‘Rooting Out’ the Issues for Biomedical Knowledge Flow: A Formal Theory for Mobile Learning Pedagogy

.


Introduction
This scientific research paper stresses the importance of getting the technology right and knowing how to use it to its best advantage. In the field of biomedical sciences this ultimately results in better healthcare. A recent review discovered a trend in the most highly cited papers i.e., 'comparing different mobile learning modes and finding more effective mobile learning approaches Lai, et al. [1]. However, some take a different approach: they compare the impact of new mobile learning approaches with traditional instruction.
In many senses this paper combines both. It compares learning modes in pursuit of effective approaches and assesses the impact of learning within these. It does this by using evidence-based research as a basis, and grounded theory to explain why the most effective approaches should be based on real life impact (regardless of whether it is based on traditional or non-traditional methods). This paper is centred around ever-progressing 'Mobile' learning which unsurprisingly has a large range of definitions. Because devices continually evolve, definitions evolve too. A deliberately many conceptual issues prevent effective planning and delivery of the learning. For example: many educators don't understand or know how to use mobile learning models and this is especially true if that learning is on the move i.e. 'mobile learning' Ginkas, et al. [3]. These conceptual issues therefore need to be identified, 'rooted out', and explained if we are to progress beyond these issues, and the limitations they impose on our understanding. This way we can understand in more depth why certain barriers to applying good quality research to mobile learning exists. However, this is not always an easy task. Historically, there was a flurry of pedagogical activity between 2004 to 2011 with many pedagogical approaches being discussed for medical mobile learning. Calbraith, [4] went some way to defining important points to consider (i.e., original generic principles and the part they could play in mobile learning formats, the importance of learner input and perceptions about their learning, and how branch and loop learning systems shape learner customisation). The participants were medics, nurses, Biomedical scientists, and Science Teachers based on two sites (Midlands and Eastern UK) with full ethical approval. Here, members of the general public (i.e. 'laymen') were effectively used as a control group. Mixed methods methodology was used to collect data on top-performing pedagogical approaches based on real-life impact (drawn from a systematic review which was then assessed via a modified version of 'Kirkpatrick's model ' Kirkpatrick,et al. [5,6,7]. Please see (Table   1). This construct was used as this allowed unhindered observation and examination of the underlying pedagogical aspects of mobile learning. It allowed promotion of cognitive activity and provided the freedom to 'test and engage' through testing assumptions (making hypotheses), adjusting variables (experimenting) and introducing content (modifying). Its basic premise allowed pedagogy to be seen and what the successful elements are without other aspects getting in the way. Data was collected on how learners used different online learning approaches during observed usability studies.  The data was verified via semi-structured interviews. This work highlighted 'learning routes' or 'pathways' that learners' chose to take through their learning packages, and importantly allowed the plotting of learners' 'knowledge flow'. This resulted in faster learning and development of clinical reasoning. Implicit meanings that learners attached to knowledge were seen to become explicit, and therefore enabled learning pathways to be plotted.
This process was essentially the learners' online 'breadcrumb trail'. When individual routes from each learner were compared with many other learner routes, they were seen to become part of a cohesive whole i.e. they were a well-trodden 'learning pathway' that all learners took. This was an original finding. Calbraith and Dennick [8] developed these concepts into a mobile learning model. This emphasised the importance of models being flexible enough to allow relevant and immediately useful information to be delivered to the learner through guided and informed reasoning.
The combination of methods was successful in uncovering some previously undiscovered underlying pedagogies for mobile learning in medical sciences. This was because a) The learners were an integral part, b) The impact was measurable, and  [11] believes pedagogical underpinnings for mobile learning are "much needed and long over-due" due to how little is known about 'technology use and learning interactions'.
Aagaard is not alone. Many authors feel that learning more about learners' technology use is a key issue but cannot fully explain why Zhang, et al. [12,13]. Removing these barriers would mean more effective learning would be possible. Consequently, the next step is to examine Calbraith's hypotheses and substantive theory then compare this to current approaches to see whether the field has developed enough for the theory to achieve formal theory status.
This would be a significant step, as it would also remove many barriers to development. These hypotheses will therefore form the basis for the novel thinking in this paper to develop conceptual understanding (and therefore the shape of online learning) in this field. This paper goes further than others because it explains why some current pedagogical approaches 'work' and some do not; it provides the crucial factors involved, identifies what elements are involved, and why barriers exist in the first place. These aspects are really important as they further our understanding as to why some approaches to mobile learning are more effective than others. If we look at the background of pedagogical and knowledge flow issues for mobile learning, two issues are immediately obvious: a) There is a lack of applied research, and b) When researchers try to apply research, unexpected barriers arise. This paper will therefore look at these issues before looking at how well the theory explains the successes and shortfalls of popular, current approaches (which will be discussed in more detail in the findings section).

