Article review

Internet and Higher Education 19 (2013) 10–17

Contents lists available at ScienceDirect

Internet and Higher Education

Unpacking online learning experiences: Online learning self-efficacy and learning satisfaction

Demei Shen a,⁎, Moon-Heum Cho b, Chia-Lin Tsai c, Rose Marra d

a Shanghai Engineering Research Center of Digital Education Equipment, East China Normal University, China b Lifespan Development and Educational Sciences, Kent State University at Stark, United States c Department of Psychological Sciences, University of Missouri – Columbia, United States d School of Information Science and Learning Technologies, University of Missouri – Columbia, United States

⁎ Corresponding author. Tel.: +86 1 206 524 3022. E-mail addresses: demeishen@gmail.com, demeishe

1096-7516/$ – see front matter © 2013 Elsevier Inc. All http://dx.doi.org/10.1016/j.iheduc.2013.04.001

a b s t r a c t

a r t i c l e i n f o

Article history: Accepted 8 April 2013 Available online 15 April 2013

Keywords: Online learning Self-efficacy Online learning self-efficacy Learning satisfaction

Self-efficacy is believed to be a key component in successful online learning; however, most existing studies of on- line self-efficacy focus on the computer. Although computer self-efficacy is important in online learning, researchers have generally agreed that online learning entails self-efficacy of multifaceted dimensions; therefore, one of the purposes of the current study was to identify dimensions of online learning self-efficacy. Through exploratory factor analysis, we identified five dimensions of online learning self-efficacy: (a) self-efficacy to complete an online course, (b) self-efficacy to interact socially with classmates, (c) self-efficacy to handle tools in a Course Management System (CMS), (d) self-efficacy to interact with instructors in an online course, and (e) self-efficacy to interact with class- mates for academic purposes. In addition, the role of demographic variables in online learning self-efficacy was in- vestigated. Demographic variables, such as the number of online courses taken, gender, and academic status were found to predict online learning self-efficacy. Furthermore, we found that online learning self-efficacy predicted stu- dents' online learning satisfaction. Results are discussed, and implications for online teaching and learning are provided.

© 2013 Elsevier Inc. All rights reserved.

1. Introduction

Beliefs about self-efficacy determine level of motivation as reflected in the amount of effort exerted in an endeavor and the length of time persisting in a difficult situation (Bandura, 1988). Self-efficacy is defined as “people's judgments of their capabilities to organize and execute a course of action required to attain designated types of performances” (Bandura, 1986, p. 391). If a person has a low level of self-efficacy toward a task, he or she is less likely to exert ef- fort; therefore, the person will less likely achieve. Other research find- ings have demonstrated that self-efficacy is a better predictor of academic achievement than any other cognitive or affective processes (Schunk, 1991); therefore, self-efficacy is critical in learning and per- formance (Hodges, 2008).

Student self-efficacy seems particularly important in challenging learning environments, such as an online learning environment where students lack the opportunity to interact with others and as a result can become socially isolated and easily lost (Cho & Jonassen, 2009; Cho, Shen, & Laffey, 2010). Recent studies have shown that the drop-out rate among students in online learning environments is higher than in traditional learning environments (Ali & Leeds, 2009). Some re- searchers have asserted that the drop-out rate is related in part to lack

n@yahoo.com (D. Shen).

rights reserved.

of self-efficacy (Lee & Choi, 2011). Researchers have argued that with the self-directed nature of online learning, self-efficacy can be a key component of academic success in distance education (Hodges, 2008); therefore, understanding self-efficacy in online learning is critical to im- prove online education. The current study was an investigation of self-efficacy in online learning settings.

2. Self-efficacy in online learning settings

Self-efficacy is context-specific (Bandura, 1986). In terms of online self-efficacy, we need to consider at least three areas: technology, learn- ing, and social interaction; however, a majority of researchers of online self-efficacy consider only the technological aspect of online learning. Consequently, self-efficacy in the other two areas has rarely been explored.

With regard to technology, numerous studies have been conducted on the role of technological self-efficacy in online student achievement. For instance, McGhee (2010) found a significant, moderate, and positive relationship between online technological self-efficacy and the academ- ic achievement of 45 community college students. Thompson and Lynch (2003) studied the psychological processes underlying resistance to web-based instruction (WBI) and demonstrated that students with weak Internet self-efficacy beliefs tended to resist WBI.

Regarding learning, Ergul (2004) showed that self-efficacy in dis- tance education significantly and positively predicted students' academic

11D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

achievement. In addition, Artino (2008) found that students with higher self-efficacy for computer-based learning are more likely to experience learning satisfaction than students with low self-efficacy.

In terms of social interaction, Cho and Jonassen (2009) found two di- mensions of online self-efficacy: self-efficacy to interact with instructors and self-efficacy to contribute to the online community. In addition, they found that students who have high self-efficacy in interacting with in- structors and contributing to the online community are more likely to use active interaction strategies, such as writing, responding, and reflecting. According to Cho and Jonassen researchers of online learning self-efficacy should consider diverse situations that can occur in online learning contexts, such as interacting with others through discussion or collaboration. Hodges (2008) claimed that “research on self-efficacy in online environments is in its infancy” (p. 10); in fact, how self-efficacy manifests in online learning contexts deserves additional research and studies. Although diverse learning settings are assumed, little empirical research on self-efficacy has been conducted with a focus on all three set- tings in online learning environments.

3. Variables of self-efficacy in online learning settings

Three types of variables relating to student self-efficacy in online learning environments are prior online learning experience, gender, and academic status. Existing empirical research on relationships among the three variables and self-efficacy shows different findings; therefore, the current empirical study contributes to filling the void.

