Student Acceptance of Tablets in Secondary Education: A Longitudinal Study, Exercises of Technology

Download Student Acceptance of Tablets in Secondary Education: A Longitudinal Study and more Exercises Technology in PDF only on Docsity! Elsevier Editorial System(tm) for Computers in Human Behavior Manuscript Draft Manuscript Number: CHB-D-13-00730R1 Title: Student Acceptance of Tablet Devices in Secondary Education: A three-wave longitudinal cross- lagged case study Article Type: Full Length Article Keywords: tablet devices; secondary education; cross-lagged longitudinal research; theory of planned behavior; pre- and post-adoption Corresponding Author: Dr. Cedric Courtois, Corresponding Author's Institution: iMinds-MICT-Ghent University First Author: Cedric Courtois Order of Authors: Cedric Courtois; Hannelore Montrieux; Frederik De Grove; Annelies Raes; Lieven De Marez; Tammy Schellens Abstract: As ICT is increasingly permeating all aspects of everyday life, it is apparent that education cannot leap behind. In this article we longitudinally investigate a much-debated obligatory full-scale implementation of tablet devices in a large secondary school. We adopt a Theory of Planned Behavior (TPB) approach to verify the dynamic nature of students' acceptance of the tablet as a learning tool at three waves of data collection, both at pre- and short and long-term post-adoption stages. The results clearly indicate the evolutionary nature of the acceptance process, challenging the adequacy of cross- sectional approaches to technology adoption. In the pre-adoption stage, attitude appears as a key uptake factor, whereas three months later, due to practical and technical constraints, the attention shifts to subjective norm and perceived behavioral control. Finally, six months after introduction indicative traces of habituation appear, raising concerns on the suitability of the TPB in established post-adoption circumstances. January 6 th , 2014 Dear Editor(s), Please find attached the revised manuscript "Student Acceptance of Tablet Devices in Secondary Education: A three-wave longitudinal cross-lagged case study”. This study draws upon the Theory of Planned Behaviour to design a cross-lagged longitudinal study that addresses the over-time interplay between pre- and post-adoption key uptake factors, i.e. attitude, subjective norm, and perceived behavioural control. We have taken into account and have responded to all of the reviewer comments, as requested in the decision e-mail. We elaborate on these issues in the rebuttal, which is part of the current manuscript. We declare that this manuscript has not been published previously and that it is not under consideration for publication elsewhere. In addition, we assure that all authors have fully participated in the research and the article preparation and both of us have approved the submission. Looking forward to the results and feedback of the review procedure. Yours sincerely, Cédric Courtois, PhD Covering Letter radical innovation should be deemed substantial, as these devices are used in every class for a broad diversity of tasks throughout the entire day.” Comment: “Page 21, line #41-42: "conceptual?" remove question mark” Response: This is corrected in the current version. Comment: “Page 22, the list of references is displayed two times (starts on page 22, starts again on page 27)” Response: This is corrected in the current version. Reviewer 2 Comment: “The study adopted a Theory of Planned Behavior approach to model students' acceptance of the personal tablet device as a learning tool. The author(s) reviewed related literature well to justify their research questions, and the methodology of the study was based on survey research methods in a reasonable way. Furthermore, the author(s) tried to discuss the research results comprehensively. Overall, the methodology of the study is solid and the quality of the paper is well written. However, concerning the innovative feature of the studies "Computers in Human Behavior" usually selects, this study may be plain and bring little new information to readers in the field. The study used TPB as the framework to answer their research questions. For readers in the field, the study may lack interesting points to learn from.” Response: We understand the reservation. However, as tablets in education are considered an important game-changing innovation; a prelude to how our youth might get socialized in technology appropriation, we believe as such that it is worthwhile studying. Moreover, this study, on contrast to others in the field, open the possibility for future meta-analysis, offering an overview of what technologies are well-accepted for what reasons, as opposed to possibly different outcomes. Moreover – and this is of the utmost importance, we would like to emphasize that the revised manuscript as soon as in the abstract now – and this has changed substantially in both introduction and discussion – explicitly stresses the necessity and theoretical/conceptual merits of a longitudinal approach in