Using Photographic Data to Build a Large-scale Global Comparative Visual Ethnography of Domestic Spaces: Can a Limited Data Set Capture the Complexities of ‘Sociality’?

INTRODUCTION

This paper looks at the implications of a research approach that might appear to be un- or anti- ethnographic in some of its characteristics, and uses the conference’s focus on ‘sociality’ as a way of evaluating the potential of this approach for delivering what we expect from ethnography.

The research approach in question is a database of photographs of twenty households in each of twelve countries. It was developed to be part of a suite of mainly quantitative global consumer trends research products offered by global market research company GfK NOP. It is called the Visual Survey of Domestic Space, or VSDS.

The paper goes into some detail to describe the genesis of the VSDS. The business context for the development, sale and application of this approach frames up its potential vices and virtues. It then considers a couple of ways in which the VSDS has been used in order to look at how the ‘limited’ nature of the data set can play a role in capturing the complexities of sociality.

SOCIALITY: THE NET WE CAST

As the theme of this conference, ‘sociality’ is well chosen to strike towards both the heart of our theoretical roots and our best intentions for the effects of our practice. The fact that we are a broad community, drawing on a wide range of theoretical traditions and embracing an equally wide range of situations for practice, presents challenges for how we conduct our conversations in this academic form. Not only do we not necessarily share theoretical points of reference, but for many of us the point of being professionally situated where we are is that we have embraced the fact that we can know stuff about the world without necessarily having to find the appropriate place for that knowledge within an academic discourse. As a community of practice, we have a different relationship to theory than within the academy. We always have half an eye on our close theoretical kin, our fellow practicing ethnographers, and half an eye on the light that our theoretical conceits can shed on the preoccupations of others with whom we work who come from quite different intellectual traditions, or from none.

Being not entirely of the academy may free us somewhat from what Marilyn Strathern calls the “doubly snared perplexities” (Strathern 1988: 342) of sociological epistemology, but out in the world our role is often to convince those with whom and for whom we work that parts of the world that matter to them are more complex, and more rewardingly difficult; and that their ends will benefit from applying more of the special kind of attention that we bring. In our work we aim to ensnare the imagination of clients and collaborators, drawing on our theoretical roots for our nets.

Figurative nets, or actual ‘networks’, are probably the most prevalent metaphor that as social scientists we use in the wider world to give people a sense of what our view of the world is all about – its ‘sociality’. So for the past several years I, a dry British social anthropologist rather than a Geerzian, have used oft quoted words from The Interpretation of Cultures to communicate to clients where it is that our practice is coming from (and where it is going):

Believing, with Max Weber, that man is an animal suspended in webs of significance he himself has spun, I take culture to be those webs, and the analysis of it to be therefore not an experimental science in search of law but an interpretive one in search of meaning. (Geertz 1973: 5)

I owe my use of this quotation in the context of my work as a practicing ethnographer to my colleague Rick Robinson – in so far as I have stepped into the telling of a story that he started at E-Lab and that has run on through Sapient’s Experience Modeling discipline to our current practice of ethnography within the global market research company, GfK NOP. I cite this is an example of one type of relationship to theory that we have as practitioners of ethnography within the business world. We make use of bits and pieces provisionally. We share tit bits with one another. Whereas within the academy we might have a far stronger sense of where our colleagues sit in relation to certain discourses, and engage them on the specifics of those positions, in our broader community of practitioners we more often offer up gifts: “you might find this useful”.

Our relationship to theory is often thus more about positioning ourselves with each other, and in the world in which we work, than about advancing discourses. Our shared canon of theory is more likely to be used to show what we know about the world than, as in the academy, to test the robustness of that knowledge. In this vein we may draw on Mauss (Mauss 1990) to emphasize the shared construction of meaning through exchange, or Bourdieu (Bourdieu 1984) for the social work involved in maintaining people’s identities and differences, or Miller (Miller 1987, 1994) to point to the transformative agency of consumers. As a young and as yet only partly ‘professionalized’ profession we have a strong interest in origin stories; tales of where we have come from and where we have been along our way. So works that chart the history and evolution of our practice such as Byrne and Squires’ collection of anthropology and design collaboration stories (Byrne and Squires 2002) have a visceral appeal.

