Reflections on the Use of Visual Methods in a Qualitative Study of Domestic Kitchen Practices

Introduction

This paper reflects on the use of visual and non-visual research methods and their incorporation within a study of everyday social practices that focused on the domestic kitchen. The aim is to consider the challenges and benefits of such an approach in an applied sociological study that was underpinned by current theories of practice. We begin with a brief discussion of theories of practice before outlining the reasons that visual research methods might be used in a study that draws on such theories. Following a discussion of the study’s research design we examine what is gained from the use of visual research methods and from integrating their use with non-visual methods. The challenges inherent in this approach are considered before the paper concludes whether and how a visual research approach can contribute to an applied study of social practices.

Although the history of utilising theories of practice to examine everyday life extends some 30 years or so, (Bourdieu, 1977; De Certeau, 1984; De Certeau et al., 1998), the so-called practice ‘turn’ (Schatzki et al., 2001) in social theory is more recent and has been particularly popular within scholarship on consumption and sustainability (Halkier and Jensen, 2011; Hargreaves, 2011; Pink, 2012; Shove et al., 2012). Current theories of practice take into account three key domains: the people who might shape or perform practices (the carriers of practices [Warde, 2005]); the available resources (including technologies and ‘things’) and the meaning given to a practice through shared and individual histories, experiences, beliefs and values. Theories of practice are useful when exploring complex events as they seek to privilege neither agency (through over-emphasising individual behaviour) nor social structures (such as the influence of class, gender or ethnicity) (Halkier and Jensen, 2011). Theories of social practice inherently account for the entangled nature of relational events, things, people and the places and context in which they interact and occur, albeit via the reflexive gaze of the researcher (Linderson, 2010). This perspective can help to tease out both the tacit and the discursive elements of a practice (Reckwitz, 2002) through a close-up examination of the ‘doings and sayings’ of social life (Schatzki, 1996). Adopting this theoretical approach also reduces the potential to foreground people or places or things as each of these aspects, and others besides, are considered as constituent parts of an overall jigsaw puzzle (Wills et al., 2013). What we as researchers might observe in a participant’s home is the coming together of a multitude of pieces into one grand (yet simultaneously mundane) performance.

In order to reveal the extent, and entanglement, of social practices in an empirical study of domestic kitchens, a methodological approach was needed to facilitate an examination beyond pre-defined activities; in this study this included looking beyond ‘cooking’ or ‘cleaning’, for example, to consider any non-foodwork practices that kitchen life might entail. Our approach was also designed to dispense with pre-conceived ideas about domestic ‘food safety’ or where it might be considered that a kitchen ‘begins’ or ‘ends’. Examining social practices also meant trying to create a distance from common actions and things, such as the turning on of a tap or a kettle, to make the familiar strange and develop a robust contribution to sociological knowledge (Mannay, 2010). There was also a need to make sense of the unfamiliar (Linderson, 2010) in order to help reveal the more uncommon elements of ‘what goes on’ in UK homes for subsequent examination and analysis. In this study, the methods selected needed to help reveal the habitus of kitchen life, to help everyone in a household show and tell their own kitchen stories despite not always being aware of them (Bourdieu, 1990). In essence we wanted an approach that would help to expose the traces of everyday social practice (Pink, 2013) even though such traces are not always apparent to, or made explicit by, those involved in creating them.

