SW: Assignment 1

Journal of Infant. Child, and Adolescenl Psychotherapy, 12:228-243, 2013 Copyright © Taylor & Francis Group, LLC ISSN: 1528-9168 print S " \ Taylor «.Francis Croup

DOI: 10.1080/15289168.2013.822751

I J Routledge g ^ ^ Taylor & Francis Crou

Irreducibly Human Encounters: Therapeutic Alliance and Treatment Outcome in Child and Adolescent Psychotherapy

Cameron M. Clark

The concept of therapeutic alliance is examined in relation to treatment ouicomes in youth psy- chotherapy. The first section of the article develops an intuitiva account wherein the relationship between the therapist and youth client is purported to hold great significance foi- treattnent effective- ness in psychological interventions. The remainder of the article aitns to validate this deep-rooted intuition via close examination of how therapeutic alliance has been defined, measured, and meta- analyzed in the current literature. Methodological and philosophical challengss to the concept and implications for future research and practice are discussed. CaulSous, but optimistic conclusions are drawn affirming the intuitive position that the relationship between therapist and youth client does indeed impact treatment outcome.

Treatments within the realm of psychology bear strong resemblance to those in medicine. The analogy is so obvious that it scarcely needs to be expounded. In both cases, there is a problem related to some aspect of health in a client that requires the specialized skills of a professional in order to effect a change process with the ultimate aim of improving the health and overall well-being of that client. From this perspective, the goal of reseaich in both medicine and psy- chotherapy is to strive to conceptualize, design, and develop the most effective treatments possible for each specific problem a client might face over the lifespan. The goal of the practitioners of these fields, physicians and clinical psychologists, is simply to apply these hard-won interven- tions appropriately to populations in need. Just as a broken îeg might be mended with surgical intervention, or casting and follow-up care, so too can episodes of major depression, or anxiety, or any other mental disorder be treated with an analogous intervention and follow-up treatment designed to ameliorate the presenting problem.

However, in many important ways, psychological treatmeits must be considered categorically different from those in medicine. As intuitive as the above metaphor between psychological and medical interventions may be, over application of that metaphor eventually leads to conflicting intuitions. For example, while we intuitively expect the efficacy of the intervention for a broken leg to be wholly independent of the relationship shared between the physician and the patient, we would likely expect the opposite from interventions in psychotherapy, which are essentially rich and complex interactions between two or more human beings in addition to the structured

Cameron M. Clark, MSc, is currently a PhD student in the clinical psychology program al the University of Calgary in Calgary, Alberta, Canada. His research interests primarily fall within the donain of neuroscience and neuropsychology.

Correspondence should be addressed to Cameron M. Clark, Department of Psychology University of Calgary, 2500 University Drive, Calgary, AB T2N 1N4 Canada. E-mail: clarkcm@ucalgai:y.ca

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 2 9

application of empirically supported therapeutic techniques to client problems. Where a cold and indifferent physician wields little power to alfect the outcome of a simple medical procedure, the same traits in a psychotherapist could subvert the goals of therapy completely. In this sense, psychotherapy is not done to the client as some medical procedures might be, but rather with the client. The therapist and the client must ally in building a productive therapeutic relationship or, as it is formally called in the literature, a "therapeutic alliance." Stated simply, the relationship shared between the client and psychotherapist is paramount, if not tantamount, to the success of the intervention as a whole. If this is the case for adults in psychotheiapy, one would intuit that the relationship would be even stronger in pédiatrie and adolescent populations that rely heavily on authoritative relationships in many, if not all aspects ol' their lives, and rarely if ever refer thetnselves for treatment.

This is, at least an intuitive account of the importance of the therapeutic alliance in youth populations. The remainder of this atticle will essentially be a test of this intuition against the empirical literature regarding therapeutic alliance in relation to treatment outcome in children and adolescents. Squaring clinical (or perhaps common sense) intuition with empirical fact is particularly important in this domain due in part to the pervasiveness of the intuition, and com- parative paucity of literatute on the topic. In 1990, Kazdin, Siegel, and Bass conducted a survey of more than 1,000 clinical psychologists and psychiatrists and found that more than 90 percent of practitioners believed the relationship between the child and therapist to be important to the process of therapeutic change, whereas only apptoximately 65 percent deemed the training or experience of the therapist to be important to this process. Additionally, in reviewing the liter- ature on evidence-based treatments for the mechanisms of therapeutic change (i.e., therapeutic alliance, among other things) Weersing and Weisz (2002) found that only 63 percent of the stud- ies examined evaluated a potential mediator of therapeutic change at all, and only 9 percent of the studies they exatnined included a formal test of statistical mediation. Thus, there appears to be a large gap between therapists' intuitions about what mediates the process of therapeutic change and the empirical basis for holding such intuitions.

To reconcile clinical intuition with empirical fact regarding therapeutic alliance, I will briefly address past and current conceptions of the concept in the first section and methods of measure- ment in the second section before turning to an investigation of the latest etnpirical literature on the subject in the third section, devoting particular attention to potential mediators and modera- tors of its effect on outcome. For example, does the relationship between therapeutic alliance and treatment outcome vary actoss child gender, child age, presenting problem, or treatment modal- ity? Why or why not? Next, I will discuss some challenges to the idea of therapeutic alliance in the fourth section including pervasive methodological issues in the current literature (e.g., is ther- apeutic alliance what Kazdin might call an "arbitraty metric") and the challenge of efficacious computer-aided treatments which would seem to write the therapist out of the treatment script completely—or perhaps not. Finally, I will discuss prescriptions and proscriptions for research and clinical practice in the fifth section and make discuss conclusions and implications of this research in the sixth and concluding section.

