Distance Education, 2013 Vol. 34, No. 3, 290–301, http://dx.doi.org/10.1080/01587919.2013.835770

Self-regulation in online learning Moon-Heum Choa* and Demei Shenb a

Lifespan Development & Educational Sciences (LDES), Kent State University-Stark, North Canton, OH, USA; bShanghai Engineering Research Center of Digital Education Equipment, East China Normal University, Shanghai, China (Received 7 December 2012; final version received 17 June 2013) The purpose of this study was to examine the role of goal orientation and academic self-efficacy in student achievement mediated by effort regulation, metacognitive regulation, and interaction regulation in an online course. The results show that intrinsic goal orientation and academic self-efficacy predicted students’ metacognitive self-regulation; however, extrinsic goal orientation did not predict any type of regulation. Effort regulation and the amount of time spent in Blackboard predicted students’ academic achievement in the course, and interaction regulation predicted the amount of time spent in the online course. Results show the importance of individual students’ intrinsic goal orientation and academic self-efficacy in academic achievement. Discussion relates to current research and implications for online teaching and learning practice. Keywords: goal orientations; academic self-efficacy; metacognitive regulation; effort regulation; interaction regulation; online learning

Introduction Self-regulated learning (SRL) involves a student’s effort to manage learning processes systematically oriented to achieve goals (Zimmerman & Schunk, 2011). Often, multiple constructs explain students’ SRL (Artino, 2009; Azevedo, 2005; Cho & Jonassen, 2009; Zimmerman & Schunk, 2011). These constructs include goal orientation, academic self-efficacy, and regulations in the learning contexts (Pintrich, 2004). Skillful self-regulated learners have been reported to have higher intrinsic goal orientation and higher academic self-efficacy than less skillful students. In addition, skillful self-regulated learners better regulate and adjust their learning process in learning contexts than less skillful learners (Pintrich, 1999, 2004; Zimmerman & Schunk, 2011). Studies have reported that students’ SRL is important to determine successful learning experiences (i.e., satisfaction and achievement) in technology-mediated learning environments (Artino, 2008; Greene & Azevedo, 2009). For example, Artino (2008) found that academic self-efficacy and task value significantly explain students’ satisfaction with web-based courses. Greene and Azevedo (2009) found SRL is also related to students’ acquisition of conceptual knowledge in a web-based science course.

*

Corresponding author. Email: [email protected]

© 2013 Open and Distance Learning Association of Australia, Inc.

Distance Education

291

However, SRL is challenging to many students in a technology-mediated learning environment, especially in an online learning environment, where they may lack immediate support and feel lost or socially isolated (Cho, Shen, & Laffey, 2010; Sun & Rueda, 2012). Ali and Leeds (2009) found significant higher dropout rates in online courses than in face-to-face courses. In their review of research on dropouts, Lee and Choi (2011) found that online students’ lack of ability to self-regulate learning is a significant reason for high dropout rates, but other variables, such as personal reasons, explain a significant portion of dropout rate (Nichols, 2010). If students dropped out of online courses because of lack of SRL, these students tended to show lack of goal commitment, locus of control, and academic self-efficacy. In addition, they showed lack of coping strategies and resilience and underestimated the time required to complete tasks; therefore, SRL is an important factor in determining students’ success in online learning environments (Artino, 2008; Cho & Jonassen, 2009; Dabbagh & Kitsantas, 2005; Puzziferro, 2008). Because few empirical SRL studies have been conducted in online learning environments, the current researchers sought to determine the role of students’ SRL in their academic achievement in an online course. SRL was measured with several constructs, such as goal orientation, academic self-efficacy, and three types of regulation (effort regulation, metacognitive regulation, and interaction regulation). Thus, this study offers a comprehensive view of SRL in an online learning environment. Goal orientation Goal orientation refers to students’ intentions, implicitly set while choosing or engaging or persisting in diverse learning activities (Meece, Anderman, & Anderman, 2006; Schunk, 2012). Two types of goal orientation—intrinsic and extrinsic—have been explained by goal theorists (Pintrich, 1999). Intrinsic goal orientation refers to students’ disposition toward mastering the task, and extrinsic goal orientation refers to students’ disposition toward getting good grades in achievement situations. In general, intrinsic goal orientation is known to be positively related to students’ self-regulation and performance; and extrinsic goal orientation, negatively related (Meece et al., 2006; Pintrich, 1999; Rakes & Dunn, 2010; Sungar, 2007). Rakes and Dunn (2010) found that online students’ intrinsic goal orientation is negatively related with procrastination. Academic self-efficacy Academic self-efficacy, which refers to the confidence of students about their learning and performance, is known to be positively related to their self-regulation and academic performance (Artino, 2007; Meece et al., 2006; Pintrich, 1999). These findings seem to be replicable in online learning environments. Yang, Tasi, Kim, Cho, and Laffey (2006) found that students’ academic self-efficacy is associated with social interaction in online learning settings. They also discovered that the more confidence students have about learning and performance, the more they are likely (a) to feel comfortable sharing personal information with others (e.g., peers and instructors) and (b) to connect to instructors. In addition, Shea and Bidjerano (2010) found that the higher the academic self-efficacy, the more likely students are to regulate their effort in online learning environments.

