Meta-ethnography

ABSTRACT

Qualitative review procedures have developed significantly over the past 30 years; meta-ethnography being one. Meta-ethnography was originally located within an interpretivist world view and aimed at reviewing small numbers of qualitative studies, often involving small sample sizes. One of the defining features of meta-ethnography is that of theory generating. Whilst developed within healthcare literature, where particular procedures and processes are evident it is suited to the discipline of sports and exercise psychology. A notable gap in the healthcare literature is how meta-ethnographies honour ethnographies honour the theory generation. In order to apply the meta-ethnographic procedures, researchers need to appreciate the origins of the techniques, be aware of the impact of philosophical worldviews and critically appreciate the 7 prescribed stages. This paper seeks to; (a) provide a background to the area and overview of past meta-ethnographies within sport and exercise psychology, (b) identify the strengths and weaknesses of the approach, (c) provide a justification of assuming a social constructivist paradigmatic view to meta-ethnography, (d) provide an analysis of the 7 stages with clarity for a social constructivist and (c) provide directions for future application of meta-ethnography in sports and exercise psychology.

KEYWORDS:

• Meta-ethnography
• review
• critical appraisal
• qualitative
• social constructivism

Part 1: A critical review of meta-ethnographic review

Qualitative reviews of evidence

Qualitative review syntheses focus on the analyses of human experience which includes social and cultural phenomena (Munn et al., 2018). Currently there are 13 types (e.g. meta-interpretation, meta-study, thematic synthesis) of qualitative review (Sutton et al., 2019). The different types of qualitative reviews exist to compare, critique, develop, integrate and go beyond simply presenting the synthesised findings from individual qualitative studies. A meta-ethnographic review focuses on interpretations made by the meta-ethnographer (Noblit, 2020) and characteristically has the ability to generate hypotheses, achieve higher-level theorising, reconcile consensus and contradictions within individual qualitative studies and uniquely derive more formalised new knowledge (Zimmer, 2006). Meta-ethnography was designed and is uniquely suited to generating conceptual models and theory (France et al., 2016; Noblit & Hare, 1988). This is considered the key distinguishing feature of a meta-ethnography review (Grant & Booth, 2009; Samnani et al., 2017) and the main focus within this review article.

What is meta-ethnography?

Meta-ethnography is identified as an interpretive and inductive synthesis methodology (Atkins et al., 2008; France et al., 2019). A central component of this type of review is on the interpretivist synthesis and on the development of theory. The purpose for any meta-ethnographer/s is therefore to establish derived understanding, representations and commentary across studies, in order to further meaning and create innovation in understanding. These processes ultimately allow for the development of ideas or constructs that are representative of the ‘whole’ dataset but may not be observed in any one of the included studies (Dixon-Woods et al., 2006). Thus, meta-ethnography was designed to reveal findings beyond the original meaning identified within any one study (Campbell et al., 2011; Cunningham et al., 2019) and enables a higher level of analysis (Atkins et al., 2008). In essence it involves the exploration of findings (generated through the process of analysis; see part 2) obtained from primary qualitative studies in order to create new theories, models or processes (France et al., 2019). The most recent guidance for meta-ethnography has not placed any restriction on the methodologies used of included individual qualitative studies (France et al., 2019; Sattar et al., 2021).

How is it performed?

A meta-ethnography review involves some key principles and overarching processes. This includes identifying primary qualitative studies with the aim of undertaking a synthesis by; (1) reading each study in depth, (2) considering how studies’ findings are related to, or contradict each other, (3) identify how the studies ‘translate’ (a term used to describe the identification of common findings across different studies) into one another and (4) provide an interpretation of the translations (France et al., 2019). The synthesis processes of meta-ethnography focus on social explanation developed from comparative understanding (Noblit & Hare, 1988). Synthesis in meta-ethnography uses the process of translation; how concepts from different studies may be representative of each other. It could be considered as the process of testing transferability or inferential generalisation across studies. This is important because inferential generalisation in itself is a type of generalisation that is different from statistical-probabilistic generalisability that may be desirable and evident in traditional systematic reviews (Smith, 2018).

The logic of translation is derived by Noblit and Hare (1988) from what Turner (1980) refers to as the ‘same practice hypothesis’. This hypothesis suggests that what one group of participants (e.g. one sporting population) would do in a particular situation could be representative of another group (e.g. another sports population). This would be expressed in the following way for the analytical processes of meta-ethnography: the findings identified and interpreted by a meta-ethnographer in one study contain similar information, interpretations, shared realities or expressions as those identified in another study. Figure 1 provides an example of how different study findings from Soundy et al. (2012) could be translated (see supplementary file 3 for full ideas for translations across studies).

Figure 1. An example of how translations could be created (adapted from Soundy et al., 2012).

Translation of the findings are identified by considering; (a) verbatim quotes of participants (first order themes) found in the studies or within associated datasets (e.g. data repositories, supplementary files or raw data) (b) all other findings which represent the interpretation of the authors of the included studies (second order themes). The comparison of the coding across different studies allows for the evaluation of similarities and differences between study findings (termed reciprocal and refutational analysis, respectively). This process is iterative and more than one set of translations may be required to be representative of multiple studies (Campbell et al., 2011).

