حواس پرتی در کلاس: نقش فیس بوک و وظیفه یادگیری اولیه

Abstract While laptops and other Internet accessible technologies facilitate student learning in the classroom, they also increase opportunities for interruptions from off-task social networking sites such as Facebook (FB). A small number of correlational studies have suggested that FB has a detrimental effect on learning performance, however; these studies had neglected to investigate student-engagement in the primary learning task and how this affects task-switching to goal-irrelevant FB intrusions (distractions); and how purposeful deployment of attention to FB (goal-relevant interruptions) affect lecture comprehension on such tasks. This experiment fills a gap in the literature by manipulating lecture interest-value and controls for duration of FB exposure, time of interruption, FB material and the order of FB posts. One hundred and fifty participants were randomly allocated to one of six conditions: (A) no FB intrusions, high-interest (HI) lecture; (B) no FB intrusions, low-interest (LI) lecture (C) goal-relevant FB intrusions, HI lecture (D) goal-relevant FB intrusions, LI lecture (E) goal-irrelevant FB intrusions, HI lecture (F) goal-irrelevant FB intrusions, LI lecture. As predicted, participants were more susceptible to FB distractions when the primary learning task was of low-interest. The study also found that goal-relevant FB intrusions significantly reduced HI lecture comprehension compared to the control condition (A). The results highlight the need for recourses that will help educators increase student engagement with their learning task. Implications for future research are discussed.

. Introduction With the advent and growth in adoption of social technologies over the last decade, social networking on the Internet while working or studying has progressively pervaded many people’s daily lives. Time spent multitasking with these activities is particularly significant in the student population: data from one study estimated that students multitask for approximately 42% of class time (Kraushaar, 2010). In particular, Facebook (FB) has become a compelling source of classroom distraction. It has been reported to be the most used multitasking distractor for university students in the classroom than other technologies such as text messaging, MSN and emails (Wood et al., 2012). With more than 90% of FB users being teens and university students (Junco, 2012b), not surprisingly, researchers and academics are interested in understanding how FB interruptions impact students’ comprehension of the primary study task. So far, a small number of studies have revealed a negative relationship between FB use and academic performance (Frein et al., 2013, Junco, 2012b and Kirschner and Karpinski, 2010). The present study aims to expand on the theoretical and empirical findings on distractions and multitasking, specifically with FB, by focusing on three questions that remain unanswered: firstly, how do FB intrusions distract students? That is, do less engaging primary tasks determine attentional selection of FB distractions? This is important to understand as educators may be able to reduce student susceptibility to distractions in their classroom; by making their material more engaging and/or interactive ( Sana, Weston, & Cepeda, 2013). Secondly, which features of FB serve as particularly salient multitask distracters? An answer to this question would help us understand what types of devices or platforms promote greater attention and engagement with the distraction, and hence are more likely to divert students’ attention in the lecture theatre. Thirdly, does attention to externally driven FB-interruptions predict performance detriments on learning performance for lectures of high as well as low interest? Research on this issue of multitasking and its negative impact on task performance, as previously suggested ( Junco, 2012a and Wood et al., 2012), needs further scrutiny, as this effect may not necessarily apply to; all types of learning tasks, external interruptions (as opposed to internal interruptions) or to students who are frequent, and potentially skilled, multitaskers with FB. 1.1. Definitions: interruptions, multitasking and distractions While listening to a lecture, attention can be directed to the FB newsfeed via external interruptions, such as a FB sound-alert for a notification and the appearance of stimuli on an automatically refreshed newsfeed. Alternatively, interruptions can be internally generated, that is, they can be self-initiated task switching motivated by the propensity to temporarily abandon a task that is no longer rewarding (Adler and Benbunan-Fich, 2013, Cades et al., 2010 and Payne et al., 2007). Regardless of its source, interruptions occur when students stop their current learning activity and shift goals to perform a different task (Mark, Gonzalez, & Harris, 