292

M.-H. Cho and D. Shen

Effort regulation Effort regulation refers to students’ level of commitment to manage tasks and challenges with regard to their learning (Puzziferro, 2008). Through path analysis, Sungar (2007) reported that student effort regulation is explained mostly in terms of two variables: metacognitive regulation and self-efficacy. Furthermore, in online learning environments, Puzziferro found significant differences in effort regulation between withdrawn students and low-achieving students on the one hand and higher-achieving students on the other. She found students who acquired higher grades (e.g., A, B, or C) showed higher effort regulation than students who were either withdrawn or achieved lower grades (e.g., D or F) in online learning environments. In addition, Rakes and Dunn (2010) found online student effort regulation negatively associated with student procrastination. Metacognitive regulation Perhaps one of the most common constructs that have been extensively studied to explain SRL is metacognitive regulation, that is, students’ ability to plan, monitor, reflect, and adjust their learning process while studying learning materials (Duncan & McKeachie, 2005; Puzziferro, 2008). For example, Artino (2009) reported that students who have clear career goals are reported more likely to use metacognitive regulation in computer-mediated self-paced learning environments than students who do not have clear career goals. Puzziferro (2008) found high-achieving students are likely to show more metacognitive regulation and more satisfaction with online learning than low-achieving students. Interaction regulation Interaction regulation refers to students’ ability to regulate social interaction with others (Cho & Jonassen, 2009). Researchers have agreed that students should develop regulation skills for interaction with others in online learning environments (Cho & Jonassen, 2009; Garner & Bol, 2011). Cho et al. (2010) found that students’ interaction regulation, for example, monitoring for interaction with others is positively related to students’ perceived peer social presence, instructor social presence, connectedness to the community, and perceived learning; but self-regulation for learning tasks is not. Cho et al. concluded that in addition to self-regulation for learning tasks, multiple types of self-regulation, such as self-regulation for interaction with others, must be considered in online learning environments. Considering that interaction with others (e.g., peers) is a common task in online learning environments (Cho & Summers, 2012; Dabbagh & Kitsantas, 2005), one can easily recognize that interaction regulation is an important variable to explain online learning; however, very little research has been conducted to investigate the role of interaction regulation in online settings.

Methods Participants A total of 64 students enrolled in Introduction to Gerontology participated in the study. The course was delivered to students via Blackboard without face-to-face

Distance Education

293

meetings. All communication took place through e-mail or online discussion. The same online instructor taught two sections of the course with 34 and 30 students, all of whom participated in the study. Among the 64 participants, the majority were female (N = 58, 91%) and Caucasian (N = 54, 84%). Their average year in school, age, and number of online courses taken prior to this online course were 2.66 (SD = .86), 27.47 (SD = 9.03), and 2.08 (SD = 2.81), respectively. Measures Several measures were used to assess students’ goal orientations, academic self-efficacy, metacognitive regulation, effort regulation, and interaction regulation. Goal orientation The motivated strategies for learning questionnaire (MSLQ) (Duncan & McKeachie, 2005) were used to assess intrinsic and extrinsic goal orientation. Intrinsic goal orientation was measured with the four items (e.g., “In a class like this, I prefer course material that really challenges me so I can learn new things.”). Extrinsic goal orientation was measured with four more items (e.g., “Getting a good grade in this class is the most satisfying thing for me right now.”). A 7-point Likert scale was used where 1 denoted not at all true of me and 7 denoted very true of me. Cronbach’s alpha for intrinsic and extrinsic goal orientation was .75 and .63, respectively. Academic self-efficacy Academic self-efficacy was measured with eight items derived from the MSLQ (e.g., “I’m confident I can understand the most complex material presented by the instructor in this course.”). A 7-point Likert scale was used. Cronbach’s alpha for academic self-efficacy was .90. Metacognitive regulation Metacognitive self-regulation was assessed with 12 items derived from the MSLQ (e.g., “When reading for this course, I make up questions to help focus my reading.”). Some of the items were slightly changed for the course in which the current study was conducted. A 7-point Likert scale was also used. Cronbach’s alpha for metacognitive self-regulation was .82. Effort regulation Effort regulation was measured with four items derived from the MSLQ (e.g., “Even when course materials are dull and uninteresting, I manage to keep working until I finish.”). A 7-point Likert scale was used. Cronbach’s alpha for effort regulation was .61. Interaction regulation Interaction regulation, which consists of three regulation strategies—writing, responding, and reflection strategies—was assessed with 11 items derived from the online self-regulated learning inventory (Cho & Jonassen, 2009) (e.g., “When I write

