The Three Spaces Model for Online CPD

Introduction In recent years, an extensive body of research has been concerned with course design in face-to-face, online and blended education in different educational contexts, from primary school to university level (Nortvig, Balle & Petersen, 2018). Few have, however, addressed course design in online education for Continuing Professional Development (CPD), for which reason this is the focus of the present paper. Online education can be defined as courses that are offered completely online (Ryan et al. 2016), and the most distinctive feature of the concept is thus the absence of the physical classroom, which is replaced by web-based technologies, i.e. learning management systems (LMS) or virtual learning environments (VLE), that offer opportunities for out-of-class learning independent of time, place and pace (Bernard et al. 2014; Israel, 2015; Potter, 2015). In online learning environments, the participants’ engagement with online assignments – and the professor’s feedback – is particularly important as students and professors may never meet face-to-face. Furthermore, it is equally important that assignments for online education relate to the course participants’ professional working lives as they bring with them years of experience and learning needs that are situated in their own practice. In addition to this, the curriculum of courses addressing professionals includes both content and skills, which means that course design should be developed to support knowledge transmission as well as skills acquisition (Heinerichs, Pazzaglia & Gilboy, 2016). Ideally, in online CPD courses for professional teachers, assignments should also be exemplary in relation to the participants’ own practice in which they are also designers of assignments (usually for pupils in primary and lowersecondary school). With this in mind, the paper seeks to investigate the following research question: what characterises assignments that support CPD students’ active learning in online courses and what are the implications for the design of online courses for CPD? The paper is organised into three main sections; firstly, it explores the extent to which two widely used models for course design, Salmon’s five-stage-model (2002, 2003) and Ryan & Ryan’s TARL model (2013), are useful for course designers designing online CPD courses. Secondly, it seeks to shed light on what characterises successful course designs by comparing the two models with the underlying design patterns of a number of assignments that have been developed by experienced professors. Thirdly, on the basis of Salmon’s five-stage-model and Ryan & Ryan’s TARL model as well as the abovementioned design patterns, a new model that can support CPD course designers in designing assignments, the Three Spaces Model for Online CPD, is presented and discussed. RESEARCH


Introduction
In recent years, an extensive body of research has been concerned with course design in face-to-face, online and blended education in different educational contexts, from primary school to university level (Nortvig, Balle & Petersen, 2018). Few have, however, addressed course design in online education for Continuing Professional Development (CPD), for which reason this is the focus of the present paper. Online education can be defined as courses that are offered completely online (Ryan et al. 2016), and the most distinctive feature of the concept is thus the absence of the physical classroom, which is replaced by web-based technologies, i.e. learning management systems (LMS) or virtual learning environments (VLE), that offer opportunities for out-of-class learning independent of time, place and pace (Bernard et al. 2014;Israel, 2015;Potter, 2015). In online learning environments, the participants' engagement with online assignments -and the professor's feedback -is particularly important as students and professors may never meet face-to-face.
Furthermore, it is equally important that assignments for online education relate to the course participants' professional working lives as they bring with them years of experience and learning needs that are situated in their own practice. In addition to this, the curriculum of courses addressing professionals includes both content and skills, which means that course design should be developed to support knowledge transmission as well as skills acquisition (Heinerichs, Pazzaglia & Gilboy, 2016). Ideally, in online CPD courses for professional teachers, assignments should also be exemplary in relation to the participants' own practice in which they are also designers of assignments (usually for pupils in primary and lowersecondary school). With this in mind, the paper seeks to investigate the following research question: what characterises assignments that support CPD students' active learning in online courses and what are the implications for the design of online courses for CPD?
The paper is organised into three main sections; firstly, it explores the extent to which two widely used models for course design, Salmon's five-stage-model (2002Salmon's five-stage-model ( , 2003 and Ryan & Ryan's TARL model (2013), are useful for course designers designing online CPD courses. Secondly, it seeks to shed light on what characterises successful course designs by comparing the two models with the underlying design patterns of a number of assignments that have been developed by experienced professors. Thirdly, on the basis of Salmon's five-stage-model and Ryan & Ryan's TARL model as well as the abovementioned design patterns, a new model that can support CPD course designers in designing assignments, the Three Spaces Model for Online CPD, is presented and discussed.

