At the outset of our research project “Coming of Age on Instagram,” we noted a paradox concerning public knowledge about social media. On the one hand, users acquire expertise about platforms as they use them on a daily basis and figure out how to do so strategically. On the other hand, public debates about social media are often encased in anxious narratives of harm and risk that blot out such distributed expertise. We wanted to see what we could learn about social media by talking to users rather than merely talking about them. That sounds straightforward enough, but it turned out that, as social scientists, we face challenges in working collaboratively.

Who knows

That is because social scientists have long worked on the assumption that they have a better grasp on why people do things than the people doing those things. This goes back to the founding figures of anthropology and sociology. Émile Durkheim thought the point of social scientific inquiry was to uncover the “social facts” that were operative behind the phenomena we study, independent from anyone’s individual consciousness. Talking to the people involved in making those social phenomena happen is a distraction at best, perhaps even counterproductive. Those with a more interpretive bent, such as Max Weber, thought that understanding the meanings that actors themselves attach to what they do is important. However, the actual practice of interpretive social science often still enthrones the researcher as the one discerning the meanings guiding social action.

For many years, Grounded Theory (GT) has been the most popular approach to interpretive analysis across several social sciences, including anthropology and sociology. GT was first proposed by Barney Glaser and Anselm Strauss, and was later (after the two had a falling out) elaborated by other scholars such as Kathy Charmaz. GT is an iterative approach to building concepts and constructing theory from qualitative data. It usually entails several “passes” of coding (labeling) qualitative data, such as interview transcripts and fieldnotes, with the aim of identifying recurring themes that “emerge” from the material. After each pass, the researcher may opt to return to their research site to gather additional material on previously underexplored themes until they find they have reached a saturation point. At later stages, the researcher usually aggregates the codes assigned in earlier stages, first into “axial codes,” and later into concepts. Because the concepts are closely tied to themes identified in the qualitative material in an inductive manner, they are “grounded” in the data rather than, for instance, the researcher’s preconceptions or theoretical hobby horses. Even though codebooks, measures of intercoder reliability, and (more recently) computational techniques for pattern recognition allow the GT process to be a team effort, it generally unfolds without research participants present. Thus, despite recognizing the need for the creation of knowledge about the social world to be grounded in first-hand observations, the separation between data collection and data analysis upholds the epistemically privileged status of the researcher.

Theorizing together

GT is undisputedly the mainstream in qualitative data analysis – The Invention of Grounded Theory by Glaser and Strauss has been cited more often than volume 1 of Marx’s Capital – but it is not the only game in town. Research traditions like Kurt Lewin’s action research, marxian workers’ inquiry, or Freirian and feminist participatory methods seek to break down the division between researchers and researched. The more recent trend of citizen science usually involves nonacademic communities in data collection in their own environments, while data analysis and theorizing often remain the tasks of academic experts.

In our project, we wanted to advance collaborative theorizing. One way we did so was through a session at the Digital Methods Initiative’s 2024 Winter School. The DMI at the University of Amsterdam has a history of almost two decades of “repurposing the web for social and cultural research”. Summer and Winter Schools consist of week-long “sprints” during which groups of participants complete research projects centered on what Richard Rogers calls “natively digital” datasets.

Collaborative methods

Our session involved both pre-existing qualitative data (transcripts of in-depth interviews conducted in 2022) and autoethnographic data provided by participants in our session. For the latter, participants were asked to send DMs (direct messages) to themselves twice a day over the course of one week before the start of the Winter School to record how they used Instagram to document, recall or narrate parts of their lives. Similarly to diary studies in communication research or “experience samples” used in the psychology of everyday life, these DMs provided granular, first-person insight into an underexplored aspect of how people use Instagram: not just as a means of presenting themselves to others, but also as a way of assembling their biographies. These messages became the basis for our collaborative autoethnography.

On the first day of the Winter School, after introducing our group topic, the twelve participants interviewed each other about their diaries. In this manner they gained insights into similarities and differences between their individual experiences. Because at times participants recorded quite intimate details from their lives, analyzing them in a collaborative context was ethically challenging. We addressed this challenge with the group of participants, and actively worked on creating a safe and open research space. We shared regular check-in moments to discuss how it felt to share one’s stories with the group.

Equipped with these initial insights, they spent the afternoon studying extant literature on memory, biography, and life writing in search of concepts that resonated with their incipient ideas. This was followed by a group discussion and an initial mind map to develop a shared analytic vocabulary. We repeated this kind of joint discussion throughout the sprint, gradually sharpening our concepts through constant comparison. The following day, participants read transcripts of in-depth interviews with Instagram users of a similar background (international students pursuing master’s degrees). This provided further opportunity to reflect on their own experiences and put them in a wider context. They were instructed to write a vignette based on the interview transcript that conveyed the interviewee’s experiences with Instagram as a memory device. Through this exercise, we discovered new patterns that further enriched our analysis.

Theoretical and pedagogical lessons

As we were preparing to share the results of our sprint with other Winter School participants in the form of a poster, we struggled to condense our ideas into concepts. As facilitators of the process, our main challenge was to give up control of the process and allow our participants to decide on a way to organize our ideas. The GT process, as described above, usually unfolds over a longer duration, and our approximation of this analytic process in a condensed time frame at times felt chaotic and unstructured. Our participants, however, reported something very different. Overwhelmingly, they felt that the process led them toward a “logical conclusion” as their ideas gradually took shape through an abductive process.

Just as importantly, the process was pedagogical as well. Participants experienced the sprint as a space of intimacy and the discussions as therapeutic. They shared their vulnerabilities with relative strangers as they worked together to make sense of common issues pertaining to their Instagram use. This allowed them to put their personal wins as well as their troubles into perspective, prompting them to reflect on themselves in relation to the wider society. Some students initially felt insecure about research methods, and they gained confidence and curiosity as a result of our work together. The combination of autoethnographic methods and collaborative theorizing proved a valuable tool to teach students about qualitative methods and the practice of theorizing.

Collaborative theorizing allowed us to study youth and digital culture from young people’s perspective. Instead of analyzing data using our own framework, we learned from participants which insights they found most relevant. This took our thinking and writing on what it means to come of age on Instagram in new directions.

While this exercise in collaborative theorizing took place in a classroom and did not take us out of the academic world, we clearly experienced the benefit of giving up control over the process of theorizing to co-researchers who, in drawing on their own experiences, formulated surprising conclusions.