technology adoption research, as opposed to commonplace cross-sectional applications of the TPB. As such, it is not yet another TPB study as there are perhaps too many with limited conceptual/methodological appeal. More specifically, beyond the subject matter of tablets in education, we raise the following points of attention we believe are of interest for a general readership in the domain of the human-computer nexus: - Technology adoption is a dynamic, not a static process. It requires strategic research planning, carefully selection moments of data collection: in our specific case, we argue for and demonstrate pre-adoption, and short-term and long term post-adoption as valuable sense-making sample moments. - Failing to adopt a longitudinal approach renders research susceptible for considerable bias: our results show an evolutionary pattern, tied to each of the pre/post-adoption phases. In most studies, there is no clear rationale for selecting one or the other, especially in post-adoption. As we demonstrate, after six months, the TPB building blocks hardly explain any variance beyond prior use, indicating a habituation. A one-shot study at this point would likely lead to invalid interpretations (i.e. effects of the TPB measures, although these would be cancelled out by previous experience). This is also conceptually very important: post-adoption research should include proper measures of habit, as the habit-goal interface becomes much more relevant that the conscious factors that are included in the standard TPB framework. For that reason, we explicitly refer to psychological work on the habit-goal interface. - We also raise explicit awareness for the underestimated issue of attrition, which as we argue cannot be ignored in longitudinal technology adoption research. We demonstrate that those with a prior negative point of view are more likely to abandon the study, which has implications for the overall interpretation of results. This is the case in our study, but most likely also holds up for future studies. In light of these revisions, we sincerely hope they address your previous concerns, rendering the current manuscript suitable for publication in Computers in Human Behavior. Yours sincerely, The author(s) Highlights:  Uses Theory of Planned behavior to model pre/post-adoption of tablets in education  Demonstrates value of longitudinal cross-lagged analysis in technology acceptance  Design appears methodologically appropriate frame, sensing key uptake factors  Shows over-time interplay of attitude, subjective norm and perceived behavioral control *Highlights (for review) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 appropriation of such technologies in classrooms. Still, the fruitful implementation of digital learning tools, overcoming this chasm, remains a difficult issue. Recently, debate has sparked on the potential of tablet devices as educational means (Peluso, 2012). While in public discussions proponents praise the supposed motivating character of tablet technologies, fuelled by the many easily accessible affordances they potentially offer (Alvarez, Brown, & Nussbaum, 2011; F. Ferrer, Belvis, & Pamies, 2011; Henderson & Yeow, 2012), critics however frame it as a too expensive and inefficient manifestation of technological determinism, inspired by the alleged hype-factor that dominates the discourse on the issue. Such concerns not only surface in mainstream opinion and press coverage, but also in academic literature (e.g. Ifenthaler & Schweinbenz, 2013). In this article, we aim to subscribe and contribute to this debate by focusing on the important issue of user acceptance, not in the least by the most important stakeholders, i.e. the students themselves. After all, before the crucial assessment of potential learning effects, it is imperative to verify whether there is a bottom-up support for a continued implementation of such devices in secondary schools. Hence, the present study involves a longitudinal analysis of the acceptance process – both pre and post-launch – of the tablet as a learning tool in a relatively large Belgian secondary school that decided for a full-scale personal implementation of the tablet for all of its students and teachers. In this article, we abandon one-shot applications of user acceptance models by embracing a longitudinal approach, as also considered problematic by Sivo, Pan, and Brophy (2004). Despite calls to make this a common practice, most research efforts focus on cross-sectional inquiries. Drawing upon this study, we argue and demonstrate that this can be misleading, and that there is an apparent need to adopt a longitudinal approach that combines both pre-adoption 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 uptake determinants, as well as both short term and long term follow-up measures. As such, we are able to model over-time changes, which allows getting hold of the possible stability or dynamic interplay of uptake factors and how they develop through time. Three research waves, i.e. distinct moments of data