A study of student satisfaction in a blended e-learning system environment

Jen-Her Wu a ,Robert D.Tennyson b,*,Tzyh-Lih Hsia c

Department of Information Management,National Sun Yat-Sen University,70Lien-Hai Road,Kaohsiung,80424,Taiwan b

University of Minnesota,56East River Road,Minneapolis,Minnesota 55455,United States c

Department of Information Management,Chinese Naval Academy,P.O.Box No.90175Tsoying,Kaohsiung 813,Taiwan

a r t i c l e i n f o Article history:

Received 18March 2009

Received in revised form 23December 2009Accepted 31December 2009

Keywords:e-Learning Satisfaction Learner control Internet

Teacher-directed Learner-directed Synchronous Asynchronous Face-to-face

a b s t r a c t

This study proposes a research model that examines the determinants of student learning satisfaction in a blended e-learning system (BELS)environment,based on social cognitive theory.The research model is tested using a questionnaire survey of 212participants.Con?rmatory factor analysis (CFA)was per-formed to test the reliability and validity of the measurements.The partial least squares (PLS)method was used to validate the measurement and hypotheses.The empirical ?ndings indicate that computer self-ef?cacy,performance expectations,system functionality,content feature,interaction,and learning climate are the primary determinants of student learning satisfaction with BELS.The results also show that learning climate and performance expectations signi?cantly affect learning https://www.360docs.net/doc/3317219542.html,puter self-ef?cacy,system functionality,content feature and interaction signi?cantly affect performance expec-tations.Interaction has a signi?cant effect on learning climate.The ?ndings provide insight into those fac-tors that are likely signi?cant antecedents for planning and implementing a blended e-learning system to enhance student learning satisfaction.

1.Introduction

Classroom learning typically occurs in a teacher-directed instructional context with face-to-face interaction in a live synchronous envi-ronment.In contrast to this form of instruction,is an approach that promotes learner-directed learning.With emerging Internet commer-cialization and the proliferation of information technologies,online or electronic learning (e-learning)environments offer the possibilities for communication,interaction and multimedia material delivery that enhance learner-directed learning (Wu,Tennyson,Hsia,&Liao,2008).Although e-learning may increase access ?exibility,eliminate geographical barriers,improve convenience and effectiveness for indi-vidualized and collaborative learning,it suffers from some drawbacks such as lack of peer contact and social interaction,high initial costs for preparing multimedia content materials,substantial costs for system maintenance and updating,as well as the need for ?exible tutorial support (Kinshuk &Yang,2003;Wu et al.,2008;Yang &Liu,2007).Furthermore,students in virtual e-learning environments may expe-rience feelings of isolation,frustration and confusion (Hara &Kling,2000)or reduced interest in the subject matter (Maki,Maki,Patterson,&Whittaker,2000).In addition,student satisfaction and effectiveness for e-learning has also been questioned (Piccoli,Ahmad,&Ives,2001;Santhanam,Sasidharan,&Webster,2008).

With the concerns and dissatisfaction with e-learning,educators are searching for alternative instructional delivery solutions to relieve the above problems.The blended e-learning system (BELS)has been presented as a promising alternative learning approach (Graham,2006).BELS refers to an instructional system that combines multiple learning delivery methods,including most often face-to-face class-room with asynchronous and/or synchronous online learning.It is characterized as maximizing the best advantages of face-to-face and online education.

While BELS has been recognized as having a number of advantages (e.g.,instructional richness,access to knowledge content,social interaction,personal agency,cost effectiveness,and ease of revision (Osguthorpe &Graham,2003)),insuf?cient learning satisfaction is still an obstacle to the successful BELS adoption (So &Brush,2008).In fact,research ?ndings from Bonk and colleagues have shown that learn-ers had dif?culty adjusting to BELS environments due to the potential problems in computer and Internet access,learners’abilities and

*Corresponding author.

E-mail address:rtenny@https://www.360docs.net/doc/3317219542.html, (R.D.Tennyson).

Computers &Education 55(2010)

155–164

Contents lists available at ScienceDirect

Computers &Education

j o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /c o m p e d

156J.-H.Wu et al./Computers&Education55(2010)155–164

beliefs in the use of technology,blended course design,participant interaction,and blended environments integration(Bonk,Olson, Wisher,&Orvis,2002).These?ndings imply that an effective BLES environment should consider the human and technology factors that affect learner satisfactions with BELS,such as individual attitudes,participant interaction,educational technologies,and course design(Wu et al.,2008).Thus,more careful analysis of learners,educational technologies,and social contexts in BELS environments are needed(EL-Deghaidy&Nouby,2008).

The adoption of BELS in supporting learning has made it signi?cant to probe the crucial determinants that would entice learners to use BELS and enhance their learning satisfaction.The degree of student learning satisfaction with BELS courses plays an important role in eval-uating the effectiveness of BELS adoption.Hence,comprehending the essentials of what determines student learning satisfaction can pro-vide management insight into developing effective strategies that will allow educational institution administrators and instructors to create new educational bene?ts and value for their students.Because BELS environments differ from typical classroom and virtual e-learn-ing,a review of previous research in learning technology shows that there is a lack of studies that have examined the crucial factors that determine learning satisfaction with BELS,such as individual cognition,technological environments,and the social contexts,as stated above.There is a need for more in-depth research to understand what determines student learning satisfaction in a BELS environment and to investigate how the determinants in?uence student perceptions of BELS contexts and their correlations.This study,therefore,pro-poses a research model,based on the social cognitive theory(Bandura,1986),to investigate the primary determinants affecting student learning satisfaction in a BELS environment.We also empirically validate the proposed model and examine the relationships among those latent variables.

2.Basic concepts and theoretical foundation

2.1.Blended e-learning system

Blended learning is described as a learning approach that combines different delivery methods and styles of learning.The blend could be between any form of instructional technology(e.g.,videotape,CD-ROM,CAI,web-based learning)with classroom teaching.Recently there has been an increasing movement toward blending e-learning and face-to-face activities with students participating in collaborative learn-ing and interaction with their instructors and classmates.This is called‘‘blended e-learning”or‘‘blended e-learning system”(Graham, 2006;Singh,2003).

