Resea ch A icle
Complexi y in he Accep ance o Sus ainable Sea ch
Engines on he In e ne : An Analysis o Unobse ed
He e ogenei y wi h FIMIX-PLS
Ped o Palos-Sanchez,
1
Felix Ma in-Velicia ,
1
and Jose Ramon Sau a
2
1
Depa men o Business Adminis a ion and Ma ke ing, Uni e si y o Se ille, Spain
2
Depa men o Business Economics, Rey Juan Ca los Uni e si y, Spain
Co espondence should be add essed o Jose Ramon Sau a; [email p o ec ed]
Recei ed 31 May 2018; Re ised 7 Augus 2018; Accep ed 16 Augus 2018; Published 9 Oc obe 2018
Academic Edi o : Ana Meš o ić
Copy igh © 2018 Ped o Palos-Sanchez e al. This is an open access a icle dis ibu ed unde he C ea i e Commons
A ibu ion License, which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided he o iginal
wo k is p ope ly ci ed.
This pape analyses he complexi y o use beha iou when acing he challenge o using sus ainable applica ions, such as In e ne
sea ch engines. This pape analyses an accep ance model using ex ended TAM (Technology Accep ance Model) wi h T us as an
added ex e nal a iable. I was sugges ed ha T us indi ec ly influences he final In en ion o Use wi h he pe cep ions o U ili y
and Ease o Use. To es he p oposed model, a su ey was ca ied ou wi h use s om diffe en geog aphical a eas o Spain (n= 445).
The second aim o his s udy was o unde s and he complexi y o ma ke ing segmen a ion by sepa a ing he applica ion use s in o
diffe en use g oups. Use s we e g ouped by hei p e e ence o a o i e In e ne sea ch engine. Unobse ed he e ogenei y was
s udied using FIMIX-PLS, and h ee diffe en use beha iou s wi h sea ch engines we e iden ified. These co esponded o he
numbe o inhabi an s who li e in he use a ea. In his way, he impac ha he en i onmen has on use choice, accep ance,
and use o his ype o sus ainable applica ions was shown. The esul s we e checked using PLS-SEM and showed ha he model
o he adop ion o sus ainable sea ch engines is explana o y and p edic i e because confidence and accep ance o his TAM
we e alida ed. The conclusions a e in e es ing o de elope s o en i onmen ally sus ainable and esponsible applica ions which
wan o coincide wi h cu en ends o ensu e ha use s p e e hem.
1. In oduc ion
Many companies ha e had o adap hei business o gani-
za ions o new echnological de elopmen s in he In e ne
[1]. In a wo ld ha is inc easingly global and in e con-
nec ed, finding in o ma ion ha can en ich a company
and allow i o ob ain a compe i i e ad an age is becoming
inc easingly impo an [2]. In addi ion, hese echnological
changes ha e also affec ed use s and ha e significan ly
changed consume s’li es.
In his global con ex , he inc easing complexi y o busi-
ness en i onmen s has led o he in oduc ion o new business
models, an imp o emen in global con ac s and ela ionships,
wi h easie access o in o ma ion. Businesses need o know
how o ake ad an age o hese new oppo uni ies. One way
o do his is o companies o use In e ne sea ch engines o
find in o ma ion abou diffe en p oduc s, se ices, ac i i ies,
o any in o ma ion ha is equi ed. The use o hese echno-
logical ad ances has changed use s’habi s and ways o acces-
sing in o ma ion as well as inc easing c ea i i y when sol ing
s a egic ma ke ing p oblems [3, 4].
Companies ha e ealized ha hey need o de elop effec-
i e ma ke ing s a egies in o de o ake ad an age o new
ends in consume beha iou . One o he ools ha can be
used o do his is he sea ch engine. Sea ch engines a e web-
si es ha index in o ma ion on he In e ne and o ganize i
acco ding o i s quali y o a use ’s sea ch c i e ia. Today,
he mos widely used sea ch engines wo ldwide a e Google
wi h 78.78% o he o al ma ke sha e, Bing wi h 7.65%,
Baidu wi h 7.33%, and Yahoo wi h 4.70% [5]. Each o hese
Hindawi
Complexi y
Volume 2018, A icle ID 6561417, 19 pages
h ps://doi.o g/10.1155/2018/6561417
sea ch engines has diffe en ea u es ha can be used o ana-
lyse and imp o e ma ke ing s a egies.
As well as conside ing global de elopmen s when c e-
a ing new business models, companies a e also ying o
use he plane ’s esou ces mo e sus ainably and use niche-
ma ke ing s a egies. The complexi y o hese ma ke ing
s a egies mus be analysed in o de o unde s and he new
consume [6, 7]. Resea ch has been done on a ious sus ain-
able sea ch engine ini ia i es called G een Sea ch Engines in
some s udies, which a e a s a egic mic o niche wi hin he
sec o [5, 8, 9].
Many ene gy-consuming compu ing esou ces a e
needed o a sea ch engine o be able o find in o ma ion
om anywhe e in he wo ld. These esou ces gene a e high
empe a u es ha can be mi iga ed wi h ai condi ioning ha
also consumes elec ici y. In ac , Google says ha each que y
equi es a ound 1 kJ o 0.0003kWh o ene gy [10].
Sus ainable sea ch engine is he name gi en o a sea ch
engine ha gi es pa o all o i s p ofi s o sus ainable social
and en i onmen p ojec s. A sea ch engine’s p ofi s usually
come om ad e ising in he sea ch esul s [5, 11].
Wi h he amoun o in o ma ion ha exis s on he In e -
ne , new sus ainable business models ha e been de eloped
o sea ch engines in ecen yea s [12]. I is impo an o no e
ha he well-known echniques o SEO, Sea ch Engine Op i-
miza ion, and SEM, Sea ch Engine Ma ke ing, a e widely
used. The fi s echnique op imizes he in o ma ion gi en
in he esul s o any sea ch eques , and he second echnique
is used o p oduce economic benefi s. These economic bene-
fi s a e ea ned om sponso ed sea ch esul s (paid sea ch)
ha a e financed by ad e ise s using CPC (cos pe click)
o CPM (cos pe housand imp essions) o any o he ype
o paymen me hod [13, 14]. This ype o ad e ising also
allows effec i e ma ke ing s a egies o be used, since i col-
lec s usage and na iga ion da a abou new consume ends,
which can be used o modi y ma ke ing.
As a gene al ule, SEM-sponso ed sea ch esul s finance
and sus ain diffe en ypes o sus ainable p ojec s [15]. Some
sus ainable sea ch engines a e desc ibed below. Ecosia allo-
ca es 80% o SEM ad e ising e enue o ee e o es a ion
p ojec s a ound he wo ld. Solyda helps sus ainable de el-
opmen p ojec s as well as sec o s o he socially disad an-
aged popula ion. Goodsea ch encou ages use s o hei
sea ch engine o accumula e $5 uni s o c edi ha hey
can la e dona e o sus ainable de elopmen p ojec s. Lilo
is a sea ch engine ha dona es d ops o wa e ha a e accu-
mula ed by use s e e y ime hey sea ch using his sea ch
engine. Benefind makes a dona ion o 0.5 cen s each ime
someone sea ches using hei sea ch engine. Fo es le dona es
90% o i s p ofi s o sus ainable and social de elopmen
agencies o p ojec s such as T eeho, which is simila o he
Ecosia model o plan ing ees as a esul o using he sea ch
engine [5, 16, 17].
The pu pose o his s udy is o use he TAM wi h T us as
an ex e nal a iable o iden i y diffe en g oups o sus ainable
sea ch engine use s. A FIMIX-PLS analysis and a pos hoc
analysis a e ca ied ou in o de o iden i y diffe en beha -
iou when use s adop a sus ainable sea ch engine and o
de ec new ends in consume beha iou . In his way, a la ge
amoun o in o ma ion can be used o commen on how ma -
ke ing can ace u u e echnological challenges as use s ake
ad an age o en i onmen ally sus ainable echnologies.
2. Theo e ical Backg ound
O e he las decade, esea che s ha e ollowed a ious lines
o esea ch in he a eas o sea ch engine accep ance, use s’
eelings abou diffe en sea ch engines, and he diffe en
op ions a ailable in he ma ke (see Table 1 [18]).
Sánchez e al. [19] in es iga ed he e olu ion o sea ch,
he numbe o sea ches made, and he consis ency o any
expec ed esul when using diffe en sus ainable sea ch
engines. Likewise, Ma ínez-Sanahuja and Sánchez [20] ca -
ied ou esea ch on sea ch engine sus ainabili y o disco e
how sus ainable p og ams affec he use s’opinion and also
e iew he main ini ia i es o sus ainable sea ch engine
since 1994.
In he esea ch by Hahnel e al. [26], bo h adi ional and
sus ainable sea ch engines we e s udied o find he ac o s
which influence use s’choice o sea ch engines. Liaw and
Huang [27] sugges ed a model o in es iga e he me hods
used o find in o ma ion wi h sea ch engines and iden i y
how hese sea ches can be made mo e efficien .
Fo una i and O’Sulli an [28] showed he impo ance o
new media and new echnologies ha a e p o ided by digi al
al e na i es. Sus ainable social de elopmen was s udied wi h
special impo ance placed on how use s beha e wi h hese
new digi al al e na i es in o de o find ways o imp o e hem
Table 1: Rela ed wo ks.
