How do you know? Seven theses on qualitative sociology as theory and method

Introduction

These are notes about the practice of qualitative sociology; this is not a technical report on the best ways to conduct fieldwork, a reflection on the pairing between themes and sites, a soliloquy on positionality, on the relationship anonymity and replicability, or on what should count as an ethnography and what should, per opposition, be consequentially excluded. Doing qualitative sociology is about producing valid knowledge about a particular site or group of people, aiming to imagine how that site and/or people say something meaningful and significant about a sociological topic or issue, and advance a compelling argument about it through data.

In the pages that follow, I will develop seven theses about how knowledge is qualitatively produced, adjudicated, validated, written, and justified, emphasizing the distinctive character of qualitative sociology vis-a-vis idealized deductive and variable centered approaches. The lessons I describe here come from years as a card-carrying qualitative sociologist (having published articles and books based on participant observation, interviews, archival work, and ‘go along’ investigation of routines and spaces); teaching courses and workshops on qualitative methods at the graduate level; and working for the last few years as the co-editor in chief of Qualitative Sociology, one of the main qualitative journals in the US sociology. What I describe here is not just my ‘take’ on the field, but rather a concise guide to best practices, as seen in research published, submissions to the aforementioned journal, as well as in selected recent books and articles on how to do qualitative sociology. ¹ It does not pretend to be an exhaustive literature review, or an attempt to provide coherence to a wealth of definitions on what is indeed qualitative about qualitative sociology. The latter, in fact, seems to be a forced endeavor, which would aim to reduce the richness of world of practices of inquiry, all loosely sustained and connected by this shared moniker.

In advancing its seven theses, this article discusses: (a) how different qualitative sociology is from other approaches; (b) the role of ‘casing’ in generating both units of analysis and settings; (c) the theoretical and empirical work of adjudicating what some emergent phenomena is a case of; (d) the ‘modelization’ through writing of our case as a research object; (e) the rhetorical construction of causality, and the central role of ‘puzzles’ on it; (f) the reflexive epistemological vigilance about the role of the participant observer in producing the knowledge they generate; and (g) the reconceptualization of qualitative sociology as a type of epistemological package, and of theorization, in consequence, as a kind of practical activity.

Thesis 1: Qualitative sociology is not just a method

Historically, the instruction of qualitative sociologists was mostly a training in ethnography, captured best by the metaphor of ‘sink and swim’, in which learning happened by doing. Training new apprentices into the craft, was accomplished by sending them with few instructions into a location, and ask them to perform a series of note taking exercises, loosely tied to a theme, which itself emerged inductively from the interaction with the site and the people in it. This learning by doing training is still a part of how new scholars first approach the entrance to the field, but has become destabilized as the only and de facto way of approaching social phenomena qualitatively

This earlier stability was partially achieved by a double reduction practiced by some sociological ethnography: to make qualitative sociology equal to participant observation; of ethnography an area of inquiry made completely out of urban ethnographies; to give this name to works that are actually just ethnographies of interaction in urban settings, and for interactions to be main unit of analysis as well as the currency of what constituted good scholarship. There are multiple reasons for this destabilization. Some are endogenous, like the extension of ethnography into other topics beyond the early focus on urban settings and the consequent recalibration of the scope and potency of the method in relationship to other foci and theories; some exogenous, mainly the incorporation of considerations about race and gender, which in turn transformed the topics investigated, as much as the relationship with macro background categories, and the role of the researcher themselves.

The destabilization of this equalization of ethnographies as reports of interactions in particular places, opened multiple avenues to re think what qualitative sociology is actually about. And in this lack of standardization, what replaced the combo interaction-grounded theory-emergent themes is less a new reduction on what is to be done, than a pluralism that made evident that reporting how people inhabited places, was an active construction, in which what mattered was to identify what was puzzling or ambivalent about a phenomena, and how that enigma could be better made sense of. If not by focusing on interactions, was it better to try to capture how dispositions produced strategies for action? Shall we imagine what is getting accounted for as the result of history unfolding, and compare it to previous configurations? Focus on the mediating mechanisms and processes? Organize our questions around how culture motivates actions and its justifications? On how the distribution of people in neat categories resulted in particular kinds of hierarchical relationships?

