Introduction

Meta-ethnography is an approach to qualitative research synthesis developed in the late 1980s. While there are several other approaches (cf. Thorne, Jensen, Kearney, Noblit, & Sandelowski, 2004), meta-ethnography has become arguably the most popular approach. Most meta-ethnographies are in health sciences, even though the approach originated in education. Nevertheless, there have been developments in education in the early 21st century that are noteworthy.

This article first reviews the original proposal for meta-ethnography, published as a SAGE monograph in 1988. While there have been several developments of note since then, the basic approach remains the standard. Second, an example meta-ethnography is summarized so the reader may have a better understanding of what a meta-ethnography looks like in practice. Third, the history of meta-ethnography in the health sciences is briefly noted and argued to have been fueled by concerns over evidence-based practice. While evidence-based practice has largely fueled the popularity of meta-ethnography, there are reasons to be concerned about this development and what it means for this approach to qualitative research synthesis. Fourth, the use of meta-ethnography in education is reviewed, with special emphasis on what is argued to be a furious phase of development after 2015. Fifth, there are several elements of the original proposal that have been less fully implemented and the reasons and significance of these are discussed. The final section outlines future directions for meta-ethnography.

An Example

An example may help orient the reader to this method. Martin and Locklear (2018) conducted a meta-ethnography of Native American youth identity. Their phenomenon of interest, Native American youth identity, was driven by their positionality as Native Americans, and as a result, they sought out studies that countered the all too common colonial narratives and deficit perspectives often employed in such studies. To identify studies, they searched many databases using a set of keywords such as Native American, youth, qualitative, and identity, which yielded over 8,000 articles. Narrowing their search to only refereed articles reduced the number to over 5,500 articles. Specifying cultural identity and identity theory led to a reduction to just over 100 studies. In reviewing these, they then focused on high school students and tribes in the United States. In reading these studies, they noted that the vast majority of them employed deficit perspectives and did not use culturally appropriate research methods. When deficit studies were excluded, only five studies remained—a significant finding in and of itself—all based in the Southwestern region of the United States. These studies were read repeatedly and coded and recoded for themes and storylines that characterized each study. In their contribution, they summarized each article’s storyline. They created first a reciprocal translation of the studies that involved language, culture, and adult–youth relationships, which they discussed in some detail. Yet because the studies all addressed something distinct from the others, Martin and Locklear pressed on and created a line of argument synthesis that included: language—“the missing piece” (p. 219); indigenous knowledge systems—“fighting White man’s education” (p. 220); and social identity—“the scars of racism, discrimination and colonialism” (p. 222). The line of argument synthesis presented a fuller interpretation and critique than any of the studies did individually and established a new theory of Native American youth identity from a non-deficit perspective.

This study is a good example of meta-ethnography but it should be noted that there are many ways to (a) identify studies to be included and excluded; (b) review the studies; (c) conceptualize the phenomenon of interest; (d) approach the synthesis process; and so on. Meta-ethnography is and will continue to be a developing and changing research synthesis practice.

The Original Proposal for Meta-ethnography

Dwight Hare and I wrote the original monograph for SAGE (Noblit & Hare, 1988). We developed the ideas for it over the course of three years, presenting papers that addressed aspects of the ideas. The book was published in the SAGE Qualitative Research Methods series. Our interest was broadly based in developing an interpretive approach to synthesis. An interpretive approach was juxtaposed to the aggregative approach, which characterized the more quantitative synthesis efforts, such as meta-analysis. The aggregative approach, we argued, is context stripping as data are pulled from studies from different contexts to make a large database. The interpretive approach, we proposed, is to consider whole studies in relation to other whole studies and to focus not on data, but on the interpretations made by the authors of the studies and how these various interpretations were related to one another. The result, then, is another interpretation—one that encompasses the interpretations of the primary studies and how the set of interpretations are related—or not. In this, the intention was to build knowledge even though we saw applications to practice and policy. As discussed in this article, using meta-ethnography for practice and policy raises a set of questions that, as of 2019, I do not think are adequately addressed. It is one thing to develop an interpretation close to a given set of studies, but it is quite another to propose that one should do something in other contexts based on that interpretation. Finally, the original formulation limited the approach to ethnographies per se, but over the years scholars have found this an unnecessary limitation and have applied meta-ethnography successfully to a range of qualitative approaches.