THE THERAPEUTIC ALLIANCE DEFINED

So what exactly is therapeutic alliance? Much of the empirical work on the concept has centered on its role in psychotherapy within adult populations, despite a noticeable increase in interest in

230 CLARK

its extensions into the child and adolescent literature in recent decades (Elvins & Green, 2008). Indeed, even as late as 2002, when Division 29 of the American Psychological Association (APA) produced a large-scale review of empirically supported relationship variables (Norcross, 2002), research on children and adolescents was completely omitted. However, it should be mentioned that the subsequent version of this work (Norcross, 2011) did include a single chapter on ther- apeutic alliance and relationship variables in youth. Thus, a historical review of the concept of therapeutic alliance necessarily places emphasis on the adult literature. However, modeling the chronological progression of the concept itself, a basic understanding of the early conceptualiza- tion of the concept in adults will prove indispensable in elucidating its downward application to the child literature over the last several decades.

In adult populations, the importance of the therapeutic relationship in the treatment of mental disorder and illness can be traced back to the work of Freud himself. In his early writings, Freud spoke of making the patient a "collaborator" in the treatment process and acknowledged the role of affection and friendliness as "the vehicle of success in psychoanalysis" (Freud, 1912/1953, as cited in Safran & Muran, 2000, p. 7). Needless to say, Freud's work proceeded to inspire multiple generations of thinkers and practitioners, all of whom have emphasized the importance of understanding and working with the relationship between patient and therapist as critical to the therapeutic process. Though the details of this historiography are beyond the scope of this article, through the work of many generations of psychoanalysts, psychodynamic practitioners, psychologists, psychiatrists, and social workers, Freud's initial insight would eventually play a key role in the work of such practitioners as Carl Rogers (1957) who suggested that the emotional quality of the therapeutic relationship was in fact the primary agent of psychotherapeutic change in the treatment process (Green, 2006). Rogers' client-centered approach then served to further enshrine the value of positive, warm, and productive relationships between clients and therapists as essential for client persistence in therapy, motivation to work on problems, compliance with therapy tasks, and ultimately positive outcomes (Hawley & Weisz, 2005).

These views proliferated during the early second half of the 20th century and were reviewed, summarized, and elaborated on by Edward Bordin in 1979, whose work serves as the basis for the notion of therapeutic alliance broadly used in the literature to this day. Bordin (1979) con- ceptually distinguished between three unique varieties of alliance, each present to some degree in all aspects of psychotherapeutic relationships: agreement on the goals of therapy, collaboration with regard to specific tasks of therapy, and the joint development of personal or affective bonds between the therapist and client. These three aspects of the special psychotherapeutic relationship, goals, tasks, and bonds, Bordin argued, where jointly responsible for the overall effectiveness of any psychotherapeutic endeavor. A brief overview of each of these domains of alliance will be instructive for framing methodological and conceptual issues discussed below.

Bordin (1979) argued that the goals of therapy, are often implicitly conceptualized before the first meeting of patient and therapist, and are central to the quality of the subsequent relation- ship. Thus, these "prior bargains" should not be overlooked in treatment. He argues that because traditional psychotherapy assumes that the client's maladies are to a large extent the result of maladaptive ways of negotiating conflicts presented by the environment, the goal of therapy is to develop an enhanced method of negotiating that conflict. Without agreement on this basic tenet, Bordin argued, psychotherapeutic techniques are not likely to succeed.

Collaboration between patient and therapist regarding the specific tasks of treatment is also central to strong therapeutic alliance and, therefore, a positive treatment outcome. Bordin (1979)

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 3 1

draws attention to the concrete exchanges that occur between thetapist and client. Not only must the therapist and client agree on the specific tasks designed to facilitate therapeutic change (which may vary widely across treatment modality), the therapist and client must also execute their responsibilities within the therapeutic relationship. The client must pay the therapist for his ser- vices, and the therapist in turn must be a therapist via designating therapeutic tasks, effectively communicating, interpreting, and empathetically understanding the client and his or her behavior.

Finally, perhaps the most intuitive aspect of Bordin's conceptualization of therapeutic alliance is that of the bond between the therapist and the client. Bordin (1979, p. 254) notes: "Some basic level of trust surely marks all varieties of thet apeutic relationships, but when attention is directed towatd the more protected recesses of inner experience, deeper bonds of trust and attachment are required and developed." Safran and Muran (2000, p. 12) summarize the bond aspect of thetapeutic alliance as the "affective quality of the relationship between patient and therapist (e.g., the extent to which the patient feels understood, respected, valued, and so on)." Each of these three aspects of alliance is thought to dynamically interact such that agreement on goals and tasks of treattnent influence the type of bond that might exist between patient and thetapist, and that relationship then reciprocally influences the desire for collaboration in treatment goals and tasks (Bordin, 1979; Safran & Muran, 2000).

One additional effort to define the concept of therapeutic alliance is worthy of mention here, for its empirical backing and influence on subsequent literature. Hourgaard (1994), drawing heav- ily on the work of Bordin (1979), empirically validated a generalized form of the theiapeutic relationship via factor analysis of more than 600 manualized treatment sessions with adults suffering from depiession. The results supported a two-factor model, encompassing all of the areas accounted for in Bordin's conceptualization, and more. His model, depicted in Figure 1, essentially differentiates two main factors of therapeutic alliance between client and therapist: the personal alliance roughly cortesponding to the interpetsonal relationship between client and therapist, and the task-related alliance addressing the contractual elements of the therapeutic rela- tionship. Although Hougaatd's work seems to simply recast Bordin's (1979) three-factor model into a simplified two-factor version accounting for essentially the same variables, Hougaard notes that empirically garnered intercorrelations between dimensions favor his two-factor model.

Hougaard (1994) also listed several aspects of the relationship that the therapist and client might individually and jointly contribute to the relationship. For example, the therapist might contribute to the personal alliance via authenticity, warmth and acceptance, and so on, and to the task-related alliance via expertness and engagement. Conversely, the patient might contribute to the personal alliance via confidence, friendliness, and compliance and to the task-related alliance via motivation and positive expectations among other things. Finally, the therapist and the client might jointly contribute to personal alliance via mutual understanding and to the task-related alliance via agreement on goals and tasks of treatment. (See Figute I for a representation of these diffeiential contributions.) The strength of Hougaard's model is that each of these individual and joint contributions to overall therapeutic alliance can be tested empirically, and all mea- sutes of thetapeutic alliance can be scrutinized for which patticular domains of the therapeutic alliance they are actually measuring. Essentially, Hougaard's model provides the most expansive framework of the concept, allowing for greater perspective when considering the definition or particularly the tneasurement of therapeutic alliance and serves as the best answer to the question that began this section: therapeutic alliance is an umbrella term roughly corresponding to all the relationship variables outlined by Hougaard's comprehensive model (Elvins & Green, 2008).

