You Can Take a Horse to Water But You Can’t Make it Drink’: Exploring Children’s Engagement and Resistance in Family Therapy

Children’s engagement and disengagement, adherence and non-adherence, compliance and non-compliance in healthcare have important implications for services. In family therapy mere attendance to the appointments is no guarantee of engaging in the treatment process and as children are not the main initiators of attendance engaging them through the process can be a complex activity for professionals. Through a conversation analysis of naturally occurring family therapy sessions we explore the main discursive strategies that children employ in this context to passively and actively disengage from the therapeutic process and investigate how the therapists manage and attend to this. We note that children competently remove themselves from therapy through passive resistance, active disengagement, and by expressing their autonomy. Analysis reveals that siblings of the constructed ‘problem’ child are given greater liberty in involvement. We conclude by demonstrating how therapists manage the delicate endeavour of including all family members in the process and how engagement and re-engagement are essential for meeting goals and discuss broader implications for healthcare and other settings where children may disengage.


Introduction
Children and adolescents' disengagement from clinical services is a significant problem with cancelled appointments, failure to attend and drop-out all being costly for health services (Kazdin et al. 1997;Wang et al. 2006), and frustrating for therapists (Werner-Wilson and Winter 2010). Typically children are not the main initiators of help-seeking and neither are they the main determinants of attendance (Wolpert and Fredman 1994), as it is usually the parents who take responsibility to bring the child to therapy (Hutchby 2002) and make treatment decisions (Tan et al. 2007). In essence, there is an institutional expectation in therapy to speak about one's problems and this incitement to speak depends on the client's willingness to comply (Silverman 1997). Although the parent can physically bring the child to therapy, whether that child will engage with the therapeutic process and work towards goals and resolution is not so straightforward.
Non-compliance of children in medical and therapeutic contexts is prevalent (Richman et al. 1995), with non-completion rates being quite high, for example in child psychotherapy (Pina et al. 2003). The accomplishments of therapeutic aims, therefore, are dependent upon the child's cooperation in the production of talk about therapeutically relevant issues (Hutchby 2002). Child engagement requires a commitment from both the parent and the child (Day et al. 2006). This is because although research illustrates that the greater the involvement of the child the greater the therapeutic change (Chu and Kendall 2004), parents need to be actively involved to sustain any change (Boggs et al. 2004).
Mental health treatments for young people are usually delivered within the context of families (Tan et al. 2007), with family therapy being one arena for families to work through their problems. Concerns have been raised however about the increase in the number of families dropping out of family therapy and failing to receive the services they need (Topham and Wampler 2008). Ostensibly a key focus for family therapy is to provide a forum through which the child's perspective can be aired (Strickland-Clark et al. 2000) but problematically children and adults have different levels of cognitive and linguistic competence and this creates a challenge for mutual exchange (Lobatto 2002). Lobatto argues that it is difficult therefore for the therapist to create an atmosphere which is inclusive of all parties as therapy tends to be predominantly adult led, and has potential to contribute to attrition rates.
Research illustrates that children want to be included in therapy in a meaningful way (Stith et al. 1996) but the presence of their parents can inhibit their conversational contributions (Beitin 2008;Strickland-Clark et al. 2000). For example, children in family therapy speak less than their parents (Mas et al. 1985), are interrupted more frequently (O'Reilly 2008), and yet when interrupting are treated in negative ways (O'Reilly 2006). Research indicates that young people are particularly difficult to engage in therapy and creating an alliance with them is especially challenging (Thompson et al. 2007). In family therapy the parents and the therapist may seek to engage in the institutional tasks of therapy such as identifying and finding solutions to the problems presented, but notably children may not understand or wish to go along with this, and may actively seek to avoid participation (Hutchby and O'Reilly 2010). Alliance between clients and therapists is, therefore, considered essential to the therapeutic process (Aspland et al. 2008), and has been an area of interest in relation to establishing reliable methods of measurement (Pinsof et al. 1994). Understanding therapeutic alliance is considered particularly important for understanding treatment outcomes (Thomas et al. 2005). Unlike didactic therapy situations, family therapy invokes additional challenges as the therapist considers how to foster alliances with multiple members with different motivations and problem definitions (Escudero et al. 2008). If therapists base their decisions on input from the parents alone, however, they risk missing problems that matter to the child and may alienate or fail to engage the child (Hawley and Weisz 2003).
This disengagement or resistance to therapy is potentially averted by increasing therapeutic alliance (Frankel and Levitt 2009), but if alliance is not maintained then rupture in the relationship may occur. Ruptures in the therapeutic alliance are defined as the deterioration in the relationship between the therapist and the client which may lead to dropout and treatment failure (Safran and Muran 1996). It is important to understand dropout in order to reduce an inefficient use of resources in mental health (Masi et al. 2003), and the ruptures that frequently precede attrition. Ruptures can be recognised predominantly in changes of behaviour such as withdrawal and confrontation (Safran et al. 2001) and may arise from unvoiced disagreements about the tasks and goals of therapy (Aspland et al. 2008). Therefore, if the therapy is to progress, the therapist needs to attend to both the parental and child perspectives, because if one party perceives the therapist to not understand them and their problems they may disengage (Hawley and Weisz 2003).
Although family therapists have developed strategies for engaging children in the therapeutic process we have a limited evidence base for how children experience therapy or how they engage with it (Strickland-Clark et al. 2000) or disengage from it. Analysis of the behaviour of children and families in therapy can be useful for predicting therapeutic outcomes (Kazdin et al. 2005). The aims of this paper, therefore, are to explore how children's behaviour is an indicator of engagement and disengagement patterns thus enabling recognition of when and how these patterns occur in practice. Additionally we investigate how therapists manage any potential ruptures in alliance with children and consider how they reinstate engagement. Exploring the disengagement strategies of children in family therapy has potential to facilitate the recognition of early indicators of potential ruptures in alliance and both prevent and manage their occurrence.