Au ho s Desc ip ions
Chao e al. [21] P esen an in es iga ion o he pa icipa ing agen s when use s sea ch o in o ma ion wi h sea ch engines,
especially s udying he belie s and isks ha a e aken in hei sea ches
Palos-Sanchez and
Sau a [5]
Analyse he Ecosia sus ainable sea ch engine using he Unified Theo y o Accep ance and Use o Technology
(UTAUT) and hen analysing he esul s wi h PLS-SEM (Pa ial Leas Squa es-S uc u al Equa ion Modelling)
Rangaswamy e al. [22] Resea ch he diffe en s a egic pe spec i es o sea ch engines om he poin o iew o sus ainabili y and
sus ainable de elopmen
Kei s ead [23] In es iga es he sea ches made wi h sus ainable sea ch engines and he use beha iou
Liaw e al. [24] Use he TAM wi h PLS o find wha he use s eel abou he in o ma ion ound wi h diffe en sea ch engines
Kamis and S oh [25] De elop a PLS model o de e mine he impo ance o sea ch engines when an online pu chase is made, using
ac o s such as pu chase decision, us o pe cei ed u ili y, and he beha iou o diffe en use s
2 Complexi y
[29]. Jaca e al. [30] showed he impo ance o businesses o
conside ing socie y and use s’ espec o he en i onmen .
They poin ed ou ha sus ainable de elopmen can be
unde s ood by analysing use beha iou o sus ainable o ga-
niza ions [23].
In addi ion, Hi su [29] in es iga ed he cul u al ac o s
ha influence he choice o sea ch engine o diffe en
sea ches made by use s. The beha iou o diffe en ypes o
use s o sea ch engines was in es iga ed in o de o de e -
mine beha iou pa e ns and consequen ly p edic hem [28].
3. Resea ch Model and
Hypo heses De elopmen
A e analysing diffe en models and heo ies o echnologi-
cal accep ance, he TAM wi h T us as an added ex e nal a -
iable was chosen o his in es iga ion. TAM was chosen
because i has been shown o be a eliable model o measu -
ing he accep ance and use o echnologies as well as o he
beha iou o use s. The main cons uc s in he model explain
use s’a i udes owa ds using echnology, and he TAM has
been used o in es iga e use s’a i udes owa ds al e na i e
echnologies. Re iewing diffe en esea ch ha used he
TAM, wi h added ex e nal a iables, o accep heo ies
helped in he choice o his model. In he nex sec ion, he e
is an explana ion o each o he a iables and ela ionships
used in he model o analyse he hypo heses.
4. Technology Accep ance Model
(TAM) Va iables
The TAM es ablishes casual ela ionships be ween pe cei ed
use ulness (PU), pe cei ed ease o use (PEOU), a i ude
owa d using (ATU), and in en ion o use (USE) [31]. Fol-
lowing he esea ch o Da is [32], in which he model was
p oposed o he fi s ime, pe cei ed use ulness (PU) and
pe cei ed ease o use (PEOU), ha a e no implici ly
included in TAM, a e expec ed o influence a i ude owa d
using (ATU) and beha iou al in en ion o use [33]. In his
s udy, he ex e nal a iable T us was also included. T us is
defined as he confidence ha use s ha e in echnology and
links he eliabili y o hei implici ac ions wi h echnology
when hey use i [5]. ATU e e s o a use ’s posi i e o nega-
i e eelings owa d he use o any gi en echnology, while
BUSE is he amoun o p io use gi en o he echnology
[34]. PU is defined as how much an indi idual belie es ha
using a pa icula sys em will imp o e hei pe o mance
[32]. I is a measu e o he subjec i e likelihood ha a
po en ial use will inc ease hei wo k pe o mance in an
o ganiza ion when using he echnology [35]. The PEOU
a iable measu es how much an indi idual belie es ha
using a pa icula sys em is effo - ee. Diffe en au ho s
ha e also p e iously used he ex e nal a iable, T us , in
he TAM [36]. T us is an ex e nal a iable o he model
and has been defined by p e ious esea ch in a a ie y o
ways, bo h heo e ically and ope a ionally.
Palanisamy [37] demons a ed and de eloped a model
o he accep ance o diffe en sea ch engines and linked
he influence o PU wi h USE. Liaw and Huang [27] s udied
he influence o PU on ATU o unde s and use s’a i udes
owa ds using sea ch engines and he pe cei ed u ili y o
he diffe en sea ch engines. Using he s udies abo e, we p o-
pose he ollowing hypo hesis.
H1 Pe cei ed use ulness (PU) influences in en ion o use
(USE) sus ainable sea ch engines on he In e ne .
Lim and Ting [38] de eloped a echnology accep-
ance model o sea ch engines ha a e used in e-
comme ce web pages. A clea ela ionship was ound
be ween PU and ATU when using he sea ch engine.
Kou a is [39] s udied use beha iou when making
que ies wi h hese sea ch engines and in es iga ed
he ela ionship be ween PU and ATU when accessing
a web page as a esul o using a sea ch engine. Using
hese s udies, we p opose he ollowing hypo hesis.
H2 Pe cei ed use ulness (PU) influences a i ude
owa ds using (ATU) o sus ainable sea ch engines
on he In e ne .
Mo osan and Jeong [40] used he TAM o s udy he
adop ion o sea ch engines o booking ho els and
es au an s and esea ched he influence o he
PEOU and PU a iables when using hese sea ch
engines o achie e a elle s’goals. Yang and Kang
[41] showed he influence o USE and PU a iables
o sea ch engines in Thailand and used hem in
he UTAUT (Unified Theo y o Accep ance and
Use o Technology) model. Using his li e a u e, we
p opose he ollowing hypo hesis.
H3 Pe cei ed ease o use (PEOU) influences pe cei ed
use ulness (PU) o sus ainable sea ch engines on he
In e ne .
Hsu and Wal e [42] in es iga ed he ela ionship o
he ease o use and he pe cei ed use ulness o sea ch
engines when looking o con en on web pages. They
p oposed a ela ionship be ween PEOU and ATU
using he echnology accep ance model. Chi-Yueh
e al. [43] explo ed he in en ion o use s o use
sea ch engines o find audio and ideo con en on
he In e ne and analysed he influence o PEOU on
ATU. Using hese in es iga ions, we p opose he
ollowing hypo hesis.
H4 Pe cei ed ease o use (PEOU) influences a i ude
owa d using (ATU) sus ainable sea ch engines on
he In e ne .
Moon and Kim [44] and Ge en e al. [45] used he
TAM o s udy sea ch engines and online s o es on
he In e ne . In his esea ch, he influence o a i ude
owa d using (ATU) on in en ion o use (USE) In e -
ne sea ch engines was s udied. Following hese in es-
iga ions, in which he TAM was adap ed o sea ch
engines, we p opose he ollowing hypo hesis.
H5 A i ude owa d using (ATU) influences in en ion o
use (USE) sus ainable sea ch engines on he In e ne .
3Complexi y
Hsu and Wal e [42] adap ed he TAM o sea ch
engine use by adding he T us a iable and hen
linking his o PU. To do his, he influence ha a i-
ude has on use, when a use us s he sea ch engine,
was measu ed [44]. Palanisamy [37] also included
he T us a iable in he model, in o de o find he
eliabili y o sea ch engines and hei echnological
accep ance. Using his esea ch on sea ch engines,
we p opose he ollowing hypo hesis.
H6 T us influences pe cei ed use ulness (PU) o sus-
ainable sea ch engines on he In e ne .
Lim and Ting [38], Palanisamy [37], and Hsu and
Wal e [42] also analysed he influence ha T us ,
as an ex e nal a iable, has on PEOU in accep ance
models and e ealed he influence o bo h a iables
o sea ch engines [5, 45]. The e o e, he ollowing
hypo hesis was p oposed.
H7 T us influences pe cei ed ease o use (PEOU) o
sus ainable sea ch engines on he In e ne .
5. He e ogenei y and Segmen a ion
In Social Sciences, i is difficul o gua an ee ha he whole
sample fi s he same p obabilis ic dis ibu ion. Howe e , wi h
PLS, segmen a ion can be used wi h he s uc u al model,
which means ha diffe en pa ame e s a e used o sepa a e
he sample in o g oups [46].
He e ogenei y in he da a may o may no be obse ed.
He e ogenei y is obse ed when he diffe ences be ween
wo o mo e g oups o da a a e caused by obse able cha ac-
e is ics, such as sex, age, o coun y o o igin. On he o he
hand, unobse ed he e ogenei y a ises when he diffe ences
be ween wo o mo e da a g oups do no depend on any
obse able cha ac e is ic o combina ions o cha ac e is ics.
Howe e , he e can s ill be significan diffe ences in he ela-
ionships be ween da a g oups in he model, when he o igins
o hese diffe ences canno be a ibu ed o any obse able
a iable such as age, gende , educa ional le el, o any o he
ype [47].
In ou s udy, hese obse able cha ac e is ics we e used o
di ide he da a in o sepa a e g oups o in es iga ion and
hen analysed wi h a g oup-specific PLS-SEM me hod. To
do his, he a iables used o he g ouping o he sample
had o be ound. Once hese we e iden ified, he ela ionships
be ween hese g oupings could be es ablished and analysed.
The e a e es ablished echniques o his p ocess, bu
p e ious esea ch has shown ha adi ional g ouping ech-
niques do no wo k e y well o he iden ifica ion o g oup-
ing diffe ences [48]. Me hodological esea ch wi h PLS-
SEM has esul ed in a mul i ude o diffe en echniques,
commonly e e ed o as la en class echniques, o iden i y
and ea unobse ed he e ogenei y. These echniques ha e
p o ed o be e y use ul o iden i ying unobse ed he -
e ogenei y and g ouping he da a acco dingly [47].
TAM was used o his in es iga ion in o he adop ion
o sus ainable sea ch engines on he In e ne , and he
numbe o segmen s was es ablished so ha i was small
enough o gua an ee pa simony and la ge enough o gua an-
ee s a egic ele ance [49].
