Researchers regularly deal with ethical considerations when writing up their research. However, such considerations are rarely addressed in existing research ethics literature. Here, the focus tends to be on ethical considerations that arise in relation to research proposals and “protocols,” that is, during the process of research and data collection. Nevertheless, ethical issues can and do arise when formulating research questions, preparing data for analysis, conducting analysis, and when writing up one’s findings. Using decision-making junctures from a study on discourses of higher education in Sri Lanka, I aim to show that ethical considerations can shape the architecture of writing research. I present a number of specific questions that arose in the presentation of qualitative data, derived from the use of ethnographic methods and discourse analysis. I argue that the overarching consideration here should be the balance between content and confidentiality or, to put it another way, the “amount” of context necessary to give the appropriate meaning to the content being presented. I suggest that this should be considered a major ethical consideration in qualitative research. Adopting this view entails the consideration of a number of specific questions, including the following: How can the analysis be presented when the data are “sensitive?” Which examples should or can be chosen? And how can contextual descriptions be crafted without endangering participants’ anonymity? Drawing on the writing stage of my doctoral research, as well as later presentation and rewriting of the same data, I present a series of narrative examples and use them to explore the ethical dilemmas that emerged. I suggest that attending to these decision-making junctures that arise in the course of writing up qualitative research should be considered useful moments of reflection for researchers using ethnographic methods and discourse analysis.
By the end of this case, students should be able to
- Understand ethical dilemmas that arise when writing and disseminating research
- List methods that can be used in the pre-writing phase that help to resolve ethical quandaries during the writing stage
- Identify strategies that may help resolve these ethical dilemmas while writing your research
Project Overview and Context
In this case, I discuss the ethical difficulties that arise during the writing phase of my research and how I resolved these dilemmas. The case study is based on my doctoral study as well as two other writing events.
My doctoral research was conducted in post-war Sri Lanka. The civil war in Sri Lanka ended in 2009 and, almost immediately after, academics began a campaign for higher salaries. By 2012, when I started collecting data for my dissertation, this campaign had morphed into a struggle for more equitable policies in higher education. The country was yet to undergo any process of demilitarization, and academics felt that they were under constant surveillance by the state.
At this point, I was enrolled in a doctoral program in the United States and was excited to research events that Sri Lankan universities were experiencing as part of my doctoral work. I collected data on the academic trade union struggle, their campaign against privatizing universities, and militarism in universities. For the dissertation itself, I focused on the militarization of universities, a subtle but evident process from my perspective. I made use of a discourse analytic framework to examine how academics perceive militarization (or lack of it) in relation to their work environments.
I visited Sri Lanka from 2012 to 2014 for a period of 3 months each year. I spent the time conducting interviews, making observations, and attending events that were significant for my research. The first year of data collection, 2012, coincided with the height of the protest campaign by academics. I faced some practical complications, two of which are detailed below:
- As a researcher, I was not continuously present in the “field”: in this case, I visited Sri Lanka for 3-month periods every year. However, U.S. doctoral programs do not cohere with the need for continuous presence at the site, which meant that I had to follow some significant events from afar.
- As someone who was a member of the researched community, I found it difficult to share written drafts of the work: as a Sri Lankan academic, I was a part of a community that would be able to identify participants I wrote about. This limited the peer comments I could request from my colleagues in Sri Lanka.
My doctoral research was a qualitative study using ethnographic methods and discourse analysis. For this project, I interviewed more than 20 academics over a period of 3 years, and some of them were interviewed multiple times. I attended public speeches and pickets arranged by various academic unions and wrote field notes on those events. During these 4 years, I also collected material related to higher education in Sri Lanka which included newspaper articles, posters, informative flyers, blog posts, and video updates of union work, all of which resulted in a large amount of data. Due to methodological and ethical considerations, some of which are discussed below, I decided to focus only on interviews in the dissertation.
Ethical Dilemmas and Resolutions
The ethical dilemmas presented in this section are from the writing stage of the dissertation and subsequent texts (a conference paper and a manuscript for a journal) that used the same data.