The findings of a meta-ethnography review are presented as either a thematic synthesis (or a meta-study) of qualitative findings (Finfgeld, 2003) or within the creation of new discipline-specific understanding as a theory, process or model. The approach should illustrate how multiple studies point to a process, model or theory that is not found in any one study (Campbell et al., 2011). The mapping and representation of the synthesis can be considered as; (a) a framework combining concepts identified by individual studies within a larger process or (b) a meta-thematic analysis represented by the meta-ethnographers’ own interpretations (Brookfield et al., 2019). Figure 1 provides an extract from Soundy et al. (2012) illustrating what themes could be represented around the concept of social support.

The creation of new meaning and understanding is referred to as a line of argument synthesis (Noblit & Hare, 1988) which is, the ability to reveal what might be hidden within individual studies and enables discovery of ‘a whole’ from a combination of the parts. This shares similar goals to formal grounded theory (Finfgeld, 2003) and this stage of the process has the potential to allow for analytical concept generalisation and theoretical generalisation, where the models and processes developed are identified as generalisable (Smith, 2018). The pragmatics of this process would include a questioning-idea testing process to create new understanding.

The questioning-idea generating process would create the line of argument synthesis by the researcher asking questions of the findings, most often the third order interpretations. This includes interrogating associations, processes and influencing factors in order to create and test new knowledge. Emergent ideas and the creation of new knowledge would be critically considered and tested in order to develop a deeper understanding of the processes and factors, which influence and shape the phenomena of interest by deeper examination of the included studies. Critically and for transparency researchers must document the outcomes to all questions, ideas, reasoning generated as findings to the original posed questions are identified. The documentation of this process is crucial to allow the reader an ability to follow the development of ideas by demonstrating how support was gather from past evidence mostly from the literature reviewed but also from using existing theory and knowledge (if the paradigmatic stance permits). The meta-ethnography needs to identify where researchers own reasoning or ideas work and do not work in order to refine understanding. Moreover, the most defendable findings must be shown with clear logic and representation. Defendable findings must demonstrate the stages of idea development, and how the idea has been arrived at. This could be visual illustrations of how findings in the current review fit a particular idea or association between identified themes or sub-themes. Alternatively, it could be how findings demonstrate or are associated with past theories, models or processes. Defendable findings will use different techniques to demonstrate this, for instance open coding may identify an idea that is observable across multiple studies which are included in the meta-ethnography. The meta-ethnographer may then choose a technique to examine this, like a spider diagram or a matrix. A spider diagram may be used to view the idea, identify visually alternatives for the idea, or fit the idea to related theory or other concepts. A matrix may be used to provide illustrations of associations between study components, for instance forms of received social support and psychological outcomes.

Understanding how to ensure quality in theory generation is critical and this understanding may benefit from being situated within a paradigmatic stance. Social constructivism being just one paradigm and being well suited for the theory generating aspects of meta-ethnography with compatibility to highly analytical processes used, such as analytical memos (Hernández-Hernández & Sancho-Gil, 2017) or constant comparison analysis (Toye et al., 2014). Considering this, principles identified by Charmaz (2017) may be used to support the questioning and testing of ideas stage of the analysis (See Figure 2).

Figure 2. A process of question-idea generation to create a line of argument synthesis.

Why do it?

There are many justifications for using this type of review. First, the reduction in duplication of research and consolidation of knowledge. Qualitative reviews are highly suited to highlighting areas where data saturation across studies has been achieved. Secondly, a meta-ethnology review considers and identifies areas where studies do not agree, whereas traditionally a focus on differences has been limited (Finfgeld, 2003). Finally, this approach may highlight areas of focus for future research and the derivation of new questions (Atkins et al., 2008). Furthermore, meta-ethnography can help understand the experiences people or in this instance sportspersons have of interventions (acceptability, feasibility or acceptability) and understand how they work. This in turn will increase knowledge about sport or exercise or physical activity practices, which can be a stimulus for change or creation in policies (Booth et al., 2013).

Whilst we are not aware of any completed meta-ethnographic reviews within sports and exercise sciences before 2006 (Weed, 2008), more recently evidence has emerged of meta-ethnographies developing models and processes that directly and significantly impacts and challenges current understanding in the field of sport and exercise psychology. These include a model of physical activity and post traumatic growth (Chen et al., 2020), a model considering minority ethnic groups and factors which influence physical activity at an organisation, community and individual level (Koshoedo et al., 2015), a model identifying the determinants of physical activity using a social ecological model of physical activity for individuals in a low socioeconomic context (Rawal et al., 2020) and models of behaviour change and psychosocial determinants of physical activity for people with severe mental illness (Quirk et al., 2020; Soundy et al., 2012, 2014).