The echnique chosen o s udy unobse ed he e ogenei y
was FIMIX-PLS [50], ex ended by Sa s ed e al. [51]. FIMIX-
PLS is he mos used la en class app oach o PLS-SEM [52]
and is an explo a o y ool ha esul s in he app op ia e
numbe o segmen s in o which he sample should be
di ided. The FIMIX-PLS echnique allowed decisions o be
made abou he numbe o segmen s using p agma ic eason-
ing and p ac ical issues iden ified in cu en esea ch [53].
FIMIX-PLS is he mos widely used echnique and has
been used in a ious a eas o esea ch, such as en i onmen-
al posi ioning o businesses [46], In e ne usage by SMEs
[54, 55], ou ism managemen [56], s a egic ma ke ing
managemen [57], co po a e epu a ion [48, 58], mobile
shopping [59], and lea ning sys ems [60].
FIMIX p oposes an es ima ed pa h model using he
PLS-PM algo i hm. The esul ing la en a iable alues a e
used in he FIMIX-PLS algo i hm o find any unobse ed
he e ogenei y in he es ima ed pa ame e s o he in e nal
model ( ela ionships be ween la en a iables).
6. Da a and Me hodology
Table 2 shows he demog aphic cha ac e is ics o he sam-
ple (n= 445). I can be seen ha mos o he sample a e
young people aged 19–30 (81.3%) who a e s uden s
(75.3%) a uni e si y (73.0%) and use sea ch engines wi h
sma phones (91.2%). The pe cen ages o men (41.6%)
and women (56.8%), as well as he habi a s, we e mo e
equally p opo ioned.
The da a collec ion echnique chosen o his s udy was
he su ey, which is a quan i a i e echnique. In his case, i
allowed us o iden i y he use s’a i udes and beha iou
when using sus ainable and unsus ainable sea ch engines
on he In e ne . A 15-i em ques ionnai e abou a i udes
and beha iou and 5 classifica ion ques ions we e used. The
classifica ion ques ions we e abou gende , age, job, habi a ,
educa ion le el, and he de ice used o In e ne access. The
ques ionnai e was di ided in o 3 sec ions.
The fi s sec ion deal wi h he ques ions o he TAM
[32] abou In e ne sea ch engine echnology and he use s’
eelings, a i udes, and beha iou o he adop ion and use
o sus ainable sea ch engines. This sec ion was composed o
12 ques ions abou PU (3), PEOU (3), ATU (3), and USE
(3). The TAM a iables we e measu ed using adap ed i em
scales [32].
The second sec ion consis ed o a block o ques ions on
diffe en aspec s o T us and sus ainable sea ch engines.
These ques ions we e g ouped in o he 3 i ems in he T us
cons uc . The beha iou al i ems abou sus ainable sea ch
engines on he In e ne we e adap ed om p e ious esea ch
on T us , in which he T us a iable e e s o how much a
use belie es in he sa e y, eliabili y, efficiency, compe ence,
and alidi y o a sus ainable sea ch engine [5]. The beha -
iou al i ems o sus ainable sea ch engines e e o he
momen when a use finds a se ice o be un eliable and
in e ac s less wi h he sea ch engine, con en , o in o ma ion.
4 Complexi y
In he s udy o he beha iou al in en ion o use a sea ch
engine, T us is defined as he gene al belie ha hese
sea ches will be made [37, 38, 42, 44, 45].
The e we e 20 i ems in he esea ch ques ionnai e (see
Table 3). All he i ems, excep o he classifica o y ques ions,
we e measu ed using a Like 5-poin scale ha anged om
o al disag eemen [61] o o al ag eemen [62].
O e all, 445 ques ionnai es we e collec ed om he use s.
Google Fo ms was used because he ques ionnai e could be
p oduced online and hen dis ibu ed on social ne wo ks.
Nonp obabilis ic and con enience sampling was used, and
a pilo su ey was ca ied ou o check he alidi y and eli-
abili y o he scales. In his way, he ques ions could be
efined and addi ional commen s on he con en and s uc-
u e o he ques ionnai e we e ob ained. All he pa icipan s
in he su ey we e asked o wa ch he ideo ha accompanied
he ques ionnai e.
The PLS-SEM me hod was used o he analysis. This is a
s a is ical analysis echnique based on he S uc u al Equa-
ion Model, which is a ecommended me hod o explo a o y
esea ch as i allows he modelling o la en cons uc s wi h
indica o s [63] o analyse he collec ed da a. PLS is
app op ia e o he analysis and p edic ion o ela i ely new
phenomena [64]. Fo his s udy, we used he Sma PLS 3
so wa e [65]. The esul s we e handled wi h he s a is ical
package SPSS 24, which was used o calcula e equency
ables, CHAID ee, ANOVA, and sample s a is ics.
To find he minimum sample size o PLS modelling,
Hai e al. [66] ecommend using he Cohen ables [67].
These ables we e used wi h he G∗Powe so wa e package
[68] o find he dependen cons uc s, which a e hose ha
ha e he highes numbe o p edic o s. In his case, hey we e
PU, ATU, and USE. The ollowing pa ame e s we e used o
he calcula ion: he es powe (powe = 1 −βe o p ob. II)
and he size o he effec ( 2). Cohen [69] and Hai e al.
[70] ecommend a powe o 0.80 and an a e age size o he
effec 2=015. In ou case, he e we e 2 p edic o s, which
we e he cons uc s ha ha e causal ela ionships wi h USE
(see Figu e 1). The e o e, om PLS, he USE cons uc es ab-
lished he minimum sample size as 107 o a powe = 0 95
and c i ical F=308. The e o e, he sample used is ade-
qua e because i is mo e han ou imes he ecommended
minimum o ob aining alid and eliable esul s wi h he
es ablished pa ame e s.
7. Analysis o Resul s
7.1. Measu emen Model E alua ion. Be o e he PLS anal-
ysis was ca ied ou , he alidi y and eliabili y o he mea-
su emen model we e calcula ed wi h he ollowing es s:
indi idual eliabili y o each i em, in e nal consis ency (o
eliabili y) o each scale (o cons uc ), con e gen alidi y,
and disc iminan alidi y.
7.1.1. The Indi idual Reliabili y o he I ems: Cons uc Loads
(λ). In his phase o he in es iga ion, he indica o s’loads (λ)
we e calcula ed, wi h he minimum accep ance le el o pa
o he cons uc λ≥0707 [71]. The e o e, a alue λ≥0 707
indica es ha each measu emen ep esen s a leas 50%
(0.7072= 0.5) o he a iance o he unde lying cons uc
[72]. The indica o s ha did no each he minimum le el
we e dis ega ded [73].
The magni ude and impo ance o he ela ionships
be ween la en a iables we e calcula ed using he s anda d-
ized pa h coefficien . The ule es ablished by Chin [74] s a es
ha his alue mus be a leas 0.2 (see Figu e 1).
C onbach’s alpha and he composi e eliabili y (CR, com-
posi e eliabili y) we e hen calcula ed o find he eliabili y o
each cons uc . This e alua ion measu es he consis ency o a
cons uc based on i s indica o s [75], ha is, he igo wi h
which hese i ems a e measu ing he same la en a iable.
The lowe limi o he accep ance o he cons uc eliabili y
using C onbach’s alpha is usually be ween 0.6 and 0.7 [76].
Causali y is ound om he loads o he indica o s and he
composi e eliabili y (CR) [77] which mus ha e a minimum
le el o 0.7 [62, 78, 79].
Table 4 shows he esul s o all he eliabili y coefficien s.
As can be seen, all he coefficien s had much highe alues
han he necessa y minimum limi s, which confi ms he high
in e nal consis ency o all he la en a iables.
Table 2: Demog aphic cha ac e is ics o he sample (n= 445).
Classifica ion a iable F equency Pe cen age
Gende
Female 253 56.8%
Male 185 41.6%
O he s 7 1.6%
Age
18–30 362 81.3%
31–45 49 11.0%
46–55 27 6.1%
56–65 6 1.4%
>65 1 0.2%
Job
Unemployed wo ke 13 2.9%
Sel -employed wo ke 24 5.4%
Con ac ed wo ke 58 13.7%
S uden 335 75.3%
Housewi e 8 1.8%
Re i ed 4 0.9%
Habi a
Town wi h mo e han
100,000 inhabi an s 142 31.9%
F om 20,000 o
100,000 habi an s 153 34.4%
Less han 20,000
habi an s 149 33.7%
Educa ion le el
Basic s udies
(O-le els) 77 17.3%
P o essional
aining/A-le els 40 8.7%
Uni e si y deg ee 325 73.0%
Access o
In e ne om
Sma phone 406 91.2%
Table o iPad 117 26.3%
Lap op 270 60.7%
Pe sonal compu e 47 10.6%
5Complexi y

Pe cei ed
Use ulness
(PU)
A i ude
owa d using
(ATS)
In en ion
o use (USE)
Pe cei ed ease o
use (PEOU)
T us (T)
H5
0.272
H3
0.062
H2
0.408
H4
0.215
H1
0.482
H6
0.323
H7
0.574
0.810 0.841 0.845 0.845 0.815 0.799
0.853
0.840
0.819
0.892
0.889
0.875
0.874 0.904 0.826
Figu e 1: P oposed esea ch model and PLS esul s.
Table 4: Measu emen model.
Reliabili y o each cons uc Fo nell & La cke c i e ion
Cons uc s C onbach’s alpha ho_A CR AVE ATS USE PEOU PU TRUST
ATS 0.756 0.757 0.860 0.672 0.820
USE 0.830 0.833 0.898 0.746 0.572 0.864
PEOU 0.703 0.700 0.828 0.616 0.535 0.461 0.785
PU 0.756 0.756 0.860 0.673 0.383 0.496 0.325 0.820
TRUST 0.862 0.862 0.916 0.784 0.386 0.423 0.364 0.637 0.885
Table 3: I ems and scale.
