Modelling the impact of mHealth service quality on satisfaction, continuance and quality of life

This article was downloaded by: [University of Birmingham]On: 15 November 2014, At: 10:31Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Behaviour & Information TechnologyPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/tbit20

Modelling the impact of mHealth service quality onsatisfaction, continuance and quality of lifeShahriar Aktera, John D'Ambrab, Pradeep Rayb & Umme Hanica School of Management and Marketing, University of Wollongong, NSW, 2522, Australiab Australian School of Business, University of New South Wales, NSW 2052, Australiac Sydney Graduate School of Management, University of Western Sydney, NSW 2751,AustraliaAccepted author version posted online: 01 Nov 2012.Published online: 14 Jan 2013.

To cite this article: Shahriar Akter, John D'Ambra, Pradeep Ray & Umme Hani (2013) Modelling the impact of mHealthservice quality on satisfaction, continuance and quality of life, Behaviour & Information Technology, 32:12, 1225-1241, DOI:10.1080/0144929X.2012.745606

To link to this article: http://dx.doi.org/10.1080/0144929X.2012.745606

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/tbit20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/0144929X.2012.745606

http://dx.doi.org/10.1080/0144929X.2012.745606

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Modelling the impact of mHealth service quality on satisfaction, continuance and quality of life

Shahriar Aktera*, John D’Ambrab, Pradeep Rayb and Umme Hanic

aSchool of Management and Marketing, University of Wollongong, NSW 2522, Australia; bAustralian School of Business,University of New South Wales, NSW 2052, Australia; cSydney Graduate School of Management, University of Western Sydney,

NSW 2751, Australia

(Received 7 June 2011; final version received 17 October 2012)

Understanding the impact of service quality on economic and social outcomes is critical to extend the focus of ITservice research. This study evaluates the impact of quality on both these dimensions in mHealth using a crossdisciplinary approach. The conceptual model is rooted in the traditional cognition–affective–conation chain butexplicitly incorporates convenience, confidence, cooperation, care and concern as the primary dimensions ofmHealth quality. The model is validated in the context of a business-to-consumer mHealth services using partialleast squares path modelling. The results confirm that service quality has both direct and indirect impact oncontinuance intentions (i.e. economic outcome) and quality of life (i.e. social outcome). In this relationship,satisfaction plays the key mediating role, whereas service quality does not have any moderating effect. Researchimplications point to scale and sustain this new healthcare paradigm by linking service quality to satisfaction,continuance intentions and quality of life.

Keywords: mHealth; service quality; satisfaction; continuance; quality of life

Introduction

The global economy is becoming characterised byservices with more than 70% contribution in GDPfrom the services sector (Ostrom et al. 2010). Healthcare is one of the fastest growing sectors in the serviceseconomy. The growth of this sector is expected to besustained through a critical evaluation of its impact onthe success of firms, the well-being of societies and thequality of consumers’ lives worldwide (Bitner andBrown 2008). Though health care is arguably the mostimportant service with a pervasive impact on daily life,it is a deeply troubled sector (Berry and Bendapudi2008). The healthcare system in the world is on adepressing path, with a deadly combination of limitedaccess to care, uneven quality and high costs (Porterand Teisberg 2006). In this context, the introduction ofinformation and communication technologies (ICT),especially the application of mHealth, has created thepotential to transform healthcare delivery by making itmore accessible, affordable and available. Accordingto the World Health Organization Report (2008), ‘Theworld has better technology and better information toallow it to maximize the return on transforming thefunctioning of health systems’.

‘mHealth’, a new paradigm of the emerging ITartefact, is the application of mobile communications –such as mobile phones and PDAs – to deliver right

time health services to customers (or patients). In thehealthcare sector, mHealth is a transformative ITservice for shifting the care paradigm from crisisintervention to promoting wellness, prevention andself-management (Kaplan and Litewka 2008). As atransformative service, mHealth centres on ‘creatinguplifting changes and improvements in the well-beingof both individuals and communities’ (Ostrom et al.2010). Although this service creates positive changes,growing concerns revolve around perceived quality ofsuch services due to lack of reliability of the serviceplatform, knowledge and competence of the provider,privacy and security of information services and, aboveall, their overall effects on patient satisfaction (SA),continuance intentions (CI) and quality of life (QOL)(Akter and Ray 2010, Dagger and Sweeney 2006,Dagger et al. 2007, Ivatury et al. 2009, Mechael 2009,Varshney 2005). As such, the impact of perceivedquality on critical service outcomes becomes a criticaldimension to determine the success or failure ofmHealth platform.

However, research is scant in IT services that havemodels to analyse these relationships (Ostrom et al.2010). A review of the literature reveals that most ofthe research in this domain (i.e. mHealth) remainslargely anecdotal, fragmented and atheoretical (Akteret al. 2011, Chatterjee et al. 2009). Thus, this studyaims to model the impact of perceived service quality

*Corresponding author. Email: [email protected]

� 2013 Taylor & Francis

Behaviour & Information Technology, 2013Vol. 32, No. 12, 1225–1241, http://dx.doi.org/10.1080/0144929X.2012.745606

Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

(SQ) on SA, CI and QOL in the context of a business-to-consumer (B2C) mHealth services. This modellingextends the scope of technology–business alignment intransformative IT services research by developing anmHealth quality model and framing its overall impacton individual, economic and social outcomes.

The paper is organised as follows: Conceptualframework and research hypotheses section providesan overview of the model, definitions of the keyconstructs and a detailed development of the hypoth-eses. Methodology section describes the researchdesign, including sampling and the instrument tomeasure the conceptual model. Findings sectionpresents the results of the hypotheses testing, andfinally, Discussion section highlights research implica-tions, limitations and future research directions.

Conceptual framework and research hypotheses

The proposed research model explains the dynamics ofmHealth service quality by encapsulating its compo-nents and consequences (see Figure 1). The modelextends knowledge by conceptualising mHealth SQand modelling its overall impact on SA, CI and QOL.Although SQ is a critical parameter in any healthservice, there is a paucity of research, which hasadequately focussed on mHealth SQ dynamics (Akterand Ray 2010, Earth Institute 2010, Ivatury et al. 2009,Kahn et al. 2010, WHO 2011). Thus, this study fills thisvoid by developing a multidimensional mHealth SQmodel and framing its effects on critical serviceoutcomes. The proposed model substantially differsfrom the extant theoretical frameworks (e.g. Andaleeb2001, Brady and Cronin 2001, Dagger et al. 2007,Fassnacht and Koese 2006, Jia et al. 2008, Parasura-man et al. 1988, 2005) by articulating service quality in

mHealth and evaluating its overall effects on individualbenefit (i.e. SA), economic return (i.e. CI) and societalwelfare (i.e. QOL).

The conceptual model (Figure 1) is based on theliterature in marketing, information systems andhealth care as the study focusses on a technology-mediated health service platform. In service research,such an interdisciplinary approach is important andnecessary to adequately address the challenges andopportunities (Ostrom et al. 2010). The conceptualmodel elucidates an overview of associations in termsof cognitive–affective–conative framework (Bhatta-cherjee 2001, Cronin and Taylor 1992, Chiou et al.2006, Dagger et al. 2007, Oliver 1997, Taylor andBaker 1994, Woodside et al. 1989).

The model links consumer beliefs, affect andintention within the traditional consumer attitudestructure. This relationship simplifies quality dominantdecision-making process for a transformative ITservice platform (e.g. B2C mHealth care) with aneffect on economic (i.e. CI) and social (i.e. QOL)outcomes. The model conceptualises SQ as a higherorder construct, which has an influence on SA, CI andQOL. In this relationship, SA plays the key mediatingrole between SQ–CI and SQ–QOL. This study alsoexplores the moderating effects of SQ in the SA–CIlink as well as SA–QOL link. In the following sections,the study defines each construct and presents justifica-tion for all the hypotheses with further elaborationregarding the proposed relationships.

Service quality

Service quality is an important and particularly relevantconstruct in virtually all service business (Voss et al.2004). It is a powerful concept because of its strong

Figure 1. Research model.