Rationale for Incorporating Visual Methods in a Study of Kitchen Life

Speech and text-based methods, including digitally recorded face-to-face interviews and focus groups, often subsequently transcribed into text documents, have much to commend them and are commonly used or incorporated within studies of social practices. Narrative-based spoken or written techniques, in particular, allow participants to talk or write about everyday habits, including food provisioning, preparation and consumption, within the broader realms of everyday life (Evans, 2012; Wills, 2012). Power (2003) notes, however, that when talking about the everyday, participants often have to forego providing a full narrative because they cannot find the ‘right’ words. They often fail to ‘translate’ their narrative for a listening researcher, resorting instead to phrases or rhetoric such as ‘you know what I mean’ to fill in the gaps between what can be articulated and that which cannot. In relation to food and eating, its sensory nature – the touch, smell, taste and sound of food – may also be particularly difficult for participants to convey simply through talking to a researcher (O’Connell, 2012). More broadly, relying on talk to reveal the complexities of domestic life is also likely to exclude aspects of the ‘sensory home’ such as how a room feels to a user (Pink, 2004b) and how this relates to context and meaning (Shove et al., 2012). Relying on techniques such as interviews or diaries also inherently masks both the display or performative aspects of a practice (Halkier and Jensen, 2011) and the intricacies that might exist between those individuals undertaking the performance (Goffman, 1959). As Martens (2012) contends, relying on talk about practices – dishwashing, for example – tends to elicit the meaning and significance of a practice to an individual, as they perceive it, without necessarily revealing more about the practice itself, including the nuanced aspects of how this may vary in different contexts.

Visual methods, which include but are not limited to the use of photography and video or film-making, are considered helpful in studies attempting to lay bare phenomena which are mundane, taken for granted or difficult to articulate (O’Connell, 2012; Power, 2003). A renewed interest in visual research methods within the social sciences (Sweetman, 2009) has coincided with the turn towards drawing on theories of practice to investigate areas of consumption and everyday life. Applying theories of practice empirically requires using methods that can reveal multi-faceted social phenomena; visual methods offer a ‘way in’ to reveal such events. There are an increasing number of studies that draw on visual research methods to investigate domestic practices and consumption (Pink, 2012). Some of these studies have been particularly helpful in relation to designing and implementing the study of kitchen practice reported here and these are therefore discussed below, before we move on to describe our own study design.

First, Martens and Scott’s (2004, see also Martens, 2012) study was innovative in its use of CCTV technology and in its desire to explore kitchen (rather than food-related) practices. The study highlighted the challenges of collecting, and then attempting to analyse, vast quantities of visual data. Whilst few of the households in Martens and Scott’s study agreed to have CCTV cameras installed, a substantial amount of visual data was nonetheless still collected. Their efforts to view and subsequently categorise and quantify the practices which emerged from the viewed footage, by, for example, classifying practices as ‘cooking’ or ‘dish washing’ and so on, proved challenging not least because of the time needed to do this systematically across the dataset. The ‘breaking down’ of practices was something we wished to avoid in our bid to more fully understand ‘kitchen practices’ rather than ‘tasks in the kitchen’ and Martens and Scott’s study influenced our subsequent decisions about finding an alternative and altogether more holistic approach to data collection and analysis, to resonate more roundly with our theoretical framework.

Second, the methods used in a study undertaken by Meah and colleagues (Meah, 2014; Meah, in press; Meah and Jackson, 2013; Meah and Watson, 2011) convinced us to use the ethnographic go-along technique. ¹ The go-along enables researchers to accompany participants in their everyday life, becoming aware of their subjects’ ‘stream of experiences and practices as they move through, and interact with, their physical and social environment’ (Kusenbach, 2003: 463). This technique has a natural fit with visual methods because it is often inherently difficult for researchers to take in everything they see, hear and experience with their participants, as it happens in the field. This often leads to an over reliance on the retrospective writing of (incomplete) fieldnotes. Meah et al. video-recorded and photographed the guided kitchen tours and provisioning go-alongs they conducted with participants though their emphasis was on food-related activities and the dynamics of family life, rather than on the kitchen itself and the wider range of activities that take place there.

Lastly, O’Connell’s study of families with young children (2012), which incorporated drawing, photographic vignettes and photo-led discussions (photo elicitation) using photographs taken by participating children, explored domestic food practices and the influence of parental employment on what young children eat. O’Connell reports the analytic strategies employed in drawing together data collected by different methods and espouses the use of visual data to corroborate, elaborate on and contradict data collected using interviews and other non-visual techniques (O’Connell, 2012). She illustrates how, for example, photographs taken by a child confirm what the child’s mother said in an interview about whether the pair eat together. O’Connell argues further that, because these data reveal additional information to the mother’s interview, the visual data provide not just corroboration but also a richer picture of the family’s life overall. This approach informed our own analytic strategy.