Strengths and challenges of meta-ethnography

Many strengths of meta-ethnography have been identified, including the ability to; (a) expose experiential data that brings together more information than a singular study (Grant & Booth, 2009; Samnani et al., 2017), allowing synthesis of common findings across studies and highlighting gaps in understanding, (b) identify factors which influence the experience or perceptions of individuals to a particular phenomenon (Grant & Booth, 2009; Samnani et al., 2017), (c) enhance interactions between clinicians/practitioners and participants by understanding the participant’s perspectives (Munn et al., 2018) and to inform decisions to support different services/practices when the effectiveness of each is equivocal (Grant & Booth, 2009), (d) reveal why interventions may not be effective or be adopted despite evidence of effectiveness (Munn et al., 2018), and finally (e) generate theory/ies and novel ideas to advance the field. Meta-ethnography, vitally has the potential to broaden the generalisability of qualitative findings (Whittmore, 2005), a recognised limitation when drawing on findings from studies with small samples or from discrete and ‘unique’ populations. It is, however, important that the type of generalisation is considered, alongside the ontological and epistemological views of the meta-ethnographer. For instance, the most likely types of generalisation from a meta-ethnography would include transferability of findings, as well as analytical concept generalisation, rather than statistic-probabilistic generalisation (Smith, 2018). One of the most exciting outcomes is the generation of theory and new findings; adding creativity and impetus to advancing research and understanding in sport and exercise psychology. However, it is important that researchers consider what techniques and processes are included to achieve this and how paradigmatic considerations can play a role in this. We acknowledge this review cannot detail all considerations for varying philosophical viewpoints, rather we have chosen to illustrate the uniqueness of the social constructionist point of view. The paradigmatic position could influence the search, appraisal of literature and synthesis of literature. For instance, some meta-ethnographers may opt for a systematic search strategy with an extensive identification of articles, whilst others may opt for systematic search processes that looks to obtain articles that can generate theory and achieve theoretical saturation and concept development (this could be similar to the selection of participants for a grounded theory study, requiring two types of sampling approach; namely purposive and theoretical), meaning the sampling and selection of articles can differs within the review processes (Grant & Booth, 2009). However, more guidance is required to identify how theoretical saturation is determined within qualitative synthesis (Dixon-Woods et al., 2004; France et al., 2019). Similarly, the goal of critical appraisal and use of an a-priori theory could vary. One meta-ethnographer could seek to identify strength of confidence in the resultant evidence, focusing purely on evidence generated by included studies, whilst another meta-ethnographer who believes reality is socially constructed may have less concern about specific methodological criteria in each individual study, but look to use methods of constant comparison, memo writing and any past theory and evidence to support the development of their own new theory (Lewin et al., 2018).

Several challenges face a meta-ethnographer, including past diverse application of the methodology and an absence of reviews that detail the analysis process, meaning that the replication of processes is extremely limited (Cahill et al., 2018; France et al., 2014). It is possible that during the process of analysis the findings become interpreted so much that they lose their grounding or association with the included studies (France et al., 2019). There is a danger too that the process of synthesis just becomes a synthesis of common findings re-coded or categorised (Finfgeld-Connett, 2014) and that there is a lack of theoretically informed synthesis (Thorne, 2017). Also, it may be possible that the studies included provide limited depth or are regarded as data which is ‘descriptively thin’ (Sattar et al., 2021), which may be a result of how the studies were selected (Britten et al., 2002). These problems may mean the resultant meta-ethnography reflects a thematic synthesis of evidence rather than a conceptual development of evidence, resulting in meta-ethnographies that do not reach the intended level of analytical depth (Thorne, 2015). It is possible through the process of searching that the meta-ethnographer misses articles that challenges thinking, this is illustrated by the number of meta-ethnographies that simply focus on common findings (Britten et al., 2002) or that lack consideration of chronology or temporality (Thorne, 2017).

Future topical advances: the philosophical paradigmatic positions and their impact on decisions for processes within meta-ethnography

The first future topical advance is related to the importance of the meta-ethnographer’s world view, is highlighted in a more recent contribution to literature on meta-ethnography (Cunningham et al., 2019). For the purposes of analytical concept generalisation, philosophical positions are needed that help create robust theories or models. One such position, having a very clear direction for the use of review procedures is social constructivism. As noted above, the search processes, appraisal and synthesis with social constructivism could be very explicit, as principles from social constructivist grounded theory could be applied within the seven steps of meta-ethnography. In it of note that the most recent reporting guidelines (eMERGe) (France et al., 2019) doesn’t detail how specific philosophical positions could be integrated across the 7 stages of a meta-ethnography review. We would suggest this a valuable recommendation for future studies and publication of review findings.

Noblit and Hare (1988) ‘firmly’ based meta-ethnography in the interpretivist paradigm (p. 11). The interpretivist paradigm is situated as having an ontology of relativism. Relativism constructs reality by the meaning and understanding developed through experience and social encounter, which are unique. This means the researchers are not able to separate themselves from what they study. Advances in meta-ethnography point towards philosophical positions that contain some form of realism. For instance, subtle realism (e.g. Mays & Pope, 2000) identifies that an external reality exists and is represented through common understanding or relatable experience. This establishes a distinct way to view quality in qualitative research and could in turn impact what critical appraisal tools are used within a review process and why. However, social constructivists believe that reality is constructed through human activity and that knowledge is socially and culturally constructed (Amineh & Asl, 2015). It is made distinct from interpretivism because it illustrates that concepts, however, they are socially constructed relate to something real in the world. This is also different from an external independent reality because it is created and expressed by human mental activity (Schwandt, 1998). Social constructivism is considered compatible with critical theory and pragmatism (Charmaz, 2017). Social constructivism identifies that all knowledge wherever it has come from is useful for generating theory. Further to this, social constructivism places importance on developing, testing and refining theory to ensure quality. These particular qualities provide implications for the different stages of meta-ethnography as detailed in the below sections. Given the above, the advances in meta-ethnography should consider how different worldviews impact on a meta-ethnography review process. As identified previously we cannot consider all implications of different paradigms on the steps of meta-ethnography. Rather our paper and included review focuses on the implications of assuming a social constructivist world view to further support the application of methods in the field of sport and exercise psychology.