1226 S. Akter et al.

Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

relationship with customer satisfaction (Andaleeb 2008,Cronin and Taylor 1992, Dabholkar et al. 2001, Oliver1993, Taylor and Baker 1994), purchase intention(Dagger et al. 2007, Dagger and Sweeney 2006) andfirm’s performance (Fassnacht and Koese 2006, Sousaand Voss 2006). Research in this arena still remains‘unresolved’ (Caruana et al. 2000, Jia et al. 2008) due toits ‘elusive’ nature (Ma et al. 2005, Parasuraman et al.1985, 2005). Indeed, this concept remains difficult toconceptualise (Brady and Cronin 2001, Chiou et al. 2006,Dagger et al. 2007) and ‘far from conclusive’ (Athanas-sopoulos 2000, p. 191). This study defines service qualityas a consumer’s judgment of, or impression about, anentity’s overall excellence or superiority (Dagger et al.2007). In health care, customers or patients play a criticalrole in defining quality and designing the service deliverysystems (Donabedian 1992, Jun et al. 1998). According toO’Connor et al. (1994), ‘It’s the patient’s perspective thatincreasingly is being viewed as a meaningful indicator ofhealth services quality and may, in fact, represent themost important perspective’ (cf. Andaleeb 2001, p. 32).

The extant literature shows that the dimensions ofservice quality vary according to the context, such astwo (e.g. Gronroos 1984), three (e.g. Brady and Cronin2001, Rust and Oliver 1994), five (e.g. Parasuramanet al. 1988) and even 10 (e.g. Parasuraman et al. 1985).As such, there is no standard agreement as to thenumber of dimensions of this concept (Brady andCronin 2001). However, most of the studies generallyconfirm that service quality should be multidimen-sional (Gronroos 1984, Parasuraman et al. 1988),hierarchical (Brady and Cronin 2001, Rust and Oliver1994) and context specific (Dagger et al. 2007).

Satisfaction

Satisfaction becomes an important cornerstone forservice-oriented business practices around the world(Szymanski and Henard 2001). In healthcare, patientsatisfaction is a major indicator in measuring theeffects of quality or overall service performance(Dagger et al. 2007, Saila et al. 2008). Satisfactionalso leads to favourable results, such as higher rates ofpatient retention and higher profits (Zeithaml 2000).As such, customer (or patient) satisfaction must be anintegral part of healthcare organisations’ strategicprocesses (Andaleeb 2001). Donabedian (1992) sug-gests that satisfaction should receive equal importanceas SQ in order to design and manage the healthcaresystems effectively.

Satisfaction is an ‘affective response’ (Giese andCote 2000) though scholars report this construct fromdifferent viewpoints, such as a fulfillment response(Oliver 1997), an overall evaluation (Fornell 1992),psychological state (Howard and Sheth 1969), global

evaluative judgment (Westbrook 1981) and summaryattribute phenomenon (Oliver 1993). Whereas servicequality is a cognitive construct, satisfaction is anattitudinal construct (e.g. Brady and Robertson 2001,Cronin and Taylor 1992, Gotlieb et al. 1994). Thus, theextant literature identifies satisfaction as an affectiveresponse to the cognitive service quality approach(Oliver 1997, Taylor and Baker 1994). This distinctionsuggests a casual model that identifies service qualityas an antecedent to satisfaction (Choi et al. 2004). Inhealthcare settings, numerous studies support thiscausal linkage between service quality and satisfaction(Andaleeb 2001, Dagger et al. 2007, Woodside et al.1989). Thus, given the important link between servicequality and satisfaction, this study models satisfactionas a function of perceived service quality in the contextof mHealth:

H1: Service quality has an impact on satisfaction inmHealth services.

Continuance intentions

The success of a technology-mediated service platform,such as mHealth, depends a lot on the ongoing usagerather than initial acceptance (Bhattacherjee 2001,Limayem et al. 2007). As such, an increasing body ofresearch in this domain depends on continuance theory(Akter et al. 2011). This study defines continuance as ausage stage when technology-based service use (e.g.mHealth) transcends conscious behaviour and be-comes part of normal routine activity. Continuancedecision is similar to consumers’ repurchase decision,which is primarily based on satisfaction of a particularproduct or service (Anderson and Sullivan 1993, Oliver1980, 1993). Bhattacherjee (2001, pp. 351–352) high-lights the importance of continuance in IT services bystating that ‘long-term viability of an IS and itseventual success depend on its continued use ratherthan [its] first-time use’. Thus, continuance behaviouris a highly relevant construct from a practical pers-pective because service usage obviously continues wellbeyond the initial adoption (Montoya et al. 2010).

The continuance theory posits that consumers’satisfaction with a service is the primary motivation forcontinuance intentions (e.g. Bhattacherjee 2001). How-ever, research is scant in exploring the impact of bothservice quality and satisfaction on continuance inten-tions. The work by Dagger et al. (2007) implies thatboth service quality and satisfaction influence one’sinclination to continue using health services. Mosthealthcare platforms recognise these relationshipsbecause the level of continuance intentions indicatestheir overall financial performance or economicviability (Bernhardt et al. 2000, Eskildsen and Kris-tensen 2003). Hence, this interest creates a need for

Behaviour & Information Technology 1227

Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

developing a deeper understanding of the relationshipbetween these constructs.

H2: Customer satisfaction has an impact on continu-ance intentions.

H3: Service quality has an impact on continuanceintentions.

Quality of life

Quality of life refers to the well-being and happiness ofan individual (Ferrans and Powers 1992). QOLperceptions, therefore, determine an individuals’ eva-luation of the life and the positive or negativeattributes that characterise the life condition, includinghealth status. Thus, this study views QOL as asubjective, individual and experiential concept. Giventhe healthcare context of the present study and thesignificance of health care as a vital component inQOL, this study defines QOL as a sense of overall well-being in health (Dagger and Sweeney 2006). Strauband Watson (2001) indicate that any technology-basedservice platform should focus on increasing the qualityof its users’ lives. Researchers in both marketing (e.g.Dagger and Sweeney 2006) and information systems(e.g. Choi et al. 2007) report QOL as a critical outcomevariable. Also, studies in the healthcare industrysuggest exploring the impact of any new technology-based service on QOL (Sirgy 2001). In this study, theQOL concept indicates that people have a variety ofhealthcare needs, and the more they satisfy these needsusing mHealth services, the more they feel good abouttheir quality of lives (Heisel and Flett 2005). Here, thestudy designates QOL as an alternative outcomevariable and intends to explore how overall SQ andSA contribute to quality of (health) life of anindividual. However, no study yet frames the directimpact of overall SQ on QOL and indirect impactthrough SA in mobile health care. In addition, researchis scant in measuring the critical impact of QOL on CI.Thus, the study hypothesises that:

H4: Satisfaction has an impact on quality of life.

H5: Service quality has an impact on quality of life.

H6: Quality of life has an impact on continuanceintentions.

Mediating effects of satisfaction

Satisfaction is a major driver of positive QOLperception and continuance intentions, and, therefore,achieving high consumer satisfaction is a key goal ofservice dominant businesses (Oliver 1997, Bhattacher-jee 2001, Chiou et al. 2006). This study definessatisfaction as a mediator because, first, service quality

(predictor) influences satisfaction (mediator); second,satisfaction influences continuance intentions andquality of life (criterion variables) and, finally, servicequality influences the criterion variables in the absenceof the mediator’s influence (Barron and Kenney 1986).In addition, satisfaction as a mediator or an ‘affective’attitude between ‘cognitive beliefs’ (e.g. SQ) and‘conative’ constructs (e.g. CI and QOL) draws muchattention in psychology (Ajzen and Fishbein 1980),marketing (Bansal et al. 2005, Dagger and Sweeney2006) and information systems literature (Bhattacher-jee 2001). Thus, the mediating role of satisfaction inthe high-involvement mHealth services is important toexplore:

H7: Satisfaction mediates the relationship betweenservice quality and continuance intentions.

H8: Satisfaction mediates the relationship betweenservice quality and quality of life.