Methods
For this research we utilise a qualitative framework to explore the different ways children attempt to disengage from family therapy.

Recruitment and Participants
Our data for this project was provided by a team of systemic family therapists based in the United Kingdom. Actual family therapy sessions were video-recorded, totalling approximately 22 hours of therapy with four different families. These families have been assigned the pseudonyms of Clamp, Niles, Bremner and Webber. Two therapists took part in the research and were assigned the pseudonyms of Joe and Kim. The four families included in the data corpus were White British, from the Midlands and typically from lower socio-economic groups.
A convenience sampling method was employed with the first four families with capacity and providing consent being recruited to the study. The only exclusion criterion was parents with mental health problems that were judged to impair capacity to consent. Sampling occurred within the allocated 9 months for data collection. Sampling was appropriate to the methodological framework and issues of saturation are not intrinsic to the approach with its deductive discursive epistemology (O'Reilly and Parker 2012a). As a deductive mode of enquiry the premise of CA is that the micro-mechanisms of talk in the smallest sample can shed light on general principles of all aspects of language. This means that the notion of saturation is not inherent in this methodology. Contemp Fam Ther (2013) 35:491-507 493 The Clamp family constituted, the father (Daniel/Dan), the mother (Joanne), the uncle (paternal sibling Joe), and three children; Phillip (aged 13) the referred child, Jordan (aged 9) having both physical and mental health difficulties and Ronald/Ron (aged 6) having a learning disability. Member of the Bremner family were, the mother (Julie), the maternal grandmother (Rose), and two children; Bob (aged approximately 8 years) the referred child with Asperger's syndrome and Jeff (approximately 6 years) who had developmental delay. The Niles family consisted of the mother (Sally), Alex (father to two, step-father to two children) and four children; Steve (14 years) the referred child, suspected ADHD, Nicola (12 years), Lee (8 years) and Kevin (3 years). Members of the Webber family were, Patrick (Step father to two, father to two children), the mother (Mandy), and four children; Daniel (15 years) the referred child with special educational needs, Adam (19 years), Patrick (10 years) and Stuart (8 years).
Each of these four families remained in family therapy and with mental health services more generally after the data collection period was completed. The actual outcomes of treatment, therefore, were not actively pursued as relevant to the research question. The data were transcribed in accordance with the analytic method and Jefferson guidelines were followed (Jefferson 2004). See Table 1 for detail.