2005), where a goal can be defined as ‘‘a mental representation of an intention to accomplish a task,… or take some mental or physical action’’ (Altmann & Trafton, 2002, p. 39). Multitasking, specifically dual-tasking, occurs when the FB and learning goals are activated concurrently (Altmann & Trafton, 2002). Although multitasking is synonymous with divided attention, unless a task is automated, multitasking should more accurately be understood as switching attention between concurrent tasks, with only one stimulus attended to at any given time (Jeong & Hwang, 2012). For this reason, it has been argued that multitasking is an imperfect form of attention, as information from one task may be undetected while attending to another and therefore, subject to dual-task slowing (Bowman et al., 2010, Coens et al., 2011, Junco, 2012a and Sana et al., 2013). In contrast, distractions are intended to be ignored. That is, when FB is open in the background and is not an activated goal, the primary learning task can be selectively processed whilst ignoring irrelevant stimuli presented on FB, as intended; however, when the individual allocates limited processing resources to the newsfeed, attention is shifted making FB the onset of the target processing (Parmentier, Ljungberg, Elsley, & Lindkvist, 2011). This inability to reject goal-irrelevant stimuli (irrelevant processing) marks a failure of focused attention and results in distraction (Pool, Koolstra, & Van Der Voort, 2003), causing interruptions to the primary learning task, however brief (Groff, Baron, & Moore, 1983). 1.2. Theories and models of selective attention: how is attention allocated? In order to understand how students get distracted in the lecture theatre, it is important to understand attention. When attention is focused on a certain location (i.e. a learning task), involuntary (bottom-up) and voluntary (top-down) selection mechanisms are involved in shifting attention to distracting stimuli (Vandierendonck, Demanet, Liefooghe, & Verbruggen, 2012). Specifically, novelty-driven and goal-driven mechanisms are prominent models of attentional selection, which suggest that FB intrusions are selected due to its physical salience and by current selection goals, respectively (Awh et al., 2012 and Theeuwes, 2006). According to the novelty-driven mechanism of selection, attention may be “captured” by FB stimuli in an involuntary, stimulus-driven manner (Parmentier et al., 2010 and Parmentier et al., 2011). Neurological studies have advocated that the registration of novel events trigger an automatic novelty-detection response (Picton, Alain, Otten, Ritter, & Achim, 2000), followed by an involuntary re-orientation negativity response (RON), when the participants are engaged in a focal task (Berti and Schroger, 2003 and Berti and Schroger, 2004). This novelty distraction mechanism has also been observed when the target and distracter were temporally decoupled and presented in different sensory modalities, specifically, the visual and auditory modalities (Andres, Parmentier, & Escera, 2006). Thus, this model suggests that goal-irrelevant FB intrusions may be selected by attentional mechanisms involuntarily, based on its perceptual properties such as its sudden appearance (Theeuwes, 2006). On the other hand, goal-irrelevant FB intrusions may be selected based on strategic settings (Mackie, Van Dam, & Fan, 2013). Specifically, Norman and Shallice’s (1986) model of attention proposed that attention subserves cognitive control by modulating information processing in a goal-consistent manner. This suggests that attention may be voluntarily sustained to the lecture during FB external interference; or purposefully deployed to FB if it is thought to be temporarily more important than the learning goals. A wealth of research has supported this notion, showing that people can reallocate cognitive resources to support the higher priority task (Horrey, 2006 and Levy, 2008). Thus, attention may be strategically oriented to FB based on its priority. Although these two mechanisms have been well defined in the literature (Chica, 2013), it is still unclear whether goals or salience play the more dominant role in determining which stimulus is selected via attention (Anderson & Folk, 2010; Belopolsky, Schreij, & Theeuwes, 2010; Theeuwes, 2010). Moreover, these models cannot explain all cases of selection biases, for example, why a stimulus associated with reward can capture attention more readily than another equally salient stimulus that does not have a history of association with rewards, even when this selection bias contradicts current selection goals (Awh et al., 2012 and Chelazzi et al., 2013). Therefore, a growing body of literature from recent years have proposed that rewards have a direct influence on the computation of attentional priority that is independent of the novelty-driven and goal-driven mechanisms (Awh et