Regardless of what is our favorite theory, the Pandora’s box of qualitative sociology has been inexorably opened, calling attention to it, not anymore as just a ‘method’ (in order for it to be just that we would need very stable and shared definitions of the phenomena to begin with) but rather as a full-blown way to produce valid knowledge. As such, it is not there to generate early hypothesis, emulate quantitative research with a failed logic of inference – since it cannot emulate the statistical capabilities of random probabilistic samplings on a large N – or aim for confirmation.

In inhabiting that difference, qualitative sociologists learn to sample for variation, and for theoretical gains; to think less in terms of generalization and more of how good is the constructed research object as a stand in for a larger sociological question; to think of unique cases not as a problem, but actually as an opportunity; and to think of the process as neither deductive nor inductive, rather as a constantly iterative project, in which the constant back-and-forth between the empirical material and the scholarly literature helps us to contrast, define, and refine, what is it that we are studying is indeed a case of. In this process, it is important to remind ourselves pace Jeff Alexander (1982), that theory and data are not pre-existing things in the world, but rather linguistic conventions, their difference anchored less in their ontology, and more within the logic of aggregation. Theory is the word we use to refer to cases in which what we produced as data serve us to illuminate similar phenomena in other contexts that the ones we investigated.

Thesis 2: Cases are produced and ‘won’, not found

One of the consequences of this opening up of the menu of the styles, theories and warrants to produce qualitative knowledge, is precisely that data are not found but rather generated by our choices along a long fraught iterative process of attribution. How do we know what the data that we have generated talk about? What do we make of the uncertainty and anxiety that comes when asking this question as we enter into fieldwork? How do we take advantage of the early confusion? How do we answer the most common query posed by every journal reviewer when inspecting the presentation of data and its correspondence to certain larger claims: what is this a case of?

In this section, I argue less by proposing, and more by emphasizing the negative: no good qualitative project comes from a homogeneous sample (on this, see Small and Calarco, 2022); as exactly the result of what we have hypothesized earlier on the pre study; or that does not lead us into a confusion about our earlier orienting frames, and how much the occurrences we observe and report upon can easily fit with them. Rather, qualitative scholarship advances by a series of temporally somewhat discrete steps. Following Richard Swedberg’s (2014) conception of the research process, the research project gets divided into two phases. The first phase, the ‘prestudy’, consists of early field observations during which the researcher identifies a phenomenon inadequately explained by existing theory. Moreover, during the pre-study moment, qualitative scholars might be better favored not by privileging one particular path for the collection and production of data, but rather by engaging in multiple ways; taking accounts, actions, life stories, and archival work that explain the potential contexts to make sense of what might explain what is being described, observed, and analyzed. The limitations about what counts as a particular kind of phenomena (e.g. cultural distinction) are closely intertwined with particular avenues for data collection and production.

In other words, what would happen if scholars went into the field armed with a set of theoretical sensibilities that would allow us to interrogate the world, imagining what concept of the phenomena – among the myriad of existing ones – would best fit the case? I pose this in contrast to both the tabula rasa that grounded theory imagines and with the impulse toward quod erat demonstrandum more common to critical approaches, which tend to privilege one pre-established theoretical perspective, and consequently where we find only what we already know. I am building here on Timmermans’ and Tavory’s (2012, 2022) call for abduction in qualitative work. For them, qualitative work begins from extensive and broad knowledge of the pre-existing theories that allow for anomalies to be identified and for theory to be constructed. I am less interested here in the question of innovation but want to highlight the role that multiple theorizations play in allowing the researcher to engage in a recursive process of double fitting data and theories. Where this differs from Timmermans and Tavory, however, is that I envision in the false starts of the construction of the study object, not an avenue for the search of anomalies, but rather a great opportunity for the challenge of our earlier theoretical and normative pre-conceptions. I also highlight here one of the central component to the grounded approach to the context of discovery – though not its relationship to theory at large: patiently to take the false starts, the struggles in establishing the study object during the context of discovery, as part of the knowledge enterprise, almost as a mandatory step against theoretical omniscience, and to the service of better knowing.