Future topical advances: the concept of translation

Campbell et al. (2011) suggests there is no single way to undertake the process of translation and that the translation should be representative rather than literal. Noblit and Hare (1988) identify the importance of maintaining central ‘metaphors’ in relationship to other ‘metaphors’ that have been generated. They define metaphors as ‘themes, concepts, perspective, organising codes, or concepts’ (p. 14). There is a need to group and compare the metaphors to show the comparison between them. Researchers illustrate the process of translation through focusing on comparison of concepts and metaphors (Campbell et al., 2011). The process, however, should also be challenged by looking at the roots of the concept, this includes past knowledge including theory, models and previously proposed mechanisms. The idea of translation, as identified earlier, is derived from Turner’s (1980) research. The same practice hypothesis refers to the identification of an aspect, idea or phenomena that occurs in one situation and looks for evidence of it occurring in another situation. For the meta-ethnographer this would be an idea, concept or phenomena occurring in one paper and seeing if it occurs in another. The process is considered as a trial and error procedure and should be reflected in the reporting of the analyses for transparency. Turner (1980) stated that identifying instances where the hypothesis does not work is important, so that the hypothesis can be modified. It is also essential that the researcher evidences in detail and understands why it did not work. Turner (1980) suggests that this should be the bulk of a task when ‘offering a comparative explanation’ (p. 56). There is a danger that similarities are given a greater attention or focus as part of the analysis (Cunningham et al., 2019). Researchers are required to look for specific material, conduct or expressions that help explain differences across studies, and represents an essential part of the understanding created. Meta-ethnographers’ need to document the process of hypothesis testing, identifying when it doesn’t work and understand why those instances are so. For the social constructivist meta-ethnographer, this process must involve a process of questioning the data and creating ideas. Emphasis should be placed on the following concepts in order to achieve the aims of this philosophical position; the identification and justification of the idea, challenging the accuracy of the idea, identifying an alternative explanation and instances where the idea doesn’t work and, finally deciding if the idea can be used, needs to be modified or rejected. An example of what this may look like can be seen in Table 1. Publication of a meta-ethnographic review should include an audit trail of the processes for transparency and to illustrate logical progression.

Table 1. An example of how ideas are developed and challenged for a social constructivist meta-ethnographer.

Goal	Questions	Example documentation of information and techniques
Identification of the idea with justification or a rationale	What could be going on here? What connections between different study findings are there? What is identified directly by studies? What is identified by implication from the studies? What is identified by using theory to connect data or explain the study results?	A table with multiple columns. The first can illustrate the results of studies. The second can represent the meta-ethnographers open codes. The final column can represent any related ideas, comments and connections. Written answers and notes to questions posed with evidence to show the ideas and their associations. Memo writing and the use of a research diary. Exploration of ideas by; spider diagrams, or mind maps of association or a matrix table that could bring phenomena and studies together.
Challenging the accuracy of the idea	How is it possible to gain support for this idea? Do specific variables seem to be associated with others? What evidence are you using to illustrate this? Are there phenomena that seems to help explain outcomes? How much support is there across studies and how much detail is given within each? What or where the gaps in understanding? Could any past knowledge provide a key or link to aspects, processes or mechanisms identified in the results of included studies? Is there a natural explanation that can be derived from past theory?	Spider diagrams or tables that document the details given by each included studies for direct and indirect support of the idea. Challenge and support from related literature and theoretical literature in the form of a matrix with reporting detail of that support Identification of gaps in understanding using a list to detail the gap and lack of evidence for it. Illustration using a spider map or table to show how a past theory, model or process relates to, or ‘maps’ on to the idea. Consider if: the association is full or partial. the idea varies from the past theory/model/process.
Identifying an alternative explanation and instances where it doesn’t work	How does the idea relate to past theories, models, mechanism or processes? Does the phenomena that is focused on related to other phenomena? Could other connections be made between related phenomena? Can the idea be expressed in an alternative way? Did the data provide cases where the idea of connections or association did not work?	List of associated models, processes or theories and their components and consideration as to whether they explain the process or modify the idea? Documentation of alternative explanations using ideas and identified evidence from studies – could be created using a spider diagram or table. Identification of all negative cases and how they relates to the idea or part of the idea? Justification is needed as to why this is not problematic.
Deciding if the idea can be used or if an alternative is needed?	Could a case for theoretical sufficiency be made? Is any modification of the idea needed? Is a new idea needed?	Identify a defendable case for theoretical sufficiency being achieved. Provide a clear tabular summary that illustrates the supporting evidence.