Conversation Analysis
A distinct feature of conversation analytic (CA) work is its focus on the action orientation of talk (Hutchby and Wooffitt 2008). Through analysis, the sequential organisation of talk is explored to explicate the social actions being performed (Sacks 1992). For example the Arrows in this direction show that the pace of the speech has slowed down () Where there is space between brackets denotes that the words spoken here were too unclear to transcribe (()) Where double brackets appear with a description inserted denotes some contextual information where no symbol of representation was available.

Under
When a word or part of a word is underlines it denotes a raise in volume or emphasis : When an upward arrow appears it means there is a rise in intonation ; When a downward arrow appears it means there is a drop in intonation ?
An arrow like this denotes a particular sentence of interest to the analyst CAPITALS Where capital letters appear it denotes that something was said loudly or even shouted Hum(h)our When a bracketed 'h' appears it means that there was laughter within the talk = The equal sign represents latched speech, a continuation of talk :: Colons appear to represent elongated speech, a stretched sound semantic sentence 'what are you doing this evening?' could perform a variety of social actions depending on the context. It may be a simple question or it could be performing the social action of a pre-enquiry to an invitation or request. Social processes are revealed through close attention to sequential analysis of conversational turns which illuminates the way in which the participants in the interaction respond to prior turns. The reliability of this method is not constituted in the analysts' interpretations of the participant's talk, but in line with ethnomethdological principles, is grounded in the participants own responses. This method has great potential for illuminating insights into healthcare interactions as it enables the identification of patterns of behaviour (Drew et al. 2001). As CA has grown in popularity it has illustrated some of the fundamental organisational features and interactional processes in medical settings (Pilnick et al. 2010) and is used to examine the ways in which clinical processes are interactionally constituted in therapy (Georgaca and Avdi 2009). For this paper the two authors initially independently scrutinised the data corpus for the identification of social actions pertinent to the research question. During the second phase these social actions were jointly explored through a more detailed sequential analysis to secure inter-rater reliability. This process allowed the authors to explicate the emergent patterns of social process requiring further analytic attention, as is consistent with the CA methodology.

Ethics
During this project we employed the Principlist approach to ethics, incorporating the four core principles of autonomy, beneficence, non-maleficence and justice (Beauchamp and Childress 2008). What this meant in practice was that informed consent was collected from all necessary parties, anonymity was maintained, confidentiality assured and data were stored securely.

Analysis
By using conversation analysis to investigate the performative actions in institutional talk, our analysis revealed four social processes at work within the dynamics of the family unit during the practice of family therapy. First children display passive and active disengagement from the therapeutic agenda. Second, children attempt to express autonomy and evade adult impositions. Third, siblings are afforded greater liberty in their attempted disengagement. Finally, therapists use validation as a technique to reinstate engagement in the therapy process. Social Process One: Passive and Active Disengagement from the Therapeutic Agenda In this section we provide a series of extracts which present a continuum of social actions displayed by the children as a way of disengaging from therapy. These range from a behavioural passivity through to direct active verbal resistance. We illustrate that children passively disengage (through inattention), passively resist (when they do not attend to a direct question, or attempt at engagement), and actively resist (when they directly refuse to answer, or fail to comply with a request).