Our interest in the situations in which we as a community of practice have operated is apt, because it is in the sociality of the contexts for our work that our theories and our methods have life. And thus this paper is partly about what flows from the decision to use a particular method – predominantly visual data collection – within a particular context for the practice of business ethnography – global ‘consumer trends’ market research.

BUSINESS CONTEXT OF THE VSDS: ETHNOGRAPHY AND MARKET RESEARCH

The team of business ethnographers with whom I work has recently moved from the application of ethnographic research in technology design consultancy to the world of market research. There are broad differences in these contexts that are relevant to the application of the method described in this paper. In the former situation, as is often the case in design research, the ethnographers were almost invariably the only researchers involved in the process, meaning that the ethnography that they practiced was synonymous with research. In the team’s current, market research, context several different research methods co-exist, the main lines being drawn between quantitative and qualitative research, with ethnography playing a not entirely comfortable role as the exotic cousin within the qualitative fold. Ethnography is a more recent entrant to the world of market research than it is to product and service design or work within organizations, and though there are certainly examples of its thoughtful and appropriate incorporation within the repertoire of qualitative market research techniques (Mariampolski 2001), it is often proposed merely as a method for delivering ‘consumer closeness’ rather than for its analytical strengths.¹

It was partly in recognition of the fact that the structural and analytical center of gravity of the market research world lies in large scale comparative quantitative studies (and partly from the having experienced the frustration of the difficulties of effectively integrating ethnographic and quantitative research when working in a design research context) that the newly arrived ethnographic practice at GfK NOP embraced the opportunity to develop a market research product that would attempt an unprecedented integration of ethnographic and large scale quantitative approaches.

Working closely with researchers operating within quite different paradigms and epistemologies begs questions of who controls the eventual interpretation of the range of research methodologies applied to a given problem; questions that are less pressing when ethnographers are the only researchers around. In market research, as can be the case in ethnographic studies of organizations (in mental health contexts, for example) ethnographers do not have the monopoly as professional describers and theorizers of social life and personhood. (Pulman-Jones 2001) Turned to unintended ends, our methods may undo the best intentions that we have towards capturing ‘sociality’. The danger when integrating an ethnographic approach with a large-scale global quantitative research approach is that data produced by the ethnographic methods will get applied within the quantitative paradigm.

Global consumer trends experts at GfK NOP map differences in the ‘values profile’ of consumers in the global consumer trends subscription product, Roper Reports Worldwide, which surveys a thousand people in each of thirty countries, annually. Value profiles are arrived at through a questionnaire that asks people to rank a standard set of over sixty-five value statements, either as personal priorities or as associations with a range of global brands. Responses are mapped to a classic consulting two-by-two with corners that oppose ‘People’ and ‘Power’ on one axis and ‘Fun’ and ‘Tradition’ on the other. The data can be cut many ways to demonstrate broad differences by country or by consumer profile, revealing, for example, that consumers of a particular brand tend to prioritize creative and altruistic values. For global marketing and product design strategists this tool provides a powerful high level scan of their market terrain, identifying specific populations for further investigation. It also provides a commentary on trends in consumer values, the 2005 data showing, for example, that the values profile of consumers in rapidly growing emerging economies such as China is, as might be expected, shifting from an emphasis on Tradition to an emphasis on Striving. (Striving is defined by prioritization of wealth, power and status.)

Powerful as this tool is, from our perspective as ethnographers interested in the sociality of local contexts it might be seen as collapsing sociality into individuals, with the complexities of local sociality rendered as globally referenced attributes of individuals as, for example, in reported incidence of ‘traditional values’ noted above. As the ethnographic component of Roper Reports Worldwide, the Visual Survey of Domestic Space, was thus conceived as an appropriate complement to the quantitative study, providing a lens from the global comparative level of the quantitative data sets back into the particularity of local contexts.

