, between tens [e.g 40] and low
Inmereds (eg, 100-300)) (Snyder, 2019). th that sense, systematic
literature reviews are better suited for confined (eg, customs
‘engagement on social media) or niche research ateas (eg, social medi
Influencer marketing). Unlike systematic literature reviews that tend to
rely on qualitative techniques, which could be marred by interpretation
Dias from Scholars across different academic backgrounds (McCown,
1098), bibliometric analysis and meta-analysis rely upon quantitative
techniques and thus ean avoid or mitigate that bias
Since meta-analysis and bibliometric analysis are both quantitative
In nature, the distinction between the two methods can be confusing to
‘sone scholars. To shed light on this dstinetion, business scholars should
note that their quantitative methods are relatively different in terms of
usage, though they can both handle Inge amounts of literature. Spe:
cifically, meta-analysis concentrates on summarizing empirical evidence
by anelyzing the direction and strength of effects and relationships
‘among varinbles and is “useful fn addressing open research questions
‘with date that are closer to definitive than those reported in any single
primary study” (Carney, Gedallovie, Heugens, Van Essen, & Van Oos
‘erhout, 2011, p 438) tis performed to throw light on mixed empirical
findings and Doundary conditions. Thus, meta-analyses are often used as
Jeena f Buns Rater 198 (2021) 285.206
theory extension tools (Combs, Ketchen, Crook, & Roth, 2011). in
contrast, bibliometric analysis summarizes the bibliometric and intl
lectual structure of afield by analyzing the social and structural re
lationships between different research constiments (eg, authors,
countries, institutions, topics).
In summary, the use of any ofthe three review methods discussed
herein is dependent upon the goals of the review and the magnitude and
nature of the literature being reviewed. Nonetheless, these review
methods remain complementary to one another and they oer nique
Advantages fo scholars who are interested in using them, Table 1 pre
sents the methodological comparison of bibliometric analysis, mete
analysis, and systematic literature reviews across different iterla to
help anhors make informed decisions with respec tothe selection ofthe
appropriate review method
3. The bil
iometrie analysis technique toolbox
‘The cechniques for bibliomesric analysis manifest across wo eate
sgoties: (1) performance analysis and (2) science mapping. In essence,
Performance analysis accounts for the conrhurons of research consti
liens, whereas science mapping focuses on the relerionships between
research constituents, The next sub-sections sheds light on the tec
rhigues available for perfomance analysis and scieace mapping, whieh
ace illustrated in ig, 2
3.1. Performance analysis
Performance analysis examines the contributions of research consti
uwents toa given fled (Cobo, Lipez- Herrera, Herrera Viedina, & Herrera,
2011; Ramos Rovrigue & Rutz Navarro, 2004). The analysis, whieh is
descriptive in nature, isthe hallmark of bibliometric studies (Dost,
Reinartz, Kumar, & Patnaik, 2020), Performance analysis can be found
in most reviews, even in those that do not engage in science mapping,
bese irs standard practice in reviews to present the performance of|
diferent research constituents (eg, authors, nstcutions, countries, and
journals) in the field, which is akin ta the background or profile of
Darteipants that is typically presented in empirical research albeit more
analytically.
