Theorising Social Work Research
Researching the Social Work process 11th July 2000 Luton
Child and Family Social Work: Theorising the process June Thoburn Director Centre for Research on the Child and Family University of East Anglia
Mapping the territory
Research which may be helpful for child and family social workers potentially encompasses a wide range of studies. Each of us could 'map the territory' in our preferred ways. I shall leave to one side studies from other disciplines such as developmental psychology or sociology which make significant contributions to case analysis and practice and concentrate of studies which can more squarely be described as 'social work research'. One might classify social work studies in terms of the 'threshold' for intervention, using Hardiker's (1999) model (based on the medical model of interventions at primary, secondary, tertiary and quaternary levels). At the UEA Centre for Research on the Child and Family we are currently involved in research at all these thresholds with some studies covering more than one of them. Research methodologies tend to go from the more quantitative at primary prevention level to the more intensive, interview-based studies at 'quaternary' level, but this is not necessarily the case. To give some examples:
- A national study of Breakfast Clubs - an intervention targeted at all children living in 'deprived areas', but which does not specifically target individual 'needy' families within these areas, can be regarded as a study of an intervention at the primary prevention level. The study uses a quantitative 'trial' methodology to seek to ascertain whether any improved outcomes are associated with attendance at a club. But it also uses ethnographic case studies of a sample of schools to learn about process; semi-structured interviews to learn about user satisfaction; and standardised well-being schedules to learn about short-term outcomes such as attention-span and behavioural change. Similar studies of local Sure Start groups (providing enhanced services for children under 4) also use a range of methods but, because of the more complex nature of the interventions in local programmes serving small populations, a quantitative trial methodology would be inappropriate. These studies use the 'satisfaction' of those living in the Sure Start areas as the main outcome measure, but will link with the national evaluation, about to be commissioned, which will use aggregate data to provide child outcome data based on quantitative methodologies.
- At the other end of the continuum, for a 'fourth level' intervention - placing children in care with permanent substitute families when restoration has proved not to be possible - quantitative methodology was also used, although in the form of a 'natural experiment' rather than using random allocation to different interventions. The outcomes (measured in terms of whether the placements lasted or broke down for a period of between three and five years) were compared for those who were placed in long-term foster care and those who were adopted (Fratter et al, 1991). The size of the sample (over 1100 placements - a total cohort of those placed by the major voluntary agencies during the early 1980s) made it possible to use 'logit' statistical techniques to control for key variables, including disability of the child; age at placement; whether it was a foster or adoptive placement; and whether or not there was continuing contact with the birth family after placement. Useful data can be obtained suggesting associations between key variables and outcome, provided that large enough numbers can be obtained. In my opinion, any attempt at a randomised control trial, allocating cases randomly to one intervention rather than another, would, at the third or fourth level of intervention, be both unethical and likely to produce overly-simplistic and misleading results. Social work practice with children who have been so seriously harmed that removal from home is necessary, is based on careful assessment of the child and his or her family situation, as well as of the wishes of family members. Whilst there are many gaps in our knowledge, there is sufficient evidence from child development research and from empirical studies of placement to guide the social worker towards one choice of placement and legal status for this particular child rather than another. To randomly allocate, for example, one eight-year year old who has been sexually abused to foster care, and the next to adoption, would be contrary to social work ethics as well as to the legislation and guidance. The eight-year olds, even though 'matched' on paper, will in reality present different problems to their carers. Even when it is possible to find naturally occurring experimental situations with large enough numbers (as in the 1100 cohort), the explanations for any results obtained which point to associations between certain variables and better or worse outcomes can be hard to explain. For this reason, quantitative and qualitative methods involving interviews are usually combined. For a study of a sub-sample of the 1980s cohort involving 294 children of minority ethnic origin, a logit statistical analysis to explore whether any variables were associated with outcome was complemented by detailed interviews with the parents of 51 of them and with some of the young people, between 11 and 16 years after placement (Thoburn at al, 2000). When variables were held constant, there was no difference in breakdown rates between those placed with white families and those in 'matched' families. However, when gender was added into the equation, a higher proportion of girls did well when placed in black families than was the case with boys and the corollary was the case for those placed with white parents. If only the quantitative analysis, using placement breakdown as the only outcome measure, had been available, there would have been few clues to pass on to practitioners as to why this might have been the case. However, the qualitative interviews did provide indications as to why a higher proportion of boys than girls in ethnically-matched families appeared to hit problems. The quantitative data also allowed outcome measures other than 'breakdown rates' to be considered. These showed that, whilst a majority of the white families as well as the black families were successful parents across the range of outcome measures, white families had additional hurdles to overcome in an already hazardous enterprise. A conclusion supporting the Children Act 1989 policy of requiring strenuous efforts to be made to place children in 'matched' families was reached from careful analysis of the interview data.