Graham(2006)de?ned BELS as a mixing of instruction from two historically separate learning environments:classroom teaching and full e-learning.The term emphasizes the central role of computer-based technologies(e-learning systems)in blended learning,focusing on access and?exibility,enhancing classroom teaching and learning activities,and transforming the way individuals learn.From a course de-sign perspective,a BELS course can lie anywhere between the continuum anchored at opposite ends by full face-to-face and virtual e-learn-ing approaches(Rovai&Jordan,2004).Kerres and De Witt(2003)identi?ed three critical components of BELS that considers the content of the learning materials,the communication between learners and tutors and between learners and their peers,and the construction of the learners’sense of place and direction within the activities that denote the learning environment.This is an important distinction because it is certainly possible to enhance regular face-to-face courses with online resources without displacing classroom contact hours.Accordingly, we de?ned BELS as the combination of online and face-to-face instruction and the convergence between traditional face-to-face learning and e-learning environments.

Several BELSs,such as WebCT(https://www.360docs.net/doc/3317219542.html,)and Cyber University of NSYSU(https://www.360docs.net/doc/3317219542.html,.tw)have developed systems that inte-grate a variety of functions to facilitate learning activities.For example,these systems can be used to integrate instructional material(via audio,video,and text),e-mail,live chat sessions,online discussions,forums,quizzes and assignments.With these kinds of systems, instructional delivery and communication between instructors and students can be performed at the same time(synchronously)or at dif-ferent times(asynchronously).Such systems can provide instructors and learners with multiple,?exible instructional methods,educa-tional technologies,interaction mechanisms or learning resources and applying them in an interactive learning environment to overcome the limitations of classroom and e-learning.As a result,these online learning systems may better accommodate the needs of learners or instructors who are geographically dispersed and have con?icting schedules(Pituch&Lee,2006).As BELS emerge as perhaps the most prominent instructional delivery solution,it is vital to explore what determines learning satisfaction in a blended e-learning environment.

2.2.Social cognitive theory

Social cognitive theory(Bandura,1986)serves as an initial foundation in this study for exploring what determines student learning sat-isfaction in a blended e-learning environment.Social cognitive theory is a widely accepted and empirically validated model for understand-ing and predicting human behavior and identifying methods in which behavior can be changed.Several studies have applied it as a theoretical framework to predict and explain an individual’s behavior in IS settings.The theory argues that the meta progress of a human being occurs through consecutive interactions with the outside environment and the environment must be subjected to one’s cognition process before they affect one’s behavior.It proposes that a triadic reciprocal causation among cognitive factors,environmental factors, and human behavior exists.Behavior is affected by both cognitive factors and environmental factors(Wood&Bandura,1989).Cognitive factors refer to the personal cognition,affect and biological events.Environmental factors refer to the social and physical environments that can affect a person’s behavior.

Environments in?uence an individual’s behavior through his or her cognitive mechanisms.Hence,social cognitive theory posits two critical cognitive factors:performance expectations and self-ef?cacy that in?uence individual behavior.It gives prominence to the concept of self-ef?cacy–de?ned as one’s judgments and beliefs of his/her con?dence and capability to perform a speci?c behavior–recognizing that our performance expectations of a behavior will be meaningless if we doubt our capability to successfully execute the behavior in the ?rst place.It can enhance human accomplishment and well-being,help determine how much effort people will expend on a behavior,how long they will persevere when confronting obstacles and how resilient they will be in the face of adverse situations.The theory further

J.-H.Wu et al./Computers&Education55(2010)155–164157 argues that self-ef?cacy in?uences performance expectations and performance expectations also in?uence behavior.Thus,self-ef?cacy and performance expectations are held to be the principal cognitive determinants of individual behavior.

Regarding environmental factors,there is ample educational literature and research that shows the learning environment affects a learners’behavior and performance.Traditionally,a learning environment was de?ned in terms of the physical and social environments in a classroom setting.Piccoli et al.(2001)expanded the traditional de?nition of learning environment and identi?ed?ve environmental factors that clarify how an e-learning environment differs from classroom-based education,including technology,content,interaction, learning model,and learner control.These factors can be classi?ed into two categories that particularly are relevant to BELS-speci?c envi-ronments.The?rst category relates to the technological environment that includes system functionality and content feature.The second category relates to social environments that include interactions(between learners and instructors or between learners and other learners) and learning climate.

3.Research model and hypotheses

Based on the foregoing theoretical underpinnings,we consider that the social cognitive theory is applicable to the BELS learning context. Accordingly,three factors:learners’cognitive beliefs(self-ef?cacy and performance expectations),technological environment(system functionality and content feature),and social environment(interaction and learning climate)are identi?ed and elucidated as the primary dimensions of student learning satisfactions with BELS,as shown in Fig.1.

3.1.Cognitive factors

Cognitive factors refer to the learners’cognitive beliefs that in?uence their behaviors in using BELS.Two main cognitive variables:com-puter self-ef?cacy and performance expectations are believed to be the most relevant factors affecting human behavior in using an infor-mation system(IS)(Compeau&Higgins,1995;Compeau,Higgins,&Huff,1999;Venkatesh,Morris,Davis,&Davis,2003).The social cognitive theory de?ned performance expectations as the perceived consequences of a behavior and further noted they are a strong force guiding individuals’actions.The performance expectations are derived from individual judgments regarding valuable outcomes that can be obtained through a requisite behavior.Individuals are more likely to perform behaviors that they believe will result in positive bene?ts than those which they do not perceive as having favorable consequences.

Performance expectations are de?ned as the degree to which a learner believes that using BELS will help him or her to attain gains in learning performance.The de?nition is similar to the concepts of perceived usefulness,based on Davis’s(1989)technology acceptance model(Venkatesh et al.,2003).The in?uence of performance expectations on individual behavior of using computer systems has been demonstrated by Compeau and Higgins(1995),Compeau et al.(1999)and Venkatesh et al.(2003).Prior research in education or com-puter-mediated learning has found that performance expectations are positively related to students’learning performance(Bolt&Koh, 2001)and satisfaction(Martins&Kellermanns,2004;Shih,2006).

