A phenomenographic approach to research in medical education


Terese Stenfors-Hayes, Department of Learning, Informatics, Management and Ethics, Karolinska Institutet, Berzelius Vag 3, Stockholm 17177, Sweden. Tel: 00 46 8 524 837 37; E-mail: terese.stenfors-hayes@ki.se


Context  Phenomenography is a qualitative approach to research which has revolutionised the way that researchers and teachers think about the processes and outcomes of learning in higher education. Phenomenography has also been used successfully in medical and health care research for the last 20 years. Phenomenography provides a lens through which to view certain types of research question. It also provides direction for how to empirically carry out the research.

Methods  This paper introduces phenomenography as a viable qualitative approach for use in medical education research.

Results  A phenomenographic study maps the qualitatively different ways in which people experience a phenomenon. This type of study can have an important impact on, for example, patient communication, clinical practice and health care education.

Conclusion  We suggest that a phenomenographic approach can be used to explore many medical education research issues, and can facilitate more solid links between research and educational development and change.


Interest in qualitative methodologies in medical education research has increased substantially in recent years. 1–3 The aim of this paper is to introduce phenomenography as a viable qualitative approach in medical education research. Phenomenography is the empirical study of the different ways in which people understand various phenomena in the world around them and how these ways of understanding are related to one another.4 Although phenomenography represents a well-established approach in the fields of education and health care research, very few studies using a phenomenographic approach have been published in medical education journals. In this paper, we will provide examples of how phenomenography has been applied in studies related to medical education and give an introduction on how to conduct a phenomenographic study.

Qualitative research often concerns the development of concepts that help us to better understand social phenomena in natural settings by emphasising the meanings, views and experiences of their participants.5 Qualitative and quantitative research are not opposites, but, rather, complementary approaches in that they seek to address different types of question and provide different kinds of answer. Qualitative research commonly involves methods of data collection such as interviews, focus groups, case studies, observations and textual analysis. However, these are just tools to gather data. These methods or tools need to be linked to theory and a research methodology1,3,6,7 so that the philosophical underpinnings of the study (such as epistemology and ontology) can be understood.8 The lack of a theoretical underpinning has been a cause of criticism of qualitative research within medical education. 2,5 In a recent overview of research paradigms in medical education research, Bunniss and Kelly6 outlined four major paradigms: positivism; post-positivism; interpretivism, and critical theory. Phenomenography fits within the interpretivism paradigm, which acknowledges that there are multiple, diverse interpretations of reality. In this paper, we present the phenomenographic approach in order to help expand the methodological toolbox and widen the theoretical and philosophical horizons of qualitative medical education research.

What is phenomenography?

Phenomenography was developed as an empirical approach in educational research by Marton, Säljö, Dahlgren and Svensson at the University of Gothenburg in Sweden through a series of studies of learning in higher education carried out in the 1970s. In phenomenography, learning has been described as a transition from one way of understanding a phenomenon to another. 9 More recent theorisations derived from phenomenographic research have articulated the variation theory of learning. 10,11 The variation theory emphasises the learner’s experiences of variation and discernment as conditions necessary for learning to take place. A transition from one way of understanding a phenomenon to another hence means a change in the features that are noticed and focused on simultaneously. In a seminal paper introducing phenomenography, Marton wrote:

‘Phenomenography is a research method for mapping the qualitatively different ways in which people experience, conceptualise, perceive, and understand various aspects of, and phenomena in, the world around them.’12

There are similarities between phenomenography and phenomenology in that they both focus on human experience and awareness as the object of research. Other similarities are that human experience and awareness of phenomena in the world are explored through people’s oral and written descriptions of the phenomena. Phenomenography and phenomenology, however, differ in terms of their underlying ontology and the outcomes of their analyses. In phenomenography, the outcome focus is on the variation in experiences of a given phenomenon. The result of a phenomenological analysis is the description of the essence of the lived experience of a given phenomenon. 10,13–15 Further explorations of the differences between phenomenography and phenomenology can be found elsewhere. 15 For readers interested in the differences between phenomenography and the various tenets of grounded theory, we recommend the texts of Cresswell16 and Kennedy and Lingard.17

How can phenomenography be used?

