We would like to reflect further on the methods discussed in Appendix A1 here—how do you approach a stranger and ask them to share their thoughts around identity, obtaining identity credentials, practices on privacy and other identity-based experiences? As one respondent told us in Delhi when we asked how he felt about sharing his data with the government:
well I don’t know you and you are also asking me all these questions, and I don’t know what you will do with it. Here we discuss some of the issues we faced as a research team, including “untranslatable” terms, issues of interviewee confidence, trust and recall, how to capture and store PII (personally identifiable information) and challenge of writing up 150 interviews, including addressing our own cognitive biases. Researching identities also made us as researchers reflect about our own identities and identity credentials. We hope others conducting qualitative research on new practices of identities in a connected age may benefit from these thoughts.
The “untranslatables”—digital identity, privacy, agency, dignity
As we began the project we drew on our experiences as social researchers, promising to step back from the exclusive focus on a single artifact, or indeed on a single identity system. Even in our initial conversations, the array of components of identity was large and heterogeneous enough to preclude us from speaking about identity in the singular. We agreed, from the start, that we would be open to discussing multiple elements of identities with respondents, ascribed and constructed, digital and non-digital, state-based and private-sector.
This wide lens immediately raises important questions, such as what happens when pre-digital forms of identity become codified in a digital form? Eventually, of course, we had to discern
what do we mean by identity at all and how to translate these concepts in our interviews. We talked about
how would you introduce yourself to others (gurtoo, kandoo hidiyedoo in Kannada or pehchaan in Hindi, porichoy in Assamese) or
how would you like others to see you—all of which struggled to capture the multi-dimensional nature of the term (and we also had varied responses from responses as to
well, it depends on what context—the other doesn’t have to know everything about me).
Respondents often equated privacy (
what do you see as information which is only for you) to financial privacy, and in some cases bodily privacy, but name and address were often seen as in the public domain, and not personally identifiable information (we discuss this further in essay I1). It was also marked that privacy was seen as the owner’s responsibility, e.g., for women not to post “wrong” information online.
Assam was the last state in which we conducted research and following feedback at workshops and a review of our preliminary findings, we took a deep dive into privacy. This presented the greatest challenges in formulating questions, particularly those that moved beyond understandings of privacy as PII to broader perceptions and experiences of privacy. Reframing privacy from PII to probing around harms caused by privacy breaches, particularly in a digital context, provided insights into how identity technologies might cause harms through reducing individual ability to control what others know about them. To do so, we built two specific hypothetical questions around health and financial information. Respondents had not specifically chosen these as areas of concern, but when we set up scenarios, they agreed these were sensitive and would be of concern if shared in conjunction with other data. We ensured these data points were not treated as first-hand user experiences in themselves but rather as opinions or user perceptions and as clear artificial constructs to hypothetical questions.
A final digital/non-digital distinction was that while we were focusing on identity as credentials—and not on the material (card) nature of it—every respondent talked of identity in terms of a card, rather than a digital concept. The jump here from “Aadhaar card” to “Facebook” as a digital identity was therefore also problematic for some respondents—“what does Facebook have to do with Aadhaar?” This led us to hypothesize that identity as an overarching concept was difficult for many to understand—whereas identity which serves a purpose (identification for government, self-identification for social media) was more digestible.
In addition, two significant challenges we faced were firstly that the research team felt uneasy in many sites of identity-based transactions such as government offices (private sector environments such as SIM card providers felt more comfortable). This was partly because of the environment of government offices (similar to what many interviewees felt), but also because members of the team sometimes felt they were adding another layer of questioning to interviewees who sometimes already appeared hassled. Secondly, it took a while for many of the research team (again like the interviewees) to accept the research aims, as these seemed extraneous (
that’s just how it is). Finally, questions on caste and religious groups were largely seen as as too intrusive and so it was a challenge for us to directly explore potential issues around identity credentials and these.
Confidence, trust and recall
As with any in-depth interview process, we encountered many subjects who were hesitant or refused to speak with us. In the research design, we initially considered compensation in the form of mobile airtime, but we decided against this in case it artificially skewed responses just for compensation. However, as discussed in Appendix A1, we did buy goods from traders as an opening interaction (and kept a photographic archive of them)—these also formed a useful material exhibit on the history of the interviewees as well as personal recollections. But here we outline challenges specific to this research topic that exacerbated common challenges. The first was that identification is so routinized that many thought it immutable (
it’s like this, we can’t do anything about it, why talk about it—also generally regarding bureaucratic processes). Again, this perspective was sometimes shared by the research team, some of whom found the questions harder to conceptualize and operationalize—the transaction story lens in particular was tricky to apply and we found it needs further development in the field.
