Web-based information service adoption:A comparison of the motivational model and the uses and grati fications theoryMargaret Meiling Luo a ,⁎,Sophea Chea b ,Ja-Shen Chen aa College of Management,Yuan Ze University,Taiwan,ROCbDepartment of Management Information Systems,Temple University,United Statesa b s t r a c ta r t i c l e i n f o Article history:Received 21April 2009Received in revised form 7August 2010Accepted 17November 2010Available online 26November 2010Keywords:Technology adoption Information servicesUses and grati fications theory Partial least squaresThis study compares two user acceptance theories:the motivational model (MM),and the uses and grati fications (U&G)theory.While MM arises from the field of information systems and the U&G theory was developed in the field of communication,both are focused on explaining user acceptance of information technologies using intrinsic and extrinsic motivations.We discuss the theoretical roots of the two theories,and use partial least squares (PLS)analysis to test each in an empirical setting.A comprehensive comparison of the results is also presented,including a discussion about the relative strengths and weaknesses of each model under both theoretical/research and practical contexts.©2010Elsevier B.V.All rights reserved.1.IntroductionAdvances in the Internet have enabled some businesses to transform their products into information services that can be personalized and shipped to customers without the usual logistics costs and delays.With Web technologies,there is tremendous opportunity for businesses to create value by providing customers with additional services,or transforming from a product-based to a service-based company.In pursuing digital channels to sell and distribute products,the major challenge these businesses face is evoking user acceptance of the information service once the digital channels are in place.Information services provide clients the capability to generate,acquire,store,transform,process,retrieve,utilize,and/or make available information via telecommunications [46].Today,Web-based information services (WIS)are growing rapidly,and range from entertainment services (e.g.,video-on-demand service,online computer games and gambling)to business-to-consumer (B2C)services,such as online banking [21].These WIS empower users by enabling them to personalize services,compare and select products,and communicate instantly with others.Recent research on the adoption of WIS has taken a motivational perspective,positing that both hedonic and utilitarian needs drive an individual's information technology (IT)adoption decision.Twomotivation theories,the motivational model (MM)[17]and the uses and grati fications (U&G)theory [32],are increasingly being used to investigate the adoption of WIS (e.g.,[20,35,44,57]).The studies that have employed these two theories have helped advance our understanding of the reasons that people adopt WIS.MM and U&G are derived from similar theoretical roots [19,39].Both assume that hedonic and utilitarian needs motivate an in-dividual's adoption of information technology.Yet,recent research findings have not been discussed from a cross-disciplinary perspec-tive,and a comprehensive comparison of these two theories has not been made.A comparison and examination of the boundary conditions of the two theories will provide important information to researchers and practitioners about which theory best explains and predicts the adoption of WIS.We reviewed the literature on the two theories to provide the necessary background information for comparing the models and examining the boundary conditions and methods typically used to test these models.In addition,we tested and compared the theories under an empirical setting of information service acceptance in a non-workplace environment.A partial least squares (PLS)analysis was performed on the results to examine the psychometric properties of scale.2.Theoretical background 2.1.Motivational modelIn the motivational model [17],motivations related to intrinsic personal goals are contrasted with those related to extrinsic goalsDecision Support Systems 51(2011)21–30⁎Corresponding author.135Yuan-Tung Road,Chungli,Taoyuan,Taiwan,ROC.Tel.:+88634638800x2632;fax:+88634630377.E-mail addresses:luo@.tw (M.M.Luo),sophea@ (S.Chea),jchen@.tw (J.-S.Chen).0167-9236/$–see front matter ©2010Elsevier B.V.All rights reserved.doi:10.1016/j.dss.2010.11.015Contents lists available at ScienceDirectDecision Support Systemsj o ur n a l h o m e p a ge :ww w.e l s e v i e r.c o m/l o c a t e /d s sassociated with job advancement issues.Davis et al.