Materials and Methods
Glaser, et al. [9] constant comparative grounded theory method will be used to a) Examine any conceptual issues for learning, pedagogical or knowledge flow, and b) Develop the substantive theory.
Glaser and Strauss recommend three ways to do this in their original instructions. As Calbraith, [4] had already performed the first stage regarding theory development (i.e., applying the formed hypotheses to diverse groups and situations), this paper starts from the second stage. The aim is to push variables to their limit: 'to undertake direct data comparisons from other substantive areas in the researchers experience, or in the literature'. This was to find the extent to which the resulting theory can explain, prove, or disprove instances in current popular approaches at formal theory level, to test emergent theories, and generate theory. As the grounded theory process for the original for the substantive theory is already published and is not the focus of this paper, it will not be repeated here in depth. Each unique observation and verbatim comment from the original usability studies were coded incidentby-incident and grouped according to subject and whether they were positive, negative, or neutral statements. They were tested until saturation point, which was 6 regardless of the person using it and the pedagogy being tested. According to Kuzel [12] a saturation point of 6 indicates that the 420 participants were homogenous.
The demographics, however, indicate otherwise. Instead, it was the freely formed comments that were homogenous making 20 units unnecessary.

Results and Discussion
In the research study that forms the basis of this paper Calbraith, [4], the theoretical saturation point was achieved after 6 units, with 12 core categories and 11 pedagogical codes found. Meanings and associations were seen because. Hypotheses 1. ' A good mix of elements result in increased knowledge, which results in an ability to judge progression -therefore learning errors are identified and increased awareness of learning direction and needs result' 2. ' A good mix of elements result in increased interaction leading to increased active learning, knowledge and interest which achieves a good element-interaction balance by allowing a) a greater application of learning, b) greater linkage of knowledge which develops reasoning, and c) decreased feelings of information overload' 3. ' A good mix of elements allows comparison which increases active learning, which focuses the mind, which provokes increased reflection' 4. 'Labels, the right structure, and level of/timing of elements results in clarity and simplicity, which shows learners what is going on. This increases active learning by allowing learners to digest learning without distraction or 'overload' feelings. This results in increased reading, understanding, and reasoning speed because the learner feels in control of the learning, resulting in good navigation' i) the construct and learning packages did not impose extraneous variables which allowed the whole process to be mapped, and ii) attributed meanings were first-hand from learners. This is important. There is a tendency to focus on pedagogical 'parts' or 'stages' in the literature which effectively renders some interactions 'invisible'. This study showed that more associations can be seen when looking at the whole process simultaneously, which may explain why some authors and studies have found it difficult to explain why 'pedagogical deconstruction' techniques often fail. This paper is based on the 4 Hypotheses found (see Table 2) but before analysis of the wider mobile context can be made, there is a need to look at the issues for biomedical mobile pedagogy in more depth to see if the theory can explain them. The first major pedagogical issue is the sheer lack of applied research

Am J Biomed Sci & Res Copy@ Davina Calbraith
for mobile learning and tool use Gikas et al. [3]. A typical example is Baez et al. [13] pedagogical model which they themselves describe as 'too utopian' and "unsatisfactory for guiding lecturers in their pedagogical tasks'. This is due to 'conceptual' shortcomings and "a lack of awareness of models and strategies, and tools to implement those models" (p436). In other words, the learners' knowledge flow or 'learning pathway' could not be plotted due to conceptual and lecturer-based 'stumbling blocks'. This 'bottom-up' recommendation may be a confirming instance of the hypotheses, as capturing learners' views was an intrinsic part of these.
Due to the concept/construct used learners were happy trying out ideas so knowledge flowed easily due to the freedom to guess answers. The hypotheses imply that when mixed elements are used this also results in increased knowledge, interest, motivation, and interaction'. Could 'mixed elements' be a key component of the theory, knowledge flow, or learning pathways? It is possible as risktaking with no consequence may explain why learners found the learning packages so enjoyable, why the learning motivated them to learn more, and why they perceived the learning as 'having a good content and interaction balance'. As participants were largely drawn from medical fields where real-life risk-taking is positively discouraged (as it may result in harm to a patient if the correct clinical decision is not taken) this may have added to the appeal.
Sung, et al. [15] called for "in-depth experimental research" to remove these aforementioned barriers and provide pedagogical solutions. Pedro, et al. [16] responded to this call by providing some best practice: 'Collaborative-driven versus data-driven practices'; 'Informing students about the learning process and 'how to be focused'; and training educators to implement these. A crucial point is whether the substantive theory and hypotheses can explain, prove, or disprove these best practices. (See Table 3). As the theory matches well to Pedro et al best practice, this could be seen as a 'proving instance' of the theory, which is promising. A second major issue for mobile learning pedagogy is the unexpected limitations that crop up when researching this area. One example is Jones, et al. [17], who found unexpected limitations when attempting to explain pedagogical issues when using a conceptual model for data analytics. As Jones' model is purely theoretical Jones is likely to be unaware of the full potential or real-life impact this has on learner response. Consequently, Jones called for future research to 'test mock interfaces that simulate (learner) information controls'.
This implies that a bottom-up, student-led focus is needed for biomedical mobile learning. In instances where successful research has been achieved, unexpected limitations have then hampered its full application. Table 3: Pedro et al. [16] best practice for mobile learning.