Individual attitudes are a function of beliefs,including the behavioral beliefs directly linked to a person’s intention to perform a de?ned behavior(Ajzen&Fishbein,1980).User acceptance is an important indicator that measures a user’s positive attitudes toward the IS and predicts their behaviors while using the system,based on theory of reasoned action(Taylor&Todd,1995).Satisfaction is a good surrogate for user acceptance and is often used to measure learners’attitude in computer-mediated learning studies(Chou&Liu,2005;Piccoli et al., 2001).Thus,we conceptualize the student’s attitude toward BELS as the learning satisfaction with the BELS–de?ned as the sum of stu-dent’s behavioral beliefs and attitudes that result from aggregating all the bene?ts that a student receives from using BELS.Therefore,the following hypothesis is proposed.

H1:A higher level of performance expectations for BELS use will positively associate with a higher level of learning satisfaction with BELS.

The second cognitive factor to be applied in this research is self-ef?cacy.In general,it refers to an individual’s beliefs about his or her capa-bilities to successfully perform a particular behavior.According to social cognitive theory,individuals form their perceptions of self-ef?cacy toward a task based on cue they receive from the four information sources:(1)past experience and familiarity with similar activities,(2)vicar-ious learning,(3)social support and encouragement,and(4)attitudes toward the task.Bandura(1986)noted self-ef?cacy is task-speci?c and its measures should be tailored to the targeted domain context.Accordingly,several studies have investigated self-ef?cacy beliefs towards tasks such as computers and IS-related behaviors(Compeau&Higgins,1995;Compeau et al.,1999).Derived from the general de?nition of self-ef?cacy,computer self-ef?ciency was de?ned as the individual ability to use information technology to accomplish computer-related tasks or jobs(Marakas,Yi,&Johnson,1998).Computer self-ef?cacy was also validated as a determinant of IS acceptance and use.

We de?ne computer self-ef?cacy as the con?dence in one’s ability to perform certain learning tasks using BELS.Prior research has shown that increases in computer self-ef?cacy improve initiative and persistence,which lead to improved performance or outcome expec-tations(Francescato et al.,2006;Johnston,Killion,&Oomen,2005;Piccoli et al.,2001),including attitude and behavioral intention(Venk-atesh&Davis,2000).In the context of computer-mediated learning,empirical evidence indicates that increases in computer self-ef?cacy improve students’con?dence in their computer-related capabilities,which in turn leads to a perception of positive performance expecta-tions to the learning courses(Bolt&Koh,2001;Jawahar&Elango,2001;Santhanam et al.,2008;Shih,2006).That is,computer self-ef?cacy could reduce learning barriers in using BELS.If students have higher computer self-ef?cacy and can control BELS,they will perceive the systems’usefulness and value,which in turn motivates their intention to use BELS.Accordingly,the following hypothesis is proposed: H2:A higher level of individual’s computer self-ef?cacy will positively associate with a higher level of performance expectations for BELS use.

3.2.Technological environment

The quality and reliability of an e-learning system,as well as easy access to appropriate educational technologies,material content,and course-related information are important determinants of e-learning effectiveness(Piccoli et al.,2001).Thus,system functionality and

158J.-H.Wu et al./Computers&Education55(2010)155–164

content features are identi?ed as critical technological environment factors for BELS.They are expected to in?uence the learner to use and accept BELS.Prior research has shown that system functionality signi?cantly affected user beliefs in various computer-related contexts (Igbaria,Gamers,&Davis,1995;Venkatesh&Davis,2000).For instance,research?ndings showed that speci?c system functionality is a critical factor that in?uences e-learning system usage(Hong,Thong,Wong,&Tam,2002;Pituch&Lee,2006).Pituch and Lee(2006)de?ned system functionality as the perceived ability of an e-learning system to provide?exible access to instructional and assessment media. Accordingly,we de?ne system functionality as the perceived ability of BELS to provide?exible access to instructional and assessment med-ia.Such media,for example,allows students to access course materials and content,turn in homework assignments,complete tests and quizzes online.

In general,content is used to identify various divergent formats and types of information.In this study,content refers to technology-based materials and course-related information that may provide value for learners in the context of BELS.BELS achieves its goals of shar-ing and delivering course content through various forms of media such as tutorials,online discussions,or web-based courses.Due to the diversity of delivery methods,it is a considerable issue that how to design and represent the hybrid content in appropriate formats or types best suited to delivery or access by BELS(So&Brush,2008).Appropriate BELS content features,as well as effective design,representing hybrid course content and transparent content knowledge transfer,are core components of BELS design(Piccoli et al.,2001).Drawing on the previous research(Zhang,Keeling,&Pavur,2000),we de?ne content feature as the characteristics and presentation of course con-tent and information in BELS.Text,hypertext,graphics,audio and video,computer animations and simulations,embedded tests,and multi-media information are some examples of content features in BELS environment.

System functionality and content feature have the potential to directly affect perceived usefulness of IS(Hong et al.,2002;Pituch&Lee, 2006)that are thought to be similar concepts in performance expectation.Several empirical evidences have argued that both content fea-tures(Zhang et al.,2000)and system functionality(Pituch&Lee,2006)affects the effectiveness of computer-mediated learning.That is to say,learners perceiving a higher level of system functionality and content features in BELS will lead to a higher level of performance expec-tations for BELS use.In addition,in the BELS environment,the diverse content features can be delivered and accessed depending upon the support of appropriate system functionality BELS facilitated(Pituch&Lee,2006;So&Brush,2008).Thus,we consider that the content fea-ture highly depends on the power and quality of system functionality of BELS.Therefore,the following hypotheses are proposed: H3:A higher level of system functionality of BELS will positively associate with a higher level of performance expectations for BELS use.

H4:A higher level of content features in BELS will positively associate with a higher level of performance expectations for BELS use.

H5:A higher level of system functionality in BELS will positively associate with a higher level of content features in BELS.

