Scholarly article on topic 'The indirect relationship of media multitasking self-efficacy on learning performance within the personal learning environment: Implications from the mechanism of perceived attention problems and self-regulation strategies'

The indirect relationship of media multitasking self-efficacy on learning performance within the personal learning environment: Implications from the mechanism of perceived attention problems and self-regulation strategies Academic research paper on "Psychology"

CC BY-NC-ND
0
0
Share paper
Academic journal
Computers & Education
OECD Field of science
Keywords
{"Media in education" / "Computer-mediated communication" / "Learning communities" / "Interactive learning environments" / "Post-secondary education"}

Abstract of research paper on Psychology, author of scientific article — Jiun-Yu Wu

Abstract Media multitasking, characterized by simultaneous engagement in multiple media forms, is prevalent among university students within the personal learning environment. However, those who think they are capable of multitasking usually overestimate their ability to perform the actual tasks. This study examined university students' learning performance from the perspectives of their media multitasking self-efficacy, perceived attention problems, and self-regulation strategies using the revised Online Learning Motivated Attention and Regulatory Strategies scale. Participants were 696 university students (275 males, 39.51%) in Taiwan. The author developed the media multitasking self-efficacy scale through open-ended interviews and pilot tested the measures using an exploratory factor analysis. The confirmatory factor analysis verified the uni-factor structure of the instrument. Second-order confirmatory factor analysis validated the two orthogonal higher-order constructs of perceived attention problems and self-regulation strategies as well as their subscales. Results from the multilevel structural equation model revealed significant negative indirect relationship between media multitasking self-efficacy and learning performance via both students' perceived attention problems and self-regulation strategies. Study findings have implications for prevention and intervention of university students' media-related attention problems and poor regulation strategy use within the personal learning environment.

Academic research paper on topic "The indirect relationship of media multitasking self-efficacy on learning performance within the personal learning environment: Implications from the mechanism of perceived attention problems and self-regulation strategies"

Accepted Manuscript

The indirect relationship of media multitasking self-efficacy on learning performance within the personal learning environment: Implications from the mechanism of perceived attention problems and self-regulation strategies

Jiun-Yu Wu

PII: S0360-1315(16)30196-8

DOI: 10.1016/j.compedu.2016.10.010

Reference: CAE 3083

To appear in: Computers & Education

Received Date: 16 May 2016 Revised Date: 15 October 2016 Accepted Date: 21 October 2016

Please cite this article as: Wu J.-Y., The indirect relationship of media multitasking self-efficacy on learning performance within the personal learning environment: Implications from the mechanism of perceived attention problems and self-regulation strategies, Computers & Education (2016), doi: 10.1016/j.compedu.2016.10.010.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Running Head: MEDIA MULTITASKING SELF-EFFICACY & LEARNING PERFORMANCE

The Indirect Relationship of Media Multitasking Self-Efficacy on Learning Performance within the Personal Learning Environment: Implications from the Mechanism of Perceived Attention Problems and Self-Regulation Strategies

Jiun-Yu Wu

Institute of Education, National Chiao Tung University, 1001 University Rd., Hsinchu, 30010, Taiwan Email: jiunyu.rms@gmail.com

Correspondence concerning this article should be addressed to Jiun-Yu Wu, Institute of Education, National Chiao Tung University, 1001 University Rd., Hsinchu, 30010, Taiwan R.O.C. Email: j iunyu. rms@gmail. com / Tel: +88635731843

The Indirect Relationship of Media Multitasking Self-Efficacy on Learning Performance within the Personal Learning Environment: Implications from the Mechanism of Perceived Attention Problems and Self-Regulation Strategies

MEDIA MULTITASKING SELF-EFFICACY & LEARNING PERFORMANCE 2 Abstract

Media multitasking, characterized by simultaneous engagement in multiple media forms, is prevalent among university students within the personal learning environment. However, those who think they are capable of multitasking usually overestimate their ability to perform the actual tasks. This study examined university students' learning performance from the perspectives of their media multitasking self-efficacy, perceived attention problems, and self-regulation strategies using the revised Online Learning Motivated Attention and Regulatory Strategies scale. Participants were 696 university students (275 males, 39.51%) in Taiwan. The author developed the media multitasking self-efficacy scale through open-ended interviews and pilot tested the measures using an exploratory factor analysis. The confirmatory factor analysis verified the uni-factor structure of the instrument. Second-order confirmatory factor analysis validated the two orthogonal higher-order constructs of perceived attention problems and self-regulation strategies as well as their subscales. Results from the multilevel structural equation model revealed significant negative indirect relationship between media multitasking self-efficacy and learning performance via both students' perceived attention problems and self-regulation strategies. Study findings have implications for prevention and intervention of university students' media-related attention problems and poor regulation strategy use within the personal learning environment.

Keywords: media in education; computer-mediated communication; learning communities; interactive learning environments; post-secondary education

1. Introduction

The Internet has expanded the boundaries of higher education and made it more accessible and cost-effective (Perna et al., 2014). For example, the Personal Learning Environment (PLE) premised on social media (e.g., Facebook or Twitter) enables students to share, communicate, and collaborate with others for both formal and informal learning (Dabbagh & Kitsantas, 2012; Gillet, Law, & Chatterjee, 2010). Despite the rich informational resources and social opportunities it offers (MOOCs: Massive Open Online Courses, YouTube, wikis, etc.), the PLE is a self-directed environment (Kop & Fournier, 2011) and, therefore, requires that students self-regulate their attention to achieve good learning results (Gillet et al., 2010; Johnson & Sherlock, 2014). Attention occurs prior to cognitive information processing. How learners can stay focused and remain engaged, therefore, is fundamental to the onset of major cognitive learning activities (Petersen & Posner, 2012). To be more specific, learners' awareness of and willful control/regulation of their attention are essential for achieving a focused attention (Reisberg & McLean, 1985).

Grounded on theories of attention and metacognition, Wu (2015) explored university students' meta-attention, that is the awareness and regulation of their attention, during online learning using the scale of Online Learning Motivated Attention and Regulation Strategies (OL-MARS). He found that student academic achievement and learning-related outcomes were negatively associated with perceived media-related attention problems and positively associated with self-regulation strategies within the PLE. Although the results suggest that the OL-MARS seems an effective tool to monitor students' awareness and regulation of their attention, the scale needs to be extended to fully reflect perceived attention problems (PAP) and self-regulation strategies (SRS). In this study, we regarded PAP and SRS as broader constructs of engagement in

multitasking, because higher PAP and poor SRS reflect students' awareness of their attention problems and poor regulation to focus on a single task.

Failure to regulate attention can lead to concurrent engagement in multiple media applications that are irrelevant to learning (e.g., texting, online chatting, non-homework-related Internet use). This phenomenon has been referred to as media multitasking and is widely prevalent among university students (Carrier, Rosen, Cheever, & Lim, 2015; Junco & Cotten, 2012; Karpinski, Kirschner, Ozer, Mellott, & Ochwo, 2013). However, multitasking has been found to be negatively correlated with learning performance (Bowman, Levine, Waite, & Gendron, 2010; Junco, 2012; Kirschner & Karpinski, 2010; Kraushaar & Novak, 2010; Rosen, Carrier, & Cheever, 2013) possibly due to the limited capacity of our cognitive resources (Rosen, Lim, Carrier, & Cheever, 2011; Wood et al., 2011).

People's media multitasking self-efficacy (MMSE) may be the culprit of their attention problems and multitasking behavior. According to Bandura (2006), "perceived self-efficacy is concerned with people's belief in their capabilities to produce given attainments" (Bandura, 2006, p. 307). In the same vein, people with higher MMSE may engage in more media multitasking. However, those who think that they are capable of multitasking may overestimate their ability in performing the actual multiple tasks at once (Sanbonmatsu, Strayer, Medeiros-Ward, & Watson, 2013).

Moreover, the mechanism between MMSE and actual learning performance is still unclear. Therefore, this study aims to fill this gap by examining the indirect relationship between MMSE and learning performance via PAP and SRS. We postulated that if people obtain higher MMSE, it is likely that they will engage in more multitasking activities (evidenced in higher PAP and poorer SRS), which in turn is correlated with poorer learning performance within the PLE. This indirect

relationship will be tested in the study.

1.1. Research purposes

Drawing on the theories of attention (Petersen & Posner, 2012), meta-attention (Wu, 2015), and metacognition (Schraw & Sperling Dennison, 1994), this study aimed to (a) validate the theoretical and measurement structures of media multitasking self-efficacy and meta-attention using confirmatory factor analyses (1st-order and 2nd-order CFA), and to (b) explore the indirect effects of MMSE on learning performance in the PLE through the pathways of meta-attention constructs. In the following sections, we provided the theoretical framework and reviewed relevant literature regarding the proposed hypotheses.

1.2. The theoretical framework of meta-attention within the PLE

To avoid distraction and to stay focused on online learning, selective attention plays an important role (Broadbent, 1958; Dayan, Kakade, & Montague, 2000). Learners need a good understanding of their attention state as well as good strategy use to regulate their attention. These two components are parallel to the knowledge and regulation components in metacognition (Schraw & Sperling Dennison, 1994). A plethora of studies have investigated how metacognition, conceptualized as awareness/knowledge of one's cognition (Lee & Wu, 2013; Wu, 2014) as well as regulation of one's cognition (Hathorn & Rawson, 2012; Kauffman, Zhao, & Yang, 2011; Lee, 2015), can contribute to online learning. Nevertheless, before being engaged in a learning activity, students need to direct their attention to the online task and stay focused. Therefore, meta-attention, that is students' awareness of their attention and regulation of their attention, is critical to the start and continuation of the learning activity (Wu, 2015). Within a PLE, awareness of attention may be conceptualized as perceived attention problems (PAP), which reflects students' perception of their media-related attention problems. On the other hand,

regulation of attention may be conceptualized as self-regulation strategies (SRS), which reflects students' strategy use to regulate their attention. PAP and SRS can be seen as broader constructs of engagement in multitasking, because higher PAP and poorer SRS reflect students' awareness of their distracted attention and poor attention regulation to focus on a single task.