al., 2012, Chelazzi et al., 2013, Engelmann et al., 2009, Engelmann and Pessoa, 2007 and Krebs et al., 2010) —what Anderson has referred to as ‘value-driven attentional selection’ (Anderson, 2013). According to the value-driven mechanism, stimulus selection operates by maximizing rewards and minimizing losses (Anderson, 2013). From an evolutionary perspective such a system is necessary in order to promote the survival and wellbeing of an organism (Anderson, 2013). By modulating information processing in a reward-driven manner, this attentional system allows survival-promoting stimuli to reach awareness and become available to resource-limited cognitive systems, such as working memory and decision-making, so that it can be subsequently acted upon (Anderson, 2013). This model of selection has been evident in recent research, which have revealed that rewards exert a strong influence on stimulus-processing and response-selection pathways (Engelmann and Pessoa, 2007 and Engelmann et al., 2009). In particular, a recent study conducted by Krebs et al. (2010), revealed that participants had lower error rates and faster responses in correctly naming word ink colors in a Stroop task when the ink color was associated with monetary incentives than trials with non-rewarded colors. This finding was further supported by associated increases in neural activity in the reward-anticipation response area of the brain (i.e. the nucleus accumbens) (Krebs, 2011). It was also found that irrelevant reward associations (i.e. word meaning related to reward- predicting ink colors) impaired performance, which suggested a transfer of reward-related saliency to the task-irrelevant dimension, thereby representing the reward-driven nature of attention in selection and stimulus-processing (Krebs, 2011). Further evidence has indicated that individuals are able to rapidly choose items associated with monetary rewards from a brief display containing several distractors, regardless of the type of motor response used to express the choices (Navalpakkam, Koch, Rangel, & Perona, 2010). Overall, these studies suggest that attentional resources have access to the overall priority map and can be systematically oriented toward the spatial locations that are associated to the maximum reward. In light of this, goal-irrelevant FB intrusions may be automatically selected by attentional mechanisms due to its history of learned associations with reward by students, who are frequent FB users. Alternatively, FB may be selected only when individuals are disengaged in the learning task, to gain momentary rewards (Fisher, 1998). This notion is supported in Pempek, Yermolayeva, and Calvert’s (2009) survey study about why students use FB, which revealed that students rated having fun (38.46% “quite a bit” and 41.30% “a whole lot”), taking a break from work (31.52% “quite a bit” and 39.13% “a whole lot”), and fighting boredom (31.52% “quite a bit” and 39.13% “a whole lot”) as some of the main reasons they use FB (Pempek et al., 2009). This suggests that students’ self-initiate task switching between FB and learning tasks based on the intrinsic rewards of each activity; and it is conjectured that students selectively allocate attentional resources in the same way during instances of external interference. 1.3. Which features of Facebook are particularly distracting? Whilst it is important to understand what determines task switching to distractors, it is also interesting to know which features make FB particularly distracting. One major finding in the Wood et al. (2012) study was that FB served as a more salient multi-tasking distractor than email and text messaging, however; the researchers did not examine the specific features of FB that cause users to be distracted by FB more than these other technologies. It can be argued that interesting features, such as the pictures, make FB particularly distracting when compared to those less dynamic or interesting platforms that rely primarily on verbal information presented in a plain background (Wood et al., 2012). This has been supported in Pieters and Wedel’s (2004) study using an eye tracking device, which revealed that pictorials were superior in capturing attention during ad exploration compared to text based information, regardless of its size (Pieters & Wedel, 2004). Hence, pictorial features on FB may command more attention than text elements such as status updates. Research has also suggested that images increase attentional capture, but only when accompanied by a text message about the image (Brown et al., 2013, Perego et al., 2010 and Wiebe and Annetta, 2008). This is evident in a recent study conducted by Brown et al. (2013), who found that attentional biases were demonstrated for cigarette warnings only when they included a text caption about the image. This supports the argument that, when text and graphic stimuli serve a unified instructional goal, processing is faster ( Wiebe & Annetta, 2008) and relatively effortless ( Perego et al., 2010), which consequently facilitates responses to those stimuli. Attentional biases to FB photos may therefore only occur in the presence of a photo comment. 