The second phase of Swedberg’s research process, the ‘main study’, consists of a more focused research design oriented toward answering the research questions iteratively raised by the prestudy. Orienting our projects toward a more stable and coherent set of concepts and definitions, is a long fraught struggle, that happens in multiple stages. I see this as a dialectical activity, in which certain occurrences in the field point us to their potential conceptualization or classification as a particular kind of phenomena, even if not in the realm of ‘our favorite theory’ (Burawoy, 1998; Eliasoph and Lichterman, 1999) still within alternative options our colleagues in the particular sub-discipline we partake in have mobilized.

In this process, variation within the sample itself plays a central role. There is an interplay between internal variation and external validity. The more variation of people, actions, locales that make up the site; the more we have elements to establish a series of descriptive inferences to better think what are our data a case of. We can start thinking about how variation occurs, and also about what explains it. We also avoid the fate of sampling on the dependent variable, and make of the world we observe and describe to others, the result of pre-existing forces outside the purview of the research we have undertaken. If we probe enough, and for a relatively long period of time, we will see how much we can start noticing nuances, internal distinctions, and we can make what we thought at first as a homogeneous entity into something heterogeneous, that needs and calls for an explanation. At that moment, ‘theory’ is invoked again, helping us to elucidate – to quote the words of Jack Katz’s (2002) – how to go from how to why, via the comparison of what we have found so far to ideo-typical instances from the literature of what we imagine to be the same phenomena. While casing is a heuristic device, it is also a social fact of sorts, which has to be ‘won’ by the variegated and puzzling data we encounter, produce, and present.

Thesis 3: What a phenomena is a case of is adjudicated iteratively

As qualitative sociologists, we partake in effective symbolization using homologies that jump between different levels of analysis, from particular actions and utterances to what is being said; we reconstruct both the context under which certain actions become meaningful and the cultural structures that organize the experience. Aiming to find homologies between disparate situations (like Bourdieu does, e.g.) is a pragmatic exercise, which aims to weave disparate materials into larger signification structures as well as its role in practice and interaction. Different contemporary traditions have justified particularly effective symbolization practices, conceptualizing them in three different yet closely associated ways: thinking of them as the work of analytic induction, the result of discarding data by searching for negative cases, or by focusing – following the logic of pragmatist philosopher Peirce – into the role of abduction.

While I discussed the latter earlier on, analytical induction is an exercise in theoretical refinement and subtraction. The idea behind it is to specify the conditions under which certain parts of social life emerge (Katz, 2001). Attached to this is a claim to causality, in which the refinement of the explanandum leads also to the transformation of our understanding of the explanans. Going against the search for confirmatory cases, its impetus is to seek new encounters with an increasing range data to revise our explanations, and aim to find a varied set of cases for which our ad hoc theories would also apply. Relatedly, and following Deener (2017), we could say that qualitative analysis advances in three sequential operations: The first one consisting on how we isolate phenomena trying to tag what we observe to similar phenomena pre-existing in scholarly work; the second one slowly marking spatially and conceptually boundaries around the internal order of the phenomena studied – as in the delimitation of what our site is, and how that contributes to understand what our study is a case of; and third, taking advantage of instances that a priori do not make sense, are puzzling, paradoxical, and how much these are ambiguous occasions that necessitate further theoretical elucidation.

Alternatively, we can arrive to the effective symbolization I mentioned earlier, by pursuing a strategy that defines, focuses, and refines our cases through the work of negative hypothesis, what we do is to ask a different, pointed, and more modest version of ‘what is this a case of’. What we do is to ask what kind of data do we need to build support for our current explanation? What is the necessary data to have at hand to gather support for the alternative explanation? And – moreover – what would be the data necessary to discard alternative explanations to the one we are giving? Have we probed for it?