[17]associated perceived usefulness with performance as a consequence of use(i.e., as an extrinsic motivation).Intrinsic motivation,in contrast,is con-cerned with the reinforcement and enjoyment of the process of performing a behavior.System use,therefore,is determined not only by extrinsic motives(i.e.,perceived usefulness)but also intrinsic motives(i.e.,perceived enjoyment).Several studies have confirmed that both intrinsic and extrinsic motives are significant predictors of behavioral usage[1,3,37,44,55,56,63–65].The conceptual model of MM is shown in Fig.1.es and gratifications theoryThe uses and gratifications theory is a media use paradigm that inquires into the reasons why people use certain media and the gratifications derived from usage and access.It posits that media consumption is purposive,and that users actively seek to fulfill their needs via a variety of uses[32].U&G has been considered a useful approach for understanding users'motivations in the context of traditional media,such as TV and radio[8,43,53].Its application to interactive media,such as the Internet,is promising[38,54].A large body of literature suggests that motivations predict media usage and that media usage influences gratifications/satisfaction[7,22,35,36,49–51,53].The motivations identified in U&G studies are more diverse than those in MM studies.As can be seen in Table1,intrinsic motivations(engaging in an activity out of pleasure and satisfaction) and extrinsic motivations(engaging in an activity to obtain something outside the activity[62,p.273])of media/IT usage include the need to search for information,to interact socially,and for escapism.In particular,recent studies have reported the following motivations of using the Internet:convenience,information seeking,interpersonal utility,to pass time,entertainment,escapism,peer identity,and social interaction[11,33,50].Thesefindings suggest that using a media perspective to investigate the WIS use may potentially broaden theview of IS researchers who commonly take an information system perspective to examine Web adoption because today a WIS has characteristics and features of not only an information system but also a medium.Recent U&G studies on the Internet have identified a set of common underlying dimensions of Internet usage motivations that reflect the inherent interactivity and user-directed nature of Internet media.Thesefindings also suggested that the U&G approach provides the theoretical framework for understanding what specifically drives user adoption of the Internet.Hence,the U&G approach is particularly well suited for examining Internet use in a customer context.Among the U&G studies,Kaye[33]used motivation items of TV and identified six factors:entertainment,to pass time,escapism,social interaction, information seeking,and Web site preference.Ferguson and Perse [22]suggested that people use the Internet for entertainment,to pass time,relaxation,and social information.Papacharissi and Rubin[50] developed a scale of Internet usage motivation consisting offive motives:entertainment,to pass time,interpersonal utility,informa-tion seeking,and convenience;these draw from interpersonal,media, and new technology motives,supporting informative and interactive capabilities of the Internet.The most salient use of the Internet was information seeking,which reflected an instrumental orientation of media use(Fig.2).The Internet possesses unique characteristics,including machine interactivity,telepresence,hypermedia,and network navigation,that distinguish it from tradition media and some other interactive multimedia[45]these characteristics gratify entertainment and information needs[54,58].Because of these unique characteristics, researchers have stressed the need to develop a rigorous researchFig.1.The motivational model.Table1Intrinsic and extrinsic motivations of technology use.Authors Medium Intrinsic ExtrinsicBerelson[5]Newspaper Source of security Searching forinformationCeremonialRitualisticTool for everyday living Social contact forconversationalpurposeEscapeSocial prestigeRubin[53]TV Pass time/Habit InformationEntertainment CompanionshipEscapeKaye[33]WWW Entertainment Social interactionPass time InformationEscape Web site preference Papacharissi andRubin[50]Internet Entertainment Information seekingPass time ConvenienceInterpersonal utilityFerguson andPerse[22]TV-relatedweb surfingmotivationsEntertainment Social informationPass timeRelaxationStafford andStafford[57]Commercial Entertainment Search factorWeb sites Cognitive factorNews and uniquefactorsCharney andGreenberg[11]Internet Division entertainment Keep informedPeer identity CommunicationGood feelings Sights and soundsCoolness CareerKo,Cho,andRoberts[35]MarketingwebsiteEntertainment