Extract one: Clamp family
Disengagement from therapy can be simply inattention to the process. By removing themselves from the therapeutic conversation, children display passive resistance to the social process. The children's laughter and jumping on chairs (lines 5 and 6) occasion the father to suspend therapy to attend to Ron and Jordan. Sequentially this rupture affords an opportunity for the therapist to initiate a topic shift (Jefferson 1984) and to make the behaviour of the children therapy-relevant (line 12).
Extract two: Bremner family This extract illustrates that children display more active strategies for inattention than simply passively disengaging themselves from the conversation. Here Bob's attention actively moves from the therapy process to an alternative activity, playing with children's building blocks. By actively attending to the building blocks and the on-going dispute with his brother, Bob passively resists attending to the question posed by the therapist 'Bob, how was it at Christmas?' (line 8, 11). Notably the therapeutic conversation involved negative descriptions of Bob's behaviour toward his mother (lines 1-3) from which Bob disengaged by actively verbally diverting the adults' attention to the play. This, like in extract 1, results in a topic shift as they discuss possession of the toy blocks.
Extract three: Clamp family Extracts one and two illustrated that the continuation of therapy is displayed as the primary objective of the adult parties, and disruptions to this process are treated as interference. Here the continuation of therapy requires the child to leave the therapeutic space due to the delicate nature of the topic (paedophilia 1 ). Research illustrates that delicate inappropriate topics require careful management in the therapeutic conversation (O'Reilly and Parker 2012b) and here the therapist works to remove the child from the overhearing position he is currently in. Interestingly when the child answers the question with the dispreferred response (Pomerantz 1984) 'nah' (line) both the mother and the therapist question this. They repeat the response 'nah?', 'no?' but the questioning intonation implies that the response ought to be revised. This occasions a downgraded, less emphatic version of the refusal as Ron shakes his head. Although acknowledged by both the therapist 'alright then' and the mother 'nah', the therapist enforces his original request from line 1, by actively and physically taking the child out of the room (line 10).
Extract four: Bremner family There are occasions in therapy where a therapist will use active engagement strategies to involve the children in the process and here the therapist uses first person selection 'Bob' (line 1) to directly address the child. Ostensibly saying 'would you like' offers Bob a choice to provide an explanation for the mother's visually obvious negative affective state. Notably, because the therapist is looking at Bob, addressing him by name, and emphasising 'you', it is problematic for Bob to display passive inattention, and therefore necessitates a more active response. In this case, Bob interrupts the therapist during her question and actively refuses to comply with the request 'no' (line 2) offering a justification 'I'm not in the mood' (line 2) and a candidate alternative respondent 'mummy can' (line 3). Although Bob references the mother as the next speaker, her distressed state occasions a minimal refusal 'mummy can't say anything' (line 5) which is audibly quieter, and in turn precipitates a self-selected answer to the question from Bob's sibling, Jeff.

Social Process Two: Expressing Autonomy and Evading Adult Impositions
There are two ways in which children express their wish for autonomy to disengage from the therapy. First they attend to the present interaction, making requests to cease participation, and second, they orient to future sessions by expressing desire not to continue attending. Building upon the previous analysis we demonstrate examples of children displaying active resistance to the process of therapy by initiating requests to disengage.
Extract five: Niles Family In this extract Steve's request to turn on his mobile telephone is an attempt to actively disengage from the therapy. This potential alternative activity is rebuffed by the parents who collaboratively account for the refusal by orienting to institutional rules imposed by hospitals. By illustrating to Steve that there are potentially severe consequences of his action 'you could kill someone' (line 6), they not only provide good reason not to allow the phone to be turned on, but also mitigate parental responsibility for the denying the request. Notably this account does not attend to the potential social action being performed by Steve, of active disengagement. This intersubjective misalignment occasions a second attempt to disengage from Steve, 'can't I just' (line 7) and 'can we go home' (line 9). At this point this is simply declined without any explanation 'no' (line 11). Parental imposition is not always without explanation and in extract six the parents position the child himself as the reason why disengagement is not possible.
Extract six: Niles family In this extract the child actively expresses autonomy to disengage from the therapy by requesting that the family leave 'can't we just go?' (line 1). The father's signal for not hearing the request, affords the opportunity for the child to reiterate it. However the request is upgraded by the footing shift (Goffman 1981) from 'we' to 'I', and the removal of the minimiser 'just'. The direct way in which the child's expressed choice is reformulated 'I want to go' (line 3) not only occasions a refusal, but also an account from the father. This account positions Steve as the problem which necessitates Steve's attendance.

