ethnographic masking in an era of data transparency
Demands for data transparency are sweeping across the social sciences. Calls for the sharing of survey instruments, code, databases, and even interview transcripts, have grown increasingly louder. How will ethnographers respond? One place we might start is by rethinking the long-held convention of masking—the practice of hiding or distorting identifying information about places, organizations, and people.
We frame this essay as a conversation because, over the last few years, the two of us have been having a vigorous discussion about masking as we each grappled with methodological and theoretical challenges that masking was posing for our respective ethnographies in two different small towns. This conversation has evolved beyond our own work and resulted in an article that probes larger issues around the ethical and scientific tradeoffs of masking and challenges the usual justifications given for this practice. We want to use this forum to share some of our experiences and thinking on this matter and invite others to join us in the exchange.
Murphy (M): Our starting point is that masking has become virtually a ritualistic practice in ethnographic writing. When ethnographers sit down to write, they commonly assign fictitious names to the people and places they study. This is often justified as an ethical necessity, to protect our participants’ privacy and/or prevent them from being harmed from their participation in the research. But concealing identities is a slippery slope. Pseudonyms alone are seldom sufficient to protect confidentiality. And so ethnographers frequently engage in more extensive masking by, for example, altering identifying characteristics about people (e.g., changing a person’s gender or occupation) and places (e.g., altering historical events or census data), omitting primary source references, and/or creating composite characters. These practices run counter to the growing expectation for data accessibility and for making it possible for others to triangulate data sources and compare cases.
Jerolmack (J): It seems to me that one way masking does this is by burying sociologically significant information that can be used by the scholarly community to independently evaluate the ethnographer’s analysis and consider alternative interpretations. We discovered a great example of this in the book Forgive and Remember, an examination of the training of surgeons and the organization of their work in one hospital. In the book, Charles Bosk changed the gender of a female surgical resident in an effort to ensure her confidentiality. Almost 30 years after its publication, Bosk revealed that this participant was the only woman resident in the group. By altering her gender, Bosk made it impossible for a reader to consider how gender shaped hospital interactions and promotion practices. In hindsight, Bosk admits that gender was actually very central to understanding the social dynamics of the hospital, but it was something he did not notice at the time because of how his own gender was privileged in the field.
I don’t think ethnographers acknowledge the extent to which decisions about what and how to mask are inherently theoretical choices. Though ethnographers usually claim they are only making “minor” changes, to other scholars—especially those with different theoretical interests—they may be quite major and necessary pieces of information for them to come to their own conclusions about the role of particular biographical or situational factors in the analysis. As Bosk writes, “My problem with changing Jones’ gender is that it makes the critique I did not make impossible for others to make.”
M: Another problem with masking is that it makes generalizing beyond the ethnographic case difficult. Scholars often give places fake names, for example, to highlight how their case generalizes (e.g., “Middletown”) and to help move readers away from the case’s idiosyncrasies. But in my estimation, masking often removes the details a reader needs to specify what to generalize a case “to.” For example, in my work on suburban poverty, I have been documenting the ways that the built environment significantly shapes the social networks, neighborly relations, and survival strategies of people who don’t have cars or access to public transit. In analyzing my case, I turned to Carol Stack’s All Our Kin, a classic ethnography of survival strategies and kinship networks in an “African American ghetto” in the “Midwest” that she calls “The Flats.” Why, I wondered, do my findings about kinship networks differ from Stack’s? Unfortunately, I’m not able to explore whether differences in the built environment of the communities we studied account for some of these discrepancies because, in addition to concealing the location, Stack gives few details about the spatial configuration of the community or the networks she observed. She does note that her participants are spread out and that no one lives more than 3 miles apart, but what I have found is that living two blocks versus three miles apart or living in walking distance to businesses makes a significant difference in how these ties help people make ends meet. What kinds of communities can we generalize her findings to, then? To build a broader, more specified theory about poverty, race, place, and social networks, we need sufficient detail to answer questions central to making sociology a cumulative social science.
J: I’m also concerned with how masking can preclude replication, falsification, and comparison. The ethnographic revisit, for example, has been touted as an important way to study change over time. By revisiting someone else’s fieldsite years or decades later, ethnographers can use the original study as a kind of “baseline” comparison in order to specify how observed changes in interaction and the social order of the setting are the result of intervening historical forces. But when the scholarly community is denied the identity of places, organizations, and even people, making these types of revisits may be impossible for anyone but the original ethnographer, who remains the sole gatekeeper of the fieldsite and the knowledge produced about it.
Even if an ethnographer should serendipitously stumble upon a pseudonymous site that has been previously studied ethnographically, the original masking of the site constrains the later ethnographer’s ability to use the first study as a historical record. I am dealing with this dilemma in my current research project. When I moved to Williamsport, Pennsylvania in 2013 to study how shale gas extraction (“fracking”) is transforming community life, I was delighted to discover that an ethnography had been written about the area in the years just before fracking commenced. Since I had not been in the field before fracking began, having this earlier portrait of the community would enable me to specify how this industry has changed this town. The problem? In addition to using fake people- and place-names, the earlier ethnography masked numerous details about civic leaders, historical events, and organizations. That move now hampers my ability to directly compare changes over time among particular community groups.
Revealing place names can enable other kinds of revisits as well. For example, although my colleague Eric Klinenberg gave pseudonyms to the victims he wrote about in his book Heatwave, his decision to name the city of Chicago and the two neighborhoods he compared allowed other scholars to quantitatively test his qualitatively-derived thesis about how differences in neighborhood-level factors contributed to disparate mortality rates across neighborhoods. Such comparisons enabled researchers to examine how representative Klinenberg’s neighborhoods were to other neighborhoods in the city and test whether his claims about his own case studies were supported with other kinds of research techniques and data. All too often, however, such virtual or quantitative revisits are foreclosed by the ethnographer’s decision to mask place.