The subscales in the PAP construct are derived from theories of attention (Petersen & Posner, 2012; Posner & Boies, 1971; Posner & Petersen, 1990). From a cognitive point of view, attention includes three subsystems: (a) The executive control system selects the target information for further processing, (b) the orienting system directs learners' attention to sensory stimuli, and (c) the alerting system keeps learners alert and vigilant for priority events. Each of the subsystem can correspond to a subtype of PAP. Within the PLE, we propose that three typical questions guide PAP: (a) "Do I know that I tend to select and process information irrelevant with online learning?" (b) "Do I know that I am easily attracted to social media notifications or signals from digital devices?" and (c) "Do I know that I am expecting something unrelated with learning while studying?" The three questions correspond to self-awareness of the executive control, orienting, and alerting systems within the PLE.

In Wu (2015), the nature of Perceived Attention Discontinuity (PAD) is the awareness of the executive control system, such that PAD reflects university students' awareness of selecting and processing learning-unrelated things. The Social Media Notification (SMN) construct is the awareness of the orienting system, which measures students' awareness of being distracted to sounds, signals, and vibrations from social media or digital devices. However, in Wu (2015), a scale to assess university students' awareness of the alerting system was not available. In the present study, therefore, awareness of the alerting system was operationalized as the Lingering Thoughts (LT) that university students have regarding expecting things to happen or remaining

vigilant for an event on the Internet during online learning. Table 1 summarizes the functioning of the meta-attention components within the PLE with operational definitions and guiding questions as well as task-relevant/-irrelevant attention states and successful/unsuccessful attention regulation.

The construct of SRS may include subscales to assess learners' strategy use that can prevent them from being distracted. SRSs are generally categorized as either explicit or implicit. Explicit strategies are behavioral strategies characterized by enforcing behavioral control on Internet use, such as access denial and removal of devices for Internet connection (Quan-Haase, 2010). Implicit strategies are motivated inwardly to achieve success or to avoid guilty feelings. Among the implicit strategies, outcome appraisal (Weiner, 1985) is the major implicit activity in this genre. Outcome appraisal is characterized by a sequence of linking the perceived success or failure with a positive or negative emotion. For example, if the learner finds himself engaged in the intended online learning activity, he feels happy, but if he finds himself distracted from the intended online learning, he feels sad and frustrated. Causal attribution, which follows the outcome appraisal, may have implication for university students' regulation of their attention. Thus, following outcome appraisal, learners attribute their success or failure based on three dimensions: locus of causality (internal vs. external), stability (stable vs. unstable), and controllability (controllable vs. uncontrollable), which are associated with a specific emotion (Weiner, 1985).

In the current study, we focused on learners' self-esteem (pride) and guilt when they find that they can or cannot engage in online learning. Self-esteem (pride) is linked with an internal locus of causality, such as ability, which is stable but uncontrollable. If learners ascribe a focused attention on online learning to their ability, they may have more positive self-esteem, but if they

ascribe failure to lack of ability, they may have low self-esteem. Guilt is derived from a controllable cause, such as efforts, which are directed inward but are unstable. For example, learners may feel guilty when they cannot control their online behavior because they recognize that they are responsible for distractions from online learning. More positive structures of thinking (causal attribution), that is, linking failure with lack of effort and linking success with ability, are associated with increased persistence in a task (Robertson, 2000). Both self-esteem and guilt are intrinsically directed emotions, which may affect learners' motivation through causal attribution and activate adaptive behaviors (Weiner, 1985) for regulation of attention. Therefore, we investigated students' behavioral strategies as well as outcome appraisal so as to derive implications for instructional practices within the PLE.

In this study, we conducted 2nd-order CFAs to validate the theoretical factor structure of PAP and SRS along with their subsuming scales. As shown in Figure 1, we posited in Hypothesis 1 (H1) that meta-attention consists of two constructs, namely PAP and SRS along with their subsuming scales.

1.3. MMSE and engagement in the multitasking

University students' MMSE on the PLE may account for their simultaneous engagement in multiple media forms based on Bandura's self-efficacy theory (Bandura, 1977). According to Bandura (1977), people's belief about their ability to complete a task can affect their amount of efforts and extent of involvement in the endeavor. For example, students with higher Internet self-efficacy spent more time per week on the Internet compared to those with lower Internet self-efficacy (Tsai & Tsai, 2010). In addition, higher self-efficacy to concentrate on homework was related with less frequent multitasking behavior (Calderwood, Ackerman, & Conklin, 2014).

Likewise, in a sample of 310 university students, Sanbonmatsu, Strayer, Medeiros-Ward,

and Watson (2013) showed that university students who think they are capable of multitasking more often than others engage in simultaneous tasks; however, the students tended to overestimate their ability to satisfactorily accomplish the simultaneous tasks. As noted earlier, PAP and SRS can be regarded as broader constructs of engagement in multitasking as they reflect people's awareness of their distracted attention and poor regulation to focus on a single task. Therefore, higher MMSE may be associated with more engagement in multitasking, which could be evidenced in higher PAP and poorer SRS. This relationship would be tested in Hypothesis 2.1 (H 2.1): Higher MMSE is correlated with higher PAP and poorer SRS.

1.4. Direct relationships of PAP and SRS with learning performance

Attention is fundamental to selecting information for "focal (conscious) processing" (Posner & Petersen, 1990, p. 25). Unfortunately, the capacity of human attention is limited due to the constraints of working memory capacity (WMC) so that it cannot attend to multiple sensory inputs at the same time (Ibos, Duhamel, & Hamed, 2013; Pashler, 1998). Experimental studies have revealed that participants with high WMC demonstrated better attentional control in dichotic listening tasks (commonly used to investigate selective attention within the auditory system) for both selective attention (Conway, Cowan, & Bunting, 2001) and divided attention (Colflesh & Conway, 2007). Specifically, in the selective attention condition, participants with low WMC tended to detect highly pertinent information (e.g., hearing their own name) in the message that they were told to ignore and thus had difficulty in blocking out irrelevant information. In the divided attention task, compared to low WMC participants, participants with high WMC were better able to divide their attention to shadow (repeat aloud) messages in one ear and detect their name in the other ear, suggesting they had appropriate control of attention to goal-relevant information.

The divided attention condition may be viewed as a type of multitasking. Though people with high WMC can do better in the divided attention task in laboratory experiments, the cost of multitasking is dependent upon the level of problem representation, task difficulty, and amount of mental information carried over from one task to another (Iqbal & Bailey, 2006). A common multitasking scenario for university students consists of studying for a test and instant messaging friends on social network sites (SNS) at the same time (Ravizza, Hambrick, & Fenn, 2014). According Salvucci, Taatgen, and Borst's (2009) unified model of multitasking continuum, this typical multitasking could extend the designated task over hours of work, causing potential delay or tasks undone. Thus, in Hypothesis 2.2 (H2.2), we posited that higher PAP and poorer SRS as broader constructs of engagement in multitasking may be associated with poorer learning performance within the PLE.

1.5. The indirect relationship between MMSE and learning performance

Researchers demonstrated that university students with higher self-efficacy in multitasking may be engaged in more media multitasking; nevertheless, more frequent multitasking was correlated with poorer multitasking ability in performing the cognitive task (Sanbonmatsu et al., 2013). More recently, Brooks (2015) tested if multitasking computer self-efficacy moderated the relationship between social media use and task performance. His result showed that social media use negatively predicted task performance but no moderation effect was observed. Evidence from the reviewed studies suggested that self-efficacy in multitasking would be positively related with concurrent engagement in multiple tasks but may have a negative relationship in the task performance or no effect in moderating the relationship between engagement and task performance.

It is evident that self-efficacy in multitasking may not agree with the actual performance in

the task. Those who think they are capable of multitasking tend to overestimate their ability in the actual multitasking tasks (Sanbonmatsu et al., 2013) or assume they can do them without the loss of efficiency and productivity (Kirschner, Sweller, & Clark, 2006). It is likely that the two simultaneous activities compete for the limited cognitive resources available, with the result that performance of the two tasks interferes with each other (Karpinski et al., 2013). However, the mechanism between MMSE and task performance was still unknown. MMSE would be correlated with higher PAP and poorer SRS (i.e., broader constructs of engagement in multitasking), which in turn would correlate with learning performance within the PLE. Therefore, in Hypothesis 2.3 (H2.3), we posited that there would be negative indirect relationships of MMSE on learning performance through the two meta-attention constructs.

2. Method

2.1. Participants and Procedure.

Participants were 748 Taiwanese undergraduate and graduate students from 22 psychology, education, statistics and measurement related courses across seven universities. Instructors were recruited from a professional development community for higher education. All instructors had experiences in using Facebook to facilitate their teaching and agreed to participate in this project. Data were collected in the spring and fall semesters of 2015 school year. The instructors assigned readings and coursework online and created private learning communities for the courses on Facebook. The intent behind creating these online learning communities on Facebook is to provide a platform for students to collaborate and communicate with each other due to the popularity of Facebook among university students and the prevalence of portable devices.