Moderating effects of service quality

This study defines a moderator as ‘a variable thataffects the direction and/or strength of the relationbetween an independent or predictor variable and adependent or criterion variable’ (Barron and Kenney1986, p. 1174). In fact, moderation occurs whenpredictor (SA) and moderator (SQ) have a joint effectin accounting for incremental variance in criterionvariables (CI and QOL) beyond that explained by themain effects (Cohen and Cohen 1983). This studyassumes service quality as a moderator, which mayhave an influence on the links between SA–CI and SA–QOL. As such, the variation in service quality mightinfluence the strength or the direction of these links(Barron and Kenney 1986). Surprisingly, research onthe moderating role of service quality on both theseassociations is non-existent. This study finds thisomission intriguing in order to explore complexinterdependencies among latent variables (Chin et al.2003, Homburg and Giering 2001). Besides, ananalysis of moderating effects is of high relevance ascomplex relationships are typically subject to contin-gencies in the causal network of consumer attitudes.Thus, the study hypothesises that:

H9: Service quality moderates the relationship betweensatisfaction and continuance intentions.

H10: Service quality moderates the relationshipbetween satisfaction and quality of life.

Methodology

Research context

This study focusses on mobile telemedicine services inBangladesh, which is one of the leading mHealthservice providing developing nations (Akter and Ray


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

2010). In recent years, this particular mHealth plat-form has become very popular in the developing world(e.g. India, Bangladesh, Pakistan, Mexico, SouthAfrica, Peru, etc.) and serves millions by deliveringright time medical services at an affordable cost(Ivatury et al. 2009, Akter et al. 2011). Currently,more than 24 million people in Bangladesh have accessto such B2C mHealth services provided by the leadingmobile operator Grameen phone. Under this platform,a customer (or a patient) can access health service atanytime by dialing ‘789’ from his/her own mobilephones and receive services in the form of medicalinformation, consultation, treatment, diagnosis, refer-ral, treatment and counselling from registered physi-cians. In addition, customers who do not have theirown mobile phones can access this mHealth servicefrom local mobile phone kiosks, which are widelyavailable at every corner of the country.

Qualitative research

This study obtained qualitative data from three focusgroup discussions conducted with mHealth consumersin Bangladesh. A total of 24 participants, eight perfocus group, were involved in the focus group sessions.Participants ranged in age from 18 to 62 years, andboth genders had equal participation. Each session wasconducted by two moderators, which lasted about 90min. In addition, 10 in-depth interviews were con-ducted to explore users’ insights on the researchagenda. In both cases, participants were recruitedusing convenient sampling in order to ensure produc-tive findings and the richest data for scale development(Dagger et al. 2007). In each case, respondents wereasked to evaluate their mHealth experiences. The studyasked the following questions to identify the servicequality dimensions:

In your opinion, what makes mHealth different fromother health services?

What are the major merits and demerits of mHealthservices?

Any positive or negative experience that you had whilereceiving mHealth services?

The answers were recorded, synthesised andsorted into different categories to identify the coredimensions and their link to outcome constructs. Inthe qualitative study, service quality was frequentlyidentified as a multidimensional and context-specificconcept. Users expressed their opinion on differentservice-level attributes (e.g. ‘I can access mHealthplatform whenever I want’, ‘The physician showssincere interest to solve my problems’, or, ‘‘I feel safewhile consulting with Physicians’, or ‘It is worthwhilehaving service from this platform’) under multiple

dimensions. Throughout this process, the studyfound support for five primary service qualitydimensions in mHealth, that is, convenience, con-fidence, cooperation, care and concern on privacy.

Instrument development

The questionnaire consists of previously publishedmultiitem scales with favourable psychometric proper-ties and items from qualitative research (see Table 1).All the constructs in the model, except satisfaction,were measured using 7-point Likert scale (e.g. stronglydisagree–strongly agree). Satisfaction was measuredusing bi-polar semantic differential scale (e.g. verydissatisfied–very satisfied). The study developed theprimary version of the questionnaire in English andthen translated the measures into the local language(Bangla). The local version was retranslated andconfirmed by a panel of judges that both versionsreflect the same content. Before the final study, thestudy conducted a pretest over 15 convenient samplesto ensure that the question content, wording, sequence,format and layout, question difficulty, instructions andthe range of the scales were appropriate. In response tothe pretest, context-specific adjustments were made torefine the final version of the questionnaire.

Sampling

Data were collected from Bangladesh under a globalmHealth assessment project from 7 January to 17March 2010. In the absence of lists for drawing arandom sample, about 600 interviews were plannedfrom two urban areas and three rural areas using areawise cluster sampling. Areas were selected in a mannersuch that different socioeconomic groups were repre-sented. From each area, firstly, thanas were selectedrandomly; then, streets/villages were selected fromeach thana; and finally, residential homes were selectedfrom each street/village. In order to obtain a prob-ability sample, systematic random sampling wasapplied so that each sample unit/element had an equalchance of being selected. The population was definedas the customers who had experience of using mobiletelemedicine services in the past 12 months. In allsurvey interactions, interviewers were given a letter ofintroduction from a reputed university containing thephone number for respondents to see that the studywas authentic. Those who agreed to be interviewedwere explained the academic purpose of the study withadequate assurance of anonymity and freedom of notanswering particular questions or withdrawing fromthe interview at any stage. A total of 623 respondentswere approached, of which 480 (77%) surveys wereultimately completed. Of the total number of


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

completed surveys, seven were considered problematicand excluded, because of excessive missing data, do notknow answers or NA answers and response biases.Finally, 473 surveys were analysed.

The demographic profile of the respondents repre-sents a diverse cross section of the population (seeTable 2). The respondent group ranged in age from 18

to 62 years, were 59% male, 58% lived in rural areas,47% had income less than US $70 per month,employed in a wide range of professions (students,professionals, self-employed, academics, farmers,housewives, day laborers and retirees) and hadvarious educational levels (from illiterate to doctoraldegrees).

Table 1. Operationalisation of constructs.

Constructs Operational definitions Measures

Convenience (Akteret al. 2010)

The degree to which the mHealthservice platform is available‘anytime’ and ‘anywhere’ basis.

CV1. I can access to mHealth whenever I need.CV2. I can access to mHealth wherever I need.CV3. mHealth platform does not have long waitingtime.

CV4. mHealth platform is always available.Confidence (Dagger

et al. 2007,Parasuraman et al.1988)

The degree to which mHealth serviceprovider has the ability to serve thepatients.

CF1. Physicians at mHealth platform are competent inproviding services.

CF2. I feel safe while consulting with Physicians atmHealth platform.

CF3. The behaviour of physicians at mHealth platforminstills confidence in me.

CF4. Physicians have the knowledge to answer myquestions.

Cooperation (Daggeret al. 2007,Parasuraman et al.1988)

The degree to which mHealth serviceprovider is willing to help patientsand provide prompt service.

CO1. Physicians at mHealth platform provide meprompt service.

CO2: Physicians provide the service by a certain time.CO3. Physicians are never too busy to respond to myrequests.

CO4. Physicians are willing to help me.Care (Dagger et al.

2007, Parasuramanet al. 1988)

The degree to which mHealth serviceprovider shows caring andindividualised attention topatients.

CA1. Physicians give me personal attention.CA2. Physicians give me individual care.CA3. Physicians understand my specific needs.CA4. Physicians have my best interests at heart.

Concern on privacy(Parasuramanet al. 2005)

The degree to which mHealth serviceplatform reduces concerns bymaintaining patients’ privacy.

CN1. mHealth platform protects information about mypersonal problems.

CN2. mHealth platform does not share my personalhealth information with others.

CN3. mHealth platform protects information about mypersonal identity.

CN4. mHealth platform offers me a meaningfulguarantee that it will not share my information.

Continuanceintention(Bhattacherjee2001)

Users’ intention to continue usingmHealth services.

CI1. I intend to continue using mHealth.CI2. My intention is to continue using this service ratherthan use any alternative means (e.g. going to localclinics).

CI3. I will not discontinue my use of this service.Satisfaction (Spreng

et al. 1996)Users’ affect with (or, feelings) about

prior mHealth services use.How do you feel about your overall experience of

mHealth service use:SA1. Very dissatisfied/very satisfiedSA2. Very frustrated/very contended.SA3. Very displeased/very pleasedSA4. Absolutely terrible/absolutely delighted.

Quality of life(Dagger andSweeney 2006,Choi et al. 2007)

QOL refers to a sense of overall wellbeing in health.

QOL1. Getting services from this platform have enabledme to improve my overall health.

QOL2. In most ways, my life has come closer to myideal since I started using this service.

QOL3. The conditions of my health life have improvedbecause of this service.

QOL4. I have been more satisfied with my health life,thanks to this service.