3.1. Prior online experience and self-efficacy

Although little research has been conducted to investigate the rela- tionships between prior online learning experience and self-efficacy, a reasonable hypothesis is that the more students experience online, the more they are likely to have higher levels of online self-efficacy. An- other possible hypothesis is that prior online experience is not related to online self-efficacy. In their recent study, Cho and Kim (2013) found that the number of online courses students took is not related to their self-regulation for interaction with others. They viewed other factors, such as task structures for interaction and requirements for in- teraction, including quality and the number of online interaction may be associated with self-regulation for interaction with others. Although Cho and Kim's study is not directly related to online self-efficacy, their findings imply that prior online experience may not necessarily predict online self-efficacy. Because we have two reasonable but contrasting hypotheses and because little research has been done to investigate the relationship between online experience and self-efficacy, our re- search findings will contribute to the expansion of understanding that relationship.

3.2. Gender and self-efficacy

Gender difference in self-efficacy has been reported in many empir- ical studies. For example, Wesley (2002) studied 400 community col- lege students and found no significant difference in the self-efficacy of male and female students, but students 25 years old and older exhibited higher levels of self-efficacy than younger students. Li (2007), collecting data from 306 Taiwanese students at a technical col- lege, found that male students had higher level of general self-efficacy and computer self-efficacy than female students, and senior students had a higher level of both of the two types of self-efficacy than under- class students. Fletcher (2005) found that gender and previous online experience influence online learning self-efficacy, with female students having greater self-efficacy; however, more recently, Hung, Chou, Chen, and Own (2010) found no gender differences in computer or Internet self-efficacy or online communication self-efficacy. Because of mixed research results, more empirical study is necessary.

3.3. Academic status and self-efficacy

Billings, Skiba, and Connors (2005) compared differences between undergraduate and graduate nursing students who took web-based courses and found that undergraduates perceived higher levels of fac- ulty–student interactions than graduate students, and undergraduate reported higher levels of perceived connection with other students and instructor. Using the independent samples t test, Artino and Stephens (2009) compared undergraduate and graduate students en- rolled in several online courses with regard to their academic motiva- tion and self-regulation strategies. They found that undergraduates had more online learning experience, took a greater number of online courses, showed significantly greater levels of task value beliefs, and were more likely to continue to take online courses in the future than graduate students; graduate students showed significantly higher levels of critical thinking. They found no statistical differences between the two groups in self-efficacy beliefs. More empirical re- search will contribute to identifying relationships between academic status and online self-efficacy.

3.4. Students' satisfaction with online learning

Self-efficacy has been reported as a consistent variable in predicting students' learning satisfaction in online learning environments. Womble (2008), who investigated the relationship between e-learning self- efficacy and e-leaner satisfaction among 440 government agency em- ployees in training courses, found significant and positive correlation be- tween them. Lim (2001) examined the relationships among computer self-efficacy, academic self-concept, satisfaction, and future participation of adult distance learners. Findings indicated that computer self-efficacy was a significant predictor of both the satisfaction of online learners and their intention to take future web-based courses. Lin, Lin, and Laffey (2008) investigated students’ task value, self-efficacy, social ability and learning satisfaction. Among participants from 11 online courses in a dis- tance learning program, the researchers found that self-efficacy, task value, and social ability significantly impacted online learning satisfaction.

4. Research questions

The overarching research question in this study was designed to investigate the role of self-efficacy in online learning environments. More specifically, the following three research questions were exam- ined in this study.

1. What are the dimensions of online learning self-efficacy? 2. What variables are related to students’ online learning self-efficacy? 3. To what extent is self-efficacy related to students’ online learning

satisfaction?

5. Method

5.1. Participants

The participants in this study were students who were enrolled in an online course at the time the study was conducted. Response rate was not calculated because students were not required to report their course information. A total of 406 online students participated in the study. Among them, 301 (74.1%) students were female, and 104 (25.6%) students were male. The majority (N = 351, 86.5%) of the participants were Caucasian. More than 50% of the participants were in pursuit of a graduate degree (N = 244, 60.1%), but undergraduates were also included in the pool of respondents (N = 151, 37.2%). See Table 1 for detailed demographic information.

Table 1 Description of participants.

Demographic variables N %

Gender Male 104 25.6 Female 301 74.1 No response 1 .2

Ethnicity/citizenship African/African American 13 3.2 Asian, Asian American, or Pacific Islander 17 4.2 Latino/Latino American 14 3.4 Caucasian/Caucasian American 351 86.5 Biracial 9 2.2 No response 2 .5

Degree pursuing Undergraduate 151 37.2 Graduate 244 60.1 Other 11 2.7

Total 406

12 D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

5.2. Measures

Two instruments were used for this study: one for identifying di- mensions of online learning self-efficacy and the other for measuring online learning satisfaction.

5.3. Demographic variables

Participants were asked to self-report demographic information, including gender, academic status, and number of online courses taken. Gender was a two-category variable with male coded as 1 and female coded as 0. Academic status was reported as graduate stu- dent or undergraduate student. Participants were asked to provide the number of online courses they had taken up to the time of the survey.