As part of the course requirement, students were asked to join the online communities, discuss the course-related topics, share related learning materials and comments, complete all the

required online readings, return the homework online, and take the paper-and-pencil midterm and/or final examinations. A semester lasts 18 weeks long in Taiwan. Students signed the consent form at the beginning of the semester and answered a 30-minute paper-and-pencil questionnaire in the classroom in the mid of semester. They were asked to answer the items based on their experiences over the past two months. Students could decide not to answer the questionnaire or not to grant the access to their final course grades. They were informed that their decision would not affect their rights and final grades in the course. Extra course credit was given to students who completed and returned the questionnaire. For those who chose not to return the survey, alternative assignment (i.e., a one-page reflection page about the course) was provided to earn the extra credit.

Of the 748 students, 696 (275 males, 39.51%; 546 undergraduate students, 78.4%) returned the questionnaire with valid responses (return valid rate = 93%). Therefore, data of the 696 students were used for analysis in this study. Among the participants, 36.1% was from education-and human development-related departments, 22.1% from psychology- and social science-related departments, 20.4% from science-, technology- and engineering-related departments, 14.1% from language-related departments, 5.3 % from business- and management-related departments, and 1.9 % from humanity- and art-related departments. The mean age of the sample was 21.75 years old (MIN=18, MAX=48 and SD = 3.77 years old).

All students possessed at least a terminal to access the Internet, had experiences in using the Internet to complete course-related assignments, and had an active Facebook account. They reported visiting Facebook primarily using smartphones (46.1%) and laptops (44.7%), compared with 18.0% reporting using desktops. 2.2. Instrument

The questionnaire consisted of two parts. The first part required students to report their

demographic information and time spent on the Internet and Facebook, including average time spent on the Internet per day, average time spent on Facebook per day, average number of Facebook visits per day, and average time spent on Facebook per visit. In order to avoid extreme values, responses were measured on a Likert scale. The average time spent on the Internet was rated from 1 (below 2 hrs) to 5 (above 8 hrs). The years of FB use ranged from 1 (below 1 yr) to 5 (above 4 yrs). The average time spent on Facebook was rated from 1 (below 30 min) to 6 (above 4 hrs). The average number of Facebook visits per day was rated from 1 (below 3 visits) to 5 (above 15 visits). The average time spent on Facebook per visit was rated from 1 (below 5 min) to 6 (above 2 hrs). The second part of the questionnaire required students to report their media multitasking self-efficacy and their perceived attention state and regulatory strategies in online learning. We provide a description of these measures below. 2.2.1. Media multitasking self-efficacy scale (MMSE)

Bandura (2006) denoted that there is not an all-purpose self-efficacy scale because an all-purpose test deviates from the specific situations and may have little to do with the designated domain functioning. Two measures relevant with the current study were Sanbonmatsu et al's (2013) "perceived ability to multi-task" and Basoglu et al's (2009) multitasking computer self-efficacy (MTCSE). The former asks participants to rate their ability to multitask relative to that of others without a specific context. The latter assesses a person's ability to switch between multiple tasks on a single central processing unit. We created our own media multitasking self-efficacy (MMSE) scale because we argued that Sanbonmatsu et al's (2013) lacked specific multitasking contexts. On the other hand, Basoglu et al's (2009) measures focused on multitasking on a single digital device, which may not account for the prevalent simultaneous tasks on multiple media forms among today's university students. Thus, we developed the MMSE based on

interviews with twelve heavy Internet users recruited from the Internet. Of the twelve students, nine were undergraduate students and three were graduate students.

The common theme that arose concerning their multitasking experiences during learning included concurrent use of Facebook, instant messaging, online shopping, web surfing, and video or audio calls using more than one possible terminals. Therefore, we generated five items of MMSE based on the interview results. The items were constructed in accordance with Bandura's guide in constructing self-efficacy measures (Bandura, 2006). The five questions were "I can surf the Internet for non-academic purposes while studying and still study sufficiently," "I can make video or audio calls (e.g., line, facetime, google hangouts) with friends while studying and still study sufficiently," "I can instant message friends while studying and still study sufficiently," "I can do online shopping while studying and still study sufficiently," and "I can use Facebook for non-academic purposes while studying and still study sufficiently." Responses were rated on a 6-point Likert scale with 1 indicating not confident at all, and 6 very confident. 2.2.2. Revised online-learning motivated attention and regulation Scale (OL-MARS v.2).

As shown in Figure 1, the OL-MARS v.2 based on theory of meta-attention included two major constructs, including PAP and SRS. PAP consisted of three subscales, including perceived attention discontinuity (PAD), lingering thoughts (LT), and social media notifications (SMN). SRS consisted of two subscales, including behavioral strategies (BS) and outcome appraisal (OA). Responses were rated on a Likert scale with 1 indicating extreme disagreement and 5 extreme agreement. The description about the measures was provided below.

Perceived attention discontinuity (PAD). PAD measures students' awareness that they often select and process information irrelevant to their online learning. PAD was measured by eight items. Sample questions included "I visit websites or open software that are irrelevant to my

learning when using the Internet for my project or study." and "I often click the links of interesting ads, pictures, or articles unconsciously when using computers to search information for my project." Factor loadings ranged from .609 to .806 with an internal consistency of .887.

Lingering thoughts (LT). LT assesses students' awareness that they often expect or stay vigilant for things irrelevant with their online learning. LT was measured by four items. Sample questions included "When studying, I often feel that there is something interesting happening on the Internet." and "When using the computer for studying, I can't help but feel like to play mobile games unconsciously." Factor loadings ranged from .498 to .736 with internal consistency of .724.

Social media notification (SMN). SMN measures students' awareness that they are easily attracted to sounds, signals, and vibrations from social media or digital devices. In the original OL-MARS, there is only one item for SMN (i.e., I constantly watch out for the sound or pictorial signals from smartphones, Line, Facebook, or other devices and applications). In the current study, three items derived from the original item were developed to improve the psychometric properties of the scale. Questions included "When I see, hear, or feel the signals, sounds, or vibrations from my mobile devices (e.g., cellphone and tablets), I will check them immediately.;" "When I see or hear notifications from social media (e.g., Line, Facebook), I can't wait to check them.;" and "When studying, I can immediately notice the signals from instant message software (such as Line, What's app)." Factor loadings ranged from .683 to .730 with an internal consistency of .774.

Behavioral strategies (BS). Students' behavioral control in regulating their attention was measured by six items. Sample questions included "I ask myself to complete the scheduled assignment first before using the Internet for non-academic purposes" and "When studying, I log out my Facebook account or close instant message software, so that I can focus on my work."

Factor loadings ranged from .491 to .716 with an internal consistency of .774.

Outcome appraisal (OA). Students' act of linking the outcome of their online learning to a specific emotion was measured by three items. Sample questions included "If I focus on what I should do when using computer (e.g., studying), I will feel happy and proud of myself' and "When I notice that I am browsing unrelated sites or playing computer games, I will feel guilty." Factor loadings ranged from .446 to .720 with an internal consistency of .694. 2.2.3. Learning performance within the PLE.

Participants had to complete the required coursework by reading assigned online reading, sharing course-related information in the clubs via Facebook, and collaborating with their classmates within the PLE. Their course grades were determined by their performance in the in-classroom paper-pencil midterm and/or final exam, team projects, individual assignments, and participation in the online and face-to-face discussion. Therefore, the final course grades were a composite measure of students' learning performance collected both within and outside the PLE. Students signed the consent form to grant the use of their final course grades for research purposes and were promised that only the average course grade would be shown to readers and that no personal information would be revealed. A total of 643 students (92.38%) granted permission to use their final course grade for analysis in the study. Due to the classroom-level differences in the content covered in each course, scoring standards, assessment formats, and class member composition, Multilevel Structural Equation Modeling (MSEM, Preacher, Zhang, & Zyphur, 2011; Wu & Kwok, 2012) was utilized to deal with the nested structure of this learning outcome to reflect students' original performance within each classroom in the student-level. 2.3. Data Analysis

Under the framework of Structural Equation Modeling (SEM, Joreskog, 1970; Kline,

2010), the Confirmatory Factor Analysis (CFA) and second-order CFAs (Brown, 2006) were used to examine the measurement structure of MMSE and OL-MARS v.2 scales according to their theoretical constructs respectively.

As for the relationship among study variables, MSEM technique was used to test our research hypotheses concerning the interplay among MMSE, PAP, SRS, and learning performance within the PLE. An unconditional multilevel model was first constructed to examine the extent of between-level (i.e. classroom-level) variation due to the classroom difference to the overall variation of raw course grades, considering the design effect (Kish, 1995) and Intra-class correlation (ICC: Bengt O. Muthen, 1994; Shrout & Fleiss, 1979). The design effect evaluates the degree of underestimating the standard errors in a complex sample. The ICC quantifies the amount of dependency in the data by dividing the between-level variance ( too) with the sum of between-and within-level variance (t00 + S) (Hox, 2010). Design effect larger than two (B. O. Muthen & Satorra, 1995) or ICC larger than .05 (Kreft & De Leeuw, 1998) might indicate the need to construct the multilevel models. Secondly, a random intercept model with within-level hypothesized indirect effect model was constructed incorporating the within-level (i.e. student-level) factors and raw course grades. Students' gender and age were included as covariates for PAP, SRS, and the learning performance to control for their influence on the study variables. Due to the non-significant linear association between MMSE and learning performance within the PLE, the indirect effects of MMSE on learning performance via PAP and SRS were investigated using the Sobel test (1982) with Mplus v7.4 (Muthen & Muthen, 2010). A model-fit chi-square test and related model fit indices, including CFI, RMSEA, and SRMR (Hu & Bentler, 1999), were used to assess the adequacy of the hypothesised models. Variance explained and standardized path measures were used as effect sizes to evaluate the practical significance of the

hypothesized structural model and the indirect effects. The full information maximum likelihood (FIML, Arbuckle, 1996) estimation was chosen to handle the incomplete dataset by considering all the available information of variables from the whole sample (Hancock & Mueller, 2010) in addressing the missingness of students' final scores. Moreover, to accommodate the nesting nature and possible non-independence of our data (i.e. students nested within 22 courses), TYPE=COMPLEX routine and TYPE=TWOLEVEL routine with robust procedure (MLR) in Mplus v7.4 was utilized to yield the consistent standard error estimates and statistical inferences of CFA and MSEM analytical results (B. O. Muthen & Satorra, 1995; Wu & Kwok, 2012).