A second way of classifying research studies is to look at whether their primary aim is:
- to improve understanding of the situations of the children and parents who use child welfare services or those such as foster parents who provide a resource, or
- to provide information which helps to understand and evaluate child welfare interventions.
As with the first 'typology', in many studies these overlap, and again a range of research methods will be used. However, for studies which seek to contribute to practitioners' understanding of those using services, qualitative methods are most often used. To provide an account of the impact on children of having been sexually abused by a parent; on a mother of having a child placed for adoption against her wishes; on a childless couple of facilitating contact with their adopted child of a birth mother who seriously assaulted her as a baby, it is necessary to use interviews conducted by highly skilled and sensitive researchers. Theories of child development and also of socialisation both inform the design of the research methods, and are evaluated in terms of their possible usefulness for practitioners in the context of the issues and themes which arise from the detailed analysis of the interview data.
Studies which concentrate on the process of intervention may also be qualitative and descriptive. However, if the interventions and outcomes can be sufficiently tightly described, and there is specificity about those to whom a service is provided, attempts to find associations between methods of intervention and outcomes by using quantitative methodologies sometimes produce accounts which can help the practitioner to match the style of intervention to the circumstances of the case. From my experience of asking social workers to describe their methods of intervention I conclude that observation and interviews by researchers who are familiar with social work approaches are necessary if social work interventions are to be sufficiently carefully described for evaluative purposes. A survey instrument which asked 'what methods did you use?' and got the answer 'crisis intervention' would not necessarily tell the researcher whether the principles of the crisis intervention method were followed, or whether this was a social worker who described as 'crisis intervention' any work done in response to a crisis (a tendency with which readers who mark students' case studies will be all too familiar). A research interviewer familiar with crisis intervention as a social work method is able to ask the supplementary questions which will make it clearer whether that description can be used for the work undertaken. Better still, but even more expensive for the funders and difficult to negotiate with those whose work is being studied or to whom services are being provided, is observation by the researcher of the actual interactions between worker and client.
How can research on 'helping' be helpful to child and family social workers?
For the second half of this paper I shall concentrate on the 'interventions' research. Child and family social workers intervene in two main ways.
- They assess and make or recommend decisions- to leave this child at home, to place this child for adoption.
- They spend time with parents, children and carers and use the relationship which is built up to provide help directly, which will most often be broadly supportive in character but may sometimes be described as 'therapy'. (I take exception to the reservation of the adjective 'therapeutic' for descriptions of interventions provided by social workers and other professionals in secondary or clinical settings, since it appears to imply that what area team social workers do is not therapeutic. Many research interviews with parents and children have taught me that what those providing 'therapy' do is not always 'therapeutic' whilst much of the 'run of the mill' support and practical help provided by area team workers is valued and seen as highly therapeutic in the sense of bringing much-needed relief from stress.
I shall concentrate on the second of these since there is more extensive research base on the first, and confine myself to one comment on the evaluation of placement decisions. Unless the study is of short term placements made to achieve a specific purpose, there is often a period of years between the intervention- the choice of placement- and the outcome as measured by the well-being and life chances of the young adult which the child placed at the age of two months, two or twelve will become (a point explored in Sellick and Thoburn 1996). With large enough numbers we can predict the likelihood of the placement lasting for a child of a certain age with a certain profile of early experiences and special needs. However, many of the variables we might study overlap so finding associations between variables and specified outcomes, rather than 'causation', is the most we can aim for. I have a particular interest in the impact of continuing birth family contact on long-term outcomes for children placed with permanent substitute families. From a range of studies it appears that continuing contact is associated either with a lower rate of placement breakdown or no association is found either way. However, from qualitative studies, it is clear that those adoptive or foster parents who are able to facilitate comfortable post-placement contact also have many of the less tangible characteristics which a range of US and UK researchers have linked with more successful outcomes. In other words, the existence of continuing birth family contact may act as a 'proxy' for the ability of the adopters to perform well the other tasks of adoptive parenthood. It may be these less measurable characteristics, such as the ability to empathise with the birth family and find the right balance between the acceptance of difference and emphasising difference in adoptive parenthood, that contribute to more successful outcomes, and not the contact itself. Nevertheless, for the practitioner, if willingness and the ability to facilitate contact is a proxy for other attributes of successful adoption or long term foster parenting, it is worth giving this prominence in the list of required attributes for substitute parents.