collection, are analyzed: the pre-adoption expectation in September 2012, followed by post-adoption experiences measured at November 2012 and March 2013. To do so, this study is based on the Theory of Planned Behavior (TPB) as a guiding framework (Ajzen, 1991; Montaño & Kasprzyk, 2008; Taylor & Todd, 1995). TPB, incorporating the elements of attitude, subjective norm, and perceived behavioral control to explain use (intention), is especially relevant to model technology acceptance in constrained environments. This was the case at the studied school, as students had no choice whether to adopt, which was in itself a ground for debate. Theory of Planned Behavior: origins and form The Theory of Planned Behavior (TPB) is a seminal theory connecting belief systems with actual behavior, aiming at its explanation and even prediction (Ajzen, 1991). It has a rich history, having its origins in other prior theories. Most important though, is its roots in the Theory of Reasoned Action (Montaño & Kasprzyk, 2008; TRA). This theory aims to explain voluntary behavior, based on a conscious decision of the actor. Basically, TRA envisions behavior as a function of behavioral intention, which is in turn based upon positive relations with the interface of attitude and subjective norm. The former element refers to affective responses, i.e. a positive or negative stance, towards performing certain behavior. The substrate of an attitude is the beliefs held towards the behavioral outcomes and the extent to which these are valued. Subjective norm comprises how significant others feel about the actor’s 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 behavior, as perceived by that actor. Of course, the motivation to comply with these persons is of equal importance. In information systems research, TRA has been adapted to what is called the Technology Acceptance Model (TAM), which was initially composed by explanatory elements such as perceived ease of use and perceived usefulness (Davis, 1989; Legris, Ingham, & Collerette, 2003), and later on supplemented – among other constructs – with subjective norm (Schepers & Wetzels, 2007). TPB adds to TRA by incorporating perceived behavioral control, which is relevant in situations in which the actor might not have complete volitional control over the situation at hand. It involves possible facilitators or barriers that might aid or endanger posing the behavior. Again, this is a function of perception of control attributes and the importance of possessing these attributes. The literature counts various applications of TRA, TAM, and TPB in educational research on technology acceptance. Still, most of these studies focus on teachers, rather than students (cf. infra). It could be considered somewhat odd to leave these primary stakeholders out of the equation. Hence in this study, we explicitly focus on student acceptance of using tablet devices for school on a day-to-day basis. In most studies, attitudinal factors have shown relatively consistent in explaining either intention or actual use. For example, direct effects of teachers’ attitude were found on the usage (intention) of technology-supported teaching (Hu, Clark, & Ma, 2003), learning management systems (De Smet, Bourgonjon, De Wever, Schellens, & Valcke, 2012), web-based learning (Gong, Xu, & Yu, 2004), and digital games in the classroom (De Grove, Bourgonjon, & Van Looy, 2012). Research on learners of variable ages revealed a similar pattern, for instance on the topic of taking e-learning courses (Liu, Chen, Sun, Wible, & Kuo, 2010; Park, 2009), the 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 epitome of a practical and easily accessible technology (Ant Ozok, Benson, Chakraborty, & Norcio, 2008). H3: Perceived behavioral control positively explains use intention and actual use at each wave. The longitudinal component: pre- versus post-adoption As amply demonstrated in empirical applications of acceptance models, attitude, subjective norm, and perceived behavioral control play a central part in explaining behavioral intention and actual use. However, an obvious limitation of purely cross-sectional designs is the neglect of the potential over-time changes of these factors’ roles. This inevitably obfuscates researchers’ understanding of technology adoption, as it is a dynamic process. For example, prior research on information systems’ continuance has indicated that attitude, a core component (Yang & Yoo, 2004), is likely to differ over time (Bhattacherjee, 2001). A plausible explanation, partially supported by empirical research (Karakhanna, Straub, & Chervany, 1999), is that pre-adoption attitudes are usually based on second-hand information and perhaps give rise to inaccurate or unrealistic beliefs, whereas post- adoption attitudes are rooted in actual first-hand, repeated experience. Continued follow-up on the development of such a key variable is of the utmost importance to assure adoption continuance. Still, as Venkatesh and Morris (2000) argue, supported by evidence from both technology acceptance