0

E 0 Y T 1 M: MOD LI
I L ETWORKI
ITE

I L MEDI

R '

b

Milad Fathi

THE I
UBMITTED IN PARTIAL ULF ILLM
THE REQ IREME T F R TH DE RE
MATRF
I
E
IN
B INE ADMIN I TRA TI

NIV R I Y

RTH RN BRITI H
ugut 20 15

© Mil ad athi, 20 15

L MBIA

DOPTIO

Abstract
ocial Networking
dyna1ni c

it

(

of the w ay pe pl

) hav

bee me

conllTiuni cate.

f fundam ental imp rtance in haping th e
pnmary obj ctive of

practitioner i to ~ nnulate trategie th at can lead t

hi gh er numb r of ad ption rate

among u er . The aim of thi re arch i to tud y ad ption f
fa ctor

cial m edia

s and to hed li ght n th

that influence u er pr ference . T o th at end , fir t, by borr w ing th ories from

business ecosy tem , platform bu ine , the technology acceptance model (T AM ), and
hedonic and utilitari an benefits an initial et of potenti al m ea ures is reached . Nex t, the
m easures are used to create a quantitative survey whi ch i co mpl eted b y a sample of 100
university stud ents. An exploratory fac tor analysis perform ed on the coll ected data yields
four dimen ions: 1) platform, m easured by control over privacy and ease of use 2) u ser
benefits 3) network, m easured by number of fri end s and m ember

and 4) contri butor

benefits. Con sequentl y, the results of a conjoint anal ysis based on un covered compon ent
highlight the considerable importance of control over privacy and ease of use from a u ser
persp ective. Moreover, findings show that for u sers, an optimal SNS where other users
share mostly entertaining content, contributors share m ostly u eful content, app lication
are mostl y fun, control over privacy of posts exists, a good number of friends are registered
and accessible, and is easy to u se. Result

also show that content shared by external

contributors is almo st as impmiant as content

bared b y users in shaping p references .

These findings are exp ected to be of value to both scholars and social media and
communications practitioners.

11

TABLE OF

ONTE NT
..

Ab tract

11

Table of ontent

lV

Li t of Tables

v

Li t of Figure

Vl

Dedication

Vll

Acknow ledgem ent

Vlll

..

C hapter One: Introduction

1

C hapter Two: Literature revi ew

7
7

2 .1 Business ecosystem
2 .1.2 Bu siness Eco y tem a a Perspecti ve for Studying
2.2 The Socia l Networking Ecosyste1n Conceptual M odel

ocial Networking Sites

8
11

2.2. 1 The Leader and the Platform

12

2.2.2 Users

15

2.2.3 Contributors

16

C hapter T hree: R esearch Methodology
3.1 M ea urem ent of Constructs and H yp otheses

18
18

3.1.1 Platfonn-related Measures

19

3 .1.2 Users-related M easures

20

3 .2.3

ontributors-related M easures

3.2 Research Design

21
21

3 .2.1 xploratory Factor Analys i

22

onj oint A nalys is

24

3.2.2

111

3.2. Ju tificati n of

pproa h

25

perati naliza ti n

27

.1 Pil t tudy

2

3.4 ampling

29

.5 Data Analy i

0

.5. 1 Fact r Analy i

30

3.5.2

31

onj int

nal y i

Chapter Four: F indin g
4 .1 tudy 1: xpl rat ry Factor Analy i

32
32

4 .1.1 ample Demographi c

32

4.1.2 De criptive tatistic

33

ample Adequacy

34

4 .1.4 Factor Analy i Finding

35

4.1.3

4 .2 Study 2:

onjoint Analy is

36

4.2 .1 Attributes and Levels

37

4 .2.2 Respo ndents' Demo graphics

37

onjoint Analy i Finding

3

Chapter Five: Discuss ion and Conclu sion s

41

4 .2.4

5.1 Di scussion

41

5.2

onclusion

45

5.3

ontributi ns

46

5.4 Limitation

4

I\

LI T OFTABL
Table 3.1: M ea ur

for plat£ liD u er and c ntribut r

19

Table 4.1: ummary f r pond ent d m ographi c

32

Table 4.2: D criptive tati tic for tud y 1

33

Table 4.3: KM

and Barl ett ' t t re ult

34

Table 4.4: Patten1 M atrix

35

T able 4 .5 : Attribute and lev ls fo r conj oint analy i

37

Tabl e 4 .6 : Re pondent demographi c

38

Table 4.7: Conj oint analy i results

39

Table 4.8: Top 3 optimum N S profile

40

\'

7. Dedication

I w uld lik t
pr
m mcer gratitude t m up
r. un gchul h i and r.
Waqar 1-laqu D r th
ntinu u upp rt f m m t r' tud and related re arch , D r
th ir patience m ti ati n, and imm n kn wl dg . Th ir guidance h 1 d me in all th
tim f r ear h and ritin g f thi the i . I uld n t ha imagin d ha ing better
advi r and m nt r .
I w uld al lik t thank my c mmitt e m mb er r. J alil a fa i [! r hi in ightful
comm nt and nc urag m nt, and the re earch pp rtunitie that h ha pr vid ed me
with.
La t but n t l a t, I ould like to thank my wife, ilo far £1 r alway being th er D r me,
my parent , hahnaz bba i and Mu tafa athi [! r upp rting m thr ugh ut my lifl , my
brother, Dr. li Fathi ~ r b in g 111 ''b ig br th er", and 111 uncle and ld friend , Habiballah
Abba i, who I mi d arly.

\II

8. Acknowledgement
I would like to d dicat my work to thr
x tra rdinary wom n: m y grandm th r, wh
alwa y encouraged m t b tt r my if thr ugh ducati n my m ther, an award -winning
cience teach r wh i th r a n ~ r my e i tenc , and my wife, il fa r, for supp01iing
m e through thick and thin and acrific ing more than one c uld imagin to m ake my dream
become reality.

VIII

hapter On e: Introduction

ocial media i defin ed by K aplan and H aenlein (2 01 0) a
application that build

n th

that allow the cr ati on and

"a gr up of Intern t-ba ed

ideo! gical and t elm 1 gical D und ati n
change

f u er-g n rat d cont nt".

of ocial m edia, including new p rta l , e-c mm rc

web it

f Web 2.0, and

h r ar variou form

, Wiki , bl g , me sagmg

application , and ocial n etworking ite .

In particul ar, Social N tworking

ite (

public profile w ithin a defin ed

y tern, have li t of conn ecti n

) all w u r

to create publi c or parti ally
or friends w ho they

interact with and brow e their own connections and those of other users (Boyd & E lli son,
2012). Through these pl atforms, users communicate and share content in the fo nn of text ,
photos, videos or audio . Examples of such platforms are global SNSs (e.g. Face book,
Twitter, Instagram, G oogle+ and You Tube) as well as locall y known ones (e.g.

ma

W eibo in China, VKontakte in Russia and C loob in Iran) . Regardl ess of their type and
location, SNS s h ave fund am entally ch anged the way people and businesses cmm ect and
communicate. They act as platforms with active ecosystem s of u sers and contributors
around them (Zhu & Iansiti, 2007) that fa cilitate creation and global di ffu sion of conten t
and often consumption of appli cations (Jones, R amanau , Cross, & Healing

20 1O·

Subralunanyam & Greenfield, 2008).

Whil e m any large SN Ss are created by developers, sta1i-ups and large enterprises on a
continuous basis, a relatively sm all number of them are cunentl y popul ar and have gained
u er ' rna s ad o pti on. Thu , a qu estion that ari ses, and one th at i the mai n subject of this
research, is: What characteri ti cs must a N

po ess to enh ance u er · ad ption ? [n other

word , how do the charact ri tic
preference for that

omp n nt that [! rm a

ucce

fact r fl r ocial m dia, including de ign and

?

Previou re earch ha found ar1 u
u er-friendline

( garwal & Venkate h 2002; B nbunan- ich 2001; Lavi & Tractinsky,

2004· Palmer, 2002), u er ati facti n (Horan
& Yoon, 2002·

impact a u er'

f the

ugianto & Tojib, 2006·

Bil el Buytikozkan & Ruan, 2006;
Lin, 2007; Park, Gretzel &

bhichandani, & Rayalu, 2006; McKinney

zyman ki & Hi e, 2000), quality (Barnes, 2005;

ao, Zhang &

eydel, 2005;

. Kim &

toel, 2004;

irakaya-Turk, 2007) and content (BalogJu & Pekcan, 2006;

Cheung & Huang, 2002) . Variou studies have al o focu ed on

N

in particular. For

instance, Ross et al. (2009) inve tigate Facebook usage from a u er perspective using the
Five-Factor Model of personality (McCrae & John, 1992) . Dwyer, Hiltz, and Passerini
(2007)'s study is mainly focused on perceived tru t and privacy among Facebook and
MySpace users . In a study on adoption of ocial media, Curtis et al. (20 10) applied the
Unified Theory of Acceptance and Use of Technology (UTAUT) (Venkatesh, Morri ,
Davis, & Davis, 2003) using non-profit organizations as the sample. However, despite the
emergence of SNSs as a notewotihy stream of theory, what is surprisingly overlooked in
studies on SNS adoption is a system view, which would not only take into account
technology level factors and user characteristics, but would also explain network effect ,
importance of user engagement and the role of third party contributors and their offerings.
The impact of network effects as a competitive advantage has been studi ed exte

ively by

vari us scholars including Katz and Shapiro (1994) . Al o con iderable lit rature exi t on
the role f third pa1iy contributors, and their interaction with u ers in defining th fate of
teclmology platforms (Boudreau & Hagiu, 2009 ; David

Evan , 2003; Gab z wicz &

Wauthy, 2004 · Ian iti & Levi n, 2002· M or , 19 6· P lt ni mi 2006; W yl 2010).
Henc

it i only natural t

p ct th e a tor and th ir p rceived attribut
and ultimat ly ad pt r n t ad pt them .

impa t n the way u er

1 tream of th

Bu ine

co y t m i a n

which ha

been ex ten i

bu ine e

du e mainly t it dynami c y t m view .

ly u d £ r

ry und r the rubric

platforms. Wherea

f tra t gic m anagement

tud ying and m d ling lnform ati n Technology
defined by Jam es F Moore ( 1993)

and Anggraeni, Hmi igh, and Z gveld (2007) bu ine
network of busine

t have an

eco y tem (herewi th B ) i

a

and ind ividu al that co-ev lve by interacting thro ugh technology
BE a

a theory ha

imilaritie

with pl atfo rm bu siness (Gawer &

Cusumano, 2002; Zhu & Iansiti, 2007) and multi -sid ed network (Boudreau & Hagiu,
2009 ; DavidS Evan s, 2003; W eyl, 201 0) , it is particularly useful fo r stud ying social medi a
due mainly to its emphasis on not only the technology platfo nn it elf, but also the ro le of
users, third party developers and external cont1ibutors. As Li (2009) in a stud y on C isco's
technological BE

highlights, symbiosis, co-evoluti on

and

platfo rm

are

defining

characteristics of business ecosystem, all of which are arguabl y the defining characteristics
of a typical SNS . H ence, what this research pri1narily po tulates is that perceived
characteristics of components that fon11 the business eco ystem of a SNS can, to a good
extent, expl a in users' preference for th at SNS .

Several steps are taken to stud y this posit. First, based on the li terature, a conceptual mod el
dubbed Social Networking Eco ystem (SNE) is introdu ced. The model identifie three
categori e of measures for ocial networkin g sites, namely, platfom1 -related mea ure ,
u er-related measure , and contributor-related meas ure .

u b equently, it m

expected to fonn each category are extracted from the literature of bu ine

that are

eco y tem,
3

platfonn bu ine

the techno! gy ace ptance m d 1 (T M) , h d me and utilitarian

ben fit , etc. The m a ure are then empirically te ted in a tw - ta ge tudy. In tudy 1, a
ample of 100 uni

r ity tud nt wa a k d t rate the pr po ed m a ure in term of

i1nportance in cho

ing and u ing a ocial n tworking ite. The r ult

Factor Analy i

n the c ll ect d data r

characte1i tic

(contr 1 ver pri acy and ea e

fan

xplorat ry

eal d fl ur c mp nent , namely, platform
f u e), u er benefit

(fun and utility) ,

network (number of friend and members), and c ntribut r benefi ts (fun and utility). In
study 2, ba ed on the 4 component

extracted in stud y 1, a conj oint analysi with 6

attributes (each with 2 level ) and 8 profile wa perform ed utilizing a sampl e of 110
university students. The re ults highlight the consid erable importance of control over
privacy of posts as well as ea e of u e from a u er perspective. Taking into account the
considerable amount of literature on hedonic and utilitarian benefits in the technology
adoption domain, an interesting finding of the conjoint analysis is where the source of
these benefits are. Based on the findings, users' favourite source of hedonic benefit are
other u ers' posts as well as co ntributors' applications, while contributors' co ntent i the
favored source of utility. Overall , according to respondents, an optimum social networking
site is one in which users share mostly entertaining content, contributors share mostly
useful content, developers develop mostly entertaining applications, users have control
over privacy of their posts, the SNS is easy to use, and lastly, a good number of f1iends are
registered on it.

This stud y make significant contributions. It addresse a clear need for a holi tic model
that would explain how NS characteri sti c can impact user ' preferences towards ocial
networking sites.

iven the multidi sciplinary nature of thi

tudy, it contribute to two

4

1 and und r tudied

n

tr am

nam 1 , bu in

f th

tern in pm1i cular ha gain d gr und in

cial media (In[! tmati n
n erning th IT indu try v r

tudi
de pite pr

.L.
th

1

u u e

f th

t rm

f

n t rel on the lite ratur

). H wever

t m · by R. Hanna, A . R hm and

cia l Media

cial m di a b th c nceptually and

m nti ned tud y d

re ( 19

rk i th fir t t e tend th n ti n f bu ine

ritt nd n (20 11 ), thi
d mam

it intr du ti n b M

111

mpiri ca ll y, a

f bu me

cc

ec

y t m t

the auth r

y tem [! r their

f th e
nceptual

w rk.