Meta-ethnography has seven phases, although these are to be seen more as guidance than requirement (Noblit & Hare, 1988). Meta-ethnographers have modified this list to fit their specific studies (cf. Beach, Bagley, Eriksson, & Player-Koro, 2014). The seven phases are as follows:

The actual phases employed, though, should be adapted to the requirements of the particular meta-ethnography. If this is the case, the rationale for the phases actually employed should be made explicit so others can both understand the process and potentially learn how to make adaptations for their own work. For example, Beach et al. (2014), in tailoring the phases to their project, employed four steps instead of seven.

The Development of Meta-ethnography

Meta-ethnography was proposed in 1988 but was slow to develop. Eventually the volume of qualitative studies being published led to interest in talking about what a collection of qualitative research had to say. Uny, France, and Noblit (2017) document the development of meta-ethnography in both health care and education. Meta-ethnography in health care was spurred by concerns about evidence-based practice and policymaking, especially in the United Kingdom. Beginning in the 1970s, health care researchers developed what came to be called systematic reviews. Initially these were focused on quantitative, controlled trial studies about various aspects of medical treatment. In 1993, the Cochrane organization was established to house the volume of systematic reviews being done. In the 1990s as well, there was increasing interest in qualitative studies and what they may add to the evidence base in healthcare, and qualitative synthesis methods were being considered. By 2004, the Cochrane Qualitative Research Methods Group was established. As Uny et al. (2017) discuss, the net result has been a “steady growth” (Uny et al., 2017, p. 247) in the publications of meta-ethnographies in health research. Indeed, France et al. (2015) found that 118 meta-ethnographies were published in 2014 and 2015 alone. For meta-ethnography in health care research, there were several notable developments. First, the volume of studies to be synthesized was much larger than Noblit and Hare’s examples in their book. This was due to both the sheer number of studies available and to the capabilities afforded by electronic search and database software that developed since 1988. Second, although this was not solely the product of healthcare meta-ethnographies, the inclusion of a wide range of types of qualitative studies (beyond ethnographies) was demonstrated to be possible without adverse effects. Third, it became evident that systematic reviews were but one base (and possibly overly positivistic) for meta-ethnographies. Iterative search, for example, was also fully reasonable as a search strategy, depending on one’s intent for the meta-ethnography (cf. Finfgeld-Connet & Johnson, 2013). Iterative search differs from systematic reviews in that the former uses keywords to identify all possible studies as a first step. In iterative search, one starts with one (or a small set of studies). Then reading these leads to a preliminary specification of the phenomenon of interest, which then drives the search for more studies. The iterative process continues until the authors conclude they have the final specification of the phenomenon of interest and the set of studies relevant to that specification. Fourth, with the increased volume of studies, a concern with study quality emerged and led to quality appraisal processes for inclusion and exclusion of studies from the synthesis. Quality appraisal remains controversial, in part because of the wide range of genre in qualitative research. Finally, it became evident that meta-ethnographies varied in terms of how close they adhered to Noblit and Hare’s method. Some of this was because the translation of studies into one another was difficult, and some of this was because standards had yet to be developed for good practice. In health-care, studies by Pound et al. (2005) and Campbell et al. (2003) demonstrated that a fuller explanation of processes used in the synthesis was essential. This in turn led to sustained efforts to improve reporting guidelines for meta-ethnographies (cf. France et al., 2014; France et al., 2015; France et al., 2019).

Meta-ethnography in education research has had “delayed growth” (Uny et al., 2017, p. 250). There are many reasons for this but probably most important has been the politicization of evidence-based practice and policy in education. Educational policy and practice are much more subject to shifts in politics and cultural beliefs than it appears healthcare is. Borgnakke (2017) shows how, in the Danish case, the evidence-based policy movement in education reduced synthesis to merely a technique to feed political agendas. Second, Davies (2000) notes that schools are complex and changing environments year to year. Specific contexts then have to be addressed when discussing applicability of synthesis findings. Further, qualitative research in education during this time period has been developing alternative epistemological and theoretical (especially critical) approaches. This has inhibited the growth in number of meta-ethnographic studies but education’s concern with qualitative methods has also led to improvements in the meta-ethnographic method itself.

