From snapshots of practice to a movie: Researching long-term social work and child protection by getting as close as possible to practice and organisational life

Research into social work and child protection has begun to observe practice to find out what social workers actually do, however no such ethnographic research has been done into long-term practice. This paper outlines and analyses the methods used in a study of long-term social work and child protection practice. Researchers spent 15 months embedded in two social work departments observing organisational practices, culture, and staff supervision. We also regularly observed social worker’s encounters with children and families in a sample of 30 cases for up to a year, doing up to 21 observations of practice in the same cases. Family members were also interviewed up to three times during that time. The paper argues that a methodology that gets as close as possible to practitioners and managers as they are doing the work and that takes a longitudinal approach can provide deep insights into what social work practice is, how helpful relationships with service users are established and sustained over time, or not, and the influence of organizations. The challenges and ethical dilemmas involved in doing long term research that gets so close to social work teams, casework and service users for at least a year are considered.


Introduction
While a large research literature exists on social work very little of it has got close to practice and explored what social workers and service users actually do and the impact of organisational culture and support on practice. Virtually no such research has been done into long-term social work, an absence that is all the more remarkable in the area of child protection given that in cases where children have not been protected, families were worked with over long periods and usually years (Jay, 2014). There has also been little research into how social work practitioners work with parents in effective ways to create change and are supported by their organisations to promote children's safety and well-being over the long term.
This paper is based on research that sought to make a significant contribution to filling these gaps in knowledge. It outlines and analyses the methodology used in a study of long-term social work and child protection that used participant observation of practice and social work organisations. . The study involved 15 months of fieldwork on two sites, spent shadowing practitioners and managers in the office and on the move when seeing service users. We will argue that observation and a focus on time through a longitudinal approach enables deep insights to be gained into organizational life and the nature of practice and how relationships with children and families are established and sustained over time, or not. Such sensitive ethnographic and mobile methodologies can get at the underlying forms of experience, sense making, skill and decision-making that make social work what it is as it is practiced in real time. . Our aim is not to present the research findings, but to set out the methodology and analyse what we learned from applying it, so that the potential we feel it has to inform research into all areas of social work may be realised.

Research into practice: Snapshots of social work
There is a rich tradition of ethnographic studies that have used participant observation to study what goes on in social work organisations (Dingwall, 1983;Pithouse, 1987;Scourfield, 2001;Broadhurst et al, 2010;Gillingham and Humphreys, 2009;Helm, 2016;Gibson, 2016, Saltiel, 2016. These show the impact of audit culture, performance management and the pressure on social workers to complete computerised records and work to tight timescales. A significant emerging theme is the impact of increasingly prevalent office designs such as hot-desking on organisational culture. Jeyasingham's (2016) ethnography of a traditional small office design with allocated desks and a hot-desking 'agile working' environment found that the latter offices were experienced by workers as less supportive than the former. Another ethnography of social work offices also found that open plan designs with few spaces for artefacts such as toys were experienced by service users as alienating (Author's Own). While engaging in long-term institutional ethnography, none of these studies researched the effects of organisations on the quality of face to face practice. Similarly, the literature on staff supervision in social work illuminates different approaches to and processes of support (Davys & Beddoe, 2010), and has only begun to explore the link between supervision, the quality of practice and service-user outcomes (Wilkins, et al, 2018).
Research into social work has been dominated by methods that have kept a distance from practice, but there is increasing interest in studying face to face practice as it is going on. Hall and colleagues (2014) gathered audio-recordings of social worker-service user communications in fixed rooms in family homes, without using observation. Symons (2017) also used conversation analysis of audio-recordings of telephone encounters between service users and professionals and shows that without doubt, speech and questioning styles are important. A 'practice-near' research approach considers in particular the emotional and unconscious aspects of social work (Froggett, 2012). This is done either through researchers getting physically close to practice encounters in real time, including on home visits (Henderson, 2018;Noyes, 2018), or by not observing the practice but conducting 'reflective case discussions' with social workers after the work is done (Ruch, 2014).
Author 's Own (2014; research got close to social workers' practices in a systematic way by observing and audio-recording their encounters with service users. Six months of fieldwork was conducted, split between two local authorities in England. While time was spent in the social work offices, the primary focus was on practice and the study observed and audiorecorded 87 practice encounters (71 home visits and 16 office and school visits). This method produced insights into how and where children, parents and other carers were related to, and visits in which social workers were observed not relating to children at all.