Wherea

the co nceptual model and it

appli cation can pave th e way ~ r future r lated

re earch the outcome of the expl oratory fact r anal y i un cover th e internal dynamic
and making of a typi cal ocial netwo rking ite. There ult

f th e conjoint analy i , n th e

other hand provide a c lear compari on between th e importance of tho e component th at
form SNSs from a u er perspective, for th e fir t time .

ltimately, the findin g

of th e

conjoint analysi uncover what the characteri ti c of the optimum social netwo rkin g ite i
for the users and where the ource of variou benefits are in a
are developed each year by enterpri ses,

tart-up

S . While co unt!

and developer , very few

ucceed m

reaching ma s adoption by u er . The results of factor analy i pre ented in thi paper can
be readily u ed for social media trategi t a a trategic framework for ocial netw rkin g
site analysi . Moreover,

ocial m edia practitioners can find the finding of the conjoint

analysis valuabl e as they clarify which attribut

of N

are more imp rtant from a u

r

per pective, and can there[! re be u ed a a guideline [! r directing their focu and re ourcc
to the right a pcct . Finding
meth d logy developed for thi

a ide, re

ar her

re carch to b

in thi

d mam may find the novel

of va lue a

well. The combina tion of

5

mea ure development thr ugh conceptual and empirical mod lling and th n utilizati n of
tho e mea ure for a c njoint analy i i
site re earch domain which al o pr
xtend

topic

eemingly unpreced nt d in the
ed t

b

ial netw rking

r bu t and fruitful. Finally, thi

the i

uch a multi- id d bu ine , hedonic and utilitarian benefit , network

efD ct and TAM in the context f ocial netw rking ite. In particular, mea urement of
hedonic and utilita1ian benefit

eparately fl r u r and contributors wa another novelty of

thi re earch.

6

hapter T wo: Literature r eview

In thi chapt r a br ad r

w

f the lit rature

y tern and

f b th bu ine

pr vid d.

2. 1.1

Bu m e

The on
ha

me

pt

eco y tern

f bu in

ec

wa fir t intr du ced by J. . M

y t m (

r ceived c n id rable attenti n fr m b th acad m1 c
er th term "bu ine

2006). De pite a lack f c n n u

ec

) and
( dner,

and bu in
y tern" (Zhang &

1ang,

2011 ), it can be defined a a n twork of c rnpanie and individual whi h w rk on the
arne technology platfonn , co-evolve together and ultimate ly hare th e arne fate (Kilam ,
Hanunouda, Mikkonen & Aaltonen 20 11 · J.F. M ore, 1993 ; P lt ni emi , 2006; Zhang &
Liang, 2011 ). BE a a theory overlap with variou other re earch domain

(Kinnunen,

Malvalehto, & Haapa alo, 2011). In particular, it ha many imilariti e with multi - id ed
network

(D.S. Evans, Hagiu , & Schmalen ee, 2006), platfom1 bu ine e

(Gawer &

Cusumano, 2002; Katz & Shapiro, 1994; Zhu & Ian iti , 2007) and value chain (Pmier,

1998), as they all emphasize the role of partner hip among numerou

organizations in

innovation and bringing solution to the end u er. In pired by biological eco y tern , what
differentiates BE is that it offers a dynamic, nonlinear, y tern view (Moore, 2006) which
not

nly includes va lue chain , but al o take into account entitie with rather indir ct

role , such a

bu ine e

pr ducing comJ lem ntary

olution , out ourcing

regulatory ag ncie , financia l institute , re earch in titut

, media, uni er iti

competitor (Anggraeni et al., 2007; Bo

h J., 2009; M . Ian iti

Moore, 199 ; M

Zha , 2011 ).

re, 1996; Yu,

techn logy platform

( lan iti

1,

ompame ,
and

ven

, R. Levien, _004; J.F.

i en th

' L vi n, 2002), it a! o highlight

empha i

of BE on

the importance of
7

ynerg1

b tw

n u r and third party c ntribut r m c -e oluti n f compl x bu ines

network .

T day doing bu ines i no longer a
with the 'healthier'
Intere tingly a

c

litary road a c mp etition take plac between

y tern having the competitive edge (Hearn & Pace, 2006).

ueguen and I ckia (20 11) highlight, in the e B

war the level of

member exclu ivity i generally 1 w m eaning that companie often take part in vari us
BE , and therefore, it i difficult to id entify where the b rd er

fa B

are to be defined .

H nee, bu ine es governing BE compete to attract not only more cu tomers, but al o
third party contributor

that can add co mplementary capabilitie

to their offerings

(Adomavicius, Bockstedt, Gupta, & Kauffman, 2006; de Reu ver & Bouwman, 20 12;
Peltoniemi, 2006). According to Marco Iansiti and Roy Levien (2004b ), in order to be
successful a BE needs to stay robust, innovative and productive, which can only be
achieved through strong business network ties . All in all, in comparison to playing a lone
hand, being part of a healthy BE opens doors to new opportunities for creating value
(Bosch J., 2009).

2.1.2

Business Ecosystem as a Perspective for Studying Social Networking Sites

Du e mainly to its dynamic view, BE has been applied in numerous, mo tly technology
related, contexts . In fact the majority of more recent studies on business ecosystem are
dedicated to expansion of the theory to other areas of applicability. In their well-known
book, Marco Iansiti and Roy Levien (2004a) apply concepts from BE to a range of
businesses from oftware, biotechnology to internet industries. Quaadgras (2005) u es BE
to stud y the compl exities of the RFID (radio freq uency ID) industry while Vuori (2005)
find s B

the mo t appropriate tool to model knowledge intensive

erv1ce (KI )
8

organization . In an attempt t
Vi ch r (2006) int:roduc

mod 1 bu in

their B

indu try. Rong, Hu Lin, hi , and

c

y tern health d n Hartigh , Tol, and

fram ework and m a ur
u (2 0 15 ) rea n that

th health of the
1

mor

utch IT

uitabl e for mod eling

the emerging concept f Intern t-of-Thing c mp ar d t the cla 1c upply chain view . B
and e pecially it
diver e notion

plat~ rm

related capaciti

ha e al

such a open inn va ti n (Harland , Wu t

been fl und id eal for

tud ying

Dedehayir, 20 14; Xiaoren,

Ling, & Xiangdon g 2014), entrepreneur hip edu cati n (Bru h, 2014), electri c mobility
(Gie ecke 2014; C . Lu, Rong, You &

hi, 2014), and big data (Yoo ,

hoi, & Lee, 2014).

Throughout the above m entioned tudi e , a well as numerou other example , certain
components and characteristic of busine ses eco y tern have been emphasized that m ake
the concept an ideal theoretica l backbone fo r modelling social networkin g sites. Below
some of these attributes have been discussed.

In a business ecosystem , one or more organizations take the lead and enforce the direction
of the entire network through controlling key resources and establishing regulations (Jam es
F Moore, 1993). According to Marco Iansiti and Roy Levien (2004a) the ri ght strategy fo r
BE leaders is to find ways to share value with other actors within the busine s network in
ord er to reach sustainabl e gro wth. A BE lead er fa cilitates innova tion by simplifying the
connections betw een businesses within a network as well as busine es and end u ers
through technology platfmms (de Reuver & Bouwman, 2 01 2 ~ Nikou, Bouwman, & de
Reuver, 2014; Peltoni emi, 2006) . Qu ality, improvem ents and openne

t

third paiiy

agents are a number of platfo rm characteristics that are contro ll ed by the ecosy tern leader
(Bosch J., 2009).

9

ne of the mo t important characteri tic
their popularity a a bu ine

f plat£ 1m

and p rhap the main reason for

m del i th ir tw - ided nature in that, n t nly a platform

fa cilitate creati n f omplementary produ ct and

rv1ce

it i al

u ed a chokehold , a

cu tom r gateway (Moore, 2006) thr ugh which cu t m er interact with each other a well
a

the BE lead r and contribut r

(B udreau & H agiu , 2009;

avid

vans, 2003;

Gab zewicz & W authy, 2004 ; W yl, 20 10). M bile applica ti n t res are per hap the m o t
prominent example of two- id d netwo rk , but car manufac turing indu trie and h pping
centers are other in tance where platf01m have proven critical (David

vans, 2003 ).

The two-sided (or multi- ided) nature of pl atform as a central building b1 ck of bu ine s
ecosystem autom atically make both users and third party developers especially irnp011ant.
As Katz and Shapiro ( 1994) addre s, users' effect on business networks can be both direct
and indirect. Users often make assumptions about populari ty of BEs and tend to choose the
one perceived to have the highest number of consumers, becau e they assume that a more
popular network will give them access to more applications. Apart from direct revenu e,
more customers can also result in more applications for the respective pl atfonn due to
higher demand (Eisenmann, Parker, & V an Alstyne, 2006).

While users form one side of the pl atform , third pa11y contributors complete the other side.
Importance of developers has been extensively studied in the context of software and
mobile industry in particul ar ( ee fo r exampl e, D.S.

vans et al. (2006) ; Gonc;alves and

Bailon (20 11 ); de Reuve r and Bouwman (20 12); N ikou et al. (20I4); Bergvall-Karebom
and Howcroft (20 II ); Tee and
technologie

awer (2 009)) . T hrough interactions with u ers and using

provided to them by the leader via the platform, d

elopers offer

complementary produ ct or ervices that not onl y bri ng in revenue, but al o e tend the
10

capabiliti

an e ampl in an

of th eco y t m (Zhu & Ian iti, 2007).

Facebook provid e a platfl nn that u r u
API (Applicati n Pr gratruning Int rface)
appli cation fl r it platform or make
both con umed and har ed by u er

fl r haring c nt nt.
ac b

i ting one ace

k

nable

N

c ntext,

dditi nall y thr ugh it
develop r

to create

ibl e. Applicati on and content are

who in turn ha e a hand in crea ti n f cont nt which

hared publicly or w ith their peer network. In a nut bell , ach act r reache it goal, in
that, Facebook get an eco y tem that i c n tantly extend d by developer and u ers,
developer create applicati on u ing the platform and reach a higher number of user
benefitin g the cu tom er chokehold (M oore 1996) that Facebook is, and user get to create
and share content, and consume the applications provid ed to th em through the pl atfo rm . As
a fruit of thi symbiosis between various actor and a healthy co-evo luti on, Faceboo k's
Social Networking Ecosystem has been adopted by over 1.19 billion active user as of June
201 5 (TheNextWeb, 201 5).
2.2

The Soci al Networking Ecosystem Conceptual M odel

The aim of the Social Networking Ecosystem conceptual model (herewi th, SNE model) is
to envisage and visualize the structure of a typical network fo rmed around a social medium
from a business ecosystem perspective.

11

Figure 2.1:

ocial N etworking

co y tern Mod el

Contributors

Platform

Users
As fi gure 2. 1 above illustrates, the

NE leader provid s th e criti cal online pl atfo rm . The

pl atfo rm provides contributors and u ers with the neces ary too l , th e prim ary fun cti o na lity
of whi ch is enabling and simpli fy ing the processe of creating. conn ectin g and

harin g.

The synergies that occur as a result ultimately lead to m ore adopti ons and co-evo lut ion of
the netwo rk. Below, vari ous components of the model have been di scus ed in more deta il.

2.2. 1

The Lea der and th e Pl atform

M oore ( 1996) is of the view that gove rmng the interacti ons in busines
mo stl y performed throu gh quas i-democrati c mechani sm

eco ystem s i

and community gove rn ance

systems. In hi s studi es on the iss ues related to gove rn ance, M oore co mpared ecosyste m
govern ance with mark ets and hi e rarchi c . Moo re (2006) beli eve that wha t happens in the
ecosystem i internali zation of the bu ine s sy tern s and the markets based o n \\ hi ch the)
are connected unde r the influe nce of the leaders of the co mmunity. As me ntio ned above.
12

Marco Ian iti and R y Levien (2004a) al

tat d that c mmon fa t

eco y tern . Yo (2006) i

m ane

f th

i w that g

f bu in

g

co y te1n

rn

bu in

uppli

th

m mber with a r admap ba d on which th y co p rat t achi v a c ffilnon g al, and a
fre dom t r ach the obj

tive of th

y t m ba d n p r nal initi ativ

no inter~ renee hind r th ir m tivati n .
eco y t rn '

capability in d alin g

in a way that

t the am tim e, enl1an ing the 1 v 1 f bu iness
ternal f rc

ith

f

hange

and m ainta ining th

internal rate of innova tion the gov m ance utilize mechani ms of controlling the m embers
to mak e ure that their perfo rmance i not in c ntradicti n with the coffi111 on bj ecti ve .

imilar to conventional bu ine
the entity that govern

cosy tem , the

the netwo rk,

et

ocial N etw rking

co ystem leader is

the neces ary regulati on

and controls k ey

network resources, including the pl atfo rm (Marco Ian iti & R oy Levien, 2004a) . Some
social media platfo rms are initiated by individu al developers, some by start-ups (e.g.
Instagram) and som e by large enterprise (e.g. Google+ ). But regardless of the size and
origin of SNSs, the strategies and tandards that the SNE leader incorporates and enfo rces
are critical in reaching optimum adoption rate. Ap art from design characteristics, openness
to third party developers (Kilamo et al. , 2011 ), security, privacy and data pro tection
policies (Dwyer et al. , 2007; Iachello, Sn1ith, Consol vo, Chen, & Abowd, 2005), as well as
users' co ntro l over leve l of di sc los ure (Feijoo, Pascu, Misuraca, & Lusoli, 2009) are all
gove1nance related decisions that are made by the ecosystem leader and can impact the fate
of an SNE.

