The problem statement of the study is exploratory and open-ended, and requires an exploratory design. The research method used to develop the E-learning Circle is Grounded Theory (Glaser & Strauss 1967, Strauss & Corbin 1998). The inductive, theory discovering approach of Grounded Theory, allowing the grounding of theory in empirical data, is appropriate for exploratory studies.

The selection of research sites followed the technique of theoretical sampling, described as “data gathering driven by concepts derived from the evolving theory and based on the concept of ‘making comparisons’, whose purpose is to go to places, people, or events that will maximize opportunities to discover variations among concepts and to densify categories in terms of their properties and dimensions” (Strauss & Corbin 1998). The empirical data, upon which this article is based, were collected through interviews, focus groups, and expert groups.

Grounded Theory provides the researcher with analytical tools when analyzing the data in the iterative research process. “As grounded theorists, we begin our analyses early to help us focus further data collection. In turn, we use these focused data to refine our emerging analyses” (Charmaz 2005). In the processes of open, axial and selective coding (Strauss & Corbin 1998) tools like questioning, constant comparison, diagrams, and memos were valuable.

The “learning objectives” sector illustrates the connection between learning objectives and assessment tools (Figure 2). The learning objectives are divided into four types; skills, knowledge, attitudes and meta-learning. The use of the terms skills, knowledge and attitudes are inspired by well-known taxonomies (Bloom 1956; Kratwohl 1964; Dave 1970). Meta-learning is the state of “being aware of and taking control of one’s own learning” (Biggs 1985) and is added as a learning objective because also in e-learning settings it is important to focus on students’ ability to ‘learn to learn’.

The learning objective “Knowledge” is based on Bloom’s taxonomy for the cognitive domain (1956), and has the following subcategories; Knowledge, Comprehension, Application, Analysis, Synthesis and Evaluation. Several articles (Conole 2004; Conole 2005) have been formulating verbs that belong to each level in Bloom’s taxonomy e.g. the verbs reproduce, arrange, and memorise, which belong to the knowledge level, while the verbs categorize, combine, create and design belong to the ‘synthesis’ level. Similar categorization of verbs can also be found for other taxonomies, e.g. the taxonomies for the affective and the psychomotor domains.

Figure 2 - The learning objective sector. The sector describes how the learning objectives via taxonomies are concretized into technological solutions.

The outer circle presents existing assessment tools to cover the assessment needs of different learning objectives on specific levels. A procedure to find existing assessment tools covering the different learning objectives is to consider each verb mentioned at each level of the four learning objective types and then connect assessment tools to these, e.g. the verbs reproduce, arrange and memorise (from the Knowledge level of Bloom’s taxonomy) have assessment tools like short answer, match and memory connected to them. This exercise is performed on each level of all the taxonomies and the results are shown in the E-learning Circle. Doing this exercise we find that it is necessary to develop more assessment tools within e-learning systems in order to assess the different learning objective levels.

This sector contains elements that describe the student group’s heterogeneity. The “student” sector is divided into the main categories “multiple intelligences”, “proficiency stages” and “culture” (Figure 3). To be able to individualize e-learning to a heterogeneous student group we need to quality assure the production of learning objects based on all categories mentioned.

Figure 3 - The student sector. This sector describes how learning objects need to take into account the various needs of a heterogeneous student group.

The ”multiple intelligences” theory (Gardner 1985) provided a contribution to the discussion about who the learner is. The theory defines eight different intelligences: the visual / spatial intelligence, the verbal / linguistic intelligence, the logical / mathematical intelligence, the bodily / kinaesthetic intelligence, the musical / rhythmic intelligence, the interpersonal intelligence, the intrapersonal intelligence and the naturalistic intelligence. The idea is that all persons have eight intelligences, where some intelligences are more developed than others. The student sector describes what production tools the different intelligences require. Producing learning objects for different intelligences requires a variety of media types and production tools. Learning objects for the visual intelligence can be produced in tools like presentation software, mind maps, graphics tool (raster and vector graphics) and motion graphics tool (animation tool, screen capture tool, motion graphics tool and video editing tools). For the verbal intelligence we have tools like word processors, web editors, and tools to record audio. The logical intelligence needs tools like spread sheets, databases and modelling software.