5.4. Online learning self-efficacy

A new scale was developed to measure students' online learning self-efficacy. Following a literature review, online learning self-efficacy was conceptualized into six types of self-efficacy: (a) self-efficacy to complete an online course, (b) self-efficacy to interact with classmates, (c) self-efficacy to interact with an instructor, (d) self-efficacy to self-regulate in online learning, (e) self-efficacy to handle a course man- agement system, and (f) self-efficacy to socialize with classmates. An initial item pool consisting of 120 items was generated to assess six types of online learning self-efficacy. Each item was evaluated through an expert review conducted by five doctoral students and two faculty members with extensive experience in online teaching and scale devel- opment. Each of the experts received six review forms and rated each item in terms of relevance to one type of online learning self-efficacy, for example, self-efficacy to complete an online course, based on a scale of 1 through 3, where 1 was “not relevant at all,” 2 indicated “rel- evant but needs minor alteration,” and 3 was “very relevant.” They were also asked to write comments about the items. The research group discussed and compiled the review results and then selected and re- vised 35 items as the final scale. An 11-point Likert scale was used. Par- ticipants were asked to report how confident they were when engaging in various actions in an online course. They responded on a scale of 0 to 10, where 0 indicated “cannot do at all,” 5 indicated “moderately confi- dent can do,” and 10 indicated “highly confident can do.” High scores in- dicated higher levels of online learning self-efficacies.

5.5. Learning satisfaction

Learning satisfaction was measured with five items on a scale of 1 to 5, where 1 denoted “strongly disagree” and 5 denoted “strongly

agree.” They were adapted from Lin’s (2005) research. Cronbach's alpha for learning satisfaction in this study was .88.

5.6. Procedure

The data were collected from two universities in the Midwestern US. The authors of the current study contacted online instructors and asked for permission to conduct the study in their online courses. With their approval the researchers posted the recruiting letter and link to the on- line survey on a message board. The instructors also encouraged stu- dents to participate in the study. After students filled out the online consent form, they were directed to fill out the online survey on the website. This study was approved by the IRB and conducted ethically.

6. Results

6.1. Question 1: what are the dimensions of online learning self-efficacy?

An exploratory factor analysis was conducted to identify dimen- sions of online learning self-efficacy. The principle axis factoring (PAF) extraction method with oblique rotation was applied (in this case Promax). We chose an oblique rotation method because each self-efficacy dimension was assumed to be correlated with one anoth- er. Four criteria were used to arrive at the solution. First, item load- ings should be above .40. Second, any discrepancies between cross-loadings with an absolute value less than .15 were deleted; third, no item was cross-loaded highly to more than one factor, for example, greater than .40. And fourth, each factor should have at least three items (Pett, Lackey, & Sullivan, 2003). Our results sug- gested five factors of online learning self-efficacy. The Kaiser– Meyer–Olkin measure of sampling adequacy (KMO) value was .947, and the chi-square value of Bartlett's Test of Sphericity was x2

(595) = 15121.91, with p b .001. Both of the tests indicated a high level of appropriateness for a factor analysis to proceed.

The five factors explained 74.20% of the total variance, and each fac- tor explained 50.62%, 10.23%, 5.95 %, 4.22%, and 3.18% of the variance, respectively. In accordance with the items included in the factors, we named the five factors as follows: (a) self-efficacy to complete an online course, (b) self-efficacy to interact socially with classmates, (c) self- efficacy to handle tools in a Course Management System (CMS), (d) self-efficacy to interact with instructors in an online course, and (e) self-efficacy to interact with classmates for academic pur- poses. Cronbach's alpha for each dimension of online learning self-efficacy was .93, .92, .93, .94, and .93, respectively. See Table 2 for more detailed information about factor loadings, items, and Cronbach's alpha.

6.2. Question 2: what variables are related to students' online self-efficacy?

In order to answer Research Question 2, descriptive statistics and multiple regressions were conducted using IBM SPSS Statistics 20.

6.2.1. Descriptive statistics of variables Descriptive statistics, including mean and standard deviations,

were acquired and have been presented in Table 3. Information about how many online courses students took prior to participating in the study was also collected. Students reported that on average they took 5.43 online courses (SD = 3.89). Students tended to have high levels of self-efficacy beliefs, and the mean value of all the five self-efficacy factors were above 7 out of 10. Their learning satisfaction score was high as well, with a mean value of 4.32. The correlation of factors indicated all the dimensions of online learning self-efficacy significantly (ranged from r = .295 to r = .749) correlated to one an- other, and they all significantly correlated to learning satisfaction (ranged from r = .320 to r = .562). To conduct the following analy- ses, gender was treated as a dichotomous variable, with 0 indicated

Table 2 Results of exploratory factor analysis.

Factor loadings

1 2 3 4 5

Factor 1: Self-efficacy to complete an online course (Eigenvalue = 17.72; 50.62% of variance explained, alpha = .93)

How confident are you that you could do the following tasks in the Online Course? Complete an online course with a good grade .849 −.034 .036 .065 −.126 Understand complex concepts .867 .101 −.123 .156 −.233 Willing to face challenges .684 .104 −.041 .197 −.047 Successfully complete all of the required online activities .874 −.034 −.038 .072 −.043 Keep up with course schedule .808 −.008 −.096 −.110 .211 Create a plan to complete the given assignments .516 .039 .185 −.013 .130 Willingly adapt my learning styles to meet course expectations .629 .097 .025 −.007 .199 Evaluate assignments according to the criteria provided by the instructor .551 .103 .235 .051 .026

Factor 2: Self-efficacy to interact socially with classmates (Eigenvalue = 3.58, 10.23% of variance explained, alpha = .92)

How confident are you that you could do the following social interaction tasks with your CLASSMATES in the ONLINE course?

Initiate social interaction with classmates −.022 .812 .045 −.010 .067 Socially interact with other students with respect −.120 .567 .095 .164 .187 Develop friendship with my classmates .099 .860 −.045 −.094 −.058 Apply different social interaction skills depending on situations −.017 .899 .057 −.020 .024 Pay attention to other students’ social actions .081 .945 .033 −.093 −.084

Factor 3: Self-efficacy to handle tools in a CMS (Eigenvalue = 2.08; 5.95% of variance explained, alpha = .93)

How confident are you that you could use the following tasks while using online course TOOLS in the course management system?

