Moving Beyond Templates: A Bricolage Approach to Conducting Trustworthy Qualitative Research

Researcher Confusion

The problem of researcher confusion emerges from the most attractive feature of templates. Templates simplify in ways that make the process of doing qualitative research easier and can be an important bridge to doing qualitative research for those who do not have access to mentors who are seasoned qualitative researchers. Although templates provide an apparently easy way to dive into the field with limited formal training, doing so leaves these scholars in the precarious position of not fully understanding what they are doing or why.

The simplifying nature of templates can cause confusion for a variety of reasons. First, some users of templates fail to recognize that no matter how well a template is specified, it is inherently incomplete. No template can comprehensively describe all of the methodological steps that need to be taken in any research project. As abstract descriptions of research processes, templates cannot accommodate the particularities of specific research projects and the subtle ways projects differ. Even projects that can be described as having followed a template require a lot of work on the part of the researchers to achieve that effect. The analogy of the map and the terrain is useful here. A template is at most a map; the project is the terrain and infinitely more complex than the map.

Second, and relatedly, novice researchers may assume that when they have done everything identified in the template, they have completed the work needed to produce a study suitable for publication. For example, a central feature of the Gioia method is the three-column data structure, which is based in the relationship between open, axial, and selective codes in grounded theory (Strauss & Corbin, 1998; cf. Charmaz, 2014). A three-column data structure may result from a grounded theory analysis, but these columns are the outcome of the analysis—not the analysis itself. In other words, there is much more to grounded theory than listing and categorizing codes. To begin, it involves some specific ways of designing your study and gathering data. Additionally, it involves immersing yourself in the data, understanding and appreciating both their fine-grained properties and their larger whole, creatively combining codes, abstracting from the data, dialoguing with existing theory, and so on. Although authors may summarize the iterative data analysis process in their methods sections, few summarize the process of surprise and discovery involved in generating these codes (Locke, Golden-Biddle, & Feldman, 2008). Although it is neither possible nor illuminating to describe all of the myriad thoughts, tangents, and blind alleys that go into and ultimately result in the insights presented in the article, a template can give the impression that the research can be conducted without these other, often essential, parts of the research process. Moreover, a template will not offer guidance on the myriad methodical choices you need to explain in order to produce a compelling account of the process.

Finally, following a template does not ensure that researchers understand the assumptions underneath their choices or that they can explain what they did. Methods sections not only have to describe what a researcher has done but also outline why the researcher made those choices. For example, one can state that they “initially consult[ed] with existing literature, with suspension of judgment about its conclusion to allow discovery of new insights” (Gioia, et al., 2013, p. 25) without really knowing what this means or being able to articulate the role of theory in shaping the initial design and data collection. In our terms as editors, we have each seen researchers who are able to reference phrases from templates but who nonetheless do not communicate that they understand why their choices made sense in the context of the particular study they conducted. For example, authors may refer to first-, second- and third-order codes without understanding how these three relate to the generation of an explanation or even to each other.

These misunderstandings of a template may result in further confusion: Manuscripts that follow the template may be rejected without the authors understanding why. Indeed, in our roles as editors, we find that applying templates often instills in authors a sense that they have “done it right,” and thus they are perplexed when receiving a decision letter that raises substantial problems with their methodology.

Narrowing Methodological Choices

A different kind of problem with templates emerges from the limited conditions under which they are best suited. Templates simplify the research process by identifying a particular way of doing it. But any template also can limit the theoretical stories that can be told. Moreover, any given template is better at depicting some types of qualitative analyses than others. Even broad types of templates represent specific ways of enacting that type of research. The Gioia method, for instance, is based on grounded theory; but even within grounded theory, there is a diversity of perspectives. After the publication of the foundational book, Discovery of Grounded Theory (Glaser & Strauss, 1967), several versions of grounded theory have emerged (Locke, 2001). Compare, for instance, the views of grounded theory expressed by Glaser (1992, Basics of Grounded Theory Analysis) and by Strauss and Corbin (1988, Basics of Qualitative Research). And these perspectives differ from Charmaz’s (2014, Constructing Grounded Theory) more recent take on the method. Similarly, Eisenhardt’s take on case methodology is premised largely on Yin’s (2003, third edition of Case Study Research: Design and Methods) work, but there are other perspectives on how to do cases (e.g., Stake’s, 1995, The Art of Case Study Research) or how to compare cases (Vaughan, 2004, 2014). Put another way, it is implausible to expect a template to reflect the diversity of even a single qualitative methodology such as grounded theory or case analysis, much less be able to represent the wider range of qualitative methods that exists in our field. As a result, the use of templates is destined to limit methodological choices, thereby potentially limiting the theoretical work that arises from empirical studies.

Another hidden but potentially dangerous effect of templates is that they may “select out” certain moves that researchers found to be helpful to their research but were not mentioned in their articles because they did not fit the template. ² To illustrate, the Gioia method talks about the importance of a “well-specified, if rather general research question” (Gioia et al., 2013, p. 19). However, common to many inductive studies, research questions may change. Indeed, Pratt, Lepisto, and Dane (2019) noted that their initial questions were about how firefighters made intuitive decisions at the fireground but learned during the course of their data collection that this was not a central problem for the firefighters. As a result, their research question shifted dramatically to one examining the role of faith in trust. A similar shift in research question occurred for Feldman (2000) in the course of analysis when she realized that the “stable” routines she had studied over 4 years had actually changed in both small and large ways. Such moves are often critical in conducting the research, but they are not explicitly accounted for in the Gioia method or any template. Leaving these moves out to hew closer to a template would be misleading and would deprive readers of knowledge important to doing inductive research.

We have argued that although the elegance of a simple template makes the process appear less messy and more apparently predictable, it can mask the complexity of qualitative research and the potential of the data. It might also guide the researcher into a less promising part of the data that errs on oversimplification—leading to a paper that lacks enough insight to produce a publishable manuscript. At a minimum, the application of a template may lead to very different manuscripts than would occur if a template were not applied.

Take, for instance, written works based on using narratives as a way of analyzing the data, which are often used because they create a richly interconnected way of understanding and interpreting the data. The method has its roots in many disciplines, and there is a lot of variation in how it is enacted (Holstein & Gubrium, 2011; Riessman, 1993). As such, reducing narrative analysis to a template is difficult. Of course, the proponents of templates might argue that is precisely the point—that we need a more streamlined and less variant use of methods. However, our perspective is that heterogeneity is precisely what makes narrative analysis, or for that matter, qualitative research, a powerful and helpful framework. More specifically, it empowers researchers to customize the analysis in ways suited to their research question and data, avoiding a “one-size-fits-all” approach to data that inevitably takes many shapes and sizes. More problematically, trying to fit a narrative analysis into a more established template would deprive the field of results that are uniquely derived from a narrative method. This can happen when readers, reviewers, and editors take templates for the proper way to do qualitative research rather than a representation of one way to do that research.

To illustrate our point, consider the potential consequences that might have followed from one of our own articles had it followed the Gioia method as a template. The article was written by Sonenshein and published in the Academy of Management Journal in 2010. He explored and explained how managers and employees interpret change processes. In this illustration, we compare how the author’s analytical moves compare with those suggested by the Gioia method. We refer to specific methodological choices made by the author as analytical moves or simply, moves.