Phenomenography has revolutionised the way in which both researchers and teachers think about the processes and outcomes of learning in higher education. 18 Phenomenography has been used to explore approaches to learning, 19,20 and the multiple ways in which medical students come to understand scientific phenomena presented in medical school21 and how they experience interactions during a patient encounter. 22 Phenomenography has also been used to explore approaches to teaching,20,23 including how different teacher roles, such as those of lecturer, preceptor and mentor, and the relationships between them are variously conceptualised.24–26 One of the most well-known findings of a phenomenographic study resulted in the metaphor of deep and surface approaches to learning and studying.27 This metaphor has been referred to as a foundation stone for much research, theory and practice in higher education,28 proving central to helping both staff and students, and is often used to encourage more student-centred learning and teaching. The metaphor provides a powerful, simple idea which conveys complex pedagogical principles in a readily accessible way.29 Many studies during the last 30 years have provided further empirical evidence supporting these original studies.

In health care, patients understand their illnesses in different ways, as has been documented in patients living with chronic renal failure,30 and, similarly, doctors may differ in the way they understand a treatment, such as for asthma.31 Phenomenography has also been used to explore health care professionals’ different understandings of diabetes care,32 as well as to elucidate the differences in how patients and health care professionals understand the implications of having a stroke.33 A recognition of the different ways in which a phenomenon is understood by different groups or within a group (of clinicians, patients, administrators, etc.) can have an important impact on patient communication and education, interprofessional care and collaboration, health care, health maintenance, clinical practice, theory and health care education.15,34 Medical teachers will also benefit from developing a better understanding of how their students conceptualise phenomena that they teach about and how students understand medical practice.35 Phenomenographic research findings illuminate differences in ways of understanding, and how these different ways of understanding are related to one another. Knowing about these differences and their relationships with one another can help teachers to better support development by building on students’ pre-existing understanding. This is likely to facilitate and improve communication between teachers and students and provide us with tools with which to improve education. Similarly, in health care, a doctor will benefit from knowing about the different ways in which his or her patients conceptualise their illnesses or their treatments.

The phenomenographic method

The intent of phenomenography is not only to explore the different ways in which a certain phenomenon is understood by a particular group, but also to examine how these ways of understanding are structurally related to one another. Often, but not always, the categories represent different breadths or complexities of awareness and are thereby hierarchically inclusive in their relationship. (This means that category B includes category A, and category C includes both A and B, but not vice versa.) For example, many studies have shown that the way teachers understand what it means to be a teacher in higher education range from a teacher-centred to a student-centred perspective, where the student-centred perspective is conceived of as being more inclusive and representing a broader understanding of teaching than the teacher-centred perspective.23,36–38 This perception of structural relationships among the categories is one of the epistemological assumptions of phenomenography.10 The sum of all categories (i.e. the range across which a phenomenon is understood) is referred to as the ‘outcome space’.39 In other words, people understand the world in different ways, but these differences can be described and form the outcome space in a phenomenographic study. Identifying the internal and structural relationships among the categories in the outcome space is an important additional part of the analysis in phenomenography, which is often not included in other qualitative methods of analysis such as content or thematic analysis.

Phenomenography builds on a non-dualistic ontology, which means that ways of experiencing a phenomenon represent a relationship between the phenomenon and that which is being experienced. As Marton and Booth have explained:

‘There is not a real world “out there” and a subjective world “in here”. The world (as experienced) is not constructed by the learner, nor is it imposed upon her; it is constituted as an internal relation between them.’10

Marton9 further clarified that it is not possible to describe a world that is independent of our descriptions or of ourselves as describers. This is what leads to the assumption that different ways of understanding are related through the phenomenon being experienced.


Interviewing is the most commonly used method for studying beliefs and conceptions today and is the primary method for phenomenographic data collection. However, written text, such as answers to free-text questions in questionnaires, can also be used (such as in Dall’Alba’s work on how medical students understand medical practice40). As the aim in phenomenography is to capture the range of possible ways in which a certain phenomenon is understood within the group, the sampling of participants is mostly theoretical to maximise variation. For example, if we are to explore ways of understanding what it means to be a good clinical supervisor or preceptor, we will want the respondents to vary in age, gender, medical specialty, teacher training and teaching experience. It is, however, not easy to predict the real differences according to surface impressions or formal characteristics. For example, it has been shown that the respondents’ own experiences as students can have significant impact on the way they understand teaching,41,42 but this would be difficult to use as a selection criteria. The number of respondents in a phenomenographic study usually ranges between 10 and 30. The sample needs to be large enough to ensure sufficient variation; however, too much data will make it impossible to manage the data because a very thorough method of analysis is used.43