‘Table 1
Compassoa of major sevew methods
Reiwope Goal Wes we Woes soto Sop Demet Ane
‘Wloneric + Summa lng quantce of» When he nope fveviowis + When the spe ofvoiow Bread lange + Quanto
aie Wiener date pes the brow Eapecie (eelunon nd
ne ofthe tlic suture + When he ant to lange» When ete all inerpecon)
fd enering wens of locales nd manageable eoush + Quatawve
"esearch opto el hac ts comtene ante limereeraon
sanuly eiewed cain)
Meceanaiyis + Suman the nit =» Whenthefousofeview nto + Whenetidesin fe feld Broad Large + Quant
idence ofelmonsipbeeveen”sunmartesisrebe tan” re betrogeneou SSpecne 5 Small bur” (vauon and
‘rahe while uneven tecngne with cote + When te numberof Mdequte —_lerpeation)
sng ste, ‘peste, telly ow
+ When sit nthe eld are» When the umber a Hh
only der tenty
ek
sSyvumae + Suamaines debe the» When the ope fteiew = When espe oevew « Specie = Small» Quainive
Tiere Endings ering arr on pete ort (eralun nt
a reek tpi ei «When te data tno ‘serpenin)
foot a be manual
renews
Inge fer unl re Jeena f Buns Rater 198 (2021) 285.206
ior aah
Man esha Torchnt chu
formance aoe = Necro ab
— ;
Fublaion ltd metrin ‘lation nabs Neeork mt
Tosa pebtcaon TP) 4+ Relations ang pubisons + Degecofeaiy
Naber fcoetting sth (NCAY Montini penis 3 eweecncealty
2 Sonor poston A) 2 Egemectrceaiy
+ Comedies CA) Corcaton aay Clone cetaliy
Namber race year of pb (NAY) 4+ Reaontipeaneng ced bene stank
(Cintewrted meen graph eupting Explor ftr amas
+ Tonloaion 10) latrines irre tig
fee cianNC) 1 Pei ren hemes Mini
Gio paticatonre Connor aes aes
5 eee + Extn re rantisamions Site conen pin
1 Calaorai oeet(CC) tenis a
2 Namba pans CP) + Ween tet (vet) + BiwenarcR = Sein
potion ated pbs PCP) 7 ee) tsa
Cis hd pbnen CoP , Sezateripenas on
[oon mies eraiensamne | | | So
finer + thors end tor iio $ vciser
Tings 10 10,200) eS 2 vosvewer
Fig. 2. The bibiomettic analysis totbo.
Myriad measures for performance analysis exst. he mst prominent
measures are the muuber of publications snd citations per year or per
research constiruent, wherein publication is @ proxy for productivity,
Whereas citation isa measure of impact and influence. Other measures
suchas citation per publication and frindex combines both citations and
publications to measure the performance of research constituents. The
analysis, despite being descriptive, recognizes the importance of
dlifferent constituents in @ research fel Table 2 presents sample of
mettes sultabe for performance analyst
2, Selence mopping
Selence mapping examines the relaionships berween research con
stituents (Baker, Kumar, & Pandey, 2021; Cobo et al, 2011; Ramos
Rodrigue J Rie Navarra, 2004). The analysis pertains tothe intellec-
tual interations and strvetural connections antong research, consti
‘ents. The techniques for scence mapping include citation analysis, co
citation aralyss, bibliographic coupling, co-word analysis, and co
whorship analysis. Suc techniques, when combined with network
‘analysis, are instromental in presenting the bibliometric stractare and
the intellectual strecure of the research field (Saker, Pandey, Kua, &
Haldar, 2020; Tunger & Fulerieh, 2018), Table 3 presents a summary of
the different techniques fr selence mapping with focus on thelr usage
‘ancl datn considerations
12.1, Cttation anabss
CCtation analysis is a basic technique for science mapping that
‘operates on the nssumption that citations reflect intellectual linkages
‘ervicen publications thac are form when one publication etes the other
(pio, Cesaroni, & Di Minin, 2014). In this analysis, the impact of &
publication i determined by the number of eitations that itreceives. The
‘analysis enables the most influential publications in a research field to
be ascertained, Though there are a variety of methods (e.g, network
rmettes) to determine the importance of publications in research field,
the most objective and straightforward measure of its impact Is its
citation (Pieters & Baumgartner, 2002; Stremersch, Verniers, & Verhoef,
2007). Therefore, using citations, one ean analyze the most influential
Dublicadons in a research Reld to gain an understanding of the intl
lecrual dynamics ofthat field
$2.2. Cositation anadysis
Co-citation analysis is technique for sefence mapping that ass
publications that are cited rogether frequently are similar thematieally
(Giorland, 2013). The analysis can be used to reveal the inelletal
structure of a research field (Rossetto, Bernardes, Borni, & Gattaz,
2018), such sits underlying themes (Li, Yin, Li & Dunford, 2015), kn
1 cocitation network, two publications are connected when they co
‘occur inthe reference list of another publication, The beneft of using
co-citation analysis is that, in addition to nding the most influential
publications, business scholars can also discover thematic clusters. Here,
the thematic clusters are derived based on the cited publications. How
ver, co citation analysis concentrate ouly on highly-cited publications,
tnd leaves publications that are recent or niche ont of is thematic
clusters, In that sense, co-citation analysis is suitable for business
Scholars who wish to uncover seminal publications and knowledge
foundations
8.2.3 Bibliographic coupling
Bibliographic coupling is a technique for seience mapping that
operates on the assumption that two publications sharing common ref
ferences are aso similar in their content (Kessler, 1963; Weinberg, 1974).