QOL5. So far, this service has helped me to achieve thelevel of health I most want in life.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Data analysis

Service quality in this study serves as a second-orderreflective construct, which contains convenience, con-fidence, cooperation, care and concern as the first-order components. According to Bagozzi (2010, p.209), ‘The second-order factor approach is most validand conceptually meaningful when the first-orderfactors loading on the second-order factor can beinterpreted as subdimensions or components of a moreabstract, singular construct’. In the similar spirit, thisstudy estimates the second-order mHealth servicequality construct model applying the component-basedstructural equation modelling (SEM) or partial leastsquares (PLS) path modelling (PLS-PM) technique.The PLS approach to SEM, also known as PLS-PM, isregarded as a component-based SEM (Chin et al. 2003,Tenenhaus 2005) to model multiple causes and multi-ple indicators of a single latent variable and to latentpath models (Wold 1975, 1982, 1985). In contrast tocovariance-based SEM (maximum likelihood ap-proach), PLS-PM is a favourable technique forestimating hierarchical models with moderating andmediating effects because it can ensure more theore-tical parsimony and less model complexity (Chin 2010,Edwards 2001, Law et al. 1998, MacKenzie et al. 2005,Wetzels et al. 2009).

Justification of the analytical approach

This study applies component-based SEM (or PLS)because, first, this approach is consistent with theobjective of the study, which aims to develop and test atheoretical model through explaining and prediction(Chin 2010, Hair et al. 2011). Second, this approachestimates a hierarchical model with more theoreticalparsimony and less model complexity (Bagozzi and Yi

1994, Edwards 2001, Wetzels et al. 2009). Third, thisapproach can effectively handle various constraintswith regard to the distributional properties (multi-variate normality), measurement level, sample size,model complexity, identification and factor indetermi-nacy (Chin 1998b, 2010, Fornell and Bookstein 1982,Hair et al. 2011, Hulland et al. 2010). Fourth, thisapproach works better when the model is relativelycomplex (e.g. hierarchical model) and the phenomenonunder study is new or changing (Chin and Newsted1999). Finally, this approach is suitable for the studybecause PLS-PM provides more accurate estimates ofmediating and moderating effects by accounting for themeasurement error that attenuates the estimatedrelationships and improves the validation of theories(Chin et al. 2003).

Operationalisation of the approach

This study applies PLS-PM using PLS Graph 3.0(Chin 2001) to estimate the reflective, second-order SQmodel through the repeated use of manifest variables(Wold 1985). As per the guidelines of hierarchicalmodelling (Chin 2010, Wetzels et al. 2009), manifestvariables were used repeatedly to estimate the scores offirst-order constructs (i.e. convenience, confidence,cooperation, care and concern) as well as for thesecond-order SQ construct (see Appendix 1 for de-tails). According to Wetzels et al. (2009), ‘Thisapproach also allows us to derive the (indirect) effectsof lower-order constructs, or dimensions, on outcomesof the higher-order construct’. Using this approach,this study created the second-order SQ construct thatrepresents all the manifest variables of the underlyingfirst-order latent variables (convenience, confidence,cooperation, care and concern). Table 3 outlines the

Table 2. Demographic profile of respondents.

Items Categories %

Gender Male 59Female 41

Location Urban 42Rural 58

Income (per month in US $) 5$70 47$71–$141 22$142–$212 10$212þ 21

Age 18–25 2526–33 3234–41 2142–49 1750þ 5

Occupation Working full time 38Working part time 34Housewife 16Others 5

Table 3. Estimation of mHealth service quality as a second-order, reflective model.

First-order model(dimensions of Healthservice quality)

Second-order model(mHealth service quality

construct)

yi ¼ Ly. Zj þ eiyi ¼ manifest variables(e.g. items/ indicators)

Ly ¼ loadings of firstorder latent variable

Zj ¼ first order latentvariable(e.g. convenience,confidence,cooperation,care and concern)

ei ¼ measurement errorof manifest variables

Zj ¼ �. xk þ zjZj ¼ first order factors(e.g. convenience)

� ¼ loadings of secondorder latent variable

xk ¼ second order latentvariable (e.g. mHealthservice quality)

zj ¼ measurementerror of first orderfactors


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

equation for estimating the second-order, reflective,SQ construct model in mHealth. For instance, theequation for the first-order model specifies first-orderlatent variable (Zj), its indicators (yi), loadings (Li) andan error term (ei). The equation of the second-ordermodel specifies the first-order factors ((Zj ) in terms ofthe second-order latent variables (xk) and error (zj) forthe first-order factor and second-order latent variablesloadings (�). Therefore, applying this equation, thestudy estimates the research model and presentsempirical findings in the next section.

Findings

Assessment of the first order model

The study assesses the psychometric properties of thefirst-order measurement model by examining reliabil-ity, convergent validity and discriminant validity. Thefirst-order model consists of eight constructs in theresearch model, that is, convenience, confidence,cooperation, care, concern, SA, CI and QOL. Initially,the study calculated all the item loadings, which exceedthe cut-off values of 0.7 and significant at p 5 0.001.The higher average of item loadings (40.7) and anarrower range of difference provide strong evidencethat respective items have greater convergence inmeasuring the underlying construct (Chin 2010). Assuch, the study removed CV3, CF4, CO2, CA1, CN4,QOL2 and QOL5 as their item loadings were lowerthan 0.7 (see Table 1 for the complete list of items).

A complete picture of the first order model emergesin Table 4 after applying the testing criteria of itemloadings and eliminating the items that damage thesoundness of the criteria. The study also calculatedaverage variance extracted (AVE) and composite reli-ability (CR) (Chin 1998a, Fornell and Larcker 1981) toconfirm reliability of all the measurement scales. AVEmeasures the amount of variance that a constructcaptures from its indicators relative to measurementerror, whereas CR is a measure of internal consistency(Chin 2010). Basically, these two tests indicate theextent of association between a construct and itsindicators. The study shows that the CR and AVE ofall scales are either equal to or exceed 0.80 and 0.50 cut-off values, respectively (Fornell and Larcker 1981).Thus, the study confirmed that for all the item loadings,CRs and AVEs exceed their respective cut-off valuesand ensure adequate reliability and convergent validity(Chin 1998a, Fornell and Larcker 1981).

In addition, in Table 5, this study calculates thesquare root of the AVE that exceeds the intercorrela-tions of the construct with the other constructs in themodel to ensure discriminant validity (Fornell andLarcker 1981). This test indicates that the constructs

Table 4. Psychometric properties of the first-order con-structs.

Constructs Items Loadings

Compositereliability(CR)

Averagevarianceextracted(AVE)

Convenience CV1CV2CV4

0.820.810.83

0.86 0.67

Confidence CF1CF2CF3

0.900.860.81

0.89 0.73

Cooperation CO1CO3CO4

0.950.940.93

0.96 0.89

Care CA2CA3CA4

0.930.920.86

0.93 0.82

Concernon Privacy

CN1CN2CN3

0.900.940.95

0.95 0.87

Satisfaction SA1SA2SA3SA4

0.950.940.940.93

0.94 0.89

Continuanceintentions

CI1CI2CI3

0.930.910.95

0.95 0.86

QOL (qualityof life)

QOL1QOL3QOL4

0.910.930.90

0.95 0.83

Table 5. Mean, standard deviation (SD) and correlations of the latent variables.

Construct Mean SD CV CF CO CA CN SA CI QOL

Convenience (CV) 5.78 0.98 0.82*Confidence (CF) 5.75 1.12 0.69 0.86*Cooperation (CO) 5.64 1.18 0.73 0.65 0.94*Care (CA) 5.74 1.17 0.64 0.72 0.57 0.90*Concern (CN) 5.75 1.18 0.47 0.44 0.44 0.42 0.93*Satisfaction (SA) 5.68 1.11 0.68 0.66 0.64 0.58 0.49 0.95*Continuance (CI) 5.59 1.26 0.65 0.58 0.63 0.53 0.43 0.7 0.92*Quality of life (QOL) 5.53 1.14 0.68 0.67 0.62 0.6 0.51 0.72 0.7 0.91*

*Square root of the AVE on the diagonal.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

do not share the same type of items and they areconceptually distinct from each other (Chin 2010). Inother words, each construct and its measures in theresearch model do a great job in discriminatingthemselves from other constructs and their corre-sponding measures. Thus, the study ensures a validmeasurement model with the evidence of adequatereliability, convergent validity and discriminant valid-ity. This process also paves the way for testing all thehypotheses and proving the research model.