2 3 2 CLARK

"Ci»

Tl» 'f I>Ç2) warmtli & ucc«|>t.uiec •(•|»(3) empathy

'Vu&k alltance

CiXI) ugrtenient on g<iiah! <.'T{2)i»i|ir<«n«nt on k

t tiâion ú>paíi0tu{ Musée'

FIGURE 1 Components of the therapeutic alliance (adapted from Hougaard, 1994).

Armed with this better definition of therapeutic alliance, the juxtaposition between medical interventions and psychological ones gains a degree of clariiy: though the mending of a broken leg likely does not require any agreement on the goals of treatment, collaboration in tasks of the intervention, or a deep personal bond between physician and patient, treatment of mental disorders is indeed categorically different. Clients of psychological treatments must be persistent in therapy, remain motivated to work on problems, and comply with therapy in general, all of

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 3 3

which are intuitively influenced by the relationships between therapist and client. So, given this framework, what is the relationship between it and treatment outcome in child and adolescent populations? Before answering this central question, I will briefly discuss several of the ways in which therapeutic alliance has been measured over the last several decades.

MEASURING THE THERAPEUTIC ALLIANCE

A recent empirical review of the conceptualization and measurement of therapeutic alliance (Elvins & Green, 2008) revealed a varied history of measurement of Ihe concept evidenced by the existence of 32 independently designed and noimed rating scales created between 1962 and 2007. The authors of the review note that the conceptual reworking of the concept over time (briefly touched on above), in combination with a lack of consensus within the field regarding a gold standard measure, has contributed to the proliferation of tests over this 45-year period. Unfortunately, despite the efforts of Bordin (1979) lo forge a coherent conceptualization of ther- apeutic alliance, a content analysis of the 32 measures designed both before and after Bordin's work reveal that these alliance scales measure widely divergent dimensions of the alliance con- cept, and even largely difter wilh regard to which party in the therapeutic relationship ought to make the ratings of alliance (i.e., therapist, client, or observer). Perhaps more disconcerting to the present analysis, Elvins and Green (2008) note that of these 32 measures of therapeutic alliance, only two were specifically designed for use with the child and adolescent population: the Family Engagement Questionnaire (FEQ) (Kroll & Green, 1997) and the Therapeutic Alliance Scales for Children (TASC) (Shirk & Saiz, 1992). One recent ostensible exception here is the Therapy Process Observational Coding System-Alliance Scale (TPOCS-A) (McLeod & Weisz, 2005), designed with the expres.sed purpose of ameliorating shortcomings of prior scales. However, it too was designed from a mixture of items from other scales originally designed for use in adult populations.

Although several scales have been adapted for use with youth populations, several researchers have argued that even these are inappropriate for use with children and adolescents. Therapeutic alliance, Ihey argue, between child clients and therapists is inherently different that that shared between adults and therapists due to the vastly divergent contexts by which they find themselves in therapeutic environments. Intuitively, relationship variables in child and adolescenl therapeutic relationships may be even more important ihat in adult populations, as children rarely refer them- selves to treatment, may lack awareness of or insight into their clinically significant problems upon entering treatment, and may also be in conflict with their parents or guardians (Shirk & Saiz, 1992; Elvins & Green, 2008). For these and other reasons, Di Giuseppe, Linscott, and Jilton (1996) aigued that contrary to Bourdin's three-factor model or Hougaard's two-factor model of alliance (both primarily conceptualized in regard to adult therapeutic relationships), alliance in youth should be conceptualized as a single-factor. Their reasoning was that because children and even adolescents face specific and general developmental constraints, they may not naturally or easily distinguish between different conceptual aspects of the relationship, effectively collapsing all distinguishable factors salient for adult therapeutic relationships into a single category. If this is true, it follows that failure to establish any one aspect of the alliance with youth populations could effectively result in failure to establish any semblance of alliance at all.

234 CLARK

Despite these ongoing difficulties with the concept and measurement of therapeutic alliance, it will be helpful to see the actual content of the some of these scales. To this end. Table 1 lists items corresponding to each of the components of Hougaard's (1994) general model represented in Figure 1. The fact that items had to be taken from several of the most prominent measurement scales to cover all of the elements of the model is a testament to the fact that each of these scales is measuring quite different domains of therapeutic alliance, whether knowingly or unknowingly. Despite the several decades of research and thought mentioned earlier, Elvins and Creen (2008, p. 1184) end their summary of the empirical literature on therapeutic alliance on a rather dis- mal note: " . . . the alliance concept has remained essentially at the descriptive level, with little rigorous research as yet into the underlying processes behind its formation."

With these definitional and measurement cautionary notes in mind, we can now turn to an investigation of the recent empirical findings regarding therapeutic alliance in youth populations so that we might attempt an answer to the central question of this article: Is the relationship between therapist and youth client important for positive treatment outcome?

DOES THERAPEUTIC ALLIANCE AFFECT TREATMENT OUTCOME?

Considering that the majority of work regarding the definition and measurement of the therapeutic alliance has been in relation to adult populations, it should come as no surprise that empirical studies assessing its relationship to treatment outcome have generally followed the same pattern. A brief tour through the main findings of this literature wiB be helpful in contextualizing the discussion of the relationship in the context of youth therapeLtic alliance that follows.

Though several early reviews and reports of specific findings of the adult literatute pointed to significant effects of therapeutic alliance on treatment outcone, it was not until Martin, Carske, and Davis's (2000) authoritative meta-analytic summary of 79 separate studies that the finding was considered robust—and the field consequently began to take notice. Their review concluded that there was in fact a positive although modest association between therapeutic alliance (mea- sured via a variety of the scales discussed above) and treatment outcome, with a combined effect size of .22, falling within Cohen's (1988) criteria for a "small ' effect. Although this main finding was not particularly surprising given the intuitive importance of the therapeutic relationship, what was surprising was the lack of support for several of the many key moderator variables hypoth- esized to impact the relationship. Martin and colleagues (2000) noted that the relation between alliance and outcome remained constant regardless of whether the alliatice ratings were com- pleted by the therapist, the client, or a neutral observer, and even held regardless of when the alliance measures were taken in treatment.