RESEARCH
The Three Spaces Model for Online CPD -a model for designing assignments for online courses in Continuing Professional Development Malene Erkmann, Anne Kristine Petersen and Pernille Lomholt Christensen The paper explores the challenges of designing assignments for online learning environments and looks into the use of models as analytic thinking tools for course designers. The paper opens with a discussion on challenges central to designing assignments for online learning environments in higher education. Subsequently, two widely used models for course design, Salmon's five-stage-model (2002Salmon's five-stage-model ( , 2003 and Ryan & Ryan's TARL model (2013), are explored with the aim of evaluating their usefulness in Continuing Professional Development (CPD) for teachers and pre-school teachers, a context which has received relatively little attention in terms of research on course design. A number of assignments that have been used in online CPD courses for (pre-)school teachers are analysed with the aim of identifying design patterns, i.e. examples of how recurring pedagogical problems can be solved and, on the basis of this, a new model that can support CPD course designers in designing assignments, the Three Spaces Model For Online CPD, is presented and discussed.

Theoretical Framework
In the paper, we draw on case study methodology, following Yin's (1994) definition: 'A case study is an empirical inquiry that investigates a contemporary phenomenon within its real-life context, especially when the boundaries between phenomenon and context are not clearly evident'.
The real-life phenomenon to which the present study relates are assignments for online CPD courses designed to make learning social and to bring it closer to the participant's own environment or own practice. The most important aspects of teachers' learning (CPD) are discussed by Putnam and Borko (2000), who argue that it is important to include a high degree of situated learning, either in the participating teachers' own practice or by bringing their practice into the course room, for instance via video recordings. Another crucial aspect are authentic activities that are similar to the activities the teacher conducts in his or her own practice. Social interaction, and particularly social interaction with people in the teachers' professional practice, are mentioned as key elements in effective learning in CPD courses (Putnam and Borko, 2000).
As one of the defining characteristics of online education is the fact that the students and their professor do not meet face-to-face in the physical classroom, we are particularly interested in examining assignments that are successful in supporting peer-to-peer interaction as well as interaction between the participants and people with whom they collaborate in their professional practice (e.g. colleagues, principals, pupils, etc.).Consequently, the professors' accounts of unsuccessful assignments fall beyond the scope of the present study and are not discussed here.
The empirical data collected are treated as design narratives, which are further described below and include assignments that have been used for online CPD courses for (pre)school teachers carried out at a Danish university college in the period of 2016-2017. The overall aim of the courses is to provide learning experiences that are closely related to the participants' own practice, as stressed by Putnam and Borko (2000). The term assignment is used here to refer to assignments that are shared online and have been designed by a professor to increase the students' learning outcome. The assignments that are included for analysis in the paper have been chosen on the basis of the following criteria: They are described as meaningful by the professors based on their course evaluations, they have resulted in a positive learning outcome and they support knowledge transmission as well as skills acquisition. It should be stressed that the assessment of the students' learning outcome and their experience of meaningfulness is solely based on the objective evaluation of the professors, which is based on feedback from their students and data from learning management systems.
The collection of empirical data in the form of assignments is theoretically based on three basic concepts for articulating design knowledge: design patterns, design narratives and design scenarios (Mor, 2013).
A design pattern can be described as a pedagogical pattern (Laurillard, 2012), that is, a semi-structured description of the way in which an experienced teacher has solved a problem related to a specific teaching and learning context. The aim of a design pattern is to externalise knowledge (Goodyear, 2005;Mor & Winters, 2007) so that it can be shared, criticised and redesigned by a community of teachers (Laurillard, 2012).
Design patterns can be deduced through analysis of design narratives (Mor, 2013). A design narrative is a teacher's account of critical events in her own practice narrated from a personal perspective. It describes a problem in a specific domain, the actions taken to resolve it and their unfolding effects. Thus, it articulates knowledge about how a design works in practice (Warburton & Mor, 2015). Our design narratives, which are further discussed below, are deduced from assignments developed by professors at a Danish university college and tested in online CPD courses for (pre-)school teachers.
A design scenario can act both as a scientific category for analysis and as a practical design tool (Mor, 2013). As a scientific category, a design scenario can be used for validating the statements that have been deduced from a design narrative and subsequently have been generalised as a design pattern. This is done through analysis of how effective a pattern is perceived in a number of hypothetical scenarios. The validity of a pattern increases if it is perceived as meaningful not only in the context from which it was originally deduced, but also in hypothetical future contexts. Both design narratives, design patterns and design scenarios constitute a description of a context (a challenge, a solution and its expected results), but a design scenario is formulated as a postulate (if x is done in z way, y will happen). In this way, a design scenario can both validate reflections on the past and can be used to design the future (Warburton & Mor, 2015).