Rice (2002) and Doyle (2003) were the first meta-ethnographies in education after Noblit and Hare. Both stayed rather close to the primary studies and did not offer significant new interpretations. Rice (2002) synthesized 20 qualitative case studies of collaboration in professional development schools, summarizing many themes across process, situational, structural, and relational dimensions. She concluded the relational dimension was dominant in the collaborative process. Doyle (2003) selected four school ethnographies to study educational leadership, and from these synthesized three major themes: commitment to visions; power with an essence; and congruency of actions. Moreover, Doyle articulated an important role for positionality of the person performing the synthesis. She argued that a meta-ethnographer needs to explain “one’s place in the text” (p. 331), especially in terms of creating the synthesis translation. She found that journaling this process was helpful in avoiding a tendency to aggregate instead of translate the studies. Savin-Baden and Major (2007) and Savin-Baden, McFarland, and Savin-Baden (2008) pushed Noblit and Hare’s insistence that the meta-ethnography is interpretive into a fuller claim for an interpretive meta-ethnography that offers more than the original studies did. They incorporated positionality into their syntheses, employed critique, and problematized the emphasis on similarity over difference in meta-ethnographies. They referred to these as a set of “honesties”: (a) situating ourselves in relation to our participants; (b) voicing our mistakes; (c) situating ourselves in relation to the data; and (d) taking a critical stance for research (Savin-Baden & Major, 2007, p. 837; see also Noblit, 2018 on related points).

Other meta-ethnographies that followed these studies did not always see positionality, difference, and critique as essential. For example, Tondeur et al. (2012) used 19 studies to discuss the preparing of preservice teachers in the use of technology. They reported themes as distributed across studies—emphasizing similarity over difference as a result—and created a model for preparing preservice teachers. The model was essentially descriptive. Positionality of the authors and a critical perspective were not emphasized. Jamal et al. (2013) started with over 60,000 potential qualitative studies and, through review, reduced the number to 19 studies that addressed how school context affects students’ health behaviors and how students’ backgrounds relate to the context and behaviors. They discovered four “meta-themes” (“performance, collective identity, and bonding; the social importance of space; staff–student relationships; and the influence on health and ways of ‘escaping’ the school environment”) (Jamal et al., 2013, p. 5). Their line of argument synthesis was seen to have extended existing theory. Again, positionality of the authors and a critical perspective were not emphasized. They also did not problematize an emphasis on similarity over difference.

However, Beach et al. (2014) did address positionality, critique, and difference—and in the process added an explicit emphasis on theory. Beach et al. (2014) used meta-ethnography to better understand changes in teacher education in Sweden. The authors had been conducting studies relevant to this topic for many years (which were part of the 34 selected studies) and had developed a critical perspective about these changes about which they were using the meta-ethnography to address. They focused on differences as well between student teachers’ reflections on their preparation and practicing teachers’ needs for professional knowledge as compared with policy intentions. They found, among other things, that it was the institutional and policy actors who had constructed a discourse about the failure of progressivism to enable an “ideological project of change in teacher education that actually lacks scientific support” (p. 165). Beach et al.’s meta-ethnography (2014) addressed education policy, revealing it to be essentially ideological. Baker and Harter (2015) employed meta-ethnography of six studies to address a published high-profile critique of differentiated instruction. In this, their perspective and positionality were also evident. The result was a systematic rejection of the critique of differentiated instruction.

In health education, Bearman and Dawson (2013) explicitly built on Savin-Baden and Major’s study (2007) for their synthesis but also made an argument for three criteria for rigor in meta-ethnographies: specification of researcher positionality, transparency of how the synthesis was conducted, and triangulation of perspectives. These could be accomplished, for example, by documenting the negotiation of multiple perspectives represented on the research team as part of the synthesis process. Park et al. (2015) conducted a meta-ethnography of undergraduate medical education in the United Kingdom, which led to two themes: (a) the importance of interpersonal interactions for learning, and (b) the significance of sociocultural spaces for learning. They did not build on the methodological developments discussed here but did demonstrate that medical education, like other forms of education, is highly contextual.