3.3.Social environment

In computer-mediated instructional design,there is an increasing focus on facilitating human interaction in the form of online collab-oration,virtual communities,and instant messaging in the BELS context(Graham,2006).From the group interactions perspective,social environment factors,such as collaborative learning(Francescato et al.,2006),learning climate(Chou&Liu,2005)and social interaction (Johnston et al.,2005)are important antecedents of beliefs about using an e-learning system.Prior research(Pituch&Lee,2006)shows that social interaction has a direct effect on the usage of an e-learning system.The interactions among students,between faculty and stu-dents and learning collaboration are the keys to learning process effectiveness.In addition,the emotional learning climate is an important indicator of learning effectiveness.

Interaction is de?ned in our study as the social interactions among students themselves,the interactions between instructors and stu-dents,and collaboration in a BELS environment.Learning climate is de?ned as the learning atmosphere in the BELS context.Johnston et al. (2005)argued that contact and interaction with instructors and learners is a valid predictor of performance.A positive learning climate encourages and stimulates the exchange of ideas,opinion,information,and knowledge in the organization that will lead to better learning satisfaction(Prieto&Revilla,2006).That is,when learners believe that BELS provides effective student-to-student and student-to-instruc-tor interactions and improves learning climate,they will be more satis?ed with BELS.Therefore,the following hypotheses are proposed: H6:A higher level of interaction will positively associate with a higher level of performance expectations for BELS use.

H7:A higher level of interaction will positively associate with a higher level of learning climate.

H8:A higher level of learning climate will positively associate with a higher level of learning satisfaction with BELS.

4.Method

4.1.Instrument development

To develop the self-report instrument,a number of prior relevant studies were reviewed to ensure that a comprehensive list of measures were included.All measures for each construct were taken from previously validated instruments and modi?ed based on the BELS context.

J.-H.Wu et al./Computers&Education55(2010)155–164159 For instance,the measures for learning satisfaction were selected from Chiu,Hsu,and Sun(2005)and Wu and Wang(2005).Measures for computer self-ef?cacy and performance expectations were taken from Compeau and Higgins(1995).The measures for content feature were adapted from Zhang et al.(2000)and Molla and Licker(2001).The measures for functionality were taken from Pituch and Lee (2006).The measures for student and instructor interactions were taken from Johnston et al.(2005),Kreijns,Kirschner,and Jochems (2003),and Pituch and Lee(2006).Finally,the measures for the learning climate were selected from Chou and Liu(2005).Supplementary material lists the de?nition of each construct,its measures,and the references.

The questionnaire consisted of two major parts including a portion for the respondent’s basic data and another for the responses to our research constructs.The basic data portion recorded the subject’s demographic information(e.g.,gender,age,highest education,computer experiences,and so forth).The second part recorded the subject’s perception of each variable in the model.It includes items for each con-struct.All items are measured via a7-point scale ranging from1(strongly disagree)to7(strongly agree).

Once the initial questionnaire was developed,an iterative personal interview process with professionals,instructors,and students from blended learning courses(including four instructors and?ve students from three different universities)was conducted to verify the com-pleteness,wording,and appropriateness of the instrument and to con?rm the content validity.Feedback from the interview processes served as the basis for correcting,re?ning,and enhancing the experimental scales.For example,scale items were eliminated if they rep-resented the same aspects with only slightly different wording and modi?ed if the semantics were ambiguous in order to enhance the psy-chometric properties of the survey instrument.At the end of the pre-test,there were seven constructs with21items in total to be used for the survey.

4.2.Participants

The empirical data were collected using a cross-sectional survey methodology.Participants for this study were students that had the opportunity to take courses via BELS.We distributed518paper-based and online questionnaires to target universities.The target univer-sities were purposively selected for the universities or colleges actually implemented BELS courses in Taiwan.Because of the applications of BELS are still at an early stage in Taiwan,the target universities are relatively rare.Data were collected via snowball and convenient sam-pling.Due to the conventional expectation of low survey response rates in survey studies,we endeavored to?nd a speci?c local contact person for each target university who was placed in charge of distributing the questionnaire.Three hundred and seven-six questionnaires were returned.Sixty-four responses were incomplete and had to be discarded.This left212valid responses for the statistical analysis,and a valid response rate of40.93%of the initial sample.Among the valid responses,84responses were received from physical classrooms and 128responses were gathered from online learning environments.The potential non-response bias was assessed by comparing the early versus late respondents that were weighed on several demographic characteristics.The results indicated that there were no statistically signi?cant differences among demographics between the early(the?rst semester)and late(the second semester)respondents.These re-sults suggest that non-response bias was not a serious concern.The respondent pro?les and the non-response bias analysis results are shown in Table1.

5.Results

Partial least squares(PLS)method was applied for the data analysis in this study.An analytical method is,in general,recommended for predictive research models emphasized on theory development,whereas Linear Structural Relationships(LISREL)is recommended for con-?rmatory analysis and requires a more stringent adherence to distributional assumptions(J?reskog&Wold,1982).PLS performs a Con?r-matory Factor Analysis(CFA).In a CFA,the pattern of loadings of the measurement items on the latent constructs was explicitly speci?ed in the model.The?t of this pre-speci?ed model is then examined to determine its convergent and discriminant validities.This factorial valid-ity deals with whether the loading patterns of the measurement items corresponds to the theoretically anticipated factors(Gefen&Straub, 2005).Convergent validity is shown when each measurement item correlates strongly with its assumed theoretical construct,while dis-criminant validity is shown when each measurement item correlates weakly with all other constructs except for the one to which it is the-oretically associated.The evaluation of the model?t was conducted in two stages(Chin,1998;Gefen&Straub,2005).First,the measurement validation was assessed,in which construct validity and reliability of the measures were assessed.The structural model with hypotheses was then tested.The statistical analysis strategy involved a two-phase approach in which the psychometric properties of all scales were?rst assessed through CFA and the structural relationships were then validated using bootstrap analysis.