“The nnalysis concentrates on the division of publications into thematic
clusters based on shared references, and is best used within a specific
timeframe (Zuple & Cates, 2015), Here, the thematic clusters are Formed
based om he citing publicarions, and this, recent and nice publications
can gain visibility dhrough bibliographle coupling (unlike co-citation
snalysis). In that sense, bibliographic coupling is stable for business
Scholars who wish to uncover a broad spectrum of themes and is latest ‘Table 2
Metis for perfomance analysis
‘Mee Desion
Toul pubieaons (TP)
-ublemion or sesdemin 7.
Pubeaton ondary (TP)
‘Tot puto of esearch consent
‘Total publeaton of eseach conser fom
Tot ptbieaton of esearch construen fom
tench mite
Tot umber of eo auhred puletons by
Pubeatons fo sede
"nde elaboration (TPAD
inca)
18a)
‘co-eutord publintons (C8)
pubic A) seco a poleton
odie scive year o€ TPT NAT
bleston PAY)
‘ot craton (Te)
Avengers (46)
Average tts per plistion, per
eon pepe) of eon soot
alteration ter (2) (WC TP) Tike, heen of
‘elaboration free snc)
TGP NCA) (he tannin tbe erento
stor caliabraon beeen and)
atboatin otic (0)
incr) shat ar ee
"Pcr)
(cer)
index enum of publication ted tet ines
Ges suc of ene)
index ® {number of publtions ecg a est
Staton (hes east of pee)
umber of beans ed test Cmes
(a. 10,10, 20, ee)
dex (40, 6400, 200)
‘Note(s: Compilation base on author expeteace and experi inbblioneate
analysis. Metres canbe compited for eath researc consent (6, authors,
Insiutions, counties journals) as an aggregate (cg esearch consti) orin
specific (eg, research constituent per publiedon, per year, or pet peta)
‘depending on information needs (x, aggtegtes for overviews, species fr
teens obsevation),
developments. The analysis ean thetefore provides representation ofthe
present ofthe research fl
12.4. Co-word analysis
While the previous three techniques for seience mapping focus on
publications, the unit of analysis for co-word analysis Is “words.” In
‘other words, unlike citation analysis, co-citation analysis, and biblio
‘graphic coupling, which employs either eted or citing publications asa
focal point o a proxy, the co-word analysis sa technique that examines
the actual content of dhe publication itself, The words in a co-word
‘analysis are often derived from “author keywords", and init absence,
notable words ean also be extracted from “ace ies," “abstracts,” and
“fll ext forthe analysis eg, Baker, Kumar, & Panviey, 2020; Burton,
Kumar, & Pandey, 2020; Dont, Gremler, Knmar, & Patti, 2020;
Emieh, Kumar, Lu, Notder, & Pandey, 2020; Liu, Mai, & MacDonald,
2018), Similar to co-citation analysis, the eo-word analysis assumes that
words that frequently appear together have a thematic relationship with
‘one anoth
The usage of words as a unit of analysis, however, has is downsides
For example, certain words are used in multiple contexts, and thus, (ee)
‘reading of publications becomes necessary to understand the meaning of
Jeena f Buns Rater 198 (2021) 285.206
‘Table
‘echiques for science mapping and ther usage, unit of analysis, and data
requikements.