I did this very explicitly, for instance, in my own work (Benzecry, 2011) dedicating a whole chapter to show how much the experiences and knowledge opera fans accumulated did not turn into a form of capital, as Bourdieu has so thoroughly explored to understand the reproduction of inequality through cultural consumption. To do this, I explored how much those who defined themselves as opera fans communicated this experience to others around them, hoping to probe for the conversion of this into symbolic capital, only for the participants to explain to me how little they shared of their intense attachment to others because of the fear of stigmatization, and all the contextual work they needed to perform to explain that attending opera several times a week, was in each case a different endeavor, instead of obsessive behavior. I inquired about how many friends they made at the opera house, thinking that would mean I could find the conversion into social capital, only to witness and find reports of people who, while friendly and attentive to others, rather enjoyed the music alone, surrounded by strangers; I started the fieldwork with the fantasy of people breaking bread together before or after, only to find folks returning home alone to slowly take in what they had witnessed and experienced. I then tried to see whether their intense knowledge brought them any kind of connections, claims of status that would lead to upward mobility in their life, only to find again no currency conversion. In doing this, I was hoping to convince the reader that the Bourdieusian framework was not enough to make sense of what I had witnessed during fieldwork, and had to rely instead on a different toolkit.

Part of the work of generating qualitative analysis rests on the exercise of subtracting ‘theory’ – as what we call authors and their citations, complete theoretical frameworks – and focus instead on how the different bits and pieces of what we are studying connect. Under what condition does instance x appear? If I see instance x, will I immediately see instance y happening? When is a phenomena absent? And after we play with this ‘removal’ of theory with capital T from the text, we need to complement that with abstracting what we have encountered in the field or archive, in a movement of aggregation that slowly gets us closer to think beyond the particularities of our site, and into a more precise definition of what our case is, and how it relates and contributes to the literature. We do this by thinking of what is unique and distinctive from our findings abstracted. In consequence, theory – with a small t – means abstracting from the data, case, and phenomenon. Paraphrasing Abbott (2011) and his notes on theorizing, abstracting then means removing particular things from the case, to enable us to see what other kinds of phenomena it resembles. It means thinking about what kinds of things are comparable to your phenomenon and why.

But there is not just one type of casing strategy. Sometimes, we can think of a case to see it as the most extreme version of a particular phenomenon – as Erin O’Connor did in her ethnography of glassblowing, which shown the embodied character of skill acquisition, and the perils of doing so under the extreme conditions of literally playing with fire. Sometimes, we produce our data as a ‘transparent case’ (on this, see Zussman, 2004), as Arlie Hochschild (1983) did when showing the ‘emotional labor’ of airline stewardesses, which was something particularly visible in that profession. The connection to other similar cases was less its representative character, but rather that it allowed us to see a dimension present in other white-collar service works, but was until then under theorized. In my own archival work (Benzecry, 2014), I pursued a ghost/paired case strategy, which used the conflicting history of the foundation of the Colón Opera House of Buenos Aires, to illuminate dimensions that were missing in what was the paradigmatic case of cultural entrepreneurship (DiMaggio, 1982a, 1982b): the establishment of classical music not as entertainment but as high culture. While his work on the Brahmins emphasized their ability to produce social closure, it missed certain key dimensions present at the Colón Opera House, including intra elite conflict, the conflict with subaltern populations, as well as the absence of the state as a player in producing and solidifying cultural classifications. ² The pairing resulted in a new avenue for theorizing variegated pathways into high culture institutionalization, encompassing most other cases in the literature.

Thesis 4: To write the case in you need to clutter and shed!