Download instructional materials .061 −.075 .717 .032 .087 Post a new message in a discussion board −.128 −.043 .997 .011 .009 Reply to others' message in a discussion board −.110 −.063 .912 .048 .018 Submit assignments .094 .058 .767 .041 −.102 Open files within the course management system .068 .055 .822 .040 −.017 Send email to others with or without attached files .022 .089 .784 −.055 −.105

Factor 4: Self-efficacy to interact with instructors in an online course (Eigenvalue = 1.48; 4.22% of variance explained, alpha = .94)

How confident are you that you could do the following tasks while interacting with your INSTRUCTOR in the ONLINE Course?

Clearly ask my instructor questions .029 −.070 .033 .958 −.041 Timely inform the instructor when unexpected situations arise −.028 −.026 .180 .658 .101 Initiate discussions with the instructor .014 .032 .002 .768 .123 Express my opinions to instructor respectfully .198 −.056 −.071 .721 .101 Seek help from instructor when needed .058 .036 −.008 .902 −.074

Factor 5: Self-efficacy to interact with classmates for academic purposes (Eigenvalue = 1.11; 3.18% of variance explained, alpha = .93)

How confident are you that you could do the following tasks while interacting with your CLASSMATES in the ONLINE course?

Actively participate in online discussions .131 −.073 .020 .060 .737 Effectively communicate with my classmates −.029 .218 −.033 .072 .722 Express my opinions to other students respectfully .028 .005 .022 .341 .521 Respond to other students in a timely manner .033 .097 −.036 .063 .751 Provide help to other students when assistance is needed .015 .307 −.085 .000 .684 Request help from others when needed −.146 .301 −.030 .170 .582

Note: the bold emphasis indicates the loading value of an item and that the item has loaded on the identified factor.

Table 3 Descriptive statistics of variables.

Variables 1 2 3 4 5 6 7 8 9

Academic status – Gender −.13 – Number of Online Courses .339⁎⁎ −.090 – Factor 1: Self-efficacy to complete an online course .116⁎ −.167⁎⁎ .205⁎⁎ – Factor 2: Self-efficacy to interact socially with classmates .037 −.060 .074 .452⁎⁎ – Factor 3: Self-efficacy to handle tools in a CMS .150⁎⁎ −.146⁎⁎ .118⁎ .596⁎⁎ .295⁎⁎ – Factor 4: Self-efficacy to interact with instructors in an online course .037 −.099⁎ .115⁎ .675⁎⁎ .550⁎⁎ .538⁎⁎ – Factor 5: Self-efficacy to interact with classmates for academic .066 −.145⁎⁎ .144⁎⁎ .705⁎⁎ .671⁎⁎ .529⁎⁎ .749⁎⁎ – Learning Satisfaction −.079 −.146⁎⁎ .038 .562⁎⁎ .377⁎⁎ .320⁎⁎ .506⁎⁎ .428⁎⁎ – M 5.43 9.08 7.53 9.68 9.17 8.85 4.32 SD 3.89 1.21 2.36 0.80 1.35 1.51 0.80

⁎ p b .05. ⁎⁎ p b .01.

13D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

14 D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

“female,” and 1 represented “male.” Similarly, we used 0 to represent “undergraduate”, and 1 to indicate “graduate”. Gender and number of online courses both weakly correlated with four self-efficacy factors except self-efficacy to interact socially with classmates. Table 3 shows the result of descriptive statistics.

6.2.2. Multiple regressions To investigate how demographic variables are associated with stu-

dents' online learning self-efficacy, five separate multiple regressions were performed. The demographic variables included the number of online courses students took prior to participating in the study as well as gender and academic status.

6.2.3. Self-efficacy to complete an online course The number of online courses, gender, and academic status together

accounted for approximately 6.5% of the variance in self-efficacy to com- plete an online course (Radj

2 = .065), F(3, 391) = 10.07, p b .001. The number of online courses was a significant predictor of self-efficacy to complete an online course, t (391) = 3.48, p b .01, which accounted for 3% of the variance in self-efficacy to complete an online course not accounted for by other self-efficacy factors (pr = .173) and uniquely accounted for 3% of the variance in self-efficacy to complete an online course (sr = .170). Holding other variables constant, we found that as one more online course was taken by students, their self-efficacy to com- plete an online course was estimated to increase by about .06 point (95% CI: .03, .09, Beta = .18).

Gender was also a significant predictor of self-efficacy to complete an online course, t(391) = −3.13, p b .01, which accounted for 2% of the variance in self-efficacy to complete an online course not accounted for by other self-efficacy factors (pr = −.156) and uniquely accounted for 2% of the variance in self-efficacy to complete an online course (sr = −.152). Holding other variables constant, we found that males had .43 point less than their female counterparts (95% CI: −.69, −.16; Beta = −.15).

Academic status was not a significant predictor of self-efficacy to complete an online course, t (391) = 1.07, p > .05.

6.2.4. Self-efficacy to interact socially with classmates The number of online courses, gender, and academic status did not

significantly predict self-efficacy to interact socially with classmates, F (3, 391) = 1.01, p > .05.

6.2.5. Self-efficacy to handle tools The number of online courses, gender, and academic status together

significantly predicted self-efficacy to handle tools in a CMS, and accounted for approximately 4.2% of the variance (Radj

2 = .042), F(3, 391) = 6.75, p b .001.

The number of online courses was not a significant predictor of self-efficacy to handle tools in a CMS, t(391) = 1.25, p > .05.

Gender was a significant predictor of self-efficacy to handle tools in a CMS, t(391) = −2.85, p b .01, which accounted for 2% of the variance in self-efficacy to handle tools in a CMS not accounted for by other self-efficacy factors (pr = −.143) and uniquely accounted for 2% of the variance in self-efficacy to handle tools in a CMS (sr = − .141). Holding other variables constant, we found that males had .26 point less than their female counterparts (95% CI: − .44, − .08; Beta = − .14).