If the relationship between alliance and outcome is present, albeit small, in the treatment of adults, intuition holds that the relationship must also be present in the treatment of children, and likely to a greater degree. As outlined briefly above, a variety Df reasons have been posited for the strength and tenacity of this intuition throughout the literature: children rarely refer themselves for treatment, may lack critical insights into their own clinical condition via lack of experience or a variety of other developmental variables, may present with concurrent problems related to the primary caregiver, and alliance formation can be a particulaily challenging problem with youth (Shirk & Saiz, 1992; Elvins & Green, 2008).

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 3 5

TABLE 1 Operationalizations of tfie subcomponents of the Therapeutic Alliance (adapted from Hougaard, 1994)

TP(lj: Therapist authenticity: Therapist is leal and genuine with the patient' Therapist nearly always expresses what he is feeling and thinking as he says it'

TP(2): Therapist wantith and acceptance: Therapist respects patient as a person' Therapist is friendly and warm with patient'

TP(3): Therapist unconditionality: How much therapist likes or dislikes patient is not altered by anything that the patient tells him about himself' (-) Depending on patient's behavior, therapist has a belter opinion of the patient sometimes than he has at other times'

TP(4): Therapist empathy: Therapist understands the whole of what patient means' Therapist appreciates exactly how the things patient experiences feel to him'

Tr(l): Therapist expertness: Therapist conveys the idea that he is competent to help with patient's problems^ Therapist focuses consistently on understanding the patient with the goal of resolving his problems^

TT(2): Therapist engagement: Therapist commits himself and his skills to help the patient to fullest extent possible^ Therapist is alert and engaged in the interaction

TT(3): Therapist efforts to strengthen patient expectation.'!: Therapist conveys a sense of hopefulness that treatment goals can be achieved' Therapist expresses hope and encouragement, a belief that the patient is making (or can make) progress^

PP(I): Patient confidence: Patient trusts therapist's commitment and motives'* (-) Patient acts in a mistrustful or defensive manner toward the therapist^

PP(2): Patient friendline.<!s:

(-) Patient acts in a hostile, attacking, or critical manner toward the therapist^ Patient is friendly and agreeable

PP(3): Patient compliance: Patient is responsive to therapist's effort to help' (-) Patient engages in a power struggle with therapist^

PP(4): Patient receptiveness of entpathy: Patient appreciates being understood (-) Patient held therapist at arms length with floods of words'

PT(I): Patient working capacity: Patient explores his or her contribution to problems' Patient self reflects on thoughts, feelings, and behaviors'

PT(2): Patient motivation: Patient indicates a strong desire to overcome his problems^ Patient is committed to work with his problems in therapy

PT(3): Patient piKsitive expectatiotts: Patient believes therapy is helpful' (-) Patient is worried about the outcome of therapy*

CP{1}: Mutual liking: Therapist and patient like each other Therapist and patient are warm and friendly toward each other

(Continued)

2 3 6 CLARK

TABLE 1 (Continued)

CP(2): Mutual understanding: Therapist and patient understand each other' The patient feels a rapport with the therapist, he or she feels understood [and accepted]^

CP(3): Agreement on the degree of intimacy: Therapist and patient agree on the level of intimacy in the relationship Patient and therapist aie satisñed with the emotional quality of the relationship

CP(4): Agreement on the degree of directiveness: (-) Patient would prefer more clear-cut directives and advice fron" therapist Therapist is responsive to patient's need for autonomy and self-determination in therapy

CT(1): Agreement OR goats of therapy: (-) Therapist and patient have different ideas on what the patients problems are* Patient and therapist shares similar conceptions about the etiology of his or her problems'

CT(2): Agreement on the tasks of therapy: Therapist and patient agree on what is important for the patient tc work on in therapy* Therapist and patient are working together in a joint effort^

'items from Barrett-Lennard Relationship Inventory (Barrett-Lennard, 1986). ^Items from The Vanderbilt Therapeutic Alliance Scales (Hartley & Strupp, 1983). ^Items frotn The Penn Helping Alliance Scales (Alexander & Luborsky, 1986). ••items from The Menninger Alliance Rating Scale (Allen et al., 198̂ )̂. 'items from The Califomia Therapeutic Alliance Rating System (Marmar, Weiss, & Gaston, 1989). *Items from The Working Alliance Inventory (Horvath & Greenberg. 1989).

This commonly held intuition largely remained so until 2003 when Shirk and Karver con- ducted a meta-analysis of the 23 available studies composed over a 27-year period that examined some aspect of therapeutic alliance and treatment outcome spanning a wide variety of present- ing problems, therapeutic techniques, client ages, and therapeutic alliance measures. The study sought to investigate the overall strength of relationship between therapeutic variables and treat- ment outcomes in youth populations across these variables and determine if, and to what extent, these and other intuitive variables moderated that relationship. Their findings indicated a mean weighted effect size of .20, quite similar to that found in the adult literature by Martin et al. (2000). However, the most surprising finding, again similar to Martin and colleagues was the lack of significant moderating variables. Shirk and Karver found that alliance impacted treatment outcome fairly consistently across diverse types and modes of treatment Moderating variables included both substantive variables (i.e., related to the client) and methodological variables (i.e., related to the study).

First, the only substantive variable found to moderate the relationship between alliance and outcome was type of patient. Stronger associations were found for youth presenting with exter- nalizing problems versus internalizing problems. Shirk and Karver (2003) point to literature suggesting that client hostility can subvert alliance, and note that for this reason forming a rela- tionship with an externalizing youth can be both more difficult and more crucial for good outcome compared with children with internalizing problems.