If the previous two theses referred to our strategies to make sense of what our data are a case of, the following two discuss in depth the work of writing in, aiming to convince readers about the analysis we are advancing. They happen as we move from the context of discovery toward the context of justification. Monika Krause (2021) inquired about the work of modelization in the life sciences from a sociology of knowledge perspective, to show how case studies practices like the ones I describe here can be modelized into best practices. From her work, I take two key concepts, research objects, and epistemic targets, to underscore the tension between our sites and themes as strategic research materials and how they are stand in for more encompassing questions. The material research object is what we aim to access through particular traces we call data, produced by our instruments and observations; the epistemic target is what we formally are trying to understand better. In these conceptualization, our sites, people, and places, are our ‘objects of experience’, while our bets on what something is a case of, and how we are going to specify it, is what we think the material research object is actually ‘about’. If on the previous two sections, I discussed how a case is won during the research process, in this thesis and the next, I want to show best practices in how to write in such a way in which our claims about our immediate findings, are also claims about our epistemic target.

I share the next paragraph to highlight the difficulties to reduce or ‘slim down’ a case. If I warned us before against too little variation, what we confront here is the opposite problem, instead of scarcity, what we are confronting is the management of excess, of abundance. In the paragraphs that follow I describe the feeling of exhilaration I felt toward the end of my fieldwork in South China for my book (Benzecry, 2022) on creativity and globalization in the shoe industry, in which the latter was a strategic research object to ask questions about the embodied and mediated dimensions of craft, the centrality of repair and maintenance when thinking of scale making practices, and how collaboration at a distance is achieved.

Every scene opens up a potential new path for research, and with it a whole series of questions: Shall I learn how to do technical work on a shoe, like one of the senior technicians suggested? Should I follow the sourcing agents to the markets? Once in the markets, was it smart to try to understand how leathers make it there? And if I did that, should I have accepted an invitation from an Italian tannery, the showroom of which I visited in Houjie, to go back to Arzignano, and see how they process the material? And once there, should I have taken advantage of my Argentinean contacts, and gone to Luján to see where the leather carcasses that are treated in Italy and then used in Dongguan come from? Every new contact, every new observation opened up a new avenue for exploration that led to one important question for ethnography: where does one stop? Why? How?

One of the challenges that accompanies this kind of qualitative scholarship is how to best capture and reveal the bewildering childlike excitement of what happens without resulting in an incomprehensible collection of details without an order or a conceptual story to tell. To translate evidence into writing is to a certain extent to let some of the evidence go. We are not writing to exhibit some extraordinary piece of data, but rather to justify our excursion by explaining what we learned, and how we can communicate it to others. After the heroic moment of accumulation, comes the ascetic and sober exercise of renouncing and winding down the data that we have. Being an author is something authorized by data, and it also means standing somewhere, choosing which avenues to take, and which ones to discard.

The key to understand how to confront this issue, as advocated by my co-editor Andrew Deener (2018), might be to explain not the process of empirical inclusion (as we see in the Appendixes where we read about the heroic character of the incredible amount of fieldwork we have done, resulting in thousands of pages of notes and transcriptions) as the only way to demonstrate the reliability of our data, but rather to focus on the process of narrowing down the case. Doing so allows us to better understand the research process as a collective and institutional project that involves colleagues, students, reviewers, mentors, and editors. And I want to quote Deener verbatim here:

How researchers learn to exclude certain subjects, points of data, and alternative analytic themes is difficult to recount and assess, but it is of equal importance to constructing cases. Empirical errors occur when ethnographers misstate or misidentify basic facts as they relate to subjects, situations, events, and locations. Empirical errors are different than the observational and interpretive omissions necessary to narrow down and hold constant the units and levels of analysis. Ethnographers should be very concerned about getting the facts right, but they should be equally concerned about getting the case right.

In terms of writing – and in the spirit of Science and Technology Studies (STS) that also animates Deener’s and Krause’s work – I learned a lot during my fieldwork from the practices of designers as they put together a collection. Designers operate through a logic I have termed ‘cluttering’. Cluttering comes from the verb to clutter, and I use this term to call attention to the accumulation of objects without an apparent order or syntax. These accumulated objects contextualize or generate a background, which results then in the production of difference. Through the grouping and comparison of shoes and other items bought in Europe for inspiration, designers can think of what stands out. If we think of the different groupings of objects as starting from a relatively static state, cluttering allows us to see how the introduction of a new piece brings up properties to forms that were inactive without that particular object in the series, reconfiguring what previously existed. It is also through the myriad sketches that get drawn (and the many avatars they can generate) that designers decide which shoes belong together, form a series, or appear as something that could be narrativized via trends and mood boards.