Assessment of the second-order model

This study also estimates the second-order, reflectiveSQ construct, which consists of five first-ordercomponents (convenience, confidence, cooperation,care and concern), representing 15 (3 6 5) items.The CR and the AVE of the second order SQ constructare 0.943 and 0.524, respectively, providing evidence ofreliable, higher-order measures. The results indicatethat SQ gets its meaning indirectly through themeasures of five first-order components (see Figure2). The results also confirm that the hierarchical SQconstruct has a significant association (p 5 0.001)with all the primary components, that is, convenience(b ¼ 0.869), confidence (b ¼ 0.861), cooperation(b ¼ 0.849), care (b ¼ 0.825) and concern(b ¼ 0.661). The study analyses the implications ofthese results in the ‘Discussion’ section.

Structural model

In order to assess the research model, this studyestimates the impact of overall mHealth service qualityon satisfaction, continuance and QOL (Figure 3A).Initially, the study estimates the SQ–SA–CI link andthe results give a standardised beta of 0.753 from SQ toSA, 0.315 from SA to CI and 0.231 from SQ to CI.Based on these findings, this study confirms thatoverall SQ has both direct and indirect impact on CI,which proves H1, H2 and H3.

Furthermore, the study estimates the SQ–SA–QOLlink and the results give a standardised beta of 0.364from SA to QOL and 0.480 from SQ to QOL. Theseresults again confirm the direct and indirect impact ofSQ on QOL, thereby proving H4 and H5, respectively.This study also assesses the impact of QOL on CI(path ¼ 0.303), which confirms H6. Overall, thevariance explained by the mHealth SQ model in termsof R2 is 0.567 for customer SA, 0.598 for CI and 0.626for QOL, which are significantly large (f2 4 0.35)according to the effect sizes defined for R2 by Cohen(1988).

Mediation analysis

In Figure 3A, this study analyses the mediating effectof SA on both SQ–CI link and SQ–QOL link. Beforeanalysis, the study adequately confirms the criteria formediation analysis (Barron and Kenney 1986) as

Figure 2. First-order dimensions.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Figure 3. (A) Main effects model. (B) Interaction model (with moderation analysis).


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

follows: first, the predictor (SQ) influences the med-iator (SA). Second, the mediator (SA) influences thecriterion variables (CI and QOL). Third, the predictor(SQ) influences the criterion variables (CI and QOL) inthe absence of the mediator’s influence. In Figure 3A,‘a’ refers to the SQ–SA path, ‘b’ refers to the SA–CIpath, ‘c’ refers to the SQ–CI path, ‘d’ refers to the SA–QOL path and ‘e’ refers to the SQ–QOL path. Thus, toestablish the mediating effect of SA, the indirect effectof a 6 b has to be significant for SQ–CI link anda 6 d has to be significant for SQ–QOL link(Iacobucci 2008). Here, If the z-value exceeds 1.96(p 5 0.05), the study can accept H7 and H8, becausethe results indicate that overall SQ has an indirectimpact on both CI and QOL through SA (Sobel 1982).The study estimates the z-value as follows (see Figure3A):

ZSQ�CI link ¼a� bffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

b2 � s2aþa2 � s2bþs2a � s2b

q

ZSQ�QOL link ¼a� dffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

d2 � s2aþa2 � s2dþs2a � s2d

q

The z-value for SQ–CI link is 5.70 and for SQ–QOLlink is 6.77, which supports the mediating effects of SA.These findings support H7 and H8, which imply thatSQ has an indirect impact on CI and QOL. To estimatesize of the indirect effect, this study uses the varianceaccounted for (VAF) value, which represents the ratioof the indirect effect to the total effect. The resultsindicate that indirect effects (or customer SA) explainabout 51% of the total effect of SQ on CI and about36% of the total effect of SQ on QOL:

VAFSQ�CI link¼a� b

a� bþ c¼ 0:753� 0:315

0:753� 0:315þ0:231¼ 0:506

VAFSQ�QOL link¼a�d

a�dþe¼ 0:753�0:364

0:753�0:364þ0:480¼0:363

Moderation analysis

In Figure 3B, this study presents the moderationanalysis applying PLS product-indicator approach(Chin et al. 2003) to detect the moderating effect ofSQ on the relationship between SA–CI and SA–QOL.To test the moderating effects, first, this study multi-plies SA (predictor) and SQ (moderator) to create aninteraction construct that predicts both CI and QOL,respectively. In this study, SA is a simple latentconstruct representing four items, and SQ is a

second-order construct representing 15 items; thus,the interaction construct represents 60 (4 6 15) items.Second, this study estimates the influence of thepredictor (SA) on the criterion variables (CI andQOL), the direct effect of moderator (SQ) on thecriterion variables and the influence of the interactionvariable (SA 6 SQ) on the criterion variables (Figure3B). Now, the study can confirm the significance of themoderator (SQ) if the interaction effects (paths p andq) are meaningful, independent of the size of the otherpath coefficients (Chin et al. 2003). In Figure 3B, thisstudy estimates the standardised path coefficientsof 70.040 and 0.028 to predict the impact of interac-tion construct on both CI (p) and QOL (q). The resultsindicate that these interaction effects are not significantat p 5 0.05. The sizes of the interaction effects are asfollows:

f2CI ¼R2

i � R2m

1� R2i

¼ 0:600� 0:598

1� 0:600¼ 0:005

f2QOL ¼R2

i � R2m

1� R2i

¼ 0:626� 0:626

1� 0:626¼ 0:000

(Here, i ¼ interaction model, m ¼ main effects model.)The results show that the sizes of the interaction

effects are small (f 2 5 0.02) (Cohen 1988) as well asthe resulting beta changes (p ¼ 70.040, q ¼ 0.028) areinsignificant (p 5 0.05) (Chin et al. 2003). As such, thestudy confirms that SQ does not have any moderatingimpact on the relationship between SA–CI and SA–QOL. Thus, the study rejects H9 and H10 (see Figure3B).

Model evaluation: structural model results

In order to assess the validity of PLS-based researchmodel, first, this study estimates the power (1 – b) ofthe model in order to assess its ability to reject a falsenull hypothesis (H0) (Cohen 1988). In other words,statistical power assesses the probability of findingsignificant associations among the latent variableswhen true relationships exist (Baroudi and Orlikowski1988). In this study, the power of the main effectsmodel is 0.99, which compellingly exceeds the 0.80 cut-off value. This high power (40.80) indicates that theresults of hypotheses testing are valid and the relation-ships are significant. Second, this study estimates thepredictive relevance (Q2) of the endogenous constructsby using sample reuse technique based on blindfoldingprocedure (Stone 1974, Geisser 1975). Q2 indicateshow well-observed values are reproduced by the modeland its parameter estimates. Using the omissiondistance of seven under a cross-validated communality


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

approach, this study obtains Q2 of 0.68 for CI (40.50)and 0.79 for QOL (40.50), which are indicative of ahighly predictive model (Chin 2010). Finally, this studyestimates the global fit measure (GoF) to assess theglobal validity of PLS-based research model (Tenen-haus et al. 2005). GoF refers to the geometric mean ofthe average communality and average R2 for allendogenous constructs. The GoF value ensures globalvalidation of PLS models range between small (GoF ¼0.1), medium (GoF ¼ 0.25) and large (GoF ¼ 0.36).This study obtains a GoF value of 0.723 for thecomplete model, which exceeds the cut-off value of0.36 for the large effect size (Cohen 1988). Thus, GoFallows us to conclude that the model has a betterprediction power, which adequately validates the PLSmodel globally (Wetzels et al. 2009).

Discussion

Summary of findings

The main thrust of this study is to model the impact ofmHealth service quality on satisfaction, continuanceintentions and quality of life. As such, the studydevelops a second-order service quality model on fiveprimary dimensions (i.e. convenience, confidence,cooperation, care and concern). The study alsoconfirms the impact of overall service quality onsatisfaction, continuance intentions and quality oflife. Since the development and operationalisation ofa reliable and valid model is a fundamental goal ofscientific endeavour, the findings of the study make animportant contribution to theory and practice.