1.4. How does Facebook multitasking affect learning performance? Clearly, FB offers a variety of intrinsically interesting stimuli to view and explore, which makes it particularly distracting, however; attention to FB interruptions may affect learning performance on the primary study task. Recent research on university students have revealed that higher levels of FB use were strongly and significantly associated with lower grades (Junco, 2012b) and that that overall GPA dropped .12 points for every 93 min above the average of 106 min per day spent on FB (Junco & Cotten, 2011). More recently, Frein et al. (2013) found that high FB users (defined as being logged on for an hour or more per day) scored significantly lower on a free recall memory test than low FB users, which suggests that simply being logged on to FB and passively reading or viewing others’ posts may contribute to the relationship between FB use and reduced memory on the primary study task. It is possible that the negative relationship found in these correlational studies is an indication of the deleterious effect of trying to implement these two processes at the same time, as FB use is something that students often engage in concurrently with studying or other activities that may enhance their academic performance (Junco & Cotten, 2011). Facebook would also be a concurrent activated goal if the interruption were an important FB message or post that requires immediate attention. Wood et al.’s (2012)experiment supported these results, which revealed that students who used FB during a college lecture scored significantly lower on an exam than students who were not allowed to access FB. 1.4.1. Theories of dual-task performance Various theories have been put forward and researched in order to understand how two tasks can be performed simultaneously and hence, how they may interfere with each other as shown in Wood et al.’s (2012) study. Some researchers propose a theory of attention that views it as a single resource, regardless of which sense it stimulates or mental capacity it exercises (Kahneman, 1973). Others have advocated a multiple resource theory (MRT), which argues that humans have several independent supplies of resources that can be used at the same time (Wickens, 1984). A further development of the MRT can be seen in threaded cognition theory (TCT) (Salvucci & Taatgen, 2008), which suggests that each single resource can only execute one task at a time. Therefore interference would only occur when the need for a particular resource by more than one task produces a processing bottleneck (Borst, 2010). In applying these theories to the issue of goal-relevant FB interruptions (interruptions that require immediate attention), if more attention to text-based information on FB predicted poorer performance on a lecture-learning task, then MRT and TCT theories would gain support, as both tasks would be thought to be demanding the resource that processes verbal information. However, if attention to images on FB predicted poorer learning performance, this would suggest that interruptions from within different cognitive resources of the task can decrease a task’s performance, and so, a single recourse theory would seem to be a more accurate account of the interruption effect. 1.4.2. Practice effects However, some researchers have demonstrated conditions under which these effects can be overcome. For example, Meyer (1995) proposed an Executive-Process/Interactive-Control (EPIC) model of dual-task interference, where practice plays an important role. Specifically, this model posits that practice facilitates the conversion of declarative knowledge into procedural knowledge, which subsequently allows two tasks to be performed simultaneously (Meyer, 1995 and Schumacher et al., 2001). Alternatively, practice can allow tasks to be processed more rapidly and thereby facilitate more efficient multitasking (Dux et al., 2009). Specifically, Dux and colleagues revealed that training reduced reaction times and errors in completing behavioral tasks under both single- and dual-task conditions. Training also increased the speed of information processing in these conditions in the prefrontal cortex. Thus, the slowing of performance associated with a cognitive bottleneck can potentially be circumvented due to practice effects attained by students, who are frequent FB multitaskers. 