What I am describing and conceptualizing is an integral part of managing excess, working through an apparent disorder as a way to build a working syntax that will slim down the potential ecology of objects for consumers. In fact, a key part of the work of designers is to allow for small islands of meaning to be put together in what would otherwise be an infinite landscape of objects. The resulting ecology of products comes from designers working through excessive information, slimming down the glut, allowing for consumers to have wide yet relatively limited choices. Andrew Abbott (2014) has called this strategy for managing excess and the overload of alternatives ‘value contextuality’, explaining how items are not considered in terms of their unique and fixed contribution, but rather as part of a holistic combination, paring choices down not by looking at items as things on their own but rather by understanding their context vis-à-vis the other items around them.

The work of designers inspires the idea of how to work in the writing process through the related logics of cluttering and shedding. We clutter by pairing sociological problems, descriptions of a scene, provisional hypothesis about the relationships between concepts; tie them to a citation, a testimony, or a character; and evoke some vague reference to the literature and the problem with it we are apparently solving. Of course, this is just a first generative step, which aims to work through the disparate materials we collect when trying to make sense of puzzling data. In this process of adjudication, the work is oriented toward the production of an outline as a simplifying tool, but without ignoring the work of choosing and adjudicating every step along the way. When working on our writing through cluttering, what we do is to also conceptualize it as an assemblage, and ask ourselves: what are the options I am taking in each case? What does it lead me to? While doing, it is also important to remind ourselves that what we are pursuing is not a multiverse like organization of what we saw and heard through writing, but a more thorough understanding of how we focus the material through a few themes – and what we leave behind when we take certain roads, but not others.

We could say, following the works of Becker and Faulkner (2009), Becker et al. (2006), that what we can learn about making shoes for qualitative sociology is to observe how our practices are similar, since they both operate via ‘shedding’, helping us to see the ‘variance increasing activities’ at work in helping us to select, from a varied landscape, what is going to be considered an ‘original’ production. Much like shoe design, qualitative sociology involves a paradoxical relationship between routine practices and originality, in which actual choices are better explained by the context and background-producing procedures, and in which the creation of value is best understood not as a deviation from established routines (Stark, 2009) but rather as an exercise built on routine practices and well-known conventions that signals – through constant repetition and accumulation – that which might emerge as original, and thus distinguishable.

Thesis 5: Causality is built rhetorically

In a seminal article from 1986, late Argentinean writer and essayist Ricardo Piglia explained the appeal of the short story in this way:

In one of his notebooks, Chekhov recorded the following anecdote: ‘A man in Monte Carlo goes to the casino, wins a million, returns home, commits suicide’. The classic form of the short story is condensed within the nucleus of that future, unwritten story. Contrary to the predictable and conventional (gamble–lose–commit suicide), the intrigue is presented as a paradox. The anecdote disconnects the story of the gambling and the story of the suicide. That rupture is the key to defining the double character of the story’s form. First thesis: a short story always tells two stories.

I take Piglia’s first thesis, on how the art of the short story writer consists in knowing how to encode Story 2 in the interstices of Story 1 to reflect on how much doing good qualitative sociology resembles, in manifold ways, the writing of short stories at large, and of crime stories in particular (on this, see Boltanski, 2014). I emphasize here the central place of writing when it comes to making sense of qualitative sociological data and its organization. Unlike the Galilean method with its emphasis that every piece of evidence be directly observable and repeatable, the resolution of crimes, and the work of the sociologist are conjectural. Following cultural historian Carlo Ginzburg (1980), what I mean by this is that these case both are methods of following and presenting clues. As such, the logic in place is one in which we are not looking for general laws but rather by how a singularity can be explained. Looking for clues, for traces, becomes the common way to see how is it that a series of surface phenomena authorizes us to produce a deeper reading, far more revelatory than the description of just one detail. ³ Our knowledge as qualitative scholars is conjectural, made out of signs, and scraps of evidence, we use an interpretative method, taking marginal and irrelevant details as revealing clues for what reality is really about.