In particular, the findings suggest that all theprimary SQ dimensions have a significant positiveassociation with overall SQ. Among these dimensions,‘convenience’ emerges as the strongest component,suggesting that patients must have instant access tothis healthcare delivery platform. In fact, this factorhighlights the right time availability of this platform sothat anyone can receive health services at anytime fromanywhere. In the context of a low resource setting, thisubiquity is a central element in the promise of mHealthto transform the healthcare delivery system (Akter andRay 2010). Then, the study identifies ‘confidence’ as akey component of overall SQ, suggesting that thebehaviours of the provider must establish confidenceamong the patients. This finding is consistent withother satisfaction studies, indicating that more assur-ance from physicians can improve the level of overallquality perception (Andaleeb 2001). Then, ‘coopera-tion’ and ‘care’ emerge as significant components ofoverall SQ. Though the magnitudes of their effects aresmaller than the effects of other dimensions, theyshould receive equal importance to improve theperception of overall SQ. Cooperation suggests that

health professionals must provide prompt service andbe available always and care indicates that they mustunderstand the diversified needs of the patients and becaring and helpful. Finally, ‘privacy-related concerns’emerge as a significant component of SQ, suggestingthat adequate protection of patients’ information canlead to greater gains in patient SA and CI.

The study confirms that overall SQ is a significantpredictor of SA (explaining 57% of variance), CI(explaining 60% of variance) and QOL (explaining63% variance). This finding is consistent with theservice-dominant logic (Vargo and Lusch 2004), whichimplies that exchange process in business should focuson economic (i.e. continuance) and social outcomes(i.e. QOL). The findings also confirm that satisfactionis the key mediator between SQ–CI and SQ–QOL.Because satisfaction is a stronger predictor of CI andQOL relative to SQ, dissatisfied patients may dis-continue this service, despite having positive percep-tions of overall SQ. In other words, satisfaction is thenecessary condition for CI and QOL, and for thisreason, the moderating effects of service quality are notsignificant in the relationship between SA–CI and SA–QOL. Overall, these findings suggest that mHealthservice providers should consider ‘satisfaction’ as animportant strategic objective to predict the impact ofoverall SQ on CI and QOL.

Contribution to theory and practice

The mHealth service context emerges as an example ofbusiness and technology alignment in transformativeservices research, which aims to create upliftingchanges of both individuals and communities throughcontinued consumption. Since transformative service(e.g. mHealth) is a new area in IT service research,scholars still strive to frame its impact on criticalservice outcomes. Research is scant in this sector interms of quality model and its impact on economic andsocial outcomes. Thus, this study extends the scope ofIT service research by developing an mHealth servicequality model on five dimensions (convenience, con-fidence, cooperation, care and concern) and framing itsoverall impact on SA, CI (economic outcome) andQOL (social outcome).

The implications of this research are highly relevantto practitioners. For managers of mHealth services, thefindings of the study improve an overall understandingof how customers evaluate mHealth SQ. In particular,the findings suggest that managers should focus onimproving five SQ dimensions (convenience, confi-dence, cooperation, care and privacy) in order to havea positive impact on ultimate service outcomes. Thesefindings also make it clear that increased servicesatisfaction provides a way for managers to ensure


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

positive CI and quality of life. As continuance affectsprofitability and quality of life influences socialoutcome, gaining insights on these behaviours is ofcritical importance to mHealth service providers. Thus,the findings of the study can help managers achievepatronage for firms, better health outcomes forpatients and, above all, an improved quality of lifefor the community.

Limitations and future research directions

This study has several limitations. First, the context ofthe study is single provider, single country based.Future research could examine the sensitivity of thefindings over multiple service providers in a cross-country setting. There is always a difference withregard to the demographic variables. Information onthese differences across cultures might be of consider-able interest and significance to both researchers andpractitioners for critical managerial decision making(Reynolds and Smith 2010). Second, the study is basedon cross sectional design, which contains typicallimitations associated with this kind of researchmethodology. Future studies could undertake long-itudinal study to unfold the impact of service qualityon outcome constructs over time. Future studies couldalso explore the impact of contextual factors, such asdemographic variables (income, education, gender,etc.) and situational constructs (usage frequency,cost, etc.) on the research model.

Conclusions

The study seeks to model the impact of service qualityon critical service outcomes in a transformative ITservice research, directly applicable to B2C mHealthservice. The findings of the study support the researchmodel, thus lending confidence to the critical role ofquality as a key decision-making variable to predictsatisfaction, continuance intentions and quality of lifein an emerging healthcare paradigm. The findingssuggest that quality and associated economic andsocial outcomes should be mandatory in studies of ITservice research. Aligned with the findings, Ostromet al. (2010) state, ‘Service is not only about increasingrevenues and profits at for-profit firms but also abouthow to advance service in a way that delivers higher-order, societal outcomes’.

Acknowledgements

The authors appreciate and gratefully acknowledge thefinancial support provided for the field study by theAsia Pacific Ubiquitous Healthcare Research Centre(APuHC), Australia. Comments by Wynne W. Chin of

University of Houston and James Nelson of Universityof Colorado at Boulder were helpful in revising thepaper. The authors also thank data collection teammembers of WHO global mHealth assessment project(Bangladesh Chapter) comprising Benzir Shaon, SaidaMona, Waheduzzaman Adnan, Ismat Ara and Sha-fayet Ullah for their invaluable help.

References

Ajzen, I., and Fishbein, M., 1980. Understanding attitudesand predicting social behavior. Englewood Cliffs, NJ:Prentice-Hall.

Akter, S. and Ray, P., 2010. mHealth – an ultimate platformto serve the unserved. IMIA Yearbook of MedicalInformatics, Germany: Schattauer.

Akter, S., D’Ambra, J., and Ray, P., 2011. Trustworthinessin mHealth information services: an assessment of ahierarchical model with mediating and moderating effectsusing partial least squares (PLS). Journal of AmericanSociety for Information Science and Technology, 62 (1),100–116.

Andaleeb, S.S., 2001. Service quality perceptions andpatient satisfaction: a study of hospitals in a developingcountry. Social Science and Medicine, 52 (9), 1359–1370.

Andaleeb, S.S., 2008. Caring for children: a model ofhealthcare service quality in Bangladesh. InternationalJournal of Quality in Healthcare, 20 (5), 339–345.

Anderson, E.W. and Sullivan M.W., 1993. The antecedentsand consequences of customer satisfaction for firms.Marketing Science, 12 (2), 125–143 (Spring).

Athanassopoulos, A.D., 2000. Customer satisfaction tosupport market segmentation and explain switchingbehaviour. Journal of Business Research, 47 (3), 191–207.

Bagozzi, R.P., 2010. Structural equation models aremodelling tools with many ambiguities: commentsacknowledging the need for caution and humility intheir use. Journal of Consumer Psychology, 20, 208–214.

Bagozzi, R.P. and Yi, Y., 1994. Advanced topics instructural equation models. In: R.P. Bagozzi, ed.Advanced methods of marketing research. Oxford,England: Blackwell, 1–51.

Bansal, H.S., Taylor, S.F., and James, Y.S., 2005. Migratingto new service providers: toward a unifying framework ofconsumers’ switching behaviours. Journal of Academy ofMarketing Science, 33 (1), 96–115.

Baron, R.M. and Kenney D.A., 1986. The moderator-mediator variable distinction in social psychologicalresearch: conceptual, strategic, and statistical considera-tions. Journal of Personality and Social Psychology, 51(6), 1173–1182.

Baroudi, J.J. and Orlikowski, W.J., 1988. A short-formmeasure of user information satisfaction: a psychometricevaluation and notes on use. Journal of ManagementInformation System, 4 (4), 44–59.

Bernhardt, K.L., Donthu, N., and Kennett, P.A., 2000. Alongitudinal analysis of satisfaction and profitability.Journal of Business Research, 47, 161–171.

Berry, L.L. and Bendapudi, N., 2008. Healthcare: a fertilefiled for service research. Journal of Service Research, 10(2), 111–122.

Bhattacherjee, A., 2001. Understanding information systemscontinuance. An expectation–confirmation model. MISQuarterly, 25 (3), 351–370.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Bitner, M.J. and Brown, S.W., 2008. The service imperative.Business Horizons, 51, 39–46 (January–February).

