From the Editors: for the Lack of a Boilerplate: Tips on Writing Up (and Reviewing) Qualitative Research

Academy of Management JournalVol. 52, No. 5 From the EditorsFree AccessFrom the Editors: For the Lack of a Boilerplate: Tips on Writing Up (and Reviewing) Qualitative ResearchMichael G. PrattMichael G. PrattBoston CollegePublished Online:30 Nov 2017https://doi.org/10.5465/amj.2009.44632557AboutSectionsPDF/EPUB ToolsDownload CitationsAdd to favoritesTrack Citations ShareShare onFacebookTwitterLinkedInRedditEmail For want of a nail, a shoe was lost For want of a shoe, a horse was lost For want of a horse, a rider was lost For want of a rider, a battle was lost For want of a battle, a kingdom was lostQualitative research is only one of the methods that are appropriate for our journal, but over the past several years we at AMJ have worked diligently to increase the number and quality of the qualitative research papers we review and publish. Just this year, one of our qualitative papers won the award “Best Paper in Organizational Behavior” from the OB Division of the Academy of Management (Margolis & Molinsky, 2008). Our efforts to increase high-quality qualitative work in the Journal stems, in part, from our mission to publish research that has the highest impact. Qualitative research certainly fits this bill, as work in this area has won multiple best paper awards in AMJ and Administrative Science Quarterly, and qualitative research was overrepresented (in terms of the total number of studies published) in AMJ's survey regarding the most interesting management-related articles published in the past 100 years (Bartunek, Rynes, & Ireland, 2006).Qualitative research is great for addressing “how” questions—rather than “how many”; for understanding the world from the perspective of those studied (i.e., informants); and for examining and articulating processes. Just as quantitative research encompasses many ways to collect data, such as survey and lab studies, and many ways to analyze it, such as ANOVAs, multiple regression, and factor analyses, so does qualitative research. For example, one may choose to collect data using a case or an ethnographic method; however, when analyzing these data, one can continue to employ case (Yin, 2003) or ethnographic methods (Spradley, 1979), respectively, or even employ grounded theory (e.g., Eisenhardt, 1989; Pratt, 2000). Moreover, qualitative research can be either inductive or deductive (see Bitektine [2008] for a discussion of deductive designs) or, in very rare circumstances, a combination of both (see Ross and Staw [1993] for an example). Finally, it is possible to analyze qualitative data quantitatively, just as we analyze quantitative data qualitatively when constructing stories around the numbers we present.For the purpose of this editorial, I follow Gephart in discussing qualitative research that includes both qualitative data and qualitative analysis: “Qualitative research starts from and returns to words, talk, and texts as meaningful representations of concepts” (2004: 455). Moreover, the focus of this editorial is on one type of qualitative research—inductive qualitative research— and the challenges that face the qualitative researcher, and the qualitative reviewer, during the review process for a top-tier journal. More specifically, I hone in on what appears to be at the heart of these challenges: that there is no accepted “boilerplate” for writing up qualitative methods and determining quality.A “boilerplate” refers to standardized language, and here also refers to an accepted template for writing up qualitative research. Unlike quantitative findings, qualitative findings lack an agreed-upon “significance level.” There is no “magic number” of interviews or observations that should be conducted in a qualitative research project. What is “enough” depends on what question a researcher seeks to answer. To illustrate, if a researchers wanted to study Supreme Court justices' decision making, he or she would be limited to a very small sample. However, to examine how three cohorts of physicians changed their identities over the life of their residency programs, my colleagues and I had to conduct well over a 100 interviews (Pratt, Rockmann, & Kaufmann, 2006).This lack of a boilerplate begets other challenges for publishing qualitative research. One could rewrite this editorial's epigraph to read:For the lack of a boilerplate, there was little direction Because there was little direction, the author was lost Because the author was lost (and the reviewer went along) the contribution was missed Because the contribution was missed, perceived impact was low Because perceived impact was low, the paper was rejected.But rewriting the epigraph in this way would make one assume that the obvious way to facilitate the publication of qualitative research would be to agree upon a set of standards for its evaluation. However, given the diversity of methods—and the range of epistemological and