Pedro et al. [16] Best Practice Developing Theory Hypotheses How Well do They Match?
Collaborative-driven versus datadriven practices The developing theory fits well as methodological collaboration allows the learner to interact (ultimately resulting in good element-interaction balancesee second hypothesis)

Good match
Informing students about the learning process User-centredness was a central feature of the concept and hypotheses as findings came from the learners themselves, suggesting that learners should not just know about the learning process but be part of it.

Good match
Informing students about how to be focused The element mix in the top-performing pedagogies 'allow comparison, which focuses the mind' (see third hypothesis). Good match Therefore, it is possible that this second issue has perhaps caused the first. Some believe this lack of applied research originates from educators lacking guidance Baran, et al. [18,15] or failing to keep students interested Kuznekoff, et al. [19]. Others maintain it is 'the nature of the beast' i.e., pedagogy cannot keep pace with technology Kurzweil, et al. [20]. Clearly a combination of these issues are to blame. Addressing pedagogical problems and discovering key elements are therefore challenging. It is unsurprising that researching this area (and attempting to apply the research to practice) often results in some educators focusing in on the individual problems (e.g., lack of student engagement), and looking for approaches to solve those problems. As a result, many focus on learner motivation as their investigative starting point, and look at this through the lens of 'gaming' Rankin et al. [21]. Gaming, or

The Wider Mobile Context
Both successful and unsuccessful elements within current concepts and approaches were basically 'unpicked' and analysed to see i) the extent to which they can explain their own underlying pedagogy and knowledge flow, and ii) the extent to which any learning pathways can be identified. This is presented here as 'What works' and 'What doesn't work' for each context, followed   Table 2). This is therefore a further

Artificial Intelligence (AI) and Rhizomatic Learning
What works -The concepts of 'Learning Analytics' and ' AI Learning' are sometimes confused. Both are often 'theoretically explored' but seldom applied in a significant way Ferguson et al. [29][30][31][32][33]. The term 'Rhizomatic learning' has been coined from 'Rhizomes' which grow with root networks, rather than one established root Bogue, et al. [34]. These pedagogies are based on the work of Deleuz and Guattari Gilles, et al. [ derived from the learners themselves and is tried and tested to saturation point in the field before theory is even applied.
In summary, successful elements for AI and Rhizomatic learning are user input, timing, interaction, learning transformation, reallife scenarios, individualisation, knowledge transfer, and humandigital interaction. These accord with the hypotheses and theory. In Calbraith [4] research, students felt the environment was adapted to their needs, and the developing theory is able to explain why. This is important. The hypotheses showed that 'feelings of perpetual adaption to learner needs' feeds the 'perpetuated motivation to learn' that was seen. This crucially means that not all learning preferences have to be included in the mobile learning package, providing it engenders these feelings and knowledge flows easily.
Not all learning preferences were built into the learning packages, yet 100% of learners felt catered for. These feelings can also be explained by the causal relationships found in Calbraith's, [4] mapped learning pathways. This may also explain why personalised learning concepts are currently so popular as they engender the same feeling. Diaz, et al. [37] model is comprehensive from a technological perspective and shines an important spotlight on 'elements' and 'learning pathways'. It does not, however, specifically explain knowledge flow to the extent that the theory does.