In this ethnography as short story writing, it is important not just for the locales and people to ‘come alive in the writing’ (Narayan, 2012) but to show the data that make the connection between our case and our claims. The narrative avenue I am describing here makes of the opening scene – regardless of whether for a book or for an article – an important building block to anticipate (a) how puzzling our data are and (b) how much we are working on coding into the story about the disparate materials we found on the field, the analytics we will be advancing throughout our written work. In this relationship between surface and depth, we establish early on what we are showing, what we are rendering visible, what we are leaving behind on the background (more on this later) and how the opening vignette connects to what we will be describing at large, and developing as a research question.

In this vein, I deem exemplary Auyero’s (2012) continuing line of investigation about the urban poor in Argentina (Auyero and Swistun, 2009), in which he shows how poor populations are produced as patients of the state and how they are made confused about the toxic waste they inhabit by diverse state and non-state agents, as well as the recent work of one of his former students, Pablo Lapegna (2016), who shows the concerted work of provincial political leaders, state bureaucrats, and local social movement leaders in actively producing the de-mobilization of peasants against the toxic consequences of the soybean boom in northeast Argentina. In both cases, their work begins with a puzzle or paradox that needs to be unraveled. The conundrum in both cases has the shape of a paradox, they ask: how come when exposition to toxicity is so transparent and obvious, those suffering from it have such a hard time mobilizing against it? The opening scene invites – yet again – for a causality that is rhetorically built in the work of upsetting our expectations, much like in the example of the suicide at the Monte Carlo casino – and pushes us to make sense of a scene that looks at first glance counter intuitive. ⁴

There is – at least – a second way in which we can link through our writing the connections between pointed data and our larger sociological claims. Data presentation works better when shown, not skimping on evidence, than when just used as confirmation of our larger analytical points (as in the adagio ‘show, don’t tell’). But, this also brings in the question of when and how to foreground and background different themes in our work, until a focus emerges. Using again my recent work as an example of these exercises, this meant that sometimes the book provided copious detail, but that in others it presented the process abstracted instead. Some of these choices have to do with the tension between opening up the description too much versus using the material to advance an argument; sometimes it was about the dullness and opacity of doing ethnographic work on these settings and how little some of the scenes ‘communicate’ on their own. In fact, this is what happened at the beginning of the project, as I was trying to make meaning in traditional ethnographic ways of designers sitting in a room, looking at their screens while listening to podcasts and music on headphones. Without access to their exchanges, there was little to do; without being present in their meetings, there was little to communicate. The same thing happens – albeit differently – in the scenes where shoes are being fitted by technicians on fit models: there is nothing paradigmatic, revealing, or ‘charismatic’ about those moments; it is only the accumulation of scenes and their construction as a process that helps the reader understand how fitting works and why it matters to projects of scale-making. Thinking both thick and thin means that sometimes the aim is not the richness of data but rather what Brekhus et al. (2005) call ‘thin description’, aiming analytically to extract general forms from particular contents, committing the ethnographer to focused observation and developing analytic aims on the available data.

Thesis 6: Who the researcher is and what they do is a central vector in knowledge production

What would happen if we thought of ethnography as a particular kind of encounter, that between self and other, between a teller of tales and a listener of stories (Tsing, 2005), and we double down on this take, and think of it then as a folie a quatre? In it, two selves represent themselves and each other in an asymmetrical exchange. We would then need to seriously think about the role-played in the production of knowledge by how we represent both others and ourselves. In our role as ethnographers’ symbolization participates in how we are seen, in what ideas we bring with ourselves, in the performative dimension of how we are able to present ourselves (friendly, secretive, and ethical, as Gary Fine (1993) wrote 30 years ago) and on how we imagine the others (given the logical impossibility to know another person’s experience, pace Weber we can only reconstruct the public meaning of actions, accounts, and gestures).