The 'Talking and Listening to Children' project Ruch et al, 2017) also researched what happens in everyday, live, communicative encounters between social workers and children by conducting fieldwork across England, Northern Ireland, Scotland and Wales. The researchers spent six to eight weeks in each of the four sites in eight social work teams, and accompanied social workers at 82 encounters with children, 57% of them in the home and 24% at schools, observing and taking notes of the encounter, but not audio-recording. Their findings 'indicate that given the complex, contingent and context-specific nature of communicative encounters, it is impossible to create a definitive list of factors that facilitate communication' (Winter, et al, 2017, p. 13). Another approach, led by Donald Forrester, has used observations of practice on home visits to develop a tool for measuring social workers' skills and how they communicate with parents. Quantitative methods are then used to rate practice, interviewing skills, the relationship (or working alliance) between social worker and service user and levels of perceived engagement (Forrester, D. et al, 2019;Killian et al., 2017).
Whilst there are methodological differences in approach to studying face to face practice, what this body of research shares is a focus on individual encounters, on home visits and elsewhere. The subject of inquiry is snapshots of practice encounters between social workers and service users in isolation from detailed examination of organisational practices and what happens in casework over time. While this work has produced significant insights, what is needed now is research into long-term social work practice.

Time, experience and research methods
The core research questions of our study were how do social workers establish and sustain long term relationships with children and parents in high risk child protection cases? And what is the influence of organisational cultures, office designs and forms of staff support and supervision on social workers and their relationships with children and families? As the research was the first to study organisational life and long-term social work practice by shadowing them, a methodological aim was to explore how -and indeed if -such long term ethnographic research can best be done into social work.
A range of mobile and ethnographic research methods were used within an overall qualitative longitudinal research design (Buscher, et al, 2011;Pink, 2015). In moving beyond the 'snapshot' approach of conducting observations of one-off encounters to one that shadows long-term practice, our study took a longitudinal approach that 'offers a movie rather than a snapshot' (Neale, 2012). This enables attention to be given to the temporal dimension of experience, in terms of process, causality, dynamics, continuity, change, transitions, and turning points, discerning 'change in the making' (Neale, 2012;Saldana, 2003). There is debate in the methodological literature about how long is 'long' in longitudinal research and as Saldana (2003, p.3) argues, context is crucial to determining this. This dilemma is quite easily resolved in social work when research is based, as ours was, in what are pre-defined as 'long-term' teams. But it is important to be clear that the snapshot approach can be taken to researching practice in long-term cases by observing an episode of practice in cases that are well known to the service (Author's own). The key distinction is between research that involves observing snapshots as in single practice encounters in new referrals or in existing 'long-term' cases, and an approach that observes practitioners working with the same families on multiple occasions to explore how practice is carried out as it unfolds over time. It was the latter approach that we took and indeed pioneered, as this is the first study to undertake such ethnographic long-term research into social work practice. It involved us repeatedly going back to see the same families with social workers and doing up to 21 observations of practice in the same cases over the course of a year, while also gathering data in the social work organisation. We adopted a time-frame of 15 months of fieldwork because of our desire to study up to a full year of practice and draw out change -or its absence-in the making. We allowed an initial three months to build a sample of cases to observe for up to a year (months 4-15). We also followed work in anthropology that argues that 15 months of immersion in a research site is needed to properly make sense of the culture and practices (Miller et al, 2016). Our aim was to shadow a sample of 30 child protection cases over the course of 12 months fieldwork (15 at each site), a number that it was felt could provide a range of types of cases and enable the depth of inquiry needed. We also sought to interview service users in the same cases at up to three points during the 12 months.
As well as a focus on time our research was informed by theoretical and methodological work that enables exploration of the mobile, sensory and emotional dimensions of everyday experiences that are not captured using traditional methods, such as interviews conducted while seated / still (Ingold, 2011;Ross et al., 2009). We shadowed social workers everywhere they went, in cars, on foot, around offices, and into and around service user's homes, paying attention to the influence of atmospheres and feeling of places (Pink, 2015), the dynamics of relationships, emotional experience and use of self in practice (Davys and Beddoe, 2010;Ruch, et al, 2010). . We also used GPS [Global Positioning System] devices to trace social workers' mobilities and explore their use of office space, home working and visits to families, which we have written about elsewhere (Author's own, 2019).