Next, Shirk and Karver (2003) noted several distinct methodological variables that moderated the alliance-outcome relationship. First, perhaps not surprisi.igly, ihough contrary to findings in the adult literature, measures of alliance taken late in therapy were more strongly associated

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 3 7

with outcome, which is inconsistent with findings from the adult literature. So either children develop relationships more slowly in thetapy than adults or, mote problematically, measures of alliance late in treatment with youth populations are simply confounded with the outcome itself. For example, if a child, parent or therapist feels that the sessions have gone well and the youth has made significant progress in sytnptom reduction or social function, any of these parties might be more apt to rate the alliance more favorably. I will return to this critique I in the next section.

Second, and again perhaps not suiprisingly, alliance ratings made by treatment providers were found to be tnore strongly associated with outcotnes than those made by either the children or adolescents themselves. Shirk and Karver note that this is again opposite of what has been found in the adult literature, and there is some evidence in the literature to support the idea that youth ratings of alliance demonstrate a restricted tange, clustering around the positive end of the spec- trum, thereby biasing their ratings towards discordant relationships with outcome in cases of poor outcome. Finally, studies in which the same person rated both the alliance and outcome measure had significantly greater effect sizes than those which garnered ratings from different people.

Despite these much-needed efforts to clarify the nature of relationship variables in youth psy- chotherapy. Shirk and Karver noted that there are still many aspects of the therapeutic relationship left ambiguous and unexplored. For example, their techniques lacked the ability to determine if the therapeutic alliance affected treatment outcome directly, or via one or more other interven- ing variables. More problematic, they were not able to conclusively determine the direetion of causality in the relationship to a large extent, leaving the possibility of confounding between the two concepts of interest an open po.ssibility.

Subsequent large-scale reviews have revealed similar results to those of Shirk and Karver (2003), with a few notable inconsistencies. McLeod (2011) reviewed 38 studies in a meta- analysis, including seven doctoral dissertations, and arrived a weighted mean effect size of .14, slightly lower than previous estimates of the alliance-outcome relation in both the adult and youth literature. McLeod suggests that his effect size estitnate is likely a better representation of the true mean effect, as it includes a greater number of studies than previous meta-analytic reviews and appears to be mote robust with regatd to publication bias critiques. However, McLeod also acknowledges that his may be a relatively conservative estimate because it recorded nonsignifi- cant results from several studies as having an effect size of zero, negatively biasing the results if some small effect size was in fact found. Contrary to Shirk and Karver (2003), McLeod found that child age did moderate the relationship, with children younger than 13 years of age show- ing higher associations between our consttucts of interest than those above age 13. Interestingly, McLeod also found that referral source impacted effect sizes for the alliance-outcome relation, with youth who referred themselves to treattnent showing significantly higher effect sizes than those who were referred to treatment.

One final meta-analysis is important to mention here. Shirk and Karver (2011) re-reviewed the literature some eight years after their initial publieation and arrived at a mean weighted effect size of .22 on the basis of 16 studies. Consistent with iheir prior work, they found some familiar moderating variables, including timing of alliance measure in treatment (with later measurement of alliance having higher effect sizes), child age (with children under 12 showing greater effect sizes than children over 12), and treated problem (with higher effect sizes in youth with exter- nalizing disordets that those with internalizing disorders). The authors note that their estimate of mean effect size tnay be slightly inflated due to the inclusion of studies that only measured

238 CLARK

alliance near the end of therapy, thus leaving open the possibility of confounding the proposed predictor and criterion.

So, after this extensive review of the literature from several authoritative researchers, this is what we know: in accordance with intuition, the relationship shared between youth client and the therapist does indeed appear to have a small impact on the outcome of that treatment. This relationship generally appears to be more important for younger children and those with external- izing types of problems, but interestingly appears unaffected by the particular type of treatment modality utilized. However, this extensive literature still leaves the field with many as of yet unanswered questions. These meta-analytic reviews do not shed light on whether alliance has its effects on outcome directly or via an intervening variable, or elaborate on the conditions under which alliance might thrive. And, perhaps most concerningly, these results do not provide suffi- cient evidence to conclusively state that it is in fact alliance influencing outcome rather than the other way around. McLeod (2011, p. 614) sums up the situation:

The alliance explain[s] a small proportion of the variance in clinical outcomes. However, the uneven approach to alliance measurement, the variability in the quality of the studies, and the presence of substantive moderators raise questions about the accuracy of the current estimate. Given the number of theoretical and methodological moderators, it appears premature to conclude that the alliance plays a minimal role in youth psychotherapy.

For these, and other reasons some authors have critiqued and challenged the notion of therapeu- tic alliance. I'll discuss some of these critical perspectives in the next section before turning to prescriptions for clinical practice and concluding remarks.

CHALLENGES TO THE CONCEPT OF THERAPEUTIC ALLIANCE

Aside from some of the methodological issues mentioned between meta-analytic findings above (i.e., common rater and time of measurement confounds; summarized well by Kazdin & Nock, 2003), some fairly serious criticisms have been leveled against the concept of therapeutic alliance. First, Elvins and Green (2008) note that the concept of therapeutic alliance matches the descrip- tion of what Kazdin (2006) would call an "arbitrary metric," that is, a latent variable assessed via a measurement wherein relationship between the two is not well understood. The problem here is that small changes on any of the numerous measurement scales are not well understood in terms of their meaning regarding where the client then stands on the underlying variable. In essence, "to what extent do small differences in scalar alliance ratings by patient or therapist ecological or conceptual validity?" (Elvins & Green, p. 1181). The present literature simply does not have an answer to this question.

Potential third factor confounds also present a problem for the validity of the alliance-outcome relationship. Where the traditional intuitive view holds that better alliance building on behalf of the therapist results in better or worse outcomes for youth in therapy, Elvins and Green (2008) discuss a small literature that suggests that at least a small portion of observed alliance is due to factors the youth brings to the table. Kazdin and Whitley (2006) found baseline social functioning to explain some of the alliance, and Kivlighan, Patton, and Foote (1998) found that interpersonal style of the client is also moderately predictive of alliance. Elvins and Green (2008, p. 1181 ) note:

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 3 9

In child therapy there are few rigorous studies attetnpting to delineate the relationship between alliance and outcome controlling for patient characteristics prior to symptom change. Establishing a causal relationship between alliance and outcome in the tuture will therefore need careful control of the contribution of symptom change to therapeutic alliance . . .