ontological assumptions underlying these methods (Morgan & Smircich, 1980)—constructing a single boilerplate is not something qualitative researchers strive to achieve. In fact, Tierney (1995) argued forcefully for the use of experimentation in qualitative research. Many would agree that the creative nature of qualitative research is one of its key strengths.Although AMJ is open to a variety of qualitative methodologies (and, to accommodate such methodologies, is even is flexible, within reason, on qualitative papers' page lengths), this “equifinality” can make it extremely difficult to both write and evaluate qualitative research. So how does one encourage experimentation and creativity in the craft of qualitative research, while also providing some guidance to authors (and reviewers) about writing and assessing qualitative research? It is this delicate balance that I wish to navigate in the service of helping authors write a compelling and focused account that (1) honors the worldview of informants, (2) provides sufficient evidence for claims, and (3) significantly contributes to extant theory—all within the confines of a journal article. I have argued elsewhere that just because there are many paths to good qualitative research, this does not mean that that all paths are good ones, especially when publishing in a top-tier journal is sought (Pratt, 2008). Thus, I humbly attempt to provide some illustrations of both dangerous and good paths to tread. As a secondary purpose, it is my hope that this “From the Editors” may also assist evaluators of qualitative research.My experience at AMJ, and the accumulated wisdom of others, suggest some dangerous paths to follow—paths that will limit an author's ability to publish her or his qualitative research.Wandering Down Dangerous PathsThere are two major perilous paths: (1) lack of balance between theory and data and (2) making qualitative research appear quantitative. Along these paths are multiple subpaths. For example, with regard to balance, here are the two main branches to avoid:Telling about data, not showing it. I find this problem to be the most common in multiple case studies. Rather than show any raw data, authors give only their interpretation of the data. Golden-Biddle and Locke (2007) referred to this phenomenon as too much “telling” and not enough “showing” (also see Lofland and Lofland [1995] on “analytic vs. descriptive excess”). “Too much telling” is problematic because no clear chain of evidence shows how the researchers moved from their data to their interpretations.Crafting elaborate tables to organize data, and relegating all your data to these tables, is a variation of this problem. Ideally, one should place at least some data within the body of the paper. Separating all of the data from one's arguments not only requires reviewers to engage in the tedious task of frequently moving from text to table, but also imposes on them the heavy burden of applying the data in just the place the author intended. As an aside, a lack of data in the body of a paper also makes it less interesting to read!Showing too much data, and not interpreting it. Though “thick description” has a venerable history (e.g., Geertz, 1973), and though it can ultimately contribute to theory, limiting an analysis to simply describing what one found is not likely to be enough to achieve publication in a top-tier management journal. Journals like AMJ require the articulation of a significant theoretical contribution as well. Thus, one can err by showing too much data, with too little interpretation. Simply put, you cannot skimp on theory. It is critical that scholars communicate what theoretical conversations they want to enter, and what the current states of those conversations are—otherwise, they are likely to “reinvent the wheel.”Several strategies can lead to being more descriptive. For example, organizing findings around research questions often leads to the use of quotes and other data to “answer” research questions, but often fails to go much beyond those “answers” to discuss theoretical contributions. Crafting typologies may also lead to being overly descriptive in certain circumstances. If you are simply sorting themes in your data, without explaining how this classification scheme leads to new theory or new theoretical insights, you are being too descriptive.The next three subpaths involve the specific issue of trying to make qualitative data appear, or actually be, more quantitative. Doing so may involve some rather superficial but nonetheless problematic tactics, such as using rhetorical tactics to make one's study sound more quantitative. Others tactics may involve the quantification of a small sample of data. The most fatal tactic in this regard is inappropriately mixing quantitative elements into one's qualitative study design.Using deductive “short hand.” Using rhetorical tactics to make qualitative research seem more quantitative is a common way of making qualitative research more