InformationSocial interaction ConvenienceDiddi andLaRose[20]Internet news Escapism SurveillanceEntertainmentHabitNews quizzesPass timeFig.2.The uses and gratifications of internet usage.22M.M.Luo et al./Decision Support Systems51(2011)21–30framework for both theory and practice[25].A comparison of the two theories(two lines of investigation)would show to what degree IT and its intrinsic qualities correlate with the pursuit of certain gratifications that lead to the acceptance of IT.Among the media/IT acceptance theories(e.g.,diffusion of innovation[52],technology acceptance model[15],and theory of planned behavior[2]),U&G is distinct in its applicability to media of all kinds[38].The U&G paradigm's comprehensive nature enables us to understand electronic communication in a media environment where computers have not only home and business applications but also work and play functions [58].In contrast,IT acceptance models such as MM and the technology acceptance model(TAM)regard information systems like the Internet as tools of improving user performance at work.Applying MM to the Internet is therefore limited in scope.Therefore,while it could be argued that U&G brings a new perspective to understanding the adoption decision in a customer context,such an argument should be made only with supporting evidence from a fair comparison of the two models.In the following section,we compare the two models and discuss how the comparison contributes to our understanding of the nature of the two theories.parison of MM and U&GA comprehensive comparison of MM and U&G has not yet been conducted,but such a comparison is plausible and valuable for the following reasons.First,both theories posit that intrinsic and extrinsic motivations drive an individual's behavior.MM associates these motivations with the use of IT,while U&G similarly links them to media usage behaviors.This similarity provides a baseline for comparing the models,since they use the same approach,namely an information-process approach[23,48],to explain IT usage.Secondly,the two theories propose distinct constructs to explain the use of IT.For example,the intrinsic motivation“to pass time”is unique to U&G studies and not discussed in MM ing media/ IT may sometimes be habitual[33,53].This is a whole new perspective to MM because the model neglects hedonic factors of system use where a ritual pattern of usage is plausible.Another example is entertainment,which is an intrinsic motivation in U&G that explains the substantial variability in Internet usage[50].The entertainment motive is somewhat different from the perceived enjoyment motive in MM,but they are both intrinsic motivations associated with fun and playfulness inherent to the adoption decision process.According to its conceptual definition,the entertainment motive drives media use when customers get pleasure from consuming media content and the process of that consumption[53].Perceived enjoyment,by definition, is pleasure from experiencing the process,but not the specific use of the media/IT application content.Several studies have suggested that this motive has a significant impact on behavior intention to use personal computers and the Web[17,30,31,63].Note that these findings of MM concern the application,but not the content of the application.Given that the entertainment motive of U&G and the perceived enjoyment motive of MM share a similar attribute,an empirical comparison of these intrinsic factors of IT usage will enhance our understanding of Web-based services adoption.Thirdly,both theories have been tested extensively under various empirical settings,and their scales have been shown to be reliable and valid.Given the maturity of this research domain and rigor associated with the methodological details of the two theories,the results of each model are comparable.While there are some inherent boundary conditions imposed by the empirical setting that we used,the comparison should be meaningful.The conceptual definitions of the main constructs of each theory are listed in Appendix A.The models were compared using the three criteria used by Mathieson[40].First,we examined how well the two models predict behavioral usage of an information service.If the factors of each model largely account for the observed variance,we can argue that these factors provide an accurate picture of the issues that system developers should consider to increase system paring the models'respective levels of variance as explained also provides evidence on which model is superior.Secondly,we compared the value of the information provided by the models.Under the assumption that the comparison should not be biased to favor one model or