Brady, M.K. and Cronin, J.J., 2001. Some new thoughts onconceptualizing perceived service quality: a hierarchicalapproach. Journal of Marketing, 65, 34–49 (July).

Brady, M.K. and Robertson, C.J., 2001. Searching for aconsensus on the antecedent role of service quality andsatisfaction: an exploratory cross-national study. Journalof Business Research, 51 (1), 53–60.

Caruana, A., Ewing, M.T., and Ramaseshan, R.B., 2000.Assessment of the three-column format SERVQUAL: anexperimental approach, Journal of Business Research, 49(I), 57–65.

Chatterjee, S., et al., 2009. Examining the success factors formobile work in healthcare: a deductive study, DecisionSupport Systems, 46, 620–633.

Chin, W.W., 1998a. Commentary: issues and opinion onstructural equation modeling. MIS Quarterly, 22(March), vii–xvi.

Chin, W.W., 1998b. The partial least squares approach forstructural equation modeling. In: G.A. Marcoulides, ed.Modern methods for business research. Mahwah, NJ:Lawrence Erlbaum, 295–336.

Chin, W.W., 2001. PLS – Graph User’s Guide Version 3.0.Houston, TX: Soft Modeling, Inc.

Chin, W.W., 2010. How to write up and report PLS analyses.In: V. Esposito, et al., eds. Handbook of partial leastsquares: concepts, methods and application. Germany:Springer, 645–689.

Chin, W.W., Marcolin, B.L., and Newsted, P.R., 2003. Apartial least squares latent variable modelling approachfor measuring interaction effects: results from a MonteCarlo simulation study and an electronic-mail emotion/adoption study. Information System Research, 14 (2),189–217.

Chin, W.W. and Newsted, P.R., 1999. Structural equationmodeling analysis with small samples using partial leastsquares. In: R. Hoyle, ed. Statistical strategies for smallsample research. Thousand Oaks, CA: Sage Publications,307–341.

Chiou, J., Droge, C., and Hanvanich, S., 2006. Doescustomer knowledge affect how loyalty is formed?Journal of Service Research, 5 (2), 113–124.

Choi, K.S., et al., 2004. The relationships among quality,value, satisfaction and behavioural intention in healthcare provider choice: a South Korean study. Journal ofBusiness Research, 57 (8), 913–921.

Choi, H., et al., 2007. Contribution to quality of life: a newoutcome variable for mobile data service. Journal ofAssociation of Information System, 8 (12), 598–618.

Cohen, J., 1988. Statistical power analysis for the behaviouralsciences. 2nd ed. Hillsdale: L. Erlbaum Associates.

Cohen, J. and Cohen P., 1983. Applied multiple regression/correlation analysis for the behavioural sciences. 2nd ed.Hillsdale, NJ: Erlbaum.

Cronin, J.J. and Taylor, S.A., 1992. Measuring servicequality: a re-examination and extension. Journal ofMarketing, 56, 55 – 68 (July).

Dabholkar, P.A., Shepherd, D.C. and Thorpe D.I., 2001. Acomprehensive framework for service quality: an inves-tigation of critical conceptual and measurement issuesthrough a longitudinal study. Journal of Retailing, 72 (2),139–173.

Dagger, T.S. and Sweeney, J.C., 2006. The effect of serviceevaluations on behavioural intentions and quality of life.Journal of Service Research, 9 (1), 3–18.

Dagger, T.S., Sweeney, J.C., and Johnson, L.W., 2007. Ahierarchical model of health service quality: scaledevelopment and investigation of an integrated model.Journal of Service Research, 10 (2), 123–142.

Donabedian, A., 1992. Quality assurance in health care:consumers’ role. Quality of Health Care, 1, 247–251.

Earth Institute, 2010. Barriers and gaps affecting mHealth inlow and middle income countries: a policy white paper.Washington, DC: mHealth Alliance.

Edwards, J.R., 2001. Multidimensional constructs in organi-zational behavior research: an integrative analyticalframework. Organizational Research Methods, 4 (2),144–192.

Eskildsen, J. and Kristensen K., 2003. The predictive powerof intangibles. Measuring Business Excellence, 7 (2), 46–54.

Fassnacht, M. and Koese, I., 2006. Quality of electronicservices: conceptualizing and testing a hierarchicalmodel. Journal of Service Research, 9 (19), 19–37.

Ferrans, C.E. and Powers M.J., 1992. Psychometric assess-ment of the quality of life index. Research in Nursing andHealth, 15, 29–38.

Fornell, C.A., 1992. National customer satisfaction barom-eter: the swedish experience. Journal of Marketing, 56, 6–21(January).

Fornell, C. and Bookstein, F.L., 1982. Two structuralequations models: LISREL and PLS applied to con-sumer exit-voice theory. Journal of Marketing Research,19 (4), 440–452.

Fornell, C. and Larcker, D.F., 1981. Evaluating structuralequation models with unobservable variables and mea-surement error. Journal of Marketing Research, 18 (1),39–50.

Geisser, S., 1975. The predictive sample reuse method withapplications. Journal of American Statistical Association,70, 320–328.

Giese, J.L. and Cote J.A., 2000. Defining consumer satisfac-tion. Academy of Marketing Science Review, 1, 1–24.

Gotlieb, J.B., Dhruv, G., and Stephen, W.B., 1994. Consu-mer satisfaction and perceived quality: complementary ordivergent constructs. Journal of Applied Psychology, 79(6), 875–885.

Gronroos, C., 1984. A service quality model and itsmarketing implications. European Journal of Marketing,18 (4), 36–44.

Hair, J., Ringle, C.M., and Sarstedt, M., 2011. PLS-SEM:indeed a silver bullet. Journal of Marketing Theory andPractice, 19 (2), 139–151.

Heisel, M.J. and Flett, G.L., 2004. Purpose in life, satisfac-tion with life, and suicide ideation in a clinical sample.Journal of Psychotherapy and Behaviour Assessment, 26(2), 127–135.

Homburg, C. and Giering, A., 2001. Personal characteristicsas moderators of the relationship between customersatisfaction and loyalty – an empirical analysis. Psychol-ogy and Marketing, 18 (1), 43–66.

Howard, J.A. and Sheth, J.N., 1969. The theory of buyerbehaviour. New York: John Wiley and Sons.

Hulland, J. Ryan, M.J., and Rayner, R.K., 2010. Modelingcustomer satisfaction: a comparative performance eva-luation of covariance structure analysis versus partialleast squares. In: V. Esposito Vinzi, et al., eds. Handbookof partial least squares: concepts, methods and application.Germany: Springer, 307–325.

Iacobucci, D., 2008. Mediation analysis. Thousand oaks, CA:Sage.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Ivatury, G., Moore, J., and Bloch, A., 2009. A doctor in yourpocket: health hotlines in developing countries. Innova-tions: Technology, Governance, Globalization, 4 (1), 119–153.

Jarvis, C.B., MacKenzie, S.B., and Podsakoff, P.M., 2003.A critical review of construct indicators and measure-ment model misspecification in marketing and con-sumer research. Journal of Consumer Research, 30,199–218.

Jia, R., Reich, B.H., and Pearson, J.M., 2008. IT serviceclimate: an extension to IT service quality research.Journal of AIS, 9 (5), 294–320.

Jun, M., Peterson, R. T, and Zsidisin, G.A., 1998. Theidentification and measurement of quality dimensions inhealth care: focus group interview results. Health CareManagement Review, 23 (4), 81–96.

Kahn, J.G., Yang, J., and Kahn, J.S., 2010.The potentialroles of mHealth in developing economies: can they berealized? Health Affairs, 2, 254–261.

Kaplan, B. and Litewka, S., 2008. Ethical challenges oftelemedicine and telehealth. Cambridge Quarterly Health-care Ethics, 17, 401–416.

Law, K.S., Wong, C., and Mobley, W.H., 1998. Toward ataxonomy of multidimensional constructs. Academy ofManagement Review, 23 (4), 741–755.

Limayem, M., Hirt, S.G., and Cheung, C.M.K., 2007. Howhabit limits the predictive power of intention: the case ofinformation systems continuance. MIS Quarterly, 31 (4),705–737.

Ma, Q., Pearson, M., and Tadisina, S., 2005. An exploratorystudy into factors of service quality for applicationservice providers. Information & Management, 42 (8),1067–1080.