Part 2. Top guidelines. How to conduct a social constructivist meta-ethnographic review: the seven-stage process

A 7-stage meta-ethnography is advocated to bring together qualitative data, although the original intention was that the stages could overlap or occur in parallel (Noblit & Hare, 1988). Acknowledgement and retention of these stages is evidenced across literature. We have searched and identified examples of recent meta-ethnographic research used within sport and exercise psychology. (see supplementary file 1 for search methods and supplementary file 2 for the output). Twelve reviews (Brown & Irby, 2009; Carter et al., 2018; Chen et al., 2020; Garip & Yardley, 2011; Koshoedo et al., 2015; Morgan et al., 2019; Pollard et al., 2020; Porcuna & Rodriguez-Martin, 2020; Quirk et al., 2020; Rawal et al., 2020; Soundy et al., 2012, 2014) fulfilled the eligibility criteria. These reviews are included to help define and illustrate how each stage could be completed. It should be noted, however, that these reviews, as is commonly the case, did not evidence audit trails of their analysis. There is therefore an urgent need for future meta-ethnographies to include this level of detail (France et al., 2015).

Stage 1: getting started

Meta-ethnographers are required to identify an area of interest with explicit justification for the need for interpretative review (France et al., 2014), and an evidence-based rationale included (Toye et al., 2014). This may include updating a previously published meta-ethnographic review for further information (see France et al., 2016). Of the included sport and exercise psychology reviews, rationales centre of a lack of past understanding around relationships between phenomena or understanding of how or why experiences, behaviours or expressions occur (8/12 studies). The research aim should be associated with questions which consider how or why experiences or perceptions in regard to the particular phenomena (Noblit & Hare, 1988). Examples from our reviews include; (a) understanding the psychosocial factors which change as a result of participation in physical activity for people with severe mental illness (Soundy et al., 2012), (b) to understand the psychological and social factors which influence weight management and create a more comprehensive understanding of weight management (Garip & Yardley, 2011) and (c) to develop and interpret the association between post traumatic growth and physical activity (Chen et al., 2020). A systematic approach and sensitive search strategy is recommended initially to identify articles with qualitative information (see France et al., 2019). For the social constructivist, the focus should be on the ability of the findings to inform and direct further analyses of the data (Charmaz, 2017). After the initial search process, the meta-ethnographer may find their analysis focusing on a phenomena, model or process that requires further searching to become iterative (Dixon-Woods et al., 2006). Thus, searching may not be limited to the initial systematic search and can continue, if required, alongside the analysis in order to obtain theoretical saturation. As part of this process the research question can be further refined in terms of specific focus as the inductive processes evolves (Campbell et al., 2011).

Stage 2: Deciding what is relevant

Four specific and critical considerations are evident during this stage, including searching for literature, the sampling techniques, the sample size, and the critical appraisal of studies:

Searching for literature: Cunningham et al. (2019) state that searching can be exhaustive and systematic, or non-linear or purposeful where the intention is to generate theory. The original intention was to undertake exhaustive searching of literature only if there was a particular reason (Noblit & Hare, 1988). For instance, if access to information was limited by the phenomena being considered. During this stage the meta-ethnographer needs to be attentive to how to locate qualitative literature (Booth, 2016) and other forms of knowledge that may be important, for instance there are recognised search strategies specifically for locating theory (Booth, 2019). A general distinction, made by Sutton et al. (2019) is that for aggregative reviews, more traditional systematic processes are utilised, using extensive databases and various systematic techniques (Harari et al., 2020; McGowan et al., 2016). Recommendations for database searching, citation searching, hand searches and consultation with experts have all been identified as important in meta-ethnography (Cahill et al., 2018). Additionally, for theory generating qualitative reviews (such as social constructivist meta-ethnographic approaches) particular approaches detailed below can be followed (Finfgeld-Connett & Johnson, 2013).

Despite the importance of paradigmatic views on the different stages within meta-ethnography, only 1/12 reviews referenced a paradigmatic position. Soundy et al. (2012) refer to the outcome of their meta-ethnography being ‘a truth of truths’ rather than ‘the truth’ being established. However, further implications or considerations for this are not given. Within a social constructivist paradigm, searching should begin with selective or purposive sampling of studies, followed by the development of thematic ideas (hypothesis generation) and for the testing of hypothesis theoretical sampling could be used (Sutton et al., 2019). The researcher will come to the study with an awareness of the area and justification for the focus, but should remain open to what might be discovered (Weed, 2009). The process of searching may be aided by a specialist librarian (Cahill et al., 2018), with this being evident in just one study (Quirk et al., 2020) in the current review.