studies and related psychological research, attitudinal components appear significant determinants of intention, even after weeks of experience. Furthermore, despite reaffirming the important status of subjective norm, they presume that its influence drops as other people’s norms are internalized, especially when they are consistent with the own experiences. Equally important though, and unfortunately under-researched, is the over-time interplay 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 between technology acceptance factors. For that reason, following-up on the cross- sectional hypotheses, our study adds four more longitudinal hypotheses, inquiring possibly reciprocal causal relations between attitude and perceived behavioral control on the one hand, and attitude and subjective norm on the other. A first issue is the role of attitude, in conjunction with perceived behavioral control. As mentioned, we especially focus on self-efficacy, i.e. the perceived competence to handle a tablet as a learning tool. Various studies have indicated positive correlations between learners’ attitude or expected outcomes towards learning tools and the perceived mastery of them (Bates & Khasawneh, 2007; Moos & Azevedo, 2009; Ong & Lai, 2006). Although these studies assume causality by self-efficacy, giving rise to a positive attitude, this has not been unequivocally verified. Of course, as an assumption, this makes sense, albeit that a reversed trace of causality might be equally plausible. A learner could identify the merits of a tool, and foster a positive attitude towards it while not being able to operate it yet. Likewise, a sense of mastery, but a skeptical stance could over time turn into a positive attitude, as the perceived skills to handle the tool are at hand and the barriers to use it are low. As such, we propose the following two hypotheses: H4a: An earlier positive attitude serves as a substrate to develop a stronger perceived behavioral control. H4b: A stronger perceived behavioral control supports the later development of a positive attitude. Considerable research on technology acceptance has indicated social pressure, both by peers and hierarchical superiors as a strong explanatory factor in explaining the intention and actual uptake of innovations (Venkatesh & Morris, 2000). As yet demonstrated, developments in educational technology form no exception. Despite 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 scarcity in research on the over-time interplay between subjective norm and attitude on technology in education, there are nonetheless precedents. For instance, a study employing the TAM model found that earlier measures of subjective norm positively explain future attitudes, and vice versa, after controlling for measurement stability (Sivo et al., 2004). It appears that a positive attitude at one point renders students more susceptible to social influence, while a feeling of social influence also predicts an onwards-positive attitude. The former is likely due to a confirmation bias, i.e. the tendency to mainly select and be attentive of belief-confirming information (Nickerson, 1998). When a positive attitude is maintained, confirming information is more positively appraised. On the other hand, the reverse hypothesis hints that positively oriented teachers, together with the school board, would pass this belief on to the students. Although it is most unlikely that all teachers welcome the tablet device to an equal extent, it must be noted that before taking the decision to take the leap into implementation, wide support at teacher level was a prerequisite. Hence, we presume that the general teacher position to and communication concerning the school-wide adoption was generally neutral to favorable. Still, in literature, it is assumed that in the first stages of (forced) adoption, subjective norm has a much stronger influence on the uptake than later on (Venkatesh & Morris, 2000). If the technology performs well, like peers and superiors advocated, it is much more likely that these experience-matching beliefs are internalized. In this study, we already assumed that due to the forced nature of the implementation, subjective norm would play a fundamental role. Hence, we expect to encounter similar results as Sivo and colleagues (2004): H5a: An earlier positive attitude renders students susceptible for subjective norm at a later time. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 enumerations of items, adapted from Taylor and Todd (1995), are found in the article’s Appendix section. Attitude (A) (towards the iPad as a learning tool) was measured by a double four-item instrument, rated on a five-point Likert scale ranging from ‘completely disagree’ to ‘completely agree’. All four items were weigthed by the extent to which the attitudinal beliefs are considered important (i.e. five-point Likert scale ranging from ‘not at all important’ to ‘very important’). This measure demonstrates good internal consistency across all three waves. Subjective Norm (SN) comprised a double measure, in total comprising four items. The first two items inquire the extent to which (a) teachers and (b) the school board considers the iPad as a useful learning tool, rated on a five-point Likert scale, ranged ‘completely disagree’ to ‘completely agree’. Both items were weighted by the extent to which these two sources of subjective norm are considered important (i.e. five-point Likert scale ranging from ‘not at all important’ to ‘very important’). Both measures correlate substantially across waves, and are hence averaged. Perceived Behavioral Control (PBC) in essence comprises a measure of self- efficacy to use the iPad for school. As with Attitude and Subjective Norm, both beliefs and evaluations were measured, using the latter to weigh the former. That is, first, four items probed into efficacy beliefs, while second, these beliefs were evaluated in terms of importance. The four weighted efficacy items demonstrate a satisfactory internal consistency at all three waves Intention to use (I) was measured at the first wave in September 2012. A six- item measure inquired how often students estimated the prospective use of their iPads for school purposes, both at school and at home (cf. Appendix). The items were rated 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 on a five-point scale, ranging from ‘never’ to ‘very often’. The instrument shows good internal consistency. Actual use (U) was measured at the second and third wave in November 2012 and March 2013 respectively. It inquired retrospective estimations of actual use. At the second wave, the frame of reference was the period between the start of the school year and questionnaire administration. At the third wave, this frame ranged from the period after the midterm exams until questionnaire administration. The instrument draws upon the very same items as intention to use, employing the exact same rating scale. Both at the second and third wave, the measures demonstrate a satisfactory internal consistency. < Insert Table 1 > Results Dropout analysis As mentioned in the procedure section, this study suffered from substantial dropout rates, in the end retaining 52% of the initial respondent pool. As such, there is an apparent need to verify whether this attrition is contingent with a priori expectations towards the subject matter (Goodman & Blum, 1996). In order to shed light on this matter, a multinomial regression model is computed, employing attitude, subjective norm, perceived behavioral control, and use intention measured at the first wave as independent variables, and dropout at the second or third wave as a nominal dependent variable. This results in a well-fitting model ( 2 (8) = 24.71, p < .005), however explaining five per cent Nagelkerke pseudo-R 2 . The results, summarized in Table 2, indicate that subjective norm accounts for dropout in the both the second and third wave, whereas attitude only does in the third wave. More specifically, a sense of obligation by teachers and directors increased the odds to keep on participating in the 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 study, while a priori negative attitude explains attrition between the second and third wave. < Insert Table 2 > A subsequent analysis binary regresses the dropout between wave two and three, explaining 4 per cent of the Nagelkerke pseudo-variance ( 2 (4) = 10.21, p < .05; Table 3). This shows again that a subjective norm to use the tablet is paired with an on-going participation in the study. < Insert Table 3 > These small, yet non-surprising effects do not seem to endanger the validity of claims derived from the final sample, provided that a minimal restriction of range by dissatisfied students is probable. Moreover, reprising Table 1, we point to the very similar dispersion within measures at all three waves. Cross-lagged longitudinal path analysis To test the study’s proposed hypotheses, a path model was computed, employing all three waves’ measurements. More specifically, per wave, paths from attitude, subjective norm and perceived behavioral control toward either use intention of actual use were modeled. Next, to test the longitudinal hypotheses, the necessary paths for a cross- lagged analysis were included (Burkholder & Harlow, 2003; E. Ferrer & McAdrle, 2003). First, these comprise auto-regression stability paths, regressing a next wave’s measure onto its previous measurement. Second, a cross-lagged regression is added, which is in generic terms the effect of a variable XT1 at first time on a variable YT2 at a second later time, whereas the same logic applies for YT1 and XT2. If such paths appear significant, it represents a trace of causality. Furthermore, two types of covariance were additionally modeled. First, we modeled (residuals’) covariances 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 school and at home. As this research topic is emergent, such research is to our knowledge unprecedented. As argued, we focus on the longitudinal character of data collection, as commonplace cross-sectional