The questions in a phenomenographic interview are semi-structured, which means that they are open-ended to allow the respondent the freedom to choose the dimensions or aspects of the phenomenon on which he or she wishes to focus and elaborate.9 This also creates space for more unexpected answers and thus helps the researcher to better understand the bigger picture. The assumption is that the researcher cannot know how the question was perceived until the respondent answers it; accordingly, this method allows follow-up questions or reformulated questions to be asked. The follow-up questions may be as important as those that are predetermined for uncovering underlying meanings. Probing questions, such as those in the form of ‘Can you explain that further?’ or ‘Can you give an example?’ are often used. Abrandt used three explicit strategies for probing: repetition, requests for clarification or elaboration, and confirmation.44 It is important that the respondent feels that he or she is able to think aloud, or to hesitate before answering, and that it is made perfectly clear that the focus of the interview is not to find a certain ‘correct’ answer and that there are no right or wrong answers. Lines of discussion are followed until the respondent and researcher agree that they have reached a state of mutual understanding and the discussion is exhausted.45 This means that the researcher immediately needs to interpret what the respondent is saying in order to decide whether further questioning or probing is necessary.46 A phenomenographic interview guide often moves from action to experience and from concrete to abstract.47 This means that the interviewer may start by asking about a recent encounter with a patient or a classroom event and then ask more general questions about patient communication or teaching activities. Stenfors-Hayes et al.25 used the following questions when interviewing medical teachers about their roles as mentors: (i) ‘What do you do as a mentor?’ (ii) ‘What signifies your mentor–mentee meetings?’ (iii) ‘What does it mean to you to be a mentor?’ Larsson and Holmstrom15 used the following three questions in their interviews: (i) ‘When do you feel that you have been successful in your work?’ (ii) ‘What is difficult or what hinders you in your work?’ (iii) ‘What is the core of your professional anaesthesia work?’

In a phenomenographic interview, the researcher should have a genuine interest in what the respondent has to say and the interview style should be relaxed and friendly. The interview may be seen as a conversation devoted to a certain task.34 The 12 aspects of qualitative interviewing as described by Kvale and Brinkmann48 can also be applied in a phenomenographic interview. They describe how the interviewer seeks to interpret the meaning of the central themes in the life world of the subject, how the interview may lead to new insights and awareness, how it seeks qualitative knowledge expressed in normal language and how knowledge is produced through the interpersonal interaction in the interview. Ashworth and Lucas49 provide guidelines on how a phenomenographic research interview should be conducted (Fig. 1).

Figure 1.

 Interview guidelines as described by Ashworth and Lucas49

The researcher’s knowledge, experience and relationship with the respondent will also affect the interview situation and it has been suggested that experiences and understandings are actually jointly constituted by researcher and respondent.50 However, in phenomenographic research the researcher strives to hold back their own prejudices and theories, to fully focus on how a phenomenon is understood by the respondents rather than on how this understanding is similar (or not) to their own understanding, or how it differs from previous research findings.51 It is also important for the researcher to focus on describing the way the phenomenon under study is understood rather than trying to explain why.51 Ashworth and Lucas49 suggest that an imaginative and empathic attitude helps in capturing the respondent’s way of understanding the phenomenon under study. Using a phenomenographic approach, it is important to reflect upon how the study would have been different if the interviews had been performed by different (or several different) interviewers.


Depending on the phenomenon under study, some degree of content expertise may be necessary for the researchers. However, as a content expert in anatomy, for example, it may be more difficult to see beyond the expert view when analysing transcripts. Team members who are well versed in qualitative research should therefore ideally perform an analysis in parallel with content experts so that the findings can be compared and discussed.

In a phenomenographic analysis, all the data are viewed as one set rather than as separate sets by transcript. Each transcript is therefore interpreted within two contexts: the context of the individual script, and the context of all the transcripts.39,52 The analysis often follows a procedure such as that as described in Table 1.53 In reality, however, there is constant iteration between the steps.