As mentioned in the introduction, the primary analytic strategy of the Gioia method is around coding. In grounded theorizing, these moves are often referred to as focused or open, axial, and selective coding and begin by coding very close to the data, organizing these codes into categories and subcategories, and finally “integrating and refining the theory” (Strauss & Corbin, 1988, p. 143). As teachers, advisors, and editors, we have long promoted these moves as appropriate places to start analysis and sometimes as appropriate places to end analysis. Sonenshein’s first analytical move, however, was to construct composite narratives of managers’ and employees’ interpretations of a change process in the organization he was studying. These composite narratives were based on raw data of extensive observations and interviews. One could liken this move to the “second-order coding” of the Gioia method (e.g., it is a form of representing patterns in data, albeit in a much more contextualized form), typically depicted by the middle column of a data structure figure used with the Gioia method. The similarity between these composite narratives and second-order coding, however, is misleading. Had Sonenshein been required to use the Gioia method, it is not clear how he would have put these composite narratives in the middle column. That would have been practically difficult given their length. And what would he have put in the first column? The construction of a composite narrative is based on understanding many narratives that have overlapping parts. Showing the raw data from which a composite narrative is composed would require sharing a huge amount of data, virtually the entire data set. And how would the Gioia method accommodate the analytic process from an individual’s narrative to a composite one? In the method Sonenshein used, he immersed himself in the individual narratives and depicted the central patterns in the data by writing a composite narrative for each pattern. This was not a reductive process, as is typical when moving from first- to second-order codes whereby a primary purpose is to abstract from the raw data. Rather, he wanted to both represent the general pattern in the data while retaining its contextualized elements, situating the narrative in the specific context he was studying. As a result, it is not clear how requiring Sonenshein to fill in these columns would have been possible or how it would have enhanced the research process or the ultimate contribution.

Sonenshein’s second analytical move was to analyze the composite narratives in order to see how employees’ and managers’ interpretations differed, how they each changed over time, how employee interpretations varied across different local contexts (e.g., stores), and ultimately, how their interpretations influenced the implementation of change. His analysis showed that the interpretations as captured in the composite narratives varied along two dimensions: Managers’ interpretations were either predominantly preservational or transformational; employees’ interpretations were also either predominantly preservational or transformational but also varied as to whether they were subversive or supportive.

He could have placed the words preservational, transformational, subversive, and supportive in a third column of a three-column data structure, but it appears likely that something would be lost if he had done this. Doing so would leave the impression that these third-order categories or aggregate themes are the important part of the story. But the important part of the story is the intricacy of the interplay—that managers, broadly speaking, use two different kinds of narratives simultaneously and that employees construct evolving and multilayered narratives in response. As Sonenshein (2010) put it, “The findings suggest two narrative pathways managers simultaneously use to implement strategic change and along which employees subsequently construct their own meanings and narrate responses to change” (p. 502). In other words, by retaining the intricacies of interpretations, including their very particular features tied to the setting, Sonenshein was able to report on a more general theoretical process that was highly dynamic. For the reader reviewing these composite narratives, it is likely easier to see these dynamics emerge because they are not forced into the confines of different layers of coding.

Sonenshein used narrative moves because it allowed him to organize and analyze his data in a way that showed the rich interconnections that are ultimately the foundation of his interpretation and his contribution. Had the reviewers or the editor required him to display the data using the Gioia data structure, the outcome would have been at best distracting and confusing because the display would have been irrelevant to the findings. The result would be a less rich interpretation of an important organizational phenomenon.

Templates themselves may be viewed as a series of moves (later in this article, we discuss exemplars that demonstrate the use of templates and their various moves; see “Stepping Back: Analysis of 12 Additional Articles”). We believe that some or all of the moves utilized by the Gioia method can, in the right circumstances, be sound ones—after all, we have used some of them ourselves (e.g., Pratt, Rockmann, & Kaufmann, 2006; Rerup & Feldman, 2011; Sonenshein, 2014). For example, one could include a data structure in an article to illustrate how one moved from data to theory without using the entire Gioia methodology. But they are not the only moves—nor, depending on the data and the theory one is trying to build or extend, are they always the most helpful ones. For this reason, it is particularly harmful when authors adopt a template or when editors or reviewers strongly suggest the addition of a template without a clear reason for why a template matches what the author is trying to do or what is appropriate for the particular data.

Proposed Solution: Methodological Bricolage as an Analytical Alternative to Methodological Templates

We introduce bricolage as a metaphor for understanding an alternative way of making methodological choices and bringing forward the mindfulness entailed in these choices. Klag and Langley (2013, pp. 161-162) previously used the term bricolage to describe the process of “conceptual leaping” in qualitative research. Here, we expand the use of the metaphor and develop it as an approach to doing qualitative research. We use the term methodological bricolage in contrast to methodological templates and refer to it as an approach rather than a method because we consider it a way of thinking about how you do your methods (i.e., it is metamethodological). Inherent in this approach is the potential to improve how a researcher understands, articulates, and enacts the methodological choices for a study. Although perhaps most explicit in methods sections, methodological choices can also be found throughout an article, for example, in choices about how findings are presented, which findings are abstracted and why, how to develop and/or elaborate theory, and what methodological limitations to highlight in the discussion. Our overarching goal is to foster research that instills a sense of competence, integrity, and benevolence to insights gained in the field while also acknowledging that there are many ways to achieve these ends. We acknowledge that there are many possible ways of conducting high-quality qualitative research, and the metaphor of methodological bricolage provides a framework for thinking about this diversity of approaches in ways that fosters trust in the methods used.

We draw on bricolage because it makes salient various assumptions about the process and purpose of methods. Even if an author ends up following moves that approximate a methodological template, viewing methods from a bricolage approach provides benefits to the authors (and readers) of a manuscript. Specifically, this process involves an active choice of moves from a broad set of methods, which provides the basis for an active explanation of moves taken. The bricolage metaphor allows us to refer to researchers as bricoleurs, which draws attention to their agency, creativity, and craft (see also Klag & Langley, 2013) in the process of designing, conducting, and presenting research. There is always a researcher or team of researchers making active choices about research design, data collection, and data analysis (who to interview, what questions to follow up on, how to code data, etc.). Referring to a template at best hides or at worst strips away the agency of the researcher. Moreover, the bricolage metaphor provides a way of thinking about how to make methodological choices more transparent. We detail in the following three central elements of bricolage: making do, utilizing the resources at hand, and combining resources for new purposes (Baker & Nelson, 2005). These elements help create an “effective arrangement” (Duymedjian & Rüling, 2010, p. 141), contributing to the trustworthiness of the research.

Bricolage is, first of all, a process of making do. It is not predefined but is also not simply whatever is produced. The bricoleur’s competence “resides in his [sic] know-how and ability to ‘cobble things together’” (Duymedjian & Rüling, 2010, p. 141). Although making do is often empirically examined in the context of resource constraints (e.g., Baker & Nelson, 2005), it can occur in a variety of resource environments (Sonenshein, 2014).

Successful bricolage produces an arrangement that works for the purpose at hand (Duymedjian & Rüling, 2010). In the case of methodological bricolage, researchers “cobble together” or create a coherent argument that connects a research question with a research answer through engagement with data and theory in a manner that creates trust between the author and the reader. This echoes the sentiments in our opening quote that suggests that what is necessary is creatively ensuring that one’s story coheres. Like most creative processes, the process of bricolage is likely to be messy, involving trial and error. It also matches a widely accepted understanding that qualitative research is very iterative (Charmaz, 2014; Locke, 2001; Spradley, 1979).