5.1.Measurement validation

For the?rst phase,the analysis was performed in relation to the attributes of individual item reliability,construct reliability,average variance extracted(AVE),and discriminant validity of the indicators as measures of latent variables.The assessment of item loadings,reli-ability,convergent validity,and discriminant validity was performed for the latent constructs through a CFA.Re?ective items should be uni-dimensional in their representation of the latent variable and therefore correlated with each other.Item loadings should be above 0.707,showing that more than half of the variance is captured by the constructs.The results indicate that all items of the instrument had signi?cant loadings higher than the recommended value of0.707.As shown in Table2,all constructs exhibit good internal consistency as evidenced by their composite reliability scores.The composite reliability coef?cients of all constructs and the AVE in the proposed model (see Fig.1)are more than adequate,ranging from0.821to0.957and from0.605to0.849,respectively.

To assess discriminant validity:(1)indicators should load more strongly on their corresponding construct than on other constructs in the model and(2)the AVE should be larger than the inter-construct correlations(Chin,1998).AVE measures the variance captured by a latent construct,that is,the explained variance.For each speci?c construct,it shows the ratio of the sum of its measurement item variance as extracted by the construct relative to the measurement error attributed to its items.As a rule of thumb,the square root of the AVE of each construct should be larger than the correlation of the speci?c construct with any of the other constructs in the model(Chin,1998)and should be at least0.50(Fornell&Larcker,1981).As the results show in Table3,all constructs meet the above mentioned requirements.The

values for reliability are all above the suggested minimum of 0.7(Hair,Anderson,Tatham,&Black,1998).Thus,all constructs display ade-quate reliability and discriminant validity.All constructs share more variance with their indicators than with other constructs.Thus,the convergent and discriminant validity of all constructs in the proposed research model can be assured.

Table 1

Respondents pro?le and the results of non-response bias analysis (N =212).Variables

Classi?cation

Total (%)

Early

respondents (%)

Late

respondents (%)v 2(Sig.)

Gender Male 1060.500730.344330.1560.022(0.50)Female 1060.500720.340340.160Age

18–301010.476480.453530.500 1.344(0.855)

31–40820.387410.387410.38741–50230.108140.13290.08551–6040.01920.01920.019>6120.00910.00910.009Types of Jobs

Student

80.03830.01450.024 4.806(0.440)

Industry

300.142120.057180.085Manufacturing 570.269270.127300.142Service 100.04750.02450.024Finance 590.278360.170230.108Others

480.226230.108250.118Education level

Senior high school

00.00000.00000.0008.824(0.32)

College (2years)100.04710.00590.042University (4years)1160.547600.283560.264Graduate school

860.406450.212410.193BELS experience

Pure physical classroom experience 150.07170.03380.0380.371(0.946)

Pure virtual classroom experience 420.198200.094220.104Physical experience more than virtual experience

1050.495530.250520.245Virtual experience more than physical experience 500.236260.123240.113BELS experience:participating in BELS (years)

<0.5years 350.165180.085170.080 2.695(0.747)

0.5–1years 950.448500.236450.2122years 480.226250.118230.1083years 110.05260.02850.0244years 40.01920.00920.009>4years 190.09050.024140.066BELS experience:participating in BELS (times)1times 440.208240.113200.094 4.710(0.452)

2times 430.203220.104210.0993times 300.142150.071150.0714times 130.06190.04240.0195times 100.04760.02840.019P 6times 720.340300.142420.198BELS experience:spending time in the BELS (1week)<1h

620.29233

0.156290.137 4.729(0.450)

1–3h 750.354330.156420.1983–5h 430.203220.104210.0995–7h 200.094100.047100.0477–9h 60.02840.01920.009>9h

60.0284

0.019

0.009

Average years of computer usage experience

11.79(years)

13.7(years)

10.7(years)

27.076(0.133)

Table 2

Results of con?rmatory factor analysis.Construct

Items Composite reliability AVE Computer self-ef?cacy (CSE)30.8210.605System functionality (SF)30.9050.761Content feature (CF)20.8900.802Interaction (I)

30.9150.782Performance expectations (PE)30.9400.838Learning climate (LC)30.9260.807Learning satisfaction (LS)

0.957

0.849

160J.-H.Wu et al./Computers &Education 55(2010)155–164

J.-H.Wu et al./Computers&Education55(2010)155–164161

Table3

Correlation between constructs.

CSE SF CF PE I LC LS

CSE0.778a

SF0.5390.872

CF0.4920.6090.896

PE0.5270.5340.5960.916

I0.3890.5070.6080.6620.884

LC0.4250.5130.5930.7610.7270.898

LS0.440.5340.6010.7980.6140.740.921

a The shaded numbers in the diagonal row are square roots of the average variance extracted.

5.2.Hypotheses testing

In the second phase of the statistical analysis,the structural model was assessed to con?rm to what extent the relationships speci?ed by the proposed model were consistent with the available data.The PLS method does not directly provide signi?cance tests and path coef?-cient con?dence interval estimates in the proposed model.A bootstrapping technique was used to estimate the signi?cance of the path coef?cients.Bootstrap analysis was performed with200subsamples and the path coef?cients were re-estimated using each of these sam-ples.The parameter vector estimates was used to compute parameter means,standard errors,signi?cance of path coef?cients,indicator loadings,and indicator weights.This approach is consistent with recommended practices for estimating signi?cance of path coef?cients and indicator loadings(L?hmoeller,1984)and has been used in prior information systems studies(Chin&Gopal,1995;Hulland,1999).

Hypotheses and corollaries testing were performed by examining the size,the sign,and the signi?cance of the path coef?cients and the weights of the dimensions of the constructs,respectively.Results of the analysis for the structural model are presented in Fig.2.The esti-mated path coef?cient(standardized)and its associated signi?cance level are speci?ed next to each link.The R2statistic is indicated next to the dependent construct.The statistical signi?cance of weights can be used to determine the relative importance of the indicators in form-ing a latent construct.We found that all speci?ed paths between constructs in our research model had signi?cant path coef?cients.The results provide support for our model.