As highlighted earli er, the online pl atfo m1 is the ma m building block of a social
networking bu siness ecosystem . In an IT context

isenmann et al. (2006) define platfonn

as a et of too ls or components that provide the n cessary building blocks for application
13

pro id r . F 11 wing a bu in s eco y t m anal gy the platf01m attract and
u er and third party c ntribut r
nee

ary t ol t ha

numer u

Ian iti , 2007).

(Zhu

fri nd li t (B yd

lli

Andr w R hm, & Victoria L
oftware product that of£ r
ontributor ar
ophi ticated

lution t

an

and audi

(Richard Hanna,

nd u er (Ian iti & Richard

al . pr vid d with content creati n and baring to l
uch a Facebo k al

har cont nt in

app lica ti n in tum i d fin d a a

ritt nden 20 11 ).
a

ide

b th

pr vided with th

n, 2012) and r ate an

m ji,

£ rmat , including text, imag

er ar

rv

2006).

ow ver more

allow £ r creati n f app li cation thr ugh their

API (Application Programming Interfa ce) ( joka,

iriviano , Mark poul u, & Yang,

2008). Bo ch J . (2009) highlights a number ucces fact r £ r nline platform , including
ease of application dev lopment, con tant improvem ent of the platform feature , and
installed base.

The extensively studied Technology Acceptance Model (TAM) has been used in the
context of technology platforms by numerous scholars (M.-C. Lee (2009); H .-P . Lu and
Yu-Jen Su (2009); Y . Lu, Zhou, and Wang (2009); Moon and Kim (2001); Yen, Wu,
Cheng, and Huang (201 0)). The two main constructs ofT AM, usefulness and ease of use,
are present in majority of studies on adoption, quite often accompanied by con tructs from
additional theories. One can therefore expect usefulness and ease of use to be important
attributes of SNE platforms from a user perspective. Moreover, hedonic benefits, referred
to gains petiaining to pleasure, and utilitarian benefits, gain

which ha e regards to

usefu lne s (Babin, Darden, & Griffin, 1994) have also been applied in studie concerning
adoption of Infonnation Techno logies (H.-W. Kim,

han, and Gupta (2007); Van der

Heijd n (2004)).

14

2.2.2

U ers

f c -evolution i
characteri tic of bu ine

ec

n

f th d finin g

y tem a a th ory (Marco Ian iti & R y

evi en, 2004a ;

Qu aadgra , 2005 ; Zhang & Liang, 2011 ) thi r 1 i perhap m o t prominent in the contex t
of ocial n tw rking ite . Kaplan and Haenl in (2 0 10) r fer t u er and u er generated
content a main bui ldin g block
as Boyd and

f

cial m di a in their definiti on.

lli on (20 12) al o gi e u er

social networking

lli on (2 007) a well

imilar 1 vel of attention in their de cription f

ite . The importance of u er

ha

al o been inve tigated from a

marketing per pective in recent tudi e on viral marketing and word -of-m outh (Brown,
Broderick, & Lee, 2007; Chu & Kim, 2011 ; Mangold & Fauld , 2009 · Thackeray, Neiger,
Hanson, & M cKenzie 2008) . U er create content and/or hare them over SN s either
with their direct fri end li t or publicly. On a stud y concerning news sharing on social
media platforms, C . S. Lee and M a (2 0 12) and P ai and Arnott (2 01 3) fo und that u ers
driven by gratifications of information seeking, socializing, and status seeking were more
likely to share new in social m edia platforms. Prior experience with social m edia was also
found to be a signifi cant determinant of news sharing intention . As Von Hippe! (2005)
discusses, the major advantage of thi s u er-centered innovation (or in other words
democrati zati on of innovati on) is that th e outcome is far better tail ored to user ' ta te .

Whereas creation and sharing of content is considered the u ers' direct effect, a group of
scholars have found users to also impact business networks indirectly (Katz &

hapiro ,

1994 ). Install ed base is referred to a the number of unit of a ystem in use (Ei

nmatm et

al. , 2006). According to Zhu and Iansiti (2007) a larger install d ba e lead to a larger
suppl y of applications and thereby impact the capabiliti

of a bu ine s eco ystem
15

r c n umer e p

platfl rm . M r

th

ch

platfl rm whi h they beli

m re applicati n (Zhu

Ian iti , 2

The pri ate/public tru cture
f u er

Papachari

which a p r

f

ha th

hi gh

t numb r

n an

mak

it imp rtant t di tingui h betw een tw

herew ith r fitTed t

a

indi idual d

"memb r ". A

typ e

nt an individu 1' e 1 tin g ocial ti e ,

r ar referred t a "fri nd. ". The

but th

f u r and th r by

7.

i, 2009) . Th fir t gr up r pre

eu

r b c me

fa plat.G nn an t nd

n kn w in real life b [! re b fri nding th em n

Lampe, 2007 . Th
regi tered

h r th r le f u

mak rati nal a umpti n ab ut th in tall d ba

imp rtant.
t

tati n i an th r area

( lli

teinfi eld , &

c nd gr up i u er th at are

n t ne e aril y kn w p r
the lit rature

n,

mpha ize

nall y. The e are

it

imp rtant t

di tingui h between the e two group of u er , and mor unp rtantl y, how th ei r perceived
characteri tic can impact the adoption deci ion of indi v idu al

2.2.3

Co ntributors

For a bu ine

eco ystem to be ucce ful , it require di ver e contributor that con tantly

provide application which are of hi gh quality and variety (Bo ch J., 2009; D . . Evan et
al. , 2006 ; Zhu & Iansiti, 2007) . In a two- ided bu ine

eco y tern pl atform, u er account

for one ide and contributors (also referred to as developer ) form th e other (D .. Evan et
al. , 2006; Nikou et al. , 20 14; Quaadgras, 2005) .

In an SNS etting, th e role of co ntributor are diver e. The e include e temal entitie ,
uch a news agencte and co ntent provider that u e th e
to u er .
Applicati n

oftware applicati n d ve loper

to get their me ag

are anoth r type of

r Int rfa c (API ) pr vided by ophi ti ca ted

acr

ontributor . They u e
uch a

aceb

k and

16

Twitter to d velop

ftwar applicati n . In fa ct a highlight d by

much of Faceb ok

ucc

joka et al. (2008),

aid to c m from the fa ct that it open d the door

nline platf01m t appli cation and gam dev 1 p r wh

e c ntributi n hav mad

f it
nline

cial n tworking richer and more enj oyable t u er (Xu, Ryan, Prybut k & W en, 201 2).
lntere tingly thi relation hip i tw -way in that, devel per

uch a Zynga and King have

tran form ed them elve from mall tart-up to publicly li t d enterpri e m ainly through
the ucce

of game develop d for nline

cial n twork (M acMillan, Bunow , &

nte,

2009). Other exampl e includ e companie that provide ocial networking analy is ( N A)
olution which are u ed fo r social medi a li tening and analy i of

N

performance,

network tructure and content reach for both individual and bu ine se . xampl es of uch
software appli cations are iGraph, Paj ek and Gephi ( ombe, Largeron

gyed-Z igm ond , &

Gery, 2010; Diakopoulos, Naaman, & Kivran-Swaine, 201 0).

17

hapter Three: Re earcb Methodology

In the previou chapter, th
pr ented . Th

c nceptual rn d 1 f a

mod el identifi ed th.r

at gorie

cial
f

tw rking
m a ure

ed N

u er

tudi

mpiticall y te t d throu gh tw

c n equ nt

m a ure w re u d to create a quantitati ve urv y, whi ch wa
univer ity tud ent .

y tern wa

nan1ely, platfonn-

r lated u er-related and c ntribut r-relat d. The impact f the pr p
ad pti n wa

c

m a ure on
. In

tudy 1

mpl t d by a ample of

ext, a fa t r an aly i wa p r~ rmed on the c llected data to extract

any unob erved c mpon nt . La tl y, in ord er to valuate and c mpare the imp rtance of
extracted component

in shapin g u ers' preference of

a conj oint analys i

was

conducted.

3.1

Measurement of Constructs and Hypothe es

In this ection, potential m easures for each of the three categori es ar e presented together
with the major studie from which they were adopted. T abl e 3.1 below show the item s
that are expected to measure the impact of platfmm, users and contribu tors on SN S
preferences.

18

Table 3.1:

Mea ure for pl atform u er and contributor

ry
2001 ; Y n t

m za

u rahman

(M

Kim 2001 ; X u t a!. , 201 2)
(Hwang Thorn , 1999)
(X u et al. , 20 12)

contribut r gen rated
cont nt
tility obtained fro m
c ntribut r generated
cont nt
Fun obtained from
contribut r generated
application
tility obtained from
contributor generated
applications
Variety of contributor
generated content

(M

n & Kim , 200 1; Xu eta !. , 20 12)

(Xu et al. , 20 12)

(Moon & Kim , 200 1; Xu et al., 20 12)

(Bo ch 1.
2010; D ..

Evan,

Variety of application
developed by contributor

3.1.1

Platform-related Meas ures

ne of the most ex ten ively tudied topic in technology adopti n i the teclmology
acceptance model (TAM) (M.- . Lee, 2009; Moon & Kim, 200 1; Y n et al., 20 10).
hr ugh ut th
and i there[! re

e studie , ea c f u e ha been regard d a a main c n truct of adoption,
p

ted t impact on um r ' N adopti n a well. Th

econd

19

plat£ nn-relat d item i privacy. According to Dwyer et al. (2007), privacy c n ern
highly impact u er ' ocial m dia b havi ur.

n that th

of c ntent i on of the ba ic regulation enforc d by the

tent of c ntrol over priva y
cial n tworking ec

lead r, th item i expected t influ ence u r pr £ r nee [! r

y te1n

. ubrahmanyam &

Greenfield (200 ) addr s that ultimat ly conn cting and rec nnecting with oth r u er
the primary m tivati n £ r

cial media u age. Hen e, perceived c nnecting capabilities f

a ocial networking platfonn ma y impact an individual'

valuati n of an

Bo ch (2009) i of the view that rate and quality of improvement to a

. a tly,

ftware platform

can con iderably impact its performance. The extent to which thi impact is felt by the
user and whether it is as ociated with their ad ption decision are qu estion that thi

tudy

addre se . Overall, it is po tulated that perceived platfonn characteri tic , as m easured by
the above-mentioned items, impact users' choice of social networking site

3.1.2

Users-related Measures

Five items were expected to measure the impact of users on SN preference. The literature
of business ecosystem and platform business both emphasize the role of installed ba e, and
perceived popularity of a platfmm among users. When it com e to social networking sites,
as highlighted by Papacharis i (2009), it is critical to distinguish between perceived
number of members (SNS user that the individual does not necessatily know in real life) ,
and .friends (real life friends who are also registered on an SNS) . Hence, the two fon11
separate user-related items. Next come the fun and utility that a user expects to obtain from
an SNS. Given the extensive use of hedonic and utilitarian benefits described in the
literature as two main constructs of technology adoption, both fun and utility are exp cted

20

to be r levant for c nt nt har d by u er . Th la t u r-r lated item i u r engag m nt,
fIT y tern (Hw ang & Thorn, 1999).

which i an integral ingredi ent for ucc

3.2.3

Contributor -r elated Mea ures

Apart fr m u er , contribut r al o ha
naturally, ne can

pect th b n fit

preD renee D r

a hand in creation f

cial1nedi a content. Hence,

btained from contribut r t impact individual '

highlighted in chapter tw

apm1 fr m c nt nt crea t r ,

developer are another group of contribut r that pr vid u er with fun and utility by
creating applicati ons. As a re ult, in order to mea ure the impact of contributor offerin g ,
two item mea ure th benefit obtai ned from co ntributor ' content, and two item m as ure
the benefits obtained from co ntributor' appli cati on . Thi approach is different fro m
previous research, where hedonic and utilitari an benefit are treat d a a sep arate vari able
fo r the entire product or service. Additi onall y, whil e openn ess to co ntribu to rs' con tent and
appli cations is important, the vm·iety of these offerings ha been found to be of significance
(Bosch J., 2009; Bosch & Bosch-S ijtsema, 2010; D.S. Evans et al. , 2006 ; David

Evans,

2003). Thus, the vmiety of contributor ' content, and the variety of applications are
measured by two separate item s. Overall , it is postulated that perceived characteristics of
contributors, measured by the above menti oned attributes are assoc iated with individua ls'
choice of SNSs.

3.2

Research Des ign

A total of 15 items ( ee tabl e 3.1) were listed as potenti al measures for the three propo ed
mea ure types (pl atfo rm-related, user-related and contTibutor-related) . The envisag d
measure were used as the basis for a qu antitative survey in which individual were asked
to rate the item s in te1ms of imp011ance in respondent ' choice of social networking ites

21

on a cale of 1 to 7 (1 not important at all , and 7 v ry imp rtant) . here ult of the urvey
w r then u d for an Exploratory Factor Analy is (Thomp
mea ure the relati e imp rtance f each of the unc
wa per£ rmed. Figur 3. 1 below illu trat

n, 2004) . inally, in order to

er d comp nent , a conjoint analy i

the t p that defin e the meth dology of thi

re earch.
Figure 3.1:

Resea rch de ign diagram

Id ntification of
mea urement
item

3.2.1

urvey ba ed
on identified
mea ure

xplo ratory
Factor
naly i of
collected da ta

onjoint
analysi baed
on identified
fac tor

Exploratory Factor A nalysis

Exploratory Factor analy is (EFA) is a popular stati tical method u ed fo r uncovering the
underlying structure of a relatively large set of variabl es, findin g unobserved latent
variables, and reducing the number of factors to an optimum number (Thompson, 2004 ).
As highlighted by Bryant and Yam old ( 1995), EFA enables researchers to find the linear
factors which best fit the data .