Deciding What to Write and How to Write
Some of the most difficult decisions to be made during the doctoral research process were “What do I write about? What data do I use? What form should my text take?” Unlike ethical dilemmas that arose during data collection (e.g., being put on the spot during an interview), these problems did not necessitate decision-making at that moment itself. These were dilemmas that one could ruminate over, but were nevertheless quandaries.
The most urgent issue I faced when contemplating writing was the amount of data I had amassed (as described in the research design section). This is a common consideration for most researchers using similar methods. One reason to use interview transcripts as primary data was methodological—the literature review exposed a methodological gap that discourse analysis was not commonly used in militarization studies. Rather than an analysis of large-scale data or data using public sources, which reduces the need for contextualizing individual stories (see Van Dijk, 1989), I was drawn to the close analysis of talk—How did academics talk about militarization? How did they define it? My interest in conducting a close analysis of transcripts limited the amount of data I could include and, therefore, present in my writing.
The second reason for using interviews, but limiting the scope, was ethical in nature. During the dissertation proposal defense, my Committee came up with some solutions and cautions:
- I may want to draw out some data more than others, but at the same time the data must not appear to be “cherry picked.”
- I should keep in mind that the data are already “selected” in a sense, as I chose the participants, and they chose to speak to me and decided what to say to me.
The Committee therefore stressed the need for a systematic or theoretically informed method of selection and analysis that would subsequently help me choose illustrative excerpts in an organized manner. A possible solution was to do a “regimes of discourse” or a mapping of the discourses available on militarization. This would take away the need to focus on participants, thereby diminishing the need to refer to or portray individuals. Another solution was to use “aggregated individuals” as illustrative cases. This would mean the presentation of a merged “type” made up of several similar participants (or discussions).
In the end, my resolution was to use all the relevant interviews for analysis, but to use a limited number of individual interviews as exemplars. Given the lack of close analysis of discourse in militarization studies, I presented the excerpts as cameos. For instance, in Chapter 5 (Figure 1) of my thesis (the second chapter presenting analysis), I discussed discursive changes in the interviews of participants over a period of 2 to 3 years.
Figure 1. Structure of Chapter 5.
For each of the sections, I chose excerpts from two interviews. In doing so, I chose not only what would illustrate the analysis presented in the chapter but also excerpts that had less identifiable characteristics than others. In Chapter 4 (the first chapter presenting analysis), I used “representative segments” from seven participants. By using representative examples rather than excerpts drawn from all interviews, I limited the contextual information I needed to give about the participants.
Deciding What Context Means
At the comprehensive examination, the members of the doctoral committee raised a concern that there was not “enough context” to a preliminary analysis I had presented to them. Ethnographic writing usually provides contextually rich accounts. In my case, sufficient contextual framing of the excerpt was necessary if the reader was to understand and assess my detailed discourse analysis of the interviews I had conducted and the excerpts I provided (see Blommaert, 2004; Schegloff, 1992 on context). Yet, the more contextual information provided, the easier it would be to identify the participants discussed in the text. Much of the discussion at the exam centered on this ethical quandary: How do I provide “enough” but not “too much” context so that the reader is able to understand the analysis but not identify the participant? We all agreed that, given the sensitive nature of the issues discussed by participants, this was a significant problem.
Participant anonymity and confidentiality were addressed in two stages of the study. During transcription, I highlighted what I felt were identifiable information. These included references to the participants’ ethnic or religious affiliations, academic position or rank, the names of their own universities, names of friends or close colleagues, as well as references to place descriptions that could give away location.