The kinaesthetic intelligence has until now had few useful tools for e-learning, since simulators (e.g. a flight simulator) have been too expensive to use in e-learning settings. TV-games etc use cheap technology for kinaesthetic in-data with motion sensitive tools, where a web camera captures the user movements, which the system interprets and the user interacts with the system “waving” his hands. The e-learning systems also need to look to TV-games to satisfy the needs of the musical intelligence e.g. karaoke-systems that interprets if the singer hits the right tone. In addition there are midi tools to produce music. The interpersonal intelligence is covered by communication, coordination and cooperation tools (Studio Apertura 2006), while the intrapersonal intelligence is covered by tools like mind maps. The naturalistic intelligence is represented by tools like databases and hypertext editors.

The students also differ when it comes to proficiency stages; novice – advanced beginner – competence – proficiency – expert (Dreyfus 1998). The novice has different needs, e.g. need the help provided by a wizard, compared to a student on the advanced beginner stage, where e.g. a toolkit is useful, or the competence stage, where e.g. a framework is sufficient.

The cultural context also needs to be considered in e-learning. The E-learning Circle uses Hofstede’s (2001) five cultural dimensions; Power distance index, individualism, masculinity, uncertainty avoidance index and long-term orientation. In addition, “Location” is added to the E-learning Circle because one of the main features of e-learning is the opportunity for learning “anywhere”.

The teacher facilitates e-learning, and will use a variety of pedagogical methods based on different theories of learning; Behaviourism, information processing theory, cognitive constructivism and socially oriented theories of learning (Koschmann 1996). Heinich et al. (2002) have specified ten main pedagogical methods. In this article two of them; problem solving and discovery, are merged. The teacher sector (Figure 4) illustrates the nine methods’ relations to different theories of learning. Pedagogical methods are traditionally reusable, and the “teacher” sector shows how the pedagogical methods are implemented in an e-learning environment as learning activities, such as procedural simulation (Alessi & Trollip 2001), chat, animation etc.

The teacher sector illustrates that it is possible to vary the pedagogical methods also in an e-learning setting, and can make it easier to understand and accept that an e-learning system not necessarily needs to be dedicated to one learning theory (e.g. socio-constructivism) if the goal is to vary according to the different needs of the student group and the specific subject’s characteristics. The teacher sector also makes it possible to detect what learning activity tools an e-learning system requires.

Figure 4 - The teacher sector. This sector implies an eclectic view of learning theories within e-learning, and emphasizes that different pedagogical methods must be used when designing learning activities.

The E-learning Circle is a tool dealing with complexity. Vendelhaven (2002) describes how mistakes, information loss, work duplication and misunderstandings are typical problems when the responsibility of segments is moved between persons. The E-learning Circle provides a systematic overview and merges both pedagogical principles and technological solutions, and as a design tool in the process of designing e-learning systems, this is useful dealing with complexity.

In the design of e-learning systems IMS-LD (IMS 2003), a method for modelling learning processes, has been important the last years. IMS-LD critics concern reusability and the teachers’ difficulties to use the specification (Downes 2003, Griffiths & Blat 2005). IMS-LD is not bridging the gap between teachers and software developers, as it is based on specifications from the software engineering field. “New tools and representations are needed if teachers are to intervene in editing and creating units of learning” (Griffiths & Blat 2005). The E-learning Circle is an attempt to provide a tool, which can work as a common language between the teacher and the software developer, which in turn also will be helpful to succeed with participatory design.

Best practice is a process or method that is more effective delivering a specific outcome than other processes or methods. The E-learning Circle has best practice embedded, e.g. the pedagogical methods and proficiency stages are best practice within instructional design.

The E-learning Circle is not a prescriptive method, but a design tool, which may be used to support several processes, e.g. development of requirement specification, user modelling, interface design and choice of system architecture.

The use of the E-learning Circle is not connected to one specific software design model, but it supports different software design models in several phases. In agile methods, the E-learning Circle can be useful in costumer collaboration, which is valued in the Agile manifesto (Beck et al 2001). Using the Unified Process (Booch et al 1999) the E-learning Circle is useful from the perspectives of use case and design views, and specifically within the workflows of Requirements and Analysis and design. The ‘Grimstad model’ (Crossley and Green 1985) was an early software design model with focus on teachers as developers mainly focusing on smaller applications (lessonware), and in this model the E-learning Circle may be useful in phases of idea generation and goal formulation, as well as the phases of metaphor and market design. In prototyping the E-learning Circle is useful for e.g. requirements analysis, user modelling, choice of architecture and interaction.

The use of the E-learning Circle as a design tool of course depends on the software design model in use. The following sections describe the use of the E-learning Circle in two prototyping projects.