Cons uc I ems
A i ude owa d using (ATU)
(ATU1) My a o i e sea ch engine p o ides access o mos da a.
(ATU2) My a o i e sea ch engine is be e han p e ious sea ch engines.
(ATU3) My a o i e sea ch engine p o ides accu a e in o ma ion.
(ATU4) My a o i e sea ch engine p o ides in eg a ed, up- o-da e, and eliable in o ma ion.
Pe cei ed ease o use (PEOU)
(PEOU1) In e ac ion wi h my a o i e sea ch engine se ices is clea and easily unde s ood.
(PEOU2) Wo king wi h my a o i e sea ch engine does no equi e much men al effo .
(PEOU3) My a o i e sea ch engine se ices a e easy o use.
(PEOU4) I can easily find wha I wan in my a o i e sea ch engine.
Pe cei ed use ulness (PU)
(PU1) Using my a o i e sea ch engine allows asks o be comple ed mo e quickly.
(PU2) Using my a o i e sea ch engine imp o es wo k pe o mance.
(PU3) Using my a o i e sea ch engine inc eases wo k p oduc i i y.
(PU4) Using my a o i e sea ch engine imp o es wo k effec i eness.
In en ion o use (IU) (IU1) I am going o use my a o i e sea ch engine.
(IU2) I expec he in o ma ion p o ided by my a o i e sea ch engine o be use ul.
T us (T)
(T1) My In e ne sea ch engine is us wo hy.
(T2) My In e ne sea ch engine akes i s use s’ideas in o accoun .
(T3) My In e ne sea ch engine has good in en ions.
6 Complexi y
7.1.2. Disc iminan and Con e gen Validi y. AVE (a e age
a iance ex ac ed) is defined as he mean ex ac ed a iance
and measu es how much a iance he indica o s o a con-
s uc ha e compa ed o he amoun o a iance due o he
measu emen e o [80]. The ecommenda ion o hese
au ho s is ha AVE is ≥0.50. The ho_A coefficien [81]
shows ha in all cons uc s i is ≥0.7.
The disc iminan alidi y shows how much one cons uc
is diffe en om ano he . A high alue indica es weak co e-
la ions be ween cons uc s. Fo his es , he Fo nell &
La cke [80] c i e ion is used, which e ifies i he squa e oo
o he a e age a iance ex ac ed (AVE) o a cons uc is
g ea e han ha o he ela ionship be ween he cons uc
and he es o he model’s cons uc s. This condi ion was
me as can be seen on he igh side o Table 2.
Table 5 shows he esul s ha we e ob ained, whe e i can
be seen ha all he HTMT ela ionships o each pai o ac-
o s a e <0.90 [82, 83]. The ulfilmen o all hese c i e ia and
measu emen s means ha he alidi y and eliabili y o he
model a e confi med.
7.2. Assessmen o he S uc u al Model. The ollowing analy-
ses we e used o s udy he s uc u al model, he explained
a iance o he endogenous cons uc s (R2), he p edic i e
capaci y Q2, he pa h coefficien s (β), and he selec ion o
c i ical alues o he dis ibu ion o S uden ’s - alue [84].
Hensele e al. [72] conside he explana o y powe o
R2 alues o 0.67, 0.33, and 0.19 o be subs an ial, mode a e,
and weak, espec i ely. In Table 4, we can see ha PU
(R2=0416), ATS (R2=0335), and USE (R2=0417) ha e
a mode a e explana o y powe , while PEOU has a weak
explana o y powe (R2=0133).
7.3. Model and Hypo hesis Tes ing. The model was hen
analysed using he boo s apping echnique. Using his ech-
nique, he s anda d de ia ion o he pa ame e s and he
S uden - alues a e ound. F om hese, he simple eg ession
coefficien s o he componen s a e calcula ed, and he esul s
o he ela ionships be ween he la en a iables o he
hypo heses a e ound.
A his s age, he hypo heses we e es ed o see i he ela-
ionships es ablished in he p oposed model we e confi med
[84]. Fi s ly, all he ela ionships be ween cons uc s had a
significan impac on he beha iou al in en ion o use he
sea ch engine (see Table 6). The e o e, he p oposed TAM
was suppo ed oge he wi h he ex e nal T us a iable.
All he hypo heses we e suppo ed wi h a 99.9% confidence
le el, excep H3. The ela ionship be ween PEOU →PU was
he leas significan wi h a 95% confidence le el (β=0107,
=2628).
The ela ionships ha s ood ou mos s ongly we e, in
o de , H7: TRUST →PU (β=0598; =14622) and H2:
PEOU→ATS (β=0459; =10675).
7.4. Resul s o FIMIX-PLS: S udy o Unobse ed
He e ogenei y. FIMIX-PLS calcula es he p obabili y o
belonging o any gi en segmen in which each obse a ion
is adjus ed o he p ede e mined numbe o segmen s by es i-
ma ing sepa a e linea eg ession unc ions, which gi es a
g oup o possible segmen s. Each case is assigned o he seg-
men wi h he g ea es p obabili y.
The es is done in ou s ages: fi s ly, he numbe o
op imal segmen s is calcula ed wi h FIMIX. Then, he la en
a iables ha jus i y hese segmen s a e ound, in o de o
finally es ima e he model and i s segmen s.
FIMIX was used o di ide he sample in o diffe en seg-
men s. The fi s p oblem encoun e ed was he selec ion o
he app op ia e numbe o segmen s. I is ypical o epea
he FIMIX-PLS p ocedu e wi h consecu i e numbe s o
la en classes. In ou case, gi en he sample size n= 445,we
calcula ed o k=5,k=4,k=3, and k=2. The esul s
ob ained we e compa ed using diffe en in o ma ion c i e ia
p o ided by he FIT indices. The ollowing we e compa ed,
Akaike (AIC), he con olled AIC (CAIC), he Bayesian
in o ma ion c i e ion (BIC), and he s anda dized en opy
s a is ic (EN). The esul s ob ained o he FIT indices a e
shown in Table 7.
Fi s ly, he FIMIX es was used o find he numbe o
segmen s in o which he sample can be di ided. The algo-
i hm was configu ed o he size o he sample so ha
PLS-SEM could be applied wi h 10 epe i ions. This con-
figu a ion was done using he expec a ion maximiza ion
algo i hm (EM). The EM algo i hm al e na es be ween pe -
o ming an expec a ion s ep (E) and a maximiza ion s ep
(M) [47]. S ep E e alua es and uses he cu en es ima ion
o he pa ame e s. S ep M calcula es he pa ame e s maxi-
mizing he loga i hmic egis a ion p obabili y ound in s ep
E. S eps E and M a e applied successi ely un il he esul s a e
Table 5: HTMT and explana o y and p edic i e capaci y o he
model.
Cons uc s ATS USE PEOU PU R2
(wi h effec le el) Q2
ATS 0.335 (mode a e) 0.211
USE 0.716 0.417 (mode a e) 0.293
PEOU 0.574 0.528 0.133 (weak) 0.070
PU 0.393 0.509 0.292 0.416 (mode a e) 0.266
TRUST 0.479 0.570 0.378 0.692 ——
Table 6: S a is ical hypo hesis es .
Hypo heses Pa h βpa h coefficien s
( - alues)
p
alue Suppo ed
H1 ATS USE 0.448 (8.877)∗∗∗ 0.001 Yes
H2 PEOU ATS 0.459 (10.675)∗∗∗ 0.001 Yes
H3 PEOU PU 0.107 (2.180)∗0.029 Yes
H4 PU ATS 0.234 (5.395)∗∗∗ 0.001 Yes
H5 PU USE 0.324 (6.851)∗∗∗ 0.001 Yes
H6 TRUST
PEOU 0.364 (8.704)∗∗∗ 0.001 Yes
H7 TRUST PU 0.598 (14.622)∗∗∗ 0.001 Yes
No e: Boo s apping wi h 5000 samples based on he S uden -dis ibu ion
(499) in single queue: ∗p<005 ( 0 05 ; 499 =164791345); ∗∗ p<001
( 0 01 ; 499 =2333843952); ∗∗∗p<0001 ( 0 001 ; 499 =3106644601).
7Complexi y
s abilized. S abiliza ion is achie ed when he e is no subs an-
ial imp o emen in he alues ob ained.
Table 7 shows he esul s a e unning FIMIX wi h
diffe en numbe s o kpa i ions. Since he numbe o seg-
men s was unknown a p io i, he diffe en segmen num-
be s we e compa ed in e ms o sui abili y and s a is ical
in e p e a ion [85, 86].
A pu ely da a-based app oach was aken, which only p o-
ided an app oxima e guide o he numbe o segmen s ha
should be selec ed. Heu is ics, such as he in o ma ion c i-
e ia and he EN, a e allible because hey a e sensi i e o
he da a and he cha ac e is ics o he model [47].
The diffe en c i e ia ob ained we e hen e alua ed.
Sa s ed e al. [51] e alua ed he effec i eness o diffe en
in o ma ion c i e ia in FIMIX-PLS o a wide ange o
da a cons ella ions and models. Thei esul s showed ha
esea che s should conside AIC 3 and CAIC. As long as
hese wo c i e ia indica e he same numbe o segmen s,
he esul s p obably poin o he app op ia e numbe o seg-
men s. In Table 6, i can be seen ha in ou analysis hese
esul s do no poin o he same numbe o segmen s. The e-
o e, AIC was used wi h ac o 4 (AIC 4, [87]) and BIC. These
indices usually wo k well and, in ou case (see Table 6), hey
indica ed he same numbe o segmen s, which was k=4.
O he c i e ia showed his as a p onounced o e es ima ion,
al hough MDL5 indica ed he minimum numbe o segmen s
k+1, which in his case would indica e 3 [47].