Visual methods, like others within the social scientific toolkit, are not a neutral mechanism for finding things out (Kindon, 2002; Pink, 2004a); they cannot by default be privileged over other ‘go to’ methods (Evans, 2012). When intended for use in a multi-method study like the one reported here, consideration needs to be given to whether ‘more methods’ would lead to ‘more knowledge’ about the phenomena being investigated (O’Connell, 2012). In order to evaluate whether and how incorporating visual methods with other qualitative techniques was conducive to achieving the aims of the study, however, we must first detail the study’s research design and approach.

Research Design and Approach

This paper draws on a qualitative study commissioned by the UK Food Standards Agency (FSA) to reveal ‘what goes on and why’ in domestic kitchens and to inform thinking about how to reduce the burden of foodborne disease originating in the home (Wills et al., 2013). The study was undertaken because there are currently gaps in knowledge and understanding regarding what people do, what they say about what they do and what they know about food safety (Wills et al., 2013). The aim was to undertake a mixed-method qualitative study to complement the findings of other work the funder had commissioned, including a survey and a review of relevant literature (Greenstreet Berman Ltd., 2011; Prior et al., 2011, 2013).

In keeping with an ethnographic approach, which typically uses several methods to collect data (Hammersley and Atkinson, 1995), we were interested in trying to incorporate visual research methods to flesh out the multi-dimensional aspects of kitchen practices rather than relying on ‘just words’ (Harper, 2003). Situating interviews within an observational and visual approach (Evans, 2012) was particularly important given that the funder wanted evidence of actual as well as reported practices. A proxy for what actually happens in the kitchen (Murcott, 2000), through a reliance on verbalised accounts, was therefore not going to be sufficient. Rather than using visual methods to merely provide an illustration or visual description (Harper, 2003) of kitchen practices, however, we wanted to incorporate a visual approach to inform the overall data collection, analytic and interpretation processes (De Certeau, 1984). Visual methods, namely kitchen maps, photography, photo elicitation and video observation were integrated with interviews, kitchen go-along tours and the occasional use of written diaries, logs and scrapbooks. We detail all of the methods used here in order to highlight how visual methods were incorporated into the overall study design and to then evaluate their role. Before this, we first outline who took part in the study and the consent processes that we used.

Twenty households were recruited to participate in the study from a database of respondents who took part in an earlier FSA survey about food safety. These individuals had agreed to be re-contacted about participation in future research projects commissioned by the FSA. Age (less than 60 years; 60–79 years; 80+ years) and pregnancy status were the key selection criteria as the funder was interested in groups thought to be particularly vulnerable to foodborne illness. Following a pilot phase, fieldwork with each participating household involved 3–4 visits lasting 1–4 hours. In acknowledgement of the time commitment needed, households were told at the outset that they would receive a voucher for £100 from a store of their choosing, at the final fieldwork visit. The participating households each included 1–4 individuals aged 2–87 years. Two women were pregnant with their second child. Approval for the study was obtained from the University of Hertfordshire Health and Human Sciences Ethics Committee.