Extract seven: Bremner family
The literature on preference organisation in adult-to-adult interactions illustrates that when questions such as the one offered by the family therapist are asked, they are designed to elicit a 'yes response' (Pomerantz 1984). Pomerantz notes that when adults offer a dispreferred response, it is notably marked by pauses, prefaces and accounts. Although Bob's response is semantically congruent with the therapist's turn in the sense that he applies the same modal verb, 'will you come' (line 1) 'I will not' (line 5), his response lacks any normative social conventions of a dispreferred response. While the therapist's question has the illusion of offering choice 'will you come back again' (line 1) her next turn 'oh I think so' (line 4) dispels this possibility as she orients to the expectation of his return. This illustrates the adult's imposition of expected attendance overriding the child's autonomy to choose disengagement from further sessions. The restriction of autonomy to choose to attend future sessions is expressed more explicitly in the following extract.
Extract eight: Niles family In this extract not only does Steve express a preference to disengage from the current therapy session, but he also expresses a clear desire not to attend any future sessions 'I don't want to come anymore' (line 4). This attempt at autonomy is met with two different types of responses from the adults in the room. Initially the therapist affirms his desire for Steve to attend 'I would really like you to come' (line 5), which indicates a personal preference. In contrast, the mother's response imposes a restriction of his liberty 'you don't 'ave much choice' (line 7) and enforces her parental authority 'I'm bringing ya' (line 7). Notably, the mother does provide a caveat to the imposition by demonstrating a time limit on attendance 'til we get to the bottom of this' (line 8). Despite this account, Steve's option for choice becomes further limited by the therapist aligning with the parents. Therapeutically, alignments between therapists and all parties, including children, are important for therapeutic processes (Parker and O'Reilly 2012), but here the therapist has actively disaligned from the child which is strengthened with the category use of 'adults'. Social Process Three: The Negotiable Liberty of the Sibling Illustrated previously, despite active and passive attempts at disengagement, parental imposition has dictated that the child identified as requiring help continues to attend therapy. However the necessity for siblings to attend appears to be something open to negotiation with the therapist. This demonstrates that it is not simply the category of 'child' in contrast to 'adult', or 'therapist' in relation to 'client' that defines the direction of autonomy and authority. The other children within the family are afforded a different degree of choice regarding engagement than the 'problem child'.
Extract nine: Niles family At the end of this therapy session the therapist offers a candid closing comment 'we'll see you in four weeks then' (line 1). The assumptive element of this closing statement problematises the pronoun 'you' by raising the possibility of Nicola's non-attendance 'she said she don't want to come again' (line 2). The father here legitimises the possibility of Nicola's non-attendance by voicing her preference, and notably the other siblings, Kevin and Lee, use the opportunity to attempt to express their autonomy. By interrupting the children, the therapist focuses attention on responding to the older sibling (Nicola), directly. He acknowledges her choice 'it isn't what anyone would choose' (line 8) and validates the value of her contribution 'I find it helpful what you say' (line 6). By saying 'it'd be nice if you'd come' (line 10), the therapist maintains the scope for autonomy but clearly defines a preference for attendance. This contrasts significantly with previous extracts where the 'problem child' is clearly given no choice in the matter of attendance. Extract ten: Webber family As in extract nine, the father here raises the issue that one sibling in the family has a preference not to attend the therapy. The father's account hinges on the discrepancy between being physically present and actual engagement in the therapeutic process. What he highlights is that even if they brought Adam to therapy, he would not actively engage by communicating with the therapist about events relevant to the 'problem child', Daniel 'he won't never ever speak about that' (line 3). Interestingly this account for possible nonattendance is not utilised for the situations where the 'problem child's' attendance is questioned or raised. Although in this extract the therapist states that therapy is not 'compulsory for anybody', the lack of choice for some children is clearly marked with parental imposition, as highlighted earlier.
Social Process Four: Validation as a Technique to Create or Reinstate Engagement Problematically, where parents impose attendance on their children and those children resist or disengage from therapy, it can create difficulty for meeting therapeutic goals. There is an onus therefore on the therapist to take responsibility for recognising the probability that children may not be willing participants, and to utilise strategies to create or facilitate their engagement. One of the ways in which this can be achieved is the circumspect use of validation as a clinical intervention. By acknowledging and validating the potential challenges for the child such as boredom, the unpleasantness of listening to certain descriptions and events particularly when related to them and their behaviour, and the uncertainty of what might happen, the therapist creates a space for the child which enables them to feel accepted.