The second was to provide biographical data about participants in separate spaces. To this end, disaggregated data on academic rank and ethnic and religious affiliations were provided typically in the Methodology chapter. Rather than providing a description of each participant’s biographical and professional characteristics (e.g., in a table), I provided ethnic and religious affiliations, academic rank, or gender separately. Below is such a paragraph from my dissertation:
I also refrain from giving specific information on ethnic and religious affiliations of the participants discussed in Chapters 4 and 5 for this reason [to maintain anonymity]. In terms of ethnicity, the majority of the 20 participants are Sinhalese, and 4 participants are from minority ethnic groups. The group also includes 6 participants who are Christian, Muslim, and Hindu. Six of the 20 participants are male academics. (p. 75)
In the analysis chapters, I provided biographical characteristics in two separate instances: as a composite description at the beginning and later when introducing the participants whose excerpts were being analyzed. For instance, Chapter 5 begins with a reference to the participants discussed in it: “The representative samples are drawn from participants who belong to the majority and minority ethnic and religious groups” (Perera 2015, p. 121). When introducing the four individuals whose interviews were used in the chapter, however, I refrained from mentioning their ethnic or religious affiliations and removed such identifiable data from the excerpts. An example of this is as follows:
Bhagya and Hirusha are mid-career academics in the Social Sciences and Humanities respectively, from two universities. Bhagya is an active member of her trade union, whereas Hirusha, though she has participated in major trade union events (e.g., protests), had not obtained trade union membership by Time 2. (pp. 122-123)
Later on, when analyzing Hirusha’s interview, I introduce her as “an academic who has extensive experience in the field of education” (p. 123). The sum total of contextual information provided about Hirusha here is her gender, disciplinary affiliation broadly, field of experience, stage of career, and some sense of her activism.
A difficulty when providing names for participants in the body of the chapters is that, especially for Muslim participants, their names may give away their religious identities. The dilemma is twofold because not using a Muslim name also results in overwriting their identity, possibly with another ethnic identity. In this instance, I chose to use names that were not immediately recognizable as being recognizably identified with the Muslim community. In terms of academic position, I consistently used the descriptor “mid-career academic” instead of academic rank, which indicates long-term work experience while reducing identifiable information.
I proceeded in a similar manner when presenting the conference paper on power relations in interviews. Here too, I grouped certain characteristics together, for example, “The four episodes I present are from interviews with academics from two different universities, affiliated to programs in the disciplines of Science and Humanities” (Perera, 2014). The conflation of such characteristics, while detracting from building an image of each profile, allows for the protection of their identities.
This quandary between the need to provide context while maintaining confidentiality also brings out the contradictions inherent in a qualitative study of this kind. The literature shows the need for locally specific, contextualized research, yet this is difficult to provide in situations where participants are discussing sensitive issues or providing information they consider risky. Leaving out rich contextual detail due to considerations of safety and confidentiality means that some of that sensitive context is necessarily taken out.
Deciding What and How to Disseminate
Ethical dilemmas also appear when deciding on dissemination. This appears counter-intuitive because as researchers we are also ethically obligated to disseminate our findings. However, my own experience of (a) preparing a paper for a conference presentation and (b) writing a manuscript to be submitted to a journal made me understand that these are additional decision-making junctures. A conference presentation creates different pressures to writing a paper. The conference presentation is limited in terms of time and location: it is shorter, given at a specific venue for only the audience sitting in front of you. It reaches a wider audience if published or disseminated later. In contrast, a published text is available typically to subscribers usually in academic institutions who have access to databases. It is kept in a repository and reaches its audiences over a longer time period. Given that my participants, and our colleagues, are academics, there is a probability that a research article on Sri Lankan issues will be accessed by them. Disseminating either of these in (open-access) digital fora ensures that my writing is made available to a wider audience over a longer time span. Although the audience of the conference presentation was probably less than 50 people, at the time of writing the written version of the conference paper has 214 views and seven reads on Academia.edu and 30 reads on ResearchGate. Because of these concerns, I have felt most under pressure while writing for publication and disseminating my work online.
I will briefly discuss a few of the precautions I took to safeguard against inadvertently revealing participants’ identities.
One precaution I considered was to rename the participants for each “output.” For the conference paper and manuscript, I used interviews that were part of the doctoral research. As the interpretive demands in each instance may result in different contexts being written for each text, using different aliases each time would ensure that readers would be unable to piece together a participant’s identity even if they read all such references. I used this method for the conference paper. However, I used the same pseudonym for the manuscript as it was a revised version of a section of the dissertation and used the same excerpts.