Consent to participate was considered to be a process rather than a one-off action (Dewing, 2008; Lawton, 2001) so most participants signed a consent form at the outset but consent was also discussed at subsequent visits. We took the position that it is good practice to consider the capacity of an individual adult or child to give their own consent to participate, based on them having sufficient understanding of the research and what is expected of them (Alderson, 2007; Wiles et al., 2005). At the first fieldwork visit the study was explained to every individual in each household and they were invited to ask questions before written consent to participate was obtained. Children aged 7–17 years gave their own written consent, in addition to consent being received from their parents. Three children aged 2–6 years did not have the capacity to understand the consent form and written consent was therefore only obtained from parents. These younger children were, however, able to articulate their agreement or otherwise regarding participation, through refusing to be filmed or not speaking in front of the researcher, for example. At the second and subsequent fieldwork visits the researcher asked all participants if they were happy to continue in the study and whether they had any additional questions. Hand written additions were made to the signed consent forms when specific issues arose during the study; when participants filmed their own video footage, for example, which was not foreseen at the outset. Having some awareness of the challenges associated with the dissemination of visual images (Schatzki et al., 2001; Wiles et al., 2011), households were informed at the outset that data would not be published if faces were visible or if the household could be identified in any other way from the data. This was revisited with participants in regard to obtaining consent to show selected identifying images to the funder during the reporting process. Further emergent issues regarding consent and the value and use of publishing images that identify participants and their homes are discussed later in the paper.

We wanted to find a way of ‘getting to know’ participants whilst examining the kitchen and its meaning and place in each household in order to understand more about the ‘micro implications and possibilities’ (Linderson, 2010: 31) of their everyday practices. Whilst the kitchen space might be a room or rooms (or parts of a room) in a dwelling and might include cupboards, work surfaces, a sink, oven and other technologies, how individuals inhabit these spaces (when and how a space is given meaning and becomes a place [Kusenbach, 2003]) was of interest within our overall aim to examine social practices. The first visit to each household therefore included a tour of the kitchen and its associated spaces, in which all members of the household were invited to be involved. This go-along activity helped to build rapport between the researcher, the participants and their homes. The kitchen tour also enabled the researcher to look at and inside cupboards, drawers, fridges and freezers with participants and to begin to ask questions about ‘what went on and why’ in the kitchen space. Further, because we wanted to hear how individuals explained or accounted for their practices (Scott and Lyman, 1968), each visit incorporated a series of informal interviews (Linderson, 2010) with those present. This meant that sometimes individuals were spoken to alone and sometimes collectively.

We made sketches with accompanying notes and annotations, to help with the production of a kitchen map for each household. The researchers also took photographs, particularly but not exclusively at the first visit. The photographs helped document the space, layout, things, people, pets and resources of the kitchen. Photographs were selectively used to elicit participants’ accounts (Shove and Pantzar, 2010) about kitchen life at subsequent visits. Households were given disposable cameras, a notebook, pens and some coloured pencils so that they could capture information they thought was relevant between our visits (Ison, 2009; Monrouxe, 2009). Drawing on findings from the pilot phase of the project, households were given examples of how they could use the notebook, including as a diary, to log photographs they had taken, or as a scrapbook. Photographs taken by participants with the disposable cameras were processed and a set of prints used to elicit further talk about practices.

Each household was informed that, as well as direct observation, we wished to use video to record observations about some aspects of kitchen life. Determining what, when and who to record/observe/document within each household was an important issue (Power, 2003). We did not want to record extensive footage which would be impossible to view or analyse within the confines of the project timetable (Martens and Scott, 2004), but we wanted to record a range of practices, moments and interactions involving all those within a household (pets as well as people). The initial kitchen tour, non-participant observation in the kitchen and talking to participants helped to identify suitable periods for video recording (Paterson et al., 2003). This included identifying periods when different household members would be present (individually and in different combinations). In one household, for example, the researcher was present during and after a young child had a daytime nap to film different kitchen practices being performed throughout these periods.

What is Gained from Collecting Researcher- and Participant-generated Visual Data?