Extract eleven: Niles family
This extract demonstrates the complexity of using validation as an engagement technique. Paradoxically the therapist here does not initially attend to the overtly expressed feeling conveyed by Steve 'I'm bored' (line 2), but does attend to the implicit implication that Steve is finding therapy uncomfortable by directing his question specifically to Steve. Notably the child's two attempts to disengage from the therapy, 'I'm bored' (line 2), and interruptively, 'I wanna go home' (line 6) are not attended to by the therapist as he pursues his line of enquiry. While children's interruptions are typically ignored (O'Reilly 2006), the validating social action of the therapist's turn in this instance is designed to address the potential difficulty for the child in hearing the negative descriptions of his behaviour. This redress of a potential social breach (Parker and O'Reilly 2012), of repairing the imminent rupture created by talking about Steve in a negative way, takes precedence over attendance to the process of the child's interruption. Validation of the child's difficulties in engaging in the process of therapy can be in itself a way of engaging the child. Contemp Fam Ther (2013) 35:491-507 501 Extract twelve: Clamp family In this therapy session where multiple family members are present including the parents, three children and the uncle Joe, the use of recipient selection 'you Phillip' (line 1) may be significant in securing the child's attention. This may function to prohibit other members from contributing and selects Phillip as the intended audience. The therapist uses a series of conversational processes, beginning with acknowledgement of the family's discussions about Phillip, validation of the difficulty for Phillip in listening to those discussions and culminating in attempts to reengage him in the therapy. The therapist begins with a reformulation of the series of negative ascriptions of Phillip and his behaviour that have characterised the preceding conversation. The therapist acknowledges his contributions to this talk by stating 'we did a lot of talkin' about some of the things that YOU do' (lines 3-5) which is an inclusive footing position. However, there is a footing shift (Goffman 1981) immediately following this as the therapist positions the judgement of Phillip's behaviour with his parents 'yer mum and dad aren't too happy about' (line 6). This sequential shift in alignment from talking with the parents moves from 'we' (the three adults), to 'they' (the parents), to an alignment with Phillip as he moves to engage Phillip more directly by acknowledging how he might feel about those discussions 'It's really difficult t' sit there and listen' (lines 6-7).
Extract thirteen: Webber family The same three processes of acknowledgement, validation and engagement, are also visible in this extract. The therapist displays an interpretation of Daniel's non-verbal behaviour as indicative of his affective state 'you're looking uneasy already Daniel' (line 2). This is followed up with the use of validation as the therapist comments on the difficult nature of the conversation and the difficulty Daniel may experience in contributing 'this isn't easy stuff for you to talk about' (line 4). The encouragement to engage Daniel is presented inclusively with a statement that it is 'important for us all to be able to talk' (line 10).