Factor analyses differ based on the type of factoring and rotation method and the criteria
for determining th e number of factors varies from one researcher to another. The most
cmmnon type of fa ctoring, and the one utilized in thi s stud y, is Principal Component
Analysis (P A) . In this method, fa ctoring i continued until the minimum number of
fa ctors with maximum vari an e is reached (Jolliffe, 201 4). Given thee pl oratory nature of
this study, one maj r advantage of P A is that it incorporates fewer a umption about the

und rlying tructure of the model. Anoth r advantage of P A i
of th

1a t

quare

approa h t

aid to be the robu tne

approximating the covariance or COlT lation matrix

(Jolliffe, 2002).

The next tep i to rotate the re ulting fact r m del. R tati n i a method of maximizing
high loading

and minimizing low loading

ill

rd er to reach the simpl e t po ible

tructure. Rotation can be categorized to either oblique, where factor can conelate, or

orthogonal. In thi

re earch, a direct oblimin rotati n m ethod was u ed, which is

categorized under blique rotation method (J nnrich &

amp on, 1966). This wa mainly

due to the fact that factor were expected to be partially con elated . Nonetheless, as the
re ult in the next chapter will

how , minimal difference was ob erved in the results

obtained from alternative methods .

Once the results of the factor analysis are obtained, an important question is to dete1mine
which factor to retain. The following rule of thumb have been suggested by Field (20 13) :

1.

Only factors with eigenvalue larger than 1 should be retained.

2.

Retained factors should account for at least 70% of the variance.

3.

In reference to the scree plot of all factors , only factors before the breaking point

are to be retained. The scree plot graphs the eigenvalue of each factor against the
factor number, and is a typical output for factor analysis (see figure 5 for an
example) .

Traditionally, once the factor structure and patten1 coefficients are determined, fact r are
named for the purpose of clarification.

23

3.2.2

In

onjoint Analysi

rd r to evalu ate and compare the relati e imp rtanc

conj oint tud y wa perfo rm d.

nJ int analy i refer t a m arket r ear h teclmiqu e that
~ r

is u ed to det nnm e h w indi idu al ' preference
m ea ur th trade-off: that a con um r mak
ca e a ocial n tworking ite ( r
fundam ental a ump ti n

a pr du ct i d v loped, and to

wh n ad pting a product or ervice, in this

n, Kri ger & W ind 200 1; Nikou et al. , 20 14).

f conjoint analy i

u er I S a combinati on of all the uti liti
Hence, the primary purpo

f extrac ted components a

1

that the

verall utility of a produ ct for a

a ociated with each attribute of that produ ct.

of conjoint analy i

IS

to decompo e the utilities of each

product attribute, and re pective level . Thu , in marketing re earch, where conj oint
anal ysis is most wi dely utilized, product attributes are the independ ent variables and
overall product evaluati ons repre ent the depend ent variable of the statistical model.
Overall , in compari son w ith so li citin g subj ects ' evalu ati on of sin gle product attributes,
conjoint analysis does a better job at simulating the real-life product adoption dec ision by
accounting fo r trade-offs betw een all product attributes at the sam e time.

Two approaches to conj oint analysis previously applied by scholars are the direct choice
approach (stated) and indirect discrete choice (revealed). In the indirect approach,
respondent' s actual behaviour and/or usage is observed and recorded for the purpose of
data analysis, while in the direct approach respond ents are a ked express their evaluation
of product attribute . While the indirect approach has its own u ages, it main drawback i
said to be lack of tatisti cal precision. Moreover, the indirect method is said to be only
suitabl e for well -defined (as oppo ed to hypothetical) and distinctively known product
attributes (Bradl ey & Kroes, 1992) .
24

Approa he to conjoint analy i al o vary depending on the number of product profile and
the method of data recording. Full-profile i the traditional form of conjoint analy i , in
which all combination

f attribute and level ar u ed for creation of product bundle and

re pondent are a ked t evaluate all bundl e . The main probl em a

ciated with the full -

profile appr ach i that, a the number of attribute and their r pective levels grow pa t a
certain point, the number of re ulting profile may become t o m any for respond ents to
con ider and evaluate. A a remedy, a f ractiona l fa ctorial de ign ( un t & Mason, 2009)
typicall y utilized which u e

adequ ate number of all combination

that would

appropriately represent all po ible combination with negligible error . As highlighted in
more detail in the nex t ection, in thi study given the number of attributes and levels, a
fractional factorial design with a total of 8 profiles i used.

Moreover, conjoint anal ysis m ethod vary in terms of method of data coll ection. Two
m ethods in particular are more popul ar: rank, or rate (Gu stafsson, Herrmann, & Huber,
201 3). In the rank m ethod, respondents are a ked to provide a ranking fo r bundles which
ranges from 1 to the total number of bundles present. In the score approach, subjects are
asked to assign a Likert scale (Brooke, 1996) value to the bundle that would appropriately
represent their preference. In this research, the rate method was utilized, and re pondents
were asked to rate the bundles on a scale of 1 (not favourable at all) to 10 (very
favo urab 1e).

3.2.3

Ju stification of Approach

Previou research has expl ored va rious methodologies to study the constructs that impact
adoption in different contex t . A group of studi es ( uch as Vanna Citrin, Sprott, ilverman,
and Stem Jr (2000) and

as tin (2002)) mak use of I a t square regre sion analy i to

25

d t rmine econom tric model for acceptance of pr duct . nother wid ly utilized
approach i

tructural Equation M delling (

M) ( e for e ample, hau ( 1996) and Wu ,

Wang, and Lin (2007 ). Throu gh a en e of t p , including expl rat ry and confi1matory
fa ctor analy e

EM enable re earcher to uncover th fac tor that define a theoretical

model. It al o d tennin

the r lati n hip b tween depend nt variable , a well as

between dependent and indep ndent variabl
The approach utilized in thi r

arch i

at th

arne time.

omewhat different as it makes u e of exploratory

fa ctor anal y i and c njoint analy i over two eparate studi e to determin e the attributes
that hape consumer preferences. There are everal ju tifica tions for the elected approach.
Fir tly given that the literature of N
adoption is quite mall and there i

is still at it infancy, the number of studies on S

eemingly no con en us over th e ri ght direction for

investigation of the topic. Hence, in order reach a comprehensive model fo r SNS adoption
it is necessary to not only account for variou constructs previou ly studied in this domain,
but also rely on theoretical works in other streams of theory with explanatory capacity.
Since this can automatically result in a high number of measures, exploratory facto r
analysis (study I ) becomes pa11icularly useful , as it enables redu ction of mea ures to an
optimum number. EF A can also help find the underlying stru cture within all items and
reach a set of components that can account for most of the variation in the collected data .
Moreover, a primary obj ective of this research is to investigate the relative in1portance of
various constructs whi ch the exploratory factor analysis yields. Conjoint analysis can
acc01nmodate this need better than other methods, as it provides clear analytics about the
relative imp011ance of each attribute in addition to optimum product bundl es from a u er
perspective. Specificall y, the conjoint instrument ha several advanta ge . Firstly, it allow

26

fi r a c mparativ ly more reali tic deci ion model, a it forces respondent t e pre

their

prefi renee toward an N pr file that con ist of everal attribute . As uch, by
detennining deci ion model for ach respond nt, c njoint analy i provide a decision
model that aggregat

the evaluation obtained from all re p ndent (Louvi re, 1988).

econdly, a Hair (19 4) highlight , the conjoint in trum nt makes no a sumptions in
regards with the nature of the relation hip betw en dependent and independent variable
( uch a linearity), which make it ideal for tudie that are of explorat ry nature. Thirdly,
conjoint analy i can accommodate both metric and non-m etric variables with nominal or
ordinal cale. La tly, not only conjoint analysi mea ure consumer preference for attribute
level variables, it can also determine the impact of level of each attribute on preference
(Bajaj , 1998).
Overall, wherea the output of the factor analysi on data collected from study 1 i
notew011hy from theoretical viewpoint, these findings are complemented by the conjoint
analysis which offers practical insights that can be readily used by practitioners, while also
contributing to the body of know ledge.

3.3

Operationalization

In order to operationalize the research design, in study 1, a research questionnaire

(Appendix 1) was used to collect quantitative data from subj ects. The questionnaire
consisted of 4 sections. Section 1 included four demographic item , namely, age, gender,
nationality and edu cation level. In Sections 2 to 4 respond ents were asked to rate platf01m,
u er and contributor related items in tenns of relative importance in their deci ion to
adopt a social networking site. The results of the survey were then used for an exploratory
factor analysis, results of which would forn1 the basis for the conjoint analysi in study 2.

27

The de · gn and operationalization of conj int analy is was perfonned in reD r nee with the
procedur propo d by
re ult

reen and riniva an (1978).

hown in Tabl 4.5 ba ed on the

f th factor analy i (Table 4.4) , a t tal of 6 attribute , each with two level were

con idered to create the fracti nal factorial de ign. There were a number of m tivations for
using binary choice level for each attribute. Fir t, it wou ld allow for an accurate, clear-cut
compari on betwe n u er ' choi e (for in tance, between fun-oriented and u cfulnes oriented u er content) . econdly given that ix attribute can be a rather large number for a
conjoint analysis, and consid ering the intangibl e nature f the product (i.e. SN ), binary
attribute would allow for keeping the number of conjoint profil es at 8 and th ereby
preventing re pondent fati gue.
Also , wherea in tudy 1 contributors' content and applications form ed a single
component, the two were intentionally treated as two separate attributes. The reasoning
behind tllis decision wa to allow for a better comparison between the imp01iance of
applications, and content created by SNE contributors. This seemed necessary, especially
considering the fundamental differences between applications and content as contributor
offerings, and also the organizations that create them. Moreover, having a separate
attribute for applications would shed light on the importance of openness to third party
contributors for SNSs fr01n a user perspective.
By making use of text and figures, each profile was pre en ted on a separate card ( ee
Appendix 2). As Carlsson, Frykblom, and Lagerkvist (2005) highlight, an introductory
script about the process of conjoint analysis and the necessity of providing realistic
responses may not only facilitate data coll ection, but can al o result in more accurate
re ponses. A such, respond ents were a ked to stud y in truction about thee perim nt

_g

b fore ompleting the urvey. After an introduction about th attribute and how th y
w uld change from on bundl to an th r, 110 univ r ity stud nt were a ked to rate each
card on a cale of 1 to 10 in tenn

f pr D renee ( 1 meaning not favo urabl e at all and 10

meaning very favourable).
3.3.1

Pilot tud y

For both tudie , pilot data coll ecti n wa conducted. F r tud y 1, re earch qu e tionnaires
were di tributed am ong a ampl e of 10 participant who were asked about their
under tanding and interpretation of que tion . Ba ed on participant feedb ack, wording of
one of the survey que tion was modifie d.
The same procedure wa fo llowed for tud y 2, the conj oint analysi . A total of 10
respondents were a ked to evaluate 8 cards each repre enting an NS profile. Based on
respond ent fee dback, cards were modified and pictorial illustrations were al o added in
order to reduce the cognitive chall enge fo r respondents (Appendix 2).
3.4

Samplin g

In line with numerous studies in this domain, Xu et al. (20 12) are of the view that

university students are proper subj ects fo r research on social networking sites, because of
their high usage of such websites. Following the same view, undergraduate and graduate
students at the University of N orthetn British Columbia were chosen as the ample for this
study. A table displ ay was set up and stud ents pa sing by were a ked for their
participation. Printed research qu estionnaire were distributed among the sample by the
author. For the firs t tage of the study a sampl e of 100 students was reached . For the study
2 (conj oint analy is) a total of 80 students participated. Sampling for the econd tage of

29

the tudy wa perform d in imilar mann r. In that, the campu c01nmunity w ere a ked to
participate in th

urvey.

3.5

Data A nal ysis

3.5.1

Factor Analy is

ormally the fir t t p before fac tor analy i

th Kai er-M eyer- lkin and Bart lett' tests

(Thomp on 2004). Kai er-Meyer- lkin (KM ) i a m ea ure of sample of adequacy. As a
rule of thumb , there ult of the te t is expected to be above 0.5 in order to proce d to the
fa ctor analy i (Dziub an &

hirkey, 1974 ). Al o Bartl ett 's te t of sph eri city examines th e

overall significance of all the correlation w ithin the correlation m atrix , and its results
require to be tati ticall y significa nt before m oving on to the fac tor analysis (Jack on,
1993) . Both tests were perfo rmed a p art of the fac tor analysi in IBM SPSS .

In regards with the outcome of factor analysis, the first step is to investigate the correlati on

m ahix, which presents the intercorrelations between all m ea ures . According to Field
(20 13), the procedure fo r redu cing the dilnensionality of the correlation m atrix is to look
for item s that correlate highly w ith a group of items but correlate poorl y with the rest.
Those meas ures that do co rrelate hi ghl y represent a ' factor', whi ch creates a new
dimension "that can be vis ualized as class ifi cati on axes along wh ich measurement
vari ables can be pl otted" (Field, 20 13) . Two score that merit special attention are factor
scores, whi ch are "the scores of a subj ect on a [ ... ] factor" (Rietveld & Van Hout, 1993 ),
and factor loadings, which show the conelati on of original variables with a fa ctor. F actor
loadin gs are part icul arl y helpful fo r determinin g the "substantive importance of a pm1icul ar
vari abl e to a factor" (Field, 2013) by squaring the factor loading and thereby dete1111ining
the amount of vari ance explained by a fact r.