Research design
The study took place between 2016 and 2018 and was funded by the Economic and Social Research Council (Grant Number ES/N012453/2). It was approved by the research ethics committees of the participating universities and social work agencies. It was based in two local authorities in England, over100 miles apart, chosen to provide diversity in their office and system designs, degrees of mobile working and service users. ity At one site teams covered a large area that included some urban, town and rural life and hot-desking was used in a large open-plan room which accommodated 60 staff. This meant that staff did not have their own allocated desk but used whatever work-station was available each day. . The second site was urban and the organisational design more traditional, with practitioners based in small team rooms that accommodated around five or six staff. . All the staff had their own desk and the organisation had moved to a model of the co-location of team managers with social workers, family support workers and admin workers in smaller units. This replaced the traditional approach of a team manager supervising 8-10 social workers and being based in a separate room to them. This research design then provided for a comparison of two quite different organisational models and their effects on practice with service users.
Two research fellows were based in the offices of the social work teams, one at each site, for the 15 months of fieldwork. The academic staff members of the research team also did some fieldwork, with two in particular being allocated to a site each working regularly alongside one of the research fellows. The research team spent a total of 402 days in the field, 201 days at each site. We observed office routines and practices, computer use, staff interactions and relationships, supervision, social worker's practice on home visits and elsewhere and multiagency meetings. At both sites the social workers were based in long-term teams, whose primary role was child protection. Referral and initial assessment was done elsewhere and social workers in the study picked up new cases that were referred into them while mainly holding cases they worked with for months and even years. This enabled us to shadow some cases from the time they came in as new referrals (n16). In addition, we purposively sampled existing long-term cases, adopting the inclusion criteria of: different lengths of time known to the service; degrees of cooperation in social worker-family relationships; child characteristics, covering the age range from babies to older children, involving a range of concerns. 'Retrospective longitudinal interviews' (Neale, 2012) were carried out to establish the casework already done and future plans.
Our commitment to shadowing practitioners doing what they normally do meant adopting mobile methods and travelling with workers to see families, interviewing them on their way to home visits or meetings like case conferences about their plans and feelings. Social worker's interactions with children and families were then observed and audio-recorded on home visits, or elsewhere, in schools, hospitals, court rooms and meetings. Observational data on the encounters was taken on the spot, including detailed attention to what was said, tone of voice, questioning styles, movements, the timings when things happened, non-verbal communication, moods, atmospheres and the feeling of the home and the encounter. Drawing on a material culture approach (Miller, 2010), we also noted objects of importance, such as the presence or absence of furniture, toys, mobile phones. Afterwards, workers were interviewed about their experience of the encounter and we observed their interactions with managers and peers on returning to the office and of formal supervision sessions which deepened our understandings of the nature of staff support. So as well as studying relationships and communication between social workers and service users, we researched the nature of office designs, culture and forms of staff support and supervision in their own right in two very differently designed social work organisations. And we examined their influence on social workers, their relationships with children and families and their capacity to be helpful.
Observations were only done with the consent of practitioners and service users. Access was negotiated through social workers, who asked for parents' consent for the researcher to accompany them. On meeting the family, the researcher explained the research and formal consent was sought to audio-record as well as observe the encounters. A copy of the signed consent form was left with service users, with contact details should they wish to withdraw at any time. This happened in three cases: in two, the parents consented to observation of an initial home visit and then decided they no longer wished to be involved; in the third, after observing a second home visit, the parents felt our presence was causing their child who was regarded as hyperactive to become uncontrollable in response to the novelty of the researcher being in the home so we ceased observing. The families, local authorities and all professionals involved were promised complete anonymity. To achieve this, while the case examples and organisational issues referred to in everything we write reflect actual events and findings, details have been changed to protect the anonymity of all participants and places. The limits to confidentiality were made clear in that if the researchers had reason to think a child was left at risk then social work managers would be informed. It was never deemed necessary to do this.