One indicator of the predictive power of path models is to examine the explained variance or R2values(Barclay,Higgins,&Thomson, 1995;Chin&Gopal,1995).R2values are interpreted in the same manner as those obtained from multiple regression analysis.They indicate the amount of variance in the construct that is explained by the path model(Barclay et al.,1995).The results indicate that the model ex-plained67.8%of the variance in learning satisfaction.Similarly,37.1%of the variance in content feature,55.1%of the variance in perfor-mance expectations and52.9%of the variance in learning climate were explained by the related antecedent constructs.The path coef?cient from computer self-ef?cacy to performance expectations is.229and from interaction to learning climate is0.727.The magni-tude and signi?cance of these path coef?cients provides further evidence in support of the nomological validity of the research model.Ta-ble4summarizes the direct,indirect,and total effects for the PLS analysis.

As for the cognitive factors,Hypotheses H1and H2,effectively drawn from computer self-ef?cacy to performance expectations and per-formance expectations to learning satisfaction are supported by the signi?cant path coef?cients,respectively.That is,students who had higher computer self-ef?cacy will have higher performance expectations,which in turn will lead to higher learning satisfaction.

As for the technological environment factors,with the signi?cant path coef?cients,the analysis results also provide support for the hypotheses H3and H4,effectively drawn from system functionality and content feature to performance expectations.In addition,Hypoth-esis H5,effectively drawn from system functionality to content feature is also supported by the signi?cant path coef?cients.However,it is interesting to note that the indirect effect of system functionality on performance expectations was stronger than its direct effect(see Table 4).This seems to indicate that system functionality alone may not be suf?cient for improving performance expectations when the BELS content features are not well-matched or designed.

162J.-H.Wu et al./Computers&Education55(2010)155–164

Table4

Standardized causal effects of PLS analysis.

Dependent latent variables Independent latent variables Standardized causal effects T-statistics

Direct Indirect Total

Content feature System functionality0.6090.60911.849*** Performance expectations Computer self-ef?cacy0.2290.229 3.717***

System functionality0.0920.1040.196 1.358*

Content feature0.1710.171 2.011**

Interaction0.4220.422 5.203*** Learning climate Interaction0.7270.72718.849*** Learning satisfaction Computer self-ef?cacy0.1280.128 3.693***

System functionality0.1090.109 2.307**

Content feature0.0950.095 1.802*

Interaction0.4650.4657.175***

Performance expectations0.5570.5577.006***

Learning climate0.3150.315 3.804***

*P<0.05.

**P<0.01.

***P<0.001.

As for the social environment factors,hypotheses H6and H7,the paths from interaction to performance expectations and learning cli-mate are supported.That is,interaction apparently in?uences the performance expectations and learning climate,respectively.Hypothesis H8,effectively drawn from learning climate to learning satisfaction is also supported by the signi?cant path coef?cients.That is,learning climate in?uences learning satisfaction.Overall,both performance expectations and positive learning climate have a direct effect on learn-ing satisfaction;performance expectations provide the greatest contribution(total effect)to learning satisfaction.

6.Conclusion

BELS environments have become the most prominent instructional delivery alternative when employed in e-learning systems.This study presents a theoretical model that was based on social cognitive theory for investigating the key determinants of student learning satisfaction in a BELS environment.The results provide strong evidence for the nomological validity of each construct and the effects on learning satisfaction,as shown in Fig.2.The estimate of0.551for the performance expectations construct(R2=55.1%)for these paths pro-vides good support for the hypothesized impact of computer self-ef?cacy,system functionality,content feature,and interaction on the dependent variable,performance expectations.In addition,the estimate of0.371for the content feature construct(R2=37.1%)for the path provides support for the hypothesized impact of system functionality on the content feature.The0.529estimate for the leaning climate construct(R2=52.9%)for these paths provides support for the hypothesized impact of interaction on the dependent variable,learning cli-mate.In addition,the0.678estimate for the learning satisfaction construct(R2=67.8%)denotes that the learning satisfaction as perceived by learners is directly and indirectly mediated by the performance expectations and learning climate.Therefore,as a whole,the model has strong explanatory power for the student learning satisfaction with BELS.

The signi?cant path coef?cients,effect size and the value of the R2reinforce our con?dence in the hypotheses testing results and provide support for the association with learning satisfaction in the BELS setting.The results demonstrated that the BELS learning satisfaction is affected by the interaction among cognitive,technological environment,and social environment factors.We con?rmed that technology alone does not cause learning to occur.It is consistent with the theoretical perspective of social cognitive theory:human behavior as a re-ciprocal interplay of cognitive factors,environment,and behavior(Bandura,1986).

J.-H.Wu et al./Computers&Education55(2010)155–164163 The empirical results indicate that performance expectations and learning climate are two strong determinants of learning satisfaction with BELS.The computer self-ef?cacy,system functionality,content feature,and interaction provided an indirect contribution to learning satisfaction via the above determinants.Thus,as students become more con?dent and capable of learning with BELS and more accustomed to the BELS learning environments,they will likely expect more bene?ts from the use of BELS,foster positive learning climate,and,overall, be more satis?ed with the BELS learning.These?ndings provide initial insights into those factors that are likely signi?cant antecedents for planning and implementing BELS to enhance student learning satisfaction.The contributions and implications of this study include the following:

A BELS environment should enhance students’performance expectations and foster positive learning climate.Our?ndings indicate that per-formance expectations provide the most contribution to learning satisfaction.This suggests that instructors should take advantage of BELS effectiveness in designing and teaching courses to enhance students’beliefs that they would be able to achieve improved outcomes with BELS.A positive learning climate signi?cantly affects students’learning satisfaction.This suggests that both instructors and learners should foster and motivate the positive learning atmosphere within the BELS learning context.Consequently,if students believe that using BELS is worthwhile,valuable and simple,they will be more likely to accept it resulting in greater satisfaction.

Education institutions should provide incentives and supports to enhance students’computer self-ef?cacy.The empirical results demonstrate that computer self-ef?cacy have a signi?cant positive in?uence on performance expectations.This implies that learners should have the computer competence necessary to exploit BELS and control over his/her learning activities.Therefore,educational institution administra-tors and instructors should provide suf?cient incentives and administrative supports to encourage students to actively participate in BELS courses and to enhance their computer self-ef?cacy.BELS should provide built-in help to?t various learners’needs in different learning circumstances.