Video footage and photographs, whether produced by participants or researchers, are not a taken-for-granted record of everyday life. They are a representation, a version of events (Heath et al., 2010) co-produced by the filmmaker, photographer and the viewer and subject to interpretation by each of these (as well as other audiences), as with any other source of data (Gibson, 2005). Images also do not become data until layered with interpretation and analysis (Meah and Watson, 2013). Some have questioned whether using technology to record practices does little more than add a further layer to the analysis, leading researchers to be ‘less present’ during interactions with participants than if they were to undertake fieldwork with a notebook and pencil (Travers, 2009). We found that in a multi-researcher–participant-method project, video and photography provided a useful record of phenomena that could be shared within the research team and with participants during fieldwork. These data could be mulled over, repeatedly viewed and reflected upon as the analysis proceeded, and this benefited the project as a whole (O’Connell, 2012) and helped us to make a more robust contribution to knowledge. This can be illustrated through examining who generated what visual (particularly video) data, how this impacted on the range and depth of data collected overall and how this contributed to the findings that were subsequently generated.

Many participants were particularly interested in our use of video and were quick to offer suggestions regarding what we could film. The ‘everyday’ nature of what we were interested in seemed to appeal to participants, perhaps helping them to feel relaxed and not feel a need to perform or display only ‘special’ events (Kindon, 2002). Many participants were willing to use the video recording equipment themselves, in between our visits, which we had not anticipated at the outset. We discussed with participants the range of things we were interested in, based on our prior observations and the kitchen tour, and we stressed our interest in both the mundane aspects of kitchen practices and things that were not necessarily food-related. Participants who self-filmed had the freedom to narrate their footage, to remain silent or to have conversations with others in the kitchen whilst they were filming; they could include ‘action’ shots of a range of activities; they could also decide when to turn the video camera on and off as well as where to position it. Households were thereby enabled as co-creators of data, which helped to reveal elements of everyday life that we would not otherwise have had access to at the data collection stage. Participant-generated data differed to those produced by the research team in terms of the content and timing of footage. The flow of practices and dynamics between individuals within households were also more fully revealed than was possible when a researcher was present, as we go on to describe in the next section.

Unlike the research team, few participants filmed or photographed ‘back stage’ (Goffman, 1959) aspects of their kitchen (such as the inside of their fridge), focusing instead on ‘front stage’ practices and capturing the display and performance of life in the kitchen – spouses filmed each other washing up or preparing dessert, for example. In one household a teenaged boy became engaged with the study through directing the filming of his parents and questioning them about their kitchen practices (about their purchase of specific brands, for example) from behind the camera. When the researcher was present he was more likely to leave the room, perhaps to avoid being filmed. These data also revealed different aspects of the interrelationships within this and other families. For example, while they were together and their son was absent, two parents criticised their son’s practices to the researcher (the way he loaded the dishwasher, for example), but on occasions when mother and son were alone, a softer, more affectionate relationship was sometimes glimpsed from the video footage.

The research team, often uncomfortable with being silent observers or videographers, tended to ask participants to explain what they were doing whilst they were being filmed. This perhaps unnaturally ruptured the tacit nature of the practice being performed and we cannot know the extent to which this influenced how activities were subsequently conducted. The footage produced by participants thereby helped to create a balance in what was generated. Footage filmed by households included participants’ morning kitchen activities, whilst still dressed in nightwear, unpacking groceries after a late night trip to the supermarket and mopping the kitchen floor or brushing the dog before going to bed. This approach was inclusive as even participants who were initially concerned that they were not technically competent or physically capable enthusiastically engaged with the easy-to-use technology and the open remit about what to film; age was not a barrier to participants being image makers (Hinck, 2004).

The participatory approach to filming video footage gave participants an opportunity to present their world-view of their practices and helped us ‘look alongside’ rather than ‘at’ (Kindon, 2002) their kitchen lives, which helped with the production of more thorough and robust data, thereby strengthening the overall reliability of the research findings. The nuanced differences in what was filmed or photographed by participants and by the research team revealed insights that subsequently benefited the analysis and helped to create a different knowledge to that we might have gained if we had solely drawn on researcher-created data (Kindon, 2002).

What is Gained by Using Visual and Non-visual Qualitative Methods in a Study of Social Practices?