3. Results

3.1. The measurement models

The item statistics of MMSE and OL-MARS v. 2 scales were tabulated in Table 2. The item mean for MMSE ranged from 2.54 to 3.47. Most of the item means for the OL-MARS were between 3.17 and 3.87. Only LT and BS had item means less than 3. We summarized the analytical results of measurement models below. 3.1.1. MMSE scale validation

The MMSE were pilot tested in a sample of 175 university students. The Exploratory Factor Analysis with Maximum Likelihood extraction was performed to examine the factorial structure of MMSE. The solution of one unified factor explained 58.83% of total variance in the five items, with KMO = .823 and Bartlett's test statistic = 286.088, df= 10, p<.01. The standardized factor loadings ranged from .656 to .755 with a 48.62% of average variance extracted for all items and an internal consistency of .824. The confirmatory factor analysis (CFA) revealed a good model fit to the data (C= 5.641, df = 4, p =. 28, CFI = .983, RMSEA = .032, SRMR = .024) for the current sample. Standardized factor loading ranged from .593 to .756 with an internal consistency of .809.

3.1.2. OL-MARS v. 2 scale validation

To validate the scales of meta-attention, we fitted two second-order CFAs in a consolidated model. The results indicated a lack of fit of the model to the data due to the orthogonal relationship between the two higher-order factors (i.e., the correlation coefficient between perceived attention problem, PAP, and self-regulation strategy, SRS, was close to zero). This further suggested that awareness of one's attention problem has no linear relationship with regulation of one's attention. Therefore, two separate second-order CFAs were fit to the data, one for PAP (see Figure 3) and the other for SRS (see Figure 4). Both models indicated adequate fit to the data [Hypothesis 1]. The chi-square was significant in both models (£= 2837.081, df = 85, p<.01 for PAP; C= 40.838, df = 24, p = .02 for SRS). Other fit indices showed good fit for both models (CFI = .918, RMSEA = .056, SRMR = .061 for PAP; CFI = .956, RMSEA = .035, SRMR = .062 for SRS). The standardized second-order factor loadings ranged from .787 to .888 for PAP and from .707 to .717 for SRS. R2was 61.9% for PAD, 69.5% for LT, 78.8% for SMN, 50.1% for behavioral strategies (BS), and 51.4% for outcome appraisal (OA). Internal consistency was .908 for PAP and .772 for SRS. The standardized first-order factor loadings ranged from .498 to .806 for PAP and from .446 to .720 for SRS.

Two correlated item residuals were made for the second-order CFA of PAP and an additional correlated residual was made for the second-order CFA of SRS. Both items PAD7 (I often visit Facebook or other social network sites unconsciously when using computers to do my homework or to learn online) and SMN1 (When I see or hear notifications from social media (e.g., Line, Facebook... etc), I cannot wait to check them.) involved impulse control problems of conscious or unconscious desires. Same was for SMN3 (When I see, hear, or feel the signals, sounds, or vibrations from my mobile devices (e.g., cellphone and tablets), I will check them

immediately) and PAD8 (If I encounter difficulties when using the Internet for studying, I will open other programs, websites, or check my smartphone unconsciously). An additional correlated residual was made between BS5 (If I postponed what I should do because of using the Internet, I try to avoid this next time) and BS6 (When I notice that I am browsing unrelated websites or playing games, I ask myself to turn back to what I should do (e.g. writing paper, learning, or searching information) for the second-order CFA of SRS. Both items involved the wording of learning-unrelated tasks. These correlated residuals are in line with Brown's perspective that correlated residuals may imply similarities in content and phrasing, allowing correlated residuals is often necessary for model improvement and interpretation (Brown, 2006). 3.2. Descriptive statistics and zero-order correlations among constructs and criterion variables.

Table 3 exhibits the descriptive statistics and correlations among study variables. Subscales within PAP were positively highly correlated with each other, and same for subscales within SRS. In agreement with the result from the second-order CFA, PAP and SRS showed no linear relationship between each other. That is, university students' perception of their attention problems may not correspond to their efforts in regulating their attention. Moreover, OA was modestly and positively correlated with PAP and the subscales (rs = .194~.221) while BS had no correlation with PAP and the subsuming subscales. MMSE was positively correlated with PAP and the subscales (rs = .177~.195) but was negatively correlated with SRS as well as OA and BS (rs = -.167~-.195). However, students' course grades was negatively correlated with PAP and the subscales (rs = -.115~-. 120) but was positively correlated with SRS as well as OA and BS (rs = .086~.116). Additional correlational analyses showed that PAP and the subscales were positively correlated with time spent on the Internet or Facebook, the number of Facebook checks per day, and average time spent on Facebook per visit (rs = .163~.337, p<.05), while BS and

subscales were negatively correlated with measures of Internet and Facebook-related use (r = -.102 ~ -.166, p<. 05). Finally, there was no statistically significant correlation between MMSE and course grades (r = -.078); therefore, we would test the indirect relationship of MMSE on learning performance via PAP and SRS. 3.3. The structural model

An unconditional multilevel model of raw final course grades was first constructed to separate the estimations of its between- and within-level variance components. The ICC of raw final course grades was 0.292 (S= 39.967 & to=16.509 with ps<.001, M=84.213) with an average cluster size of 31.636 students in 22 classrooms, and the design effect (Kish, 1995) was equal to 9.946, which is larger than the threshold of 2 in Muthen & Satorra (1995). That means more than 29.2% of total variance of final course grades resulted from the differences between the classrooms (e.g. covered contents, scoring standards, assessment formats, and class member compositions) and that the use of MSEM was necessary to have efficient and effective parameter and standard error estimates for our hypothesized model given the nested data structure.

With the criterion of raw final course grades, we then constructed a random intercept model with our hypothesized indirect effect model in the lower level. After separating the classroom-level variation of learning outcome, we fitted a structural model in the student level where MMSE was correlated indirectly with learning performance via PAP and SRS. Due to the orthogonality between PAP and SRS, the factor scores of these two higher-order components were used in the analysis.

The results of the structural model are exhibited in Figure 5. As illustrated, the hypothesized model fit the data well (C = 48.232, df = 26, p <.01, CFI = .919, RMSEA = .035, SRMR = .035).

As expected, MMSE positively predicted PAP (/3MM^SE®I>AP = .178, p<.01) and negatively

predicted SRS (J3MMSE®SRS = -.178, p<.01) [Hypothesis 2.1]. PAP negatively predicted raw final course grades (bPAP ®Grade = -.114, p<.01) whereas SRS positively predicted raw final course

grades ( fisRS®Grade = 127, p<01) [Hypothesis 2.2].

The modest standardized indirect effect of MMSE on raw final grades was -.020 via PAP ( bMMSE®PAP®Grade = -.020, p = .04) and -.023 via SRS ( f^MMSE® SRS®Grade = -.023, p = .02) [Hypothesis 2.3]. Nevertheless, the standardized direct effect from MMSE on raw final grades was not statistically significant (JbMMSE ®Grade =.002, p = .981). The variance explained was 4.3%

for PAP, 5.3% for SRS, and 5.4% for raw final grades with statistical significance atp < .01 level.

4. Discussion and implications

Technological distractions, such as social websites and media, have become a major challenge for university students' learning (Rosen et al., 2013). This study explored the relationship between MMSE and learning performance within the actual PLE from the perspectives of university students' PAP and SRS while studying using the OL-MARS v.2. Within a PLE, students may be easily attracted to perceptual signals such as social media notifications, fun events, news, videos, and games on the Internet. They may also have the lingering thoughts that something is happening on the Internet and thus their mind wanders. In fact, environmental distraction and mind-wandering are the major causes of attention failure among university students (Unsworth, McMillan, Brewer, & Spillers, 2012). Such disruptions in attention may lead to incomplete coverage of the learning and longer task processing time (Bowman et al., 2010). Therefore, even though we expect that we are capable of multitasking, our memory store may keep an incoherent and only partial representation of the world due to

deficiencies of attention (Noe, Pessoa, & Thompson, 2000; O'Regan, 1992; Rensink, 2002; Rensink, O'Regan, & Clark, 1997; Simons & Levin, 1997).

For those reasons, we validated the theoretical framework of revised OL-MARS using rigorous statistical approaches and investigated university students learning within the PLE through their MMSE and their PAP and SRS, in an attempt to provide a means of practical prevention and intervention. Below we discuss our findings with respect to our two research purposes and respective hypotheses, along with implications for theory development and instructional practices.