palatable to nonqualitative reviewers. Golden-Biddle and Locke (2007), for example, discussed how qualitative articles will often mimic the structure of quantitative articles by labeling sections as introduction, methods, findings, and conclusion. (Note: only use this tactic if it fits the logic of your analysis.) However, one can go too far in this direction. In a paper I recently reviewed, the author (who gave me permission to use this quote) described his/her maximum variation sampling in the following way: “I wanted to control for other variance to eliminate other possible confounds.” Someone who does deductive work might easily understand this statement, yet it might also trigger frames for evaluating the paper that were not intended. A deductive frame was not appropriate here, as the author was conducting an inductive narrative analysis.Quantifying qualitative data. Although perhaps not problematic in archival data analysis or when a scholar has a large number of “units” to analyze, quantifying the data does not serve most small-sample qualitative studies well. There are a few reasons for this: (1) it may trigger a quantitative/deductive mind-set among reviewers; (2) it may be misleading (e.g., small changes in responses corresponding to large changes in percentage counts); (3) it may overlook “taken-for-granted meanings”; (4) it may do “violence to experience,” inadequately representing the voices of the individuals studied; and (5) it may simply create the “worst of all worlds”: not enough of a sample for a statistical test, and too anemic a representation to adequately represent rich data (Pratt, 2008).Inappropriately mixing inductive and deductive strategies. Sometimes, rather than simply describe what they did in deductive terms (as in number 3 above), researchers will actually incorporate deductive elements into their inductive study design. By way of contrast, I am not talking about doing mixed-methods research (see Creswell, 2003); I am referring to situations in which inductive and deductive elements of a study are combined inappropriately. For example, some researchers will engage in random sampling rather than theoretical or purposeful sampling when attempting to build theory. Some will exclusively use a theory or two to provide codes for their data—which is problematic if they are using grounded theory techniques. When authors go down this path, which goes beyond how the qualitative research is written up, the problems that arise are extremely difficult to “fix.”Finding Better PathsWhat then, would I suggest for researchers who are submitting qualitative research? And what do I think reviewers of qualitative research should be looking for? Below are some ideas that have helped me. I offer three disclaimers, however: First, one need not incorporate all these ideas in any single study or manuscript. In the spirit of equifinality, note that various subpaths can take you to the same place, but in different ways. Second, not everyone will endorse each of these ideas. For example, I received feedback on this editorial from well-known qualitative researchers who varied in their career stages and their approaches toward qualitative methods. I found the “including the basics” suggestion to be relatively uncontroversial. However, the use of organizing figures was a bit more polarizing; all agreed that they could be useful, but not all agreed that figures were usually done well or were always needed. Third, though I endorse the ideas below, I cannot claim ownership of them. In that I have the privilege of working with wonderful colleagues and reading some fantastic qualitative research, these ideas are certainly not unique to me:Make sure your methods section includes “the basics.” In my experience as an associate editor with AMJ, I have found that authors and reviewers are often not sure what should go into a qualitative methods section. Moreover, I find a fair number of authors devote a lot of space to things that are not that critical (e.g., a running commentary on every element of data collection), but then overlook what might be considered basic elements of a methods section. In reviewing a number of award-winning qualitative papers, I found that each one mentioned the following, although necessarily in any order (see Pratt, 2008): Discuss why this research is needed. This account may go in the methods section, or it may be more explicitly set up in the theoretical review. Essentially, you need to explain what is motivating your study and why your methods are appropriate. For inductive studies, articulating one's motivation not only involves reviewing the literature to illustrate some “gap” in prior research, but also explaining why it is important to fill this gap. The latter is often forgotten. Simply “doing what no one else has done” is not sufficient. To my knowledge, no one has studied leaders' sock preferences, but it isn't clear why anyone should. Rationales are necessary. Remember: What might be