The value of the approach we took to exploring social practices was that the intersection of participants’ actions, accounts and kitchens – and more besides – were laid bare. We found that participants were neither consciously aware of some of their actions, nor of the contradictions in their accounts about what they were doing, because households tended to enact their practices through having a ‘feel for the game’ (Bourdieu, 1990). When accounting for their actions (i.e. verbally), that flow is often disrupted and the reasons recalled or articulated may or may not relate to the ‘actual’ reason for something being undertaken in a particular way (Garfinkel, 1964). People tend to demonstrate discursive and practical consciousness (Giddens, 1984) – only accounting for or explaining the origin, presence or relevance of some of the things that they are aware that they do at that particular time – and combining visual with non-visual research methods allowed a more rounded knowledge about practices to be developed, as we discuss further in this section.

A key finding to emerge from the study was the extent and ways that social practices are entangled and therefore impossible to pull apart into separate commonly understood components such as ‘cooking’ or ‘cleaning’ (Wills et al., 2013). Simply being in the kitchen represented part of a practice that incorporated a constant flow or sequence of related events. Viewing visual data and then asking participants about what we had viewed or observing them at home was integral in helping reveal a nuanced picture of tacit yet complex social practices. For example, looking at a range of data helped us to document that ‘cooking’ can also involve giving dogs a treat to stop them whining, putting meat packaging into the bin, rearranging a bin liner, washing hands, soaking chopping boards in a sink full of soapy water, searching online for a recipe, weighing out ingredients and answering a child’s questions. If we had simply asked households to talk about ‘cooking’ they may have discussed some of the more ‘obvious’ actions related to ‘cooking’ but they are unlikely to have discussed all the entangled elements revealed about the overall practice through our mixed-methods approach. Yet it is the entangled nature of kitchen practices that is likely to contribute to cross-contamination and lead to foodborne disease, therefore the knowledge generated through this approach was important for an applied sociological study of domestic kitchen practices.

Using a variety of methods also meant that all household members could contribute according to their particular competencies or preferred ways of engaging with the research, and in the time that they had available; this was a considerable advantage when trying to understand complex practices. Through making several return visits we were also able to develop rapport with participants and so elicit a wealth of ideas, thoughts and narratives which we would not otherwise have had access to (Wills et al., 2013). Discussing with participants the contents of diaries and scrapbooks and the photographs and video footage they had recorded created an opportunity to examine what was important to them in ways that might not be immediately obvious, or captured, via other means (O’Connell, 2012). We found, as others have (Belin, 2005), that employing photo elicitation, in particular, meant it was the images themselves rather than the researcher’s questions that became the focus of discussions. Elicitation drawing on the visual data gave participants an opportunity to account for events and ‘stories’ relating to everyday life and the factors that they believed shaped them over the life course (Czarniawska, 2004; Meah and Watson, 2011; Wills et al., 2008, 2011). Informal interviews gave both the participating households and the research team opportunities to question and ‘make sense’ of the ways in which practices developed, and to reveal some of the embedded factors that shaped these experiences. Talking with participants helped identify key moments when practices, or a participant’s interpretation of practices, might have undergone a shift or rupture at transition points during the life course (Meah and Watson, 2011; Polkinghorne, 1995), such as following bereavement or during pregnancy. Furthermore, analysing video data and then interviewing participants in each household allowed us to more fully investigate the performative aspects associated with a practice (Goffman, 1959), that is, how practices are displayed in a particular way when verbally articulated by participants versus how they are displayed or ‘seen’ in visual data (Housley and Smith, 2010; Mauthner, 1997; Warin et al., 2007). This was important as the performativity of social practices is often lacking in empirical research (Halkier and Jensen, 2011). The performance of practices was particularly striking in households with children and teenagers where sibling rivalry and parent/child negotiations were displayed in front of the camera; this required visual and non-visual methods in order to be made sense of. The overall research design revealed social practices in more meaningful ways than possible with a single method (Pink, 2004a). For example, one woman filmed herself baking cakes and also photographed the ‘end product’ before the cakes were put into plastic food bags and frozen. Subsequently discussing the footage and images with her and her husband during an interview led to a conversation about where the recipes she used originated and their joint views on the ‘mess’ created when baking. Relying on the interview alone, with no visual prompts, would have produced a rather more abstract account of the purpose of and background to this household’s baking activities. An interpretation or account of the baking activity would have been foregrounded through reliance on an interview whereas the combination of visual and non-visual methods revealed more about the overall practice of baking, which involved the household’s values, beliefs, resources, technologies, material objects and evidence of the action, as well as the account about the baking activity itself.