4.1. Construct and criterion validity of meta-attention

For the first research purpose and H1, we validated the constructs of meta-attention (shown in Figure 1) by replicating and expanding Wu's (2015) meta-attention framework. The PAP is a second-order factor that subsumes LT, SMN, and PAD. The SRS is the other second-order factor that subsumes OA and BS. Both second-order CFAs showed an adequate fit of the model to the data and provided reliability and validity support for the revised OL-MARS scale. The results of the correlational analysis also provided construct validity as well as criterion validity for the scale. OA had modest correlations with PAP and the subscales. These results provide evidence to support OA in evaluating the emotional outcome per success or failure in attention, which may have implication for intervention from the perspective of attribution retraining discussed in section 4.3. BS had no correlation with students' PAP and the subscales. This finding was in accordance with the complexity of the online learning profile compositions in Wu (2015). Using &-means clustering analysis, Wu (2005) revealed two congruent (i.e., the Unaware and the Hanging on) and three incongruent (i.e., the Motivated strategic, the Non-responsive, and the Self-disciplined) clusters among university students in terms of

awareness and regulation of their attention within the PLE. For example, students in the Motivated strategic cluster may have few perceived attention problems and use more strategies to regulate their attention. Those in the Hanging-on cluster may be aware of their attention problems and use some strategies to regulate their attention. Still others in the Non-response cluster may know well their attention problems but use very few strategies to regulate their attention. Examples of these media-related attention profiles could help explain why an orthogonal relationship exists between PAP and SRS because each group of individuals may have a specific way to respond to their PAPs. Therefore, no linear relationship exists between PAP and SRS.

Results of the correlational analysis among subscales of PAP and SRS with MMSE and with Internet/Facebook uses provided additional criterion validity. Higher level of MMSE was related with higher PAP but lower SRS. Moreover, more time spent in Internet and Facebook-related uses was correlated with higher PAP and lower SRS, providing support for our assertion that PAP and SRS are broader constructs of engagement in multitasking. 4.2. The relationship among MMSE, PAP, SRS, and learning performance

For the second research purpose, we investigated the structural model among MMSE, PAP, SRS, and learning performance within the PLE, as shown in Figure 5. In line with H2.1, as expected, MMSE positively predicted higher PAP and poorer SRS. These two measures of meta-attention could be seen as broader constructs of engagement in multitasking because they reflected students' awareness of their distracted attention and poor regulation to focus on a single task. Therefore, our finding was in accordance with Bandura's self-efficacy theory (1977) that self-efficacy influences people's motivation and choice to engage in an activity. According to Bandura (1993), people are readily to take challenges that they think they are capable to cope with and avoid activities that exceed their competence, thereby affecting the direction of their

personal development. Therefore, MMSE may have a profound influence on university students' attempt to multitasking while studying.

In line with H2.2, higher PAP and poorer SRS (broader constructs of engagement in multitasking) were correlated with poorer learning performance within the PLE. Consistent with previous studies, engagement in multitasking is negatively related with academic achievement (Gaudreau, Miranda, & Gareau, 2014; Junco & Cotten, 2012; Kirschner & Karpinski, 2010). Nevertheless, university students engaged in multitasking may mistakenly assume that they are performing media multitasking without sacrificing efficiency and effectiveness (Kirschner & Karpinski, 2010) or as a strategy of information management (Junco & Cotten, 2012). In the case of media multitasking, use of social media or Internet relevant to the learning tasks (e.g., collaborating with partners or searching for related information) may enhance the learning results whereas interruption of different contexts (e.g., studying while instant messaging) usually comes at cost of frustration, stress, time pressure, and efforts (Mark, Gudith, & Klocke, 2008). In fact, researchers have noted that most people are not competent at multitasking; more often than not, they are switching tasks frequently (Kiesel et al., 2010; Rosen et al., 2013). For example, Dindar and Akbulut (2016) tested undergraduate students' learning performance in concurrent multitasking conditions and sequential multitasking conditions. They found that students obtained poorer retention scores in concurrent multitasking through online chatting and watching instructional videos, compared with the control condition where students watched the instructional video only; nevertheless, sequential multitasking through watching different instructional videos or distraction videos did not interfere with retention. These studies showed that both context and timing of multitasking play a role in students' performance. The current study relied on self-reports of multitasking behavior (shown as PAP and SRS), which can reflect

the overall perception and strategy use on students' divided attention or multitasking. However, self-reported measures could not reflect the context and timing in multitasking. Further study can be conducted by using observations or experimental designs to examine students' actual engagement in multitasking and its relationship with MMSE and learning performance within the PLE.

H2.3 was also supported by the analysis result. Though higher MMSE did not statistically significantly predict poor learning performance, the negative indirect relationship of MMSE with learning performance was statistically significant through both PAP and SRS given modest effect sizes. Compared to the cognitive and metacognitive processes, students' MMSE is relatively less noticed. However, the study results suggested that an elevated MMSE posed a potential threat to online learning as evidenced by the above indirect relationship through the two intervening variables of PAP and SRS (MacKinnon, 2008). Therefore, MMSE may be a main target for prevention and intervention of distracted attention. Challenge to university students' MMSE could be provided to reduce or minimize attention problems. As shown in studies to reduce problem behavior, challenge to people's positive perception was found to be critical in the prevention and intervention of alcohol over-consumption (Mitchell, Beals, & Kaufman, 2006; Reich, Below, & Goldman, 2010) or Internet overuse (Lee, Ko, & Chou, 2015; Lin, Ko, & Wu, 2008). For example, Lau-Barraco and Dunn (2008) tested the effectiveness of a single-session alcohol expectancy challenge intervention on 217 college students and found that students in the experimental group had significantly lower alcohol consumption in the one-month follow-up assessment. In the same vein, instructors employing PLEs for student learning may challenge students' perceived competence in multitasking through lecture-based MMSE challenge intervention programs. Students can be introduced to the constraints and limited capacities of the

human attention system (Cowan, 2008) and be aware of the potential negative effects on efficiency, effectiveness, and emotions of multitasking.

4.3. Implication for prevention and intervention of attention problems within the PLE

We investigated two methods for regulating attention, namely behavioral strategies (BS) and outcome appraisal (OA). Our results revealed that both BS and OA as well as their higher-order SRS construct were positively associated with raw final scores. This finding suggests that attention regulation whether in an explicit (e.g., BS) or an implicit (e.g., OA) form was helpful in learners' learning performance within the PLE. Based on the findings related to BS, we advise that learners create an uninterrupted environment for online learning by closing unrelated windows, disabling the Internet, slowing down the upload or download capacity, or putting off the disrupted task. Specifically, learners with better SRS may put off the interfering task (e.g., social media notifications or instant messages) until the primary task is done so that they can work well with both tasks. For example, Rosen et al. (2011) found that during a video lecture, university students who could wait for four or five minutes to respond text messages did significantly better than those who respond to the messages immediately.

Outcome appraisal was also beneficial for learning performance within the PLE. By linking the perceived success or failure of attention with a positive or negative emotion, learners are engaged in causal attribution of their attention outcome based on Weiner (1985). Therefore, attribution training may be a potential intervention for university students. According to Robertson (2000), "[A]ttribution training is a process that involves improving a person's beliefs in the causes of his or her own failures and successes to promote future motivation for achievement" (p. 111). As shown in Figure 6, learners with a focused attention within the PLE may have higher self-esteem and feel proud of themselves; otherwise, they feel guilty. Following

outcome appraisal was causal attribution. In the case of a focused attention within the PLE, motivation for future improvement of attention may be enhanced by recognizing one's own ability (i.e., attributing to an internal, stable, and uncontrollable cause). In the case of a distracted attention within the PLE, on the other hand, motivation for future improvement of attention may be enhanced by recognizing one's lack of effort (i.e., attributing to an internal, unstable, and controllable cause). As a result, learners may use adaptive behaviors and strategies to adjust their attention state. When learners appraise their attention outcome as high self-esteem, they recognize their competence and ability to regulate their attention and may persist in the endeavor (Robertson, 2000). On the other hand, when learners evaluate their attention outcome as feeling guilty, they recognize that they are responsible for their own learning and should make necessary changes (e.g., increase in efforts) to adjust their attention state. Attributional retraining programs such as those described in Perry, Hechter, Menec, and Weinberg (1993) can be provided for university students, especially for those with high PAP, to foster a positive structure of attributional thinking and thereby enhance their regulation of attention.

5. Limitation and conclusion

The results of this study should be interpreted in light of several limitations. First, meta-attention emphasizes what students know and do about their attention. Therefore, the study findings were based on self-reported measures. Future research should be conducted with data from objective measures such as observational studies, experimental designs, or attentional control tasks to triangulate the results. Second, the study sample was exclusively university students. Thus, the result may not generalize to younger students. With the prevalence of online learning, more and more middle or elementary students are using the Internet for learning. Therefore, future research should be conducted on students of different age groups. Third, the

effect sizes of the hypothesized structural model were modest. The study result indicated that a bolstered MMSE may correlate with poor performance via PAP and SRS. PAP and SRS are people's awareness of their media-related attention problems and regulation of their attention. Unlike cognitive and metacognitive processes that are closer to the learning outcome in terms of temporal order, attention is the onset of the learning process; nevertheless, the cognitive learning activity will not be able to initiate and persist without a focused attention. Thus, the modest effect sizes may be not surprising; instead, results of the study help delve into the issue of digital distraction from the perspectives of MMSE and meta-attention for effective intervention and prevention of attention failure. Lastly, we collected students' responses on the questionnaire in the mid of semester, and obtained their final course grades in the end of semester. The temporal sequence of data collection enables the construction of proximal mediation models (Hoyle & Kenny, 1999) for possible causal inference in the study. However, before we have more evidence, any interpretation about causal inferences in the hypothesized relationship should be drawn with discretion. Future research can be done in a longitudinal fashion (e.g., data collection in distinct time points) to understand the causal relationship.