the other,we examined the empirical evidence on which factors better predict behavioral usage.This identifies which factors are most important to users when adopting the WIS.Thirdly,it is important that a model provides valuable information at low cost and ideally a minimum of effort.To ensure a fair comparison,we examined both procedure and distributional equivalence[14].Procedure equivalence was assessed under two conditions.First,the boundary conditions of both theories were observed.When performing a procedure equivalent test,it is important to obtain equally careful measures of each theory's constructs and equivalentfidelity to the boundary conditions of each theory.Secondly,equal attention was given to measurement. Distributional equivalence aims to ensure that the factors or variables vary over equivalent ranges of values in their respective populations ([14]).However,establishing distributional equivalence directly is difficult because knowledge of population variance is required.When population variances are unknown,establishing distributional equiv-alence is not certain.An alternative is to examine the standard deviations of variables that are not common to both models,which helps identify obvious cases of range restriction[40].An important aspect of this study is that we tested both models in the same context,using subjects sampled from the same population. All subjects faced the same adoption decision about the same information service.This allowed us to observe the differences between the models that are likely due to the models themselves, rather than differences in the contexts under investigation.To test and compare the two theories under an empirical setting of Web service acceptance in the non-workplace environment,we conducted a lab experiment to collect data and used PLS analysis to examine the extent to which the modelsfit the data.In the following sections,we present the methods,measures,and results of that experiment,and discuss the limitations of this study and the implica-tions of ourfindings.3.Method3.1.Data collection procedureThe lab experiment was conducted in Spring,2006,and involved a questionnaire to collect the data.The empirical setting was a WIS, namely,online newspapers,whose primary offering is news content. Prior to the experiment,we conducted two pilot studies to develop pertinent scales.The studies were performed at a major U.S. university.More details of these pilot studies are presented in the next subsection.The participants in the lab experiment were business school undergraduate students enrolled in two sections of statistic classes at a major U.S.university.Extra credit was granted for participation,and participants were allowed to decline to participate or withdraw from the study at any time.Participants were randomly assigned to networked computers to systematically prevent cross-talking.After afive-minute introduction,73participantsfilled out an online questionnaire,which included all of the measures of MM and U&G.All73responses were valid.Two weeks later,the same group of participants completed a follow-up paper-based questionnaire with only measures of behavioral usage.A total of63responses were deemed usable for data analysis.Thefirst survey focused on collecting data on motivation and behavioral intention variables while the second survey on reported usage data.The practice of deploying a two-stage survey is a widely accepted in technology acceptance research[15,59].Because both models need the usage data to run the23M.M.Luo et al./Decision Support Systems51(2011)21–30full model,only the63responses that were deemed usable were included in the analysis.The63participants who completed the questionnaire were comprised of32(50.8%)males and31(49.2%)females between19 and35years of age.The majority(81.5%)was between18and 23years of age,and98%were in their junior and senior years of college.Note that ourfindings might only be applicable to the information service population thatfits these characteristics.3.2.MeasuresMM includes perceived enjoyment and measures used in TAM.The measures of intention to use,perceived ease of use,perceived use-fulness,and behavioral usage were adopted from Davis[15].Items of perceived ease of use and perceived usefulness were measured using a seven-point Likert-type scale(1=Likely,7=Unlikely).Perceived enjoyment was measured by four semantic differential scales (ranging from one to seven)taken from[10]and[30]and re-adapted by van der Heijden[63].Attitude,adapted from Wixom and Todd[67], was measured by three items using semantic differential scales, ranging from one to seven.The predictors in the U&G model(i.e.,extrinsic and intrinsic