Adaptive Learning Platforms (ALP) and Learning Personalisation (LP)
What works -ALPs are just what they sound like, learning platforms which allow adaptive learning. One purported advantage from blended ALP literature is that it allows different learner-centred choice. Al-Zahrani, et al. [45] found 'motivation to learn' was a global element. Zhang, et al. [46] do not identify successful elements but imply that 'timely feedback' is a core feature and therefore recommend examining 'learners' learning emotions' It has already been shown that there is a need to look at the whole 'learner use process' (and particularly how learners choose their route through online learning) to see all important parts of the learner pathway. It is therefore asserted that when specific elements are grouped together the total becomes more than the sum of its individual parts, i.e., the mixture of specific elements take on a life of its own to elevate learning to a whole new level due to how they make learners feel. Calbraith and Dennick [8] referred to this phenomenon as 'value-added' observations.
From their psychological approach, Bartelome, et al. [48] conclude that pathways learners take depend on the discipline.
However, this is not necessarily true. Invariably both disciplinedependent and generic principles can be derived from the pathways learners take, but the actual pathways themselves seen in Calbraith's, [4] research remained the same, irrespective of discipline. This is a disproving instance for the theory. There is a small but very important distinction here between 'the steps learners take' and 'the learning pathway'. Minor variations were seen between learners and disciplines on the learning pathway which were initially interpreted as different learning pathways.
However, as work progressed (and learning paths were overlaid) it was clear these were not different pathways or deviations but were actually parts of the same pathway with the same elements.
Not all learners needed to stop at all points on the learning pathway in the same way that not every learner needs to spend as much time as others on certain aspects. This does not amount to them taking a different learning pathway but is, instead, part of learning personalisation and self-personalisation. If we think of the learning pathway as physical stepping-stones, some learners may jump a stone but still carry on the same path/direction (this is particularly true of established learners). All complete the learning. Kukulska-Hulme [50] sums up this phenomenon beautifully: "Personalized learning takes account of learners' interests, preferences, prior knowledge, competencies, movements and behaviours, but this does not imply that all mobile application designs need to take all these aspects into consideration".
In summary, the theory and hypotheses presented here affords the learner the feeling of choice when put into practice. Bartolome et al provide both a proving and disproving instance of the theory.
Although successful elements are hard to find in current 'adaptive learning platforms' and 'personalised learning literature', both mention learner importance, their motivation, and timely feedback.
As the theory accords with these too, these are likely to be important In summary, the attraction for flipped classrooms is clear as it fulfils the need to 'learn at your own pace'. However, some effective flipped studies also contain' unformed aspects' which imply caution before accepting 'timing', and 'pacing' as key elements. However, as they feature highly in other contexts this appears to be another possible proving instance. The problem-solving, higher order thinking, active learning, informative feedback, and knowledge application to real-life problems accord with the theory's hypotheses. As this is so, and flipped learning tends to create unique learning that cannot be applied to other groups, it is suggested that it is the elements or mix of elements within flipped learning that causes it to be successful pedagogically, and not necessarily the flipped approach per se.

Implications
The grounded theory approach presented here offers an evidence-based theory to explain why a mixture of specific elements is required for effective mobile learning. This concept requires a paradigm shift when looking for pedagogical answers. Instead of taking the problem as a starting point it takes learner use of the learning to elucidate underlying mechanisms. It is true that some effort must go into successful learning, and the research method this formal theory is built on is no exception. It is 'front-heavy' but pay-off comes in the lack of maintenance needed once it is up and running. Using the mix of important elements outlined here when planning mobile learning provides learning that is 'effective from the outset'. The proving and disproving instances found and discussed have major implications regarding not just the theory's potential as a formal theory of mobile learning, but also to the range of learning situations, packages and delivery formats that this can be applied to. The application appears limitless.

Conclusion
This paper has shown that sometimes the contexts or underlying concepts that researchers choose to frame the pedagogical evaluation, knowledge flow or learning pathways can actually constrain observation of how learners use the learning.
This results in under-developed answers as to why the approaches either work or do not work pedagogically. It also results in failure to map the learning pathway, explain the existence of barriers, or explain why the knowledge flow is not apparent. By contrast, the developing theory put forward here was able to explain not just why certain elements were important, but also what role they had within the learning pathways and the impact on knowledge flow. The research this theory was built upon used Sims (2006) construct. This was therefore a good basis from which to evaluate pedagogical impact because it allowed learner engagement to increase due to the freedom to test their assumptions (and adjust/ introduce new content if necessary) but without fear of failure.
This process was shown to be important to learners. The inductive stance taken meant that issues surrounding pedagogical mobile learning practice were explored unhindered and unconstrained.
Crucial elements for effective mobile learning were discovered, and underlying pedagogical mechanisms explained. Finally, 'feedback', 'knowledge application to real-life problems', 'learning at your own pace' and 'high order thinking skills' were seen as vital elements in successful Flipped Classroom approaches.
In short, pedagogies used in this research improved learner confidence because the mixed elements impacted learner feelings positively and made them feel personally catered for. 'Learner use' of the intended learning is therefore crucial as it is the glue that holds mixed elements together to form successful learning pathways, and therefore knowledge flow. It should therefore be the starting point for any pedagogical investigation, otherwise important valueadded aspects that form the underlying pedagogical mechanisms will be missed. If these are unnoticed, they eventually form barriers. Conceptual and pedagogical issues that hinder knowledge flow were explained which added further weight to the theory's formal status. As the theory and hypotheses generated were able to explain successes/pitfalls in contemporary approaches, this potentially demonstrates that use of these mixed elements could have universal application to many topics, disciplines, and approaches to transform mobile learning in any context.