Tecbleue Une Vaio Daw one
‘lation _‘Tossahzethe Docwnens Autor eame Poslot,
‘tials roe Glacoms—schenie,
tions Tile
no oil Reference
bias na
‘bee sowie Sebi ond
bento ams)
the fant
ouplg reso The (aoa)
fimong ng Jotale
beat bot
gern eh References
pent ot
evelopment of
themes
Toeaplow the Words Tie ich ea
xsrng ofr Abie (020)
aps sumer
sion pt epwonee
by oceing 00 eywores
aie socal stow Binion (2006)
mera or (eneton
note nd coy)
tid tei
ans at
elope of
theresa
fe
the relationships between words. Besides that, some words can be very
general (eg subject eld namies—eg., advertising) and ts, Ie may be
‘challenging to assign themt to aay one thematic clster.
‘To mitigate the potentinl dovinsides of co-word analysis, business
scholars are encouraged to use the analysis strategically. Here, wo
recommendations are provided. First, a co-word analysis can be used as
A supplement co enrich underseanding about che tematic clusters
derived from co-citation analysis or bibliographic coupling because the
themies formed through the commonalities in publications tend to be
relatively general (Chang, Huang, & Lin, 2019), and thus, the wse of co
word analysis can help business scholars to elaborate on the content of
teach thematic cluster. Second, #co-word analysis can be sed t forecast
ine researc inthe fel, whic ean be happen when notable “words”
from the publication's implications and future research directions are
used in the analysis. n that sense, the co-word analysis is switabe for
business scholars who wish to enrich ther interpretations of eo-eitation
alysis (past or bibliographic coupling presen) and to predict fort
coming trajectories, The eo-word analysis can therefore provide & review of the fune ofthe research Feld
12.5, Co-authorship analysis
(Co-uthorship analysis examines the Interactions among schokaes in
4 research field. Since co-authorship is a formal way of intellectual
collaboration among scholars (Acedo, Barroso, Casanueva, & Galan,
2006; Cisneros, Ibaneseu, Keen, Lobato Calleros, & Niebla Zatarain,
2018), it is therefore important to understand how scholars interact,
amongst themselves Cineluding associated author attributes such as
affiliated institutions and countries), With the inereasing methodolog
lea and theoretical complexity in research, collaborations among
scholars have econte commonplace (Aceo et al, 2006), I fa,
collaborations among scholars can Tead to. improvements. in
research —for example, contributions from different scholars can
contribute to greater clarity and richer insights (Tahamtan, Sapo
‘Aflac, & Alvandzadeh, 2016), Here, scholars that collaborate form a
‘network known as “invisible collages,” whose study ean help develop
the undertakings in the research field (Crane, 1969). For example, the
‘analysis can shed light on elustered research among. scholars feom a
‘particular region, and such insights ean be used to justify and spar neve
research among scholars in underrepresented regions. The analysis also
‘enables collaborations to be mapped across different periods oft
thereby enabling scholars ro review the cajectory of intelleetal
evelopment against collaboration networks, while equipping prospec:
tive scholars with valuable information to teach out and collaborate
‘wilh established and tending scholars in dhe research fel
4. The bibliometric analysis enrichment toolbox
Building on the core techniques of bibliometric analysis this section
presents the add-ons that can augmented to enrich the outeontes of the
‘analysis techniques applied in bibliometric studies. In total, three
‘enrichment pathsnys predicated on network analysis are sggested in
the form oF network metres, elstering, and visalization.