30

3.5.2
Th

ONJOI T

LY I

nj int analy i card u

d [! r thi

rth g nal de ign g n rated by I M
cr ate a fra ti nal fact rial d
tati tically r pr
i

The

ript u

P

. A m ntioned arli r, rth g nal d

ign

ign which c ntain a et [ c nj int pr fi le that w uld

nt all p

n the number f attribut

u ed [! r data

tudy ( ppendi 2) w re pr par d ba ed n the

f attribut
and 1

( reen &

1 th de ign re ulted in

riniva an, 1

0) .

pr fil e which were

ll ecti n.
d for creati n f the de ign in P

wa th D 11 wmg :

T )

After data coli ction, analy i of re pond ents· ratin g of th e 8 co nj o int profile wa
performed in IBM PS

version 22 . Also two u ual goodn e

of fit tests, Pea r on' R

(Bollen & Barb , 198 1) and K endall' tau (Rom esburg, 2004) tatisti c, were u ed to
evaluate the extent to which the model can account for th e variance in re pondent ·
preference ratings.

31

Chapter Four: Findings

In thi chapter finding fr m the exploratory factor analy i ( tudy 1) and the conjoint
analy i ( tudy 2) are pre nted.
4.1

Stud y 1: Exploratory Factor Analy is

In thi

tud y 100 univer ity tud ent w r

propo ed group

of

mea ure

ask d to rate 15 item relative to the three

(plat[! 1m, u er

and contributors) in tenns of

importance in choo ing to adopt a ocial networking ite. The followin g ections present
the results from this study.
4.1.1

Sample Demographics

The demographics of participants in this stud y are shown in Table 4.1. 45 percent of
respondent were Canadian nationals, and 55 percent were international stud ent . The
average age of respondents was 27.3 years old . The maj01ity of respond ents (62%) were
betw een 21 to 30 years old. Also, 58% of respondents were male and the rest were female.
Table 4.1:

Summ ary of res pondent demographics

Count

Percentage

Nationality
Canadian
In dian
Iranian
Chinese
Oth er

45
13
13
7
22

45
13
13
7
22

15 to 20
21 to 30
31 to 40

14
62
17

14
62
17
7

Abo1'e 40

7

Male
Female

58
42

Age

Gender
58
42
3_

4.1.2

Descriptive tatistics

Table 4 .2 highlight the mean and tandard deviation valu

for the 15 iteJ.n envi aged to

mea ure the tlu·ee propo ed N comp nent : platfonn, u er and contributor .
Table 4.2:

De criptive tati tics for study 1

~

Item

(1)

Pol

::::;

VC/)
.......
< Pol

(1)

-· :::

:Pol

=·
0
::::;

0...
0-1

-I

0...

6.091

1.016

5.91

1.476

apabilitie

5.673

1.199

Platform Improvement

4.765

1.353

Fun derived from users ' co ntent

5.357

1.333

Utility derived from u er ' content

5.102

1.247

Number of registered friends

5.234

1.571

Number of registered members

4.418

1.804

Fun derived from contributor content

4.602

1.768

Utility derived frorn contributor content

4.857

1. 705

Fun derived from contributor applications

4.408

1.728

Utility derived from contributor applications

4.510

1.783

Privacy
Connecting

As the table highlights, ease of use has the highest mean value (mean = 6.091) followed by
privacy (mean = 5.918) . The item with lowest mean valu is fun derived from contributor
applications (mean = 4.408)

33

4.1.3

ample Adequacy

iv n the larg number of it m m a ured in tudy 1 it i important to ensure that the
ample ize i large enough to allow for the factor analy i . Table 4 .3 how there ult for
th KM

and Barlett' t t .

Table 4.3:

KMO and Barlett' te t re uJt

Kaiser-Meyer-Olkin Mea sure of ampling

Bartl tt' Test of pheri city

dequacy

Approx.

0.699

hi- quare

438 .410

1gma
As the result
the con elation

0.000

how, Bartl ett' te t of pheri city, which te ts the overall s ignificance of all
within the conelati on matrix, was

ignificant, indicating that it wa

appropriate to use the factor analytic model on thi s set of data. The Kaiser-Meyer-O lkin
m easure of sampling adequ acy indicated that the strength of the relationships among
variables was high (KMO = .69) , thus it was acceptable to proceed with the analysis.

4.1.4

Factor Analysis Findings

A factor analysis of the cunent results was p erformed using the Princip al Component
Analysis . Of the 15 proposed items presented in Table 1, 10 item s w ith eigenvalu es greater
than 1 were retained. A series of factor ana lyses were conducted which indicated that four
factors gave the most interpretable soluti on, and would explain an acceptable level of
variance. An Oblimin rotation, converged in 18 iterations, was perfonned since factors
were expected to be conelated . A

the scree plot in figure 4.1

bows, the first four

components have eigenvalu es above 1 and represent a significant portion of the varianc .

34

Figure 4.1:

cree plot
Scree Plot

3

cv
::J

"'c 2
>
Q,l

01

w

0
2

3

5

4

6

7

8

9

10

Component Number

The obtained pattern matrix IS displayed in table 4.4. Only items wi th factor loadings
above .5 0 are hown .
Table 4.4:

Pattern Matrix

Scale Items
1

Contributors - Content usefulness
Contributors - Applications usefulness
Contributors - Applications fun
Contributors - Content Fun

Component
2
3

.93
.89
.88
.86

Network - Nun1ber of registered members
Network - Number of registered friends

.80
.77

Users - Use rs' content usefuln ess
Users - Users' content fun

.91
.84

Platfonn - Privacy
Platform - a e of use
Percentage of Variance
igenvalue

4

.85
.65
7.32
3.7 2

15.42
l.543

12.52
1.2 52

11 .39
l.l 39
35

Th cumu lative variation explained by the four extracted compon nt wa found to be
76 .66 percent. Factor one accounted for 3 7.32 perc nt of the vmiation in data . This fa ctor
wa lab 1 d

ontributor , and con i ted of 4 item s dedicated t

ben fit derived from third party contributor
The previou ly envi aged u er

hedoni c and utilitarian

content and applicati n .

dimen ion wa

divided into two components labeled

N etwork and User . The Network component accounted for 15.42 percent of the vari ation
and con i ted of number of member

and real life fliends registered on the social

netw orking ite. The third component, labeled Users, con i t of hedonic and utilita1ian
benefit obtained from

N

u er and accounted for 12.52 percent of the variation. The

fourth and component, platform, accounted fo r 11 .39 percent of the variation and consisted
of two items, ease of u e and privacy.
4.2

Stud y 2 : Conjoint Analys is

The fa ctor analysi perfmmed in stud y 1 uncovered fo ur social networking si te
components, platform, users, network and contributors. But the extent to which perceived
characteristics of these components are valued by users in their decision to adopt a socia l
networking site rem ains a question. Conj oint analysis is a widely used statistical technique
that is utilized to determine the relative impmiance of attributes that indivi duals associate
with a produ ct or service, and the levels that make up those attri butes . Apart from
importance, conj oint analysis can also reveal the optimum product profile based on the
captured opinions ofu sers.

36

4.2 .1

Attribute and Level

r the c nj int de ign, th D ur pr vi u ly xtracted c mp nent w r u e d a a
d fin

pr du t attribut

Table 4.5:

t

ach with tw 1 v 1 ( ee Table 4 .5).

ttribute and le el for conjoint analy i
Level

ttribute

Fun

eful

un

U eful

ontribut r

Pl at~ rm

content' and two leve l : fun and u eful. The 'co ntribut r ' co mp nent wa repre ented by
two attribute , na m ly. 'contributor-g nerat d content' and ·applica ti on '. eac h with two
level :fun and u eful. Due to their di tin cti ve nature, th e tw 'p latfo rm ' related items,
' privacy' and · ea e of use·, were each u ed a separate attributes. The en vi aged level for
pri vacy we re "w ith co ntro l ove r privacy" a nd "a ll publi c" (n co ntro l over privacy), whi l
th e two leve l for the 'ea e fuse' attribute we re s impl y 'ea y to u e' and· ophi ti cated'
(see Appendix 2).

4.2.2

Re pondents ' Demographics

The demographi c of participant in thi

tud y is hown in Table 4 .6. The average ag of

re pondents was 26.2 year old , w ith 54.55 percent of re pondent being mal and th
r maining 45.45 percent wer female . A fo r ed u ati o n level, 59. 1 percent of re pondent
were und ergradu ate tud ent , 3 1. percent w re rna ter
percent w re Ph

stud ent . Moreover, 57 perc nt of ubj

tud ents and th e r maining 9. 1
t w re

anadian and the

r m aining 4 perc nt were int rnational tud ent .
17

Table 4.6:

Respondent demographics

Count

Percentage

Canadian
Indian
Iranian
Oth er

63
10
12
25

57.27
9.1
10.9
22.73

15 to 20

11

21 to 30
30 to 40
Above 40

9
53
12
6

66
15
8

Male
Female

60
50

54.55
45. 45

65
35
10

59. 1
31.8
9. 1

Nationality

Age

Gender

Education
Undergraduate
Masters
PhD

4.2.4

Conjoint Analysis Findings

The relative importance valu es of the six tested attri bute~ measured by the conjoint
instrument are shown in figure 8 below .
Figure 4.2:

Relative importan ce of Social Networking Site attri butes

-

40
35

-- -

-- - - - - - - -

36.164

I
r!

---

30
Q)

v

c:

i

0
0..

~--

'
.....
..."' 25 !

~-----

E 20

I

Q)

·.>
;:

"'
ex:

15

Q)

~--

-----

10
5

0

'

• user

contributor

• apps

• pri vacy

ease

netw ork

38

A the graph highlight , with a scor of 36.16, privacy wa found to have the highe t
r lative importanc . It wa follow ed by a e ofu e, which had a relative importance of
19.72. U er and contributor b nefit had relative importanc
re pectiv ly while applicati n

core of 12.73 and 11 .59

cored at 10.133 . The attribute with the mall est

importance c re was network (9 .652) . Result [! r the conjoint analy is are ummarized in
table 4 .7 below .
T able 4.7:

Conjoint anal is re ult

Attributes

Level

User-generated
content

Contributor-generated
content

Fun
Usefu l
Fun
Useful
Fun

Applications
Useful
With privacy control
Privacy

All public (no
control)
Easy to use

Ease of use
Sophi ticated

Network

Lots of registered
friends
Lots of registered
members

timates
.14
-. 14

Relative Importance

12.73

-. 15
.15

11 .59

.02
-.02

10.13

1. 30
- 1. 30

36.16

.60
-.60

19.73

.07
-.07

9.65

39

A highlight d in the previous chapter, a defining a umption of the conj oint in trument i
that the v rall utility of a product i the urn of utilities obtained from each attribute of the
product. Table 4.7 al o hi ghlight utility valu e for each level of each attribute.
In ord er to te t the g

dne

of fi t Pear on' R and Kendall' tau tati tics were used.

According to the r ult , the m odel had a good fit and there was a strong association
betw een the ob erved and e timat d utility va lu e . P ar on' R was fo und to be 0.975 and
Kendall ' tau wa fo und to be 0.96 1.
La tl y, resp nd ents' cho ic

we re analyzed to un cover th e optimum oc ial netwo rkin g site

profile. Ba ed on the result , the optimum bundl e is one where users share m ostly fu noriented po t , co ntri buto r ' po t are mainl y u efulnes -oriented, applications are fu nmiented, users are given contro l over privacy of shared posts, the pl atform is easy to use,
and a good number of real life f1iends are using the website. Tabl e 4 .8 below highlights
the three optim al SNS profiles based on the collected data:
Table 4.8:

Top 3 optimum SNS profiles

Attributes I Optimal
Product Profiles
User-generated content

Optimal SN S 1

Optimal SNS 2

Optimal SNS 3

fun

fun

fun

Contributor-generated
content

useful

useful

useful

Applications

fun

fun

useful

Privacy

with control over
pnvacy

with control over
pnvacy

with control over
pnvacy

Ease of use

easy to use

easy to use

easy to use

Network

lots of fri ends

lots of members

lot of fri end

40

Chapter Five: Discu sion, Conclusion, Contributions and Limitations

5.1

Discu sion

In thi re arch, bu ine

ec

y t m a well a the teclm logy acceptance model (TAM),

hedonic and utilitatian ben fits and a number f ther th mies were utilized to model
ocial networking ite . Based on a cone ptual model, three type of mea ure were
identified for N

nam ly, platform-related , u er-related and contributor-related. To

mea ure the impact of the e thre en vi aged type , a total of 15 items were identified ( ee
table 3.1) and were empirically te ted u ing a ample ofuniver ity students.
The factor analysi perfom1ed on the collected data highlighted ome new findings.
Hedonic and utilitarian benefit have been extensively used throughout the literature as
explanatory factors for technology adoption. Normally, the two type of benefits are
measured for the entire product or service. However, as discus ed in chapter 3, in order to
reach a better understanding of the origin of these benefits, hedonic (fun) and utilitarian
(useful) benefits obtained from users ' and contributors ' content and applications were
separately measured . Based on the results , respondents seem to have been able to
distinguish the effect of user from contributors clearly and the two were extracted as
separate components.
For the conjoint analysis, the impact of user and co ntributor benefit on individual ·
preference was tested using tlu·ee attributes each with two levels: 1) user-generated content
(fun/u eful) 2) contributor-generated content (fun/u eful) 3) contributor reated
applications (fun/u efu l) . Whereas numerou

tudi es in the literature of both bu ine

41

c

y t m empha iz th imp rtance f c ntribut r in the ucce

platfonn

thi r le ha rarely b

cc rding t th

n tudi din th d main f

of techn logy
ad pti n.

nj int analy i re ult , c ntrol ver priva y receiv d a con id rably

high r r lative imp rtance c re c mpar d t

ther attribute (36 .1 4) . Thi finding fall

111

lin with that fa tudy by Dwyer tal. (2007) in which auth r al o find privacy c ncerns
to influ nee indi idual'
Vari u pr

1

u

cia! n tworkin g b haviour c n id rably.