efforts are informative, although they might be equally misleading at the same time. As adoption is a dynamic process, evolving through phases of pre- and post-adoption, it is not always clear where to position a cross-sectional effort. In contrast, a longitudinal effort like this one sheds light on this evolution, emphasizing its relevance. Our appropriation of the Theory of Planned Behavior has proven sufficiently sensitive to grasp the evolving sentiments at hand. As such, the bigger picture is aptly drawn, which in turn incites goal-directed and properly informed follow-up research. In the following paragraphs, this evolution is further discussed. However, before discussing the results, we need to warn for a, albeit minor, restriction of range, caused by the attrition throughout the different waves. Students that displayed a more negative stance from day one were less likely to maintain participation whilst those who felt a stronger subjective norm, i.e. by teachers and the school board, are relatively overrepresented. Still, as argued in the dropout analysis section, the effects of attrition are rather small, so we were nevertheless able to proceed with a meaningful interpretation of the findings. As such, meticulous analyses of attrition patterns are an indispensible in longitudinal designs, and especially in interpreting their results. In general, our findings partially mirror the proposed hypotheses. At the beginning of the school year, in September, it was clear that students had fairly positive attitudes, which was the strongest explanatory factor of using the tablet for school, throughout the year. At that point, there was no significant effect of perceived behavioral control, hinting to the perception that there were no substantial 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 obstructions in handling the device for learning practices. As expected, a minor sense of obligation was perceived, as subjective norm – reflecting the urge by teachers and the school board – rendered a minimal, yet significant effect. Three months later, in November, the picture had slightly changed. The attitude measure at that point did not explain any unique variance in usage, despite a significant zero-order correlation. Nevertheless, we observed a significant indirect effect through the autoregressive stability path between intention and use by the attitude measured at the first wave. This suggests that the attitude prior to adoption proved accurate to some extent. Interesting though, is the direct effect of perceived behavioral control. Three months post-adoption, we learned from teachers and students that the implementation yielded some problems of variable nature. First, there were technical issues with the application used on the tablets (e.g. crashes, down-time, usability issues). Second, it proved more difficult than expected to use the device in a school context (e.g. incorporation in class, cope with distractions such as social media and games). Considering previous literature, these issues fit the evident struggles of implementing the tablet as a new technology (Henderson & Yeow, 2012; Ifenthaler & Schweinbenz, 2013). Next to considerable efforts to solve technical issues, the teaching staff took on a more restrictive stance towards these issues, while continuously motivating students to persist. This enables us to understand the more strongly felt subjective norm at that point. This is consistent with information systems literature that argues that in theory the influence of subjective norm drops and gets internalized (Venkatesh & Morris, 2000), on the condition that everything works properly, as advocated. In this particular instance, this was absolutely not the case. Finally, in March, the picture changed again. Both effects of attitude and perceived behavioral control disappeared, while the direct effect of subjective norm 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 toned down. At this point, we see that the TPB measures, despite continuously significant correlations, no longer explain use. Their effects are cancelled out by use at the previous wave, as the autoregressive stability path proved more substantial. This is plausibly explained through literature on habit formation, stating that repeated satisfactory behavior under stable circumstances eventually leads to habit build-up, toning down the effects of attitudes and subjective norm (Ajzen, 2002; Ouelette & Wood, 1998). Although previous behavior is not an undisputed index of habit (i.e. it is too restrictive, not fully representing it as the mental construct it is) (Verplanken, 2006), it does offer an indication that the use of the tablet at that point got internalized as a routine practice at school. At that point, conscious deliberations of attitude and perceived behavioral control were not that important anymore. This finding emphasizes the problematic nature of cross-sectional designs, especially when it comes to the timing of data collection. Our results clearly show an evolution towards habituation, which