Measu emen s o en opy, such as he s anda dized
en opy s a is ic (EN), we e also conside ed [88]. EN uses
he p obabili y ha an obse a ion belongs o a segmen o
indica e whe he he pa i ion is eliable o no . The highe
he p obabili y o belonging o a segmen is o a measu e-
men , he clea e segmen affilia ion is. The EN index oscil-
la es be ween 0 and 1. The highes alues indica e a be e
quali y pa i ion. P e ious esea ch p o ided e idence ha
EN alues abo e 0.50 allow a clea classifica ion o he da a
in o he p ede e mined numbe o segmen s [89, 90]. In
Table 6, i can be seen ha all he pa i ions ha e alues o
EN>0.50, al hough he highes alue is eached in k=2wi h
EN= 0.998; o k=3EN= 0.819, and EN=0.717 o k=4.
The e o e, om he p oposed solu ions, he numbe o
op imal segmen s was be ween k=3 and k=4.k=3 was
aken as he numbe o segmen s indica ed by FIMIX-
PLS, gi en ha he smalles size o he pa i ions in his
case was 12.1%.
As can be seen in Table 8, o he k=3 solu ion and a
sample n= 445, he pa i ioning o he segmen s was 58.2%
(259), 29.6% (131), and 12.1% [91, 92]. The segmen sizes
a e no small despi e he pe cen ages. The e o e, he sample
sizes a e sufficien o use PLS. The sample size can be consid-
e ably smalle in PLS han in SEM due o co a iance [47].
The e can e en be mo e a iables han obse a ions, and
he e may be a small amoun o da a ha is comple ely miss-
ing [46, 93]. Diffe en au ho s ha e shown ha in PLS he
sample can be e y small [94] and ha he minimum can
e en be 20 [64].
The segmen a ion s uc u e o he ob ained da a is p e-
pa ed in he hi d s ep o FIMIX. To do his, an ex pos anal-
ysis was pe o med [50], which means, fi s ly, assigning each
obse a ion o a segmen om he highes esul o he p ob-
abili y o belonging o ha segmen . Secondly, he da a a e
di ided by means o an explana o y a iable o a combina-
ion o se e al explana o y a iables, esul ing in da a g oup-
ing ha co esponds o ha p oduced by FIMIX-PLS.
A pos hoc analysis was ca ied ou o de e mine he
explana o y a iables ha jus i y his segmen a ion. Using
he ecommenda ions o se e al au ho s, CHAID decision o
classifica ion and eg ession ees we e used o do his [48, 95].
A CHAID decision ee [96] is a g aphical and analy ical
way o ep esen ing all he e en s ha may a ise om a deci-
sion. These ees allow he examina ion o he esul s and
isually de e mine how he model flows. The isual esul s
help o find specific subg oups and ela ionships ha migh
no be ound wi h mo e adi ional s a is ics [97]. In his
in es iga ion, his me hod was used o make he “bes ”deci-
sion om a p obabilis ic poin o iew on a ange o possible
decisions. As seen in Figu e 2, he ob ained esul s show ha
he HABITAT a iable is sufficien ly explana o y o he
choice o 4 segmen s.
Ano he echnique ha could be used in he pos hoc
analysis was o compa e he classifica o y explana o y
Table 7: Indices FIT. C i e ia o model choice.
FIT indices k=2 k=3 k=4 k=5
AIC (Akaike in o ma ion c i e ion) 3984.504 3891.831 3689.352 3683.153
AIC3 (AIC modified wi h ac o 3) 4007.504 3926.831 3736.352 3742.153
AIC4 (AIC modified wi h ac o 4) 4030.504 3961.831 3783.352 3801.153
BIC (Bayesian in o ma ion c i e ion) 4078.760 4035.264 3881.961 3924.940
CAIC (AIC con olled) 4101.760 4070.264 3928.961 3983.940
LnL (LogLikelihood) −1969.252 −1910.916 −1797.676 −1782.577
MDL5 (minimum desc ip ion leng h wi h ac o 5) 4639.783 4888.994 5028.399 5364.085
EN (s anda dized en opy s a is ics) 0.998 0.819 0.787 0.766
Table 8: Rela i e segmen sizes.
kSegmen 1 Segmen 2 Segmen 3 Segmen 4 Segmen 5
2 0.637 0.363
3 0.582 0.296 0.121
4 0.561 0.285 0.105 0.049
5 0.374 0.286 0.202 0.089 0.049
8 Complexi y
a iables using he Analysis o Va iance (ANOVA) o a ac-
o applied o he segmen assigned o each obse a ion.
In his way, he CHAID ee was cons uc ed, and he
cha ac e is ics o he segmen s we e ound using FIMIX-
PLS. The esul s ob ained o he Analysis o Va iance
(ANOVA) defined he ca ego y a iables wi h explana o y
capaci y. Table 9 shows ha no only he HABITAT a iable
has explana o y capaci y bu also AGE and FAVORITE
SEARCH ENGINE.
In a mo e de ailed analysis in Table 10, AGE was
no ound o be significan when s udying he diffe ences
o means o he 3 segmen s. Howe e , HABITAT and
FAVORITE SEARCH ENGINE we e significan .
Table 10 indica es he diffe ences in segmen s 1 and 2
be ween owns o <20,000 inhabi an s, om 20,000 o
100,000 inhabi an s, and >100,000 inhabi an s. The e a e sig-
nifican diffe ences be ween Google and h ps://www.ecosia.
o g/ sea ch engines in hese same segmen s.
The las s ep o he FIMIX analysis was o es ima e
segmen -specific models. Once he HABITAT and FAVOR-
ITE SEARCH ENGINE a iables we e ound o be he main
explana o y a iables ha jus i y he FIMIX-PLS pa i ions,
only he final s ep emained. In his s ep, he specific models
o he indica ed segmen s we e ound.
In o de o do his, a mul ig oup analysis was ca ied
ou o HABITAT, as he esul s o he CHAID decision
ee sugges ed. The 3 g oups co esponding o li ing in a
place <20,000 inhabi an s, 20,000–100,000 inhabi an s, and
>100,000 inhabi an s we e used. An analysis o a iance
ound significan diffe ences o HABITAT be ween seg-
men s 1 and 2 and also o he FAVORITE SEARCH
ENGINE: Google o Ecosia. A e applying boo s apping
again, he esul s in Table 11 we e ound.
These analyses comple e he basic s eps o he FIMIX-
PLS me hod. Howe e , o he esea ch sugges s es ing
whe he he nume ical diffe ences be ween he specific pa h
coefficien s o he segmen a e also significan ly diffe en
using mul ig oup analysis. Documen esea ch ound se e al
app oaches o mul ig oup analysis, which Sa s ed e al. [98]
and Hai e al. [47] discuss in mo e de ail. Hai e al. [47]
ecommend using he pe mu a ion app oach (Chin & Dib-
be n, 2010; Dibbe n & Chin, 2005), which has also been
implemen ed in he Sma PLS 3 so wa e.
Howe e , be o e in e p e ing he esul s o a mul ig oup
analysis, he esea che s mus make su e ha he measu e-
men models a e in a iable in all he g oups. Once he mea-
su emen in a iance (MICOM) desc ibed by Hensele e al.
[99] had been checked, an analysis was ca ied ou o find i
he e we e any significan diffe ences be ween he segmen s
using mul ig oup analysis (MGA). The esul s can be seen
in he h ee columns on he igh o Table 12.
As can be e ified om he esul s ob ained by he
nonpa ame ic es ing, he mul ig oup PLS-MGA analysis
confi med he pa ame ic es s and also ound significan
diffe ences be ween segmen s 2 and 3.
The e a e diffe ences be ween he fi s and second seg-
men s bu only k=2 and k=3 in H1 ATS →USE
(β=0642∗∗∗ ) and k=1and k=3in H1 (β=0521∗∗∗) and
H7 (β=0316∗∗∗) show a significan diffe ence.
The alidi y o he segmen measu emen model and
i s explana o y capaci y using R2is shown in Table 11
wi h he main esul s classified by segmen . I can be seen
ha k=2 has alues o CR and AVE below he limi s
(k=2, CR PU= 0.293, AVE PU=0.497). The explana o y
capaci y o each segmen (R2) was shown o imp o e in
he gene al model in all he pa i ions wi h he main depen-
den a iable USE.
7.4.1. Assessmen o he P edic i e Validi y. PLS can be used
o bo h explana o y and p edic i e esea ch as i can p edic
bo h exis ing and u u e obse a ions [100] P edic i e alid-
i y indica es ha a gi en se o measu emen s o any con-
s uc can p edic a dependen cons uc [101], as is, in ou
case, in en ion o use (IU).
P edic i e alidi y (p edic ion ou side he sample) was
e alua ed by c oss- alida ion wi h e ained samples. The
app oach sugges ed by Shmueli e al. (2016) was used in
his in es iga ion.
Using he esea ch by o he au ho s [102, 103], he cu -
en PLS P edic algo i hm in he Sma PLS so wa e e sion
3.2.7 was used [65]. This so wa e ga e esul s o he k- old
c oss p edic ion e o s and he summa ies o p edic ion
e o s, such as he oo mean squa e e o (RMSE) and he
mean absolu e e o (MAE). The p edic i e pe o mance
Sea ch
Engine
Node 0
Node 1
PLACE
> 100,000
Node 2
HABITAT
65.6%
Node 0.1
HABITAT
34.4%
PLACE
20,000-
100,000
Node 3
PLACE
<20,000
Figu e 2: CHAID decision ee.
Table 9: ANOVA esul s.
ANOVA F Sig.