tudie which applied the techno! gy acceptance m de l (TAM) to

int m t techno Iog ie ha e found ea e of u e t be a deci ive fact r 111 hapin g u er '
preference (

e ~ r example, M .- . Lee (2009)· H .-P . Lu and Yu-Jen u (2009); Y. Lu et

al. (2009); M

n and Kim (200 1); Y en et al. (20 10)) . R

thi

ult of the conjoint analy i in

tudy al o hi ghlight the impm1ance of ease of u e, a the attribute received the econd

highe t relative importance va lu e (19 .732) among the ix tested attribute .
A highlighted in figure 7, the relative importance valu e obtained for u er-generated
content (12 .73) i hi gher than that of contributor-generated content (11.59) . However, the
relative importance value found for application (1 0.133) i lower than both u er and
contributor generated co ntent. Given the fact that only a mall number f SN
open to third party application developers, thi

are truly

omewhat lower importance value doe not

com e as a surpnse.
Initially, user benefits together with number of friends/member w re xpected to fotm a
single comp nent. However, the factor analy is ex tra ct d the number of real li£1 friend
registered on the N

and the overa ll number of regi tered member tog th r a a

component. A tudy by Papachari ss i (2009) in the domain of N

parate

ha empha ized th e

importance f di tingui hing between a per. n' friend . (rea l li ~ friend . also regi . tcred on

42

the
R

) and member (regi t red
ult

m mber who the u

r do

n t n ece ari ly know) .

f tudy 1 (tab! 4.2) h w that parti cipant find the numb r of fri nd to be m re

imp rtant (m an = 5.234) than the number f m ember (m ean = 4.41 ) in their deci ion to
ad ptan

nethe l

the numb r f m emb er did al o receive an abo e avera ge

rating fr m p a11icipant . T hi finding can b a
n etw ork effe t . Z hu and Ian iti (2 007) p

ciat d w ith prev iou w rk

it that indi idual ' p erc pti n

ad pt r of a techn 1 gy ca n impact their deci i n to ad pt. T he rea

ni

n indirect
[the number of
aid to be

indi v idu al ' exp ectati n of receiving a b tter vari ety of fferin gs fro m ex tern al entiti e a
th numb er f platf01m ad pter gr w .
tud y 1 and 2 both a k re po nd ent

to eva lu ate pr duc t

attribut

. Th

mam

difference between the two appr ach e i that in tud y 1, each attri bute i rated eparately,
wh erea in tudy 2 ( c njoint), re p ondent rate entire bundl es created by all attribute . A
di cu s ed in ch ap ter 3, compared to o liciting evalu ati on of ingle produ ct attribute , a
major advantage of conjoint analys is i that it forces re pond ents to ev aluate each attrib ute
relative to all o ther attributed at the

arne tim e, and the reby capture

eva lu ation s. H ence, as expected, there are di fference

more rea li ti c

between the re ult

of the two

studi es . Firstly, w hereas, in study 1 ea e-of-u se has received the hi ghe t m ean value in
terms of imp011ance (mean = 6.09 1) fo llowed by privacy (m ean = 5 .9 18), in tud y 2, it is
privacy that has th e highe t relative imp rtance and ea e-of-u e come in th

econd p ia e.

Mu ch in the sam e m anner, network re lated attribute (nu mber of m mbe r and fri end )
sw itc h place wi th con tributor offering (aJ p and c ntent) in tud y 2 and ha e th e low e t
relati ve importance score.

43

La tly there ult of the conjoint analysi allowed for identification of the optimum ocial
n tworking ite from the per pective ofu ers. A expected having control over privacy of
hared po t , ability to connect with real life friends and being ea y to u e are among the
characteri tic prefened by u er . Perhap more importantly however, the outcome
uncovers the prefened source of utility and hedonic benefits on social networking ites.
According to there ult , the top

of choice i one where other individual

hare mostly

fun content, while exten1al contributor are preferred to share content that is mostly useful.
Considering the popularity of game application on N

uch as Facebook, and QQ in

China, it doe not come as a surpri e that participant favoured fun social networking site
applications more than u eful one . There ult from robustness te t for both studies 1 and
2 showed ideal robustness and goodne

of fit.

44

5.2

Conclusion

iven th emergence of

a a popular method f c01nmunication and elf-ex pre ion,

it i critical for practitioner and cholar to learn ab ut the entities that fonn the e online
platfonn and the way perc iv d characteristi c of th e entitie
and ultimately, adoption. In thi

hape u er 'preference

tudy, the main building blocks of a typical SNS were

ucce sfully uncovered, and the relati n hip between identified component and u er '
preference was tudied . The multi-stage tudy revealed ome maj rand novel fmding .
Through theoretical conceptualization and an empiri cal study, initially, four component
were found to form a typical ocial networkin g ite from au er perspective. These
component were platform, u er , contributor and the network. Furthermore the conjoint
tudy revealed that the two platfmm characteristic , privacy and ea e of use, had the
hi ghest relative importance in shapin g users' preferences. Results of the stu dy also show
that hed onic and utilitarian benefits from u ers' shared content as well as content and
applications created and shared by external contributors are next in line in terms of relative
importance for users. Moreover, although not as signifi cantl y as other attributes, the
number of friends and m embers registered on an SNS also impact users' preference.
Lastly, the study was able to uncover the prefeiTed source of hedonic and utilitarian
benefits for users. Based on the results, in an optimum social networking site, user receive
hedonic benefits mainly from content shared by other users, whereas content generated by
contributors is the prefeiTed source of utilitarian benefits. In conclusion, this study was
able to reach the objectives it set out at the beginning. Through a eries of conceptual and
empirical studies, SNSs were modelled, and the relative imp01iance of ocial networking
ite components wa uncovered, all from a u er perspective.

45

5.3

Contributions

Thi

tudy mak s a number of major contributi n . Fir tl y, it contributes to bu ine

co y tern (B ) a a ignificant, yet under tudied tream of theory in strat gic
management. inc it intr duction, bu ine s ecosy tem ha been applied to numerou
contexts. In particul ar everal studi

have e tend d the noti on of busin ss co ys tem to

Infonnation Technology bu ines e . However, de pite it exceptional potential, thi
research i the fir t to trul y and directl y extend business ecosystem to
general, and SN

cial media in

in particular. To be preci e, here bu ine ses ecosy tem was utilized as a

fram ework fo r identifi cation of the ntitie that form a typ ical social networking ite.
Con equentl y, items fo r measurement of tho e entities were adapted either fro m BE it elf
or related theorie , uch a teclmology acceptance model (TAM), platform bu iness and
hedonic and utilitarian benefi ts. Through a number of analyses, the framework did provid e
a robust model for user preferences towards SNSs.
Secondly, this research makes variou contributions to the literature of social media and
Social Networking Sites. Numerous stud ies have investigated SNS in terms of single
factors, such as design, content, trust and security, and characteristics of u ers. Here, not
only new dimensions of SNSs were conceptually and empiricall y modeled, but also the
majority of previously identified SNS measures were also accounted for in the proce s of
measure development. Hence, the resulting framework is both novel and holistic.
Using business ecosystem as a theoretical backbone, the factor analysis identifi d four
components that form a typical social networking site, namely, platfonn (measured by
privacy and ease of use), users (hedonic and utilitatian benefit obtain d from other u ers),
network (number registered members and friend ) and contributor (hedonic and utilitarian

46

ben fits obta ined from contributor ' application and content). There ult can pave the
way for future re earch and further expansion of ach of the re ulting components. By
clarifying the trategic role of variou

ntitie that form the bu ine

eco y tern around a

typical ocial networking ite, there ult of th expl ratory factor analy is can also be used
a a framework for ocial media analy t ~ r analy i

f N s. Moreover, the applied

methodology, in which a combination of factor analy i and conj oint instrument wa u ed,
i a novelty that can be explored fu 1i her in the domain teclmology adoption .
Re ult of the conjoint analy is fo und the two platfo rm related characte1istics, ea e of use
and p1ivacy, to be the most important attribute for social networking ite adopters. The
takeaway, especially fo r developer , is that the two above menti oned characteristics merit
special attention in order to reach the goal of maximizing adoption rates of social
platfonns. Furthermore, another finding of the conjoint analysis was that, for respond ents,
benefits obtained from external contributors are almost as important as benefits obtained
from other users. At the same time, aggrega ted captured ratings how that, while
respondents prefer SNSs where they can obtain hedonic benefit from other users, they
prefer the content shared by external contributor to be mostly utilitarian, and their
applications to be hedonic. Wha t thi s finding means for social media strategi ts is that
faci litating the engagement of contributors can be almost as important as u ers. For
contributors sharing content on social networking ites, findings of the conjoint analy is
mean that their content is prefen ed to be more utilitaria n than hedonic, while the takeaway
for app developers is that, overall , SNS applications are pre~ n ed to be more h doni e than
useful. Re ults indi cate that users, on the other hand, need to fo cu more on creation and
sharing of mo tly hedoni c content in ord er to expand their social n tworking reach.

47

Furthermore, another novelty of th re earch de ign i in how it measure h donie and
utilitari an benefit . Numerou

cholar have found hedonic and utilitarian benefits to be

key player in any form of adopti n. However, by and larg , the two have almo t always
been treated a

eparate variable mea ured for the entire produ ct/sy tern . In this tudy,

however, a choice wa mad e to mea ure hedonic and utilitarian benefit not for the entire
ystem, but to con ider them separately ba ed on the type of offering and type of entities
from which they originate.
Al o importantl y, another way in which the cunent re earch differentiates it elf from prior
re earch is in how it views NS adoption. Whereas previou research tudies factor that
impact adoption or non- adoption of SN s by bu inesse or individuals as a whole, this
study aims at uncoveri ng the mot ives behind individual ' ad option of certain SNSs over
others. As a result, it clarifi es the competitive advantages of SNS s from a user perspective,
which can be of considerable valu e to practitioners in this domain.

5.4

Limitations and Future Studies

The author is of the view that the impact of nationality, language and socio-cultural
characteristics of the local enviromnent on choice of SNSs would have b een captured
better through sampling in more than one country (perhaps in Canada and China for
instance). This was not within the scope of this study, but can be explored in the future .
Moreover, data collection for both studi es was performed at the Univer ity ofNorthern
British

olumbia only. Given the higher usage of SNSs by students, the sample may not

represent the preferences of the entire population perfectly.
Another limitation of the study was the number of levels u cd for the conjoint analysis
perfonned. In order to ensure the number conjoint in trument bundle would tay at a

4

manageable number for participant , it was decid ed to have dichotomou levels for each
attribute. U ing more level in the de ign, po ibly with middle-ground value , may lead to
different conjoint analy i re ults, albeit, with a higher number of bundle .
A po ible area of improvement for fu ture tudies in thi domain can be exploring the
impact ofu er ch aracteri tics on N preference . For instance, age, education and gender
are user related variabl es that may impact the adoption deci ion of individuals.
Lastl y, accounting fo r individual obj ectives for N u age i a direction that can be
explored in the future. Wherea pl atforms uch as Facebook and lnstagram are primarily
used fo r hedoni c purpo es, web ites such as Linkedln are mostly utilized for professional
purpo es. Users of these SNS are expected to have different obj ective for u age, which
may also impact their adoption decision.

49

Reference
dn r R. (2006). Match your innovati n strategy to your inn vati n ecosystem. Harvard busines
re1 •i w, 84( 4), 9 .
upta, AJok, & Kauffman, Robert J. (2006).
Under tanding patterns of technology e1•olution : An eco ystem per p ective. Paper
pre ent d at the yst m ciences, 2006. HI
'06. Proceedings of the 39th Annua l Hawaii
Intemati nal
nference on.

Adorna icius,

dimina , B ck tedt, Jesse,

garwal, Ritu
enkate h, Viswanath. (2002). A e sing a firm's web presence: a heuristic
valuation pr cedure for the measurement of usability. Information Systems Research,
13(2), 16 -1 6.
Anggraeni , . Hatiigh, .d. , & Zegveld, M. (2007). Business ecosystem as a per pective for
tudying the relations between firms and their business networks. presented at the ECCON

2007Annual m eting, B rgen aan Zee, The Netherlands.
Babin , Barry J, Darden, William R , & riffin, Mitch. (1994). Work and/or fun: measuring hedonic
and utilitarian shopping value. Journal of con umer research, 644-656.
Bajaj,

. (1998). ,,Technology Factors Influencing enior Information System Managers'
Decisions to dopt New Computing Architectures ", Pits burgh : The Heinz chool. H einz

School, Carnegie M lion Uni1• .. Pittsburgh, PA.
Baloglu, eyhmus, & Pekcan, Yakup A. (2006). The website design and Internet site marketing
practices of upscale and luxury hotels in Turkey. Tourism Management, 27(1), 171-176.
Barnes, Stuart J. (2005). Data triangulation in action: using comment analysis to refine web quality
metrics.
Benbunan-Fich, Raquel. (200 1). Using protocol analysis to evaluate the usability of a commercial
web site. Information & Management, 39(2), 151-163 .
Bergvall-Karebom, Birgitta, & Howcroft , Debra. (2011). Mobile applications development on
Apple and Google platforms. Communications of th e Associationfor Information System ,
29(1), 30.
Bilsel, R Ufuk, Buyiikozkan, Gtil<;:in, & Ruan , Da. (2006). A fuzzy preference-ranking model for a
quality evaluation of hospital web sites. Int ernational Journal ofIntelligent S.vstems.
21 ( 11) ' 1181-1197.
Bollen, Kenneth A, & Barb, Kenney H. ( 1981) . Pearson's rand coarsely categorized measures.
American Sociological Review, 232-239.
Bosch J. (2009). From softvt•are product lines to software ecosystems. Pap r pre ented at the
Proceedings of the 13th International oftware Product Line onference, an Franci co,
alifomia.

