21 - Psychotherapy Relationships

Summary

The therapeutic relationship constitutes the heart and soul of the enterprise. Second only to the patient’s contribution, the relationship is the most powerful predictor of, and contributor to, outcome. Its effectiveness cuts across theoretical orientations (transtheoretical) and largely across client problems (transdiagnostic). This chapter reviews evidence-based psychotherapy relationships, primarily with adults in individual treatment. We begin by defining terms and summarizing the meta-analytic evidence on effective relationship behaviors or components (what works). That is followed by a summary of ineffective or discredited relationship behaviors (what does not work). We advance therapeutic and training practices based on this research evidence. The chapter finishes with multiple caveats, concluding thoughts, and useful resources.

The patient–practitioner relationship constitutes the heart and soul of psychotherapy, healing in and of itself. Second only to the client’s contribution, the therapy relationship is the most powerful predictor of, and contributor to, successful outcomes. Even when offered as a manualized intervention and delivered via electronic means, therapy is invariably rooted in and dependent on that complex connection between the client and therapist. As such, it warrants substantial attention in any scientific compilation of evidence-based (or science-based) list of psychotherapy components.

The effectiveness of the multiple relationship factors cuts across theoretical orientations (transtheoretical) and largely across client problems (transdiagnostic). The research evidence on the relationship does not favor any single orientation; the probability of a positive client–clinician relationship or a failure in that relationship is not any more characteristic of one psychotherapy system than another. The relationship elements or components considered in this chapter have all been shown, in dozens of individual studies and in rigorous meta-analyses, to associate, predict, and contribute to success. Failure to provide these elements also predicts and contributes to poor treatment outcomes, however measured (e.g., dropout, deterioration).

In this chapter, we review evidence-based psychotherapy relationships, primarily with adults in individual treatment. We begin by defining our terms and diving into effective relationship behaviors or components (what works). That is followed by a few words on ineffective or discredited relationship behaviors (what does not work). We then advance therapeutic and training practices based on this research evidence. The chapter finishes with multiple caveats, concluding thoughts, and useful resources.

Definitions

Many spirited and unproductive debates on psychotherapy fail to operationally define their terms. Antagonists wind up speaking past one another, literally not on the same page. Cases in point are the psychotherapy relationship and evidence-based practice. We will not commit the errors of undefined terms and unidentified contexts in this chapter.

An operational definition of the therapeutic relationship is the feelings and attitudes that therapist and client have toward one another, and the manner in which these are expressed (Gelso & Carter, 1985, 1994). While this definition is quite general, it is mercifully concise, theoretically neutral, and sufficiently precise.

The therapeutic alliance represents a part of the relationship, but only a part. In fact, a pernicious error in the psychotherapy literature equates the totality of the relationship with the therapeutic alliance. In part, this mistake occurs inadvertently because the alliance is the most frequently measured and researched relationship factor in the psychotherapy literature (Horvath et al., 2016). In part, too, this mistake probably occurs intentionally to misrepresent and diminish the cumulative power of the relationship (Norcross & Karpiak, 2023). Ironically, the alliance’s association with psychotherapy success is not even the largest of the relationship factors, as discussed shortly. Conflating the entirety of the therapy relationship with only the alliance weakens the power of the therapeutic relationship empirically and clinically.

The short past of evidence-based practice (EBP) in behavioral/mental health traces back to the 1980s, originally in Great Britain and then gathering steam in Canada, the United States, and now around the globe (Norcross et al., 2017). The early stirrings of the movement trace back to the United Kingdom and Archie Cochrane’s (1979) article calling on medicine to assemble critical summaries of science-based treatments that had proven effective according to randomized clinical trials. Cochrane and others contrasted EBP with expert- or authority-based practice, the latter lacking in solid research support and typically resulting in less effective health care.

A consensual and concrete definition of EBP has emerged from the research literature and professional organizations. Adapting a definition from Sackett and colleagues, the Institute of Medicine (2001, p. 147) defined evidence-based medicine (EBM) as “the integration of best research evidence with clinical expertise and patient values.” The American Psychological Association (APA) Task Force on Evidence-Based Practice (2006, p. 273), beginning with this foundation and expanding it to mental health, defined EBP as “the integration of the best available research with clinical expertise in the context of patient characteristics, culture, and preferences.” We use the latter as our operational definition throughout.