294

M.-H. Cho and D. Shen

an online message, I try to organize my thoughts as much as I can.”). A 7-point Likert scale was used, where 1 indicated not at all true of me and 7 indicated very true of me. Cronbach’s alpha for interaction regulation was .78. Procedures In the middle of the semester, the instructor posted the research-recruiting e-mail on her class announcements. All the students taking the course participated in the study. Participation in the research was voluntary, but those participating received an extra point. Once they filled out the online consent form, the online survey was administered. The research was approved by the Institutional Review Board and conducted ethically. Results Descriptive statistics and correlations Descriptive statistics, including mean and standard deviations of variables used in the study, were acquired and listed in Table 1. On an average, students showed relatively high academic self-efficacy (M = 5.56, SD = .88) and effort regulation (M = 5.40, SD = 1.00). Students’ extrinsic goal orientation (M = 5.16, SD = 1.06) was higher than their intrinsic goal orientation (M = 4.75, SD = 1.04). The Pearson correlation coefficients of variables are shown in Table 1. Intrinsic goal orientation significantly correlated with academic self-efficacy (r = .30, p < .05), interaction regulation (r = .38, p < .01), effort regulation (r = .38, p < .01), and metacognitive regulation (r = .68, p < .01). Extrinsic goal orientation significantly correlated with only academic self-efficacy and interaction regulation, respectively (r = .34, r = .35, p < . 01). Academic self-efficacy positively correlated with all three types of regulation, such as interaction regulation (r = .50, p < .01), effort regulation (r = .32, p < .01), and metacognitive regulation (r = .43, p < .01) as well as login time (r = .36, p < .01). Student achievement in terms of total points significantly correlated to effort regulation and login time, r = 30, p < .05; r = .42, p < .01, respectively (see Table 1). Path analysis We created a conceptual path model according to the literature review and correlations among variables. The path model was analyzed using IBM SPSS Amos Table 1. Descriptive statistics and correlation of variables. Variables

M

SD

1

2

Intrinsic goal 4.75 1.04 – Extrinsic goal 5.16 1.06 .20 – Self-efficacy 5.56 .88 .30* .34** Interaction regulation 5.21 .83 .38** .35** Effort regulation 5.40 1.00 .38** .02 Meta regulation 4.44 .90 .68** .21 Login time (min.) 1556.07 789.74 .03 .05 Total points 282.46 36.65 −.09 −.08 Note. *p < .05;

**

p < .01.

3

4

– .54** .32** .43** .36** .18

– .36** .58** .29* .20

5

6

7

8

– .61** – .20 .15 – .30* .15 42** –

Distance Education

295

20.0.0. By dropping nonsignificant paths and using model fit indices, we found a good fit between the final tested model and the data. Some researchers (e.g., Y. Lin, G. Lin, & J. M. Laffey, 2008; Sockalingam, Rotgans, & Schmidt, 2011) followed Hu and Bentler (1998, 1999) and Byrne (2001), reporting a combination of absolute fit indices and relative or comparative fit indices (CFI). The fit indices include Chi-square values accompanied with degree of freedom, goodness of fit (GFI), root-mean-square error of approximation (RMSEA), and relative or CFI (Miles & Shevlin, 2007). Chi-square derives from the fit function and is sensitive to sample size. A nonsignificant Chi-square test result indicates good model fit because it assumes that variables in the base line model are not supposed to have any relationships. However, other indices should also be considered. For RMSEA, Hu and Bentler (1999) recommended a cutoff value of .06, and the smaller the number, the better the model fit. Both CFI and GFI range from 0 to 1; a value of CFI and GFI greater than .90 is considered a good fit. Compared to the recommended values of fit indices, the path model tested showed good model fit indices, with Chi-square tests nonsignificant (χ2 = 18.56, df = 18, p = .420), RMSEA = . 022, CFI = .996, and GFI = .929. The model fit criterion and fit indices are shown in Table 2. All path coefficients (i.e., standardized regression weights) were statistically significant at the .05, .01, or .001 levels. Figure 1 presents an overview of the model. Both intrinsic and extrinsic goal orientation significantly correlated with academic self-efficacy; however, extrinsic goal orientation had no direct effect on any other variables. In other words, extrinsic goal orientation did not significantly influence Table 2. Summary of the GFI indices. Model fits Recommended value Tested model

χ2

p

N/A 18.56

>.05 .420

CFI

GFI

>.90 .996

>.90 .929

.22

Intrinsic goal orientation

RMSEA