Second, bricolage involves engaging with and understanding the available resources. By resources, we mean the ways that qualitative research has been conducted and discussed, including but not limited to the use of and writings about grounded theory, ethnography, narrative analysis, coding, interviewing techniques, and so forth. We are not suggesting that any one researcher will know all of the resources—but will know some of them—and that familiarity with how qualitative research has been done and how it has been discussed provides the researcher useful tools. The more diverse the bricoleur’s toolbox, the more problems the bricoleur can potentially solve. In other words, a diversity of research projects can improve the methodological bricolage process.

Like other scholars, we recognize that our field has traditionally been methodologically pluralistic (Cunliffe, 2011; Easterby-Smith et al., 2008; Prasad, 2005). A bricoleur’s understanding of the available resources allows the bricoleur to use available resources to the best effect, even in ways that have not been previously envisioned (Duymedjian & Rüling, 2010). This focus on knowing one’s resources before engaging in what we are calling methodological bricolage is similar to an argument made by Prasad (2005):

In general, I would advocate that a researcher become both familiar with and comfortable in the tradition(s) of her choice. Once a certain level of theoretical and methodological comfort is reached, the researcher can draw on those elements of the tradition that work best for him or her and that are most suitable for the work at hand. (p. 285)

In methodological bricolage, analytical moves (for research design, data collection, data analysis, and data presentation) are essential resources. Many of these moves already exist, although any particular study may entail new ways of using them. Learning about methodological moves that are already at hand and how they have been used allows researchers to benefit from previous research and discover which moves work in their particular study. To illustrate, one important characteristic of how moves are used relates to the ontological assumptions that underlie them (see also Gephardt, 2004; Prasad, 2005). Even though some distinguishing characteristics between ontologies have been blurred (see Cunliffe, 2011) and some methodologies, such as grounded theory, can be approached from different ontological perspectives (Locke, 2011), different methodologies nonetheless are often grounded in specific assumptions about reality. In particular, research methods vary in the degree to which they adhere to more realist, interpretivist, or even critical postmodern ontology (Gephardt, 2004). As we develop in our discussion of integrity in the following, awareness of ontological differences that underlie moves is critical to knowing how to combine them effectively. By understanding better the available methodological moves, researchers can also fine-tune or create moves that are particular to their research and address the challenges of making an effective arrangement in the context of their data and research question.

Third, bricolage involves combining resources for a new purpose. Applied here, it involves utilizing existing moves to solve the particular problems and challenges posed by a specific study. In methodological bricolage, researchers make choices about what combinations of moves to use in their project. In concert with fine-tuning existing moves, combining moves helps yield creative combinations of practices crafted to suit a particular study.

Taken together, methodological bricolage is the combining of analytic moves for the purpose of solving a problem or problems tailored to one’s own research project. When done well, the result is an effective arrangement that exhibits trustworthiness. We want to reiterate that utilizing methodological bricolage warrants active choices about which moves to apply. Effective arrangements do not emerge on their own; rather, they are created by researchers, whose voice should be visible in the methods section and in the write-up of one’s findings. Methodological bricolage is motivated by the different challenges one faces with each new project or study and is a way of creatively meeting these challenges. With this approach comes experimentation and surprise (Locke et al., 2008).

Effective arrangements emerge through observing which moves yield compelling and interesting ways to identify, understand, and illuminate the patterns in the data and connections to the research question. Because methodological bricolage can encompass analytic moves from a variety of methods, scholarly articles that use this approach have unique ways of designing, collecting, analyzing, and presenting their data. This may cause a challenge for both the author and the reader: How do you write or evaluate what makes an arrangement effective? On this point, establishing trust between the researcher and the reader is especially critical. Researchers have the responsibility to establish the trustworthiness of their specific arrangement of analytical moves. In practice, this responsibility always belongs to the author, whether following a template or practicing methodological bricolage. Although following a template holds the promise of producing a prepackaged effective arrangement, as explained earlier, the arrangement is not always effective, and explaining why the template is appropriate helps establish the needed sense of trust between authors and readers. As noted, although much of the articulation of choices that helps to establish trustworthiness takes place in the methods section of an article, the arrangement is also evident throughout the article, for instance, in the framing of the article, in the presentation of the findings, and in the discussion of implications.

Knowing what is entailed in establishing trustworthiness is clearly an important part of creating an effective methodological arrangement. In the following, we engage a well-known way of conceptualizing trustworthiness to provide some guidance about how to create arrangements that are effective in creating trust between researchers and readers. For this purpose, we view an effective arrangement as trustworthy if it exhibits competence, integrity, and benevolence (Mayer, Schoorman, & Davis, 1995).

Creating an Effective Arrangement by Establishing Trustworthiness

Bricoleurs display three abilities: the ability (a) to “cobble things together,” (b) to understand and engage available resources, and (c) to use a combination of existing analytical moves to solve a particular problem (Baker & Nelson, 2005; Duymedjian & Rüling, 2010). When performed well, these abilities create effective arrangements. Explaining the logic behind the arrangement conveys trustworthiness.

In methodology, the kind of explanation that creates trustworthiness varies based on ontology and epistemology. For example, if viewing reality as concrete and taking a positivist stance toward research, then trustworthiness may be communicated by following the precepts outlined by Yin (2003) in his formulation of case studies. Yin suggested that construct validity, external validity, and reliability can be achieved in case studies by such things as data triangulation, replication logic (for multiple cases), and careful construction of a case study database. Lincoln and Guba (1985) see the world as more socially constructed and thus gather knowledge via naturalistic inquiry. Unlike Yin, they suggested the importance of credibility, transferability, dependability, and confirmability as markers of trustworthy methods. Still other signals of trustworthiness are needed in new growth areas in qualitative methodology in management, such as those utilizing computer-aided qualitative data analysis software (O’Kane, Smith, & Lerman, 2019) or engaging in “linguistic-based, critical, and poststructuralist work” (Cunliffe, 2011, p. 651).

Enacting a methodological bricolage approach may entail drawing on moves from different methodologies and even different ontologies. When moving beyond the single methodology (including its ontological assumptions), we need a rich understanding of what trustworthiness is. The analytical moves described by Lincoln and Guba (1985) and Yin (2003) or identified in templates contribute to assessments about whether a researcher has “done it right” methodologically (i.e., that the researcher is competent). But one needs to be mindful about whether and how these moves fit together, and this suggests the need for coherence or integrity. We further argue that for inductive research, one needs to assess whether the researcher has actually stayed faithful to the insights provided by informants—which we view as a form of benevolence. In the following section, we elaborate each of these criteria for creating trustworthiness in methodological bricolage: competence, integrity, and benevolence.

Competence involves knowing the range of analytical moves available to design and implement a study and when and how to use them. Competence is essential for differentiating bricolage from a haphazard recombination of elements (see also Easterby-Smith et al., 2008). To a large degree, standards of competence are externally derived and validated. If one chooses to conduct an ethnography, for example, there are some moves (e.g., engaging in some form of participant observation) that come with using that particular method. That said, creativity often stems from how one situates a method within one’s own particular research design. For example, an ethnographic study of an online community may involve “lurking” rather than actively participating in the group or collective one is studying. As enough people begin to take similar variations, new standards can arise (e.g., netnography as a type of ethnography).