1.5. Task interest-value and performance It is also important however, to consider student-interest in the primary learning task when assessing how FB interruptions affect learning performance (Silvia, McCord, & Gendolla, 2010). Student boredom within the lecture theatre has been shown to be associated with lower GPA, and tendency to adopt coping strategies such as daydreaming (75.4%), switching off (61.6%), and sending text messages during lecture time (45.5%) (Mann, 2009). More recently, Adler and Benbunan-Fich (2013) revealed that participants who experienced feelings of boredom, frustration and exhaustion with a Sudoku task stopped their work more often than participants who did not encounter these feelings, and that higher numbers of self-interruptions were associated with lower performance on the primary Sudoku task. Therefore, participant-engagement in the lecture should also be assessed when examining the impact of goal-relevant FB interruptions on learning performance. 1.6. Summary of the present study 1.6.1. Aims and rationale The current study has three specific aims: the first is to explore how goal-irrelevant FB intrusions (distractions) are selected by attentional mechanisms in an educational setting. Specifically, building on the work of Engelmann et al., 2009, Engelmann and Pessoa, 2007, Krebs, 2011 and Krebs et al., 2010 and Navalpakkam et al. (2010) the present study is designed to investigate whether attentional resources are oriented toward spatial locations that are associated with the maximum reward, which will be achieved by manipulating the interest value of the lecture. According to Anderson’s (2013) value-driven mechanism of attentional selection, attention should be shifted to FB during FB external intrusions when the intrinsic rewards of the lecture are low, and attention should be sustained to the lecture when the lecture is engaging. By investigating this effect, the present study fills a gap in the literature, which will help uncover how distractions are selected by attentional mechanisms in an educational context. The second aim of this study is to explore which stimuli of FB are selected by attentional mechanisms, as suggested by Wood et al.’s (2012) study. This will be achieved by presenting the same stimuli to all participants, in the relevant conditions, and then later testing their memory of those images and text-based posts. Both Pieters and Wedel (2004) and Brown et al. (2013) demonstrated and agreed that images command more attention; however, Brown et al. (2013) showed that this attentional bias only occurred when the image was accompanied by a text message about the image. In order to resolve the discrepancy of their results and understand which features of FB are attended to, all three types of stimuli, namely, text-only stimuli (status updates), image-only stimuli (photos) and an image accompanied by relevant text (a photo with a comment) will be presented and tested in this study. The third and final goal of this study is to investigate the effect of goal-relevant FB interruptions on memory and comprehension of the primary study task. Prominent models of dual-task performance are single and multiple resource theories, which both share the assumption that the cognitive capacity to attend to more than one task simultaneously is limited, with performance detriments as a consequence. A small number of prior studies have supported these theories, demonstrating a negative relationship between FB use and grade point average (GPA) ( Junco, 2012b and Kirschner and Karpinski, 2010); however, these studies were limited by their correlational designs and they did not investigate whether FB use was something that the students did concurrently with studying. Whilst the Wood et al. (2012) study is the first in the field to adopt an experimental design to examine multitasking with FB in the classroom; as in the above studies, they had failed to manipulate interest-value of the lecture and include a measure of time spent on FB as a control variable. To examine whether simply passively reading or viewing others’ posts on FB may contribute to the relationship between FB use and reduced memory on the primary study task, as suggested by Frein et al. (2013), this study will examine the effect of goal-relevant FB intrusions on learning performance by manipulating lecture interest-value and controlling for how much time students spend on FB. The three distinct conditions; goal-relevant FB intrusions, goal-irrelevant FB intrusions and no intrusions (controls); were created to explore attentional selection and interruption effects under conditions that students typically use FB while listening to a lecture. That is, the situations when (1) students purposefully attend to FB and the lecture simultaneously (both tasks are active goals), (2) when students have FB open in the background but do not intend to attend to FB (lecture learning is the only active goal) and, (3) when students do not have FB open at all; respectively. If conditions under which students receive goal-relevant FB interruptions do not reveal reduced lecture comprehension scores than controls, this may be due to practice