The data and how it was analysed
During the 402 days we spent in the field we observed a total of 271 practice encounters between social care staff and service users in a variety of settings, of which 146 were home visits (see Table 1). 54 staff supervisions were also observed and 54 interviews took place with families, some of which involved up to three interviews with the same families over the course of the year. This involved the initial recruitment of 53 cases to allow for subsequent attrition and we eventually met our goal of achieving a settled long-term sample of 30 long-term cases, 15 at each site. It was impossible to predict with certainty how long the families would be worked with but if they were placed on child protection plans several months of involvement was expected. A 'funnel approach' (Neale, 2012) was adopted that allowed for cases dropping out and 'progressive focusing' (Miles et al, 2014) over time on a core sample. As Table 2 shows, 12 cases were shadowed for the full 12 months, two for eleven and a total of 28 were shadowed for at least six months. That some of the sample of 30 were closed before reaching 12 months of being shadowed was a virtue for the study as it enabled analysis of the practice and changes that led social care involvement to end. Several cases remained open after the fieldwork ended.
Detailed field notes of observations of office routines and talk about cases were taken (Hammersley and Atkinson, 2007). All of the observations of practice encounters and interviews with social workers before and after them were audio-recorded and a selection were fully transcribed, while the remainder were written up in detail contemporaneously. The selection of the audio-recordings for transcription was determined by the aim to capture routine social work and particular episodes of relational practices and worker's closeness to or distance from children and parents. All of the observational and interview data was coded in NVivo and analysed thematically, using standard techniques of constant comparison (Bryman, 2012). We also extracted data from social work case files about the total work that was done over the year. We discovered that casefiles do not record the amount of time practitioners spend with children and families. So observing practice proved even more vital than anticipated because it filled this gap by revealing how time typically was spent, both on individual practice encounters and in the frequency and dynamics of relationships over time. .
Finding a way of capturing the flow, continuity -and sometimes dis-continuity -of what happened over time was crucial and the case study method proved vital to this. It involved assembling all the data that had been gathered on each family/case into a chronological narrative -field notes of the observations in the office, social worker interview data, audiorecordings and observation notes of social worker-service user encounters, staff supervisions, and interviews with parents. Triangulation of that data and the methods produced 30 case studies of long-term casework that show in enormous detail the amount and type of work that was done with the family, the nature of the relationships and what was going on in the organization at the same time. In the 12 month case studies for instance, the lowest number of observations of practice encounters was 10 and the highest 21, with many other interviews and observations of organizational practices on top and some case studies contained between 150,000 and 200,000 words. Even with the privilege of having a well-funded research team for two years, data management and analysis was a huge challenge. On the basis of what we have learned we feel that future researchers may be able to achieve the aims of such research with a focus on fewer cases but what must not be sacrificed is the depth of understanding that comes from observing regular practice encounters, while allowing enough time for keeping on top of the high volume of data.

The seasons of social work
While the purpose of this paper is not to present the research findings but to provide learning in relation to the methodology, it is important to provide an illustration of the kinds of data, analysis and insights our longitudinal ethnographic methods make possible. In a typical example of a 12 month case study 50 encounters took place between the 'Harris' family and professionals during the year, 17 of which were observed by the research (see Table 3). We also had many discussions with the workers about the case as it was unfolding over the course of the year. 'Adele Harris' was expecting her first baby and was regarded as very vulnerable due to mental health problems and previous relationships in which she was abused by men. Jamie, the baby's father, was known to have been violent towards previous partners. Professionals felt that Adele did not understand or accept the risk she was at from Jamie and in Month 2 -the Winter period of the case being shadowed -we observed the case conference at which their unborn baby was placed on a child protection plan. As the work developed into early Spring, during Months 3 and 4 we could see that Adele was reticent about social care involvement because she knew they regarded Jamie as high risk. Social work home visits to the Harris family lasted an average of 29 minutes. We observed how Adele's social worker and family support worker began and developed meaningful relationships with her and to a degree Jamie, and with the baby when he was born in the late spring, Month 5. We were present on the first home visit after the birth when Adele was in bed with the baby Ralph, and Davina the social worker sat on the bed alongside Jamie. She offered them "congratulations, the baby is beautiful, you must be so proud" and witnessed both parents nursing the baby, telling them: "You are both very natural with him".