BELS should offer appropriate system functionality and content features with multimedia presentation and?exibility.The results show that system functionality and content features have a positive in?uence on perceived expectations.These?ndings suggest that:(1)BELS should offer useful information with synchronous and asynchronous learning and content-rich design that satisfy students’needs;(2)BELS should provide various types of content presentation(e.g.,multimedia),customized functions to allow learners control over the system,and?ex-ible access to?t various students’learning requirements.It seems reasonable to note that education institutions may offer BELS-related technical training,awareness programs to the students to enhance students’comprehension of BELS.

BELS should provide effective interaction tools and instructors should motivate interaction publicly.The results demonstrate that participant interaction had a signi?cant positive in?uence on both performance expectations and learning climate.In addition,interaction has the most contribution(total effect)to the performance expectations.These?ndings suggest that when implementing BELS courses,the instruc-tors should motivate the positive interaction publicly to increase participant communication and collaborative learning via the system.In general,learning climate is a function and positive feedback of participant interaction in a BELS environment.A positive learning climate can make learning easy and natural.Thus,if BELS could support a good social environment to facilitate the students-to-student and stu-dent-to-instructor connectivity interaction(e.g.,interactive communication and collaborative learning),learners will be more likely to ac-tively participate in interaction,so as to foster better learning climate and to perceive greater BELS performance expectations and learning satisfactions.

Although our study provides insights into what determines student learning satisfaction in a BELS environment,it has several limita-tions that also represent opportunities for future research.First,the model was validated using sample data gathered from the target uni-versities in Taiwan.The fact that the participants come from one country limits the generalizability of the results.Other samples from different nations,cultures,and contexts should be gathered to con?rm and re?ne the?ndings of this study.Second,given the self-report instrument used(e.g.for measuring computer self-ef?cacy,system functionality,and content feature),therefore,the typical shortcomings associated with self-report measures must be recognized when interpreting the results.Third,this research sets a timely stage for future research in understanding the determinants of learning satisfaction in a BELS environment.It would be interesting to use a longitudinal design to examine the relationships among the identi?ed research variables might be a useful extension to the current study.Finally, the results cannot be exhaustive and future works should endeavor to uncover additional determinants of student learning satisfaction with BELS.

Appendix A.Supplementary material

Supplementary data associated with this article can be found,in the online version,at doi:10.1016/https://www.360docs.net/doc/3317219542.html,pedu.2009.12.012. References

Ajzen,I.,&Fishbein,M.(1980).Understanding attitudes and predicting social behavior.Englewood Cliffs,NJ:Prentice-Hall.

Bandura,A.(1986).Social foundations of thought and action:A social cognitive theory.Englewood Cliffs,NJ:Prentice Hall.

Barclay,D.,Higgins,C.,&Thomson,R.(1995).The partial least squares approach to causal modeling,personal computer adoption and use as an illustration.Technology Studies, 2,285–309.

Bolt,M.A.,&Koh,H.C.(2001).Testing the interaction effects of task complexity in computer training using the social cogitative model.Decision Sciences,32,1–19. Bonk,C.J.,Olson,T.M.,Wisher,R.A.,&Orvis,K.L.(2002).Learning from focus group:An examination of blended learning.Journal of Distance Education,17(3),97–118. Chin,W.W.(1998).The partial least squares approach to structural equation modeling.In G.A.Marcoulides(Ed.),Modern methods for business research(pp.298–336).

Mahwah,NJ:Erlbaum.

Chin,W.W.,&Gopal,A.(1995).Adoption intention in GSS:Relative importance of beliefs.The Data Base for Advances in Information Systems,26,42–63.

Chiu,C.M.,Hsu,M.H.,&Sun,S.Y.(2005).Usability,quality,value and e-learning continuance https://www.360docs.net/doc/3317219542.html,puters&Education,45,399–416.

Chou,S.W.,&Liu,C.H.(2005).Learning effectiveness in a web-based virtual learning environment:A learner control perspective.Journal of Computer Assisted Learning,21, 65–76.

Compeau,D.R.,&Higgins,C.A.(1995).Computer self-ef?cacy:Development of a measure and initial test.MIS Quarterly,19,189–211.

Compeau,D.R.,Higgins,C.A.,&Huff,S.(1999).Social cognitive theory and individual reactions to computing technology:A longitudinal study.MIS Quarterly,23,145–158. Davis,F.D.(1989).Perceived usefulness,perceived ease of use,and user acceptance of information technology.MIS Quarterly,13,318–339.

EL-Deghaidy,H.,&Nouby,A.(2008).Effectiveness of a blended e-learning cooperative approach in an Egyptian teacher education https://www.360docs.net/doc/3317219542.html,puters&Education,51, 988–1006.

Fornell,C.D.,&Larcker,F.(1981).Evaluating structural equation models with unobservable variables and measurement errors.Journal of Marketing Research,18(2),39–50.

164J.-H.Wu et al./Computers&Education55(2010)155–164

Francescato,D.,Porcelli,R.,Mebane,M.,Cuddetta,M.,Klobas,J.,&Renzi,P.(2006).Evaluation of the ef?cacy of collaborative learning in face-to-face and computer-supported university https://www.360docs.net/doc/3317219542.html,puters in Human Behavior,22,163–176.

Gefen,D.,&Straub,D.W.(2005).A practical guide to factorial validity using PLS-graph:Tutorial and annotated https://www.360docs.net/doc/3317219542.html,munications of the Association for Information Systems,16(5),91–109.

Graham,C.R.(2006).Chapter1:Blended learning system:De?nition,current trends,future directions.In C.J.Bonk&C.R.Graham(Eds.),Handbook of blended learning.San Francisco,CA:Pfeiffer.

Hair,J.F.,Anderson,R.E.,Jr.,Tatham,R.L.,&Black,W.C.(1998).Multivariate data analysis with readings(5th ed.).NJ:Prentice Hall.

Hara,N.,&Kling,R.(2000).Students’distress with a web-based distance education course:An ethnographic study of participants’https://www.360docs.net/doc/3317219542.html,rmation,Communication and Society,3,557–579.