Wha is you gen e? 1.091 0.337
Whe e is you cu en house? 3.858 0.022
Wha is you cu en si ua ion? 1.558 0.212
Wha is you educa ion le el? 1.649 0.193
How old a e you? 3.211 0.041
Wha is you a o i e sea ch engine? 5.415 0.005
9Complexi y
[8] K. Bu ns, “New echnology b iefing: en golden ules o
sea ch ad e ising,”In e ac i e Ma ke ing, ol. 6, no. 3,
pp. 248–252, 2005.
[9] R. F. Wilson and J. B. Pe ijohn, “Sea ch engine op imisa ion:
a p ime on keywo d s a egies,”Jou nal o Di ec , Da a and
Digi al Ma ke ing P ac ice, ol. 8, no. 2, pp. 121–133, 2006.
[10] Google, “Blogspo .com, powe ing a Google sea ch. Consul-
ado el 30 de ab il de 2018,”2009, h ps://googleblog.
blogspo .com.es/2009/01/powe ing-google-sea ch.h ml.
[11] A. Gup a, B. Saha, and U. K. Sa ka , “Eme gen he e ogenei y
in keywo d alua ion in sponso ed sea ch ma ke s: a close -
o-p ac ice pe spec i e,”Compu a ional Economics, ol. 50,
no. 4, pp. 687–710, 2017.
[12] H. Haans, N. Raassens, and R. an Hou , “Sea ch engine
ad e isemen s: he impac o ad e ising s a emen s on
click- h ough and con e sion a es,”Ma ke ing Le e s,
ol. 24, no. 2, pp. 151–163, 2013.
[13] J. Bagnall, “New echnology b iefing: sea ch engine ma ke -
ing,”In e ac i e Ma ke ing, ol. 4, no. 4, pp. 388–394, 2003.
[14] N. Yalçın and U. Köse, “Wha is sea ch engine op imiza ion:
SEO?,”P ocedia-Social and Beha io al Sciences, ol. 9,
pp. 487–493, 2010.
[15] K. Je a h, L. Ma, and Y. H. Pa k, “Consume click beha io a
a sea ch engine: he ole o keywo d popula i y,”Jou nal o
Ma ke ing Resea ch, ol. 51, no. 4, pp. 480–486, 2014.
[16] M. G ehan and J. B. Pe ijohn, “Sea ch ma ke ing yes e day,
oday, and omo ow: p omo ing he con e sa ion,”Jou nal
o Di ec , Da a and Digi al Ma ke ing P ac ice, ol. 11,
no. 2, pp. 100–113, 2009.
[17] A. V. Zakha o , “Me hods o web ma ke ing and sea ch
op imiza ion o lib a ies ha ecei e a p ofi om he use
o hei si es wi hin he amewo k o he “ eade -lib a y”
sys em,”Scien ificand Technical In o ma ion P ocessing,
ol. 41, no. 2, pp. 140–144, 2014.
[18] D. Tjond onego o and A. Spink, “Web sea ch engine mul i-
media unc ionali y,”In o ma ion P ocessing & Managemen ,
ol. 44, no. 1, pp. 340–357, 2008.
[19] D. Sánchez, L. Ma ínez-Sanahuja, and M. Ba e , “Su ey
and e alua ion o web sea ch engine hi coun s as esea ch
ools in compu a ional linguis ics,”In o ma ion Sys ems,
ol. 73, pp. 50–60, 2018.
[20] L. Ma ínez-Sanahuja and D. Sánchez, “E alua ing he
sui abili y o web sea ch engines as p oxies o knowledge dis-
co e y om he web,”P ocedia Compu e Science, ol. 96,
pp. 169–178, 2016.
[21] C.-Y. Chao, T.-C. Chang, H.-C. Wu, Y.-S. Lin, and P.-C. Chen,
“The in e ela ionship be ween in elligen agen s’cha ac-
e is ics and use s’in en ion in a sea ch engine by making
belie s and pe cei ed isks media o s,”Compu e s in
Human Beha io , ol. 64, pp. 117–125, 2016.
[22] A. Rangaswamy, C. L. Giles, and S. Se es, “A s a egic pe -
spec i e on sea ch engines: hough candies o p ac i ione s
and esea che s,”Jou nal o In e ac i e Ma ke ing, ol. 23,
no. 1, pp. 49–60, 2009.
[23] J. Kei s ead, “Feeling lucky? Using sea ch engines o assess
pe cep ions o u ban sus ainabili y,”En i onmen al Impac
Assessmen Re iew, ol. 29, no. 2, pp. 87–95, 2009.
[24] S.-S. Liaw, W.-C. Chang, W.-H. Hung, and H.-M. Huang,
“A i udes owa d sea ch engines as a lea ning assis ed ool:
app oach o Liaw and Huang’s esea ch model,”Compu e s
in Human Beha io , ol. 22, no. 2, pp. 177–190, 2006.
[25] A. A. Kamis and E. A. S oh , “Pa ame ic sea ch engines:
wha makes hem effec i e when shopping online o diffe -
en ia ed p oduc s?,”In o ma ion & Managemen , ol. 43,
no. 7, pp. 904–918, 2006.
[26] C. Hahnel, F. Goldhamme , U. K öhne, and J. Naumann,
“The ole o eading skills in he e alua ion o online in o -
ma ion ga he ed om sea ch engine en i onmen s,”Com-
pu e s in Human Beha io , ol. 78, pp. 223–234, 2018.
[27] S.-S. Liaw and H.-M. Huang, “An in es iga ion o use a i-
udes owa d sea ch engines as an in o ma ion e ie al ool,”
Compu e s in Human Beha io , ol. 19, no. 6, pp. 751–765,
2003.
[28] P. Palos-Sanches, J. M. He nandez-Mogollon, and A. M.
Campon-Ce o, “The beha io al esponse o loca ion based
se ices: an examina ion o he influence o social and en i-
onmen al benefi s, and p i acy,”Sus ainabili y, ol. 9,
no. 11, p. 1988, 2017.
[29] L. Hi su, “Tag w i ing, sea ch engines, and cul u al sc ip s,”
Compu e s and Composi ion, ol. 35, pp. 30–40, 2015.
[30] C. Jaca, V. P ie o-Sando al, E. L. Psomas, and M. O mazabal,
“Wha should consume o ganiza ions do o d i e en i on-
men al sus ainabili y?,”Jou nal o Cleane P oduc ion,
ol. 181, pp. 201–208, 2018.
[31] V. Venka esh and F. D. Da is, “A heo e ical ex ension o he
echnology accep ance model: ou longi udinal field s udies,”
Managemen Science, ol. 46, no. 2, pp. 186–204, 2000.
[32] F. D. Da is, “Pe cei ed use ulness, pe cei ed ease o use, and
use accep ance o in o ma ion echnology,”MIS Qua e ly,
ol. 13, no. 3, p. 319, 1989.
[33] F. D. Da is, R. P. Bagozzi, and P. R. Wa shaw, “Use
accep ance o compu e echnology: a compa ison o wo
heo e ical models,”Managemen Science, ol. 35, no. 8,
pp. 982–1003, 1989.
[34] R. J. Budd, “Response bias and he heo y o easoned ac ion,”
Social Cogni ion, ol. 5, no. 2, pp. 95–107, 1987.
[35] J.R.Sau a,P.Palos-Sanchez,andM.A.RiosMa in,“A i udes
o en i onmen al ac o s in he ou ism sec o exp essed
in online commen s: an explo a o y s udy,”In e na ional
Jou nal o En i onmen al Resea ch and Public Heal h,
ol. 15, no. 3, p. 533, 2018.
[36] R. Bagozzi, “The legacy o he echnology accep ance model
and a p oposal o a pa adigm shi ,”Jou nal o he Associa-
ion o In o ma ion Sys ems, ol. 8, no. 4, pp. 244–254, 2007.
[37] R. Palanisamy, “E alua ion o sea ch engines: a concep ual
model and esea ch issues,”In e na ional Jou nal o Business
and Managemen , ol. 8, no. 6, 2013.
[38] W. M. Lim and D. H. Ting, “E-shopping: an analysis o he
echnology accep ance model,”Mode n Applied Science,
ol. 6, no. 4, 2012.
[39] M. Kou a is, “Applying he echnology accep ance model and
flow heo y o online consume beha io ,”In o ma ion Sys-
ems Resea ch, ol. 13, no. 2, pp. 205–223, 2002.
[40] C. Mo osan and M. Jeong, “Unde s anding a ele s’adop-
ion o ho el ese a ion web si es,”in In o ma ion and Com-
munica ion Technologies in Tou ism 2006, pp. 394–405,
Sp inge , Vienna, 2006.
[41] K. C. C. Yang and Y. Kang, “Explo ing ac o s influencing
In e ne use s’adop ion o In e ne ele ision in Taiwan,”
Fi s Monday, ol. 11, no. 3, 2006.
[42] L. Hsu and Z. Wal e , “Sea ch engine o con en websi e?
A local in o ma ion seeking classifica ion model based on
16 Complexi y

consume cha ac e is ics and websi e pe cep ions,”In e na-
ional Jou nal o Human-Compu e In e ac ion, ol. 31,
no. 4, pp. 263–276, 2015.
[43] H. Chi-Yueh, H. Ci-Jhan, and C. Hsiu-Hui, “Using echnol-
ogy accep ance model o explo e he in e ion o in e ne use s
o use he audio and ideo fi ness eaching,”Jou nal o Engi-
nee ing and Applied Sciences, ol. 12, no. 18, pp. 4740–4744,
2017.
[44] J.-W. Moon and Y.-G. Kim, “Ex ending he TAM o a
Wo ld-Wide-Web con ex ,”In o ma ion & Managemen ,
ol. 38, no. 4, pp. 217–230, 2001.
[45] D. Ge en, E. Ka hanna, and D. W. S aub, “T us and TAM in
online shopping: an in eg a ed model,”MIS Qua e ly,
ol. 27, no. 1, p. 51, 2003.