However, applied outside the guiding intentions of the ethnographers, might the visual data set be a double-edge sword? Rather than revealing the complexities and connections that comprise sociality, might it be put to the service of the effacing of social relatedness that Strathern has lamented as characteristic of late-modern life?

GENESIS OF THE VISUAL SURVEY OF DOMESTIC SPACE

The VSDS was developed as a direct response to a specific business problem. The business problem, and the market research agency / client relationship in which it was communicated and framed up, were pivotal in defining this new research approach.

The business problem was identified in the course of ongoing quantitative customer satisfaction research for one of GfK NOP’s global clients. An important attribute of this company’s products for the home is style. GfK NOP’s research had identified an interesting group of people who have a positive attitude to the company, like the experience of shopping in its stores, but nonetheless purchase very little because, “I couldn’t find anything that matches my style”. On the face of it, here was a straightforward opportunity for the company to increase its sales. A group of customers are already well disposed to it: it merely needs to work out how to provide them with style options that meet their requirements. But what do people mean when they refer to, “my style”? And what is the relationship between these potential customers’ sense of their own ‘style’ and the repertoire of styles presented by the company in its stores?

Style definition, and style preferences had proven difficult to define through quantitative market research. The company had developed a framework for categorizing the styles of its own products, which drew on a small set of seemingly fundamental oppositions such as modern vs. traditional. They had arrived at a set of core style groups, each with several sub-types. They also specified a set of ways in which these style groups could be combined with each other. It wasn’t a simple matter for the company to come up with a framework for defining and categorizing styles that would usefully enable co-ordination of product development and design groups. But defining a set of style categories as a part of company practice and policy was a quite different problem than the problem of understanding how ‘style’ played a role in the experience and understanding of the company’s customers who, whilst they might have no notion of how to define their own style if asked, could still know with certainty, “that’s not for me”.

Discussions about this problem between the company and the account team at GfK NOP coincided with the launch of the Observational & Ethnographic Practice within GfK NOP, prompting the idea that ethnographic research might provide an appropriate approach. This use of ethnography thus has to be understood both in the context of the client’s problem and that of furthering GfK NOP’s relationship with the client through adding value to and integrating with the ongoing pieces of research GfK NOP conducted as part of the relationship with the client.

The range of benefits perceived within GfK NOP for this ethnographic addition to an existing quantitative research product is instructive, and echoes longstanding debates within anthropology about the status of visual data as illustration. As in other areas of professional business services such as technology consulting to which ethnography has been added, ethnographic research can be positioned as a novel differentiator that provides rhetorical evidence of ‘added value’ in the services marketplace. The first and most straightforward way in which this value could be seen as being added was through the ability to illustrate the quantitative trends analysis. Where, for example, a consumer trend towards regarding the home as a ‘haven’ had been identified, photographs would be available to illustrate the incidence of homes-as-havens, by country, or by a given consumer profile, lending rhetorical weight to the quantitative findings. By contrast, for the ethnographic team within GfK the appeal of the new approach lay not so much in the ability to provide illustrations to support quantitative research findings, but in the prospect of new analytical possibilities, the challenge of integrating the visual data with the quantitative data, and the potential for subverting conventional assumptions about the nature of ethnographic data sets.

The anthropological debates about the status of visual data on which the role of the VSDS touches are nicely framed up by Glenn Bowman as sitting between a position articulated by Margaret Mead, arguing for the irreducibility of images to text (Mead 1975: 3-12) and Kirsten Harstrup (1992: 8-25) arguing that the image, “is merely the surface level of phenomena which cannot ‘speak’ for themselves but must be ferreted out through textual contextualization and exegesis”. (Bowman 2001) Against these two poles of the debate, Bowman proposes an alternative possibility for the role of visual data in anthropology that, as we shall see, characterizes well the best practice application of the VSDS data:

the photograph in anthropology is as much a means of discovering information as it is of presenting that which has been found… a locus for dialoguing rather than as a source of information in itself. The value of the image in ethnographic fieldwork is here precisely in its indeterminacy insofar as that allows processes of interpretation… to go on around it. (Bowman 2001)

The methodology of the VSDS does not allow for discussion of the visual data by informants, but the VSDS photographs serve a similar function for hypothesis generation and analysis amongst the researchers. Bowman’s diagnosis of the value of the mute foil of the indeterminacy of the photograph is borne out by our experience with the VSDS data.