In our information-knowledge society, the boundary between learning and social entertainment in the online environment is blurred. For example, within a PLE, university students can study, share information, and collaborate with classmates while at the same time continuing with online social activities that are irrelevant for the learning task. Many university students use multitasking as a strategy for managing information (Junco & Cotten, 2012) and are not aware of the cost of multitasking, such as lack of efficiency (Kirschner & Karpinski, 2010) and negative emotions (Mark et al., 2008). The current study adds to the literature regarding the

significance of meta-attention (operationalized as PAP and SRS) in online learning and focuses on the mechanism of performance failure within the PLE from the perspectives of MMSE as well as students' PAP and SRS. Based on our findings that there were indirect relationships between MMSE and learning performance through PAP and SRS, we suggest that the MMSE and the OL-MARS v.2 can be used for monitoring students' competence belief in multitasking as well as their attention state and strategy use for regulation of attention. Moreover, we recommend that strategy training such as use of behavioral strategy and attributional retraining can be provided for university students with media-related attention problems. Finally, we propose that challenges to MMSE can be provided for prevention and intervention of poor media-related attention problem among university students.

6. References

Arbuckle, J. L. (1996). Full information estimation in the presence of incomplete data. In G. A.

Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 243-277). Mahwah, NJ: Lawrence Erlbaum Associates.

Bandura, A. (1977). Self-efficacy: toward a unifying theory of behavioral change. Psychological Review, 84(2), 191.

Bandura, A. (1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28, 117-148.

Bandura, A. (2006). Guide for constructing self-efficacy scales. In F. Pajares & T. C. Urdan (Eds.),. In Self-efficacy beliefs of adolescents (Vol. 5). Greenwich, CT: Information Age Publishing.

Bowman, L. L., Levine, L. E., Waite, B. M., & Gendron, M. (2010). Can students really multitask? An experimental study of instant messaging while reading. Computers & Education, 54(4), 927-931.

Broadbent, D. E. (1958). Perception and communication. London: Pergamon Press.

Brown, T. A. (2006). Confirmatory factor analysis for applied research (1st ed.). New York, NY: The Guilford Press.

Calderwood, C., Ackerman, P. L., & Conklin, E. M. (2014). What else do college students "do" while studying? An investigation of multitasking. Computers & Education, 75, 19-29. http://doi. org/10.1016/j. compedu.2014.02.004

Carrier, L. M., Rosen, L. D., Cheever, N. A., & Lim, A. F. (2015). Causes, effects, and practicalities of everyday multitasking. Developmental Review, 35, 64-78.

Chun, M. M., Golomb, J. D., & Turk-Browne, N. B. (2011). A taxonomy of external and internal attention. Annual Review of Psychology, 62, 73-101.

Colflesh, G. J., & Conway, A. R. (2007). Individual differences in working memory capacity and

divided attention in dichotic listening. Psychonomic Bulletin & Review, 14(4), 699-703. Conway, A. R., Cowan, N., & Bunting, M. F. (2001). The cocktail party phenomenon revisited: The

importance of working memory capacity. Psychonomic Bulletin & Review, 8(2), 331-335. Cowan, N. (2008). What are the differences between long-term, short-term, and working memory?

Progress in Brain Research, 169, 323-338. Dabbagh, N., & Kitsantas, A. (2012). Personal Learning Environments, social media, and

self-regulated learning: A natural formula for connecting formal and informal learning. The Internet and Higher Education, 15(1), 3-8. Dayan, P., Kakade, S., & Montague, P. R. (2000). Learning and selective attention. Nature

Neuroscience, 3(11), 1218. Dindar, M., & Akbulut, Y. (2016). Effects of multitasking on retention and topic interest. Learning

and Instruction, 41, 94-105. http://doi.org/10.1016/j.learninstruc.2015.10.005 Gaudreau, P., Miranda, D., & Gareau, A. (2014). Canadian university students in wireless classrooms: What do they do on their laptops and does it really matter? Computers & Education, 70, 245-255. http://doi.org/10.1016Zj.compedu.2013.08.019 Gillet, D., Law, E. C., & Chatterjee, A. (2010). Personal learning environments in a global higher engineering education Web 2.0 realm. In IEEE Education Engineering (EDUCON) (pp. 897-906). IEEE. Retrieved from

http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5492483 Hancock, G. R., & Mueller, R. O. (2010). The reviewer's guide to quantitative methods in the social

sciences. New York, NY: Routledge. Hathorn, L. G., & Rawson, K. A. (2012). The roles of embedded monitoring requests and questions

in improving mental models of computer-based scientific text. Computers & Education, 59(3), 1021-1031.

Hox, J. J. (2010). Multilevel analysis: Techniques and applications (2nd ed.). New York, NY: Routledge Academic.

Hoyle, R. H., & Kenny, D. A. (1999). Statistical power and tests of mediation. In R. H. Hoyle (Ed.), Statistical strategies for small sample research. Newbury Park: Sage.

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1-55.

Ibos, G., Duhamel, J.-R., & Hamed, S. B. (2013). A functional hierarchy within the parietofrontal network in stimulus selection and attention control. The Journal of Neuroscience, 33(19), 8359-8369.

Iqbal, S. T., & Bailey, B. P. (2006). Leveraging characteristics of task structure to predict the cost of interruption. In Proceedings of the SIGCHI conference on Human Factors in computing systems (pp. 741-750). ACM. Retrieved from http://dl.acm.org/citation.cfm?id=1124882

Johnson, M. W., & Sherlock, D. (2014). Beyond the personal learning environment: Attachment and control in the classroom of the future. Interactive Learning Environments, 22(2), 146-164.

Joreskog, K. G. (1970). A general method for estimating a linear structural equation system (No. RB-70-54) (p. 43). Princeton, NJ: Educational Testing Service.

Junco, R. (2012). In-class multitasking and academic performance. Computers in Human Behavior, 28(6), 2236-2243. http://doi.org/10.1016Zj.chb.2012.06.031

Junco, R., & Cotten, S. R. (2012a). No A 4 U: The relationship between multitasking and academic

performance. Computers & Education, 59(2), 505-514.

Junco, R., & Cotten, S. R. (2012b). No A 4 U: The relationship between multitasking and academic performance. Computers & Education, 59(2), 505-514. http://doi.org/10.10167j.compedu.2011.12.023

Kane, M. J., & Engle, R. W. (2002). The role of prefrontal cortex in working-memory capacity, executive attention, and general fluid intelligence: An individual-differences perspective. Psychonomic Bulletin & Review, 9(4), 637-671.

Karpinski, A. C., Kirschner, P. A., Ozer, I., Mellott, J. A., & Ochwo, P. (2013). An exploration of social networking site use, multitasking, and academic performance among United States and European university students. Computers in Human Behavior, 29(3), 1182-1192.

Kauffman, D. F., Zhao, R., & Yang, Y. S. (2011). Effects of online note-taking formats and self-monitoring prompts on learning from online text: Using technology to enhance self-regulated learning. Contemporary Educational Psychology.

Kiesel, A., Steinhauser, M., Wendt, M., Falkenstein, M., Jost, K., Philipp, A. M., & Koch, I. (2010). Control and interference in task switching—A review. Psychological Bulletin, 136(5), 849.

Kirschner, P. A., & Karpinski, A. C. (2010). Facebook® and academic performance. Computers in Human Behavior, 26(6), 1237-1245. http://doi.org/10.10167j.chb.2010.03.024

Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work: An analysis of the failure of constructivist, discovery, problem-based, experiential, and inquiry-based teaching. Educational Psychologist, 41(2), 75-86.

Kish, L. (1995). Survey sampling. Malden, MA: Wiley-Interscience.

Kline, R. B. (2010). Principles and practice of structural equation modeling (3rd ed.). New York, NY: The Guilford Press.

Kop, R., & Fournier, H. (2011). New dimensions to self-directed learning in an open networked learning environment. International Journal of Self-Directed Learning, 7(2), 2-20.

Kraushaar, J. M., & Novak, D. C. (2010). Examining the affects of student multitasking with laptops during the lecture. Journal of Information Systems Education, 21(2), 241-251.

Kreft, I. G. ., & De Leeuw, J. (1998). Introducing multilevel modeling. London: Sage.

Lau-Barraco, C., & Dunn, M. E. (2008). Evaluation of a single-session expectancy challenge

intervention to reduce alcohol use among college students. Psychology of Addictive Behaviors, 22(2), 168.

Lee, Y.-H. (2015). Facilitating critical thinking using the C-QRAC collaboration script: Enhancing science reading literacy in a computer-supported collaborative learning environment. Computers & Education, 88, 182-191.

Lee, Y.-H., Ko, C.-H., & Chou, C. (2015). Re-visiting Internet addiction among Taiwanese students: A cross-sectional comparison of students' expectations, online gaming, and online social interaction. Journal of Abnormal Child Psychology, 43(3), 589-599. http://doi.org/10.1007/s10802-014-9915-4.

Lee, Y.-H., & Wu, J.-Y. (2013). The indirect effects of online social entertainment and information seeking activities on reading literacy. Computers & Education, 67, 168-177. http://doi. org/10.1016/j. compedu.2013.03.001

Lin, M.-P., Ko, H.-C., & Wu, J. Y.-W. (2008). The role of positive/negative outcome expectancy and refusal self-efficacy of Internet use on Internet addiction among college students in Taiwan. CyberPsychology & Behavior, 11(4), 451-457. http://doi.org/10.1089/cpb.2007.0121

MacKinnon, D. (2008). Introduction to Statistical Mediation Analysis. Routledge Academic.

Mark, G., Gudith, D., & Klocke, U. (2008). The cost of interrupted work: More speed and stress. In Proceedings of the SIGCHI conference on Human Factors in Computing Systems (pp. 107-110). ACM. Retrieved from http://dl.acm.org/citation.cfm?id=1357072

Mitchell, C. M., Beals, J., & Kaufman, C. E. (2006). Alcohol use, outcome expectancies, and HIV risk dtatus among American Indian youth: A latent growth curve model with parallel processes. Journal of Youth and Adolescence, 35(5), 726-737. http://doi.org/10.1007/s10964-006-9103-0

Muthén, B. O. (1994). Multilevel Covariance Structure Analysis. Sociological Methods & Research, 22(3), 376-398.