4d. Network metrics
[Network metres can be used to enrich the assessment of bibliometric
analysis, In particular, network metres shed light on the relative
importance of research constituents (e., authors, stitutions, coun:
tries), which may not necessarily be reflected through publications or
citations. Importantly, network meties are commonly employed £0
‘enrich the diseussion of research fields in bibliometric studies (Ander
‘sen, 2019; Andrikopoulos & Economou, 2016; Baker, Kumar, and Patt
naik, 2020; Cisneros etal, 2018), and thus, they represent legitimate
method for enriching bibliometric assessments. To provide greater
larity, several exemplars of nerwork metres are provided (eg, degree
‘of centrality, Derweenness centrality, ejgenvector centrality, closeness
centrality, and PageRank), along with a sale table of publication
raking that canbe curated across different centality measures (see
“Table 4). specifically
1 Degre ofcenraty refers to the number of relational ties & research
constituent has in a network. For example, if an author in a co:
Authorship network Iias worked with four different authors, then
his or her degree of centrality would be four. This is by far the
simplest measure of centrality asi relies on the numerical count of
relational es. A variant of this measure Is the weighted degree of
‘enralcy, which is calculated by anuliplying the total auber of
relational ies withthe strength of each te. For example, i Author A
has written wo publications with Author B and one each with
Author Cand Author D, then Author A’s degtee of centrality will be
three but his or her weighted degree of centrality would be four
Though the siuplicity ofthese mensures is an advantage in itself the
measures do not preseat information on what role a researc con:
stiruent plays in the researeh fel
Jeena f Buns Rater 198 (2021) 285.206
‘Table 4
‘Simple ranking for publications according to different centrality mass
‘uee Derreof Weighed ——Reteomne: Ege,
feoulty —deice of cently cenvalty
uty
Tans =e 7 ‘aniaas 100000
eee (200)
200)
‘comabll and 157 17 coisa rasa
(2008)
2005)
ave(2007)_ 303 tos comes asa7a04
areal (2010) 95 35 Santos aes
(2008)
Nien sad Hae 94 ” oor ozo7es
(20)
ammerete 9 = comes —o.s3s
(2013)
ang etal 2007) 52 oooores ozs
Livers ona) 0m06Ge ——103t6#
Peat etal 011) 8 omer
"age 2005)
201
(201
‘Sh (2008)
"Note(s: Ganvaliy measures for plications oa board versity ced witha the
network soured fom Boker, Ka, an Partai (2020) Dee of entuality =
‘he number of reaont tes an aril ora reseaieheonstinent (eg, auth,
county, instucion, journal) has in a nerwork. Weighted degree of centralicy
the total number of velatoal is an aii ot a veseaich constituent (©.
utr, county, astitutin, jousnal) has ina network mul the suength of
fact le. Betacenness canal ~a node's ably fo erry information betwen
tinconnetet gros of nodes In network, whetein eel node cepresens an
tude of teseatch consivent (eg, autor, county, insti, journal.
Eigen cenrlity~ the importance of the node in the netwouk thai responsible
for tansliing information to otber highly-connected nodes, wherein each
ode Teptesents an aiticle of tesesich eonsinent (ey aot, county,
Insinifn, journal.
‘» Berweenness centrality refers to @ node's ability to eatry Information
Derween unconnected groups of nodes, wherein each node represents
fa research constitent. Though betweenness centrality is more
complex than the degree of centrality, the measure does present i
formation about the role played by the research constituent in a
hetwork. In particular, berweenness centrality Is measured by
caleulating the toval number of shortest paths passing throxgh
prticnlar nods (5,»(t) and dividing it by the total number of
shortest paths in the entire network (5,0).
a = oe
«© Elgevector coal Is higher for nodes that are connected to other
highly-connected nodes, wherein each node represents a research
constituent. Specifically, « higher value of eigenvector centrality is
reflection of the importance of the node in the network that is responsible for transmitting information to other highly-connected
odes. This network metric is caleulated as
1
iY
where, Mf) is the set of neighbours of i, and 4 is a constant
1 Closeness centrality refers 0 the eapability of nodes to carry infor:
‘ation effectively by being closer to other nodes in the network. The
sum of distance of such nodes from otter nodes in the network in
dicates the relative ease for those nodes to carry information
ectvely.