EVOLUTION FROM METHODOLOGY TO MARKET RESEARCH PRODUCT

Two over-riding requirements determined the way in which it was decided to provide ethnographic input to address the identified problem of style preference: the fact that style preferences needed to be understood across a global market presence; and the fact that in order to be actionable the ethnographic component needed to be effectively integrated with GfK NOP’s existing suite of global quantitative consumer trends research products.

On the one hand this presented itself as a straightforward case for the application of an ethnographic approach: a comparative exploration of the role of style in people’s everyday lives – both in terms of their mental models and the material artifacts through which style is expressed. But on the other, there were constraints that meant that an approach based on conventional in-home ethnographic studies would not be viable, being too costly, taking too long to execute, and generating too much data to allow straightforward integration with the existing qualitative data sets.

As large as possible a sub-sample of the 30 countries surveyed by Roper Reports Worldwide was desired for the study. A conventional in-home ethnographic study in a large enough number of homes in ten or more countries would be prohibitively costly. In order to achieve the required scale cost-effectively, data collection would need to be simplified in order to limit fieldwork time and complexity.

The options for simplifying data collection were through using brief, tightly-focused in-home interviews, focusing on how style is understood: or through direct visual documentation of variations in the ways in which people’s style manifested itself in their homes. A conventional ethnographic study would, of course, encompass both of these components, but the need to make the study cost-effective and scalable meant that to do both was not an option.

A tightly defined set of photographic data documenting the home was chosen because it offered the potential for close integration with the quantitative data through a relational database, and because it would allow a simple research design that could be executed by researchers with a wide range of skill and experience levels, as might be required if the data set was to be extended to larger sample sizes in a growing set of countries in subsequent years. (In the event we were able to use experienced ethnographic researchers in each of the twelve countries in which we conducted the first round of the study in 2005, through our global ethnographic network partner Social Solutions.)

The problem of ‘style preference’ had been the catalyst for this new research approach, but in order to match the broad consumer trends scope of the main quantitative study the VSDS needed to be extended to be about ‘domestic life’ more broadly. The impetus for using photographs as the primary data source had come from one company’s business problem: style preference – but the study itself needed to comprehend more than the ‘manifestations of style’ that had been the initial scope around which the concept was developed. A ‘limited’ research method (or form of data collection) needed to be optimized to provide input to as wide as possible a range of consumer research issues.

What then needed to be photographed in order to provide useful data on ‘domestic life’? At the most basic level it was clear that when designing a study that could be undertaken anywhere in the world, the guide for the researchers could not rely on specifying photographs to be taken in named rooms, given variations in how domestic space is divided up and named. So the visual data needed not to be structured around specific rooms or spaces, but on the areas within the home in which a core set of domestic functions are carried out, or ‘functional areas’. Eventually nine ‘functional areas’ were defined:

• Food preparation
• Food consumption
• Relaxation
• Home office
• Hygiene
• Sleeping
• Service and maintenance
• Where vehicles are kept
• Center

Researchers would ask participants to show them the areas within the house where the activities above take place, and then take a set of photographs of each area. The ninth category allows participants to indicate the area that they considered to be the ‘center’ of their home. The intention is that participants are not included in the photographs.

WHAT POTENTIAL DOES THIS APPROACH HAVE FOR CAPTURING SOCIALITY?

The risks inherent in this approach stem from the fact that it can assist in the pulling of individuals out of their local context to be lined up for comparison across a global set, for the aggregation of trends in individual attributes. The material culture of homes (the visible stuff) might be reduced to ‘possessions’ – tokens of the consumers, mutually identifying them as owners of a given product or brand, or falsely attributing intention and choice to the incidence of objects within the home. At it worst the material contents of the home become mere consumer inventory.