Integrity, as used here, means that the various analytic moves “cobbled together” fit with each other. That is, do the moves involving the research question, sampling, analysis, and other parts of your method cohere? For example, if your research question is about change processes, it is likely that you would design a longitudinal study, sample events or individuals that you could track over time, choose a context where change is happening, and use data-collection strategies that would be able to get at the changes you seek to explore (e.g., observing the same routine enactment or event over time or interviewing the same individuals at different points). Our extended illustration of Sonenshein’s article and his use of composite narratives, rather than a three-column data structure, also illustrates this type of integrity. Here, the use of composite narratives as an analytic move allowed the author to communicate the rich, contextualized understandings of how managers and employees interpret change processes in a relatively concise way.

One particular aspect of integrity that scholars have discussed involves the ontological underpinnings of the research (see Buchanan & Bryman, 2007; Denzin, 2010). For example, the unintentional mixing of positivist and interpretivist arguments reflects a lack of ontological coherence. Methodological bricolage does not imply a particular ontological (or epistemological) tradition. However, to achieve integrity, one must be aware of the ontological traditions of the moves one uses. A lack of integrity may be signaled by citing Miles and Huberman (1994), who take a positivist stance toward research, alongside Charmaz (2014), who is interpretivist, without acknowledging their deep philosophical differences. More problematically, in our roles as editors, we have often seen individuals who claim to be interpretive in their research but then use random sampling, which is based in a positivist understanding of the research process. Although this combination may make sense in some circumstances—for instance, if you want to move from a positivist to an interpretivist stance or vice versa in your study—the combination requires an explanation that must be tied to the research question. Put another way, we are not saying that a researcher always needs to be ontologically consistent. If you are going to be ontologically inconsistent or eclectic in your use of analytic moves, however, it is incumbent on you to explain why this is appropriate for the research you are engaged in.

We add benevolence to this mix because of the importance in maintaining fidelity to the data. In inductive, qualitative research, this is manifested by taking into account the interpretations and experiences of informants in formulating explanations. Benevolence moves focus on communicating that authors are not “doing violence to experience,” that they are able to communicate how the insights of the people or places they examined served to inform their theorizing (Pratt, 2008). As noted by Easterby-Smith and colleagues (2008), it is important to communicate the “proximity to the life of the worlds of those studied” (p. 243). To be clear, unlike the traditional trust framework, we do not mean benevolence broadly as expressions of kindness toward those studied, although we do expect that all research is done respectfully. It also does not mean that one only shows what is best about the places and people studied. Rather, the “kindness” employed here is honoring and respecting the knowledge and insights of those studied highly enough that it influences the theories we create. As with integrity, it is often the absence of benevolence that disrupts trustworthiness. For example, studies where theory seems largely divorced from the quotations or other evidence one gathered in the field may be a sign that benevolence is low. This may happen when authors come into a study with strong theoretical assumptions that “steamroll” what the data appear to say. It may also be reflected in not showing any or much source data but only evidence that is abstracted and interpreted. This is what Golden-Biddle and Locke (2007) referred to as “telling” about one’s data rather than “showing” it. By contrast, showing source data and effectively showing how it relates to one’s theorizing does show benevolence. In addition, you can bolster benevolence claims by talking about how long you have been in the field, how much time was spent with informants, and how well you know the context as evidence that—at the very least—you took the time to understand the “voice” of those studied.

This element of trustworthiness resonates with Golden-Biddle and Locke’s (1993) discussion of how ethnographies can be convincing by communicating authenticity: that one was actually in the field. It is also similar to Lincoln and Guba’s (1985) notion of “credibility” that asks whether investigators have adequately represented the multitude of perspectives in their study. Not surprisingly, both ethnography and naturalistic inquiry emphasize what one learns from one’s informants. ³ Yin (2003), by contrast, did not talk about benevolence per se, but even from a positivist stance, it is important for qualitative researchers to get data from multiple sources to help ensure that their findings better reflect insights from the field. What methodological bricolage adds to these depictions is that it is not only important to represent the voice of those studied but also to do so and to explain the doing in the context of the particular study one is engaged in, the theory one wants to build, and more generally, in the problem one is trying to solve.

Zooming in: Analysis of Three of our Own Articles

To illustrate an effective arrangement, which involves communicating competence, integrity, and benevolence, we each describe (in first person) one of our published articles and describe how we bricolaged it in Table 1. In the spirit of bricolage, we include the major methodological challenges in the article (i.e., the problem at hand), how we resolved them using methodological bricolage, and the trustworthiness moves used established to render an effective arrangement.

OPEN IN VIEWER

Table 1.

Three Authors Reflecting on Methodological Bricolage in a Published Manuscript.

Sonenshein (2010) [in Sonenshein’s own words] In Sonenshein (2010), I asked: How do managers’ and employees’ meaning constructions differ, and how do these differences matter during change implementation? Given the focus on meanings, I primarily used a narrative lens, which is particularly well suited to accounting for the multiple voices present during change. By focusing on multiple voices, my purpose was to avoid monolithic accounts of change that silence certain groups, privilege other ones, and fail to account for the dynamics across groups. Although a narrative approach was fitting for my research question, a major challenge was that I was studying a Fortune 500, geographically dispersed organization with employees across the country. It simply was not economically or logistically feasible to interview employees at the hundreds of spread out locations to collect their narratives in person and to probe them for follow-ups. Further adding to my difficulties, this research was part of my dissertation, so I had an urgent timeline to complete it. Another major challenge was that I was studying organizational change, which is an inherently temporal process. I needed to make sure that my data and analysis accounted for time and that my field engagement could observe, explain, and ultimately theorize aspects of the transforming organization—which, naturally, was on an unpredictable and uncontrollable timeline that did not necessarily match my own timeline. When studying change, this can be particularly difficult because as an outside researcher, it is often hard to anticipate how long a change will take place, where to be, who to speak with as power dynamics shift, and when the best time to exit the field is. Finally, I had collected a lot of data that made analyzing all of it inordinately complex, including 42 interviews, 115 documents, an ongoing project archive, 8 hours of observation, and 210 surveys that included qualitative, short-response questions. To create an effective arrangement, I turned to composite narratives as a major component of my analysis. This approach involves a researcher constructing an event based on other individuals’ own constructions of that same event (Dunford & Jones, 2000). It allowed me to create a manageable set of narratives that accounted for major differences in meanings across groups, drawing off of the wide range of data I collected. It also helped me consolidate my data and signaled competence to the reader, conveying familiarity with this move and establishing it as a sound practice for analyzing my data. Another form of competence came from the data itself. Drawing off of multiple sources of data and “official” interpretations from top managers to corroborate the interviews I conducted with them helped convey the level of care I took with the data collection. To construct composite narratives, I used a mixture of moves, including timelines (which were especially helpful for a temporal phenomenon such as organizational change), inductive coding, and thematic analysis of surveys. This was admittedly a hodgepodge of analytical moves but was guided by trying to account for a variety of data, in different forms, and with different degrees of richness. I relied on detailed tables to substantiate my creation of the composite narratives and block quotes in the text, both of which I used to help transport the reader into my context and to believe my own interpretations (a way of conveying benevolence). Although I typically do not collect survey data for my qualitative projects, I felt I needed to for this project to tap into hard-to-reach employees at locations I could not visit for a personal interview. Not surprisingly, the survey data was not as rich as in-person interviews, so I needed to not only do more typical inductive coding (which was most generative on my interview data) but also a thematic analysis of the short answer survey items. Given that some of my data were relatively thin—the survey responses—I believed it was critical to highlight the richer part of my data (to bolster competence). The survey data were also critical in reaching regular employees, overcoming the challenge of collecting data from geographically dispersed employees. The benevolence standard was very much in mind here but proved challenging. Informants who only answered surveys (regular employees) meant that their data were less rich and therefore might be less convincing to the reader and therefore less likely to be included in a manuscript. However, by grouping together informants who had similar interpretations through composite narratives, I hoped to allow participants with less of a rich voice to also have their views presented in the data. More specifically, the composite narrative allowed their voices to be louder as a collective interpretation based on a composite of patterns in the data. Furthermore, by presenting composite narratives for both employees and managers alongside each other, I wasn’t privileging any one group’s voice.