Five weeks later in Month 6 a scene that illustrates some of the key organizational and practice dynamics of the case study occurred on a home visit by Davina, who was handing the case over to a new social worker. This begins to show how organisational challenges such as staff turnover could affect the development -or (premature) ending -of relationships. But in this social work office regular joint visits were just one indicator of how well supported workers were by an essentially stable organisational culture. This was a vital part of the context in which Adele and the new social worker and family support worker went on to develop meaningful working relationships. . In late Summer, Month 9, on a joint home visit, they arrived to find Jamie assaulting Adele. Legal proceedings to protect the baby were now a real possibility and Adele ended the relationship. We then observed several home visits at which empathetic, authoritative work was done by social care and at multi-agency meetings that helped Adele deal with her love for Jamie and accept that she and the baby could not be with him, alongside help with benefits and housing. By the Autumn, Month 12, at the end of the research fieldwork, Adele remained apart from Jamie, social care's view was that her parenting was essentially okay and the baby was likely to come off the child protection plan.

Adele opens the door and smiles at seeing Davina. As we enter the room Adele goes and sits on her bed while both social workers sit on the floor, Davina closer to Adele. The baby is five weeks old now and asleep in the
We hope this brief illustration serves to give a sense of how the method enables bringing together analysis of what is happening in the organisation with what occurs in face to face encounters with service users. This includes creating 'scenic reconstructions' (Froggett and Hollway, 2010;Roy, 2017) out of the data and selecting scenes that illustrate key dynamics, turning-points and change in the making. Shadowing these organisational and relational dynamics long-term and repeatedly going back with workers to see families week after week, month after month, enables the research to capture the nature and ebb and flow of practice and relationships over time and their outcomes. It provides insights into the 'seasons of social work' in terms of the nature, rhythms and emotional texture of life and relationships as they are influenced by the ups and downs of organisations, staffing, practitioners' skills and knowledge, and the vulnerabilities and strengths of service users. Analysing all the cases in the sample where positive change occurred and those where it didn't enables the very nature of authoritative relationship-based social work practice and its various forms and effects to be established.

Discussion
We set out on this research project unsure about what long-term ethnographic research that stays consistently close to practice and organisations would be like. We had to discover whether being embedded within social work teams for 15 months and in families and casework for up to 12 months could be done ethically and relationships with the field sites and families successfully sustained. Our conclusion is that it can, but several challenges arise and in this final section we will critically reflect on them and further analyse the methods.
At its purest, our intention was to use ethnography to observe naturally occurring events that would have happened in the same way had we not been present. It is unlikely however that the presence of the researcher has no impact on what they are observing, but just how much and what kind of impact is debatable (Hammersley and Atkinson, 2007). Even if it could be ascertained that our presence had no effect on participants, another potential problem arises in how participant observation does not provide an exact reflection of the 'truth' of what is seen to go on because it is human beings who are experiencing and interpreting things. The data is filtered through the subjectivity and perception of the researchers and the potential influence of their values and relationships with research participants. Gaining and maintaining staff cooperation was easier with some workers than others and at one site in particular was quite difficult (for an extended discussion of this process, see Authors' Own, forthcoming). But generally we experienced great kindness from many staff and were often included in office rituals such as birthday celebrations, leaving 'dos', and at one site the office Christmas party. Over time we were often described as being 'one of the team', with offers to allocate us cases! Some staff shared their personal troubles and tears, as well as joys with us. We tried to get round the possible biases that arose in becoming absorbed into the culture by discussing our feelings within the research team and by examining our interpretations of a sample of the same transcripts. This included critical reflection on how we used our power as researchers, especially with regard to vulnerable service users and being attuned to how and indeed whether social workers exercised their power in anti-oppressive ways.
What can be said with confidence is that observation enables so much to be seen and experienced in social work that would otherwise remain invisible -and that has remained hidden, due to the distance researchers have kept from practice. What ethnographers do have some control over is the extent to which they become actively involved and participate in the events they are observing. Some seek to get as involved as possible, to be 'Participant-Observers'. Because we were researching highly sensitive and consequential encounters that sought to keep children safe, we worked very hard at keeping our participation at a minimum, at being observers, adopting the position of 'Observer as Participant' (Gold, 1958). Yet we found that we became participants in several ways. Sometimes parents included us in what they were doing, such as when showing social workers family photographs or films on their mobile phones. Children often engaged us in play, something that increased over the course of the year, for instance as we watched babies grow into mobile infants they acquired the strength and agility to reach and climb up us. On occasions we saw things we weren't supposed to. For instance, as the researcher was waiting in their car outside a family home for a social worker to arrive so that they could shadow them they witnessed a father who was not allowed to have any contact with his child due to his violence leaving the home with that child. After the social worker arrived the researcher told them what they saw and shadowed the worker in pursuit of finding the man and child, which they eventually did. Managing the complexity of our position as researchers, together with witnessing the struggles and suffering of children and families over an extended period, was very emotionally demanding (Authors' Own; Bergman Blix and Wettergren, 2014). We tried to use this emotional experience reflexively to gain further insights into how complex social work is and the stressful effects on workers and families.