Using video to observe practices along with direct non-participant observation extended the possibilities of each of these techniques and overcame some of the limitations of each individual method (Paterson et al., 2003). For example, video recording practices overcame the potential loss of finer level detail that can occur when relying on researchers to write their fieldnotes after a period of observation has taken place (Creswell, 2007). Such retrospective writing can overlook the multi-modal and synchronised action that takes place in kitchens, such as the sights and sounds associated with concurrently supervising a young child whilst preparing an evening meal and, at the same time, feeding a pet. Video recording overcame such issues and contributed greatly to the overall analysis as we could look back, freeze, replay and discuss emergent themes and the various nuances of the researcher–participant–kitchen interaction that might not have been seen in situ.

In terms of data analysis, as O’Connell (2012) also found, having visual and non-visual data analysed side by side enabled the research team to develop a more in-depth picture of each household. Each data type – transcripts from audio recorded data, audio recordings, video footage and photographs taken by researchers and by participants, along with fieldnotes, kitchen maps, diaries and scrapbooks – were viewed or read and analytical notes written and discussed amongst the team. Data from interview transcripts and fieldnotes were coded according to emergent themes and summary reports from these analyses were written up for each household. This was then elaborated further through repeated viewing of the visual data, enabling the analysis to move from the descriptive to the theoretical. The photographs and video data were not coded as such, as we wished to take advantage of continuing to ‘see’ the multi-dimensional or multi-layered aspects of kitchen life; the visual data helped to reveal the interactions, noises, emotions, facial expressions, talk, performance, physicality and rhythm of the kitchen (Bezemer and Mavers, 2011) and we tried to retain this during the analytical process. Moving across the data types and, subsequently, across data associated with different households, enabled us to document and flesh out the different components of the practices we had studied. Contradictions in data or across households encouraged us to delve deeper to ensure we were not misinterpreting what we thought we had seen. The data types were complementary and gaps, for example when a transcript did not make it clear what an individual was referring to, could be filled through referring to some of the visual data or to fieldnotes.

Whilst this analytic approach was time consuming, there were benefits in terms of ensuring a more robust interpretation of data leading to the creation of new and reliable knowledge about social practices. This avoided merely reducing or triangulating data down to a set of more narrow themes (Mason, 2006). For example, in a household with two parents and two young children, the fieldnotes written by the researcher tended to give the impression that this family were living the ‘good life’ after moving to the countryside. Some of the photographs taken by the researcher supported this, showing a bee hive in a large garden and a range cooker in the kitchen, for example. Video footage and interview transcripts revealed, however, that the cooker did not work and that the move to the country was more about finding an affordable place to live rather than an idyllic notion of family life. The video footage and photographs produced by the household also showed the more routine and challenging elements of everyday social practice and how encounters between adults, children and animals shaped and reflected ‘what goes on’, revealing something very different than if we had analysed only some of these data.