On the other hand, the limitations of the data set compared to a conventional ethnographic data set (one that provides a holistic view of a context for experience) bring with them potential benefits. The fact that the data set does not pretend to provide a total account of a social context forces an approach that recognizes the limitations of any one data collection method and requires a hypothesis driven approach. Sometimes it is useful to be freed from the tyranny of the completeness of ethnographic data sets, whose very completeness – the fact that seemingly any question can be answered by the data – can tempt away from theory and analysis towards mere reportage. (Not to mention the fact that the data sets of ‘rapid’ commercial ethnographies are rarely as ‘holistic’ as we wish them to be.)

In addition, in the context of market research, it may be a radicalizing move to remove the individual consumer from the picture and force a focus on the material culture that surrounds them and insists upon the fact that they inhabit shared contexts.

How does the balance of these benefits and risks play out in practice?

STYLE PREFERENCE

On the face of it there are a significant components missing from the data set for an appropriate understanding of the role of style in people’s lives. The visual data only show what is in the interior of the home. The only data gathered from the participants is in the course of getting them to indicate where each of the nine domestic functions is carried out within the home. The visual data itself provides no direct understanding of why items are there. Are they chosen, inherited, found, tolerated, cherished, or chalked for removal at the earliest opportunity? Are they shared, or doggedly owned and protected by just one member of the household? Are they ironic, or sincere?

But it is the very fact of what we know that is ‘missing’ from the data that forces us to develop different perspectives on style and new hypotheses for understanding how it manifests itself in people’s homes.

Awareness of what we are not able to identify from the visual data led us to focus on fundamental aspects that we might be able to identify. Given the difficulty of providing a single consistent set of style categories by which our household data could be organized, such as ‘modern’, ‘country’, etc., we focused on the basic characteristics of domestic space that provide the backdrop and context for style.

In the absence of a direct route to characterizing style preference through the VSDS data, we attempted to establish other attributes reflecting care and intention in relation to the domestic environment. Looking at photographs from different homes it was clear that several of us might be able to agree that a particular room was ‘messy’, but that messiness was a complex category that in many cases became a subject of debate: one person’s ‘messy’ is another person’s ‘lived-in’. But further analysis did lead us to realize that we could reliably identify some attributes of the way that domestic space is disposed. We discovered that it was possible reliably to categorize spaces in terms of their level of order, and in terms of object density. Spaces with high object density and high order were ‘neat’; whilst spaces with high object density and low order were ‘messy’. These were the first steps in a process of developing a framework for classifying variations between the disposition of different domestic spaces, and in building up to higher order classifications such as degree of ‘style intention’.

Whilst there was no escaping what the VSDS lacked of what would be expected of an ethnographic data set for the understanding of style preference, in this case necessity was the mother of invention, spurring us to identify relevant dimensions of care and intention in domestic interiors that we might otherwise have missed.

HOUSEHOLD TYPES

Another example of the paradoxical advantages of the limitations of the visual data set is provided by its application to interrogate the validity and effectiveness of a global household segmentation model built on a combination of household composition and lifestage for understanding the differences between household interiors.

The segmentation model in questions has rhetorical plausibility, given that it is reasonable to expect that the primary way in which household interiors would differ would be according to the composition and lifestages of the households occupying them. However, reliance on the solely visual nature of the data in the VSDS allowed a radical interrogation and reframing of the assumptions on which the segmentation had been built. Being able to talk with members of households in our data set about their lives in their homes might have reinforced the segmentation’s assumption that household composition is a primary and defining characteristic; that the home is the way it is because of the activities of household members and the dynamics between them.

What became apparent when the images themselves were interrogated was the that apart from evidence of the presence of children of different ages in the home, the contents, style and disposition of space in homes differed less by household composition and lifestage than by socioeconomic status, country, region and other cultural factors.