Several core features of EBPs become manifest in this definition. First and foremost, EBPs rest on three pillars: available research; clinician expertise; and patient characteristics, culture, and preferences. By definition, the wholesale imposition of research without attending to the clinician or patient is not EBP; conversely, the indiscriminate disregard of available research is not EBP. Second, the definition requires integrating these three evidentiary sources. The integration flows seamlessly and uncontested when the three evidentiary sources agree; it becomes complicated and contested when the three sources disagree. Third, not all three pillars stand equal: Research assumes priority in EBP. Clinicians begin with research and then integrate with their expertise and patients. Fourth and final, compared to EBM, the patient assumes a more active, prominent position in EBPs in behavioral health and addictions. “Patient values” in EBM rise to the status of “patient characteristics, culture, and preferences” in behavioral health EBPs.

Evidence-Based Therapy Relationships

The centrality of the therapy relationship has been highlighted since the origins of modern psychotherapy. Sigmund Freud described the operation of transference and countertransference, and psychoanalytic scholars developed a rich literature on the relationship. Among the foundational constructs are the establishment of a positive working or therapeutic alliance (Bordin, 1979; Luborsky, 1976) and the management of negative countertransference (Singer & Luborsky, 1977). The real relationship, characterized by realism and genuineness, was emphasized later by psychodynamic therapists (Greenson, 1967; Gelso et al., 2019). All three of these elements have subsequently proven effective in psychotherapy according to the research evidence (Table 21.1; Norcross & Lambert, 2019).

Table 21.1 Summary of meta-analytic associations between relationship components and distal psychotherapy outcomes (adapted from Norcross & Lambert, 2019; © Norcross & Lambert)

Relationship element	# of studies	# of patients	Effect size		Consensus on evidentiary strength
	(k)	(N)	r	d or g
Alliance in individual adult psychotherapy	306	30,000+	0.28	0.57	Demonstrably effective
Alliance in child and adolescent therapy	43	3,447	0.20	0.40	Demonstrably effective
Alliances in couple and family therapy	40	4,113	0.30	0.62	Demonstrably effective
Collaboration	53	5,286	0.29	0.61	Demonstrably effective
Goal consensus	54	7,278	0.24	0.49	Demonstrably effective
Cohesion in group therapy	55	6,055	0.26	0.56	Demonstrably effective
Empathy	82	6,138	0.28	0.58	Demonstrably effective
Positive regard and affirmation	64	3,528	0.28		Demonstrably effective
Collecting and delivering client feedback	24	10,921		0.14–0.49	Demonstrably effective
Congruence/genuineness	21	1,192	0.23	0.46	Probably effective
Real relationship	17	1,502	0.37	0.80	Probably effective
Emotional expression	42	925	0.40	0.85	Probably effective
Cultivating positive expectation	81	12,722	0.18	0.36	Probably effective
Promoting treatment credibility	24	1,504	0.12	0.24	Probably effective
Managing countertransference	9	392*	0.39	0.84	Probably effective
Repairing alliance ruptures	11	1,318	0.30	0.62	Probably effective

* Refers to the number of psychotherapists, not patients.

Note. In the behavioral sciences, an effect size (d or g) of 0.20 is generally considered a small effect, 0.50 a medium effect, and 0.80 a large effect (Cohen, 2013).

Carl Rogers’s legacy includes the three relational facilitative conditions for therapeutic change: empathy, positive regard, and genuineness/congruence (Rogers, 1957). Two of Rogers’s facilitative conditions are demonstrably effective and the third probably effective based on the meta-analyses (Table 21.1; Norcross & Lambert, 2019). A dedicated scientist, Rogers established psychotherapy process–outcome research, modeling how to scientifically examine the association between specific therapist behaviors and client responses to understand not just whether psychotherapy works but how it works.

Early cognitive-behavioral formulations of the therapy relationship emphasized it as a precondition of change, the soil that enables treatment methods to work, as opposed to a healing process in and of itself. Therapist and patient were to work together, akin to a student–teacher relationship, establishing rapport, cultivating positive expectations, and jointly determining treatment goals (Beck et al., 1979). Recent meta-analyses support the salubrious link between collaboration, goal consensus, and positive expectations and successful psychotherapy (Table 21.1; Norcross & Lambert, 2019).