Muthén, B. O., & Satorra, A. (1995). Complex sample data in structural equation modeling. Sociological Methodology, 25, 267-316.

Muthén, L. K., & Muthén, B. O. (2010). Mplus user's guide (Sixth Edition). Los Angeles, CA: Muthén & Muthén.

Noë, A., Pessoa, L., & Thompson, E. (2000). Beyond the grand illusion: What change blindness really teaches us about vision. Visual Cognition, 7(1-3), 93-106.

O'Regan, J. K. (1992). Solving the "real" mysteries of visual perception: The world as an outside memory. Canadian Journal of Psychology/Revue Canadienne de Psychologie, 46(3), 461.

Pashler, H. E. (1998). The psychology of attention. Cambridge, Mass: MIT Press.

Perna, L. W., Ruby, A., Boruch, R. F., Wang, N., Scull, J., Ahmad, S., & Evans, C. (2014). Moving through MOOCs: Understanding the progression of users in Massive Open Online Courses. Educational Researcher, 0013189X14562423.

Perry, R. P., Hechter, F. J., Menec, V. H., & Weinberg, L. E. (1993). Enhancing achievement motivation and performance in college students: An attributional retraining perspective.

Research in Higher Education, 34(6), 687-723. Petersen, S. E., & Posner, M. I. (2012). The attention system of the human brain: 20 years after.

Annual Review of Neuroscience, 35(1), 73-89. Posner, M. I., & Boies, S. J. (1971). Components of attention. Psychological Review, 78(5), 391. Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Review of

Neuroscience, 13(1), 25-42. http://doi.org/10.1146/annurev.ne.13.030190.000325 Preacher, K., Zhang, Z., & Zyphur, M. (2011). Alternative methods for assessing mediation in multilevel data: The advantages of multilevel SEM. Structural Equation Modeling: A Multidisciplinary Journal, 18(2), 161-182. http://doi.org/10.1080/10705511.2011.557329 Quan-Haase, A. (2010). Self-regulation in instant messaging (IM): Failures, strategies, and negative

consequences. International Journal of E-Collaboration (IJeC), 6(3), 22-42. Ravizza, S. M., Hambrick, D. Z., & Fenn, K. M. (2014). Non-academic internet use in the classroom is negatively related to classroom learning regardless of intellectual ability. Computers & Education, 78, 109-114. http://doi.org/10.1016Zj.compedu.2014.05.007 Reich, R. R., Below, M. C., & Goldman, M. S. (2010). Explicit and implicit measures of expectancy and related alcohol cognitions: a meta-analytic comparison. Psychology of Addictive Behaviors, 24(1), 13-25. Reisberg, D., & McLean, J. (1985). Meta-attention: Do we know when we are being distracted? The

Journal of General Psychology, 112(3), 291-306. Rensink, R. A. (2002). Change detection. Annual Review of Psychology, 53(1), 245-277. Rensink, R. A., O'Regan, J. K., & Clark, J. J. (1997). To see or not to see: The need for attention to perceive changes in scenes. Psychological Science, 8(5), 368-373. http://doi.org/10.1111/j.1467-9280.1997.tb00427.x

Robertson, J. S. (2000). Is attribution training a worthwhile classroom intervention for K-12 students with learning difficulties? Educational Psychology Review, 12(1), 111-134.

Rosen, L. D., Carrier, L. M., & Cheever, N. A. (2013). Facebook and texting made me do it: Media-induced task-switching while studying. Computers in Human Behavior, 29(3), 948-958.

Rosen, L. D., Lim, A. F., Carrier, L. M., & Cheever, N. A. (2011). An empirical examination of the educational impact of text message-induced task switching in the classroom: Educational implications and strategies to enhance learning. Revista de Psicología Educativa, 17(2), 163-177. http://doi.org/10.5093/ed2011v17n2a4

Salvucci, D. D., Taatgen, N. A., & Borst, J. P. (2009). Toward a unified theory of the multitasking continuum: from concurrent performance to task switching, interruption, and resumption. In Proceedings of the SIGCHI conference on human factors in computing systems (pp. 1819-1828). ACM. Retrieved from http://dl.acm.org/citation.cfm?id=1518981

Sanbonmatsu, D. M., Strayer, D. L., Medeiros-Ward, N., & Watson, J. M. (2013). Who multi-tasks and why? Multi-tasking ability, perceived multi-tasking ability, impulsivity, and sensation seeking. PloS One, 8(1), e54402.

Schraw, G., & Sperling Dennison, R. (1994). Assessing metacognitive awareness. Contemporary Educational Psychology, 19, 460-460.

Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in assessing rater reliability. Psychol Bull, 86(2), 420-428.

Simons, D. J., & Levin, D. T. (1997). Change blindness. Trends in Cognitive Sciences, 1(7), 261-267. http://doi.org/10.1016/S 1364-6613(97)01080-2

Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in structural equation

models. Sociological Methodology, 13, 290-312. Tsai, M. J., & Tsai, C. C. (2010). Junior high school students' Internet usage and self-efficacy: A

re-examination of the gender gap. Computers & Education, 54(4), 1182-1192. Unsworth, N., McMillan, B. D., Brewer, G. A., & Spillers, G. J. (2012). Everyday attention failures: An individual differences investigation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38(6), 1765-1772. http://doi.org/10.1037/a0028075 Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological Review, 92(4), 548-573.

Wood, E., Zivcakova, L., Gentile, P., Archer, K., De Pasquale, D., & Nosko, A. (2012). Examining the impact of off-task multi-tasking with technology on real-time classroom learning. Computers & Education, 58(1), 365-374. http://doi.org/10.1016/j.compedu.2011.08.029 Wu, J.-Y. (2014). Gender differences in online reading engagement, metacognitive strategies, navigation skills and reading literacy. Journal of Computer Assisted Learning, 30(3), 252-271. http://doi.org/10.1111/jcal.12054 Wu, J.-Y. (2015). University students' motivated attention and use of regulation strategies on social

media. Computers & Education, 89, 75-90. http://doi.org/10.1016/j.compedu.2015.08.016 Wu, J.-Y., & Kwok, O. (2012). Using structural equation modeling to analyze complex survey data: A comparison between design-based single-level and model-based multi-level approaches. Structural Equation Modeling-A Multidisciplinary Journal, 19(1), 16-3 5. http://doi.org/10.1080/10705511.2012.634703

Figure 1. Proposedframeworkfor the revised Online Learning Motivated Attention and Regulatory Strategies scale (OL-MARS v. 2) based on the theory of meta-attention within the PLE.

Figure 2. Hypothesized model of the relationship between media multitasking self-efficacy, meta-attention and learning performance within the PLE.

Hypothesis 2.1 (H2.1): Higher MMSE is correlated with higher PAP and poorer SRS. Hypothesis 2.2 (H2.2): Higher PAP and poorer SRS predict poorer learning performance within the PLE. Hypothesis 2.3 (H2.3): There are negative indirect relationships of MMSE with learning performance through the two meta-attention constructs. Note. PAD: perceived attention discontinuity; LT: lingering thoughts; SMN: social media notifications; BS: behavioral strategies; OA: Outcome Appraisal.

Media Multitasking Self-Efficacy within the PLE

Perceived competence to perform

simultaneous tasks in multiple media forms within the PLE

Meta-Attention within the PLE

Perceived Attention Problem (PAP)

•SMN

•PAD

Self-Regulation Strategy (SRS)

•OA •BS

Learning Performance within the PLE

Raw final course grades

Figure 3. Second-order confirmatory factory analyses for Perceived Attention Problem (PAP) with three first-order factors, perceived attention discontinuity (PAD), lingering thoughts (LT), and social media notifications (SMN). Unstandardized and standardized (in parentheses) coefficients were presented, and model fit results were %2 = 2837.081, df = 85, p<.01 with CFI = .918, RMSEA = .056, and

SRMR = .061. *p <.01. **p <.001.

.49(1.00)

Perceived Attention Problem (PAP)

,25(.38**)

.91(.79*

.22(.31**)

.99(.89**

.13(.21*)

1.00(.70**l>n PAD1

r95(.81

.91(.69**)

.83(.70**)

.93(.70PAD3 .91(.77**)

67(.50**) 32(.35**) 60(.51**) ,38(.41**)

H PAD5

82(.67**)

71(.61*~)»| PAD7

_.59(.52**) __47(.51**)

.53(.55**)

.56(.63**)

,.59(.46**) .21(.37**) 21(.51**)

,65(.65**)

97(.75**) .21(.41**) ,..60(.50**)

Figure 4. Second-order confirmatory factory analysis for Self-Regulation Strategy (SRS) of attention with two first-order factors, behavioral strategies (BS) and outcome appraisal (OA). Unstandardized and standardized (in parentheses) coefficients were presented, and model fit results were = 40.838, df = 24, p = .02 with CFI = .956, RMSEA = .035, and SRMR = .062. *p <.01. **p <.001.

MEDIA MULTITASKING SELF-EFFICACY & LEARNING PERFORMANCE 48 Figure 5. The Multilevel SEM with between-level random intercept and within-level indirect effect model of Media Multitasking Self-Efficacy on learning performance within PLE through the Perceived Attention Problem and Self-Regulation Strategy. Age and male were used as covariates. Unstandardized and standardized (in parentheses) coefficients were presented, and model fit results were x2 =

48.232, df = 26, p <.01, CFI = .919, RMSEA = .035, SRMR = .035. *p <.05. **p <.01.