[46] J. Mondéja -Jiménez, M. Sega a-Oña, Á. Pei ó-Signes, A. M.
Payá-Ma ínez, and F. J. Sáez-Ma ínez, “Segmen a ion o he
Spanish au omo i e indus y wi h espec o he en i onmen-
al o ien a ion o fi ms: owa ds an ad-hoc e ical policy o
p omo e eco-inno a ion,”Jou nal o Cleane P oduc ion,
ol. 86, pp. 238–244, 2015.
[47] J. F. Hai , M. Sa s ed , L. Ma hews, and C. M. Ringle,
“Iden i ying and ea ing unobse ed he e ogenei y wi h
FIMIX-PLS: pa I-me hod,”Eu opean Business Re iew,
ol. 28, no. 1, pp. 63–76, 2016.
[48] M. Sa s ed and C. M. Ringle, “T ea ing unobse ed he e o-
genei y in PLS pa h modeling: a compa ison o FIMIX-PLS
wi h diffe en da a analysis s a egies,”Jou nal o Applied S a-
is ics, ol. 37, no. 8, pp. 1299–1318, 2010.
[49] M. Sa s ed and E. Mooi, “Clus e analysis,”in A concise guide
o ma ke esea ch, pp. 273–324, Sp inge , Be lin, Heidelbe g,
2014.
[50] C. Hahn, M. D. Johnson, A. He mann, and F. Hube ,
“Cap u ing cus ome he e ogenei y using a fini e mix u e
PLS app oach,”Schmalenbach Business Re iew, ol. 54,
no. 3, pp. 243–269, 2002.
[51] M. Sa s ed , J.-M. Becke , C. M. Ringle, and M. Schwaige ,
“Unco e ing and ea ing unobse ed he e ogenei y wi h
FIMIX-PLS: which model selec ion c i e ion p o ides an
app op ia e numbe o segmen s?,”Schmalenbach Business
Re iew, ol. 63, no. 1, pp. 34–62, 2011.
[52] M. Sa s ed , “A e iew o ecen app oaches o cap u ing he -
e ogenei y in pa ial leas squa es pa h modelling,”Jou nal o
Modelling in Managemen , ol. 3, no. 2, pp. 140–161, 2008.
[53] M. Sa s ed , M. Schwaige , and C. M. Ringle, “Do we ully
unde s and he c i ical success ac o s o cus ome sa is ac ion
wi h indus ial goods?- ex ending Fes ge and Schwaige ’s
model o accoun o unobse ed he e ogenei y,”Jou nal o
Business Ma ke Managemen , ol. 3, no. 3, pp. 185–206,2009.
[54] M. C. J. Caniëls, H. K. L. Lenae s, and C. J. Gelde man,
“Explaining he in e ne usage o SMEs: he impac o ma ke
o ien a ion, beha iou al no ms, mo i a ion and echnology
accep ance,”In e ne Resea ch, ol. 25, no. 3, pp. 358–377,
2015.
[55] M. C. J. Caniëls, H. K. L. Lenae s, and C. J. Gelde man,
“Explaining he in e ne usage o SMEs,”In e ne Resea ch,
ol. 25, no. 3, pp. 358–377, 2015.
[56] C. Ma ques and E. Reis, “How o deal wi h he e ogenei y
among ou ism cons uc s?,”Annals o Tou ism Resea ch,
ol. 52, pp. 172–174, 2015.
[57] A. Na a o, F. J. Acedo, F. Losada, and E. Ruzo, “In eg a ed
model o expo ac i i y: analysis o he e ogenei y in
manage s’o ien a ions and pe cep ions on s a egic ma ke -
ing managemen in o eign ma ke s,”Jou nal o Ma ke ing
Theo y and P ac ice, ol. 19, no. 2, pp. 187–204, 2011.
[58] L. M. Ma hews, M. Sa s ed , J. F. Hai , and C. M. Ringle,
“Iden i ying and ea ing unobse ed he e ogenei y wi h
FIMIX-PLS: pa II–a case s udy,”Eu opean Business Re iew,
ol. 28, no. 2, pp. 208–224, 2016.
[59] M. G oß, “He e ogenei y in consume s’mobile shopping
accep ance: a fini e mix u e pa ial leas squa es modelling
app oach o explo ing and cha ac e ising diffe en shoppe
segmen s,”Jou nal o Re ailing and Consume Se ices,
ol. 40, pp. 8–18, 2018.
[60] J. A enas-Gai án, F. J. Rondán-Ca aluña, and P. E. Ramí ez-
Co ea, “Modelling he success o lea ning managemen
sys ems: applica ion o la en class segmen a ion using
FIMIX-PLS,”In e ac i e Lea ning En i onmen s, ol. 60,
pp. 1–13, 2018.
[61] I. Ajzen and M. Fishbein, Unde s anding A i udes and
P edic ing Social Beha io , P en ice-Hall, Englewood Cliffs,
NJ, 1980.
[62] R. P. Bagozzi and Y. Yi, “On he e alua ion o s uc u al equa-
ion models,”Jou nal o he Academy o Ma ke ing Science,
ol. 16, no. 1, pp. 74–94, 1988.
[63] J. Y. Son and I. Benbasa , “O ganiza ional buye s’adop ion
and use o B2B elec onic ma ke places: efficiency-and
legi imacy-o ien ed pe spec i es,”Jou nal o Managemen
In o ma ion Sys ems, ol. 24, no. 1, pp. 55–99, 2007.
[64] W. Ching and P. News ed, “Chap e 12. S uc u al Equa ion
modeling. Analysis wi h small samples using pa ial leas
squa es,”in S a is icas s a egies o sma sample esea chs,
E. R. H. Hoyle, Ed., Sage Publica ions, Thousand Oaks, CA,
USA, 1999.
[65] C. M. Ringle, S. Wende, and J. M. Becke , Sma PLS 3,
Sma PLS Gmb H, Boennings ed , 2015, h ps://www.
sma pls.com/.
[66] J. F. Hai , G. T. M. Hul , C. M. Ringle, and M. Sa s ed , A
P ime on Pa ial Leas Squa es S uc u al Equa ion Modeling
(PLS-SEM), Sage, Thousand Oaks, CA, USA, 2014.
[67] J. Cohen, S a is ical Powe Analysis o he Beha io al Sci-
ences, E ibaum, Hillsdale, NJ, 2nd edi ion, 1988.
[68] F. Faul, E. E d elde , A. Buchne , and A. G. Lang, “S a is ical
powe analyses using G
∗
Powe 3.1: es s o co ela ion and
eg ession analyses,”Beha io Resea ch Me hods, ol. 41,
no. 4, pp. 1149–1160, 2009.
[69] J. Cohen, “A powe p ime ,”Psychological Bulle in, ol. 112,
no. 1, pp. 155–159, 1992.
[70] J. F. Hai J , M. Sa s ed , L. Hopkins, and V. G. Kuppelwiese ,
“Pa ial leas squa es s uc u al equa ion modeling (PLS-
SEM): an eme ging ool in business esea ch,”Eu opean
Business Re iew, ol. 26, no. 2, pp. 106–121, 2014.
[71] E. G. Ca mines and R. Zelle , Reliabili y and Validi y Assess-
men , Sage Publica ions, Newbu y Pa k, CA, 1979.
[72] J. Hensele , C. M. Ringle, and R. R. Sinko ics, “The use o
pa ial leas squa es pa h modeling in in e na ional ma ke -
ing,”Ad ances in In e na ional Ma ke ing, ol. 20, pp. 277–
320, 2009.
[73] D. Ba clay, C. Higgins, and R. Thompson, “The pa ial leas
squa es (PLS) app oach o causal modelling: pe sonal com-
pu e adop ion and use as an illus a ion, echnology s udies,
special issue on esea ch me hodology,”Technology and
In es men , ol. 2, no. 2, pp. 285–309, 1995.
17Complexi y
[74] W. W. Chin, “The pa ial leas squa es app oach o s uc u al
equa ion modelling,”Mode n Me hods o Business Resea ch,
ol. 295, no. 2, pp. 295–336, 1998.
[75] O. Gö z, K. Lieh -Gobbe s, and M. K aff , “E alua ion o
s uc u al equa ion models using he pa ial leas squa es
(PLS) a oach,”in Handbook o Pa ial Leas Squa es,E.W.
W. V. Esposi o Vinzi, Ed., pp. 691–711, Sp inge Be lin Hei-
delbe g, Be lin, Ge many, 2010.
[76] J. Hai , W. Black, B. Babin, R. Ande son, and W. C. Black,
Mul i a ia e Da a Analysis, Cengage, UK, 8 h edi ion, 2018.
[77] C. E. We s, R. L. Linn, and K. G. Jö eskog, “In aclass eli-
abili y es ima es: es ing s uc u al assump ions,”Educa ional
and Psychological Measu emen , ol. 34, no. 1, pp. 25–33,
1974.
[78] J. F. Hai , C. M. Ringle, and M. Sa s ed , “PLS-SEM: indeed a
sil e bulle ,”Jou nal o Ma ke ing Theo y and P ac ice,
ol. 19, no. 2, pp. 139–152, 2011.
[79] J. Nunnally, Psychome ic Theo y, McG aw-Hill, New Yo k,
1978.
[80] C. Fo nell and D. F. La cke , “E alua ing s uc u al equa ion
models wi h unobse able a iables and measu emen e o ,”
Jou nal o Ma ke ing Resea ch, ol. 18, no. 1, pp. 39–50, 1981.
[81] T. K. Dijks a and J. Hensele , “Consis en and asymp o ically
no mal PLS es ima o s o linea s uc u al equa ions,”Com-
pu a ional S a is ics and Da a Analysis, ol. 81, pp. 10–23,
2015.