Academic status was a significant predictor of self-efficacy to handle tools in a CMS, t(391) = 2.48, p b .05, which accounted for 2% of the variance in self-efficacy to handle tools in a CMS not accounted for by other self-efficacy factors (pr = .125) and uniquely accounted for 1% of the variance in self-efficacy to handle tools in a CMS (sr = .122). Holding other variables constant, we found that graduate students had .22 point more than undergraduate students in self-efficacy to han- dle tools (95% CI: .05, 39; Beta = .13).

6.2.6. Self-efficacy to interact with instructors in an online course The number of online courses, gender, and academic status to-

gether significantly predicted self-efficacy to interact with instructors in an online course and accounted for approximately 1.6% of the var- iance (Radj

2 = .016), F(3, 391) = 3.16, p b .05. Number of online courses was not a significant predictor of self-efficacy to interact with instructors, t (391) = 1.94, p > .05.

Gender was a significant predictor of self-efficacy to interact with instructors in an online course in a CMS, t(391) = −2.07, p b .05, which accounted for 1% of the variance in self-efficacy to interact with instructors in an online course not accounted for by other self-efficacy factors (pr = −.104) and uniquely accounted for 1% of the variance in self-efficacy to interact with instructors in an online course(sr = −.104). Holding other variables constant, we found that males had .32 point less than their female counterparts (95% CI: −.62, −.02; Beta = −.10) in their self-efficacy to interact with instructors in an online course.

Academic status was not a significant predictor of self-efficacy to interact with instructors in an online course, t(391) = .005, p > .05.

6.2.7. Self-efficacy to interact with classmates for academic purposes The number of online courses, gender, and academic status together

accounted for approximately 3.8% of the variance in self-efficacy to in- teract with classmates for academic purposes (Radj

2 = .038), F(3, 391) = 5.21, p b .01.

The number of online courses was a significant predictor of self- efficacy to interact with instructors, t(391) = 2.23, p b .05, which accounted for 1% of the variance in self-efficacy to interact with class- mates for academic purposes not accounted for by gender (pr = .112) and uniquely accounted for 1% of the variance in self-efficacy to interact with classmates for academic purposes (sr = .111). Holding other vari- ables constant, we found that as one more online course was taken by stu- dents, their self-efficacy to interact with classmates for academic purposes was estimated to increase by about .05 point (95% CI: .01, .09, Beta = .12).

Gender was also a significant predictor of self-efficacy to interact with classmates for academic purposes in a CMS, t(391) = −2.81, p b .01, which accounted for 2% of the variance in self-efficacy to inter- act with classmates for academic purpose not accounted for by gender (pr = −.141) and uniquely accounted for 2% of the variance in self-efficacy to interact with classmates for academic purposes (sr = −.140). Holding the number of online courses constant, males had an average of.48 point less than their female counterparts (95% CI: −.81, −.14; Beta = −.14) in their self-efficacy to interact with classmates for academic purposes.

Academic status was not a significant predictor of self-efficacy to interact with instructors in an online course, t (391) = .30, p > .05. Please see Table 4 for more information about the regression results.

6.3. Question 3: to what extent is self-efficacy related to students' online learning satisfaction?

Multiple regression analysis was performed using the five self- efficacy factors as independent variables and learning satisfaction as dependent variable to detect whether the five self-efficacy belief fac- tors predict learning satisfaction (Table 5).

The five self-efficacy factors together accounted for approximately 35% of the variance in learning satisfaction (Radj

2 = .352), F (2, 59) = 28.43, p b .001. Self-efficacy to complete an online course was a significant predictor of lea rning satisfaction, t (392) = 7.28, p b .001, which accounted for 12% of the variance in learning satisfac- tion not accounted for by other self-efficacy factors (pr = .345) and uniquely accounted for 9% of the variance in learning satisfaction (sr = .294). Holding other self-efficacy factors constant, as self- efficacy to complete an online course increased by 1 point, learning

Table 4 Multiple regression analysis results for self-efficacy factors.

Variable B SE. β pr Sr 95% C.I. T Sig.

Lower Upper

Self-efficacy to complete an online course (Constant) 8.78 .12 Number of Online Courses .06 .02 .18 .173 .170 .03 .09 3.48⁎⁎ .001 Gender −.43 .14 −.15 −.156 −.152 −.69 −.16 −3.13⁎⁎ .002 Academic status .14 .13 .06 .054 .052 −.12 .40 1.07 .285

Self-efficacy to interact socially with classmates (Constant) 7.54 .24 Number of Online Courses .03 .03 .05 .051 .051 −.03 .09 1.00 .316 Gender −.34 .26 −.07 −.065 −.065 −.85 .18 −1.29 .197 Academic status .03 .25 −.01 −.006 −.006 −.52 .47 −.11 .910

Self-efficacy to handle tools in a CMS (Constant) 9.54 .08 Number of Online Courses .01 .01 .07 .063 .061 −.01 .04 1.25 .214 Gender −.26 .09 −.14 −.143 −.141 −.44 −.08 −2.85⁎⁎ .005 Academic status .22 .09 .13 .125 .122 .05 .39 2.48⁎ .013

Self-efficacy to interact with instructors in an online course (Constant) 9.06 .14 Number of Online Courses .04 .02 .10 .098 .097 .00 .07 1.94 .053 Gender −.32 .15 −.10 −.104 −.104 −.62 −.02 −2.07⁎ .039 Academic status .00 .15 .00 .00 .00 −.29 .29 .01 .996

Self−efficacy to interact with classmates for academic purpose (Constant) 8.72 .15 Number of Online Courses .50 .02 .12 .112 .111 .01 .09 2.23⁎⁎ .026 Gender −.48 .17 −.14 −.141 −.140 −.81 −.14 -2.81⁎⁎ .008 Academic status .05 .16 .02 015 015 .27 .37 .30 .764

⁎ p b .05. ⁎⁎ p b .01.