Table 1.   Seven steps of analysis 53
1 FamiliarisationReading through the interview transcripts to get a fresh impression of how the interview proceeded
In this initial phase, all data in the entire pool are given equal consideration
2 CondensationIdentifying meaning units in the dialogue and marking or saving these for the purpose of further scrutiny
The size of the meaning units identified in this step varies: some researchers claim that these chunks can be fairly small, whereas others emphasise the importance of keeping the whole transcript more or less together58
3 ComparisonComparing the units with regard to similarities and differences
4 GroupingAllocating answers expressing similar ways of understanding the phenomenon to the same category
5 ArticulatingCapturing the essential meaning of a certain category
6 LabellingExpressing the core meaning of the category
Steps 3–6 are repeated in an iterative procedure to make sure that the similarities within and differences between categories are discerned and formulated in a distinct way
7 ContrastingComparing the categories through a contrastive procedure whereby the categories are described in terms of their individual meanings as well as in terms of what they do not comprise

Using a phenomenographic approach, we assume that the categories are constructed by the researcher(s) in relation to the data54 and are not ‘discovered’ within the data. To avoid the findings representing a recreation of the researcher’s preconceived ideas, the researcher should try to find counter examples in the data and discuss the analysis with peers. Some researchers find it helpful to start the analysis with a smaller pool of data, such as a third of all the transcripts. These are later reconsidered in light of the full set. It may be useful to have several researchers read, analyse, compare and discuss the findings in this first smaller subset before proceeding. 55

In the analysis, the whole transcripts are read many times because each new query formulated by the researcher and each new perspective should be explored with reference to the entire pool of data. This means that the researcher will have a new question or idea in focus in each reading:

‘The matter of focus is important. As you read the transcript there must be, to my mind, a focus. In all of the analyses we have done, I have read all the transcripts many times – at least six and sometimes a dozen times. On each occasion, some new perspective is being sought in order to clarify what the student [respondent] means. On each occasion, the reading of the transcript is a new experience.’56

The iterative approach to analysis described in Table 1 is similar to a qualitative analysis as performed using a grounded theory approach (constant comparison).17 However, as previously mentioned, the next step of identifying relationships among the categories is unique to the phenomenographic approach.

Identifying the relationships between the categories

The identification of relationships among the categories in the outcome space can be approached at different stages of the analysis: looking at structure too soon in the analysis may cause certain aspects of meaning to be missed out, but there is also a risk in leaving any consideration of how the categories relate to one another too late because they are supposed to be a co-construct.39 It is especially important not to let assumptions about the structure among categories affect the interview in such a way that certain aspects of a phenomenon are avoided or disregarded as they may not fit into the preliminary structure of categories. Walsh provides an example of such a dilemma in her own research exploring ways in which teaching is understood.54 She found it difficult to include affective aspects of teaching in the structure and these important issues were therefore almost ignored, which resulted in a neater set of categories. These neater categories, however, did not reflect the life world of the respondents equally well.54

What the findings represents

It is important to remember that as the transcripts are pooled together, the final categories do not necessarily represent certain respondents. The descriptions are related to the group or the ‘pool of meanings’, rather than to the individual respondents.10 Marton4 describes the categories as ‘frozen forms of thought’ in that even if respondents develop and change their conceptions of the phenomenon, the categories are likely to remain, although additional research may identify further categories. The categories should be as faithful as possible to the individual’s conceptions, but are not equal to them as the conceptions are dynamic and represent a relationship between the individual and his or her context.51,52,57 Each person’s experience is unique, but a descriptive collection of such unique conceptions is considered less useful when it comes to guiding educational change, for example, than the related categories and their critical variation that a phenomenographic analysis provides.58 This means that the descriptions in a phenomenographic study are less rich or ‘thick’ when it comes to describing a certain person’s way of understanding than in a phenomenological study.

Trustworthiness of a phenomenographic study

A qualitative study requires harmony and alignment between the research question and ontological and epistemological assumptions (methodology), the nature of the phenomenon under study, ways of gathering data and the method of analysis.7 The numbers of grids or guidelines available with which to critically appraise the quality of a qualitative study are increasing59–66 and are also applicable to phenomenographic research. A review of common criteria used in social sciences and health sciences research (especially primary care and medicine) was recently published.67

In phenomenography, we cannot ask how well the findings correspond to the existence of the phenomenon in ‘reality’ because phenomenography rests on the assumption that there is no description of reality independent of the describer. Instead we assess how well the categories (the outcome space) correspond to the ways in which people understand the phenomenon. Credibility in a phenomenographic study is therefore mainly about the relationships among the categories and the data and is thus strengthened by quotations.46,68 Peer debriefing and presentations can be used to test preliminary findings and deal with questions that may arise.69 To increase the trustworthiness of the research, a phenomenographic project can be conducted by a team of researchers (investigator triangulation).61 The procedure for conducting a phenomenographic analysis as a team of researchers in order to increase trustworthiness is described by Wahlström et al.70 One of the investigators is usually assigned the main responsibility for exploring a certain domain, and the suggested categorisations are then discussed with the other investigators in the team, frequently in several rounds, until agreement is reached about the final categorisations. Wahlström et al.70 called this process ‘negotiated consensus’. This procedure replaces an inter-judge reliability test.