The final theoretical model (found in Figure 1 of the original article) was a consequence of all of the methodological tools I used. It was impossible to follow a template, or even find a comparable article to base my methods off of, because of the unique combinations of my research question, data, and context. I mixed moves that I bricolaged together and aggregated into composite narratives. To generate these narratives, I used moves from grounded theory, case study analysis, and thematic analysis in a way that brought out the richness of the data while contributing to research on sensemaking and organizational change. The overarching principle was to make sure I could depict the various meanings of change, across time and location. An ability to understand those meanings was a guiding selection criteria for selecting this effective arrangement of elements. As I reflect back on this research, I found integrity especially important, as there were few, if any, similar studies I could reference. Without a recipe, I needed to convince the reader how the different moves were coherent. The composite narratives proved very valuable here because they allowed me to pull everything together. I had rich narratives—even if drawing from less rich data—that met my research objectives of showing changes in meaning across groups and time. Instead of presenting a set of disjointed analytic moves, I organized my findings around a coherent set of composite narratives. I also tried to be transparent with the evolution of my research, presenting its twists, such as an unexpected focus on managerial discourse too.

Pratt & Rosa (2003) [in Pratt’s own words] My coauthor and I were curious as to why network marketing organizations (NMOs) appeared to magnify work-family tensions in their motivational pitches to those who sold their products (e.g., distributors or consultants). Although I had noticed the phenomenon when gathering my dissertation data from Amway, my data were not sufficient to deeply understand its impact. Moreover, we wondered if similar practices were used in other NMOs. Toward this end, my coauthor collected new data in two other NMOs: Mary Kay and Longaberger. We faced at least two main challenges, one practical and one theoretical. The first was how to combine multiple qualitative data sets that utilized different data sources and data-collection techniques. At the time, we could not find books or articles on the topic or exemplars in top journals. The second was convincing our readers that—as indicated by our data—work-family tensions, which have long been theorized to weaken members’ ties to an organization, can actually be used by leaders to increase commitment. Related to this challenge, we noticed that the process of engendering commitment that seemed evident in these NMOs differed from those we had seen in extant theory on commitment. Thus, we were potentially offering a new way to see a foundational concept in organizational behavior (O’Reilly, 1991). To address these challenges, we engaged in a series of moves that not only combined practices from a range of methodologies but also created new ones. Though implicit, we borrowed on the logic of multiple case studies to justify why we added Mary Kay and Longaberger to the data set. We also borrowed from ethnography and other qualitative sources to justify the use of our earliest data-collection methods, namely, participant observation and ethnographic interviews. We argued that to get a handle on the complex and hard-to-observe dynamics we were interested in studying (e.g., Douglas, 1976; Spradley, 1979), we needed to actually be engaging in distributor practices and then listening to how NMO “insiders” explained why we were engaging in them. Once we identified a certain set of practices we were interested in, heightening work-family tensions, we switched to a grounded theory logic to explain why we needed to narrow and deepen the focus of our interviews (thereby changing the interview protocol) in our subsequent cases and why we moved to more theoretical sampling (e.g., distributors who were mothers). To explain how we analyzed our data, we created a data analysis figure (Figure 1, p. 397) that divided our analysis into two phases. Although we label the phases with new terms, “Discovery and Narrowing” and “Enriching and Validating,” the logic we use within this phase borrows heavily on grounded theory that moves from coding close to the data toward more abstract codes and that iterates emerging theory with existing theory. Taken together, in our methods section, we drew on the strengths of different methodologies in our article: case studies to explain our sampling of cases, ethnography to explain our initial data sources that necessitated getting broad cultural insight, and grounded theory to explain why we changed our sampling of informants and interview protocol across time to fit our emerging theory and how we pulled it all together in our data analysis.

By drawing on the strengths of different methodologies, we were able to meet both our practical challenges of combining data sets and our theoretical challenges of arguing for a new way to look at work-family tensions. Upon reflection, we pulled this off by using a combination of competence, integrity, and benevolence moves to show: (a) that we knew what we were doing and thus could combine data sets in a coherent way that challenged extant thinking and (b) that these insights really drew from the people we were studying. We signaled competence by discussing the different types of data we collected, why we chose to collect them, and by utilizing well-known sources to legitimate these choices. In this same section, we signaled benevolence by explaining why I spent nearly two years as a participant observer in Amway. As we turned to our analysis, we had to justify combining data sets that utilized different types of data (e.g., different types of interview and observations in Amway versus the other contexts). Thus, in the analysis section, we shifted strongly to competence and integrity moves. We began by admitting that our analysis was not quite traditional and showed competence by discussing where we varied from existing methodological canon. Throughout our methods section, we justified how we sampled and analyzed our different sources of data in light our research question, which signaled integrity. Methodological bricolage was not confined to the methods section, however. In our discussion section, we switched back to a combination of competence, integrity, and benevolence moves. For example, to signal competence and benevolence, we drew on case study logic via a “triangulation table” (Table 1, p. 400) to show the strength of evidence of our emergent constructs across the three cases. In addition, we used the ethnographic practice of long quotes and linked them to our model to show integrity and to help illustrate that we were not doing “violence to experience.” In sum, given that we were engaging in new practices and challenging extant theory, we needed to engage in moves that communicated the trustworthiness of our arrangement of methodological moves.

Feldman (2000) [in Feldman’s own words] This study began as an exploration of the mechanisms of stability in organizational routines. I chose an organization with routines that its members assured me were extraordinarily stable. So far so good. Over the course of 4.5 years of ethnographic engagement, however, these routines changed in both small and large ways. As a result, analyzing my data involved not only the standard challenges of the enormity and complexity of ethnographic data gathered over a long period of time but also the challenge of recasting my orientation to the data and developing a new research question. Today, I would recognize this as an abductive process in which my analysis helped me engage in a process of discovery based on my interaction with the data. At the time, I simply knew that I needed to engage with my data in ways that would help me explore and develop new possibilities that would allow me to understand how the changes I observed were part of the stability that members of the organization experienced. Creating an effective arrangement required many choices in collecting and analyzing the data as well as in theorizing. Establishing myself as a competent ethnographer was important. There is certainly an element of methodological bricolage to any ethnographic study as the ethnographer takes advantage of opportunities that arise for engaging with the field site and gathering data. Planning the hours of observation, the number of interviews, and the people to interview is just a start. Some of the best information comes from unplanned opportunities. In my case, I needed to understand how people accomplished their work. While much of the work in my setting was discussed in meetings, had I only observed prescheduled meetings and interviewed organizational members, my understanding of how people accomplished their work would have been very limited. I also learned a great deal from informal conversations, hanging around and pitching in when small and large crises arose, and taking up the challenge when members of the organization suggested that I could participate in doing things like conducting trainings and helping midlevel managers articulate their concerns to their bosses. Because email had become an important part of the organizational community during the time of the study, I also asked for and received permission to collect email streams and collected over 10,000 emails. This broad base of engagement gave me a deep understanding of the way people did their work in this organization that allowed me to show competence when I wrote vignettes about how they deal with certain situations. It also provided me with a basis for benevolence. I knew these people. I understood their concerns and had experienced many of the concerns myself.