15D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

satisfaction was estimated to increase by about .31 point (95% CI: .22, .40, Beta = .46).

Self-efficacy to interact socially with classmates was a significant predictor of learning satisfaction, t (392) = 2.68, p b .01, which accounted for 2% of the variance in learning satisfaction not accounted for by other self-efficacy factors (pr = .134) and uniquely accounted for 1% of the variance in learning satisfaction (sr = .108). Holding other self-efficacy factors constant, as self-efficacy to interact socially with classmates increased by 1 point, learning satisfaction was estimat- ed to increase by about .05 point (95% CI: .01, .09, Beta = .15).

Self-efficacy to interact with instructors in an online course was a significant predictor of learning satisfaction, t (392) = 3.96, p b .001, which accounted for 4% of the variance in learning satisfaction not accounted for by other self-efficacy factors (pr = .196) and uniquely accounted for 3% of the variance in learning satisfaction (sr = .160). Holding other self-efficacy factors constant, as self-efficacy to interact with instructors in an online course increased by 1 point, learning satis- faction was estimated to increase by about .16 point (95% CI: .08, .23, Beta = .26).

Table 5 Multiple Regression Analysis Results for Learning Satisfaction.

Variable B SE.

(Constant) 1.04 .39 Self-efficacy to complete an online course .31⁎⁎⁎ .04 Self-efficacy to interact socially with classmates .05⁎⁎ .02 Self-efficacy to handle tools in a CMS −.06 .05 Self-efficacy to interact with instructors in an online course .16⁎⁎⁎ .04 Self-efficacy to interact with classmates for academic purpose −.09⁎ .04

Note. R Square = .360. Adjusted R Square = .352. ⁎ p b .05.

⁎⁎ p b .01. ⁎⁎⁎ p b .001.

Self-efficacy to interact with classmates for academic purpose was a significant predictor of learning satisfaction, t (392) = −2.16, p b .05, which accounted for 1% of the variance in learning satisfaction not accounted for by other self-efficacy factors (pr = −.109) and uniquely accounted for 1% of the variance in learning satisfaction (sr = -.087). Holding other self-efficacy factors constant, as self-efficacy to inter- act with classmates for academic purpose increased by 1 point, learning satisfaction was estimated to decrease by about .09 point (95% CI: −.17, − .01, Beta = − .16).

Self-efficacy to handle tools in a CMS was not a significant predic- tor of learning satisfaction, t (392) = −1.08, p > .05.

7. Discussion

7.1. Multiple dimensions of online self-efficacy

One of the purposes of this study was to investigate dimensions of self-efficacy in online learning. Through the exploratory factor analy- sis, we found five dimensions of online learning self-efficacy. This

β pr sr 95% C.I. T Sig.

Lower Upper

.46 .345 .294 .22 .39 7.28⁎⁎⁎ .000

.15 .134 .108 .01 .09 2.68⁎⁎ .008 −.06 −.054 −.044 −.16 .05 −1.08 .281

.26 .196 .160 .08 .23 3.96⁎⁎⁎ .000 −.16 −.109 −.087 −.17 −.01 −2.16⁎ .031

16 D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

result indicates that online learning self-efficacy is multidimensional and reflects the complex, multifaceted situation of online learning. Dif- ferent from much online self-efficacy research that focused solely on one aspect of learning setting such as technology, the current online self-efficacy study involved three aspects of online contexts, including technology, learning, and social interaction. The five dimensions of the self-efficacy scale may contribute to understanding multifaceted self- efficacy in online learning environments.

Interestingly enough, different from our assumption that “self- efficacy in self-regulation” is a separate online self-efficacy, “self- efficacy in self-regulation” did not emerge as a separate factor, at least not with our data set. The three items, which were created to assess self-efficacy for self-regulated learning, loaded to self-efficacy to com- plete an online course. These three items included the following: Create a plan to complete the given assignments, willingly adapt my learning styles to meet course expectations, and evaluate assignments according to the criteria provided by the instructor. One possible explanation is that students might have interpreted self-regulation skills, such as plan- ning, monitoring, and evaluation, as tactics leading to complete an on- line course. Future researchers should restate each item to measure self-regulation aspect of self-efficacy.

7.2. Variables related to self-efficacy in online settings

Three variables including gender, online experience, and academic status were related to online learning self-efficacy to some extent. First, gender was a significant predictor of all the self-efficacy beliefs ex- cept self-efficacy to interact socially with classmates. In general, the re- sults demonstrate that female students were likely to have higher online learning self-efficacy than male students, implying that female students may be more active, seek more help, or function better than male students. Our results are consistent with Gebara's study (2010), demonstrating that female students reported higher level of online self-efficacy than male students.

Second, online experience measured with the number of online courses was a significant predictor for two self-efficacy beliefs: self- efficacy to complete an online course and self-efficacy to interact with classmates for academic purposes. This finding indicates that the stu- dents who took more online courses were more likely to have higher online learning self-efficacy to complete an online course; in addition, they were more likely to communicate and collaborate with other stu- dents on academic tasks. However, online experience was not signifi- cantly related to self-efficacy to interact socially with classmates, self-efficacy to handle tools in a CMS, and self-efficacy to interact with instructors in an online course.