Method
The empirical data for the present study include assignments that have been used for online CPD courses for (pre-)school teachers in the period of 2016-2017. As mentioned, the assignments that are included for analysis have been chosen on the basis of the following criteria: They are described as meaningful by the course participants, they have resulted in a positive learning outcome and they support knowledge transmission as well as skills acquisition.
In the study, a group of three researchers (the authors) conducted a small-scale case study in which a number of professors at a Danish university college were asked to share one or more written design narratives that exemplify what they perceive as their most successful assignments. Subsequently, the professors were each asked to further discuss their design narratives in a semi-structured interview with one of the researchers. Based on the professors' written design narratives and the semi-structured interviews, the researchers discussed the characteristics of the assignments and deduced 10 parameters that make up the underlying design patterns of the assignments. On the basis of the 10 parameters, a design scenario is presented and visualised in the present paper in the form of a model, the Three Spaces Model for Online CPD, which can be used as a tool for designing online courses for CPD.

Ethical considerations
The data used has not been produced directly by the students, but consists of the professors' experiences with specific assignments, i.e. their design narratives. The professors' design narratives are based on systematic, ongoing written and oral course evaluations carried out with the students. All names, including student and professor names, have been anonymised for this study.

Models for Designing Courses and Assignments
Pedagogical models may serve as analytic thinking tools for course designers, both in relation to designing courses and assignments. We define a thinking tool as ' a researchbased and systematic tool, which is not normative in relation to practice in the sense of specific rules or descriptive to a certain practice' (Staunaes & Bjerg, 2013, our translation). In the following, we analyse two models for course design, Salmon's five-stage-model (2002Salmon's five-stage-model ( , 2003 and Ryan & Ryan's TARL (2013) model, with the aim of identifying conformities between them as well as potential shortcomings in relation to online CPD courses. The two models, which have been chosen because they are both widely used for designing courses in higher education in a Danish context, will serve as a starting point for our analysis of the empirical data collected.
Developed on the basis of research into online education and training, Salmon's five-stage-model aims at offering support and development to participants at 5 different stages as they build up expertise in learning online through e-tivities (Salmon, 2002(Salmon, , 2003. The model, which takes a social constructivist approach to online teaching and encourages collaboration between peers, is widely used to scaffold online courses in higher education (Lisewski & Joyce, 2003;Hughes et al. 2004).
In this context, we focus on the main themes of the model: 1) Access and motivation, 2) Online socialization, 3) Information exchange, 4) Knowledge construction and 5) Development (see Figure 1 below). The aspects of the model related to technical support are not discussed here as they are beyond the scope of the paper.
Salmon's five-stage-model serves as a framework for the teacher to design, for example, the opportunity for motivation and socialization among the participants, so they can later build up knowledge with other participants.
While Salmon's model is primarily concerned with supporting course designers in making informed choices in relation to e-moderating, Ryan and Ryan's (2013) TARL (Teaching and Assessing Reflective Learning) model is concerned with scaffolding students' abilities to reflect throughout a course or programme. Thus, they have proposed what they refer to as a 'systematic and deliberate approach […] to teaching and assessing reflective learning across whole programmes/courses in higher education' (Ryan & Ryan, 2013: 248). Their approach is illustrated in a model that is customisable and transferable to other contexts in higher education, and its main purpose is to ' describe the pedagogical 'landscape' associated with reflection so that effective pedagogic choices can be made' (ibid.: 248).
The model consists of a pedagogic field, as illustrated in Figure 2 below, which has a category-based dimension (the vertical axis) concerned with levels of reflection or application of higher ideas and a development-based dimension (the horizontal axis) concerned with the participants' reflection over time as they progress through a course.
The category-based dimension consists of two replaceable scales, which can be customised to specific courses or programmes, and a fixed scale, which we will Assessing Reflective Learning) model (Ryan & Ryan, 2013). The shaded areas highlight an assumed trend, whereby the students will be able to engage in higher levels of reflection in relation to their professional practice. focus on here. The fixed category-based scale is inspired by Bain et al. (2002) and their 5Rs model of reporting, responding, relating, reasoning and reconstructing. As indicated in Figure 2, the fixed scale increases in complexity, moving from description of and personal response to an issue or situation, to the use of theory and experience to explain, interrogate and ultimately transform practice. This is illustrated in Taylor and Ryan's (2015: 120) 4Rs Model of Reflective Thinking in Table 1, which is also discussed in the study by Ryan and Ryan (2013). The development-based dimension tries to capture the different demands as students progress through a course or program over time. This dimension also relates to the subject matter of reflective activities across time; in the beginning of a course or program the participants need to reflect on ideas and contexts that are familiar to them and then move on to what is new in terms of novel theory and practice-related knowledge and skills. Likewise, the focus moves from self, own views and one's own place in society, to the students' peers and society in general and, finally, to their (future) colleagues, clients and society in relation to their professional practice (Ryan & Ryan, 2013: 251). Moreover, it is assumed that early experiences with reflection are best carried out in a simulated context, e.g. through using scenarios or problem-based learning, while in the final periods of the course or program, reflections can be carried out in a real context, i.e. the students' professional workplace.
The strengths of the model lie in the fact that it provides a means for course designers to include deep reflection at different points across a course, so that the participants have the skills to critically engage with the theories and practices introduced along the way and, moreover, it prioritises both time and contextual space in pedagogical decision-making.
To sum up, Salmon's five-stage-model describes the levels of technical support and e-moderation needed in online learning environments and, furthermore, the model is concerned with progression in relation to students' immersion in the online space. Ryan & Ryan's TARL Model is also concerned with progression, but here the focus is on taxonomical progression in relation to levels of reflection or application of higher ideas and participants' reflection over time as they progress through a course. Both models are useful in supporting course designers in making effective pedagogical choices when designing online courses, but they do not focus on scaffolding students' collaboration with peers outside the online learning environment, which is highly relevant in CPD.