At other times, identification, especially for lower income demographics, was seen as something beyond the interviewee’s capacity to discuss (
what will I know about answering for a study? I don’t even speak English. I just do what I have to. Bank cards and such are for educated/moneyed people.) Women (especially in north India and Assam) were particularly hard to solicit experiences from, as they would sometimes direct us to male family and friends stating they
knew more about these things while they—the women—
only sit at home. An additional concern around identity was whether we were gathering information for malicious purposes. Our Assam researcher mentioned that she faced aggressive groups crowding around her when she asked respondents about citizenship issues. People, especially men, were confrontational and told her that it did not matter what they said, and suspected her of gathering data to frame them.
You people from the government are going to write whatever you want. Why ask? We are legal residents and have been living here since generations.
Despite us repeating our research perspective, rather than “checking up on people,” interviewees might have initially also felt the need to give positive responses especially in the case of Aadhaar. There was a sense of having to comply as it is a government initiative, so as to not seem like a “troublemaker.” Thus initially the answers to questions such as
why did you make an Aadhaar card?, would receive responses like
the government has said we should make it, so of course we will and
being a good citizen.
We addressed lack of confidence by reiterating that all information would be anonymous (to the level of village names), that there was no “right or wrong answer” and in the case of women, we did particularly want to talk to them, not their husbands. We also built rapport by ensuring female researchers could talk to female interviewees, while young male researchers could bond on mobile phone content and “skirting the law” with young male interviewees. Our research credentials also went a long way in assuaging concerns. Once rapport was established, almost all interviewees allowed us to audio record (but not video). In all cases, we respected interviewee privacy—where recording was not possible or where interesting issues arose in spontaneous discussions, we noted these down. From the outset, our aim was a local language speaking mixed gender research team. However, in Assam, while we spoke Assamese, we did not speak some Bengali dialects (including Dhakaia, Goalparia and others), but we found the addition of an interpreter broke rapport.
One final issue specific to our questioning about identity credentials was that of memory recall for what are typically infrequent events. For any of us, recounting the chronological sequence in which we obtained identity credentials is tricky. In addition, we often only remember the situations which present challenges as we do not recall the details around smoother processes, other than that they were easy (a common story-telling narrative is to focus on a beginning (problem), middle (challenging circumstances) and end (resolution).1 This tended to bias answers (we often only heard challenges). In addition, mid-way through the research we added a question on which identity credentials interviewees had been most asked for in the past month and whether they had been asked for one card in particular. This was extremely hard for interviewees to recall and reiterated how our identity transactions are often so ingrained that they become imperceptible to us.
Examples of interview interactions
In hindsight, despite all these challenges, eliciting such personal experiences was only possible through building rapport, speaking local languages, asking open-ended questions, and the overarching skills of the research team. One example of sensitive questioning (in Garudahalli, translated from Kannada) was:
What will you do from a study on this?
Researcher (Sarita Seshagiri):
Just to understand people’s perspective on the use of these ID cards that they possess. What do they think about the services or facilities they have or are supposed to have access to? Next, people also have their identity online in terms of Facebook or WhatsApp. Then, how are people managing their identities. In general, what could be improved for people in terms of privacy and security of data.
When you came through Mr Jagannath and you mentioned your name, and you introduced yourself as someone’s daughter and granddaughter and who you are, where you are staying. That gave us a lot of trust. We know Mr Jagannath really well. His forefathers have known my forefathers well. We could get a sense that you are someone we could trust. So for someone like you, I will have no issues telling you my personal details. If someone from somewhere just walks in and asks me, I will not bother to answer and speak.
We will not speak with such people.
Why will you not speak with others?
Somehow…we do not know what kind of people they are. We cannot guess what is in their minds and hearts.
I am glad you trusted…honoured. Tell me, you must have been to Bangalore several times?
Yes of course. My daughter married and is now settled in Bangalore. So I keep going and visiting her every now and then.
What difference do you see between Bangalore and Garudahalli?
The thing is you cannot trust the people there (Bangalore). Here in the village, the ambience is so different. You can talk to people and share things. There the people around us are not people that we know really well. They will be saying things and thinking things. Some will dupe you into trusting them.
Have you experienced this? Or, have you heard people say so? Newspapers?
We see the news and we also read. We get to hear these things.
I have studied only till the 3rd standard. I have not studied much. But I do know there are so many fraudsters in Bangalore.
Tell me about it. I want to know.
You know how things are. Fraud is a fraud.
But I want to know. Quite possible that what you see as fraud, and she (Sangeetha) thinks is different. And what (Sangeetha) sees as fraud is different from what she (Nagamma) thinks or what I think. So do tell me.
Another skillful example was in the high migrant area of Kesarpur:
Researcher (Ananya Basu):
So if anyone asks you what is your identity? Do you relate to your village or Delhi?
Village, we are from our village.
You feel like you still belong from Chhapra?