compelling or obvious to you may not be compelling or obvious to your audience.Are you building new theory or elaborating existing theory? Most inductive pieces are meant to either build or elaborate theory. The latter means that existing theory in the area exists but gaps or oversights need to be filled in (Lee, Mitchell, & Sablynski, 1999; Vaughn, 1992). For example, in my study with Rockmann and Kaufmann, we argued that the careers and socialization literatures implied that identities change during socialization, but that the process of identity change was not well specified (see also Ibarra, 1999). Thus, we needed to fill in this area of theory. For building theory (especially using grounded theory), I would refer the reader to Suddaby's (2006) excellent “From the Editors.”Why did you choose this context and this “unit of analysis?” Qualitative methods paper writers should explain the nature of the context they are examining. Often this is done in a separate subsection placed either within or right before a methods section. Authors should also justify their context from a sampling perspective—for example, answering the question, Is it a prototypical or an extreme case? In my own research, Amway was an extreme example (Pettigrew, 1990) of an organization that engendered very different types of attachment (Pratt, 2000), which made it ideal for building theory on identification management.Similarly, qualitative authors should discuss whether they are sampling people, events, cases, and the like, and why they are being sampled. Strauss and Corbin (1998), Patton (1990), and others have discussed the various sampling strategies useful for inductive qualitative research. For example, purposeful and theoretical sampling are often used when building grounded theory. What is more, when employing these sampling strategies, one's criteria for sampling may change as a study progresses—and that is not only legitimate, but expected! However, random and convenience sampling, especially if one is building grounded theory, is often difficult to justify. Similarly, I have found that individuals often misuse or fail to properly justify the use of snowball sampling.How did I get from my data to my findings? Although specific qualitative methodologies talk about them differently, most articulate a basic set of steps that one goes through in an inductive analysis. First, find out what informants say (first-order codes, provisional codes). In explaining how you arrived at what informants thought or believed, be sure to explain what data you drew upon (e.g., observations, documents, interviews, etc.). If you conducted an interview study, it is imperative that you include your interview questions or protocol in an appendix. This is critical so that readers can determine the degree to which your findings are directly linked to the questions you asked and how you asked them. Second, find out what the literature says (e.g., enfold theory, second-order codes, axial coding). Be clear in your methods as to how and in what way data were used in the analysis of your data. Finally, tell your story about how it all fits together. How you tell your story can vary (Van Maanen, 1998); but in telling it, make sure your chain of evidence is clear. The goal is to be clear about what you did so that someone else can evaluate the veracity of your methods. Some individuals confuse striving for such transparency with the goal of replication (which is not even done often in the hard sciences [Collins, 1982]).As an aside, I should point out that having someone else code your data does not necessarily make it valid. If I were coding archival data or data that I did not collect, then I would certainly expect to use multiple coders and show some measure of interrater reliability. However, at the other extreme, if I were engaged in an extended ethnography, expecting someone else to adequately code my data—when he or she knows nothing about the context or the individuals involved—makes little sense. Part of doing ethnography is gaining rich experiences over an extended period of time. This can and should change how you view the data you collect.To this list, I would add one other point that is not often well articulated in manuscripts. One should be very clear about one's “position in the field”: the relationship between the researcher and the researched (Anteby, 2008). For example, was participation involved? If so, in what form? (As an aside, I find that individuals often misunderstand what it means to be an actual participant observer; see Gold [1957/1958] for details). Did you know any of the informants beforehand? Were you ever an employee of the organization you observed? These details are important, as they inform the reader about how you approached your study.Show data—in a smart fashion. To be honest, very few, if any, of the papers that come across my desk have their theory and data well-aligned in the first round(s) of reviewing. Moreover, very few successfully