PageRank alysis isan alternative measure ofa publication's impact
(Wing etal, 2000). Though PageRank was initially designed to pri
ortize web pages in Keyword search, the method lis found its way
‘o bibliomerres. n parienlar, PageRank ean be used to calculate the
prestige of publications that have an inflience on the research field
by influencing highly-cited publications despite not being highly
cited themselves. In that sense, a publication with ahigh PageRank is,
deemed as “high quality” and thus a “nus cite” among highly-cited
publicadons. m addition, PagoRank ean also be applied in clustering,
‘which willbe discussed in the next section, to reveal the themes nm
review domain, The formula for calculating PageRank is given as
follows
aa
w
PR
(C8
mi
ana)
rs)
wtere A isthe publication cited by highly-cited publications Ts, Ts, To,
1 Tae CCT) Is the citations of publteation T,, PRCT is the publieation’s
‘Tables
‘Sample tbe for publication rankings according to PageRank,
‘tile etn Global
‘steve wad Kroeber @ooi) oon) 289 5
‘on er Wotan ingly 2005) 0.08298 117 ”
rel nd ech (2005) ooskzes 267 135
en eal 2008) 0a 8 >
ow (2007) oer m2 eo
Cimpielland Miagsvern 0.021228 530, “
(2008)
Sesame a (2009) cone 6 s
Gorerest 2010) 013828 260 1
‘Beamer a (2007) oons 96 By
Bearers (2010) oon m2 =
Frncoer el (2008) ooinee 67 ss
Mose et a (2002) ooi2its a4 °
Nice an Hare (200) pote: 192 o
Gulerat 2010) eoiests 177 ry
aie et (2007) 0.003513 90 2
Kang e007) Sano 147 &
see (2009) oss 72 =
Tage eal 2020) oo0rses 99 5
Seni et (2011 overt 155 2
Note(s: PageRank of the cop 20 publications on board diversity sourced from
Bake, Kuma, sud Patna (2020), PageRank = « mete detived ftom thee:
tatons thar an ati eeltes om othe highly-ited stiles, which indeates
the prstige of tha aril. Global tations ~ the eitatons that an aril re
coves asi (viiout filtration). Local citations =the ciation that an article
‘ecelvs fom other altiles inthe veview corpus only (wih uation —Le, the
‘evew domi). Loeal citations ave genetally lower and ean never be higher
‘han global cations, wherein the oecreace ofthe ater i an indication a an
‘erroneous enty. The comparison of global and local cations ean eric un
essandng of vsearc ipat and infuenceas they reveal the seta or tue
Stat of ass (eg. ates with high global ciatons demonstate pact and
Inence aeros disciplines, wheteas alles with high local citation indicate
limpet at iene within the espn.
Jeena f Buns Rater 198 (2021) 285.206
PageRank, d is dampening factor, and N is the size of the network
“Table 5 presents a sample of publication raukings according. to
PageRank.
4.2, Chisering
Ccustering is another enrichment technique for bibliometic analysis
whose primary goal is to create theuatic of social clusters (depending
"upon the type of analysis being conducted). Curating network esters
and observing their development can be useful for understanding how @
research field manifests and develops. For example, the thematic clus
(ers created using co-citation analysis and bibliographic coupling shed
light on the major themes underpinning the intlleewal streture and
their davelopment aver tine in the research field, Several techniques
can be used for clustering such as exploratory Faccor aaa, hierar
chica clustering, Island algorithm, Louvain method, multidimensional
scaling, and sinple centers algorithm, whieh ean be complementary to
‘one another (aupie & Cater, 2019).