The challenges to therapeutic alliance thus far have been methodological. Future research might (and likely will) address these issues over the next several decades. There is, however, one partic- ular brand of challenge to the concept from which it is conceivable that it may never recover: the challenge from technology. What if effective tteatment for mental disorders required no therapist at all? Or, at least in the sense described in in the opening paragraphs, what if therapy was some- thing the therapist could do to the client, rather than with the client? Jain (2011) outlines a small literature noting that at least in the treatment of posttraumatic stress disorder, positive and stable therapeutic relationships can be developed over the internet. Though this discussion is quickly followed by a disclaimer about the limits of the generalizability of these findings, the challenge remains.

Chu and colleagues (2004) discuss virtual reality treatments for anxiety disorders and a plethora of computer and multimedia guided cognitive-behavioral techniques for relaxation, sys- tematic desensitization, self-exposure, and cognitive restructuring. The authors (p. 50) note that the limited research available on the use of computer-assisted therapy programs for adults "are no less effective for circumscribed, structured cognitive and behavioral interventions than tradi- tional therapist guided treatments, and are acceptable to patients." Chu and colleagues also draw attention to the fact that technology in therapy has been associated with a variety of other benefits such as promoting increased access to mental health services, creating user-friendly interfaces, and increasing standardization of care. Though the literature remains insufficient to make rash conclusions about the itnportance of the therapeutic relationship, certainly in youth populations, the authors note that to the contrary, the fear of technology negatively affecting the therapeutic relationship simply has not been substantiated. One possibility for these findings is that therapeu- tic alliance may have its effects on treatment outcome via a variety of mediating variables, such as motivation for treatment, which might even be bolstered by innovative and engaging imple- mentations of technology in therapy. Though more research is clearly needed in this area, the application of technology to therapy certainly does not write the therapist or his relationship with clients out of the picture, though it does perhaps force us to think more critically of the nature of that relationship.

PRESCRIPTIONS AND PROSCRIPTIONS FOR RESEARCH AND CLINICAL PRACTICE

Given this empirical landscape, what are the researcher and clinician to do regarding therapeutic alliance? For the researcher, the aim is quite simple: to fix the above-mentioned methodological problems. Though this task is easier said than done, the statistical and conceptual tools needed to increase the rigor of alliance-outcome studies already exist; it is simply a matter of taking the comments in the limitations sections of previous reviews seriously and putting improved methodologies into practice. These are large problems, but ones familiar to research in psychol- ogy, and ones for which the mechanization of modern psychology has been designed to solve.

240 CLARK

Alternatively, it has been suggested by some that because the effect sizes for the relationship factors seem to be as large as the effect sizes observed for the efficacy of treatment programs themselves, research should proceed with equal emphasis on relationship variables and treatment variables (Green, 2006), though it seem would seem that such an endeavor should be held off until the methodologically muddy waters of the alliance-outcome relationship have been cleared by future research.

The problem facing the clinician is slightly more difficult from a prospective point of view. In light of this information, how is the clinician or therapist to amend Iheir practices to best positively influence treatment outcome? Norcross and Wampold (2011) discuss conclusions garnered from the more than two dozen meta-analyses included in the updated version of Norcross's (2002) book on relationships in psychotherapy commissioned by division 29 of the APA to investigate relationship variables in psychotherapy. Integrating findings from across both adult and youth populations, the authors make recommendations for research and prac- tice. After noting many of the methodological issues that have plagued investigations of the alliance-outcome relationship across all treatment populations and outlining suggestions for fix- ing these issues in future research, the authors turn to a discussion of what works and what does not work in psychotherapeutic relationships. Variables that worked fell into three expertly judged categories: demonstrably effective, probably effective, and promising but insufficient research to judge. Among elements considered to be in the demonstrably effective category were alliance in youth (and adult) psychotherapy, empathy, and collecting client feedback. Variables deemed probably effective were goal consensus, collaboration, and positive regard. And finally, congruence/genuineness, and repairing alliance ruptures were listed as being promising relation- ship variables for future investigation, although insufficient research exists at this time to deem them significantly related to outcome. Conversely, what does not work is the opposite of these:

One means of identifyitig ineffective qualities of the therapeutic relationship is to simply reverse the effective behaviors. Thus, what do not work are poor alliances in individual psychotherapy . . . Paucity of empathy, collaboration, consensus, and positive regard predict treatment dropout and fail- ure. The ineffective practitioner will neither seek nor respond to client feedback, will ignore alliance ruptures, and will not be aware of his/her countertransference. And less effective psychotherapists wifl rarefy tailor or customize treatment to patient characteristics beyond diagnosis. (Norcross & Wampold, 2011, p. 427)

Norcross and Wampold (2011) identify several other factors of dysfunctional psychotherapeu- tic relationships garnered from scouring the literature and expert polls. Confrontation, negative processes (e.g., hostile, pejorative, critical, rejecting, and blaming behaviors), therapist centricity (e.g., privileging the therapist's experiences over the clients), and fidelity without flexibility (i.e., structuring treatment too rigidly) have each been empirically or reportedly associated with poorer outcome.