Credibility is also enhanced by the researchers’ self-awareness of their roles as researchers and how their interpretations influence the research process (reflexivity and ‘the researcher’s interpretative awareness’).71–73 A detailed description of all steps of the research process (including changes in the interview protocol and details on decisions made in the analysis) increases credibility. As with most research approaches and methods, there are variations in how phenomenographic research is carried out and described.39 Acknowledgement of these differences in procedure and arguments for their appropriateness should be provided. The final categories identified in a phenomenographic study are a form of construction, and another research team studying the same phenomena may not necessarily find the same categories. However, the categories identified should be recognisable by others once they have been identified.9 Three criteria for reviewing the quality of a phenomenographic outcome space are commonly used: (i) each category reveals something distinctive about a way of understanding the phenomenon; (ii) the categories are logically related, and (iii) the critical variation seen in the data is represented by as few categories as possible.10

Furthermore, the categories should be defensible, useful and meaningful to their intended audience.39,74 The findings of a phenomenographic study often help the reader view a certain aspect of the world in a new way.


People can interpret the same events and situations in many different ways (even if we often feel that our own way is the only reasonable one). Phenomenography provides a way to investigate these differences to facilitate improved understanding and learning. Phenomenography can be a way of integrating or synthesising complex conceptual alternatives and stimulating critical thinking about the conceptions and assumptions that drive our thinking, feeling and action.75 It can also help to bring to the surface and define our different perspectives so that these can be compared, contrasted and scrutinised. Phenomenography makes it possible to view variation in experience holistically, through the structural relationships among the categories.58 During the 1990s, published critique of the phenomenographic approach led to the further development of the theoretical foundations of the approach. As this debate focused on mainly philosophical issues, it will not be summarised here; instead we provide references for further reading (see: 28,29,51,76,77).

In qualitative research in medical education, the tools of data collection and analysis are often emphasised. However, we should also consider the larger picture within which the tools are situated, including the epistemological and ontological perspectives, as these shape the types of questions we ask and the ways in which we interpret the answers. The research paradigm in which the study is steeped dictates what is done and how the outcomes are interpreted.6 Phenomenography provides a lens through which to explore certain types of research question and also gives support for the empirical conduct of the research. As a research approach with epistemological and ontological assumptions that emphasise change and complexity, the phenomenographic approach is well suited to inform medical education research.6 (Readers eager to learn more about the theory behind the approach should see: 10,11,13,78.) When a developmental perspective is applied in phenomenography, the rationale for the research will include how the outcome space (the categories) can be used to enable and support change, or inform and influence practice.79 In a clinical context, a more comprehensive way of ‘understanding work’ will provide doctors with a broader repertoire of action in clinical situations.15 In an educational setting, the purpose is usually to facilitate learning by using the outcomes to improve courses and programmes or train teachers.52 Marton and Pang11 have shown how knowledge of the variation in how students understand a certain topic or phenomena can be effectively used in designing education to improve learning. Like Prosser57 and others, we believe that many current problems in, for example, teaching and learning can be approached by considering the individual perceptions of those involved in the teaching and learning process.23,42,80–82 In medical education, phenomenography would represent a viable approach to explore differences in, for example, ways of understanding feedback, the role of information technology or simulation-based education in learning, professional development and professionalism. We suggest that a phenomenographic approach can be used to explore many medical education research issues, and that this can contribute significantly to improving the quality of qualitative medical education research and establishing a more solid link between research and educational development and change.

Contributors:  all three authors contributed to the study conception and design. TS-H wrote the first draft of the article. All authors contributed to the subsequent revision of the article and approved the final manuscript for publication.


Acknowledgements:  This text was written in fond memory of Lars Owe Dahlgren, one of the founders of phenomenography.

Funding:  none.

Conflicts of interest:  none.

Ethical approval:  not applicable.