Turning now to the evaluation of models and methods of intervention, as Nigel Parton (1996) pointed out in a review of Child Protection: Messages from Research, the DH overviews give too little space to the detailed description of social work practice, though a careful reading of the studies themselves turns up some very useful accounts and analyses. The more recent DH overviews such as that on residential care (DH 1998) and the about to be published The Children Act Now: Messages from Research, give more attention to process and styles of intervention.
So what issues arise for those who seek to evaluate social workers' attempts to provide direct help to children, their parents and carers, over and above those which arise when researching social work interventions with other client groups?
As with placement decisions, the evaluation of outcomes is easiest, and has most often been accomplished, if desired changes have the potential to be achieved within short time scales. There is less chance of other variables which can nor be controlled for intervening. It is for that reason that there is a tendency for research on 'helping' to concentrate on models such as behavioural work or 'task centred practice' designed to achieve specific behavioural or other changes within defined periods of time. Such studies of short term interventions of the sort most often taking place in family centres and other secondary settings are necessary if 'best value' is to be achieved from scarce resources and child care workers are to be helped to 'purchase' the appropriate packages of care. Since numbers are usually small, specificity is essential about the characteristics of the children or parents using the service and the outcome measures used. Most importantly, a detailed descriptive account is needed by the researcher of what the workers actually do (rather than what the workers say they do or what the text-book says about how the method should be delivered). Such accounts must seek to include all aspects of the social work practice, including an account of the nature of the worker/client relationships and such less tangible features as the extent of user involvement as well as the techniques and skills used. A UEA evaluation nearing completion of a US style 'intensive family preservation' service is providing guidance to two local authority social services departments and a voluntary agency about which sorts of families can be best helped by such services, and also about the lower level interventions which most appropriately follow on from the time-limited intensive services (Brandon and Connolly, 2000).
However, thresholds for intervention in child and family cases are generally so high that, by the time families are allowed across the threshold and allocated to a social work case-load, problems are far too entrenched to respond to a single, easily identifiable, intervention. US researchers have concluded that 'wrap around' services are most appropriate for those cases where children have already been harmed. Similarly, in the UK Jane Aldgate (1999) has concluded, from her study of respite care for families under stress, that the generally positive impact of the respite service itself can not be easily disentangled from the impact of the counselling, practical help, marital work, volunteer visiting and other services specially created by imaginative social workers and respite carers to meet the needs of each particular family. The whole package has therefore to be brought into the equation and it proves difficult to allocate a proportion of the positive (or negative) outcome to any one part of the package. For this reason, user satisfaction as outcome measure will always have an honourable place in child and family social work research. With complex overlapping interventions, often the best we can do as researchers is to take to the service users our descriptive account of the service provided, with the different components carefully listed. We can then ask the parent, young person, or carer if we have got an accurate picture of the intervention as they saw it, and which components of the service they found most and least helpful.
Finally, service users' perceptions of positive and negative outcomes have to be put alongside the ratings of the service providers and the results from any standardised tests. 'Triangulation' of data on outcomes is a well-established feature of child and family research, and has led to some imaginative ways of arriving at 'researcher ratings' of good, moderate or poor outcomes when different 'witnesses' have different perceptions of the outcome in a particular case. The most usual method is to have 'protocols' which spell out the criteria for inclusion in each outcome category, with a sample of cases checked for reliability by a different member of the research team. Some researchers (eg Farmer and Parker, 1991), using an adaptation of the 'vignette' research technique, have produced descriptions of the case outcomes and asked practitioners and fellow researchers if they considered this to be a successful or unsuccessful outcome.
Some researchers move from the descriptive (sometimes by way of an analysis of outcomes and sometimes not) to the prescriptive, prescribing policy change or the adoption of specific practices. The most regrettable example of this is the espousal by criminal justice policy makers of the slogan 'cognitive behavioural intervention works'. What the child and family social worker needs to know is whether cognitive behavioural (or any other) intervention appears to be associated with more successful outcomes (as measured by higher well-being in the child or parents or some other specified measure) in specified sets of circumstances. The circumstances we would need to know about would include information about the parents- their cognitive abilities, any other disabilities or mental health or addiction problems; their environmental circumstances including the extent of poverty and material deprivation, the availability of transport to get to attend a particular resource, the ages and characteristics of the children.