A second strategy I used was the member check, an accepted and encouraged strategy for protecting participant information. I used this method while writing the manuscript. I provided the participant with the final version, asking her to let me know “if you want less (or more) included” (personal correspondence, 2016). In addition, I invited her to comment on the analysis itself. In this instance, the participant did not suggest any changes, although in other instances member checks do result in changes (see Kinginger, 2004; McCormack, 2000).
However, for the conference paper, I was not keen to do a member check and therefore chose a different method for safeguarding participant identities. In the paper, I presented instances where the participants subtly but clearly intimidated the interviewer (myself). I contrasted two such interviews with two interviews that show rapport and cooperation between the interviewer and participants. As I was discussing two academics senior to me and illustrating the production of power within the interview (albeit unconsciously), I found it difficult to imagine presenting them with this analysis (see Ellis (2007) for difficulties with member check). Instead, I shared the paper with a trusted researcher who may have been able to identify the participants but was not an academic herself. I asked her to try to identify the participants using the contextual profiles (and the excerpts). My aim in doing this was to test the descriptions. She could not identify any of the participants (according to my field notes, she was “way off on all counts”).
Even though I presented this paper at the conference, it was with some reluctance that I made it available on open-access fora. Initially, I considered uploading only the abstract. Before uploading the text of the paper to Academia.edu and ResearchGate, I further edited the text. At the conference itself, I included details of my acquaintance with each participant, that is, how well I knew or didn’t know the participants, as this may be relevant to the analysis of power relations within the interview. However, when uploading the conference paper online, I decided to withhold such information because that might have helped other academics to identify these participants. These are concerns I will have to revisit if I develop the conference paper into a full paper. It is also an example of ethical considerations that need to be worked out throughout the writing and disseminating stages.
Practical Lessons Learned
- Refrain from using the participant’s name during the interview if possible. This changes the tone of the interview somewhat, making it sound somewhat formal but makes for easier transcription. This is especially so if the interview is transcribed or checked by people who did not conduct the interview. You will also notice that participants themselves at times use this strategy and refrain from naming people, for example, “a lecturer in the department of English Language Teaching told me” instead of “kaushalya told me.”
- Keep journal notes or a field diary during the writing stage. Most ethnography textbooks advise journaling before and after the interview and during transcription (these are also referred to as analytic memos). This was especially helpful in instances where I was unable to record parts of the interview, for example, if the participant asked for the recorder to be turned off at a specific point of the interview. In such instances, field notes or journal entries helped me resurrect the interview, contributing to the analysis. In addition, I find it useful to continue to journal during the writing stage as it allows for the reflection of prior data collection and analysis.
- Use fellow researchers as a resource. Having a supportive network of fellow researchers, which includes mentors and colleagues, is invaluable when trying to solve ethical dilemmas. Much of the practical “solutions” for ethical dilemmas are unwritten and discussing problematic instances with other (and experienced) researchers provides you a place to reflect on your own work. Research conferences, reading groups, and informal networks were significant moments of such sharing.
In this case study, I exemplified ethical predicaments I faced during the writing stages of three texts. First, I discussed the issues of choosing data to write about; second, I discussed the quandary between context and confidentiality when writing qualitative research; and third was a delineation of different strategies used while presenting individual descriptions of participants. These are not issues that can be disregarded by researchers, given the necessity of presenting our findings to large publics, be they scholarly or non-scholarly. Predicting some at least of such dilemmas could help us at not only while we write but at the research design stage as well.
I would like to thank Thakshala Tissera and Nimasha Malalasekera for their helpful comments on a draft version of this case study.
Exercises and Discussion Questions
- What are some alternative ways of dealing with the ethical dilemmas described above?
- Using any academic source relevant to your own research, consider the ethical quandaries the researchers of that study may have faced while writing. If it is not explained in the article, you may want to contact the researchers to find out how they resolved those ethical dilemmas.
- Look at your own work and list the ethical dilemmas that may arise when writing. What are some of the ways you could resolve those dilemmas?
- Think of ways that you would disseminate your research and consider ethical considerations related to this; list the steps you would take to resolve these issues.