Other application areas of the E-learning Circle are to evaluate existing e-learning systems and to quality assure development of e-learning courses.

The E-learning Circle uses specific pedagogical theories e.g. Bloom’s (1956) taxonomy and Gardner’s (1985) multiple intelligences. Some of these theories are controversial within the pedagogical field. The experience of this work, however, is that the theories are useful connecting technology to pedagogical theories. There is however possible to replace theories in the circle, for example the taxonomies used in the sector “learning objectives” could be replaced by Anderson’s (2001) revised taxonomy for the cognitive domain, Harrow’s (1972) or Simpson’s (1972) taxonomies for the psychomotor domain. Such a replacement will not influence the outer circle in a large extent.

Also concerning the student there are many pedagogical theories; Coffield et al. (2004) identified 71 models of learning styles. Some of these models are alternatives on how to describe the student. The pedagogical discussion of different theories, e.g. student descriptions, is therefore not the main issue. The important issue is that we are able to describe the student and able to connect technology to these descriptions. This can be regarded as both a weakness and strength of the E-learning Circle. The strength is that the circle remains the focus on the subject’s characteristics, and does not regard one pedagogical theory as the single most useful theory. The drawback is that the circle can be diluted if different users keep replacing the theories.

It can also be regarded as a weakness to use Dreyfus’ (1998) proficiency stages since these originally are used in skill acquisition. In most learning situations the learners are on different stages in the learning process, and therefore Dreyfus’ stages are used generically in the E-learning Circle. The E-learning Circle also shows that there are not many technological tools to support the fact that students are on different stages in the learning process, and indicates that more research should be conducted on these questions.

One may discuss whether meta-learning could be placed under ‘skills’. Learning to learn, e.g. learning to work in groups, learning to give constructive critics etc. can be considered skills, but the E-learning Circle keeps meta-learning as an additional learning objective to face the challenges of e-learning in life-long learning. Adding meta-learning next to the traditional learning objective types (knowledge, skills and attitudes) is one of the contributions to the e-learning field made through the E-learning Circle.

Robson (2004) claimed that “context is the friend of learning and the enemy of reuse”, and explains the problems of learning objects and reuse. The E-learning Circle suggests instead of specific context some generic student “types”, similar to “personas”, that we need to have in mind both when developing e-learning systems and when designing a course.

The E-learning Circle illustrates how learning objects must be retrievable for the students, based on different criteria like degree of difficulty, intelligence support and cultural dimensions. It is important that the learning objects are presented to the student without the danger of information overload. This is of great importance planning the human-computer interface of e-learning systems.

One aim of the study was to implement the pedagogical principle of individualization and the question “Why is not the student placed in the circle’s centre?” then becomes apparent. If the student is the main focus, many of the sectors would disappear. Focus on the subject taught is empirical-based and provides a holistic approach to e-learning, including learning objectives, assessment, student needs, learning objects, teaching methods and learning activities. The E-learning Circle implies an eclectic view of learning theories, and illustrates that different subjects have various characteristics and needs when it comes to learning theories and technology, but is still able to implement individualization.

The UNFOLD project claims activity-based learning is the next generation e-learning (Griffiths 2004). The E-learning Circle shows that learning activities are just one of many factors that are important in e-learning.

Dye et al (2005) suggests that mobile learning is the next-generation e-learning. Their argument is understandable if the technological solution is the main factor in the transition from one generation to another. The E-learning Circle focuses on pedagogy connected to technology, and illustrates how mobile learning is one of many student group heterogeneities. If mobile learning is to be considered the next generation e-learning, it is important that lessons learned from computer-based learning are remembered when moving to another technological platform. The E-learning Circle can contribute to quality assure that pedagogical principles are covered also in a mobile e-learning environment.

Wibeck (2000) argues that brainstorming is a technique which does not belong within the definition of focus groups, because it does not allow criticizing each other’s ideas. The experience was, however, that brainstorming provided valuable data, encouraging creativity and creating new ideas. Data collection through focus groups was chosen because it provides multiple user perspectives, and because the group interaction allows creative ideas to thrive. This allows the researcher to ‘step back’ to a larger extent than in ordinary interview settings, because as focus group participants share insights and ideas, each individual member will respond, interact and continue the process. This was useful in an exploratory study in order to move ahead.