0

Bo ch, J., & Bosch- ijtsema, P. (2010). From integration to composition: n the impact of
software product lines, global development and ecosystems. Journal of Sy terns and
Softvvare, 3(1), 67-76 .
B udreau , Kevin I & Hagiu, Andrei . (2009). Platform rule : Multi-sided platforms as regulators:
Cheltenham, UK: dward Elgar Publishing Limited.
Boyd, Danah M & . . lli on, Nicole B. (2012) . ocial network sites . Online Communication and
Collaboration : A Reader.
Bradley, M , & Kroes , P. (1992) . Fore asting i
m fat ed pref erence un 1ey research. Paper
pre ented at the
L TED
ADING IN TRAN PORT SURVEY
METHOD LOGY. DITED PRO
DING OF TH 3RD INTERNATIONAL
0 F
N
ON URV YMETHOD IN TRAN PORTATION , JANUARY 5-7
1990, WA HINGTON, D .
Brooke John. ( 1996) . U -A quick and dirty usability scale. Usability evaluation in industry,
189( 194), 4-7 .
Brown, Jo, Broderick, Amanda J, & Lee, Nick. (2007) . Word of mouth communication within
online communities: onceptualizing the online social network. Journal of interactive
marketing, 21(3 ), 2-20.
Brush, Candida G. (2014). Exploring the concept of an entrepreneurship education ecosystem.
Inno1 •ative Pathways for Unh ·ersity Entrepreneur hip in the 21st Centwy, 25-39 .
Bryant, Fred B, & Yamold, Paul R. (1995) . Principal-components analysis and exploratory and
confinnatory factor analysis.
Cao, Mei, Zhang, Qingyu, & Seydel , John. (2005) . B2C e-commerce web site quality: an empirical
examination. Industrial Management & Data Systems, 1 05(5) , 645-661 .
Carlsson, Fredrik, Frykblom, Peter, & Lagerkvist, Carl Johan. (2005). Using cheap talk as a test of
validity in choice experiments. Economics Letters, 89(2) , 147-152 .
Chau, Patrick YK. (1996). An empirical assessment of a modified technology acceptance model.
Journal of management information systems, 185-204.
Cheung, Wai Man, & Huang, W . (2002) . An investigation of commercial usage of the World Wide
Web : A picture from Singapore. International Journal of Infonnation Management, 22(5),
377-3 88.
Chu, Shu-Clman, & Kim, Yoojung. (20 11 ). Detenninants of consumer engagement in electroni c
word-of-mouth (eWOM) in social networking sites. International journa l of ad1'C!rtising,
30(1), 47-75 .
ombe,

avid , Largeron , Christin e, Egyed -Zsigmond , Elod, & Gery, Mathias. (2 0 10). A
comparative study ofsocial network analysis tools. Paper presented at the Internationa l
Workshop on Web Intelli gence and Virtual Enterprises.

dwards , arrie, Fraser, Kristen L, Gudelsky, beryl , Holmqui t, Jenny, Thornton,
& weetser, Kaye D. (2010). Adoption of social media for public relations by
nonprofit rganizations . Public Relation Review, 36(1), 90-92.
de Reuver, Mark, & Bouwman, Harry. (2012). Governance mechani ms for mobile service
innovation in value networks . Journal of Busin e Research, 65(3) , 347-354.
den Hartigh, Erik, Tol, Michie) , & Vi s cher, Wouter. (2006). Th e health mea urement of a business
ecosy tern . Paper pre en ted at the Proceeding of the uropean Network on Chaos and
omplexity Research and Management Practice Meeting.
Diakopoulos, Nicholas, Naaman, Mor, & Kivran - waine, Funda. (2010) . Diamonds in th e roug h:
ocial media 1'i ual ana ly tics for journalistic inquiry. Paper presented at the Visual
Analytics cience and Technology (VA T) , 2010 I E ymposium on.
Dwyer,

atherine, Hiltz, tarr, & Pas erini, Katia . (2007). Trust and privacy concern within social
networking sites: A compari son ofFacebook and M y pace. AMCIS 2007 Proceedings,
339.

Dziuban, Charles D , & hirkey, Edwin C. (1974). When is a correlation matrix appropriate for
factor analy is? ome deci sion rules. Psychological Bulletin, 81(6) , 358 .
Eastin, Matthew . (2002). Diffusion of e-comrnerce: an analysis of the adoption of four ecommerce activities. Telematics and informatics, 19(3 ), 251-267 .
Eisenmann, Thomas, Parker, Geoffrey, & Van Alstyne, Marshall W . (2006). Strategies for twosided markets. Han,ard business review, 84(1 0) , 92.
Ellison, Nicole B . (2007) . Social network sites: Definition, hi story, and scholarship. Journal of
Computer-Mediated Communication, 13( 1), 210-230.
Ellison, Nicole B , Steinfield, Charles, & Lampe, Cliff. (2007) . The benefits of Facebook "friend s:"
Social capital and college stud ents' use of online soc ial network sites. Journal of
Computer-Mediated Communication, 12(4), 1143- 1168.
Evans, D .S. , Hagiu , A. , & Schmalensee, R . (2006). Invisible engines: how sofm•are pla({orms drh •e
innovation and transform industries: The MIT Press.
Evans, David S. (2003) . Some empi ri cal aspects of multi -si ded platfonn industries. Re1·iew of
Network Economics, 2(3).
Feij6o, Claudio, Pascu, Carina , Misuraca, Gianluca, & Lusoli, Wainer. (2009). The next paradi gm
shift in the mobile ecosystem: Mobile social computing and the increasing relevance of
users . Communications & Strategies(75), 75.
Field, Andy. (2013) . Discovering statistics using IBM SPS statistic : age.
abszewicz, Jean, & Wauthy, Xavier. (2004). Two-sided market s and price competition with
multi-homing. Available at SRN 975897.
awer, A ., &

usumano, M .A. (2002). Pla(/orm leadership : Harvard Busi ness

hoot Press.

52

tes eke, Rapha l. (2014). Th lectric mobility bu in e e osystem. Paper pr sented at the
col gical ehicle and Renewabl nergJ ( V R) , 2014 Ninth Int rnati nal
onfer nee on.
J ka , Mina , iri ian
Mi cha el, Mark poul u thina, Yang, Xia wei. (200 ). Poking
.fa "book: haracterization of osn applications. Paper pre en ted at the Pr ceeding f the

fir t work h p on

nJine

ial n tw rks.

ons:alve , Vani a, & Bail on, Pi eter. (20 J J ). Adding va lue t th e netw rk: Mobile operators'
experiment with ftware-a -a- ervice and Plat~ nn-a -a- ervic m d I . Telematic and
Informatic , 28( 1), 12-21.
r

n, Paul ~, Krieger , bba M, Wind , Yoram. (200 1). Th irty yea r of conj oint ana ly is:
Refl ecti n and pro pect . Interfaces, 31(3_ uppl ement), 56- 73.

reen, Paul , & rini a an, Venkat. (1990).
nj int anal y is in marketing: new developments
with implications£ r re earch and practice. Th e Journal ofMarketing, 3-19.
reen , Paul , & riniva an, Venkatachary. (197 ). onj oint ana ly i in consumer research : i sue
and out look. Journal o_f consumer research, 103-123.
Gueguen ael, & I ckia, Thierry. (20 11). The borders of m bile hand et eco ystems: I
coopetition inevitable? Telema tics and informatics, 28(1) , 5-1 1. doi :
10.10 16/j .tele.20 10.05.007
Gunst, Ri chard , & Mason, Robert L. (2009). Fractional factorial de ign. Wiley Interdi ciplinary
R e1·iews: Computational Sta tistics, 1 (2) 234-244.
Gustafsson, Anders, Hem11ann , Andrea , & Huber, Frank. (2013). Conjoint measurement: methods
and application : pringer cience & Busine s Media.
Hair, Joseph F. (1984). Multi1 •ar Data Anafy is with Readings: Macmillan Pub.
Hanna, R., Rohm, A., & rittenden, V.L. (20 11 ). We're all connected: The power of the social
media ecosystem. Business Horizons.
Hanna, Richard, Rohm , Andrew, & 1ittenden, Viet riaL. (20 11 ). We're all connected : The pm..,er
of the social media ecosystem. Business hori::.ons, 54(3), 265-273 .
Harland , Peter , Wust, Sabine, & Dcdehayir, zgur. (20 14). fnn o1 ·ation processes in husiness
eco.\ ystems: Creating a common understanding hy requirements Paper presented at the
Management f Engi neering & Technology (PICMET), 20 14 Portland International
onference on.
Hearn , G., & Pace, . (2006). Value-creating
sys tem . foresight, 8( 1), 55-ES.

g1 : under tandmg next genera ti on busine,

Horan, Thomas , Abhichandani, Tarun, & Raya lu, Raghu vira. (2006). ,1sscssing user satisfaction
of e-gm ·crnm ent sen ·ices: de1·efopment and testing of qualit1'-iii-11Se satisfaction mth
advanced trai'C'Ier infornwtion .\ ystcms (AT!,~ Paper present -d at the , yst '111 , cienccs,
2006. JII SS'06. Proceedings of th e 9th nnualll awmi International onfcrcnce on

Hwang, Mark I, & Thorn, Ron . (1999). The effect of user ngagement on system success: a
meta-analytical integration of re earch finding . Information & Management, 35( 4), 229236.
lachello, io anni, mith, Ian onsolvo, unny, hen, Mike, & Ab wd, regory D. (2005).
D eveloping privacy guide/in for social location disclosure applications and service .
Paper pre ented at the Proce dings of the 2005 sympo ium on Usable privacy and security.
Iansiti, M., & Levien, R. (2004).

trategy as ecol gy. Harvard business review, 82(3).

Ian iti , Marco, & Levien Roy. (2002). The New Operational Dynamic of Bus in s Ecosystems:
Implications.for Policy, Operations and Technology trategy: iteseer.
Iansiti, Marco, & Levien, Roy. (2004a) . The keystone advantage: what th e new dynamics of
business cosystem mean for strategy, inno1•ation, and su tainabi/ity: Harvard Business
Press.
Iansiti , Marco, & Levien, Roy. (2004b).

trategy as ecology. Harvard busine s re1 •iew, 82(3), 68-

1.

Iansiti , Marco, & Richards , Gregory L. (2006). Information technology ecosystem:
health, and perfonnance, The. Antitrust Bull., 51, 77.

tructure,

Jackson , Donald A. (1993). topping rules in principal components analysis: a comparison of
heuristical and statistical approaches. Ecology, 2204-2214.
Jennrich, R.I. , & Sampson, P. F. (1966). Rotation for simple loadings. Psychometrika, 31(3), 313323. doi : 10.1007/BF02289465
Jolliffe, Ian . (2002). Principal component analysis: Wiley Online Library.
Jolliffe, Ian. (2014). Principal Component Analysis Wiley StatsRef Statistics Reference Online:
John Wiley & Sons, Ltd.
Jones, Chris, Ramanau , Ruslan, Cross, Simon, & Healing, Graham. (2010). Net generation or
digital natives: is there a distinct new generation entering university? Computers &
Education, 54(3), 722-732.
Kaplan, Andreas M , & Haenlein, Michael. (20 10). U ers of the world, unite! The challeng s and
oppmtunities of Social Media. Business horizons, 53(1 ), 59-68.
Katz, M .L., & Shapiro, C. (1994). Systems competition and network effects. The Journal of
Economic Perspectives, 8(2).
Kilamo, T., Hammouda , I., Mikkonen, T., & Aaltonen, T. (20 11 ). From proprietary to open
rowing an open source ecosystem. Journal of ystems and Software.
Kim, Jlee-Woong, Chan, Hock huan, & upta, umeet. (2007) . Value-based ad ption of mobile
internet: an empirical investigati n. Decision Support ystems. 43( l ), 111-126.
Kim, Soyoung, & Stoel, Leslie. (2004). Dimensional hierarchy of retail website qua lit
Information & Management, 41 (5) , 619-6
54

Kinnunen, T., Malvalehto J., & Haapasalo, H. (2011). Theoretical reflections on busines
eco y tem a e seminconductor industry.
La vie, Talia, & Tractinsky, Noam. (2004). A e ing dimensions of perceived visual aesthetics of
web it s. International journal of human-computer tudies, 60(3), 269-298.
Lee, hei ian, & Ma, Long. (20 12). News haring ins cia! media: The effect of gratifications and
prior exp rienc . Computer in Human Behavior, 28(2) , 331-339.
Lee, Ming- hi. (2009). Factors influencing the adoption of internet banking: An integration of
TAM and TPB with perceived risk and perc ived benefit. Electronic Commerce Research
and Application , 8(3), 130-141.
Li , Yan-Ru. (2009) . The te hnological roadmap of isco's business ecosy tem. Te hno1 •ation,
29(5) 379-3 86. doi : 10.10 16/j .technovation .2009 .0 1.007
Lin, Hsiu-Fen . (2007) . The impact of website quality dimensions on customer satisfaction in the
B2 e-commerce context. Total Quality Management and Busin ess Excellence, 18(4) , 36337 .
Louviere, Jordan J. ( 19 8) . Analyzing decision making: M etric conjoint analysis : Sage.
Lu, Chao, Rong, Ke, You, Jianxin, & Shi , Yongjiang. (2014). Business ecosystem and
stakeholders' role tran sfonnation: Evidence from Chinese emergin g electric vehicl e
industry. Expert System with Applications, 41(10) , 4579-4595 .
Lu, Hsi-Peng, & Yu-Jen Su, Philip . (2009) . Factors affecting purchase intention on mobile
shopping web sites . Internet Research, 19(4) , 442-458.
Lu, Yaobin, Zhou, Tao, & Wan g, Bin . (2009). Explorin g Chinese users' acceptance of instant
messaging using the theory of planned behavior, the technology acceptance model, and the
flow theory. Computers in Human Behavior, 25(1) , 29-39 .
MacMillan, Douglas, Burrows, Peter, & Ante, Spencer E. (2009). Inside the app economy.
Business Week, 4153 , 347-357.
Mangold, W Glynn, & Faulds, David J. (2009) . Social media : The new hybrid element of th e
promotion mix . Business hori::ons, 52(4) , 357-3 65 .
McCrae, Robert R, & John , Oliver P. (1992). n introduction to the fi ve-factor model and it
applications. Journal o.fp ersonality, 60(2), 175-215 .
McKinney, Vicki , & Yoon, Kanghyun . (2002). The measurement of web-customer sa ti sfacti on: An
expectation and disconfinnation approach. ll?formation . ystems research. 13(3), 296-3 15.
Moon, Ji -Won, & Kim , Young-Gul. (2001) . xtending th e TAM for a World -Wide-Web cont ext.
Information & Management, 38(4) , 2 17-23 0.
Moore, J.F. (1993). Predators and prey: a n w ecology of co mpetiti on. Han ·anl business rel'ie11 '.
71.