The sampling techniques:

Most literature supports the use of purposive sampling (Campbell et al., 2011; Doyle, 2003; Grant & Booth, 2009), which may involve maximum variation sampling (Campbell et al., 2011). Dixon-Woods et al. (2006) distinguish sampling based on review type. They identify a greater need to capture all data where the nature of the review is aggregative. Our current review identified that 8/12 studies used systematic search processes and other processes that would be similar to quantitative reviews including; a study protocol, a framework/checklist to guide content, PRISMA flow diagram to report searching etc. None of the meta-ethnography reviews within our current review of literature drew on strategies that may be reflective of a social constructivist approach. Although research convention recognises that within qualitative evidence syntheses, the search methods should be one that best suits the purpose, for theory generating reviews this could be an expansive non-linear approach (Uny et al., 2017). Dixon-Woods et al. (2006) however, name ‘theoretical saturation’ as an example which would fit within a social constructivist meta-ethnography review and which we detail in part 2. Other considerations would include the importance of sampling studies due to their richness and diversity (Booth, 2019). For the purpose of the social constructivist meta-ethnography, the sampling of papers should change within Stage 3 of a line of argument synthesis (see Figure 2). This would be considered theoretical sampling and represent the final stage of analysis. Sampling is needed that can provide justification, detail, context and contrast to proposed association and connections. Particular attention should be given to evidence that goes above and beyond the current understanding of the model, process or idea. This may be generated from; (a) studies that met the inclusion criteria which provided the greatest insight or challenge to the idea of interest generated, (b) studies that didn’t fulfil a particular aspect of the inclusion criteria (e.g. different design type, different population group) but could provide a significant insight to the area of interest, (c) studies that have provided a review of the constructs and ideas in question (e.g. review articles that consider resilience would be appropriate if the idea related to that construct) and finally (d) other forms of evidence including books or conference presentations that explain processes, models or theories in greater detail (e.g. ideas around the construct of hope was given great insight from the philosopher Marcel, 1951). When selecting papers, the meta-ethnographer should provide a defendable case for choices made around a focus of the sampling. This represents the process of obtaining articles which may be deemed most significant within area of focus, or even outside the field in order to test or refine the analysis and synthesis (Greenhalgh et al., 2005). The evidence obtained in this way would map on to the questioning process that occurs within the idea development stage (see Table 1). The above process will provide the meta-ethnographer with confidence in their ability to theorise; something described by Dey (1999) as ‘theoretical sufficiency’. A criterion is needed which describes the development of theoretical ideas which are rich or thick (Fusch & Ness, 2015). To achieve this, the meta-ethnographer should provide a defendable case that illustrates that the categories or idea are well developed in terms of their dimensions, variations or properties (Corbin & Strauss, 2008) and that represent well defined, checked and explained relationships around the categories or ideas (Charmaz, 2014). These are the central components of theoretical saturation; however, we have not used the term theoretical saturation to avoid the connotations which suggest an unquestionable completeness of the idea (Nelson, 2017).

The sample size: Historically the suggested number studies required for a meta-ethnography was between 2 and 6 (Noblit & Hare, 1988), although a maximum of 40 studies has more recently been suggested (Campbell et al., 2011). France et al. (2014) report that reviews have used between 3 and 77 research articles. Within our current review there were on average 17 ± 8 studies, ranging from 8 (Soundy et al., 2012) to 39 (Morgan et al., 2019).

Toye et al. (2014) identify the concept of theoretical saturation as important to ensure the robust development of conceptual categories. Theoretical saturation would require the researcher to continue to sample for studies until the development of the theory, process or model has been achieved. Sampling stops when the addition of new material provides no new theory building material. One of the main requirements of idea testing is to identify instances where the idea is rejected. Therefore, the sample size should be adequate, thus that the researcher is satisfied that the hypothesis is robust, instances where the hypothesis does not work have been explored and, the hypothesis is defendable to others and evidenced within an audit trail (see Table 1 as to how this may be evidenced). This is supported by research that identifies the importance of achieving conceptual depth, clarity or richness of the research (Campbell et al., 2011; France et al., 2014; Morgan et al., 2019). Researchers must ensure that enough studies are identified to achieve a saturated understanding of the reasons why and when the hypothesis in question works.

The critical appraisal of studies: The purpose of appraisal of qualitative studies is most often to either exclude research, grade research to refine the focus or meaning of findings (Dixon-Woods et al., 2006). The purpose of critical appraisal within traditional systematic reviews has been to; (a) judge quality (risk of bias) and then to (b) use this judgement to determine the impact of the quality on the findings (and recommendations) generated by the review. Within our review, only Quirk et al. (2020) provides an example of this by using of CerQual (Lewin et al., 2018). CerQual being used to report confidence in evidence of each theme. The majority of studies (11/12) within sport and exercise psychology used critical appraisal tools (most often CASP appraisal tool 6/12) to include or exclude studies. The common use of CASP is consistent with past observations (France et al., 2014), however, our review identified that using CASP could contribute to poor inter meta-ethnographer agreement (Carter et al., 2018); a well-documented problem when using numerical appraisal checklists (Dixon-Woods et al., 2007). One study in the current review identified that quality assessed by CASP did not affect the contribution of a paper to the synthesis (Garip & Yardley, 2011). Although another review was able to document confidence in the evidence of the meta-ethnography review themes from the critical appraisal process (Quirk et al., 2020). For the purpose of analytical concept generalisation (Smith, 2018), the reporting of certain aspects may limit the impact of the resultant synthesis and findings. It should be noted that poor reporting of methods does not equate to a poor study, it may simply reflect the need to discriminate content (notably the methods) in manuscript development to meet the target journal word count (Toye et al., 2014). With the move to online journal access and ability to use supplementary files, this should no longer be a limiting factor to transparency and precision in reporting. It should also be considered that techniques or methods designed to assess quality may not be compatible with a particular paradigmatic viewpoint (Sparkes & Smith, 2009). For instance, Smith and McGannon (2018) problematise particular quality techniques (e.g. member checking or inter-rater reliability).