Moving to prescriptions for the youth therapist specifically. Shirk and Karver (2011) discuss several salient findings from the literature. Creed and Kendall (2005) found that for anxious chil- dren in the early stages of treatment, presenting therapy as a collaborative or team effort between the child and therapist, and helping the child set goals for therapy were positively associated with child rated alliance measures. Conversely, therapist behaviors that were associated with lower measures of alliance were "pushing the child to talk" about anxious experiences, "emphasizing common ground" (e.g., using comments such as "Me too!"), and talking to the child in an overly

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 4 1

formal manner. Generalizing these and other findings to the entire youth population. Shirk and Karver (2011) note that setting goals early in treatment may be particularly important for youth- therapist alliance as it likely helps the child view the therapist in a different light than parents and caretakers, and also helps establish the therapist as an ally. They also draw attention to Chu and Kendall's (2009) results which showed therapist flexibility in individualizing manualized treatments was also significantly associated with higher child involvement in therapy. Relatedly, a small number of studies with adolescents have investigated the temporal patterns of therapist behaviour in relation to alliance and have generally found significantly higher alliance scores when the therapist began therapy with less structure in the first session, before progressing to highly involved cognitive behavioral modules in later sessions. From all this. Shirk and Karver (2011, p. 87) conclude:

Efforts to engage the child by pushing or praising are contraindicated. This patter n is somewhat incon- sistent with the emphasis on psycho-education as a method of tr-eatment socialization and suggests that client-centered strategies at the start of therapy may be trrore effective for alliance formation.

Additionally, Shirk and Karver (2011) note that due to the unique complexities of youth popula- tions, therapists ought to be awar'e of alliance in relationships with parents as well as youths, as parents often make the final decision as to whether or not to continue with or return to therapy.

In sum, the literature suggests the following best therapeutic practices for alliance formation in youth psychotherapy:

• Begin therapy with a minimally structured client-centered approach with an emphasis on the youth's idiographic experiences and expressed emotions.

• Be flexible in the application of a manualized tteatment plan. • Continually monitor and evaluate alliance over the course of therapy. • Be aware of and consider the conflicting interests (and therefore alliances) of both youth

and parent. • At all times, avoid pushing youth to talk about troubling experiences, and attempting to find

common ground.

CONCLUSIONS

After this tour through the literature on therapeutic alliance in youth p.sychotherapy, we can view the questions posed at the beginning of this exploration with new insight. What I termed the intuitive distinction between interventions in medicine (i.e., done to the client) and those in psychotherapy (i.e., done with the client) have certainly been vindicated by the psychological lit- erature, despite several thorny methodological problems in definition and measurement. Though the optimal conditions of its formation and maintenance are not yet known, and whether it affects treatment outcome directly or via one or more intervening mediating variables is unclear, we can say with confidence that the relationship shared by youth and therapist does indeed have an impact on the outcome of that treatment process. As for whether or not the clinician might ever be written out of the therapeutic picture completely, it would appear that while remotely feasi- ble in theory, the same is not true in practice. The few studies that have examined applications of technology in youth psychotherapy still make reference to a central human therapeutic figure

242 CLARK

around which the utilization of technology in therapy is organized, related to, and all but wholly dependent on. Novel technological methods of reducing anxieties, promoting relaxation, and cog- nitive restructuring have been proposed, however the key insight here is that these technologies still require a user that must skillfully apply these mechanisms in a way that acknowledges the uniquely human ability to engage the youth to persist in therapy, continue to work on problems, and willfully comply the tasks of therapy. In essence, these inroads of technology into the realm of psychotherapy might best be viewed as opportunities to enhance the therapeutic relationship if utilized well, rather than threats that seek to replace the therapeutic relationship, or the therapist altogether. Norcross and Wampold (2011, p. 429) conclude with conviction:

. . . psychotherapy is at root a human relationship. Even when "delivered" via distance or on a com- puter, psychotherapy is an irreducibly human encounter. Both parties bring themselves - their origins, cultures, personalities, psychopathology, expectations, biases, defenses, and strengths - to the human relationship. Some will judge that relationship to be a precondition of change and others a process of change, but all agree that it is a relational enterprise.

REFERENCES

Alexander, L. B., & Luborsky, L, (1986). The Penn Helping Alliance Scales. In L. S. Greenberg & W. M, Pinsof (Eds,), The psychotherapeutic process: A research handbook (pp. 325-366), New York, NY: Guilford Press.

Allen, J., Newsom, G., Gabbard, G., & Coyne, L. (1984). Scales to assess the alliance from a psychoanalytic perspective. Bulletin of the Menninger Clinic, 48, 383-400.

Barrett-Lennard, G. T. (1986). The relationship inventory now. In L, S, Greenberg & W. M, Pinsof (Eds.), The psychotherapeutic process: A research handbook (pp, 439-476). New Yoi"k, NY: Guilford Press,

Bordin, E. S. (1979). The generalizability of the psychoanalytic concept of the working alliance. Psychotherapy: Theory, Research & Practice, 160), 252-260. doi: IO.IO37/hOO85885

Chu, B. C , & Kendall, P. C, (2009). Therapist responsiveness to child engagement: Flexibility within manual-based CBT for anxious youth. Journal ofCiinical Psychology, 65(7), 736-754. doi: 10, lOO2/jcîp.2O582

Chu, B., Choudhury, M., Shortt, A., Pincus, D., Creed, T., & Kendall, P. (2004). Alliance, technology, and outcome in the treatment of anxious youth. Cognitive and Behavioral Practice, Jl(\), 44-55. doi: 10. l0l6/S1077-7229(04)80006-3

Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Creed, T. A,, & Kendall, P, C. (2005). Therapist alliance-building behavior within a cognitive-behavioral treatment for

anxiety in youth. Journal of Consulting and Clinical Psychology, 75(3), 49S-505. doi: 10.1037/0022-006X.73.3.498 DiGiuseppe, R., Linscott, J., & Jilton, R. (1996). Developing the therapeutic alliance in child-adolescent psychotherapy.

Applied and Preventive Psychology, 5(2), 85-100. doi: 10.l016/S0962-1849(96)80002-3 Elvins, R., & Green, J, (2008). The conceptualization and measurement of therapeutic alliance: An empirical review.