In this case the primacy of the visual data enforced by the VSDS led to recognition of fundamental aspects of sociality (material conditions; material culture) that might have received less emphasis in a data set that was dominated by the personal narratives of individual participants. Rather than focusing on meanings articulated in participant narratives, and on the consciously co-constructed aspects of the home, as would have been likely with conventional ‘rapid’ commercial ethnographic interviews, analysis had no choice but to focus on material aspects of the home that are often overlooked in studies that privilege the narrative articulation of participants. In this case there was a clear benefit from the enforced separation of visuals from participant narratives in the ethnographic data set.

Conclusion

This paper has described the development and application of an ethnographic research approach that runs counter to conventional ethnographic practice in that it uses a highly focused and selective data collection method to construct a data set consisting solely of photographs of domestic interiors.

Such a data set forces recognition of its limitations, and as such can provide a useful means for generating new perspectives and hypotheses on research problems. A forced emphasis on material culture can surface aspects of sociality that are sometimes under-emphasized in conventional ethnographic data sets dominated by participant narratives.

Whilst this paper has noted that visual data sets such as the VSDS could potentially be used to support accounts that remove individuals, as repositories of consumer choices, from their social contexts, in closing it is interesting to observe the extent to which in practice the visual data set is acknowledged to be the sole domain and property of the ethnographic research team.

As soon as the first sets of photographs started to come in from the field the ethnographic team made rapid forays through the data to pull out brief picture sets that could provide a taster of what the database might eventually provide. These were somewhat speculative stories about how domestic functions as defined by the VSDS ‘functional areas’ overlapped differently within homes in different parts of the world. One of the first quick-and-dirty stories that emerged was of laundry experiences differently inflected through overlapping with student study in a shared student house in the UK, with wine-storage in the backstage provisioning of the pleasures of a bourgeois home in northern Italy, and with purification of drinking water in a kitchen in Mumbai. These visual stories proved to be simple and compelling ways of bringing to life the basic proposition of the VSDS with clients. Yet despite the fact that the stories told around the pictures were simple and easily remembered for re-telling, the quantitative consumer trends experts with whom the ethnographic team worked were reluctant to use the stories in client-meetings when members of the ethnographic team could not be present, insisting that they couldn’t tell the story the way the ethnographers could.

Whilst as straightforward illustrations the value of the photographs might appear to non-ethnographers to be self-evident and readily available, in combination for the construction of narratives or propositions they become complex puzzles, emphasizing rather than masking the ethnographer’s skill and labor. Armed with an understanding of what can be told via the pictures’ underlying ‘functional area’ structure, for the ethnographer the photographs may thus become “technologies of enchantment” as defined by Alfred Gell – like the ornate prow configurations of Trobriand Island canoes used for Kula expeditions, at once simple and beguilingly complex:

I am impressed by works of art in the extent to which I have difficulty … in mentally encompassing their coming-into-being as objects in the world accessible to me by a technical process which, since it transcends my understanding, I am forced to construe as magical (Gell 1992: 49)

The final paradox of this ‘limited’ form of ethnographic data may thus be that, compared to participant narratives that we serve up, or especially to illustrative video clips, those with whom and for whom we work may be less likely inappropriately to co-opt the photographic data set to define people in ways that are insufficiently grounded in the complexities of the sociality of their lives.

Simon Pulman-Jones is European Director of Ethnographic Research for global market research company GfK NOP. Following work in documentary film, and a Social Anthropology PhD at the LSE, Simon joined the London Business School Future Media Programme before moving to the US to join Sapient Corp.’s Experience Modeling discipline.

Ethnographic Praxis in Industry Conference Proceedings 2005,pp. 128–139. © American Anthropological Association, some rights reserved.

NOTES

I would like to thank Nina Wakeford for her comments on earlier drafts of this paper.

¹ Dominated by the psychological paradigm of depth and interiority, conventional qualitative market research does not have easy affinities with social and cultural anthropology’s attention to networked complexities of meaning and structure. The author has described a categorical lack of affinity based in the same epistemological opposition, between the psychiatric team and the continuous care team in a therapeutic school unit for children with emotional and behavioral difficulties. (Pulman-Jones 2001: 130)