M: Many scholars may be inclined to concede that not masking identifying information would make ethnographic work more transparent—and, I think we both agree, more scientifically useful—but what would be the ethical implications of not masking, especially around issues of privacy and confidentiality? After all, it is our duty as researchers to “do no harm” to those we study. Colin, how do you think about this having already published a book where you do not mask and do not use pseudonyms?
J: My stance on naming was heavily influenced by a passage in Barbara Myerhoff’s moving portrait of an elderly Jewish community center (Number Our Days) in which she reveals that the seniors asked her to reconsider her decision to use fake names because they craved “an enduring record” of their existence. (She didn’t). The lesson for me was that I would give my participants a choice to be identified or masked, and in the end, only one person chose anonymity. There was very little that I could offer my participants for all the time they gave me, but they viewed seeing their name in print as intrinsically rewarding. When I handed out copies of The Global Pigeon, most quickly thumbed the pages looking for their name and some excitedly took photos of the printed pages they appeared on and texted them to friends and family. This has convinced me that, at least some of the time, naming may be more ethical than masking. It’s worth noting that I have gone through 3 different university IRBs, at both public and private schools, and have never had trouble getting IRB approval to reveal names for either of my ethnographies; IRBs do not in fact require pseudonyms or masking.
As for the “do no harm” ethic, it’s not clear to me that pseudonyms and masking actually protect our participants. There are numerous instances where readers have identified the people, places, and organizations depicted despite the author’s efforts to scrub identifying information. Furthermore, masking also often fails in its goal to prevent participants within a fieldsite from identifying one another. A great example of this comes from what happened after the publication of Nancy Scheper-Hughes’ Saints, Scholars, and Schizophrenics. Not only did a reporter publicly unmask the Irish village that she called “Ballybran” but villagers also easily figured out the person behind each pseudonym despite the lengths she took to “scramble certain identifying information.”
M: It’s interesting to note that this example, and others like it, occurred long before the era of Google and social media. These technologies have made it even easier to identify people and places by performing keyword searches based on information or events depicted in an ethnography. I would guess that this is all the more possible given how much contemporary ethnographers have to interact with participants online to truly understand their social worlds, tethering us to our participants in very public ways.
J: Indeed. Given all of this, it seems to me that the practice of using pseudonyms and masking offers, at best, an illusory promise to protect confidentiality.
M: That’s exactly what Scheper-Hughes concluded as well. In fact, she wrote that if she were to write her book on “Ballybran” again:
I would…avoid the ‘cute’ and ‘conventional’ use of pseudonyms. Nor would I scramble certain identifying features of the individuals on the naïve assumption that these masks and disguises could not be easily de-coded…I have come to see that the time honored practice of bestowing anonymity on ‘our’ communities and informants fools few and protects no one—save, perhaps, the anthropologists’ own skin.
So Colin, are you saying that if we are not fully able to protect participants through masking we should do away with this practice altogether?
J: Not quite. Masking may be practically necessary to carry out some research and ethically required to carry out other research. But I think the universe of cases that “require” masking is much smaller than ethnographers acknowledge. Moreover, I believe that it is unethical for ethnographers to imply to our participants that we can promise confidentiality when it cannot—with any degree of certainty—be ensured, and that we should respect the fact that sometimes participants actually want to be named.
M: That makes sense, but I would add that participants should also have a right to decide if they don’t want their involvement in research to appear online when a friend, family member, lover, or employer Googles their name years after the completion of the research. The use of pseudonyms, without the masking of other identifiers, would provide such protections.
J: So where does this leave us? It seems apparent that there are very real scientific costs introduced when the ethnographer engages in masking, costs that negatively impact the community of scholars by impeding our efforts to construct cumulative social science. It also appears that the reason ethnographers usually give for masking—confidentiality—often does not hold up in practice. Yet I can see that some degree of masking may be practically or ethically necessary for some fieldwork.
My own conclusion is that masking is no longer defensible as the taken-for-granted default position. My experience with naming in The Global Pigeon, and my frustration with how a previous ethnographer’s decision to mask Williamsport is hampering my research on fracking in the same community, have convinced me that ethnographers should make naming the default, and then mask only to the extent required to ethically carry out your research. Alex, where do you come down on this?
M: This is a question I am still struggling with. I do think we should strive to mask as minimally as possible. And I think to do this we need to have extensive, honest, ongoing conversations with our participants about the costs, benefits, tradeoffs, and limitations of masking. As I work to make these decisions for my own book I am doing just this—going back to my participants as well as other community members and stakeholders to have these discussions. Doing so has been invaluable—they’ve made me aware of concerns around naming and masking that are important to them but that I had not thought of and vice versa. I guess I would say that I don’t think there is a universal answer. Every ethnography poses some distinct ethical issues, and so these decisions have to be made on a case-by-case basis, in consultation with one’s participants. That’s where I stand.
I am glad, though, to be having this conversation; it’s an important one. I definitely think that masking is a practice worth rethinking in light of the issues raised here.
J: This is especially the case in the face of scholarly and public demands for greater data transparency. While ethnographers may be understandably reticent to share their interview transcripts and unfiltered fieldnotes, masking minimally, and only when necessary, is arguably a more practical means of addressing the transparency expectation. It also is likely to have more scholarly utility since it enables researchers to bring other perspectives and sources of data to bear on one’s case study, which is in alignment with the goals for data transparency. At any rate, ethnographers risk marginalization both within and outside the academy if they ignore these concerns.
Alexandra Murphy is in the sociology department and population studies center at the University of Michigan. She studies suburban poverty and transportation insecurity and is co-editor of The Urban Ethnography Reader.