Clinical Psychology Review, 28(1), 1167-1187. doi: IO.I016/j.cpr,2008.04,002 Freud, S., Strachey, J., Freud, A., Rothgeb, C, L,, & Richards, A, (1953). Vie standard edition of the complete

psychological works of Sigmund Freud. London, England: Hogarth Press. Green, J. (2006), Annotation: The therapeutic alliance— a significant but neglected variable in child mental health treat-

ment studies. Journal of CiiildPsycliology atid Psychiatry, 47(5), 4 2 5 ^ 3 5 . doi: 10.1111/j. 1469-7610,2005.01516.x Hartley, D., & Strupp, H. (1983), Therapeutic alliance: A contribution to outcome in brief psychotherapy. In J, Masling

(Ed.), Empirical studies of psychoanalytic theory (pp, 1-37). Hillsdale, NJ: Erlbaum, Hawley, K, M., & Weisz, J, R. (2005). Youth versus parent working alliance in usual clinical care: Distinctive associa-

tions with retention, satisfaction, and treatment outcome. Journal ofCiinical Child & Adolescent Psychology, 34(\), 117-128. doi: 10.t207/sl5374424jccp340l_ll

Horvath, A. O., & Greentjerg, L. S. (1989). Development and validation of the Working Alliance Inventory. Journal of Counseling Psychology, 36(2), 223-233. doi: 10.1037//0022-0167,36.2,223

Hougaard, E, (1994). The therapeutic alliance: A conceptual analysis. Scandinavian Journal of Psychology, 35(\), 67-85. doi: 10.1111/j. 1467-9450.1994.tb00934.x

THERAPEUTIC ALLIANCE IN YOUTH PSYCHOTHERAPY 2 4 3

Jain, S. (2011). Treating posttraumatic stress disorder via the Internet: Does therapeutic alliance matter? JAMA: The Journal of the American Medical A.<isociation, 306(5), 543-544. doi: 10.1001 /jama.2011.1097

Kazdin, A. E. (2006). Arbitrary metrics: Implications for identifying evidence-based treatments. American Psychologist, 0/(1), 42-49. doi: 10.1037/0003-066X.61.1.42

Kazdin, A. E., & Nock, M. K. (2003). Delineating mechanisms of change in child and adolescent therapy: method- ological issues and research recommendations. Journal of Child Psychology and Psychiatry, 44(&), 1116-1129. doi: 10.1111/1469-7610.00195

Kazdin, A. E., & Whitley, M. K. (2006). Pretreattnent social relations, therapeutic alliance, and improvements in par- enting practices in paient management training. Journal of Consulting and Ciinical Psychology, 74(2), 346-355. doi: 10.1037/0022-006X.74.2.346

Kazdin, A. E., Siegel, T. C , & Bass, D. (1990). Drawing on clinical practice to inform research on child and adoles- cent psychotherapy: Survey of practitioners. Professional Psychology: Research and Practice, 2/(3), 189-198. doi: 10.1037//0735-7028.21.3.189

Kivlighan, D. J., Patton, M. J., & Foote, D. (1998). Moderating effects of client attachment on the counselor experience- working alliance relationship. Jottntal of Counseling Psychology, 45(3), 274-278. doi: 10.1037//0022-0167.45.3.274

Kroll, L., & Green, J. (1997). The therapeutic alliance in child inpatient treatment: Development and initial valida- tion of a family engagement questionnaire. Clinical Child P.sychology and P.iychiatry, 2(3), 4 3 1 ^ 4 7 . doi: 10.1177/ 1359104597023009

Marmar, C. R., Weiss, D. S., 8L Gaston, L. (1989). Toward the validation of the California Therapeutic Alliance Rating System. Psychological A.'ssessment, /(I), 46-52. doi: I O.I 037/1040-3590.1.1.46

Maitin, D. J., Garske, J. P., & Davis, M. K. (2000). Relation of ihe therapeutic alliance with outcome and other variables: A meta-analytic review. Journal of Consulting atid Clinical Psychology, 68, 438-450.

McLeod, B. D. (2011). Relation of the alliance with outcomes in youth psychotherapy: A meta-analysis. Clinical Psychology Review, 31, 603-616. doi: 10. IOI6/j.cpr.2OI 1.02.001

McLeod, B. D., & Weisz, J. R. (2005). The Therapy Process Observational Coding System-Alliance Scale: Measure characteristics and prediction of outcome in usual clinical practice. Journal of Consultitig and Clinical Psychology, 73(2), 323-333. doi: 10.I037/0022-006X.73.2.323

Norcross, J. C. (2002). Psychotherapy relatiotiships that work: therapist contributions and responsiveness to patietits. New York, NY: Oxford University Press.

Norcross, J. C. (2011). Psychotherapy relationships that wotk: Evidetice-based responsivettess. New York, NY: Oxford University Press.

Norcross, J. C , & Wampold, B. E. (2011). Evidence-based therapy relationships: Research conclusions and clinical practices. In J. C. Norcross (Ed.), Psychotherapy relationships that work: Evidence-based responsiveness. New York, NY: Oxford University Press, doi: 10.1093/acprof: oso/9780199737208.003.0021

Rogers, C. R. (1957). The necessary and sufficient conditions of therapeutic personality change. Journal of Consttlting P.?ychology, 2/(2), 95-103. doi: IO.IO37/hOO45357

Safran, J. D., & Muran, J. C. (2000). Negotiating the therapeutic ailiartce: A relationai treaitttent guide. New York, NY: Guilford Press.

Shirk, S. R., & Karver, M. (2003). Prediction of treatment outcome from relationship variables in child and adolescent therapy: A meta-analytic review. Journal of Cottsuiting and Clinical Psychology, 7/(3), 452-464. doi: 10.1037/0022- 006X.71.3.452

Shirk, S. R., & Karver, M. S. (2011). Alliance in child and adolescent psychotherapy. In J. C. Norcross (Ed.), Psychotherapy relationships that work: Evidetwe-based responsivettess. New York, NY: Oxford University Press, doi: 10.1093/acprof: oso/9780199737208.003.0003

Shirk, S. R., & Saiz, C. C. (1992). Clinical, empirical, and developmental perspectives on the therapeutic relationship in child psychotherapy. Developmettt attd Psychopathology, 4(4), 713. doi: IO.I017/S0954579400004946

Weersing, V. R., & Weisz, J. R. (2002). Mechanisms of action in youth psychotherapy. Journal of Child Psychology and P.<iychiatry, 43, 3 - 2 9 .

Copyright of Journal of Infant, Child & Adolescent Psychotherapy is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.