The analysis also entailed bricolage. The first stage of analysis was writing the ethnographic description of the five routines I had followed. This was a practice that I had seen other ethnographers use. The next stage of analysis was entirely new: I used various theories (e.g., ethnomethodology, semiotics, dramaturgy, and deconstruction) as analytical tools. I call this metatheoretical analysis, and I wrote a monograph describing this way of doing analysis (Feldman, 1995). Each of the theories provided constraints that helped me disrupt the order in which the data had come to me (see Locke, Golden-Biddle, & Feldman, 2008) and provided me with new ways of seeing into the data. Writing was a next step in the analysis. I wrote vignettes of the changes I had observed. These vignettes drew on both the ethnographic descriptions and the metatheoretical analyses and helped me engage more deeply with both of these earlier moves. Describing the process of analysis is always an important part of establishing competence, but in this case, I had some difficulty. Indeed, when one of the reviewers asked that I draw a clear path between each of the metatheoretical analyses and the resulting insight, I could not do that. For me, the point of presenting information about the variety of analyses was not that each had led to a singular insight but that together they allowed me to see meaningful connections within the data. Much like Scott’s composite narratives, it is impossible to show all of these connections without taking the reader through all of the analyses and the subsequent memoing. I don’t know whether this reviewer was ever satisfied. I suspect that it was the integrity of the interpretation that emerged from seeing and showing the connections within the data that allowed the reviewers and subsequent readers to trust my analytical approach. An important part of establishing integrity was showing that my counterintuitive insight—that routines entail both stability and change—was consistent with the data and the analysis. I did this, in part, by acknowledging that my research question had changed over the course of the study. This admission established that I had not gone looking for change but had found it despite my efforts to focus on stability. Since I had created new ways of analyzing data specifically for this study, I needed to assure the reader that they were appropriate for the data. I did this, in part, by noting that the analyses “allowed me to develop new understandings of the data I had gathered without going beyond what I had actually observed or been told” (p. 615). I also established integrity in the presentation of the data by first presenting vignettes that provided an overall view of the stability and change I had observed and then showing how stability and change played out in these vignettes through the discussion and table that followed. In connecting the analysis to theorizing, I drew on a general insight that emerged from the metatheory analysis that while concepts may be separable analytically, they are not separable in the process of doing. In particular, the analysis had helped me to see that stability and change are entangled and inseparable in the enactment of organizational routines. In this part of the research process, I gravitated to theories of practice because they deal with entangled concepts. Establishing trust here required both that I explain my findings in ways that scholars who are experts in practice theory would recognize my competence with those theories and that scholars unfamiliar with practice theory would recognize the integrity of the interpretation without them having to know practice theory. Presenting the data also involved a bricolaged combination of vignettes, discussion, and a table. Vignettes were important for showing the routines in context. Each vignette illustrated multiple kinds of stability and change within the routines I had observed. I followed the vignettes with a discussion of the four kinds of change and three change responses they illustrated, and I created a table that made it easy to see that, across the routines, there were multiple instances of each kind of change and each kind of change response. The vignettes and the table were straightforward and hard to refute even though the thrust of the evidence was counterintuitive. Together, they showed that the dynamics of stability and change were in each of the three presented routines. Combining these ways of presenting the data provided layers of evidence supporting the theoretical move that had emerged from the analysis. None of these moves were new, and I’m sure there are other papers that use the same or similar combinations. I chose from among a variety of possible moves a combination of presentation modes that would assure the reader of the integrity of the interpretation I had made of the data. The detail in the vignettes also allowed me to display benevolence by showing that I understood the ideas and ideals of the people whose actions I portrayed.

Stepping Back: Analysis of 12 Additional Articles

The deep dive into our own work helped us illustrate the broad ideas around an effective arrangement that established trust between authors and readers. Although this process was very generative, we wanted to consult additional qualitative research to refine our ideas and illustrate a wider variety of analytic moves authors have made to create a trustworthy effective arrangement. We chose 12 articles published in top journals that use a variety of qualitative methods ⁴ encompassing a wide range of moves. Then we identified the critical moves authors made that produced an effective arrangement. This portrayal, unlike the portrayal of our own articles, is from the perspective of a reader.

Table 2 lists each of the 12 articles as well as the research questions posed, the analytical moves used, and the trustworthiness dimensions we believe those moves supported. When research questions were not explicit, we inferred them from each article. We include research questions because, as we explained earlier, what authors need to establish is that the reader can trust their answer and their way of answering the research question. Thus, the analytic moves here are appropriate for the specific research projects and research questions. The list provides a broad array of moves. In the spirit of bricolage, one cannot incorporate any move from the table haphazardly into any given manuscript. To illustrate, taking Bechky and Okhuysen’s (2011) analytic move in their study design phase of comparing very different contexts to show how they were actually similar is a very different move than that used by Pratt and Rosa (2003), who were explaining how similar contexts were potentially different. Rather, the goal is to create an effective arrangement that cobbles together the moves that create a trustworthy manuscript in order to answer a study’s research question.

OPEN IN VIEWER

Table 2.

Illustrating Trustworthiness Moves in 12 Management Articles: ⚠ Warning: Uncritical application of this table to your writing may result in a rejection of your manuscript and damage to your career.