In this study, where focus is on how to improve the design process of e-learning systems, expert groups of software developers and the problem solving activities were valuable in order to stay close to the studied world (Charmaz 2005) and to provide data, which cover the intersection between pedagogy and technology. The aim of the study was to bridge pedagogy and software development, and expert groups were useful in order to bring pedagogical ideas into the situation of a software design process.

How to deal with the literature within Grounded Theory is often discussed, and Urquhart (2001) claims that “one of the oft quoted misconceptions about Grounded Theory is that the researcher does not do any literature searching”, but emphasizes that the Grounded theory researcher has to relate to literature in a slightly different way to a conventional researcher. “So literature is used to help build the theory, and the substantive theory is related to the literature, but only once the substantive theory has been developed” (Urquhart 2001). The E-learning Circle includes well-known pedagogical theories, but these were included in a late stage of the development of the E-learning Circle and are based on the empirical data, where coding and comparisons revealed that the pedagogical theories were hidden in the empirical data.

Since the Grounded Theory approach originated with the work of Barney Glaser and Anselm Strauss in their book “The Discovery of Grounded Theory” in 1967, not only the two originators of Grounded Theory have moved in slightly different directions, but also other researchers have adopted and adapted the Grounded Theory, which has lead to alternative versions of Grounded Theory (Denscombe, 2003).

Smit (1999) describes the divergence between Glaser and Strauss, the two originators of the Grounded Theory method. “Strauss and Corbin mention that they set out to provide clear, straightforward, and basic information on the knowledge and procedures needed by researchers who want to build their first theory at a substantive level” (Smit, 1999). “Glaser argues that what Strauss and Corbin describe will not produce a Grounded Theory, but rather ‘a forced, preconceived, full conceptual description, which is fine, but is not Grounded Theory’ (Smit, 1999). Locke (1996) notes that there are no differences between Glaser and Strauss’s positions on the key analytical procedures (constant comparison and theoretical sampling) involved in Grounded Theory methodology. “However, they do write subsequently different renditions of researchers’ relationships to the worlds they study” (Locke, 1996). “With the Glasarian approach the researcher allows the theory to emerge from the data, whilst the Straussian approach the researcher interrogates the data in order to arrive at a theory” (Smit, 1999). This research is inspired by the Strauss’s version of Grounded Theory, based on the interpretive approach and the interrogation with the data.

Critics of the Grounded Theory approach include the epistemological positions. Denscombe (2003) describes how Glaser’s version is positivistic, while Strauss’ version is interpretive. “Glaser’s version rest on the belief that: (a) the researcher should maintain a distance and independence from the data; and (b) the meaning of the data will emerge inductively from the data if studied using a suitably neutral methodology... Contrasting with this, there is Strauss’s version, which is more in line with interpretivism, in that the role of the researcher is to go looking for the meaning that the data hold, possibly probing beyond their superficial meaning” (Denscombe, 2003).

Pandit (1996) describes some of the problems with Grounded Theory; First, the Grounded Theory research is extremely time-consuming; second, the Grounded Theory research involves long periods of uncertainty. Third, Grounded Theory research requires certain qualities of the researcher e.g. confidence, creativity and experience. Accordingly, the novice researcher are likely to find the approach more difficult than more conventional methodologies and the more experienced researcher is likely to produce better theory” (Pandit, 1996).

One disadvantage using the Grounded Theory method is that “the approach does not lend itself to precise planning” (Denscombe, 2003). Not being strictly dependent on a plan created early in the project has been interesting, and our opinion this made the results of this project better. Grounded Theory is experienced as useful in this exploratory project, and instead of regarding no precise plan as a weakness of the method; this allows an exploratory and creative approach and strengthens the research project.

Also the fact that Grounded Theory allows a variety of qualitative data collection methods (e.g. interviews, observations, document analysis) is an advantage. Together with the rich tools and techniques, the method helps the novice researcher in the process of analyzing the data.

“Grounded Theory has been increasing in popularity in Information Systems as a research method. This is evidenced by the growing literature that is either discursive on philosophy and application or detailed about the method” (Hughes & Jones, 2003). Hughes and Howcroft (2000) point out that there are four inconsistencies in the understanding and application of the Grounded Theory method in IS research. First, the projects range from those concerned with organizational change to those concerned with the practical use of the method to inform knowledge based systems design. Second, some use the method prescriptively, whilst others use some of its procedures to supplement other research strategies. Third, the underlying assumptions made explicit by the researchers range from qualitative-interpretive to qualitative-positivist and finally, Grounded Theory is used on its own or alongside other methods (Hughes & Howcroft, 2000).