Other researchers seek to generate typologies and theories for understanding groups of clients or for selecting the most appropriate methods of intervention in different circumstances. Jean Packman's (1986) 'victims' 'villains' and 'volunteereds' or Farmer and Parker's 'protected' and 'disaffected' categorisations encourage the reader not to 'lump' all children in care together. Owen categorises adopters into 'old hands' and 'novices'. The services appropriate for one group may be inappropriate for another. Disaggregation is particularly important when considering services to parents and children of different ethnic backgrounds. Jane Rowe and colleagues (1989) pointed out that although African Caribbean children were more likely to be in care in one borough than white children or those of other ethnic origin, this was a positive finding since the type of care they typically received was short term voluntary care of younger children, provided as part of a family support service. Other groups were less likely to be in out-of-home care, but if they did come into care it was more likely to be when older, through compulsory routes and for longer periods.
Some move beyond categorisation and typologies to generate 'middle range' theories which can lead to more appropriate choice of helping methods. Cleaver and Freeman's (1995) typology of families who become involved with the formal child protection system can be used as a basis for a 'triage' system. This could help hard pressed intake teams to decide which cases can be safely closed with no further action, which might best be taken on for short interventions before problems become entrenched, and which should be taken on immediately with the acknowledgement that problems are chronic and long-term intervention, albeit of an episodic nature will be necessary if there is to be any chance of the children achieving their potential. Howe and colleagues (2000) have explored the potential of attachment theory to help social workers with 'stuck' cases.
In conclusion, I come back to a point touched on earlier and make a plea for qualified social workers to continue to be recruited as researchers alongside those who have followed the more traditional route of the academic PhD. (And this will not happen unless they are paid on a scale which encourages them to do so, recognises the special skills and qualities they bring to research and provides a better career structure than is presently available.) I have already explained that their ability to 'speak the language' of practitioners and probe when answers are unclear is more likely to ensures the collection of more accurate data on process and the detail of intervention. But those who have done the work themselves also understand about the power dimensions in both practice and research. For example, they understand that, when a parent or young person decides to tell them something highly relevant to the case which appears not to be known by the social worker or recorded in the file, this is not because of their superior skills as interviewers nor the social workers' incompetence. It is simply that the researcher, who comes only to listen and understand, and has no power to change anything (whether that means providing a much needed service or taking a child into care) is in a highly privileged position. Practitioner/researchers know that there are things which clients will tell researchers which they would never tell their case-accountable social workers. They are therefore less likely to lapse, when writing up and orally presenting their findings, into 'how come that I could find this out and the social worker didn't know' mode which gives research a bad name amongst the practitioners it is intended to serve.
Aldgate, J. & Bradley, M. (1999). Supporting families through short-term fostering. London: HMSO.
Brandon, M. and Connolly, J (2000) Evaluation of Families First, Paper to Making Research Count Conference, Norwich: UEA Centre for research on the Child and Family.
Cleaver, H. & Freeman, P. (1995). Parental perspectives in cases of suspected child abuse. London: HMSO.
Department of Health (1999) Residential Child Care: Messages from Research, Chichester: Wiley
Department of Health (forthcoming) The Children Act Now: Messages from Research, London: The Stationery Office
Farmer, E. & Parker, R. (1991). Trials and tribulations: Returning children from local authority care to their families. London: HMSO.
Fratter, J., Rowe, J., Sapsford, D. & Thoburn, J. (1991). Permanent family placement: A decade of experience. British Agencies for Adoption and Fostering.
Hardiker, P., Exton, K. and Barker, M. (1999) 'Children still in need, Indeed' in O. Stevenson (ed.) Child Welfare in the UK. Oxford: Blackwell Science
Howe, D., Brandon, M., Hinings, D., and Schofield, G. (2000) Attachment theory, child maltreatment and family support, Basingstoke: Macmillan
Owen, M. (1999) Novices, Old Hands and Professionals, London: BAAF
Parton,.N. (1996)'Child protection, family support and social work: A critical appraisal of the Department of Health Research Studies in Child Protection' Child and family Social Work, Vol. 1.1: 3-11
Rowe, J., Hundleby, M. & Keane, A. (1989). Child care now - A survey of placement patterns. London: British Agencies for Adoption and Fostering.
Sellick, C & Thoburn, J. (1996). What works in family placements. Barkingside: Barnardos.
Thoburn, J., Norford, L. & Rashid, S. P. (2000). Permanent family placement for children of minority ethnic origin. London: Jessica Kingsley.