Subsequent generations of practice and research on the therapy relationship have emphasized transtheoretical conceptualizations (as opposed to theory specific) and research evidence (as opposed to clinical lore or authority). In this context, three American Psychological Association (APA) Interdivisional Task Forces were convened to identify, compile, and disseminate therapy relationships that were evidence based (Norcross, 2002; Norcross, 2011; Norcross & Lambert, 2019). The Task Force aimed, in addition, to heal some of the damage of the culture wars in psychotherapy that unproductively pit treatment methods against therapeutic relationships.

In its third and most recent iteration, Psychotherapy Relationships that Work (Norcross & Lambert, 2019; Norcross & Wampold, 2019) contains two volumes of 30 meta-analyses of therapist and patient contributions to therapy effectiveness. The strength of the scientific evidence is undeniable, and the field has (mostly) matured past polarized positions (relationship vs. method) to consider how all elements optimally operate and interact. Table 21.1 presents, from those volumes, a summary of the meta-analytic associations between relationship elements and distal (end of treatment) psychotherapy outcomes. In the following section, we concentrate on those elements that have proven effective in individual psychotherapy with adults.

What Works

Alliance

The term alliance is not easily differentiated from several other relational concepts; in the literature, the words “working,” “helping,” or “therapeutic” often appear in conjunction with it (Fluckiger et al., 2019). An early tripartite definition by Bordin (1979) emphasized (1) a warm emotional bond, (2) agreement on respective tasks, and (3) consensus on treatment goals. A more recent definition includes mutual collaboration between client and therapist on goals and tasks of psychotherapy, along with the therapeutic bond between the dyad (Del Re et al., 2021). Alliance is interpersonal – therapist and client both contribute to it – but the ability to form an alliance with an array of clients is a therapist characteristic that can be learned (Ackerman & Hilsenroth, 2003; Muran & Eubanks, 2020).

A meta-analysis of more than 30,000 clients found a moderate, but extremely robust, association between the alliance and outcome in adult individual psychotherapy (d = 0.57; Fluckiger et al., 2019). The alliance relates to, predicts, and contributes to psychotherapy success. To a lesser extent, success in therapy also strengthens the relationship.

Likewise, with a few wrinkles, the alliance in youth psychotherapy works. Across 43 studies of child and adolescent therapy (3,447 clients and parents), there was a moderate effect size between alliance and treatment outcome (d = 0.40; Karver et al., 2019). Importantly, the strength of the alliance–outcome relation did not vary with the type of treatment. Further, the effect size, or clinical impact, of dual alliances – therapist with youth, therapist with parent – was identical. Both the therapist–youth and therapist–caregiver alliance matter mightily.

The average effect size for the alliance in couples/family therapy is in the same range (0.62), based on 40 studies (Friedlander et al., 2019), as is the impact of the alliance in psychopharmacological treatment, based on 8 studies (Totura et al., 2018).

The alliance emerges in the ongoing relationship between therapist and client, and accurate measurement of the alliance at any point during the course of therapy probably requires collecting information from both (or all) participants. Most studies indicate that the client rating is the better predictor of treatment outcome than therapist or observer ratings, but all information proves valuable. Alliance measures completed by the client can be used by a therapist or supervisor to track the development of this vital competency and predictor of therapy outcome.

Goal Consensus and Collaboration

These two relational components are present across theoretical orientations and are sometimes considered part of the therapeutic alliance. Indeed, both are commonly assessed for research purposes via measures of the alliance, completed separately by clinician and client. Goal consensus refers to the agreement between the therapist and the client about the targets of their work together and how to achieve them. A large body of research documents the vital role this factor plays in treatment outcome (d = 0.49; Tryon et al., 2019). Collaboration is the active mutual engagement of the therapist and client around the work of therapy. Research also shows that collaboration is substantially associated with treatment outcome, with an effect size (d) of 0.61 (Tryon et al., 2019).

Empathy

The term empathy is widely used in the common vernacular to refer to a strong emotional response to the situation of another, and in popular use is often conflated with sympathy or compassion. For clarity, we use Rogers’s (1980, p. 85) definition as “the therapist’s sensitive ability and willingness to understand the client’s thoughts, feelings and struggles from the client’s point of view.” Empathy can be assessed, according to Barrett-Lennard (1981), as “(a) the therapist’s empathic resonance with the client, (b) the observer’s perception of the therapist’s expressed empathy, and (c) the client’s experience of received therapist empathy” (Elliott et al., 2019, p. 248).