Last, academic status was not related with most of the dimensions of online learning self-efficacy, which was consistent with other studies; for example, Artino and Stephens (2009) found no significant difference in self-efficacy between undergraduates and graduates. In the current study, academic status predicted self-efficacy to handle tools in a CMS only; in other words, graduate students tended to have higher levels of technological self-efficacy than undergraduate students perhaps be- cause graduate students had more experience with online learning technology and perhaps because more graduate level courses were de- livered online than undergraduate courses. This was verified by the number of online courses taken by undergraduate and graduate stu- dents. An independent sample t-test on number of online courses by ac- ademic status revealed that graduate students had taken significantly more online courses (M = 6.48, SD = 3.80) than undergraduates (M = 3.76, SD = 3.42), t (394) = 2.72, p b .001.

Overall, gender, the number of online courses, and academic status together did not explain much of the variance in the dimensions of on- line learning self-efficacy (R squared less than 7%), showing that other reasons may play vital roles in development of self-efficacy beliefs in online contexts. Perhaps, what really influences to students' online self-efficacy is quality of learning experience instead of demographic

information, including gender and academic status or the number of online courses students took. Recently, Cho and Kim (2013) found that instructors' scaffolding efforts to facilitate social interaction among students or between students and teacher are most significantly related to students' self-regulation for interaction with others than any other variables, such as demographic variables and the number of on- line courses in online learning settings. This shows that if students per- ceived that they received quality support from online instructors, the students might engage in more social interaction with others, perhaps leading them to develop high self-efficacy for social interactions regard- less the number of course she or he takes.

7.3. Self-efficacy and learning satisfaction

The five dimensions of online learning self-efficacy except self- efficacy to handle tools in a CMS significantly predicted students' online learning satisfaction. Among the four significant dimensions of self- efficacy, self-efficacy to complete an online course was most significantly associated with learning satisfaction; and it explained approximately 12% of variance in satisfaction. However, the power of remaining self-efficacies that explain variance in satisfaction is marginal. Self- efficacy to interact socially with classmates, self-efficacy to interact with instructors in an online course, and self-efficacy to interact with classmates for academic purpose explain variance in satisfaction only 2%, 4%, and 1%, respectively. The results show that students' self- assessment about their capabilities to complete an online course is more important in explaining satisfaction with online learning than any other self-efficacies.

7.4. Implications for online teaching

The five dimensions of online self-efficacies were identified: self-efficacy to complete an online course, self-efficacy to interact so- cially with classmates, self-efficacy to handle tools in a CMS, self- efficacy to interact with instructors in an online course, and self- efficacy to interact with classmates for academic purposes. Several implications for online teaching follow.

7.4.1. Scaffold students to participate in learning activities Self-efficacy to complete an online course explains most of the vari-

ance both in online self-efficacy and in learning satisfaction. Online in- structors can support students' course completion by providing co- regulation opportunities, for example, online instructors' regular moni- toring on students' course participation and providing support and encouragement to complete an online course (Shea, Li, & Pickett, 2006). Through regular monitoring of students' activities, online in- structors can find students' slow participation or failure to submit as- signments, especially for less experienced online students. Then, online instructors can provide immediate support and guidance for stu- dents to participate in the activities or complete the tasks.

7.4.2. Promote social interaction with others The current study shows that three dimensions of online self-efficacy

are related to social interactions among students and between students and instructors. The nature of online learning requires students to inter- act actively with both instructors and classmates. Especially those stu- dents with less experience may experience anxiety about interacting with others and may feel social isolation if they perceive lack of support from others. Online literature suggests that instructors should create so- cial presence and teaching presence to foster a sense of learning commu- nity (Yang, Tsai, Kim, Cho, & Laffey, 2006). Possible examples to promote social interactions with others include instructors' direct interactions ef- forts, such as participating in discussion boards (Artino & Stephens, 2009), providing guidelines for social interaction, recognizing students' contribution to online learning community (Shea et al., 2006), and mon- itoring students' social interaction processes (Cho & Kim, 2013).

17D. Shen et al. / Internet and Higher Education 19 (2013) 10–17

7.4.3. Provide an orientation to enhance students' self-efficacy to handle tools in a CMS

Although we found that self-efficacy to handle tools in a CMS was not a significant predictor for learning satisfaction, we found that self-efficacy to handle tools in a CMS is still a dimension of online self-efficacy. In addition, a significant number of studies consistently reported that technology-related self-efficacy promotes student mo- tivation and learning. In particular, self-efficacy to handle tools in a CMS is important for new online learners or at the beginning of the class (Cho, 2012). An orientation program will help students develop self-efficacy to handle tools in a CMS. Possible examples of orientation include scavenger activities that engage students with tools in a CMS and a video tutorial that demonstrates the use of tools in a CMS.

7.4.4. Consider gender differences in online self-efficacy Our study findings demonstrate gender differences in all dimensions

of online self-efficacy except for self-efficacy to interact socially with classmates. Female students have significantly higher self-efficacy than male students. Online instructors may need to provide additional support for male students to help them develop online self-efficacy. Possible instructional strategies include paying extra attention to male students' learning processes, providing immediate feedback and assis- tance, supporting them in the completion of tasks, and encouraging them to interact with others by sending an individual note or recogniz- ing their contributions to the development of an online learning community.