In 2017, the journal Ethnography and Education published a special issue on meta-ethnography, edited by Kakos and Fritzsche. The reason for this effort, they argued, was that meta-ethnography “promises more than its current application in the context of educational research suggests” (Kakos & Fritzsche, 2017a, p. 130). They noted the method had “possibilities . . . for a deeper and more creative interpretation of existing results” and “in its educational value for those engaged in the process” (Kakos & Fritzsche, 2017a, p. 130). Inasmuch as meta-ethnography drives toward new interpretations that characterize a field of studies and provides a new contextualization for each individual study, then such possibilities are likely. The special issue had several types of contributions. Eisenhart (2017) described three relatively recent methodological approaches that “extend contextualization and generalizability” (Eisenhart, 2017, p. 134) of ethnographies: multisited ethnography, meta-ethnography, and comparative case study. She argued the approaches are complementary and characterizes meta-ethnography as “a means for clarifying and extending ethnographic generalisations across time and space, accumulating knowledge and serving social and political purposes” (Eisenhart, 2017, p. 141). Borgnakke (2017), as noted in this section, analyzed the case of the evidence movement in education in Denmark. She found that after the year 2000, “the evidence-producing process (from empirical findings to practice-oriented statements) are formed in terms of politicized platitudes rather than research-based proposals for action” (Borgnakke, 2017, p. 194). Moreover, the evidence movement has reduced meta-ethnography from “critical meta-reflections” to “a technique” (p. 198). Bornakke compared the arguments used for qualitative research synthesis before and after the year 2000, which in Denmark is the pivotal year for the evidence movement. Before 2000, qualitative syntheses led to themes which were nuanced and “elaborated for dilemmas and characteristic contradictions” (p. 198). After 2000, qualitative syntheses were relegated to offering insights into explaining how and why quantitatively established effects worked. Meta-ethnography and other forms of qualitative research synthesis were no longer seen as producing new knowledge. Rather, qualitative syntheses were to serve the knowledge created by quantitative research. The article by Uny et al. (2017), as discussed, described the different development patterns for meta-ethnography in health care and education. Taking Uny et al. (2017) alongside Borgnakke, it is evident that evidence-based practice and policy can serve as a driver for the popularity of meta-ethnography but may also reduce qualitative research and qualitative synthesis status as a form of knowledge production.

This special issue also included five meta-ethnographies on a range of educational topics. Kakos and Fritzsch (2017b) decided to focus on two primary school ethnographies, one that each of the authors had conducted, that shared the same focus on teacher–student interactions. The meta-ethnography revealed, among other things, that inflexible interactional performances were preferred by students because flexible arrangements increased a sense of vulnerability in the classroom and that, in general, teachers desired to maintain classroom control. The German schools were less formal than the English schools and thus caring for students in German schools was more personalized, whereas caring was more located in the professional role of teachers in the English schools. In regard to meta-ethnography, they argued that the method is less linear and more complex than implied by Noblit and Hare (1988). They found that the process led to them adjusting their original interpretations. It also resulted in a discovery of relationships between the texts and the construction of a “common space” (Kakos & Fritzsch, 2017b, p. 239) as essential to translating the texts into one another. Moreover, they noted that the authors had to establish a new relationship as the process unfolded through phases of “reflection, defence and emotional catharsis” (p. 239), which, in turn, led to an expanded understanding of both their field of study and their identities as ethnographers. They noted that as authors of the original studies they had a sense of ownership and a “tendency to protect and (to some extent) overemphasize the uniqueness and distinctiveness” (p. 239) of their respective studies. The meta-ethnography was an opportunity to renegotiate their relationships with their studies as they created a new space among their studies.

Huf and Raggi (2017) also worked with two studies they had conducted in different countries, Germany and Austria, with a shared intellectual interest that moved “towards a more theoretical generalization” (Huf & Raggi, 2017, p. 166) of “how children help each other and how this is mediated in the social order of the age-mixed classroom” (p. 171). Their effort was to have meta-ethnography be a synthesis of data rather than interpretations, as Noblit and Hare proposed, and to involve the primary researchers as well. This “shared grounded synthesis” (p. 169) yielded “‘help’ as a sensitizing concept” (p. 170). Further, they were able to develop a set of new perspectives of “doing help” (p.173), including:

In creating this synthesis, they point out the value of having the primary researchers involved in the synthesis but do not wish to state that this is a requirement. Rather, it is a debate worth engaging in.

A multisited meta-ethnography that also involved studies conducted by the authors of the synthesis addressed “youth learning in and outside of schools” (Hernandez-Hernandez & Sancho-Gill, 2017, p. 178). There were a total of 10 studies to be synthesized—five by students and five by the authors. The meta-ethnography enabled a conversation between the two sets of reports. Using second-generation grounded theory, they focused both on similarities and differences, conceptualized as potentially representing a “qualitative distinction” (p. 182). Hernandez-Hernandez and Sancho-Gill (2017, p. 183) conceptualized this as a “diversity-oriented comparative approach.” The meta-ethnography was guided by a hypothesis that there is a gap between school learning and learning outside school using new literacies, and found (and elaborated on) through the synthesis a theme of “connecting in and outside learning.” Further, they argued, informed by post-structural ethnographers, that meta-ethnography required them to “work toward a language of possibility rather than a language of certitude” (p. 190).