show the contributions flowing from their research. What then prevents some papers from being rejected? One important answer is showing your data!Showing data is critical for assessing whether successful theorizing is plausible. Lofland and Lofland argued that there are “no precise rules regarding balance [of theory and data], only order-of-magnitude guidelines” (1995: 165), yet they did suggest erring slightly more on the showing data side. Especially for early drafts, this is good counsel. Ample data allow editors and reviewers to make some wonderful suggestions to authors about how to craft their theoretical stories. For example, it was an AMJ reviewer who metaphorically “hit me up the side of the head” in a paper that I ultimately published with Kevin Rockmann and Jeffrey Kaufmann on identity work (Pratt et al., 2006). Our frame did not fit the data, and the reviewer pointed out the appropriate frame to use instead. He or she could not have done that if we had not provided enough data for the “pattern” in the data to be clear.I have suggested elsewhere that it may help to have data both in the body of your paper and in tables. (Note: these should be different data; don't put the same content in body and tables.) These data may be in the form of “power quotes” and “proof quotes” (Pratt, 2008). Power quotes are the most compelling bits of data you have, the ones that effectively illustrate your points. These should be in the body of your paper. However, you should also put some additional data in tables. Ideally, you can provide multiple quotes for each point or argument you are making so that you have some “proof” of what you are saying. Although the exact number of quotes needed will vary with the size of your sample, these tables should be as complete and exhaustive as possible. Moreover, you should recognize that these proof quotes only bolster points you have already made in the body of the paper; readers should be able to understand your main arguments without referring to the tables.Think about using organizing figures. Using figures to organize your thinking, even if you do not show them in the final submission, is an effective way to help clarify your thinking. These figures may be used for different parts of your paper. I have seen and have used figures to depict how a methodological process unfolded. This is an especially handy device for depicting more complicated analyses (see Pratt & Rosa, 2003: 397). Figures are also a very good way of capturing your chain of evidence (see Corley & Gioia, 2004: 184). When done well, a figure can show visually how you moved from raw data (e.g., interview data) to the theoretical labels or constructs you are using to represent that data.In addition, a lesson I learned from Anat Rafaeli is to construct figures that help you visually represent your findings. I find figures particularly good for depicting processes. Such figures are often used to “summarize” findings (e.g., Pratt, et al, 2006). However, they can be as or even more effective at the start of a findings section. Here, a figure allows you to “walk through” findings with your readers (Pratt & Rafaeli, 1997).Finally, with regards to showing data and creating figures: Make sure that you integrate tables and figures into the text. Don't create them and then forget to refer to them in the body of the paper. Also, make sure you explain your figures in sufficient detail. Be especially careful about using traditional “boxes and arrows”—which are often used in variance models—when depicting processes. If you are using boxes and arrows to tell a process story, make sure that this is clear to the reader.Think about telling a story. A narrative way to capture what a figure does to organize your findings is to create a coherent story in which you not only describe themes, but how those themes fit together (see Spradley, 1979). A piece of advice given to me was to think of each theme as a character in a story. Who is the central character or protagonist? What obstacles does the protagonist faces? What does the protagonist hope to accomplish? This exercise helps a writer distinguish “figure” from “ground.” Just as a literary story has a focal character, so too should a qualitative story have a focus around which the rest of the content revolves. For example, in the physician study I have referenced, there were issues of professional identity, identity change, skill competence, daily tasks, and organizational artifacts. These various “characters” did not cohere into a story until we decided to make professional identity change (identity customization) the protagonist/figure. Once this choice was made, we could identify daily tasks as challenges to our informants' professional identity and thus a trigger for change; organizational artifacts, as resources that helped identity construction; and competence, as something that developed along with identity changes.A common mistake in qualitative writing is to have too many protagonists, each