Some of the Challenges Encountered

Our intentions were clear at the beginning of the study, in that we gave assurances to participants about not identifying households or the individuals within them when publishing findings from the study. Once all the data were collected, however, it became clear that the data and their analysis represented a substantial insight into people’s lives (Muir and Mason, 2012). This was more likely than if we had only collected a snapshot using one method or if we had used only non-visual methods of data collection. This led to many discussions about what does or does not identify a person or their household. We collected many images, for example, which were similar irrespective of the type or age of the dwelling or the composition of a household. ‘Under sink’ cupboards usually contained cleaning products; the interior of refrigerators looked more or less similar; drawers often contained cutlery (and other items); kettles and toasters were located on work surfaces. These findings were universal amongst our selection of UK households, even if the products and appliances differed. This raised questions about whether, for example, images of a kitchen sink potentially identify a household. If that image is accompanied by a narrative about ‘whom the sink belongs to’ (even when pseudonyms are used) from excerpts of interviews with multiple people from within that household, does that increase the likelihood that a household will be identified? The discussions we had about anonymity were ongoing and systematic and highlight what others have noted about ethnographic (particularly applied) research in sociology generally – that stringent rules about the governance of a study are sometimes inappropriate and often impossible to apply (Murphy and Dingwall, 2007). This is partly due to the impossibility of setting a-priori conditions about what participation will involve and what images might be filmed or photographed in a study such as that reported on here. The lessons learnt during this study have led some of the authors to approach consent differently in other projects involving visual research methods, asking participants at the end of data collection to more fully consider the consent they wish to give regarding the dissemination of data that identifies a household; only when a researcher or a participant knows what data have been gathered can consent be fully discussed.

Careful consideration of whether and when we are revealing ‘too much’ (Muir and Mason, 2012) about a household has led us to crop, blur and exclude some images and video stills from the outputs produced thus far. We are unable to disseminate any of the video footage as it all too often reveals people going about their kitchen business or, even when individuals are not seen, the ‘whole’ of a kitchen is revealed, which means the household can potentially be identified. This may skew the findings we present in outputs, not to mention that we set out not to privilege textual over visual data. We wanted to use different forms of data to corroborate, contradict and elaborate on our understanding (Brannen, 2005) of social practices, therefore systematically ignoring some data when disseminating the findings does not fully do justice to this approach.

Using the visual data to inform the analysis and interpretation, but not during presentation of that data (Muir and Mason, 2012), limits the contribution the study can make to sociological knowledge (De Certeau, 1984). Nonetheless, regardless of what we disseminate, the key themes that emerged from the data remain salient and can, with some skill (Muir and Mason, 2012), be written up. If, however, we choose or rely on particular images or particular kinds of images (or no images, as with this paper) to support what we write about in journal articles or conference presentations, does this misrepresent the complexity of the social practices that we aimed to investigate at the outset? What effect does cropping and blurring – effectively disembodying images from their owners and contexts – have on what they can reveal about the households from which they were collected? Such questions remain an ongoing project for the authors though, as described earlier in this section, this study has led some of us to now explicitly seek consent from participants so we can disseminate video and photographic data that reveals their identity. Ethics committees seem content with this development if different modes of potential dissemination (e.g. online journals and social media) are fully discussed with each participant during and at the end of fieldwork.

Conclusion

Whilst no study is ever perfect in its design or execution, the approach adopted in the work reported here meant that rather than looking ‘at’ participants’ practices or just some elements of a practice, we were able to view them from a perspective more consistent with participants’ own streams of experience (Kusenbach, 2003), something which would not have been possible had we relied exclusively on narrative or textual methods. This is an advantage in an applied sociological project. As we have indicated throughout this paper, this methodological endeavour was not unproblematic, both in practical, analytical and ethical terms. It would be too grand a claim to say that inviting participants to be involved in co-creating visual data was experienced in any way as empowering. However, it did provide participants with an opportunity to ‘fill in the gaps’ of our knowledge about the reality of ‘what goes on’ in everyday life. This would have been impossible if the study had relied exclusively on the recordings, observations and interpretations of the research team alone or on a single method.

Drawing on current theories of practice informed the study design as it was important to ‘de-centre’ the individual participants to ensure they were observed and seen as component parts in an overall jigsaw puzzle of everyday social practice. The approach was successful in terms of bringing together the methods and the interpretations about practices that we were able to elicit.