Ellis (2009) had termed meta-autoethnography for an author’s efforts to consider her or his own multiple autoethnographies but had not delved into synthesis of a set of autoethnographies by different authors. Staying close to the original method developed by Noblit and Hare (1988), Hughes and Noblit provided a “worked example” (Hughes & Noblit, 2017a, p. 213; Britten et al., 2002) of a meta-ethnography of autoethnographies, using a set of four autoethnographies of teaching difference in higher education, including one autoethnography by Hughes. A “reciprocal analogy” (Hughes & Noblit, 2017a, p. 218) resulted that involved the following: (a) difference is stigmatized; (b) . . .“[W]hite students are bodies of resistance” (p. 218); and (c) the academy is itself another body of resistance to teaching difference (Hughes & Noblit, 2017a, p. 218). They argued that through the meta-ethnography of autoethnography “we come to say something about the meanings of others’ lives—and our own” (p. 224), and “better access to subjectivities, when so much of education recognizes teachers, students and parents as objects…” (p. 225; italics in the original).

Policies of personalization of education in Sweden were the focus of Beach (2017), whose synthesis of studies (including some studies of his own) across levels of schooling (secondary and higher education) led to a theme of ideology, social reproduction, hegemony, class relations, stereotypes, false consciousness, and choice (“low-end choosers” and “skilled choosers” [p. 157])—and a line of argument synthesis. This led to the conclusion that personalization “actually reproduces various types of inequity” (Beach, 2017, p. 158). When education is a commodity, “personalization makes a virtue out of selfishness by rewarding it and dressing it up as something else” (p. 161). Beach theorizes that social reproduction is not automatic, allowing for the possibility that the dominant order can be challenged potentially via youth culture and through currently existing programs: “Other more just outcomes can still be struggled for and won” (p. 161).

This Special Issue signaled that meta-ethnography had finally found purchase in education. Further, meta-ethnography in education continues to develop differently from that in health studies. In health studies, the evidence-based practice logic frames meta-ethnography as largely post-positivist and providing a factual basis for practice. In education, positionality, difference, and critique frame meta-ethnography as more critical and constructivist. In education, meta-ethnography is less about facts and more about interpretation, more about critique of policy, theory, and conceptualizations and less about justifying particular policies or practices. The Special Issue continues these themes but also adds some themes as well. These include considering meta-ethnography as a way to create networks of scholars and provide them with a project that invites their collaboration and reconceptualization of their works. In these networks, people bring their primary studies to the collaboration and transform individually designed studies into multisited studies. The process of conducting the meta-ethnography yields not only new findings but also new understandings of interpretation based in the interactions of researchers, overcoming defensiveness about one’s prior interpretations, and involving levels of interpretation—moving from first-order to third-order interpretations that constituted the synthesis and translation. The researchers in this Special Issue developed new networks, new understandings, and new approaches to interpretation.

Also in 2017, the American Educational Research Association published an edited volume by Anderson-Levitt and Rockwell (2017). This volume was the culmination of an effort to compare educational ethnographies across borders and featured major Latin American and U.S. scholars collaborating to explore studies across the United States and Latin America’s many borders. This volume has one chapter that employs a meta-ethnographic approach but meta-ethnography informed the wider comparative agenda of the volume and the commentary chapter also invoked it. Rockwell and Levitt-Anderson (2107) recount that the volume was organized around the idea of comparison of both education in different contexts and how ethnographers do their work in these different contexts. They discuss meta-ethnography as an “inspiration” (Rockwell & Levitt-Anderson, 2017, p. 10) for the comparative project, citing Noblit and Hare’s argument about saliency of context and the necessity of an interpretive rather than an aggregative approach. They also argue that the volume expands the notion of translation, with a special interest in line of argument synthesis. The chapter (Levitt-Anderson & Bueno, 2017) that employed meta-ethnography focused on ethnographies of teachers’ work in Latin America and the United States in a set of three comparisons, each of a pair of studies, eschewing an exhaustive list of studies in favor of preserving a strong focus on context. The three comparisons concerned knowledge for teaching, teachers’ work and the state, and becoming a teacher. These yielded, respectively, a reciprocal synthesis, a refutational synthesis, and a line of argument synthesis. Working across the three sets of comparisons, a “larger line of argument” (p. 141) about globalization and national context was also possible but differences still remained. Globalization of education reform was found to be powerful but the economic poverty of students and teachers in Latin America meant that national context remained salient. Also, U.S. ethnographers seem less likely to name state power possibly due to “theoretical assumptions more than reality on the ground” (p. 142). Sanchez and Noblit (2017) provide a commentary on the nature of ethnography and qualitative synthesis approaches and, in doing so, argue for thinking through a more anti-essentialist approach to meta-ethnography. They call this approach “enabled meta-ethnography” that involves “radical contextualization” (p. 178) (adding context not included in the studies), “perspective taking/border dialogue” (p. 178) (seeing each study from the perspective of the studies from across the borders that are reflected in the set of primary studies), and radical heterogeneity (“a new engagement in ontology”) (p. 179) including “a commitment to critique” (p. 179). Clearly, this call is ambitious and whether others find it worth engaging remains to be seen.