MacKenzie, S.B., Podsakoff, P.M., and Jarvis, C.B., 2005.The problem of measurement model misspecification inbehavioral and organizational research and some re-commended solutions. Journal of Applied Psychology, 90(4), 710–730.

Mechael, P., 2009. The case for mHealth in developingcountries. innovations: technology, governance, globaliza-tion, 4 (1), 103–118.

Montoya, M.M., Massey, A.P., and Khatri, V., 2010.Connecting IT services operation to services marketingpractices. Journal of Management Information System, 26(4), 65–85.

O’Connor, S.J., Shewchuk, R.M., and Carney, L.W., 1994.The great gap: physicians’ perceptions of patient servicequality expectations fall short of reality. Journal ofHealth Care Marketing, 14 (2), 32–39.

Oliver, R.L., 1980. A cognitive model for the antecedents andconsequences of satisfaction. Journal of MarketingResearch, 17, 460–469.

Oliver, R.L., 1993. A conceptual model of service quality andservice satisfaction. In: T.A. Swartz, D.E. Bowen, andS.W. Brown, eds. Advances in services marketing andmanagement. Research and practice, Vol. 2, Greenwich,CT: JAI Press, 65–80.

Oliver, R.L., 1997. Satisfaction: a behavioural perspective onthe consumer. Boston, MA: Irwin-McGraw-Hill.

Ostrom, A.L., et al., 2010. Moving forward and making adifference: research priorities for the science of service.Journal of Service Research, 13 (1), 4–36.

Parasuraman, A., Zeithaml, V.A., and Berry, L.L., 1985. Aconceptual model of service quality and its implicationsfor future research. Journal of Marketing, 49, 41–50.

Petter, S., Straub, D., and Rai, A., 2007. Specifyingformative constructs in information systems research.MIS Quarterly, 31 (4), 623–656.

Parasuraman, A., Zeithaml, V. A, and Berry L.L., 1988.SERVQUAL: a multiple-item scale for measuring con-sumer perceptions of service quality. Journal of Retailing,64 (1), 12–37.

Parasuraman, A., Zeithaml, V.A., and Malhotra, A., 2005.E-S-QUAL: a multiple-item scale for assessing electronicservice quality. Journal of Service Research, 7(3), 213–233.

Petter, S., Straub, D., and Rai, A., 2007. Specifyingformative constructs in information systems research.MIS Quarterly, 31 (4), 623–656.

Porter, M.E. and Teisberg, E.O., 2006. Redefining healthcare: creating value based competition on results. Boston:Harvard Business School Press.

Reynolds, N., and Smith, A. 2010. Assessing the impact ofresponse styles on cross-cultural service quality evalua-tion: a simplified approach to eliminating the problem.Journal of Service Research, 13 (2), 230–243.

Rust, R.T., and Oliver, R.L., 1994. Service quality: insightsand managerial implications from the frontier. In: R.T.Rust and R.L. Oliver, eds. Service quality: new directionsin theory and practice. Thousand Oaks, CA: Sage, 1–19.

Saila, T., et al., 2008. Measuring patient assessments of thequality of outpatient care: a systematic review. Journal ofEvaluation and Clinical Practice, 14 (1), 148–154.

Sirgy, J.M., 2001. Handbook of quality of life research.Dordrecht, the Netherlands: Kluwer.

Sobel, M., 1982. Asymptotic confidence intervals for indirecteffects on structural equation models. In: S. Leinhardt,ed. Sociological Methodology, San Francisco, CA:Jossey-Bass, 290–312.

Sousa, R. and Voss, C., 2006. Service quality in multichannelservices employing virtual channels. Journal of ServiceResearch, 8 (4), 356–371.

Spreng, R.A., MacKenzie, S.B. and Olshavsky, R.W., 1996.A reexamination of the determinants of customersatisfaction. Journal of Marketing, 60 (3), 15–32.

Stone, M., 1974. Cross-validatory choice and assessment ofstatistical predictions (with discussion). Journal of theRoyal Statistics Society, Series B, 36, 111–147.

Straub, D., and Watson, R.T., 2001. Research commen-tary: transformational issues in researching IS and net-enabled organizations. Inform System Research, 12 (4),337–345.

Szymanski, D.M., and Henard, D.H., 2001. Customersatisfaction: a meta analysis of the empirical evidence.Journal of Academy of Marketing science. 29 (1), 16–35.

Taylor, S.A., and Baker T.L., 1994. An assessment of therelationship between service quality and customersatisfaction in the formation of consumers’ purchaseintentions. Journal of Retailing, 70 (2), 163–178.

Tenenhaus, M., et al., 2005. PLS path modelling. ComputerStatistics and Data Analysis, 48 (1), 159–205.

Vargo, S.L. and Lusch, R.F., 2004. Evolving to a newdominant logic for marketing. Journal of Marketing, 68,1–17.

Varshney, U., 2005. Pervasive healthcare: applications,challenges and wireless solutions. Communication ofAssociation of Information System, 16 (3), 57–72.

Voss, C.A., et al., 2004. A tale of two countries’ conserva-tism, service quality, and feedback on customer satisfac-tion. Journal of Service Research, 6 (3), 212–230.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Westbrook, R.A., 1981. Sources of consumer satisfactionwith retail outlets. Journal of Retailing, 57, 68–85(Fall).

Wetzels, M., Schroder, G.O., and Oppen, V.C., 2009. UsingPLS path modelling for assessing hierarchical constructmodels: guidelines and empirical illustration. MISQuarterly, 33 (1), 177–195.

Wold, H., 1975. Soft modelling by latent variables: the non-linear iterative partial least squares (NIPALS) approach.In: J. Gani, ed. Perspectives in probability and statistics:papers, in Honour of M.S. Bartlett on the occasion of hissixty-fifth birthday. London: Applied Probability Trust,Academic, 117–142.

Wold, H., 1982. Soft modelling: the basic design andsomemextensions. In: K.G. Joreskog and H. Wold eds.Systems under indirect observation: causality, structure,prediction. Vol. 2.), Amsterdam: North Holland, 1–54.

Wold, H., 1985. Partial least squares In: S. Kotz and N.L.Johnson, eds. Encyclopedia of statistical sciences. Vol. 6.New York: Wiley, 581–591.

World Health Organization. 2008. The world health report.Primary healthcare – now more than ever. Geneva: WHO.

World Health Organization. 2011. mHealth: New horizonsfor health through mobile technologies: second globalsurvey on eHealth. Geneva: WHO.

Woodside, A.G., Frey, L.L., and Daly, R.T., 1989. Linkingservice quality, customer satisfaction, and behaviouralintention. Journal of Health Care Marketing, 9 (4), 5–17.

Zeithaml, V.A., 2000. Service quality, profitability, and theeconomic worth of customers: what we know and whatwe need to learn. Journal of Academy of MarketingScience, 28 (1), 67–85.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Appendix 1: Estimation of service quality as a hierarchical reflective model using PLS path modelling (Chin 2010, Wetzels et al.2009)

Construction of a hierarchical model Service quality as a hierarchical model

Figure 1 shows the first-order latent variables(LVs), i.e. convenience, confidence,cooperation, care and concern, which arerelated to their respective manifest variables(MVs).

Figure 1. First order latent variables of service quality.

Figure 2 shows service quality as a secondorder, hierarchical, reflective latent variable,which is constructed by relating it to theblock of the underlying first-order latentvariables. For instance, service quality isconstructed by using 15 MVs (3þ3þ3þ3þ3)of five first-order constructs. This model isregarded as a hierarchical reflective model,which explains the common variance acrossconvenience, confidence, cooperation, careand concern.

Figure 2. Service quality as a second-order, hierarchical reflective model.Criteria of a reflective model (Jarvis et al. 2003,

Petter et al. 2007):. Direction of causality is from latent variable(construct) to manifest variables (items);

. Indicators are manifestations of theconstruct;

. Indicators are interchangeable, having acommon theme, and dropping of anindicator will not change the conceptualdomain of construct;

. Correlation between any two measures ishighly positive;

. Indicators and constructs have the samenomological net.


Dow

nloa

ded

by [

Uni

vers

ity o

f B

irm

ingh

am]

at 1

0:31

15

Nov

embe

r 20

14

Documents

Modelling the impact of mHealth service quality on satisfaction, continuance and quality of life