43, Visualization
‘The use of bibliometric analysis often goes hand fn hand with
network visualization sofware, which ranges from entirely graphical
user interface-based software such as VOSelewer (van Eck & Waltman,
2010) to command-based software such as Bibiomettx package in R
(sia & Cuccurullo, 2017). Other prominentl-used bibliometric sof
ware includes Bibexcel, Pajek, Geph,SeiMat, Se2, and UCINET. Fig. 8
and Fig. 4 present random exemplars of network visualization using
Vosviewer and Gephi, respectively
“Mest network visualization software are open source and free, and
thus, the choice of software reside with scholars. Each software bs its
‘awn pros and cons. For example, though Pajek aid UCINET have many
features, thee speed of development is slower as contpared to software
such as Gephi and R. The flexibility ofthe generated network is another
challenge. For example, in @ network generated using VOSviewer,
diferent forms of the same words cannot be merged, while the same ca
bee done using Gephi. One method 10 address this isue is 10 use bibio:
metric analysis and network visualization software in combination with
fone another. Infact, many bibliometric studies do take such a comple
mentary approach (o leverage on software strengths and to overcome
software shortcomings (eg., Baker etal, 20204; Dondu, Kusnar, Pan
dey, &© Gupta, 20215 Xu et al, 2018). Therefore, considerations per
taining to software feares and the flexibility ofthe resulting network
should be taken into account when making # decision ro choose a bib
ometrie or @ combination of bibliometric software for anslysis and
visualization
5. The bibliometric analysis procedure
In this section, the paper presents the steps for conducting bibl
metric analysis long with the general guidelines to be followed. Pg. 5
presents an lustration ofthe steps while Table 6 presents the guidelines
for bibliometric analysis with a focus on the specific (questions) re
ommendations regarding what scholars should ask themselves in each of
these steps.
5.1, Step 1: Define the aims and scope of the bibliomerc say
This first step isto define the sits and scope of the bibliometric
sy, which must occur before the selection of bibliometric analysis
techniques and the gathering of bibliometric data. Doing the later
before the former is risky as nsuitable aims and scope ean render bib
ometrie analysis useless, and thus, wasting precious resources that
cold be better invested with careful planning.
The ins ofa bibliometric study should relate to a retrogpection of the
performance and sclence of research fel. In terms of performance, Jeena f Buns Rater 198 (2021) 285.206
nutcoganagerent
ews
formance
“ats op@taupns and manag
Fi
3, Example of co-wotd (keyword co-occusence) network visualization using VOSviewer, Notes) Each node in @ network represents an entity atl,
‘author, county, lastitio, keyword, journal), an ln the case af Fig 3, «keyword, wheteln: (1) the se ofthe node indicates the occurence of he keyword (Le the
numberof times that te Keyword oe) (2) the nk Bewee ences represents the co-occurrence berwee keywords Keywords that eo-oeut or oceurtogete),
(2) the eickress of the fink signals the ocerence of co ecrreces ewe Reyworts (Le here of ties tat the Keywnds o-ose Foxe Ket) 4 he
igen, the greater he occurence of he keyword an (5) the ticker hein Between toes, the rater the ocurenceof ie co-occurrence eve eyo Each
calor represents a date cise, wherein the nodes and ks in tha csr canbe used to expla the tne’ (hster') coverage of pics (nodes) and te reatonsips
(dns) berwen he mes (odes) manesting ner thar theme (ser)
bibliometric studies ate often set ont to umpc the prolific research
‘constituents in the research field, which may inckide authors, in
stitutions, countries, and journals. In terms of sience, bibliometric
studies are usually designed to reveal the bibliometric structure that ex
‘capslates the networks berween research constients contebuting £0
the intellectual structure that i founded upon clusters of pertinent themes
in the research field
‘The scope for study should generally be large enough to warrant
bibliometric analysis because dhe analysis is designed 10 handle large
‘volumes of bibliometric data (Ramos Rodrigue & Ruiz Navarro, 2004),
‘To determine whether the scope of the study is adequately large,
scholars can review the number of papers that avail on the intended
‘researc eld for study. If there are considerable hundreds (e500 oF
ore) or thousands of papers, then tha researc field can be considered
to be large enough to warrant the se of bibliometric analysis f there
are only tens (g., 50) o low hundreds (e., 100-800) of papers, then
the researc field is considered to be stall ad thus do not warrant the
use of ibliontetrc analysis, as forcing the analysis on this smal compus
would be an overkil In this ease, alternative review methods such as
meta-analysis and systematie Lnererure reviews may be better suited,
5.2. Step 2: Choose the tectuques for bibliometric analysis
‘The second step is to design the bibliometric su, wherein the
techniques for bibliometric analysis are chosen to meet the aims and
scope ofthe study in the first step. One challenge that scholars often
‘encounter at this stage i the decision of whether to choose a tecique
based on the bibliometric data sought orto choose technique fest and
then prepare the bibliometric data nccording to that selected technique
thereafter. To overcome this challenge, this paper recommends the later
as the former mits the choiee of techniques that can be used by
scholars. Moreover, bibliometric data is often retrieved in a raw format,
rnd thus, scholars will need to clean and prepare that data recording to
the format that is required for the chosen bibliometric analysis tec.