Article (Alphabetical Order by First Author)	Research Question	Illustrative Analytic Moves by Research Question	Dimensions of Trustworthiness
1. Barley (1986) “Technology as Occasion for Structuring…”	Applying and testing theories of structuring to technological implementation	In analysis, mapped out structuration phases prior to looking at scripts because according to the theory he draws on, structuring should start with major internal or external events	Competence
		Length of time in the fieldFocus on close observations	Benevolence
		Given emphasis on structuring, organized data in phasesDesigned the study to get real-time data given his research question about patterns of emergent actionQuantified data and ran regressions in order to test theory given his more deductive focus	Integrity
2. Bechky & Okhuysen (2011) “Expecting the unexpected?…”	Inductive understanding of how two seemingly different types of teams (film crews and SWAT teams) handled ongoing surprises	Detailed description of data sourcesCites noted grounded theory and case methods sourcesSharing and questioning of raw data	Competence
		Detailed description of research settings	Benevolence
		Detailed description of how the cases compared with each other (including a table) fits inductive, comparative research questionUse of constant comparative method appropriate for research questionUsed multiple examples of surprise and bricolage responses across both settings that also fits inductive, comparative nature of research question	Integrity
3. Boje (1995) “Stories of a Storytelling Organization…”	Exploring the interplay of premodern, modern, and postmodern discourse in Walt Disney stories	Cites noted deconstruction sourcesNoted faults with past deconstruction effortsDescribes rationale for analytic shift in perspective	Competence
		Went to original sources in archives to get Disney storiesGathered information from CEOs, non-CEOs, and outsiders to get various “voices”	Benevolence
		Displays story data according to time period to emphasize the differences across premodern, modern, and postmodern	Integrity
4. Dutton & Dukerich (1991) “Keeping an Eye on the Mirror…”	“Generate new theory on how individual interpretation and organizational action on an issue are related over time” (p. 520)	Cites noted case study sourcesOutlines analytical steps	Competence
		Noted surprise in data-collection efforts	Benevolence
		Timelines in Figures 1 and Figure 2, along with the organization of findings across phases key to answering research question about how issue interpretation and action relate over time	Integrity
5. Eisenhardt (1989) “Making Fast Strategic Decisions in High-Velocity Environments”	Explores “how executives decide quickly” (p. 544)	Cites noted case study sourceUtilized replication logic and embedded designNotes different data sourcesOutlined rules used in data gatheringUsed “courtroom style questioning” (p. 547)	Competence
		Utilized long (90-120 minutes), open-ended, inductive interviews to get top managers’ perspectives; this information was used to map out decision processesUtilized all interview data in interview notes	Benevolence
		Table 2 directly relates to research question by showing data of the speed of strategic decision-making across contextsData organized around speed, decisions, and decision-makers (see also Tables 3-6 and propositions)	Integrity
6. Elsbach & Kramer (1996). “Members’ Responses to Organizational Identity Threats…”	Understanding responses to Business Week rankings as identity threats to business schools	Explained sampling choicesDelineated data sources and why they were usedCited relevant and established methods sourcesUsed number counts to legitimate coding categoriesDiscussed early, middle, and later iterations of data analysis	Competence
		Multiple iterations of data through multiple sources of data using multiple coders to ensure veracity of their findings	Benevolence
		Utilized archival data to get benchmarks on ratings data and various archival sources to get at responsesOrganized data in terms of identities (see also Table 2), identity threats, and responses in order to show evidence of answering research question	Integrity
7. Gioia et al. (1994), “Symbolism and Strategic Change in Academia…”	Research question: “uses of sensemaking, influence, and symbolism in launching a strategic change initiative at a university” (p. 363)	Discussed and justified interpretive stance and why collected different perspectivesExplained different insider vs. outsider roles of researchers in terms of data and analysisCites established methodological sourcesFollowed grounded theory processes of constant comparison and theoretical samplingUsed early version of Gioia “data structure” to show “chain-of-evidence”Discussed integration of insider and outsider analysesDid member checks	Competence
		Strong interpretive stance that highlights importance of informants’ voices in developing theory	Benevolence
		Layout of data and Exhibit 2 highlight different phases of how task force moves from sensemaking to strategic implementationUsed computer program and interrater reliability in coding to assist in coding given their concern for bias (integrity with epistemological assumptions regarding rigor)	Integrity
8. Isabella (1990), “Evolving Interpretations as Change Unfolds…”	“(1) How do managers construe events over time? and (2) How are those viewpoints linked to the process of change?”	Described sampling (explains use of random sampling)Discusses rationale for looking at eventsIncludes interview questionsDiscussed rationale for how she asked interview questionsCited legitimate methodological sourcesTable that outlines coding categoriesData tableDiscusses analysis
		Outlines interpretive assumptions	Benevolence
		Depicted findings in temporal stages to fit her research question (see also Tables 2 -3) and explicitly links to change (Table 4)	Integrity
9. Kaplan & Orlikowski (2013), “Temporal Work in Strategy Making”	“How do managers negotiate and resolve differences in interpretations of the past, present and future to make strategy? And, specifically, why do some interpretive linkage work and some fail?” (p. 968)	Cites relevant methods expertsJustifies deep engagement theoreticallySpecifies case designData sources tableExtended description of analysisFigure describing in depth how they analyzed their data (in appendix)	Competence
		In-depth discussion of research context “Paid attention to surprises and puzzles” in analysis (p. 968)	Benevolence
		Lays out findings in terms of past, present, and future and unpacks temporal work (see also Figure 1 and Table 3)Creates timeline of temporal work (Figures 2-7) in each organization studied	Integrity
10. Locke (1996) “A Funny Thing Happened…”	What is the role of playfulness in the provision of medical care?	Detailed discussions of note-taking and field notesDescription of multiple data sourcesCiting relevant and legitimate methods sourcesExtended description of different ways data were analyzedDebriefing with colleagues	Competence
		Extended discussion of context and occupationExtended time in the field with details about time durations of onsite visitsFrequent discussions of personal mistakes in interpretations and surprises	Benevolence
		Analyzed comedic performances to get at “playfulness”Organizes data around different types of comedic performances (see also Table 1)Relatively long quotes to emphasize comedic practices	Integrity
11. Plowman, et al. (2007), “Radical Change Accidentally…”	“In this research, we attempt to understand how and why an initial small change, whose ultimate consequences were unintended, escalated and led to radical organizational change” (p. 516)	Cites relevant and legitimate methods sourcesJustifies decision to do a case studyDelineates data sourcesDetailed discussion of various analytical types usedData source table	Competence
		Utilized multiple methods and coders to help ensure the veracity of their findingsClearly stated how their research question changed based on what they were finding in the field	Benevolence
		Created timeline to show changes over time (Table 1)Organized findings temporally as well to fit research questionUse of two researchers at each interview, multiple coders and agreement and discuss validity, and discussion of how they avoided various biases (integrity with epistemological assumptions regarding rigor)	Integrity
12. Reay, Golden-Biddle, & Germann (2006), “Legitimizing a New Role…”	“How do individual actors institute changes in established ways of working?” (p. 977)	Provided rationale for why using qualitative methodsDetailed description of data sources and analyses in text and detailed tableCited relevant and legitimate methodological sourcesDiscussed how they improved dependability of the data and reduced biasesDiscussed how they approached theory	Competence
		Detailed discussion of context and occupations in the study; in addition to detailed Table about occupationsLong time in the field	Benevolence
		Because of their emphasis on the role of individual actors in institutional changes, organized data around the microprocesses underlying this changeAlso used relatively large blocks of data and even excerpts of conversations to illustrate these microprocesses	Integrity

With the help of the 12 articles, we were able to expand on our initial formulation of methodological bricolage in several ways. First, there is a wide variety of moves that authors drew on in conducting and writing their articles. These moves were borrowed from a variety of methodologies, including grounded theory, ethnography, narrative analysis, story deconstruction, case studies, and more. This is not an exhaustive list of possible moves but simply the moves found in 12 articles chosen for potential diversity of moves. Given the sheer volume of moves evident in just 12 articles, we are confident that inductive researchers have access to a rich and plentiful trove of “resources” to borrow from when conducting methodological bricolage.

Second, just as bricoleurs can work with the same materials in different ways, so too did the authors in these 12 studies, who used many of the same (or similar) analytic moves in different ways. For example, several researchers discussed their extended time in the field (often several months or even years). The purpose of this immersion differed across articles. To illustrate, Locke (1996) explained in great detail the nature and length of her contact with people in the hospital she studied. In doing so, she signaled benevolence: Her extended experience in the field was necessary for building a grounded theory of playfulness in providing medical care. Others used the move of extended time in the field to express integrity. Kaplan and Orlikowski (2013), for example, justified their intensive observations by stating it follows from taking a practice theory perspective in their study. Still others used time in the field to signal competence. Bechky and Okhuysen (2011), for example, argued that their prolonged observations (and for Bechky, participation) in their contexts was essential for judging their competence in comparing two very different contexts: movie grips and SWAT teams.