renders the TPB building blocks that dominantly draw upon salient cognition fairly obsolete and even deceptive. Both in theoretical and methodological terms, post-adoption research efforts should therefore focus on the habit-goal interface (i.e. issues of automaticity; Wood & Neal, 2007). The present study however also takes into account the interplay between TPB measures over time. In that respect, there is a strong support for the assumption that a positive attitude at a prior instance gives rise to the development of a stronger perception of behavioral control. Students who have a favorable position towards the tablet as a learning tool are more prone to develop their sense of skill. Our results strongly indicate the prominence of perceived behavioral control. Although it is commonly assumed that tablet devices are easy to use, this might be so for entertainment purposes, but not necessarily for educational ends – next to the issue of 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 References Ajzen, I. (1991). The Theory of Planned Behavior. Organizational Behavior and Human Decision Processes, 50, 179-211. Ajzen, I. (2002). Residual Effects of Past on Later Behavior: Habituation and Reasoned Action Perspectives. Personality and Social Psychology Review, 6(2), 107-122. Alvarez, C., Brown, C., & Nussbaum, M. (2011). Comparative study of netbooks and tablet PCs for fostering face-to-face collaborative learning. Computers in Human Behavior, 27(2), 834-844. Ant Ozok, A. A., Benson, D., Chakraborty, J., & Norcio, A. F. (2008). A comprative study between tablet and laptop PCs: User satisfaction and preferences. International Journal of Human-Computer Interaction, 24(3), 329-352. Bandura, A. (1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28(2), 117-148. Bandura, A. (1994). Self-Efficacy: The Exercise of Control. New York: Freeman. Bates, R., & Khasawneh, S. (2007). Self-efficacy and College Students' perceptions and Use of Online Learning Systems. Computers in Human Behavior, 23(1), 175-191. Bhattacherjee, A. (2001). Understanding Informaiton Systems Continuance: An Expectation-Confirmation Model. MIS Quarterly, 25(3), 351-370. Bourgonjon, J., Valcke, M., Soetaert, R., & Schellens, T. (2010). Students’ perceptions about the use of video games in the classroom. Computers & Education, 54(4), 1145-1156. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Burkholder, G. J., & Harlow, L. L. (2003). An illustration of a longitudinal cross- lagged design for larger structural equation models. Structural Equation Modeling, 10(3), 465-486. Davis, F. D. (1989). Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information Technology. MIS Quarterly, 13(3), 319-340. De Grove, F., Bourgonjon, J., & Van Looy, J. (2012). Digital games in the classroom? A contextual approach to teachers' adoption intention of digital games in formal education. Computers in Human Behavior, 28(6), 2023-2033. De Smet, C., Bourgonjon, J., De Wever, B., Schellens, T., & Valcke, M. (2012). Researching instructional use and the technology acceptation of learning management systems by secondary school teachers. Computers & Education, 58(2), 688-696. Durndell, A., & Haag, Z. (2002). Computer self efficacy, computer anxiety, attitudes towards the Internet and reported experience with the Internet, by gender, in an East European sample. Computers in human behavior, 18(5), 521-535. El-Gayar, O., & Moran, M. (2007). Examining students' acceptance of tablet pc using TAM. issues in Information Systems, 8(1), 167-172. Fagan, M. H., Stern, N., & Wooldridge, B. R. (2004). An empirical investigation into the relationship between computer self-efficacy, anxiety, experience, support and usage. The Journal of computer information systems, 44(2), 95-104. Ferrer, E., & McAdrle, J. J. (2003). Alternative Structural Models for Multivariate Longitudinal Analysis. Structural Equation Modeling, 10(4), 493-524. Ferrer, F., Belvis, E., & Pamies, J. (2011). Tablet PCs, academic results and educational inequalities. Computers & Education, 56(1), 280-288. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Gong, M., Xu, Y., & Yu, Y. (2004). An enhanced technology acceptance model for web-based learning. Journal of Information Systems Education, 15(4), 365- 374. Goodman, J. S., & Blum, T. C. (1996). Assessing the non-random sampling effects of subject attrituion in longitudinal research. Journal of Management, 22(4), 627- 652. Henderson, S., & Yeow, J. (2012, January, 4-7th, 2012). iPad in Education: A Case Study of iPad Adoption and Use in a Primary School. Paper presented at the 45th Hawaii International Conference on System Sciences, Maui, Hawaii. Hsu, M. S., & Chio, C. M. (2004). Internet self-efficacy and electronic service acceptance. Decision Support Systems, 38(3), 369-381. Hu, P. J. H., Clark, T. H. K., & Ma, W. W. (2003). Examining technology acceptance by school teachers: a longitudinal study. Information & Management, 41(2), 227-241. Ifenthaler, D., & Schweinbenz, V. (2013). The acceptance of Tablet-PCs in classroom instruction: The teachers' perspectives. Computers in Human Behavior, 29(3), 525-534. Karakhanna, E., Straub, D. W., & Chervany, N. L. (1999). Information technology adoption across time: A cross-sectional comparison of pre-adoption and post- adoption beliefs. MIS Quarterly, 23(2), 183-213. Kenny, D. A. (2012, July 5th, 2012). Measuring Model Fit. Retrieved July 26th, 2013, from http://davidakenny.net/cm/fit.htm Legris, P., Ingham, J., & Collerette, P. (2003). Why do people use information technology? A critical review of the technology acceptance model. Information and Management, 40(3), 191-204. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Yang, H. D., & Yoo, Y. (2004). It's all about attitude: revisiting the technology acceptance model. Decision Support Systems, 38(1), 19-31. Table 1. The study’s measures’ means, standard deviations (min = 1, max = 5) and Cronbach’s alpha internal consistency measures. The measures were calculated each wave separately, not taking into account the attrition. + As these were two-item measures, Pearson zero-order correlations are reported instead. Wave 1 Wave 2 Wave 3 M SD  M SD  M SD  A 3.46 1.02 .92 3.19 1.12 .93 3.18 1.10 .93 PBC 3.73 0.84 .88 3.76 0.90 .90 3.79 0.87 .90 SN 2.85 0.92 .71 + 2.63 0.92 .67 + 2.58 0.93 .70 + I 3.87 0.66 .83 - - - - - - U - - - 3.94 0.71 .84 3.74 0.77 .87 Table Table 2: Multinomial dropout regression analysis explaining dropout at the second and third wave by first wave measures. Initial participation is employed as reference category. * p < .05, ** p < .005 Dropout after Wave 1 Dropout after Wave 2 B SE Wald Exp(B) B SE Wald Exp(B) Intercept .03 .69 .00 - -.02 .60 .00 - AW1 -.13 .13 .96 .88 -.22 .11 4.01* .80 PBCW1 .16 .14 1.24 1.17 .06 .12 .25 1.06 SNW1 -.08 .03 1.58** .92 -.06 .02 6.51 .95 IW1 -.01 .12 .00 .99 .14 .10 1.75 1.15 Table 5: Summary of the study’s hypotheses and their supporting evidence. Hypothesis Evidence W1 W2 W3 H1: Attitude positively explains use intention and actual use at each wave.    H2: Subjective norm positively explains use intention and actual use at each wave.    H3: Perceived behavioral control positively explains use intention and actual use at each wave.    W1-2 W2-3 H4a: An earlier positive attitude serves as a substrate to develop a stronger perceived behavioral control.   H4b: A stronger perceived behavioral control supports the later development of a positive attitude.   H5a: An earlier positive attitude renders students susceptible for subjective norm at a later time.   H5b: A stronger sense of subjective norm at an earlier time supports the development of a positive attitude later on.   Attitude (belief; ‘completely disagree’ to ‘completely agree’) To what extent do you agree with the following statements?  Using the iPad for school is fun  Using the iPad for school is enjoyable  It feels good to use the iPad for school  It is interesting to use the iPad for school (evaluation of desirability; ‘not at all important’ to ‘very important’) How important is it that…  Using the iPad for school is fun  Using the iPad for school is enjoyable  It feels good to use the iPad for school  It is interesting to use the iPad for school Subjective norm (normative belief; ‘completely disagree’ to ‘completely agree’) To what extent do you agree with the following statements?  My teachers think the iPad is useful for school work  My school’s board of directors think the iPad is useful for school work (motivation to comply; ‘not at all important’ to ‘very important’) How important is it…  To do what my teachers think I should do  To do what my school’s board of directors think I should do Perceived behavioural control (control belief; ‘completely disagree’ to ‘completely agree’) To what extent do you agree with the following statements?  It is easy to learn how the use the iPad for school  The directions to use my iPad for school are simple  It is easy for me to become an advanced iPad user  The iPad is straightforward to use for school (perceived facilitation; ‘not at all important’ to ‘very important’) Appendix How important is it that…  It is easy to learn how the use the iPad for school  The directions to use my iPad for school are simple  It is easy for me to become an advanced iPad user  The iPad is straightforward to use for school Intention How often do you think you will use the iPad (Never – Very often)  I will use the iPad during classes at school  I will use the iPad for assignments at school  I will use the iPad for homework  I will use the iPad to study  I will use the iPad to contact my classmates about school work  I will use the iPad to contact my teachers about school work Use How often have you use the iPad since (a) the beginning of the school year, (b) this semester (Never – Very often)  I have used the iPad during classes at school  I have used the iPad for assignments at school  I have used the iPad for homework  I have used the iPad to study  I have used the iPad to contact my classmates about school work  I have used the iPad to contact my teachers about school work