It is essential that the meta-ethnographer understands what impact of study exclusion would have, for instance it is possible that excluding research will decrease the confidence in concepts that require data saturation. The social constructivist undertaking grounded theory creates rigour by creating confidence in the ideas they have obtained from a source of data and through constant comparison of such (Charmaz, 2017). Additionally, the importance of a good level of description of conceptual categories within included studies is a requirement to allow translation to take place (Noblit & Hare, 1988; Toye et al., 2013). The process of analysing and obtaining new data from studies to test, check and modify hypotheses would reflect the need to establish theoretical coverage (Weed, 2009). In this regard, the social constructivist meta-ethnographer would likely ask different questions for the inclusion of articles. For the purpose of a social constructivist the following questions are suggested: (a) are the research questions focused on the same key elements or phenomena that the meta-ethnography is? (b) are the research findings relevant and useable to the focus of the synthesis? (c) do the findings provide context for the culture, environment and setting, and is this critical given the question posed? (d) are considerations made sufficiently well detailed (clarity of insight of concept identified as conceptual clarity by Toye et al., 2013) so that the concepts can be translated? Once a description of the synthesis and greater focus of findings has been established, the meta-ethnographer can then prioritise identifying studies that provide a distinct contribution to the analysis, or studies that challenge, further and refute the research findings, with the goal of establishing theoretical saturation. The social constructivist meta-ethnographer may be required to apply the critical appraisal questions to any new articles as required by this process. Justification and decision making for this stage should be provided within an audit trail. As suggested earlier some stages of meta-ethnography in practice can overlap (Uny et al., 2017).

Stage 3: reading studies

Research consensus suggests that papers should be read and re-read (e.g. Atkins et al., 2008; Britten et al., 2002), this is consistent with findings from our review, with all studies (12/12) using multiple researchers. Two studies (Carter et al., 2018; Porcuna & Rodriguez-Martin, 2020) used qualitative software (e.g. NVIVO) to assist this process, this being consistent with past reviews (Toye et al., 2014). Quality enhancement could be further made at this stage by using more than one meta-ethnographer (Cunningham et al., 2019) and/or enhanced by using electronic packages like NVIVO (Toye et al., 2014).

Following this, data extraction should be undertaken to document demographic details and main concepts (Cunningham et al., 2019). This could be achieved by identifying main concepts (Atkins et al., 2008; Campbell et al., 2011) and inputting this into tables (Cunningham et al., 2019). During this process first order (participant verbatim quotes), second order (study author reported findings) and third order constructs (meta-ethnographer interpretation) can be clearly identified (Atkins et al., 2008). The premise being that the initial stage seeks to describe thematically the phenomena of interest before conceptualising it (Weed, 2009). All studies within the current review present a thematic analysis of the meta-ethnographer’s interpretations. In sport and exercise psychology this should be considered a baseline of output at this stage.

Stage 4: Deciding how studies are related

Noblit and Hare (1988) identified distinctive processes that occur at this stage including: (a) noting of interpretative themes, concepts, perspective, organising codes, or concepts, (b) making a list of metaphors, phrases, ideas, concepts, and (c) undertaking a juxtaposition of these. These processes are supported across more recent literature (Atkins et al., 2008; Britten et al., 2002; Cahill et al., 2018; Toye et al., 2014). They should also be considered within the context and purpose of the reading of papers which could be to compare, organise, relate, map, stimulate or verify (Lee et al., 2015). Therefore, techniques used at this stage should not be limited by current examples. Within our review several techniques were identified including: (a) line-by-line coding where the meta-ethnographer codes all findings from included studies (first order and second order concepts). (b) Juxtaposition and bringing together findings by determining their relationships between studies. This is done in a way that preserves meaning (Cunningham et al., 2019). Soundy et al. (2013) provide a tabulation of literal and idiomatic representations of concepts extracted from studies. Figure 3 provides an example of this with a focus on a phenomena common to all studies.

Figure 3. An example of the literal and idiomatic interpretations around the concept of hope (adapted from Soundy et al., 2013).

The generation of new concepts is required to explain how findings are inter-connected rather than describing the data (Cahill et al., 2018). For the purpose of the social constructivist meta-ethnography the meta-ethnographer needs to document the development of their own ideas. (See supplementary file 5 for an example of a mind map). This can be undertaken by using a diary of questions which are answered by the meta-ethnographer. An example of this could include: What concepts are being discussed and expressed by the authors of included studies and how do they relate to other concepts? Is there a particular phenomenon that is prevalent across studies and what is the current understanding of it? Do the current findings challenge or contradict each other? Are there any particular factors, which may explain particular findings?