effects, as suggested by Dux et al. (2009) and Meyer’s (1995) Executive-Process/Interactive-Control (EPIC) model of dual-task interference. On the contrary, Ophir, Nass, Wagner, and Posner (2009) revealed that heavy media multitaskers performed more poorly on a test of task-switching ability than light media multitaskers. To resolve whether practice decreases performance on either task (Ophir et al., 2009), or increases performance due to better multisensory processing (Lui & Wong, 2012), the current study included a polychroncity scale and a measure of dual-task habits while studying in addition to the dependent variable of score on a test of comprehension. Facebook dual-task habits will also be correlated with FB score to investigate whether previously reward-associated stimuli exert a stronger influence on stimulus selection than novel stimuli that provide individuals with momentary rewards, in Anderson’s (2013) value-driven mechanism of attentional selection. 1.6.2. Hypotheses H1. In the conditions where participants are faced with goal-irrelevant FB intrusions; participants listening to the LI lecture will score significantly higher on the FB quiz than students listening to the HI lecture. H2. All participants that receive FB intrusions will be significantly more likely to correctly identify the presence of the image-with-photo-comment than image-only stimuli. They will also be more likely to correctly identify the presence of image-only stimuli than text-only stimuli. H3. Participants that receive goal-relevant FB interruptions should score significantly lower on the lecture comprehension quiz than students who do not receive FB interruptions, for both lectures.

Results 3.1. Preliminary analysis On a scale from 0 (never) to 21 (very often), the mean attendance to live lectures reported by participants was 14.68 (SD = 5.92) and the mean use of online lecture was 11.50 (SD = 6.36). To ensure that the two lectures differed in interest-value, a manipulation check was carried out. On a 4-point Likert scale, the results indicated that 94% of participants allocated to the HI lecture agreed or strongly agreed that they enjoyed listening to the lecture; in contrast, only 19% of participants allocated to the LI lecture rated the same level of enjoyment. The results revealed a significant difference in level of enjoyment between the two lectures F(5, 144) = 26.98, p = .000, partial η2 = .484 and difficulty in maintaining concentration F(5, 144) = 15.18, p = .000, partial η2 = .325. Moreover, 76% of participants reported to be ‘very unfamiliar’ or ‘unfamiliar’ with the lecture material, and there was also a non-significant difference in familiarity between the two lectures F(1, 44) .727, p = .395. To measure the ecological validity of the study, the relevant questionnaire responses were examined. Eighty percent of participants who received FB interruptions agreed or strongly agreed that the experience of switching between FB and the lecture was realistic of what they usually experience; and 75% of participants agreed or strongly agreed that they enjoyed viewing the FB posts. Moreover, of the 99% of participants who said they knew who Angelina Jolie was and the 98% of participants who knew who Brad Pitt was; 70% of participants agreed or strongly agreed that they were interested in what Angelina posted on FB and 76% of participants revealed the same level of interest in what Brad posted on FB. In regard to the novelty of the FB stimuli, 94% of participants reported to have never seen the image in Fig. 1, 95% of participants had never seen the image in Fig. 2, and 73% of participants had never seen the image in Fig. 3, before the study. Moreover, only three participants had read Angelina’s status update before the study, only three reported to have read Brad’s status update and only two reported to have read the photo comment before the experiment. Therefore lecture interest-value was successfully manipulated and the ecological validity of the study was good in relation to the nature of task switching, lecture familiarity, participants’ enjoyment of viewing the FB material and the realism of the FB posts. 3.2. Facebook use All participants were asked how often they have FB open while engaging in University related work. On a 5-point rating scale (1 = never, 5 = always), only 6 participants, on average, reported to ‘never’ have FB open while studying, working and listening to live and online lectures; and all participants reported that they receive FB notifications at least once a day, with 79% of participants reporting that they receive more than six notifications to their mobile device or computer while working or studying each day. As shown in Table 1, university students tend to self-initiate task switching with FB more frequently than other forms of communication such as SC, Instagram and emails. In regard to externally driven interruptions (see