Between Level

15.13(1.00**)

Within Level

MEDIA MULTITASKING SELF-EFFICACY & LEARNING PERFORMANCE 50 Figure 6. The strategic process of attention outcome appraisal and the follow-up causal attribution and regulatory behaviors within the PLE.

Attention state

Outcome appraisal

Focused attention

Distracted attention

Causal attribution

High self-esteem or pride

In the case of good attention outcome evaluation, attention

state may be maintained by I/S/UC attribution (e.g., ability).

In the case of poor attention outcome evaluation, attention

state may be enhanced by I/US/C attribution (e.g., efforts).

Desired outcome

Applying adaptive behavior and strategies for attention regulation.

The Functioning of Meta-Attention Components and Their Operational Definitions Within the PLE

Perceived Attention Problem (PAP)

Component in Meta-Attention

Operational Definition within the PLE

Guiding Question

Awareness of Task-Related Attention

Awareness of

Task-Irrelevant

Attention

Awareness of Alerting: The awareness of maintaining a vigilant and alert state for high-priority events.

Lingering Thoughts (LT): "Do I know that I am Learners are aware that

The awareness of expecting learning-unrelated events in PLE

expecting something unrelated with learning while studying?"

they can stay alert and expect things to be

covered and learned in the online learning, (e.g., a learning goal).

Learners are aware that they expect and watch out for updates from close friends' posts on the social media.

Awareness of Orienting: Social Media Notification "Do I know that I am Learners are aware that

The awareness of attention being drawn to sensory stimuli.

(SMN): The awareness of being distracted to sounds, signals, vibrations from social media or digital devices

easily attracted to social media notifications or signals from digital devices?"

their attention was directed to the sensory stimuli relevant to the learning context such as keywords/major terms, diagrams, or video lectures on the Internet.

Learners are aware that they are easily directed or attracted to social media notifications (e.g., sounds or visual signals).

Awareness of Executive control:

The awareness of selecting information for cognitive processing.

Perceived Attention Discontinuity (PAD): The awareness of selecting and processing learning-unrelated information in PLE

"Do I know that I tend to select and process information irrelevant with learning?"

Learners are aware that they tend to select the target or relevant information for further processing (e.g., a diagram illustrating the relationship of the target learning).

Learners are aware that they tend to detect a novel or interesting event on social media (e.g., friends checked in at a fancy restaurant), and then select this information for further processing.

Self-Regulation Strategy of Attention (SRS)

Component in Meta-Attention

Operational Definition Guiding Question

Successful Attention Regulation in the Learning Task

Unsuccessful Attention Regulation in the Learning Task

Behavioral Strategies: BS are characterized by enforcing behavioral control on Internet use, such as access denial and removal of devices for Internet connection.

The use of behavioral control to avoid access to media irrelevant with learning

"What behavioral strategies do I use to regulate my attention in online learning?"

Learners log out Facebook account, close unrelated windows, limit the speed of upload/download, put off disrupted events while studying.

Learners' attention was attracted to social media or webpages and applications unrelated with online learning.

Outcome Appraisal: OA is an implicit strategy that involves linking the perceived success or failure to a positive or negative emotion

The act of linking the perceived success or failure in attention regulation to self-esteem (pride) or guilt

"What is my perceived emotion following a success or failure of attention regulation in online learning?"

Learners feel happy and Learners feel guilty

proud of themselves due to their focused attention in online learning.

due to distracted attention in online learning. If learners do not feel guilty or frustrated, OA is not present.

Item descriptives of Media Multitasking Self-Efficacy (MMSE) scale and revised Online Learning Motivated Attention Regulation and Strategies (OL-MARS v. 2) scale

Note. Latent constructs are capitalized, and second-order latent constructs are capitalized and bolded. PAP= perceived attention problem, PAD = perceived attention discontinuity, LT = lingering thought, SMN = social media notification; SRS = self-regulation strategy, BS = behavioral strategy, and OA = outcome appraisal.

Scale/Item M SD a MIN MAX Skew Kurtosis

Media Multitasking Self-Efficacy Scale

MMSE1 2.544 1.359 .809 1 6 0.586 -0.576

MMSE2 3.386 1.390 1 6 -0.144 -0.891

MMSE3 2.955 1.374 1 6 0.216 -0.941

MMSE4 2.476 1.287 1 6 0.614 -0.474

MMSE5 3.459 1.395 1 6 -0.121 -0.872

Online Learning Motivated Attention and Regulatory Strategies scale

Perceived Attention Problem (PAP)

SMN1 3.349 1.102 .774 1 5 -0.321 -0.506

SMN2 3.327 1.130 1 5 -0.166 -0.815

SMN3 3.170 1.026 1 5 0.027 -0.565

LT1 2.681 1.141 .724 1 5 0.349 -0.783

LT2 2.703 1.046 1 5 0.304 -0.412

LT3 2.951 1.004 1 5 0.245 -0.586

LT4 2.402 1.137 1 5 0.417 -0.672

PAD1 3.413 1.134 .887 1 5 -0.251 -0.671

PAD2 3.320 0.957 1 5 -0.080 -0.500

PAD3 3.296 1.068 1 5 -0.198 -0.648

PAD4 3.473 0.963 1 5 -0.124 -0.474

PAD5 3.225 1.047 1 5 -0.146 -0.471

PAD6 3.230 0.964 1 5 -0.013 -0.500

PAD7 3.670 0.997 1 5 -0.540 -0.138

PAD8 3.521 0.939 1 5 -0.161 -0.479

Self-Regulation Strategy (SRS)

BS1 2.838 1.037 .774 1 5 0.128 -0.547

BS2 3.175 1.072 1 5 -0.136 -0.548

BS3 2.784 1.148 1 5 -0.004 -0.897

BS4 2.940 0.921 1 5 0.226 -0.293

BS5 3.579 0.848 1 5 -0.272 -0.067

BS6 3.458 0.851 1 5 -0.212 0.009

OA1 3.507 1.067 .694 1 5 -0.411 -0.332

OA2 3.376 1.101 1 5 -0.267 -0.641

OA3 3.872 0.933 1 5 -0.615 -0.013

Correlation and Descriptives of OL-MARS v.2 Constructs, Media Multitasking Self-Efficacy, Learning Performance, Demographic Information, and Internet and Social Media Uses within PLE

Note. Latent constructs are capitalized, and second-order latent constructs are capitalized and bolded. Statistically significant correlations at .05 level are bolded. For OL-MARS v.2, PAP = perceived attention problem, PAD = perceived attention discontinuity, LT = lingering thought, SMN = social media notification; SRS = self-regulation strategy, BS = behavioral strategy, and OA = outcome appraisal.

MMSE = Media Multitasking Self-Efficacy.

Final score = Raw final course grade, the index of student's learning performance with the PLE.

Age = participant's age.

Male = participant's sex, 1=male, 0=female.

Nettime = average time spent on the Internet per day from 1 (below 2 hrs) to 5 (above 8 hrs). FByear = years of FB use from 1 (below 1 yr) to 5 (above 4 yrs).

FBtime = average time spent on Facebook per day from 1 (below 30 mins) to 6 (above 4 hrs). FBcheck = average Facebook visits per day from 1 (below 3 visits) to 5 (above 15 visits). FBAvetime = average time spent on Facebook per visit from 1 (below 5 mins) to 6 (above 2 hrs).

11 12 13 14 15 16

1. PAD 1

2. LT .80 1

3. SMN .80 .89 1

4. PAP .90 .95 .97 1

5. BS -.02 .01 -.03 -.02 1

6. OA .22 .19 .20 .22 .71 1

7. SRS .05 .07 .04 .05 .98 .84 1

8. MMSE .18 .19 .18 .20 -.19 -.17 -.20 1

9. Final Grade -.12 -.10 -.11 -.11 .12 .09 .11 -.08 1

10. Age -.07 -.08 -.07 -.07 -.02 .04 .00 -.07 -.12 1

11. Male .08 .10 .09 .09 -.15 -.10 -.15 .03 -.17 .04 1

12. Nettime .22 .16 .17 .19 -.17 -.12 -.16 .21 -.07 .06 .02 1

13. FByear .08 .01 .04 .04 -.07 -.07 -.08 .16 -.10 .08 .05 .23 1

14. FBtime .33 .28 .33 .34 -.13 -.04 -.11 .22 -.12 .02 -.03 .46 .29 1

15. FBcheck .27 .26 .30 .30 -.13 -.09 -.12 .25 -.15 -.04 .13 .28 .25 .57 1

16. FBAvetime .22 .19 .18 .20 -.02 .07 .01 .11 .05 .03 -.10 .15 .06 .46 .16 1

Mean 84.16 21.75 0.27 3.01 3.69 3.07 2.90 2.77

SD 0.62 0.69 0.68 0.45 0.42 0.54 0.31 0.89 7.36 3.77 0.44 1.14 1.08 1.08 1.33 1.29

# of Observations 570 570 570 570 570 570 570 599 643 563 573 411 409 568 410 409

Acknowledgement

This research was partially supported by Grant 105-2511-S-009 -009 -MY3 and 104-2628-H-009 -001 -SS3 from the Ministry of Science and Technology, Taiwan. The author is indebted to the four anonymous reviewers and the editor for their insightful comments in revising this manuscript.

Highlights

• The theoretical framework of meta-attention is validated by 2nd order CFAs.

• Meta-attention includes Perceived Attention Problems and Self-Regulation Strategies.

• PAP and SRS predict courses grades in the Personal Learning Environment.

• Media Multitasking Self-Efficacy indirectly predict course grades via PAP and SRS.

• Strategy training and challenge to MMSE are suggested to avoid distraction in PLE.