Empathic responding is one of the strongest and best-supported contributors to outcome (Elliott et al., 2019). Starting with the groundbreaking research of Carl Rogers, decades of evidence now attest to its value, with meta-analytic effect sizes ranging between moderate to large (d = 0.58), from 82 high-quality studies. Even better, the skills and basic stance of empathic understanding can be practiced in everyday relationships outside of a real or simulated therapy situation (see Miller, 2018), as long as others in these relationships are willing and able to provide feedback.

Affirmation and Validation

Therapist positive regard for the client (and expression of that regard, including through affirmation) is another relationship factor originally investigated by Carl Rogers. As with empathy, positive regard has been the subject of much research, with its effect on psychotherapy outcome falling in the moderate range (d = 0.57), based on 64 studies of quality appropriate for inclusion in a meta-analysis (Farber et al., 2019).

Like empathy, the terms positive regard and affirmation are easily misunderstood – mistaken for simple compliments, shallow praise, or other concrete tactics (e.g., preceding requests for compliance). In fact, positive regard is the therapist’s genuine nonpossessive liking and expressed appreciation for the client as a unique person. This strengthens the client’s sense of agency and self. To contribute to outcome, this regard must be made evident to the client through words and nonverbals. Therapists can express on a regular basis that they value, care about, and believe in the client, ideally over the course of treatment. However, it does not need to be (and probably could not be) experienced by the therapist at every moment across treatment with any given client (Farber et al., 2019).

Congruence/Genuineness and Real Relationship

This pair of relational factors includes the last of the three Rogerian core conditions (empathy, unconditional positive regard, and congruence/genuineness) and a concept from the psychodynamic literature: the real relationship. Both have accumulated sufficient evidence to be classified as effective.

Congruence/genuineness has both intrapersonal and interpersonal features, meaning it is both a personal characteristic of the therapist as well as a quality of the therapeutic interaction. When congruent, therapists’ actions and behaviors not only fit their words but also who they are as a person – their values and identity – exuding groundedness, thoughtfulness, and genuineness. In short, they are real, not phony, distracted, or playing a role. Studies of this relational component show a moderate association with treatment outcome, making it a reliable contributor to therapeutic success (d = 0.46; Kolden et al., 2019).

The real relationship is composed of both genuineness and realism. It’s “the personal relationship between therapist and patient marked by the extent to which each is genuine with the other and perceives/experiences the other in ways that befit the other” (Gelso, 2014, p. 119). In contrast to the alliance, it refers to a subset of therapist–client interactions not directly focused on the tasks. These interactions are taken at face value in the here and now. A meta-analysis based on 17 studies and 1,502 patients revealed a large effect between the real relationship and client success (d = 0.80; Gelso et al., 2019).

Emotional Expression

Although emotion is obviously core to psychotherapy, organized research on the subject is quite recent. That evidence shows that the facilitation, experience, and expression of client emotion in session are strongly correlated with treatment outcome (d = 0.85; Peluso & Freund, 2019). Contributions to this research base come from a wider range of theoretical orientations than do the components reviewed thus far. Because of this variability, the definition of emotional expression is less precise and has failed to achieve consensus in the field. One essential and simple clarification: what is not being referenced here is the “expressed emotion” from family process and relapse prevention research with serious mental illnesses, where reducing expressed emotion is the goal.

Of all the relational elements, emotional expression is possibly the one that can most easily go astray in the absence of a clear plan. Treatment model and case formulation help determine which emotions to address, where a particular emotional expression fits into the therapeutic endeavor, and what to do with it – in other words, how to attend to and make therapeutic use of emotion versus allowing the session to deteriorate into an unfocused rant, wallow, or self-attack.

Repair of Alliance Ruptures

Ruptures are problems or strains in the collaborative relationship between client and therapist related to treatment goals, agreement on the tasks of therapy, or the emotional bond (Eubanks et al., 2019). Two main types occur in session: (1) withdrawal, in which the client moves away from the therapist and the work; and (2) confrontation, in which the client moves against the therapist by expressing anger or dissatisfaction. Although the term rupture may connote a dramatic breakdown, many studies point to subtle tensions and minor misalignments as markers.

Therapist efforts to repair alliance ruptures can be overt or indirect. Either way, research shows attending to them improves treatment outcomes. A meta-analysis of 11 studies, involving 1,318 clients, revealed that repair of alliance ruptures in individual therapy is moderately to strongly associated with outcome (d = 0.62). That is, addressing ruptures works; ignoring them does not.