Table 2), participants also reported that they would be more likely to immediately attend to a FB chat message, FB photo comment, FB wall-post and an FB image wall-post, than immediately attend to an email or instant message e.g. via MSN. Similar results were obtained for their likelihood of immediately replying to FB messages compared to other technologies. Table 1. Percentages of participants that self-initiate task switching with social media while working or studying. Technology Never 1–3 times 4–6 times 7–9 times >10 times FB 7.3 52.0 32.0 6.0 2.7 SC 48.7 22.7 8.7 .7 0 Instagram 29.3 29.3 9.3 4.0 1.3 Emails 37.3 53.3 7.3 2.0 0 Table options Table 2. Percentages of participants that immediately attend to and reply to social media intrusions while working or studying. Communication mode Action Very unlikely Unlikely Likely Very likely IM Attend 16.7 24.7 17.3 26.7 Reply 25.3 21.3 20.0 17.3 TM Attend 5.0 10.3 28.7 53.0 Reply 2.0 10.0 40.0 48.0 Email Attend 18.0 38.7 30.7 11.3 Reply 34.0 49.3 14.0 2.7 FB chat message Attend 5.3 10.0 36.7 46.7 Reply 6.7 23.3 34.7 34.0 FB photo comment Attend 10.0 24.7 32.0 33.3 Reply 22.0 41.3 24.7 12.0 FB wall-post Attend 4.7 18.7 30.7 46.0 Reply 12.0 36.0 32.0 20.0 FB image wall post Attend 3.3 9.3 28.0 59.3 Reply 9.3 26.7 43.3 20.0 Table options 3.3. Analysis model, alpha level and covariates The hypotheses of the present study were tested using a one-way analyses of covariance (ANCOVA) implemented by the GLM. The alpha level for significance for the overall test was set at .05, and because the contrasts were both orthogonal and planned, there was no further requisite to control Type 1 error rate beyond .05. Pearson correlations were used to investigate plausible covariates for FB scores and lecture comprehension scores (see Table 3). Facebook score only had significant main effects with the covariates, live lecture attendance, F(1, 93) = 9.46, r = −.32, p = .003, speed of viewing all FB stimuli, F(1, 93) = 9.39, r = .31, p = .003, and tendency to immediately attend to an FB interruption, F(1, 93) = 9.11, r = .27, p = .003, when all other significant correlates were controlled for. Accordingly, these variables were controlled for in the one-way ANCOVA with FB score as the dependent variable ( Hypothesis 1). For the same reason, ESL, F(1, 69) = 9.53, r = −.32, p = .003, and lecture difficulty, F(1, 69) = 22.79, r = −.42, p < .001, were controlled for in the one-way ANCOVA with LI lecture comprehension as the dependent variable a ( Hypothesis 3). The HI lecture comprehension, however, had no significant predictors; therefore, a simple one-way analysis of variance (ANOVA) was conducted for HI lecture comprehension as the dependent variable. Hypothesis 2 was tested using a simple Pearson’s chi-squared analysis. Table 3. Bivariate correlations of FB score and lecture comprehension. Possible covariate FB score HI lecture comprehension LI lecture comprehension 1 Age −.13 −.06 .25⁎ 2 Gender .11 −.07 −.10 3 ESL −.05 −.24⁎ −.32⁎⁎ 4 IEC frequency −.03 .17 −.03 5 FB notification frequency .03 .17 −.27⁎ 6 Live lecture attendance −.32⁎⁎ −.31⁎⁎ −.17 7 Ilecture use .14 .13 .04 8 Pause ilearn tendency .20⁎ −.26⁎ .05 9 Immediately attend FB .27⁎⁎ −.11 .10 10 Immediately attend IM −.01 .09 −.23⁎ 11 Immediately attend TM −.01 .01 −.11 12 Immediately attend email .11 −.06 −.04 13 Browse FB dual-task .07 .21 −.04 14 Browse SC dual-task .06 .07 .27⁎ 15 Browse instagram dual-task .07 .11 −.15 16 Browse email dual-task −.141 −.12 −.03 17 Speed of viewing FB posts .31⁎⁎ .09 −.25 18 Polychronicity .11 .22 .20 19 FB material familiarity .34⁎⁎ −.14 .19 20 Lecture familiarity −.08 −.07 .24⁎ 21 TV/music multitask .21⁎ .24⁎ −.13 22 FB enjoyment .42⁎⁎ .15 −.22 23 Lecture enjoyment −.15 .32⁎⁎ .07 24 Lecture difficulty .04 −.27⁎ −.42⁎⁎ ⁎ p < .05. ⁎⁎ p < .01. Table options 3.4. Data check and assumptions The assumption of normality was met for FB quiz score and the HI lecture comprehension, which were checked by inspecting the descriptive statistics, histograms, normal plots of the residuals and the significance of Shapiro–Wilk tests of normality (p > .05). Although normality was violated and skewed distributions were found for the LI lecture comprehension, the results based on the original datasets are reported, as the transformed distributions offered no improvements to the analysis outcomes. The homogeneity of variance assumption was met by the presence of non-significant p-values (p > .05) for Levene’s test for each ANCOVA and ANOVA. Moreover, the assumption of independence was met due to the random allocation of participants to conditions. The ANCOVA specific homogeneity-of-regression (slopes) assumption was met due to the fact that there was no interaction between condition and each of the covariates in the prediction of FB score and in the prediction of the LI lecture comprehension. A linear relationship between the covariates and dependent variables was confirmed by graphic analysis of scatter plots fitted with least squares regression lines. There were 25 participants in each condition, therefore the sample size was adequate for the simple Pearson’s chi-squared analysis, and all observations were independent of each other. Moreover, all expected counts in the two-by-two table were above 10; therefore the assumption of adequate expected cell counts was also met for the simple Pearson’s chi-squared analysis for Hypothesis 2. 