There have been three ways that depart from the metaphor of the encounter, aiming to cancel, or control one of the dimensions of participating in it. The first one was embraced both by post-colonial and early feminist texts. In these texts, the difference inherent to the ethnographic encounter was made into an aporetic narration of the difficulties caused by the cultural distance and power asymmetries between the researcher and ‘subaltern’ subjects and resulted both in the silent contemplation of an object whose remoteness renders it un-nameable; and in the celebration of the lack of control over communication though a whole series of reflections on writing tropes, and the self of the ethnographer. If we could not really learn about otherness, why bother! The second one, standpoint theory and strong objectivity stood like the inverted mirror of positivism, reproducing the idea that reality could be represented without distortions – trying to escape the bewitchment of language – by valuing positively everything that had been deemed negative and discarded before: intimacy, emotion, involvement, experience, and partiality. We could know, but only by aiming to dissolve the distance between subject and object. The third one has aimed to make of qualitative sociology something akin to other positivist-based science, aiming to produce theory prior to data, develop testable hypothesis, and control the terms of the encounter both at the context of discovery – by eliminating subjectivity at the door – and explanation – by producing uniform accounts understandable by most working colleagues. ⁵

Gianpaolo Baiocchi has recently advocated – to stop producing paranoid readings of the natives’ experience, in which power is always the culprit for what we find on the field. The restorative reading he proposes – following Eve K. Sedgwick – would help us to be faithful to what the locals’ say and yet be able to criticize it on its own terms, since they both know and do not know about who they themselves are, and why they do what they do. Representational practices are present also at the level of the ethnography: in the larger narrative forms and cultural tropes subjects use to present themselves in front of the researcher (see Portelli, 1997); in how they imagine us (Venkatesh, 2001); how they imagine we imagine them; but – more importantly – how they imagine we would like them to be (what Pugh (2013) has called ‘the honorable interviewee’). How we are seen is not just the product of macro categories of classification and experience (cis-male, white, elite, etc.) but rather better accomplished by reflexive situational understanding of who we are and how does that organize how we are able to produce knowledge (on this, see Lichterman, 2017); a knowledge that is always historically situated and conjectural, but not arbitrary because of that. While we engage with the empirical material in multiple potential ways, what is real resists. To paraphrase Howie Becker (2007), discussing the work of Bruno Latour, we can say the moon is made out of green cheese, but the moon has to collaborate! In our interpretations, we need to enroll the data in convincing ways, exhibiting the characteristics and behaviors we attribute to it.

All of this is not new, but I want to emphasize that thinking like this means to understand that ethnography is always an exercise in miscommunication, in which: (a) total control about what is being communicated is impossible and (b) understanding ‘what is really going on there’ is more a horizon of intelligibility than a potential to be fulfilled. Recognizing this should not stop us and force us to think of knowledge as a total impossibility but rather to push us into three directions: (i) to understand that we are aiming to objectify the words of the others, to borrow what they say and do, and systematize it; (ii) a reflection on the validity of how we construct these encounters as data; and (iii) an opening up of the menu of theoretical options available when doing ethnography. All this results in a reflexive account not of just who we are – as in the diverse versions of positional reflexivity – but rather on a deeper understanding how limited our knowledge is and how our practices can be validated, objectified, and studied. Our knowledge is productive precisely because it is limited, not despite of it!

The advantage of qualitative work at large, and of ethnography in particular might be precisely its incomplete and limited character (Candea, 2010), an admission of the impossible closure in the business of producing description about otherness. And this happens by design; we don’t record the totality of social life in a particular place. We organize a reality that is multiform, complex and contradictory, according to (and as such limited by) the questions we want to answer. It is this limitation – the fact that we are limited by theory, language, and selfhood – that actually allows us to produce this kind of knowledge. This article aimed to make explicit all the choices we make along the way. I want to emphasize here how much theory and methods usually constitute a package (Clarke and Star, 2008), in which theory relates strongly to how we conduct qualitative sociology. Each dimension followed by scholars also highlights one particular road for data production.

Conclusion