8. Conclusion

Considering three aspects of dynamic online learning environ- ments – technology, learning, and social interaction – we explored the dimensions of online learning self-efficacy and found that online self-efficacy involves five dimensions. Although more research is nec- essary to test its validity with larger samples from multiple online courses, our study provides a reliable instrument that can be used to measure diverse aspects of online self-efficacy. In addition, the study demonstrates that researchers studying online learning self-efficacy should consider multiple aspects of self-efficacy in online contexts. Different from the existing self-efficacy researchers who have explored only one or two aspects of self-efficacy in online set- tings, we explored five aspects of self-efficacy that may represent more concrete online learning contexts. Furthermore, we found gen- der differences in online self-efficacy. This finding will contribute to existing online study involving gender difference in self-efficacy. Last, our study demonstrates that self-efficacy to complete an online course most significantly explains variances in satisfaction. It shows students' self-judgment about their capabilities to complete an online course is critical for their satisfaction with an online course. Further- more, instructors' proactive approaches for social interaction, such as monitoring and encouragement for social interactions, are suggested to help students develop the self-efficacy needed to complete an on- line course.

References

Ali, R., & Leeds, E. (2009). The impact of face-to-face orientation on online student reten- tion: A pilot study. Online Journal of Distance Learning Administration, 12(4) (Retrieved from , http://www.westga.edu/~distance/ojdla/winter124/ali124.html)

Artino, A. R. (2008). Motivational beliefs and perceptions of instructional quality: Predicting satisfaction with online training. Journal of Computer Assisted Learning, 24, 260–270.

Artino, A. R., & Stephens, J. M. (2009). Academic motivation and self-regulation: A com- parative analysis of undergraduate and graduate students learning online. Internet and Higher Education, 12(3–4), 146–151.

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

Bandura, A. (1988). Self-regulation of motivation and action through goal systems. In V. Hamilton, G. H. Bower, & N. H. Frijda (Eds.), Cognitive perspectives on emotion and motivation (pp. 37–61). Dordrecht, Netherlands: Kluwer.

Billings, D. M., Skiba, J., & Connors, H. R. (2005). Best practices in web-based courses: Generational differences across undergraduate and graduate nursing students. Journal of Professional Nursing, 21(2), 126–133.

Cho, M. -H. (2012). Online student orientation in higher education: A developmental study. Educational Technology Research and Development, 60, 1051–1069.

Cho, M. -H., & Jonassen, D. (2009). Development of the human interaction dimension of the Self-Regulated Learning Questionnaire in asynchronous online learning envi- ronments. Educational Psychology, 29, 117–138.

Cho, M. -H., & Kim, B. J. (2013). Students' self-regulation for interaction with others in online learning environments. Internet and Higher Education, 17, 69–75.

Cho, M. -H., Shen, D., & Laffey, J. (2010). The role of metacognitive self-regulation (MSR) on social presence and sense of community in online learning environ- ments. Journal of Interactive Learning Research, 21(3), 297–316.

Ergul, H. (2004). Relationship between student characteristics and academic achieve- ment in distance education and application on students of Anadolu University. Turkish Online Journal of Distance Education-TOJDE, 5(2) (Retrieved from , http:// tojde.anadolu.edu.tr/tojde14/articles/ergul.htm)

Fletcher, K. M. (2005). Self-efficacy as an evaluation measure for programs in support of on- line learning literacies for undergraduates. Internet and Higher Education, 8, 307–322.

Gebara, N. L. (2010). General self-efficacy and course satisfaction in online learning: A corre- lational study. (Unpublished doctoral dissertation). Columbia, MO: University of Missouri.

Hodges, C. B. (2008). Self-efficacy in the context of online learning environments: A re- view of the literature and directions for research. Performance Improvement Quar- terly, 20(3–4), 7–25.

Hung, M., Chou, C., Chen, C., & Own, Z. -Y. (2010). Learner readiness for online learning: Scale development and student perceptions. Computers in Education, 55(3), 1080–1090.

Lee, Y., & Choi, J. (2011). A review of online course dropout research: Implications for practice and future research. Educational Technology Research and Development, 59, 593–618.

Li, H. (2007). Efficacy, computer self-efficacy, and satisfaction with e-learning courses. (Unpublished doctoral dissertation). Vermillion, SD: University of South Dakota.

Lim, C. K. (2001). Computer self-efficacy, academic self-concept, and other predictors of satisfaction and future participation of adult distance learners. The American Journal of Distance Education, 15(2), 41–50.

Lin, Y. (2005). Understanding students' technology appropriation and learning perceptions in online learning environments. (Unpublished doctoral dissertation). Columbia, MO: University of Missouri.

Lin, Y. M., Lin, G., & Laffey, J. (2008). Building a social and motivational framework for understanding satisfaction in online learning. Journal of Educational Computing Re- search, 38(1), 1–27.

McGhee, R. M. H. (2010). Asynchronous interaction, online technologies self-efficacy and self-regulated learning as predictors of academic achievement in an online class. (Doctoral dissertation). Baton Rouge, LA: Southern University and Agricultural and Mechanical College.

Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis: The use of factor analysis for instrument development in health care research. Thousand Oaks, CA: Sage.

Schunk, D. H. (1991). Self-efficacy and academic motivation. Educational Psychologist, 26, 207–231.

Shea, P., Li, C. S., & Pickett, A. (2006). A study of teaching presence and student sense of learning community in fully online and web-enhanced college courses. Internet and Higher Education, 9, 175–190.

Thompson, L. F., & Lynch, B. L. (2003). Web-based instruction: Who is inclined to resist it and why? Journal of Educational Computing Research, 29, 375–385.

Wesley, J. W. (2002). A study of academic achievement, attitude, motivation, general self-efficacy, and selected demographic characteristics of community college students. (Doctoral dissertation). Starkville, MS: Mississippi State University.

Womble, J. (2008). E-learning: The relationship among learner satisfaction, self-efficacy, and usefulness. The Business Review, 10(1), 182–188.

Yang, C. -C., Tsai, I. -C., Kim, B., Cho, M. -H., & Laffey, J. M. (2006). Exploring the relation- ships between students' academic motivation and social ability in online learning environments. The Internet and Higher Education, 9, 277–286.