Our mere presence had an effect, for instance in cases where there were changes of worker over the year, the new worker seemed to benefit from how the families knew us and we provided some continuity of presence. A small number of families told us they felt they got a better service because we were present. Some parents said they found us being there supportive, despite us studiously trying to remain neutral. As one mother on her third research interview after a year of involvement in the study epitomised it: I think in capital letters that having you both shadow what has been going on, I think has made the situation like 100 times better for me. Because everything has been recorded and they know that everything is being recorded and there is also witnesses to the conversation because of course they don't normally work like that, so for me it has been like a god send actually you know. I see you guys as basically like, you know, like angels.
We were stunned by how positive this perception was of our impact, but on reflection it made sense, given that we observed practitioners working with this mother on 21 separate occasions and we were witnesses to profoundly significant events in her life. In effect, we developed relationships with some service users, albeit peculiar ones given that as Observer-Participants we barely spoke. . When we interviewed family members on their own we would remind them of our neutrality and independence from social work staff but what we had little control over was what we came to mean for them, unconsciously as well as consciously. For some we seemed to mean little and barely be noticed, while others, like the mother who called us angels, seemed to regard us as protectors. Ironically, the social workers who worked with this mother told us that they felt she treated them better when we were present. One possible implication is that what we got to see was not practice encounters as they normally occurred but a more harmonious version. Yet in the cut and thrust of day-to-day practice it didn't look or feel like that to us and there were times when neither practitioners or service users felt they were being treated nicely. From a psycho-social perspective, what participants are more likely to have been expressing was the feeling of being emotionally contained that being observed gives (Hingley-Jones, et al, 2017).
Another dimension of such containment was the way practitioners often commented on how they found being observed and interviewed about their work while they were doing it helpful. This is typified by what this social worker said in the car following observation of a home visit: it helps me to think about things as well. Like I said to you earlier, if I was to kind of drive off and you think about it yourself but it's not the same as being asked questions and getting you to think about it in depth, so it's really helpful for me to kind of break it all down.
This finding that often the research was experienced as having positive effects supports Westlake and Forrester's argument that decisions by organisations about whether to provide access to researchers and by research ethics committees about whether to approve proposed studies should focus on their potential advantages as well as risks, on 'supporting researchers to not only 'do no harm', but to help people where possible' (Westlake and Forrester, 2016). .

Conclusion
Using longitudinal, ethnographic and mobile methods to study organizational life and face to face practice with service users over the long-term moves research forward from a focus on snapshots of practice to produce something closer to a movie by connecting scenes from week to week, month to month practices in ways that provide for detailed understandings of the complexities of organisations, social work and long-term relationships. The rich data also provides the basis for producing real movies for use in education and training and our dissemination activity includes the production of 360 degree videos in digital formats that provide an immersive experience for viewers that capture the lived experiences revealed in our findings (Author's own). While we have shown that the full impact we had on the practices we observed is debateable, what we can say with certainty is that no participants ever told us that our presence made things worse, that it had a negative impact on peoples' experiences or outcomes. In the single case where we were told our presence was over-exciting a hyperactive child we withdrew immediately. Such 'ethically important moments' (Guillemin and Gillam, 2004) will always arise in ethnographic research and it is not their existence but how they are managed in ways that ensure no harm is done that matters. We conclude that the methodology we have outlined and critiqued in this paper has the potential to illuminate all areas of social work. Qualitative longitudinal research into long-term practice provides vital insights into what social work is, how it is effective in creating change for service users and why it sometimes does not meet this aim. It fulfills a core mission of social work research to not only understand the world but provide resources for changing it.