Another ambitious effort can be seen in Urrieta and Noblit (2018a). This edited book had two distinct contributions to the larger meta-ethnography project. First, the book is an exploration of how meta-ethnography can inform theory. The book included eight meta-ethnographies of different forms of cultural identity with the explicit intent of informing and critiquing cultural identity theory. Price and Burton (2018) addressed studies of black racial identity theory. Bettez, Chang, and Edwards (2018) examined studies of multiracial identity. Latin@ identity in North Carolina (one of the New Gateway states for Latin@ immigration) is the focus of Ender and Rodriguez (2018), while Parkhouse and Pennell (2018) offered a meta-ethnography of Latina students, gender, and sexuality. Kolano, Childers-McKee, and King (2018) focused on racialized identities of Southeast Asian American youth, and LaGarry and Conder (2018) examined whiteness in preservice teachers. Martin and Locklear synthesized how Native American youth navigated their identities. Finally, Hatt (2018) conducted a meta-ethnography of how race, class, and gender were featured in the identity formation of smartness. Each of these meta-ethnographies were then used by Urrieta and Noblit (2018b) to determine their collective contributions to cultural identity theory. Noblit (2018) noted that these meta-ethnographies were explicitly critical meta-ethnographies. This culminates a move signaled by Beach et al. (2014) and Beach (2017), who were critical of educational policies. In Urrieta and Noblit’s book, the meta-ethnographies are critiques used to advance theory. In their concluding chapter, Urrieta and Noblit see the meta-ethnographies as arguing for a theory of cultural identity that is “(a) intersectional, (b) fluid, dynamic, and in process, (c) agentic and transgressive, (d) located in contextual specificity, and (e) situated within a backdrop of whiteness and white supremacy” (Urrieta & Noblit, 2018b, p. 250), which both confirms and extends existing identity theory.

The review given here of meta-ethnography as practiced in education shows that while Uny et al. (2017) were correct in their assessment that the development of meta-ethnography in education had “delayed growth” (Uny et al., 2017, p. 250) in comparison to that in health research, the recent efforts seem almost furious. Meta-ethnography in education has stimulated scholars to join in new teams for a form of multisited studies and in turn has invigorated their understanding of ethnography and of interpretation. It has also spurred a fuller consideration of meta-ethnography as an interpretive endeavor and as a form of critique to inform policy as well as theory. Finally, in this furious development period, scholars have pushed for meta-ethnography to become anti-essentialist. This engaged meta-ethnography pushes meta-ethnography into an engagement with ontology that to date has not been characteristic of the method.

The Ongoing Difficulties with Meta-ethnographic Synthesis

Meta-ethnography, as noted in “The Original Proposal for Meta-ethnography,” is based on two fundamental premises: (a) themes in qualitative studies are metaphors not literal descriptors; and (b) synthesis requires studies to be translated into one another.