Analysis of the empirical data
In the following section, we look into a number of assignments that have been developed by professors at a Danish university college and tested in online CPD courses for (pre-)school teachers. The aim of the analysis is to evaluate what in their perspectives constitutes successful assignments in CPD and to discuss the kind of support that course designers need for developing successful assignments. The analysis is, as previously mentioned, based on the professors' written design narratives and semi-structured interviews with the same professors in which they discussed and elaborated on their design narratives.
For the purposes of this paper, we focus on describing the underlying pedagogical concept of the professors' design narratives. The four assignments that we have included for analysis include: • Empirical analysis using Pepito Pearls (a type of candy) in which the participants practiced and discussed the nature of empirical analysis. • Observation of classrooms using 360-degree photos in which the participants shared their classroom observations. • Analysis of questionnaires from a public library, which was used as a starting point for the participants to design and test their own questionnaires. • Shared reading experience in which all participants were responsible for sharing their reflections on specific texts. This provided the participants with blinkist 1 resources and the professors gained insight into the participants' understanding of the texts.
Our analysis of the professors' written design narratives and the semi-structured interviews, shows that all of the above-mentioned assignments consist of several subassignments. For this reason, each design narrative was divided into a number of sub-narratives. For instance, the assignment in which participants were to observe a classroom using a 360-degree photo consisted of the following sub-assignments: 1. Observe, make notes and relate to the literature provided by the professor. 2. Share your observations in the online course space. 3. Comment on the observations made by your peers by drawing on the texts provided. 4. Share your reflections: are your observations similar or do they differ? Why/why not?
In our analysis (see Table 2) of the design narratives, we have deduced 10 parameters that make up the underlying design patterns of the assignments, i.e. the parameters serve to show how recurring pedagogical problems can be solved. The 10 parameters include: • Peer-to-peer collaboration or interaction with people from the professional practice (e.g. colleagues, principals, pupils, etc.). • Use of a common artefact.
• Reusability (the extent to which an assignment can be reused by others). • The nature of interaction (e.g. in terms of Salmon's 5 stages (see Figure 1)).
• Progression (in terms of Ryan & Ryan's 4Rs as described in Table 1 and Salmon's five stages as illustrated in Figure 1). • Learning resources (e.g. texts, videos, etc.).
• The technological platform.
• Time frame of the assignment.
The 10 parameters were used for describing the design patterns, as exemplified here in the form of an assignment.
Generally speaking, the assignments that are less complex were mostly placed at the beginning of the course. Typically, these assignments were to be solved individually, whereas assignments that are more complex were to be solved in collaboration with others, either in the course space or in the participant's professional work space. This supports previous findings about social learning (Vygotsky, 1978), but at the same time, it suggests that collaboration in online courses might take place in spaces other than the online course space. In the case of the assignment on empirical analysis using Pepito Pearls, the sub-assignments were also conducted individually by the students in the private space, followed by assignments of sharing and reflecting on findings in the course space. In other cases, however, the sub-assignments were conducted in collaboration with partners from the professional work space, as was the case for two separate assignments in which the participants interviewed their principal and were to design an intervention in their own classrooms in order to test new knowledge.
Together the assignments reflect the kind of progression illustrated in Ryan & Ryan's (2013) 4Rs model; the least complex assignment involves simple everyday tasks like reporting or repeating, more complex assignments operate further away from the participant's life in a more abstract course space, where the participants are asked to relate and reason, and the highest level of complexity is reached when participants are asked to reconstruct knowledge in relation to or even within their own professional work space.