motivations)were measured using the scales developed by Papachar-issi and Rubin[50].The reason to adopt these items is twofold.First, their validity is well-established in previous research.Second,these items are derived from previous researchfindings on interpersonal communication,mediated communication,and new technology motives,which reflect the motives of using the Internet that is similar to our research context.Furthermore,they are the most current Internet motive scales.Four items elicited by open-ended questions in thefirst pilot study,which involved23business school students in a junior-level IT course,were included.These items reflect the motives of using news WIS.We expect these items to complement Papacharissi and Rubin's Internet motive items.All motive items were measured on afive-point Likert-type scale.The second pilot study tested behavioral usage and behavioral intention items of30college students enrolled in an undergraduate junior-level IT course.A total of 29responses were collected.The pilot study tested the measures in a customer context,and the results suggested that these items were adoptable.4.ResultsA partial least squares(PLS)analysis using PLS Graph(Version3.0) was conducted to examine the reliability and validity of the mea-sures.The loading pattern of all items was highly consistent,with most loadings above0.70[61]on their intended constructs with relatively small cross loadings and significant t-values at p b0.001. However,there were two exceptions.One item in attitude towards usage loaded poorly at−0.434and was subsequently dropped from the model(Tables2and3).Another item for pass time loaded at0.502 but was kept in the model because its cross loadings on other constructs were relatively small.Note that the attitude measures loaded very high on their intended constructs(0.977and0.977), suggesting that these items were a good measure of attitude.How-ever,these items loaded relatively high(0.671and0.693)on per-ceived enjoyment.A possible explanation is that attitude consists of both affective and cognitive structures[6,29,47,68]and perceived enjoyment reflects a person's affective evaluation of system use. Therefore,the two constructs have some commonality,and conse-quently high loadings and inter-correlation(see Table4)can be expected.The major distinction between these two constructs is that the former represents the evaluation of emotion(i.e.,[2,47])while the latter the information an individual has about an object[48]. Therefore,they are still different conceptually but some similarities can be anticipated.4.1.Scale validity and instrument reliabilityWe employed the bootstrapping method(200sub-samples)to test the validity of the constructs and the level of significance of the regression path coefficients.The internal composite reliability,squareTable2Item loadings(MM).Latent constructs Item Latent constructsPEU PU PE ATT BI BUPerceived ease of use(PEU)PEU010.811⁎⁎⁎0.4530.1920.2810.1530.198PEU020.901⁎⁎⁎0.4110.2050.3280.1680.229PEU030.869⁎⁎⁎0.5050.1510.2770.1500.268PEU040.916⁎⁎⁎0.5030.1760.2900.2120.225PEU050.901⁎⁎⁎0.4930.1930.3840.2540.204PEU060.926⁎⁎⁎0.5770.2830.3960.2010.231 Perceived usefulness(PU)PU010.5510.845⁎⁎⁎0.3340.4820.3370.429PU020.4590.920⁎⁎⁎0.3100.4720.2570.384PU030.5280.905⁎⁎⁎0.2160.3320.1980.248PU040.5510.929⁎⁎⁎0.3680.5020.3260.391PU050.4470.845⁎⁎⁎0.3330.3640.2100.209PU060.3710.833⁎⁎⁎0.3110.4300.3360.240 Perceived enjoyment(PE)PE010.2510.3010.947⁎⁎⁎0.6810.5780.418PE020.0920.2970.904⁎⁎⁎0.6430.4860.359PE030.1980.3290.940⁎⁎⁎0.6750.5860.469PE040.2940.3780.838⁎⁎⁎0.5210.6110.610 Attitude towards use(ATT)ATT010.3390.5230.6710.977⁎⁎⁎0.4440.478ATT020.3860.4410.6930.977⁎⁎⁎0.4180.349 Behavioral intention(BI)BI010.1120.2410.5610.2780.894⁎⁎⁎0.356BI020.2170.3300.6320.4320.973⁎⁎⁎0.462BI030.1900.2860.6520.4300.941⁎⁎⁎0.382BI040.2650.3140.3870.5120.793⁎⁎⁎0.311BI050.1570.2350.5630.2730.898⁎⁎⁎0.346 Behavioral usage(BU)BU010.2740.4460.5050.4980.4250.877⁎⁎⁎BU020.0760.1050.3160.1970.2990.800⁎⁎⁎BU030.2820.2640.2990.1630.1210.600⁎⁎⁎Note:N=63.Item loadings higher than0.5are in bold.