Stage 5: translating studies

Translation is the generation of a common rubric that represents the identification of common categories of meaning identified across different studies (Noblit & Hare, 1988). This common rubric can represent either the study authors’ original language or the meta-ethnographer’s interpretation. Two types are considered: (1). Reciprocal translation being the identification of instances when accounts are directly comparable, (2) Refutational translation being the identification of accounts are oppositional. The meta-ethnographer may also need to translate the differences between studies that are identified (Campbell et al., 2011). The supplementary material provided by Soundy et al. (2014) provides an illustration of the development of themes and support from first and second order concepts. Apart from the examples given above, studies within our review used a variety of techniques to achieve the translation of concepts, these include (a) creating a table, grid, or conceptual map of concepts as a way to consider concepts and, for cross referencing, [an example of this can be found in Soundy et al. (2013)] (b) constant comparative analysis, (c) double coding with blinded meta-ethnographers and (d) tabulation of findings. Supplementary file 4 provides an example of tabulation to generate translations from Soundy et al. (2012). Clarity in reporting of an audit trail is needed in order to show how the final synthesis is derived, as this something that is not clear across meta-ethnographies (Lee et al., 2015).

In order to develop the process of hypothesis testing for the social constructivist meta-ethnography it is important that Turner‘s (1980) derived concept of translation is considered. Given Turner’s definition, it is important that the meta-ethnographer identifies techniques which demonstrate the following: (a) identification of associated hypotheses that are generated from the association of concepts, processes of factors. Idea identification should be documented and included in the audit trail. (b) the development of the idea should be documented alongside evidence showing examples of supporting and negating evidence; explanation should be provided for the reader. (c) If required the identification of an alternative/related idea is needed, then part a and part b should be repeated. And (e) finally the meta-ethnographer will have to document the process and resultant hypotheses.

Stage 6: synthesising translations

This stage broadly describes making ‘a whole’ from the parts of the synthesis (Noblit & Hare, 1988). Much of the literature states that the focus at this stage can be directed towards the implications from the refutational and reciprocal translations and considering how they impact the third order constructs (Atkins et al., 2008; Britten et al., 2002; Campbell et al., 2011; Toye et al., 2014). Part of this process involves comparing translations to see if some can represent others or if some can be reduced where there is overlap (Noblit & Hare, 1988). The supplementary information from Chen et al. (2020) provides an example of how first, second and third order themes relate. The process of hypothesis testing different ideas should report the process, model or unique set of ideas that represent findings from individual studies. The social constructivist meta-ethnographer may need to examine the different ideas created and provide evidence of how and when they relate together (if they do). Abduction would be required at this stage, where abductive reasoning looks to pursue observations that don’t make sense within current theory or understanding and, attempt to identify all possible theoretical explanations by going back and forward between theory and data (Charmaz, 2017). This process would be appropriate at the stage of testing and idea development. Table 1 provides an example for the social constructivist meta-ethnographer. Being able to document this stage of the process in such a way will illustrate the strength of the evidence and provide transparency behind the development of the idea, theory, process or model. These are important requirements for a meta-ethnography (Cunningham et al., 2019).

Stage 7: the expression of the synthesis

The final stage should be undertaken and make several considerations about how the synthesis is expressed, including the need to make it accessible to a selected audience (Atkins et al., 2008; Cunningham et al., 2019). For instance, Britten et al. (2002) identify the importance of considering if the meta-ethnography is appropriate for practitioners, policy makers, or clinicians. Whereas, Campbell et al. (2011) consider more the relevance linked to the cultural context. Further it is important to consider other approaches to knowledge translation other than written form e.g. as a film (Noblit & Hare, 1988; Toye et al., 2014). Given the nature of a qualitative synthesis and the volume of findings often identified, it is important to consider the utility of the findings in terms of target audiences and how any process, model or idea is expressed to enhance the uptake of critical findings and subsequent implementation of changes in practice and policy.

Recommendations for meta-ethnography

• An audit trail of the analytical phases of the meta-ethnography should be made available so good practice can be established and the field can have an abundance of examples to draw on.

• The paradigmatic position taken for the meta-ethnography should carry through all 7-stages of the approach. This may be particularly important when considering how and why particular techniques are used for synthesis and if, how and why critical appraisal is used.

Recommendations for a social constructivist meta-ethnography

• Searching for studies to include should continue until findings are identified that have developed the line of argument synthesis to a point where theoretical saturation can be identified.

• An audit trail of the analysis must be used to show the development of the synthesis and line of argument synthesis or resultant model, process or theory

• The process of question-idea generation should be used and challenged in a constant comparative way that looks for cases to support and refute the idea.

• The process of critical appraisal should be modified to reflect the social constructivist world view with a focus on the outcome of generating a line of argument synthesis.

Summary

This paper presents the current literature that considers meta-ethnography. An overview of current literature in sport and exercise psychology has been presented and problematised. This review provides a step-by step guide that will help a meta-ethnographer to assume the philosophical position of a social constructivist in sport and exercise psychology. Adoption of these ideas and techniques has the ability to considerably advance our knowledge and understanding in sport and exercise psychology, recognising the wealth of research that has been undertaken in discrete athletic populations which could be further synthesised to derive new knowledge and understanding to inform practice.

Funding details

No funding was obtained for this work.

Acknowledgements

We would like to acknowledge the insightful and critical comments and requests from all reviewers. Conceptualization; AS, Data curation; AS, Formal analysis; AS, Investigation; AS, Methodology; AS, Project administration; AS, NR, Software; None. Supervision; NH, Validation; NH, Visualization; AS, Roles/Writing – original draft; AS, Writing – review & editing; AS, NH.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability

Additional tables can be viewed upon request.