demanding ample “screen time.” The net result is that what (or who) your story is about is not clear—which makes identifying a theoretical conversation to enter and add to a difficult task.Consider “modeling” someone whose style you like who consistently publishes qualitative work. The body of qualitative research published in top-tier journals grows every year. As a result, authors have more and more “provisional selves” (Ibarra, 1999) to “try on.” You might want to find an author or two (or three…) who consistently publish in the journals in which you seek to publish and try on their styles. (By consistently publishing, I mean appearing in multiple outlets and/or publishing more than once in the same outlet.) Such modeling is helpful because these authors have likely gotten better and better at overcoming the “boilerplate hurdle.” Style here refers to how these authors write up their material. Some authors clearly divide their theory from their data (e.g., in first- vs. second-order findings). Others intermix theory with data. Some authors write very short introductions and go almost immediately into their methods and data, while others erect more elaborate theoretical frames before describing how these frames are to be filled in.As a beginning qualitative writer, I found imitating others to be essential in the learning process. I still find it helpful today. By mimicking how other authors constructed qualitative papers, I began to pick up practices that helped me do it successfully. Eventually, I was able to develop my own voice. As a note of caution, however, recall that not only do authors have voices—journals do too. Thus, be sure to read through the types of qualitative research that a specific journal publishes.I want to close with three brief points. First, I want to acknowledge and thank Michel Anteby, Peter Bamberger, Beth Bechke, Jason Colquitt, Duane Ireland, Bob Gephart, Karen Golden-Biddle, Karen Locke, Leslie Perlow, Anat Rafaeli, and Roy Suddaby for their comments on a draft of this editorial. Second, it is important to note what I have written here covers at best a small portion of what takes place when one publishes a qualitative paper. The points above are more about making valid and reliable claims than about working with editors and reviewers in the “discovery” process. The most enjoyable and exciting part of being an editor is to be able to work together with reviewers and author(s) to help the author(s) to actualize a paper's potential. Third, I want to reiterate a key point of this “From the Editors”: because of the equifinality of writing qualitative research, the lack of a boilerplate need not mean that the “kingdom is lost.” What is lost in structure is gained in the ability to be creative. Thus, one might replace a nail with superglue or other bonding agent. However, whatever you use, the burden is on you to make sure the adhesive adheres (i.e., your story coheres) and the shoe holds tight.REFERENCES Anteby M. 2008. Moral gray zones. Princeton, NJ: Princeton University Press. Google Scholar Bartunek J. , Rynes S. , Ireland R. D. What makes management research interesting, and why does it matter? Academy of Management Journal, 49: 9–15. Google Scholar Bitektine A. 2008. Prospective case design: Qualitative method for deductive theory testing. Organizational Research Methods, 11: 160–180. Google Scholar Collins H. M. 1982. The replication of experiments in physics. In Barnes B.Edge D. (Eds.), Science in context: 94–116. Cambridge, MA: MIT Press. Google Scholar Corley K. , Gioia D. 2004. Identity ambiguity and change in the wake of a corporate spin-off. Administrative Science Quarterly, 49: 173–208. Google Scholar Creswell J. 2003. Mixed methods procedures. Research design: Qualitative, quantitative, and mixed methods approaches (2nd ed.). Thousand Oaks, CA: Sage. Google Scholar Eisenhardt K. 1989. Building theories from case research. Academy of Management Review, 14: 532–550.Abstract , Google Scholar Geertz C. 1973. The interpretation of culture: Selected essays. New York: Basic Books. Google Scholar Gephardt R. 2004. What is qualitative research and why is it important? Academy of Management Journal, 7: 454–462. Google Scholar Gold R. L. 1957/1958. Roles in sociological field observations. Social Forces, 36: 217–223. Google Scholar Golden-Biddle K. , Locke K. 2007. Composing qualitative research (2nd ed.). Thousand Oaks, CA: Sage. Google Scholar Ibarra H. 1999. Provisional selves: Experimenting with image and identity in professional adaptation. Administrative Science Quarterly, 44: 764–791. Google Scholar Lee T. , Mitchell T. , Sablynski C. 1999. Qualitative research in organizational and vocational behavior. Journal of Vocational Behavior, 55: 161–187. Google Scholar Lofland J. , Lofland L. 1995. Analyzing social settings: A guide to qualitative observation and analysis (3rd ed.). Boston: Wadsworth. Google Scholar Margolis J. D. , Molinsky A. 2008. Navigating the bind of necessary ev