Hong,W.,Thong,J.Y.L.,Wong,W.M.,&Tam,K.Y.(2002).Determinants of user acceptance of digital libraries:An empirical examination of individual differences and system characteristics.Journal of Management Information Systems,18(3),97–124.

Hulland,J.(1999).Use of partial least squares(PLS)in strategic management research:A review of four recent studies.Strategic Management Journal,20,195–204. Igbaria,M.,Gamers,T.,&Davis,G.B.(1995).Testing the determinants of microcomputer usage via a structural equation model.Journal of Management Information Systems, 11(4),87–114.

Jawahar,I.M.,&Elango,B.(2001).The effect of attitudes,goal setting and self-ef?cacy on ender user performance.Journal of End User Computing Practice,13(2),40–45. Johnston,J.,Killion,J.,&Oomen,J.(2005).Student satisfaction in the virtual classroom.The Internet Journal of Allied Health Sciences and Practice,3(2).

J?reskog,K.G.,&Wold,H.(1982).The ML and PLS techniques for modeling with latent variables:Historical and comparative aspects.In K.G.J?reskog&H.Wold(Eds.), Systems under indirect observation:Causality structure and prediction(pp.219–243).Amsterdam:North Holland.

Kerres,M.,&De Witt,C.(2003).A didactical framework for the design of blended learning arrangements.Journal of Educational Media,28(2/3),101–113.

Kinshuk,D.,&Yang,A.(2003).Web-based asynchronous synchronous environment for online learning.United States Distance Education Association Journal,17(2),5–17. Kreijns,K.,Kirschner,P.,&Jochems,W.(2003).Identifying the pitfalls for social interaction in computer-supported collaborative learning environments:A review of the https://www.360docs.net/doc/3317219542.html,puters in Human Behavior,19,335–353.

L?hmoeller,J.B.(1984).LVPS1.6program manual:Latent variable path analysis with partial least squares estimation.Universitaet zu Koehn,Zentralarchiv fuer Empirische Sozialforschung.

Maki,R.H.,Maki,W.S.,Patterson,M.,&Whittaker,P.D.(2000).Evaluation of a web-based introductory psychology course:Learning and satisfaction in on-line versus lecture courses.Behavior Research Models,Instruments,and Computers,32,230–239.

Marakas,G.M.,Yi,M.Y.,&Johnson,R.D.(1998).The multilevel and multifaceted character of computer self-ef?cacy:Toward clari?cation of the construct and an integrative framework for https://www.360docs.net/doc/3317219542.html,rmation Systems Research,9,126–162.

Martins,L.L.,&Kellermanns,F.W.(2004).A model of business school students’acceptance of a web-based course management system.Academy of Management Learning and Education,3,7–26.

Molla,A.,&Licker,P.S.(2001).E-commerce systems success:An attempt to extend and respecify the DeLone and McLean model of IS success.Journal of Electronic Commerce Research,2(4),1–11.

Osguthorpe,R.T.,&Graham,C.R.(2003).Blending learning environments:De?nitions and directions.The Quarterly Review of Distance Education,4(3),227–233.

Piccoli,G.,Ahmad,R.,&Ives,B.(2001).Web-based virtual learning environments:A research framework and a preliminary assessment of effectiveness in basic it skills training.MIS Quarterly,25,401–426.

Pituch,K.A.,&Lee,Y.(2006).The in?uence of system characteristics on e-learning https://www.360docs.net/doc/3317219542.html,puters&Education,47,222–244.

Prieto,I.M.,&Revilla,E.(2006).Formal and informal facilitators of learning capability:The moderating effect of learning climate.IE working paper.WP06-09.21-02-2006. Rovai,A.,&Jordan,H.(2004).Blended learning and sense of community:A comparative analysis with traditional and fully online graduate courses.The International Review of Research in Open and Distance Learning,5(2),1–12.

Santhanam,R.,Sasidharan,S.,&Webster,J.(2008).Using self-regulatory learning to enhance e-learning-based information technology https://www.360docs.net/doc/3317219542.html,rmation Systems Research, 19,26–47.

Shih,H.P.(2006).Assessing the effects of self-ef?cacy and competence on individual satisfaction with computer use:An IT student https://www.360docs.net/doc/3317219542.html,puters in Human Behavior, 22,1012–1026.

Singh,H.(2003).Building effective blended learning https://www.360docs.net/doc/3317219542.html,cational Technology,44(1),5–27.

So,H.J.,&Brush,T.A.(2008).Student perceptions of collaborative learning,social presence and satisfaction in a blended learning environment:Relationships and critical https://www.360docs.net/doc/3317219542.html,puters&Education,51,318–336.

Taylor,S.,&Todd,P.A.(1995).Decomposition and crossover effects in the theory of planned behavior:A study of consumer adoption intentions.International Journal of Research in Marketing,12,137–156.

Venkatesh,V.,&Davis,F.D.(2000).A theoretical extension of the technology acceptance model:Four longitudinal?eld studies.Management Science,46,186–204. Venkatesh,V.,Morris,M.G.,Davis,G.B.,&Davis,F.D.(2003).User acceptance of information technology:Toward a uni?ed view.MIS Quarterly,27,425–478.

Wood,R.,&Bandura,A.(1989).Social cognitive theory of organization management.Academy of Management Review,14,361–384.

Wu,J.H.,Tennyson,R.D.,Hsia,T.L.,&Liao,Y.W.(2008).Analysis of e-learning innovation and core capability using a hypercube https://www.360docs.net/doc/3317219542.html,puters in Human Behavior,24, 1851–1866.

Wu,J.H.,&Wang,S.C.(2005).What drives mobile commerce?An empirical evaluation of the revised technology acceptance https://www.360docs.net/doc/3317219542.html,rmation&Management,42,719–729. Yang,Z.,&Liu,Q.(2007).Research and development of web-based virtual on-line https://www.360docs.net/doc/3317219542.html,puters&Education,48,171–184.

Zhang,X.,Keeling,K.B.,&Pavur,R.J.(2000).Information quality of commercial website home pages:An explorative analysis.In Proceeding of the21th international conference on information systems(pp.164–175).Brisbane,Australia.