Type of collaboration
The assignment did not involve collaboration Use of a common artefact Pepito Pearls (available everywhere in DK) Sharing/reusing Reusable as is, but difficult to redesign The nature of interaction The participants were to buy Pepito Pearls, taste them and measure them

Scaffolding elements Assignment instruction on video and in text
The professor's presence The professor was occasionally present in the online learning environment

Progression
The assignment took place at the beginning of the course and had a low level of complexity

Learning resources Video and academic texts on empirical analysis
The technological platform The learning platform used for the course and an app for timing Time frame Approximately 1 hour

Discussion
On the basis of the above analysis, we propose a new model that can serve as a thinking tool (Staunaes, 2013) for designing online courses and assignments for Continuing Professional Development. The model differs from Salmon's five-stage-model (2002Salmon's five-stage-model ( , 2003 and Ryan & Ryan's (2013) TARL model in two ways; in Salmon's five-stage-model assignments are solved in the course space, and Ryan & Ryan's model focuses on assignments that relate to the student's (future) professional practice, whereas the model proposed here emphasises the potentials of involving both the private space and, in particular, the professional work space in the participants' learning experience. The 4 four levels adapted from Taylor & Ryan (2015), i.e. reconstructing, reasoning, relating and reporting (Table 1), have been adopted in our model. Furthermore, it is suggested that participants are scaffolded to move through the three spaces illustrated in the model below (see Figure 3), i.e. from the private space to the course space and into the professional work space, while at the same time moving through increasing levels of complexity (the vertical axis), from simple reporting in a familiar setting, to relating and reasoning in relation to academic contents within the course space and to reconstructing within their professional work space. It should be noted that a common artefact provided by the professor or the participants themselves (e.g. a template for writing an interview guide or candy to be used in an empirical study) may help in creating a common ground for the participants, even though their prior knowledge as well their work situation may be very different. This can be seen as a way of familiarising and providing bridges between different cultures and learning environments (Salmon, 2016).
The model covers both course progression over time and/or progression within a specific module or theme (the horizontal axis), as well as progression in relation to the complexity of assignments (the vertical axis), which supports both knowledge transmission and skills acquisition. Also, the model embraces the fact that the course design and the assignments should scaffold immersion into the online space and support engagement in activities outside the course, e.g. in the participants' work space, thus ensuring long-lasting transfer of the knowledge and skills acquired during the course into everyday work situations.
Throughout the course, the participant will move from engaging with activities that take place in the private space and are characterised by reporting and relating, to activities that take place in the participant's work space or everyday practice and are characterised by higher complexity, such as reconstructing and reasoning.
The assignment in which participants are introduced to empirical data collection by measuring Pepito Pearls (the aforementioned candy) primarily focuses on reporting and relating and was placed in the beginning of a course. The participants primarily worked in their private space but shared their findings. Later, when participants were used to working in teams with their peers, further levels of complexity were introduced in the assignments. Finally, at the end of the course, assignments such as designing an activity for co-workers or pupils can be introduced, which situate the assignments closer to the participant's own professional practice. Figure 4 below shows how the assignments can be plotted into the model according to their level of complexity, course progression over time and the three different spaces. For instance, the assignment in which the participants observed classrooms using 360-degree photos and shared their observations in the online platform was placed at the beginning of the course and involved a medium level of complexity (reporting and relating). The assignment in which the participants were to analyse questionnaires from a public library and subsequently design and test their own questionnaires was placed at the end of the course and involved a high level of complexity (reconstructing).
The assignments that have been included for analysis here, typically consist of 3-4 sub-assignments which, when plotted into the model, spread from the left to the right and/or from the bottom to the top of the model (see Figure 5). For instance, in an assignment in which the participants were to interview a principal, the following three sub-assignments were included: 1) preparing an interview guide, 2) giving and receiving peer-feedback on the interview guide and 3) interviewing a principal using the interview guide. The three sub-assignments range from the private space (preparing the interview guide), to the course space (giving and receiving peerfeedback), to the professional work space (interviewing a principal). In addition, the three sub-assignments range from low complexity (reporting and relating to questionnaires) to medium complexity (reasoning with peers) and high complexity (reconstructing new knowledge through interviews).