I’ll tell you the truth, there’s no need to hide anything. We are around four brothers. Our father’s land is not enough
for all four to settle in the village.
We do go and visit at times. Once our kids grow up and earn a living maybe we can think of building a house.
My brothers live there, we visit once in 2/3 years for 10 days and come back.
Like you I am also an outsider here, I am from Calcutta but working here in Delhi now…
Where in Calcutta?
Do you know Sealdah station?
Tell me where exactly.
You know Calcutta?
Asif: Yes I know Calcutta very well.
I stay in south Calcutta, Jodhpur Park. I’ve grown up there so if anyone asks about me, I always relate myself to Calcutta. It is what I identify with, so I understand. But now that I’m working here in many instances I have been asked for a Delhi ID proof. Although these documents are only paper, but even so when there is a stamp to certify a residence proof, there is a voice inside that says that although this document states otherwise…
But my home isn’t here…
Yes, but my home is actually somewhere else. So this is what I want to ask you, have you felt the same as well?
The Aadhaar card was made only because it is a necessity. It is required for everything nowadays. Whether it is the company, or where we are staying on rent, or even if we are traveling to any place, the Aadhaar card is a must. It is not my identity.
Such personal interactions were only possible because our research team was humble, patient and open, and shared their own experiences and conflicts (although we were also aware of biasing responses, as we discuss below).
Privacy in photography
We were meticulous about both privacy and PII in photography in our research design. Early in the research, we agreed not to use written consent forms as signing a document presented an extra level of concern and intimidation for many of our respondents. As the “informed” element of “informed consent” was more important to us than if consent was written or oral, we instead showed reports of our work on our research mobile phone to ensure the respondents understood the purpose of the research.2 We strove to balance the goal of using photography to raise the visibility of respondents while also protecting the privacy of those for whom photos were a concern. For this reason, we have fewer photographs of women, particularly in Kesarpur and Assam (where many female respondents stated
my husband will not like it). We also maintained judgement where some interviewees were pressurized by others—for example Maqbul, the Grameen bank manager in rural Assam was happy to have his photograph taken, but when Narisa, a customer (essay V5), did not want hers taken, he insisted
let them take a photo, nothing will happen.
We respected Narisa’s privacy and only took a photo of her using the fingerprint reader for her bank account, with no photo of her face, and showed it to her to make sure she was comfortable.
A contrary challenge was when respondents wanted to use us as a voice—for example in Kesarpur, some respondents were so fed up of their conditions that even though they were aware of the privacy risks, they asked:
make a video of us and show everyone how bad our working conditions are… we don’t care. However, we had to reiterate the risks of doing so. Equally, we were extremely careful taking photographs of any cards with PII—without exception, respondents were very free sharing these (as they did not consider the details as private) but we made sure photographs were blurred (an example was of Devi’s ID cards— which all showed different dates of birth and two different names but we blurred all other data on the cards). Finally, we also conducted a responsible data audit with SIMLab who reviewed and approved our data capture and processing.
Analysis and writing up
Qualitative research is interpreted twice. First, by the interviewee (their responses, storytelling, performance and so on. For example, one question was
why do you have an Aadhaar card—how has it helped you and when a non-Indian researcher was present, the answer was
because I am Indian, but in the presence of the Indian research team, it was much more practical, about obtaining benefits from the state. Second, interview responses are interpreted by the interviewer (equal “performance” in the interview about learning but also subjectivity in analysis and writing up). We were clear that this research would not and could not be representative, but we were also aware of our own biases when designing this research and open to answers which contradicted these. For example, while initially we asked questions around caste, gender, and other “vulnerable groups,” we found that these issues were not simple, but rather intersectional and layered—a woman may find it more helpful to have a man assist in
making her ID cards or someone of a lower caste may have more benefits because of a caste certificate, rather than another individual who is poorer but of a higher caste. We specifically accounted for this by building in codes to flag data that was in contradiction to our initial thinking.
Finally, we were also aware of researcher impact, when we heard responses such as
I did not know so much about information being misused. But now having spoken to you, I understand the importance of personal data and how it can be misused (Amit, henna artist) or the Tibetan sisters who felt we were trying to tell them that the problems they were facing in changing their contact number for the bank in Ladakh was
for their own privacy (essay V2). We had to emphasize that we were only asking questions, not giving suggestions. In addition, we were conscious—as in all research—that while we had invaded the privacy, personal space and time of our very generous respondents, we did not have the chance to discuss the findings with them. We are grateful to all respondents, from whom we learned so much and on whose behalf we are working in this study.
Catherine Kohler Riessman, Narrative Analysis (Newbury Park, CA: Sage, 1993); Barbara Czarniawska, Narratives in Social Science Research (Newbury Park, CA: Sage, 2004).↩
Josh Woodard, “What Is Informed Consent in Digital Development Photography?,” ICTWorks, March 16, 2016.↩