[82] D. Benne , D. P. B. Yába , and J. R. Sau a, “Uni e si y Incu-
ba o s May Be Socially Valuable, bu How Effec i e A e
They? A Case S udy on Business Incuba o s a Uni e si ies,”
in En ep eneu ial Uni e si ies. Inno a ion, Technology, and
Knowledge Managemen , M. Pe is-O iz, J. Gómez, J. Me -
igó-Lindahl, and C. Rueda-A mengo , Eds., Sp inge , Cham,
Swi ze land, 2017.
[83] J. Hensele , “B idging design and beha io al esea ch wi h
a iance-based s uc u al equa ion modeling,”Jou nal o
Ad e ising, ol. 46, no. 1, pp. 178–192, 2017.
[84] R. F. Falk and N. B. Mille , A P ime o So Modeling, Uni-
e si y o Ak on P ess, 1992.
[85] J. Hensele , C. M. Ringle, and M. Sa s ed , “A new c i e ion
o assessing disc iminan alidi y in a iance-based s uc-
u al equa ion modeling,”Jou nal o he Academy o Ma ke -
ing Science, ol. 43, no. 1, pp. 115–135, 2015.
[86] M. Sa s ed , C. M. Ringle, D. Smi h, R. Reams, and J. F. Hai
J , “Pa ial leas squa es s uc u al equa ion modeling (PLS-
SEM): a use ul ool o amily business esea che s,”Jou nal
o Family Business S a egy, ol. 5, no. 1, pp. 105–115, 2014.
[87] H. Bozdogan, “Mix u e-model clus e analysis using model
selec ion c i e ia in a new in o ma ion measu e o complex-
i y,”in P oceedings o he Fi s US/Japan Con e ence on he
F on ie s o S a is ical Modeling: An In o ma ional App oach,
H. Bozdogan, Ed., pp. 69–113, Kluwe Academic Publishe s,
Do d ech , Bos on, London, 1994.
[88] V. Ramaswamy, W. S. Desa bo, D. J. Reibs ein, and W. T.
Robinson, “An empi ical pooling app oach o es ima ing
ma ke ing mix elas ici ies wi h PIMS da a,”Ma ke ing Sci-
ence, ol. 12, no. 1, pp. 103–124, 1993.
[89] C. M. Ringle, S. Wende, and A. Will, “Cus ome segmen-
a ion wi h FIMIX-PLS,”in P oceedings o PLS-05 In e na-
ional Symposium, T. Aluja, J. Casano as, V. Esposi o Vinzi,
A. Mo ineau, and M. Tenenhaus, Eds., pp. 507–514, PAD
Tes &go, Pa is, Pa is, 2005.
[90] C. M. Ringle, S. Wende, and A. Will, “Fini e mix u e pa -
ial leas squa es analysis: me hodology and nume ical
examples,”in Handbook o pa ial leas squa es, Sp inge
handbooks o compu a ional s a is ics se ies, Vol. 2, V. Espo-
si o Vinzi, W. W. Chin, J. Hensele , and H. Wang, Eds.,
pp. 195–218, Sp inge , Heidelbe g, Do d ech , London, NY,
2010.
[91] L. T. Hu and P. M. Ben le , “Fi indices in co a iance
s uc u e modeling: sensi i i y o unde pa ame e ized model
misspecifica ion,”Psychological Me hods, ol. 3, no. 4,
pp. 424–453, 1998.
[92] L. T. Hu and P. M. Ben le , “Cu offc i e ia o fi indexes in
co a iance s uc u e analysis: con en ional c i e ia e sus
new al e na i es,”S uc u al Equa ion Modeling: A Mul idis-
ciplina y Jou nal, ol. 6, no. 1, pp. 1–55, 1999.
[93] M. Tenenhaus, V. E. Vinzi, Y. M. Cha elin, and C. Lau o,
“PLS pa h modeling,”Compu a ional S a is ics and Da a
Analysis, ol. 48, no. 1, pp. 159–205, 2005.
[94] H. O. Wold, “In oduc ion o he second gene a ion o
mul i a ia e analysis,”in Theo e ical Empi icism: A Gene al
Ra ionale o Scien ific Model-Building, H. O. Wold, Ed.,
p. VIIIeXL, Pa agon House, New Yo k, NY, 1989.
[95] C. M. Ringle, M. Sa s ed , and E. A. Mooi, “Response-based
segmen a ion using fini e mix u e pa ial leas squa es: heo-
e ical ounda ions and an applica ion o Ame ican cus ome
sa is ac ion index da a,”Da a Mining, ol. 8 o Annals o
In o ma ion Sys ems, pp. 19–49, 2010.
[96] G. V. Kass, “An explo a o y echnique o in es iga ing la ge
quan i ies o ca ego ical da a,”Jou nal o Applied S a is ics,
ol. 29, no. 2, pp. 119–127, 1980.
[97] V. Be langa-Sil en e, M. J. Rubio-Hu ado, and R. V. Baños,
“Com aplica a b es de decisió en SPSS,”REIRE Re is a d'In-
no ació i Rece ca en Educació, ol. 6, no. 1, pp. 65–79, 2013.
[98] M. Sa s ed , J. Hensele , and C. M. Ringle, “Mul ig oup anal-
ysis in pa ial leas squa es (PLS) pa h modeling: al e na i e
me hods and empi ical esul s,”in Measu emen and esea ch
me hods in in e na ional ma ke ing, pp. 195-218, Vol.37 (8),
pp. 1299–1318, Eme ald G oup Publishing Limi ed, S a is-
ics, 2011.
[99] J. Hensele , G. Hubona, and P. A. Ray, “Using PLS pa h
modeling in new echnology esea ch: upda ed guidelines,”
Indus ial Managemen & Da a Sys ems, ol. 116, no. 1,
pp. 2–20, 2016.
[100] G. Shmueli and O. Koppius, “P edic i e analy ics in in o ma-
ion sys ems esea ch,”MIS Qua e ly, ol. 35, no. 3, pp. 553–
572, 2011.
[101] D. S aub, M. C. Boud eau, and D. Ge en, “Valida ion
guidelines o IS posi i is esea ch,”Communica ions o
he Associa ion o In o ma ion Sys ems, ol. 13, pp. 380–
427, 2004.
[102] C. Felipe, J. L. Roldán, and A. L. Leal-Rod íguez, “Impac o
o ganiza ional cul u e alues on o ganiza ional agili y,”Sus-
ainabili y, ol. 9, no. 12, p. 2354, 2017.
[103] M. Sa s ed , C. M. Ringle, G. Schmueli, J. H. Cheah, and
H. Ting, P edic i e Model Assessmen in PLS-SEM: Guidelines
o Using PLSp edic , Wo king Pape , 2018.
[104] Sma PLS, PLS P edic Sma PLS, 2018, July 2018,
h ps://www.sma pls.com/documen a ion/algo i hms-and-
echniques/p edic .
[105] N. Danks, S. Ray, and G. Shmueli, “E alua ing he p edic i e
pe o mance o cons uc s in PLS pa h modeling (Wo king
18 Complexi y
Pape Oc obe 18, 2017),”2018, July 2018, h ps://pape s.ss n
.com/sol3/pape s.c m?abs ac _id=3055222.
[106] A. G. Woodside, “Mo ing beyond mul iple eg ession analy-
sis o algo i hms: calling o adop ion o a pa adigm shi
om symme ic o asymme ic hinking in da a analysis
and c a ing heo y,”Jou nal o Business Resea ch, ol. 66,
no. 4, pp. 463–472, 2013.
[107] E. E. Rigdon, C. M. Ringle, M. Sa s ed , and S. P. Gude gan,
“Assessing he e ogenei y in cus ome sa is ac ion s udies:
ac oss indus y simila i ies and wi hin indus y diffe ences,”
Ad ances in In e na ional Ma ke ing, ol. 22, pp. 169–194,
2011.
[108] J. A. Ma illa and J. C. James, “Impo ance-pe o mance-
analysis,”Jou nal o Ma ke ing, ol. 41, no. 1, pp. 77–79,
1977.
[109] K. Ma zle , F. Bailom, H. H. Hin e hube , B. Renzl, and
J. Pichle , “The asymme ic ela ionship be ween a ibu e-
le el pe o mance and o e all cus ome sa is ac ion: a
econside a ion o he impo ance-pe o mance analysis,”
Indus ial Ma ke ing Managemen , ol. 33, no. 4, pp. 271–
277, 2004.
[110] M. Feng, J. Mangan, C. Wong, M. Xu, and C. Lalwani, “In es-
iga ing he diffe en app oaches o impo ance–pe o mance
analysis,”Se ice Indus ies Jou nal, ol. 34, no. 12, pp. 1021–
1041, 2014.
[111] A. Ri iezzo, A. de Nisco, and M. Rosa ia Napoli ano, “Impo -
ance-pe o mance analysis as a ool in e alua ing own cen-
e managemen effec i eness,”In e na ional Jou nal o Re ail
& Dis ibu ion Managemen , ol. 37, no. 9, pp. 748–764,
2009.
[112] S a is a, Accedidoen Mayo 2018, 2018, h ps://es.s a is a.com/
es adis icas/670092/cuo a-de-me cado-de-los-mo o es-de-
busqueda-po -buscado -espana/.
[113] V. Venka esh, M. G. Mo is, G. B. Da is, and F. D. Da is,
“Use accep ance o in o ma ion echnology: owa d a unified
iew,”MIS Qua e ly, ol. 27, no. 3, pp. 425–478, 2003.
[114] X. Wang and S. Zande , “Ex ending he model o in e ne
s anda ds adop ion: a c oss-coun y compa ison o IP 6
adop ion,”In o ma ion & Managemen , ol. 55, no. 4,
pp. 450–460, 2018.
19Complexi y