First, language is metaphoric in nature. Words refer to something inexactly and in relation to other concepts and other words for phenomena. In qualitative coding, data analysis, and interpretation, we draw from language used by scholars, our own experience, and people in the scene to characterize the scene. In our best studies, the words approximate experience but are not the experience itself. The themes in qualitative research, then, are “as if” characterizations. (They also are not innocent—ethnographers realize that we colonize with words and theories as much as with our bodies, armies, and economies.) This means that in comparing interpretations, we need to think broadly about the specific terms each author uses. We need to look at them in terms of relative equivalents and relative commensurability. In doing so, one should consider the adequacy of the concepts or themes in the same way we assess metaphors (Noblit, 1999). That is, what is the relative cogency (efficient integration), range (ability to incorporate other symbolic domains), and apparency (revealing of multiple connotations) of the themes offered? The concepts of one study may be able to accommodate the terms of another study while the reverse may not be as true. For some syntheses, the existing terms may not be sufficiently expansive and new concepts may need to be created that can take in a range of themes from the primary studies. This type of thinking results in a third-order interpretation that speaks across all the primary studies involved. Of course, not all studies can be synthesized solely by focusing on similarities. In the case of refutational syntheses, the interpretation of one study (or one set of studies) stands in opposition to the interpretation of another (study or set of studies). In the case of line of argument syntheses, the examination of themes reveals that each study (or set of studies) is addressing different aspects of a shared phenomenon of interest. When these interpretations are laid against one another, an expanded understanding of the phenomenon of interest is revealed. Put another way, the themes, concepts, and metaphors used in one study (or one set of studies) lack sufficient range, apparency, and cogency to subsume those of other studies in either refutational or line of argument syntheses. As the chapters in Urrieta and Noblit (2018a) aptly demonstrate, in any set of studies there can be multiple synthesis forms. Lines of argument can supplement reciprocal translations and a field of study can have both refutation and reciprocal syntheses within it.

Many meta-ethnographies treat the themes of the primary studies as literal, which reduces the synthesis effort to listing the exact themes that studies share. This is a minimal contribution to knowledge synthesis and does not advance understanding in any appreciable way over simply doing a literature review. Third-order interpretations (which work across both the primary studies’ interpretations and the second-order interpretations to produce a new theory, framework, or model) in particular are enhanced when the themes in the primary studies are understood as metaphoric. Treating study themes all too literally in effect means that a translation was not accomplished—the second major difficulty with meta-ethnographic synthesis.

Meta-ethnography is an elaboration of Turner’s (1980) argument that sociological explanation is translation. In the body of published meta-ethnographies, it is apparent that all too many qualitative researchers do not understand the implications of this. Qualitative researchers, when doing primary studies, translate that which they observe into their own experience and into what others have said about related phenomena. However, this translation is often not explicit in primary studies. Ethnographers emphasize the interpretation offered rather than how it emerges from the juxtaposition of the world of the ethnographer with the world of those studied. Nonetheless, this is a fundamental premise of ethnography—the interpretation is the ethnographer’s take on the experience of the participants. It is a translation. Geertz (1989) emphasized this when he distinguished between “being there” (where one observes) and “being here” (where one writes). One works back and forth between what one observed there and the author’s perceptions when back in her or his home culture. A translation of there to here is the result. In this, the interpretation is a translation of the researcher’s worldview with those of the participants.

In meta-ethnography, the translation is moved up a level. The effort is to effect a translation among studies. Here the effort is to translate the whole of interpretations, the complete storylines, of each study. This is in contrast to those meta-ethnographies that focus mostly on the subsets of themes, which are consistent across studies and thus do not synthesize the complete storyline of the individual studies. Admittedly it is not easy to work with the complete storylines. However, this is necessary to create a translation among the studies. For meta-ethnography, then, “one study” interpretation is like that of other studies except that it may be regarded as a general guide to constructing a translation among studies. Such complete storyline translations entail a fuller understanding of context and enable robust third-order syntheses. They also can better speak to, and critique, the theories that characterize a field of study.

The analogy between the worldview of the researcher and those in the studies remains at issue, of course. This means that positionality of the meta-ethnographers must also be unpacked as to how their worldviews are engaged in the topic and in the synthesis effort. This is as essential in synthesis as it is in primary qualitative studies. Martin and Locklear (2018), as Native Americans, used their positionality to critique the wider field of Native American identity studies as employing a deficit perspective. This then became a warrant to focus on studies of positive Native American identity and develop a new theory.