⁎:p b0.05,⁎⁎:p b0.01,⁎⁎⁎:p b0.001.24M.M.Luo et al./Decision Support Systems51(2011)21–30roots of average variance extracted(AVE),and the inter-construct correlations were all at acceptable levels(Tables4and5).In PLS, convergent and discriminant validity of the measurement were assessedfirst by examining the square root of AVE for each construct to see whether they are larger than0.7and larger than the correlation to other constructs,and secondly by comparing each item's loading on its construct against its cross loadings on other constructs to make sure the latter is relatively smaller than the former[24].Our sample met both criteria,signifying the validity of the constructs.Further-more,the internal composite reliability(ICR)of all constructs was greater than0.80(except for PT which has ICR of0.78)which is an indication of construct reliability.These results suggest that the constructs were reliable and valid.4.2.Explaining behavioral usageThe PLS analysis estimated the path coefficients of the two models. Tables6and7present the path coefficients for each relationship,their respective t-value,and the variance explained for each dependent variable.In MM,and perceived enjoyment significantly influenced attitude towards usage,whereas perceived usefulness and perceived ease of use did not have a significant impact on attitude towards usage.The causal links between attitude,behavioral intention,and behavioral usage were supported by the data.On the other hand,in U&G,information seeking,entertainment,and interpersonal utility were the most salient predictors of behavioral usage.The remaining two motivation factors did not have a significant impact on behavioral usage.In PLS analysis,a good modelfit is established with significant path coefficients,acceptably high R2,and internal composite reliability of constructs[24].As supported by the results of our PLS analysis,the data fairlyfit both models.The strength of all relationships among the MM and U&G constructs was high(standard path greater than0.2 [42]),indicating that all relationships were meaningful.We were particularly interested in the variance in behavioral usage explained by each model.We found that U&G theory explained more variance (R2=0.367)than MM did(R2=0.173).The difference may be attributed to the use of more variables in U&G to explain behavioral usage.The items of both models have comparable internal composite reliability and square root of AVE.Therefore,we conclude that both models predict behavioral usage well,though U&G explains more variance than does MM.We will further discuss the results in the discussion section.Table3Item loadings(U&G theory).Latent constructs Item Latent constructsIU C PT E IS BUInterpersonal utility(IU)M010.725⁎⁎⁎0.291−0.0300.275−0.0330.170 M020.746⁎⁎⁎0.1760.0500.3090.1190.351M030.867⁎⁎⁎0.3250.0970.2740.0530.339M040.787⁎⁎⁎0.3420.1200.2590.1170.276M050.829⁎⁎⁎0.3700.1740.4720.1560.369 Convenience(C)M160.3700.954⁎⁎⁎0.0260.3970.3040.223 M170.3390.937⁎⁎⁎0.1000.3400.3230.190 Pass time(PT)M070.097−0.0590.909⁎⁎⁎0.1630.3020.148 M090.0450.1920.769⁎⁎⁎0.1570.3800.113M08−0.1000.0050.502⁎⁎⁎0.0800.403−0.043 Entertainment(E)M180.3990.4240.3160.844⁎⁎⁎0.2440.371 M190.3210.3150.2120.911⁎⁎⁎0.3430.359M200.3820.3440.1450.927⁎⁎⁎0.3130.422M220.3150.2770.0220.772⁎⁎⁎0.4260.443 Information seeking(IS)M100.0760.2930.2860.3570.678⁎⁎⁎0.323 M110.1610.2170.3000.2100.747⁎⁎⁎0.329M120.1360.2650.3410.1960.814⁎⁎⁎0.357M13−0.0010.2410.1620.4320.828⁎⁎⁎0.363 Behavioral usage(BU)BU010.3690.2110.1840.5360.4580.866⁎⁎⁎BU020.2090.0940.0900.2320.3330.771⁎⁎⁎BU030.3370.1990.1320.2020.1720.676⁎⁎⁎Note:N=63.Item loadings higher than0.5are in bold.⁎:p b0.05,⁎⁎:p b0.01,⁎⁎⁎:p b0.001.Table4Measurement model estimation(MM).AVE/correlation Latent constructsLatent constructs ICR(1)(2)(3)(4)(5)(6)(1)Perceived ease of use0.960.89(2)Perceived usefulness0.950.560.88(3)Perceived enjoyment0.950.230.360.91(4)Attitude towards use0.980.370.490.700.98(5)Behavioral intention0.960.220.320.620.440.90(6)Behavioral usage0.810.250.370.510.420.420.77Notes:1.ICR:internal composite reliability.2.Diagonal elements are the square root of the shared variance between the constructs and their measures;the square root of the shared variance higher than0.7is in bold.Square root of the shared variance across items measuring a construct was higher than correlations across constructs shows that discriminant validity across constructs is supported.off-diagonal elements are correlations between constructs.Table5Measurement model estimation(U&G theory).AVE/correlation Latent constructLatent construct ICR(1)(2)(3)(4)(5)(6)(1)Interpersonal utility0.890.79(2)Convenience0.940.380.95(3)Pass time0.780.120.060.75(4)Entertainment0.920.410.390.190.87(5)Information seeking0.850.120.330.350.390.77(6)Behavioral usage0.820.400.220.180.470.450.77Notes:1.ICR:internal composite reliability.2.Diagonal elements are the square root of the shared variance between the constructs and their measures;the square root of the shared variance higher than0.7is in bold.Square root of the shared variance across items measuring a construct was higher than correlations across constructs shows that discriminant validity across constructs is supported.off-diagonal elements are correlations between constructs.25M.M.Luo et al./Decision Support Systems51(2011)21–30。