5

Moore, J.F. (1996). The death of competition: leadership and strategy in the age of bu ine s
eco y !ems: HarperBu ines New York.
Moor , J.F. (2006) . Business ecosystems and the v]ew from the firm. Antitru 1 Bull., 51.
Moore, James F. (1993). Pr dator and prey: a new ecology of competition. Harvard busine s
review, 71(3) , 75- 6.
Nikou, hahrokh, Bouwman , Harry & de Reuver, Mark. (2014). A onsumer Perspective on
Mobile ervice Platfonns :
onjoint Analysis Approach. Communications of the
Association f or Information y tem , 34(1) , 82.
Pai , Peiyu , & Arnott, David . (20 13) . User adoption of social networking sites: liciting uses and
gratifications through a means-end approach. omputers in Human Behavior, 29(3) , 10391053 .
Palmer, Jonathan W . (2002). Web site usability, design, and perfonnance metrics . Information
systems re earch, 13(2) , 151-167.
Papacharissi, Zizi. (2009). The virtual geographies of social networks: a comparative analysis of
Facebook, Linkedln and A mallWorld. N ew media & society, J 1(1-2) , 199-220.
Park, Young A , Gretzel, Ulrike, & irakaya-Turk, Ercan. (2007) . Measuring web site quality for
online travel agencies. Journal ofTrm•el & Touri m Marketing, 23 (1) , 15-30.
Peltoniem:i , M . (2006) . Preliminary theoretical framework for the study of bus]ness ecosystems.
EMERGEN CE MAHWAHLA WREN CE ERLBAUM, 8(1) , 10-19.
Porter, M.E. (1998). Competitive advantage: Creating and sustaining superior p elfo rmance: Free
press.
Quaadgras , Anne. (2005). Wh o j oins the platform ? Th e case of the RFID business ecosystem. Paper
presented at the System Sciences, 2005 . HIC S'05. Proceedings of th e 38 th Annual Hawaii
International Conference on .
Rietveld , Toni , & Van Hout, Roeland . (199 3). Statistical techniques/or the stud.J· a_{ lang uage and
lang uage behaviour: Walter de Gruyter.
Romesburg, Charles. (2004). Cluster analysis f or researcher : Lulu. com .
Rong, Ke, Hu, Guangyu, Lin, Yong, Shi, Yon gj ian g, & Guo, Liang. (201 5). Und erstanding
business ecosystem using a 6C framework in Internet-of-Things-based sector .
International J ournal o_f Production Economic , 159(0), 41 -55. doi:
http://dx. doi .org/ 10. 10 16/ j.i jpe.20 14. 09. 003
Ross,

rai g, Orr, -. miJy , Sisic, Mi a, Arsenea ult , Jaime M, Simmering, Mary , & Orr, R Robeti.
(2009) . Personality and moti va tion s as ociated with Facebook u e. Computers in Human
Behm'ior, 25(2), 578-58 6.

56

ubrahmanyam, Kaveri, & reenfi ld, Patricia M. (2008). Virtual worlds in development:
Implications of ocial networking sites. Journal of Applied D evelopmental Psychology,
29(6), 417-419 .
ubrahmanyam, Kaveri , Reich , tephanie M , Waechter, Natalia, & spinoza, Guadalupe. (2008).
Onlin and offline ocial networks: Use of social networking si tes by emerging adults.
Journal ofApplied D evelopmen tal P ychology, 29(6), 420-433.
ugianto , Ly-Fie, & Tojib, Dewi Roo !ani . (2006) . Modelin g user sa tisfaction with an employee
p rtal. International Journal of Bu in es and Information, 1 (2), 239-255 .
zymanski, David M , & Hise, RichardT. (2000) . £-satisfaction: an initial examination. Journal o.f
retailing, 76(3) , 309-322.
Tee, Richard, & Gawer, Annabelle. (2009) . Industry architecture as a determinant of successful
platform strategies: A ca e study of the i-mode mobile Internet service. European
Managem ent Review, 6(4), 217-232 .
Thackeray, Ro emary, eiger, Brad L, Hanson, Carl L, & McKenzie, James F. (2008). Enhancing
promotional strategies within social marketing programs: use of Web 2.0 social media .
H ealth promotion practice, 9( 4) , 338-343.
TheNextWeb. (2015). Facebook passes 1.23 billion monthly active users , 945 million mobile users,
and 757 million daily users. Retrieved 27 .01.2015 , 2015 , from
http://thenextweb.com/facebook/20 14/0 1129/facebook -passes-1-23-billion-monthly-activeusers-945-mi Ilion -mobile-users-7 57 -million-daily-u sers/
Thompson, Bruce. (2004). Exploratory and confirmatory fa ctor analysis: Understanding concepts
and applications: American Psychological Association.
Vander Heijden, Hans. (2004). User acceptance of hedonic information systems . MIS quarterly,
695-704 .
Varma Citrin, AJka, Sprott, David E, Silverman, Steven N , & Stem Jr, Donald E. (2000). Adoption
of Internet shopping: the role of consumer innovativeness. Industrial management & data
systems, 1 00(7) , 294-300.
Venkatesh, Viswanath, Morris, Michael G, Davis, Gordon B, & Davi s, Fred D. (2003). User
acceptance of information technology: Toward a unified view. MIS quarterly, 425-4 78 .
Von Hippe! , Eric. (2005). Democratizing innovation: The evo lving phenomenon of user
innovation . Journalfur B etriebswirtschc{ft, 55(1), 63-78.
Vos,

. (2006) . Business Ecosystems: Simulating ecosystem g vemance. Department of
Technology, Strategy and Entrepreneurship.

Vuori , Elisa K . (2005) . Kn owledge-intensi, •e service organi::ations as agents in a busines,
ecosystem . Paper presented at the ervices ystems and ervices Manage ment, 200
Proceedings of ICSS M'05. 2005 International onference on.

57

Weyl,

len. (2010) . A price theory of multi-sided platforms. The American Economic Review,
1642-1672.

Wu, J n-I-Ier Wang, lm- hing, & Lin, Li-Min. (2007). Mobile computing acceptance factors in
the healthcare industry: A structural equation model. International journal of medical
informatics, 76(1), 66-77 .
Xiaoren, Zhang, Ling, Ding, & Xiangdong, Chen. (2014) . Interaction of Open Innovation and
Business Ecosystem. International Journal of u-and e-Service, Science and Technology,
7( 1) ' 5 1-64.
Xu ,

henyan, Ryan , berry, Prybutok, Victor, & Wen, hao . (2012). It is not for fun: An
examination of social network site u age. Information & Management, 49(5) , 210-217 .

Yen, David , Wu , hin- han , Cheng, Fei-Fei, & Huang, Yu- Wen . (20 10) . Determinants of users'
intention to adopt wireless technology: An empirical study by integrating TTF with TAM.
Comput r in Human Beha1 •ior, 26(5) , 906-915.
Yoo

oonduck, Choi , Kwangdon , & Lee, Malrey. (2014) . Business Ecosystem and Ecosystem of
Big Data Web-Age Information Management (pp. 337-348) : pringer.

Yu , J., Li , Y. , & Zhao, C. (2011) . Analysis on Structure and Complexity Characteristics of
Electronic Business Ecosystem. Procedia Engin eering, I 5, 1400-1404.
Zhang, J. , & Liang, X.J . (2011) . Business ecosystem strategies of mobile network operators in the
30 era: The case of China Mobile. Telecommunications policy, 35(2), 156-171 .
Zhu, F ., & Iansiti , M. (2007) . Dy namics ofplatform competition : Exploring the role of insta/!ed
base, platform quality and consumer exp ectations: Division of Research, Harvard Business
School.

58

Appendix 1 - Re earch Questionnaire
ection I - Background
1. Age

2. Plea e ircle the highest year of school completed
1 2 3 4 5 6
(Primary)

7

9 10 11 12
(High chao!)

13141516
17 18 19 20 21 22 23+
(College/ Univer ity)
(Graduate School)

3. Plea e circle your nationality. lf other, please specify you nationality.
a.

anadian

4. Gender:

b.

hines

c. Indian

Mal

eO

Female

e. Other (Please spe (fy) ............. ..

d. Iranian

D

Other

D

There are numerous characteristics that one can ass ciate with social networking si tes . Please rate
the importance of the following attributes for you when choosing to join and use a social
networking si te. 1 represents very unimportant and 7 represents very important.
How important do you find the following characteristics when adopting (joining and using) a social
networking site?
Very
neutral
Very
Criteria
unimportant
important
1. Ease of Use

1

2

3

4

5

6

7

2. Privacy

1

2

3

4

5

6

7

3. Connecting capabilities

1

2

3

4

5

6

7

4 . Improvements to the website

1

2

3

4

5

6

7

Users
4 . User engagement

1

2

3

4

5

6

7

5. How fun (enjoyable) user
generated content on the social
networking si te is.

1

2

3

4

5

6

7

6. How useful user genera ted
content on the social networking
site is

1

2

3

4

5

6

7

7. The number of people I know
in real life (offline) who are on
the social networking site

1

2

3

4

5

6

7

8. Number of members who are
on the social networking site

1

2

4

5

6

7

9

App endix 1 continu ed
ontributors
ontributors are external organizations or individuals who create and share either content (news,
phot , video , etc) or applications and games on social networking sites. Please use the criteria
below to rate the following attribute r lative to contributors in terms of importance.
9. Vari ty of content (news
photo , video , etc.) provided by
third party contributors

1

2

3

4

5

6

7

10. Variety of applications
provided by third party
contributors

1

2

3

4

5

6

7

11. njoyment (fun) derived
from content provided by third
party contributors

1

2

3

4

5

6

7

2

3

4

5

6

7

12 . Utility (usefulness) derived
from content provided by third
party contributors
13. Enjoyment (fun) derived
from Applications provided by
third party contributors

1

2

3

4

5

6

7

14. Utility (usefulness) derived
from Applications provided by
third party contributors

1

2

3

4

5

6

7

60

App endix 2 - C onjoint instrument card s

This ite gives you control

Other use rs share mostly
Entertainin

over privacy o f content you

Posts

share (Who sees what)

This social network is

Businesses (Such as news

simple to use

o r ent rtainmen t co mpanies)
Social Network 1

A lot of your friends are

(

Apps you get through this site

usmg this soc1al network

are mostly Entertaining

No control over privacy of

Other users share mostly

posts. All your posts are pu blic.

Entertaining Posts

Businesses (Such as news or

This social netvvork is

enterta inment companies) share
~~CL.!:=~=.c:..:..:..:..:.. posts

geeky and difficult to use
ocial Network 2

Applications you get hrough
h1s srte are mostly Entertaining

A lo t of people are using this
sacral network

61

Appendix 2 continued

ThlS srte giVes you control over

Ot her users share mos t ly
Entertain in

privacy of con en you share (Who

Post s

seeswha)

Bus i nesses (Such as news or

This social net wo k is si mple

enter ainmen companies) share

to use
ocial Network J

of people are using this
social n twork

App lica t ions (apps) you get through
htS srte are mos ly U5eful

•
0 her users share mostly

Enterta inin

Posts

Busi nesse s (S.Jch as news or

This social networ is geeky

en ertainment companies) share

and difficult to use

mostly useful (i.e. Informative) posts

•
Applications (apps) you ge through
th1s srte are mostly useful

Socia l Network 4

Appendix 2 continued

Other users shar mostly useful
ThiS srte gtves you control o v r

(i.e . informative) Posts

privacy of con en you share (Who

•

s es what)

Bus inesses (Such as news or

This social ne twork is geeky

en er tai1men com panies) share

and di ffi cul t o u

m ost!

entertain i ng posts

Social Network 5

soCial network

Applications (ap ps) you get through

l:tlJ

h1s srte are mos ty useful

'
Other users sha1e mostly useful
(i. e. informative) Posts

•
0

Bus i nesses (Such as news or

This soc1al networ is simple

entertainment companies) share

to use

mostly entertaining posts

Social Network 6

A lot of people arp using this
Applications (apps) you get throu h
thiS srte are mostly useful

<.Oudlr twork

ppendix 2 continued

Other users share mos ly useful
Thrs srte grves you control over

(i.e. in fo rmative) Posts

priva cy of con ten you share (Who

se s wha)

'
Businesses (Such as news or

This social ne work is geeky

entertainmen compani s) share

and difficul to use

mostl

useful (i.e informative) po s

.

So ia l Network 7

•
•o

A lot of people are using th is

Appl ica t ions (apps) you get through

social etwork

has s e are mos:ly useful

Other users share mostly usefu l
(i.e informa ive) Posts

•
0

Bu si nesses (Such as news or

This social network rs simple

entertammen companies) share

to use

mostly usef ul (i.e. informative) posts

of your friE>nd~ arP usmg thts
Applica t ions (apos) you g t through
thas srte are rnos ly useful

sotlill nPtwor

i:tl.