The third major difficulty experienced by meta-ethnographers concerns how one should think about defining a field of study and consequently how one should search for studies that characterize a field of study. Meta-ethnography was designed before the advent of computer-based search engines and consequently imagined a search process that was library based and, by necessity, iterative. One searched for studies in the library and then read them to find other studies of interest. In the process, an understanding of the field of study developed. With the advent of computer-based searches using keywords and the burgeoning popularity of qualitative research, the number of potential studies is usually quite large at first and then is narrowed with successive passes through the database and as one moves from keywords to abstracts to full studies. The process focuses on exclusion of studies not relevant to the phenomenon of interest and inclusion of those found relevant. Clearly studies are still considered and read individually, but coming to understand the field of study is based on the work of inclusion and exclusion and less through the process of iterative reading. There is no research that says this shift is problematic but it is clearly different to learn a field through reading candidate studies early in the process than to learn it through inclusion–exclusion decisions with subsequent reading. The computer-based search process has led to systemic reviews being seen as the first step in a meta-ethnography. Indeed, this has become a usual practice in healthcare qualitative synthesis.

As many have pointed out, there are iterative and purposive search processes (Finfgeld-Connett & Johnson, 2013; France et al., 2014) as well but these are less common. Further, we have no research that says that seeking all relevant studies is more desirable than more selective or targeted searches (as in a focus on highly regarded or most cited studies or studies that were central to changes in theory or methods in a field). Hughes and Noblit (2017b) have argued that it is helpful to use search and early review processes as ways to understand and refine the understanding of the phenomenon of interest or field of study—a process of specification. In general, experimentation with search processes is needed. Urrieta and Noblit (2018b) offer an alternative process as a case in point based on the “expanded case study” method advocated by Burawoy (1998) and Gluckman (1961). One could adapt the extended case study method (Small, 2009) and select studies that iteratively examine the various contexts to a phenomenon of interest. Put simply, the extended case method (cf. Mitchell, 1983; Burawoy, 1998) works from a case to its contexts “which examines how the social situation is shaped by external forces” (Burawoy, 1998, p. 6). This method would mean that one would expand the search from direct studies of a phenomenon of interest by searching for studies that address the various contexts identified in the direct studies. In the case of education, one could study Latinx students, which may show how they are situated in classrooms, schools, communities, the economy, stratification systems, racial formations, and so on. This would then lead to searches for studies about each of these contexts. Conceivably, this could also lead to searching for studies of another set of contexts such as globalization and immigration, for example, which would then need to be synthesized as well. Following Small (2009, p. 20), this method may allow meta-ethnography to “. . .tell[s] us about society as a whole rather than about a populations of similar cases.” This approach “refines or reconstructs a theory rather than identifying an empirical fact” (p. 21), enabling a fuller role for meta-ethnography in social and educational research.

Adapting the extended case study approach for meta-ethnography could allow scholars to interrogate the studies (one or more) of a phenomenon of interest for what contexts are implicated in those studies, as is now the practice. But then, an additional process would be to seek studies of those contexts, and to do this repeatedly until all contexts possible have been exhausted. With this elaboration, meta-ethnography would synthesize all the contexts and the relations between the contexts and the phenomenon of interest. In addition, one may examine how interpretations of a phenomenon have shifted over time. This could potentially be related to shifts in policy contexts as well (cf. Beach et al., 2014) or in relation to critical events in society or a discipline. One could locate variations in interpretations of a phenomenon of interest in different social structural contexts and even nest these, from group context within organization context within societal contexts, for example. This approach will allow a synthesis from scene to society (or globe), from early manifestations of a social and cultural form to later manifestations of a social and cultural form, and possibly both contextualizations at once. This is an important direction for future research in the method of meta-ethnography.

Directions for Future Research

Several areas for future research have been identified. There is a need for methodological studies that compare more literal and more metaphoric conceptions of interpretations to ascertain how much of a difference this actually makes. Similarly, there is a need to compare syntheses that use selected themes versus syntheses of the fullest set of themes (what I have termed whole interpretation) to see what substantive differences result, and if these make a significant difference in our understanding of a phenomenon of interest. Experimenting with the extended case study approach to synthesis is also needed. To my knowledge no one has attempted this as a synthesis. Comparing search processes that use systematic review of all studies with more iterative and selective search processes is also needed to ascertain which achieves what, as compared to the other.

Aside from these methodological studies, Urrieta and Noblit (2018a) is the most expansive attempt to articulate the relation of meta-ethnographic synthesis and theory. However, it was focused on one area of cultural theory—identity theory. Other theories may not articulate with meta-ethnographic synthesis in the way cultural identity theory did, and so more explorations of theories and synthesis are needed.

Of course, the future of meta-ethnography is bound up with people asking the question, “What does qualitative research have to say about . . . ?.” If the recent furious pace of work on meta-ethnography in education is to be maintained, scholars will need to address this question repeatedly across all the different fields of study in education.