niques. inthis regard, the paper's recommendation of the latter will
provide seholas with a wider rather than limiting selection of bib
Hometic analysis techniques for deployment. More importantly, the
choice of bibliometric analysis techniques will depend on the aims af the
study. For example, if the study intends to provide a review of the past
present, and future of research field with a lage biblionsetric corpus,
then & combination of co-citation analysis (pas), bibllograple coupling
(presen, and co-word analysis (eg, notable words inthe implication
find fture research directions of fll texts) (future) can be selected.
Whereas, ithe study is interested to wneover the thentes in general and
over specific periods, then the latter (Le. co-word analysis) can be used
In conjunction with chor keywords to enrich the analysis the former
two Ge, co-citation analysis and bibliographic coupling). As
mentioned, the performance analysis in bibliometric studies is akin to
the profile of participants in empirical studies, and thus, by default, the
components e.g. otal publications, total citations) ofthe performance
tnalysis should be selected nov, and analyzed and reported in @
descriptive (e., what it Is) yet analytical (Le., what I¢ means) way
thereafer Jeena f Buns Rater 198 (2021) 285.206
cnn nse
etn hee
manent is
Femme
sete
oe of :
mang hair
map ToFay Ween SHA LITRE Segest
Commit
een Comme Ba
Best
xo
Nu
cam
Pete Atreny
‘Community Ceattons
‘Corporate Soci Responsibility
Descriptive Norm
Other PP
‘Theory of Plamned Behavior
Corporate Sogial Marketing
Auutience Seumentation
Fig. 4. Example ofcoword (keyword co-occurence) network visualization using Gepl. Note(s) Each nade a @uetwoukrepresens an entity (eg, ail, amor,
Você também pode gostar
- The 7 Habits of Highly Effective People: The Infographics EditionNo EverandThe 7 Habits of Highly Effective People: The Infographics EditionNota: 4 de 5 estrelas4/5 (2487)
- The 7 Habits of Highly Effective People Personal WorkbookNo EverandThe 7 Habits of Highly Effective People Personal WorkbookNota: 4 de 5 estrelas4/5 (2515)
- Art of War: The Definitive Interpretation of Sun Tzu's Classic Book of StrategyNo EverandArt of War: The Definitive Interpretation of Sun Tzu's Classic Book of StrategyNota: 4 de 5 estrelas4/5 (3321)
- Influence, New and Expanded: The Psychology of PersuasionNo EverandInfluence, New and Expanded: The Psychology of PersuasionNota: 4.5 de 5 estrelas4.5/5 (730)
- Never Split the Difference: Negotiating As If Your Life Depended On ItNo EverandNever Split the Difference: Negotiating As If Your Life Depended On ItNota: 4.5 de 5 estrelas4.5/5 (3311)
- Pride and Prejudice: Bestsellers and famous BooksNo EverandPride and Prejudice: Bestsellers and famous BooksNota: 4.5 de 5 estrelas4.5/5 (20479)
- The Perfect Marriage: A Completely Gripping Psychological SuspenseNo EverandThe Perfect Marriage: A Completely Gripping Psychological SuspenseNota: 4 de 5 estrelas4/5 (1134)
- The Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good LifeNo EverandThe Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good LifeNota: 4.5 de 5 estrelas4.5/5 (20101)
- The Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good LifeNo EverandThe Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good LifeNota: 4 de 5 estrelas4/5 (5814)
![American Gods [TV Tie-In]: A Novel](https://imgv2-2-f.scribdassets.com/img/audiobook_square_badge/626321117/198x198/22ab6b48b6/1718315036?v=1){kind=icon}