Similarly, the same move used to express the same facet of trustworthiness could nonetheless be enacted in different ways. For example, every article, to some degree, showed competence by explaining how the authors analyzed their data. However, some did this only in the text while others created either simple tables (Reay, Golden-Biddle & Germann, 2006) or elaborate figures (Kaplan & Orlikowski, 2013) that walk the reader through the various steps of the analytic process. Gioia and colleagues (1994), for example, used Exhibit 1 to talk through their informant and analytical codes and how they related to their broader theoretical categories. Kaplan and Orlikowski (2013) used their figure to walk the reader through five rounds of coding and denote where in the process they iterated with existing theory (and what that theory was).

Third, as we have argued, effective arrangements involve utilizing moves in ways that communicate competence, integrity, and benevolence. Indeed, we found evidence of moves that support each of these aspects of trustworthiness between the readers and the authors in each of the articles. Although it might appear that some aspects are more important than others because there are more moves associated with them, we caution against such an interpretation. The array of moves necessary to support competence, for example, will depend on what you are doing in the article. If you are introducing an idea that is counterintuitive or runs against what is commonly accepted either theoretically or methodologically in the field of your contribution, you will probably need a stronger competence claim. This may require more competence moves, but a stronger claim is not necessarily represented by additional moves. Barley’s (1986) competence claim, for instance, consisted of one move repeatedly executed—justifying his moves based on the theory he was drawing on.

Fourth, conveying trustworthiness was not confined to choices made in data analysis. As noted in Table 2, methodological bricolage was evident in the research design, especially with the articulation of sampling. For example, Eisenhardt’s (1989) use of replication logic is critical for signaling competence from a positivist, case study perspective. Boje’s (1995) sampling of stories according to era (premodern, modern, and postmodern), by contrast, was appropriate for his research question and his discourse analysis. Similarly, data display was often used to signal integrity. For example, the temporal organization of findings in Dutton and Dukerich (1991) showed coherence between their research question—how interpretation influenced action—and how they addressed that question.

To close, we argue that methodological choices, including research design, data collection, analysis, and presentation of findings need to answer a research question in a way that creates an effective arrangement. Explaining to the reader the trustworthiness of the arrangement is intrinsic to its effectiveness. Authors achieve this by asking themselves several questions: What kind of competences do I need to show the reader, and how do I show these competences? Do my methods and arguments have internal consistency, and how do I show the reader this integrity? What kind of benevolence is important to the kind of study I have, and how do I show it?

Benefits and Limitations of Methodological Bricolage

Methodological bricolage yields several important benefits. First, the very process of using methodological bricolage offers a deeper connection to data. It disciplines researchers to think about the various choices one has to make to meet the challenges inherent in a particular study. By focusing on selecting each move to meet these challenges, methodological bricolage spotlights action, pushing researchers to make their own choices and actions transparent in their methods. Instead of hiding behind the template, researchers’ decisions get front-staged so that a reader can evaluate them.

Second, engaging in methodological bricolage allows researchers analytic solutions that are both specific to their research and recognizable and acceptable to evaluators of this research. To use a clothing metaphor, it allows researchers the advantage of both a “custom fit” (rather than a “one-size-fits-all” template solution) and the use of well-recognized and acceptable “materials”—as well as occasional new “materials.” Trustworthiness as a principle serves as a guide to choosing the “materials.” Putting together methodological moves that establish your competence as a researcher, the integrity of your study, and the benevolence of your engagement with the data helps you to create the “custom fit” appropriate to your study.

Third, methodological bricolage builds off the pluralism that has been a hallmark of qualitative methods and enables the potential for new methodological innovations to emerge that can ultimately benefit both the researcher and the field. With regard to methodological pluralism, embracing the wide variety of tools or resources to draw on in methodological bricolage helps remind us that qualitative methods are not a single method or a small set—for example, they are not limited to case studies, grounded theory, and ethnography (Cunliffe, 2011; Easterby-Smith, et al., 2008; Langley & Abdallah, 2011; Prasad, 2005). By proposing methodological bricolage, we move beyond philosophical conversations regarding how these different methods relate to each other. We suggest a way to better draw on this methodological diversity in the face of the recent pressures that have led to the increased use of templates that could undermine this methodological diversity.

With regard to methodological innovation, by not following a template, researchers must make study-specific choices that might lead to the discovery of new ways of approaching their methods. These new ways may be new combinations or entirely new moves. For example, a researcher might find a better way of visually representing data. Methodological bricolage allows researchers to develop a wider repertoire of qualitative research skills, thereby enhancing their ability to address a greater variety of future research questions. The idea is that instead of learning “the way” to do qualitative research, a scholar using methodological bricolage becomes adept at learning how to deploy a repertoire of moves and customize them for a specific research question. This should widen the potential research questions they could study. In contrast, the constant repetition of cognitive structures such as methodological templates can entrench us (for a summary, see Dane, 2010). As a field, we may fail to develop new methodological innovations because we remain too wedded to practices and too comfortable (cognitively and emotionally) in the way we approach our methods. Methodological bricolage provides a means of always learning and experimenting with new tools to understand and analyze qualitative data.

Fourth, if adopted widely, methodological bricolage may also benefit the field by pointing out areas where existing moves are lacking. For example, in constructing Table 2 (and in our roles as editors), it became clear to us that there were not a lot of moves that illuminate how one goes from abstract codes to full-fledged theoretical models. Even when authors use a data structure table popularized by the Gioia method—a move to help authors delineate their “chain of evidence”—it is not always clear how discrete codes became combined in particular ways in building new theoretical insights. Thus, as researchers face different challenges, it may become clear where resources for bricolage are scant and where they are more plentiful. Again, this opens the opportunity for methodological innovation.

There are, of course, some potential limitations to a bricolage approach. First, utilizing methodological bricolage will require more methodological sophistication from reviewers and editors. Having had to assign reviewers for many manuscripts ourselves, we personally know this creates challenges, especially for topics that have an underrepresentation of qualitative researchers. Moreover, methodological competence in reviewers and editors are needed because some moves may be relatively easy to identify (e.g., using citations to denote competency; see Table 2), whereas integrity and the coherence it creates requires judgment. Integrity is often achieved by a combination of moves or various forms of evidence that make sense only within the context of a given manuscript. And not just authors but also editors and reviewers should be aware of disciplinary differences in what may be deemed acceptable or unacceptable combinations of moves. Holding a management ethnographer responsible for producing moves unique to ethnographers in anthropology, for example, is problematic.

Second, such an approach could lead to overly long methods sections. Although we suggest that examples of each type of move be present in a methods section, researchers who want to change the status quo might focus especially on competency moves, those that want to build from the insights of informants may have more emphasis on benevolence moves, and in studies that integrate many “moving parts” (e.g., multiple methods), authors may wish to focus on integrity moves. Of course, manuscripts may have each of these challenges to some degree. Thus, we recognize that a bricolage approach may entail somewhat longer methods sections than a template approach does, but we think the payoff in higher quality would make this worthwhile.

Third, a bricolage approach can be subverted, just as the use of templates have been. To be clear, we are not advocating a “kitchen sink” approach whereby authors throw in as many moves as they can in one manuscript in the hopes that it will please reviewers. Ironically, following such an approach would lead to the same type of mindless research that plagues the use of templates and motivated us to write this paper in the first place! As we have noted, methodological bricolage should be done mindfully to meet the specific challenges of a manuscript. Furthermore, as authors add additional moves, satisfying the integrity component of trustworthiness could get very difficult, providing a built-in check for authors tempted to add extraneous moves into their papers.