Discussion
The aims of this paper were to illuminate through empirical analysis some of the ways in which children attempt to resist and disengage from family therapy, and also which interventions from therapists are helpful in seeking to manage these processes. Our analysis revealed four social processes that relate to children's disengagement. Social process one considered how children's disengagement from therapy can be active or passive: passive disengagement was characterised by inattention to the therapeutic process; passive resistance was characterised by active attention to alternative activities; and active disengagement was displayed by verbally refusing to answer questions directed specifically to them. Social process two considered how children expressed their autonomy and evaded adult impositions. These were expressed verbally, conveying a desire to cease therapy either in the present moment or in the future, and were set up as contrary to adult expectations and wishes. Social process three considered the role of other family members in therapy, specifically exploring the more flexible obligations of attendance of siblings. Social process four explored how therapists attempt to create engagement or re-engage a child to repair any rupture that may have occurred.
Adult and children's adherence to treatments is considered to be an important aspect of healthcare (Osterberg and Blaschke 2005). Research has focused heavily on children's adherence to pharmaceutical treatment programmes with non-compliance having serious consequences for children's health (Butler et al. 2004;Osterberg and Blaschke 2005). Compliance with medical treatments has clear physical benefits to the child which become visible during the course of interventions and has potential to encourage future engagement with medical services. Importantly non-compliance in the talking therapies is less visible as the child is ostensibly present in the therapy which indicates immediate adherence. Problematically, the mere presence of the child does not guarantee their participation and this potentially renders the therapy ineffective. For example, using a medical metaphor, if a child hides medication under the tongue and later spits it out the treatment will not be effective; in therapy, without active engagement in the process of therapy, the intervention will not achieve its outcomes. Furthermore, not only will the therapeutic process be rendered ineffective, but it may also have an iatrogenic effect. As the children are listening to negative descriptions of them, which is common in family therapy (Parker and O'Reilly 2012), without recourse to contribute their own perspective, this may have a potentially damaging impact.
The literature indicates that we have a limited evidence base regarding how children engage with therapy (Strickland-Clark et al. 2000) and one way to explore this important issue is to investigate how children resist and disengage in practice. It is evident that analysis of the behaviour of children and families in therapy can be an important aspect of predicting outcomes (Kazdin et al. 2005). Our analysis illuminates the range of behavioural and verbal indicators of how children withdraw from the therapeutic process and how this is managed by the adults. Research with adult participants indicates that they withdraw or disengage from therapy when they sense something threatening developing, and use disengagement as a way of stalling discussion which may result in criticism from the therapist (Frankel and Levitt 2009). Parental criticism of children in therapy through the positioning of the child as the problem can lead to them being talked about in a derogatory way (O'Reilly and Parker 2012b). Sociological research illustrates that children possess social competencies of greater sophistication than is typically assumed (Hutchby 2002;Hutchby and O'Reilly 2010) and therefore disengagement from therapy could be understood as a mechanism for managing criticisms.
An understanding of children's contributions to family therapy through qualitative analysis facilitates an understanding of the process through which children disengage from services. This understanding of disengagement is useful in informing the broader context of attrition as cumulatively these disengaged moments can contribute to the failure of the therapy as a whole. This has important implications given that families are offered therapy to assist them when they experience violence, breakdown or juvenile delinquency (Hutchby and O'Reilly 2010) and thus failure in therapy has potential wider social consequences. To avoid dropout from family therapy it is important to consider the role the child plays. It is necessary to achieve more than just the physical presence of the children, but to prevent, recognise and manage disengagement while maintaining alliance with both the parents and children. Quantitative scales, such as the CTAS-R (Pinsof et al. 1994), have been designed to measure the possible discrepancies in strength of alliance between individuals in couples therapy (Knobloch-Fedders et al. 2004). The advantage of using conversation analysis to investigate alliances in family therapy is that it relies on observable data as opposed to self-reports and allows the analyst to examine alliance processes as they occurs in practice. Our analysis illustrates that validation as a way of recognising the difficulty for the child has potential to circumvent disengagement or facilitate re-engagement. The therapist therefore has some responsibility for attending to the passive and active disengagement strategies of the child in terms of recognising their occurrence and attending to the non-verbal indicators. This can be a complex task when the parents are especially active and it is easy to overlook the passive disengagement of quieter children.
By applying a micro-analytic approach to the social processes inherent within naturally occurring family therapy sessions, we are able to explicate the nuances of the interaction. This has allowed us to interrogate the sequential nature of therapeutic interactions in a way that highlights the process of children's resistance and disengagements. This has important implications for exemplifying wider social processes in order to broaden our understanding of approaches that may facilitate engagement. Families are an important social institution and our findings suggest that the mere presence of the child within the family unit does not necessarily equate to active involvement in family processes.
There are some limitations with the conversation analytic approach to data analysis, for example, while suggestions are made, the power to implement these recommendations lies with those who commission and practice (Antaki 2011). It can be difficult, however, for family therapists as consumers of research evidence to engage with and implement strategies due to barriers such as time and resources (Kosutic et al. 2012). Nonetheless research evidence is necessary for informing change and improving services and our analysis provides a benchmark for understanding the process of adult-child alliances in a family therapy setting. These principles also translate to other domestic situations, for example in family disputes, in terms of how children may competently resist alliance with or disengage from the family unit. Our findings also have broader implications for understanding children's compliance and engagement in other institutional settings such as education. In the classroom it may be helpful to consider similar patterns of how children's physical presence does not necessarily equate to their active engagement with pedagogy. Arguably therefore the strategies children use for resisting and disengaging from education may not be that different from therapy and thus this could be a useful area for exploration in future research.
The task for the therapist is to actively encourage engagement with the child and to circumvent disengagement and dropout regardless of the therapeutic model they adhere to. This can be a delicate endeavor as it is necessary to maintain alliances with both parents and the children, who may hold contradictory positions. It is clear that to yield the benefits of therapy, there is a requirement for children to do more than simply attend appointments, but to also be actively involved in the process.