3.5. Hypothesis 1 and 2: Facebook score Mean FB quiz scores between each of the four conditions, were analysed using a one-way ANCOVA, controlling for live lecture attendance, mean of 15.42, speed of viewing all FB stimuli, mean of 9.42, and tendency to immediately attend to FB when interrupted, mean of 12.61. The results showed that participants in condition F (LI lecture, 1 goal) had significantly higher FB scores (M = 11.28, SD = 4.49) than participants in condition E (HI lecture, 1 goal) (M = 9.52, SD = 4.88), F(1, 93) = 8.55, p = .035. Moreover, participants in condition C (HI lecture, 2 goals) had significantly higher FB scores (M = 14.60, SD = 3.61) than participants in condition E (HI lecture, 1 goal) (M = 9.52, SD = 4.88), F(1, 93) = 16.32, p = .000. Similarly, participants in condition D (LI lecture, 2 goals) had significantly higher FB scores (M = 15.76, SD = 4.48) than participants in condition F (LI lecture, 1 goal) (M = 11.28, SD = 4.48), F(1, 93) = 16.08, p = .001 (see Fig. 4). Mean FB quiz score for each condition receiving FB interruptions. Error bars ... Fig. 4. Mean FB quiz score for each condition receiving FB interruptions. Error bars represent 95% confidence intervals. Figure options Additionally, across all four conditions with FB interruptions, difficulty in maintaining concentration on the lecture (measure of interestingness) had a weak positive correlation with FB quiz score (r = .248, p = .013) (see Fig. 5). The relationship between difficulty in maintaining concentration on the lecture ... Fig. 5. The relationship between difficulty in maintaining concentration on the lecture and FB quiz score. Figure options Planned contrasts revealed that participants were no more likely to correctly identify the image-with-text stimuli than image-only stimuli, χ2(1, N = 100) = 0.22, p = .643; and there was no difference between the number of correctly identified image-with-text stimuli than text-only posts (status updates), χ2(1, N = 100) = 2.45, p = .118. On the contrary, participants were less likely to correctly identify image-only stimuli than text-only stimuli, χ2(1, N = 100) = 8.97, p = .003, (see Fig. 6). However, these results should be interpreted with caution, as there was an approximately 2-s delay in presenting the images at the same time as the status updates, due to slow Internet connectivity. The number of participants that correctly identified the presence of text-only ... Fig. 6. The number of participants that correctly identified the presence of text-only stimuli, image-only stimuli, both text and image stimuli and neither types of stimuli on the FB newsfeed. Figure options Interestingly, participants were more likely to correctly identify the presence of the status updates in the LI lecture conditions than the HI lecture conditions, χ2(1, N = 100) = 11.01, p = .001. In contrast, participants were no more likely to correctly identify the two images as present in the LI lecture conditions than the HI lecture conditions, χ2(1, N = 100) = 0.01, p = .947. 3.6. Hypothesis 3: comprehension scores The overall effect of condition on the LI lecture comprehension was non-significant, F(2, 75) = 0.77, p = .469, controlling for ESL and lecture difficulty. Thus, no follow-up tests were conducted. However, planned contrasts of mean HI lecture comprehension revealed that, participants who received goal-relevant FB interruptions reported significantly lower comprehension scores (M = 9.56, SD = 2.60) than participants who were not interrupted by FB (M = 11.48, SD = 1.782), F (2, 69) = 9.01, p = .004, (see Fig. 7). The results also revealed a significant difference in comprehension between the LI lecture controls (M = 6.84, SD = 2.56) and HI lecture controls (M = 11.48, SD = 1.78), t(48) = −7.43, p < .001. Mean comprehension score for each condition listening to the LI lecture (left) ... Fig. 7. Mean comprehension score for each condition listening to the LI lecture (left) and HI lecture (right). Error bars represent 95% confidence intervals. Figure options 3.7. Practice effects As shown in Table 4 below, neither FB dual-task habits nor polychronicity scores had significant correlations with comprehension and performance on the FB quiz. Table 4. Correlations between FB dual-tasking and polychronicity, and FB and comprehension Scores. Variable FB score LI comprehension HI comprehension FB dual-tasking .01 .03 .02 Polychronicity .11 .20 .28