The model (as illustrated in Figure 4) may serve as an analytic thinking tool for course designers involved in designing online CPD courses, as it encourages and supports course designs that cover different levels of complexity, considers course progression over time and addresses the three different spaces (the private space, the course space and the professional work space). This is particularly important in CPD for teachers because, ideally, assignments should be exemplary in relation to the participants' own practice in which they are also designers of assignments (usually for pupils in primary and lower-secondary school). The fact that there is a need for CPD to be exemplary is supported by recent studies on assignments used in primary and lower-secondary schools in Denmark.
In three consecutive studies, Bremholm et al. (2014Bremholm et al. ( , 2016Bremholm et al. ( , 2017 found that the assignments that are often used by teachers involve primarily routine and training activities that focus on reproduction and repetition of concepts or procedures already known by the pupils. Such assignments are placed in the lower left corner of  the model, and if they are not combined with other types of assignments, they will lead to a relatively low learning outcome. In addition, there is a high probability that such assignments are not perceived as meaningful by the pupils because they may not clearly relate to the overall contents of the subject (Bremholm, 2016). Moreover, in a study focusing on the primary and lower-secondary subjects Natural Sciences, Danish and Mathematics, it was found that only 20% of the artefacts produced by the pupils involved peer-collaboration (Bremholm, 2014), which is, seen from a constructivist perspective, a rather low percentage. This is supported by another study, which found that many assignments encourage group work, but that collaboration does not always take place because group work is oftentimes not sufficiently scaffolded by the teachers (Bremholm, 2016).
Finally, in a study on the quality of assignments in primary and lower-secondary school, in which data were collected through observation, interviews, classroom video recordings and collection of students' artefacts, it was found that there is limited focus on the pupils' productive work and academic contents, whereas form and technical difficulties received much attention in the classroom (Bremholm, 2016). This is referred to as professional emptying (Bremholm, 2016), which means that: 1) the academic criteria of assignments are vague, 2) scaffolding of the pupils' in their learning process is limited or does not occur and 3) the framework of assignments counteracts a professional focus, e.g. because the requirements of the assignments relate more to form than to academic contents.
These findings are relevant in relation to CPD courses for primary and lower-secondary teachers because they stress the importance of including exemplary assignments in CPD courses, and a need for a model, such as the one presented here, that can serve as a tool for course designers.

Conclusion
In the paper, we have explored the challenges of designing assignments for online courses in Continuing Professional Development (CPD), and we have looked into two widely used models for course design, namely Salmon's five-stage-model (2002Salmon's five-stage-model ( , 2003 and Ryan & Ryan's TARL Model (2013). It is concluded that the models are indeed useful in supporting course designers in making effective pedagogical choices when designing (online) courses, but as they do not focus on scaffolding students' collaboration with peers outside the online learning environment, they are less useful for designing online CPD courses.
For this reason, the Three Spaces Model for Online CPD is proposed as a thinking tool (Staunaes, 2013) for designing online courses and assignments for CPD. The model differs from other models in that it addresses both the level of complexity in assignments and course progression over time in relation to the three spaces that are involved in online CPD: the private space, the course space and the professional work space. The model is informed by analyses of design narratives that are considered exemplary by course designers at a Danish university college, and it is argued that exemplarity is particularly important in CPD targeting teachers, as the participants are also course designers in their own professional practice.
Thus, the paper demonstrates that the Three Spaces Model for Online CPD is useful as a thinking tool for analysing and evaluating course designs and assignments in online CPD. However, further empirical research is needed to test and validate the usefulness of the model as a thinking tool for course designers involved in designing assignments across whole courses.

Note
1 Blinkist is an app that summarises non-fiction books into digestible summaries.