Repairing ruptures proves valuable for all psychotherapists, but especially for therapists with less experience and training in negotiating the therapeutic alliance (Eubanks et al., 2019). Moderator analyses revealed that rupture resolutions training is more effective for cognitive-behavioral therapists, many of whom have not received explicit training in processing relationship dynamics with their clients.

Collection of Client Feedback

In collecting feedback from patients – or the more recent term routine outcome monitoring (ROM) – psychotherapists inquire directly about the patient’s progress on a regular basis, compare those data to benchmarks or norms, address the progress (or lack thereof) directly in session, and, in some cases, offer clinical support tools to identify obstacles and adapt future sessions. A dozen or so ROM systems are now available, but most of the controlled research has employed the Outcome Questionnaire and the briefer PCOMS (Partners in Change Outcome Monitoring System) feedback systems. A meta-analysis of 24 controlled trials (on more than 10,000 patients) conducted on those systems (Lambert et al., 2019; Table 21.1) found that feedback or ROM produced variable but salutatory effects (between 0.14 and 0.49) on distal treatment outcomes. A subsequent and larger meta-analysis (de Jong et al., 2021) on 58 studies, encompassing more than 20,000 patients, reported an overall d of 0.15.

Collecting feedback or conducting ROM is thus slightly effective for all patients but more effective for patients not on track in treatment or at risk of an unsuccessful outcome. ROM, in fact, reduces the risk of patient dropout by 20–25% (de Jong et al., 2021; Lambert et al., 2019). All these are additive effects to conducting standard therapy.

Overall

The scientific conclusion emerges that there is a robust, consistent association between these core relational elements and client improvement. On average, the correlation (r) is about 0.25–0.30. That translates to an effect size (d) of about 0.55 and indicates that clients receiving psychotherapy characterized by high degrees of empathy, regard, and the like will experience a decided advantage over clients that receive (or perceive) relatively lower degrees of those relationship attributes.

Although this estimate of treatment effects may seem rather modest, bear in mind the large number of complex variables that contribute to treatment outcomes, especially clients’ contributions and life events that exist before and during the therapeutic encounter (Lambert et al., 1992). Also bear in mind that the average effect size (d) between psychotherapy and no psychotherapy hovers about 0.85; any single relational behavior in Table 21.1 comes in at an impressive 0.55.

It would probably prove advantageous to both practice and science to sum the individual effect sizes in Table 21.1 to arrive at a total of relationship contribution to treatment outcome, but reality is not so accommodating. Neither the research studies nor the relationship elements contained in the meta-analyses are independent; hence, the amount of variance accounted for by each element cannot be simply added to estimate the overall contribution. For example, the correlations between empathy and therapeutic alliance are as high as 0.70 (Watson & Geller, 2005). The intercorrelations between the person-centered conditions are also high: in an early research review on client-centered conditions empathy correlated 0.53 with positive regard, 0.62 with congruence, and 0.28 with unconditionally (Gurman, 1977). Unfortunately, the degree of overlap between all the measures (and therefore relationship elements) is not available but is bound to be substantial.

Despite the overlap in relational elements, the best scientific estimates of relationship effects are reliable and robust. The therapeutic relationship contributes as much, and probably more, to client outcomes than the particular treatment method. The effect sizes for relationship behaviors (Table 21.1) of 0.39–0.72 are higher than the effect sizes of 0–0.20 attributable to different treatment methods found in bona fide comparisons (Wampold & Imel, 2015). Although we deplore the mindless dichotomy between relationship and method in psychotherapy, we also need to publicly proclaim what decades of research has discovered: The relationship can heal.

What Does Not Work

Translational research is both prescriptive and proscriptive; it tells us what works and what does not (Norcross et al., 2017). Here, we highlight those practitioner relational behaviors that are ineffective, perhaps even hurtful, in psychotherapy (Karpiak & Norcross, 2022; Norcross & Karpiak, 2023).

Of course, we could simply reverse the effective behaviors identified in the meta-analyses (Table 21.1) to identify ineffective qualities of the therapeutic relationship. What does not work, for example, are poor alliances, paucity of collaboration, and inadequate efforts at empathy. The ineffective practitioner will not seek nor be receptive to client feedback on progress and relationship, will ignore alliance ruptures, and will not promote their patients’ emotional expression. “One doesn’t have to operate with great malice to do great harm,” warned Charles Blow. “The absence of empathy and understanding are sufficient.”

Another means of identifying ineffective qualities of the relationship is to scour the research literature for clinician behaviors frequently associated with negative outcomes and premature discontinuation (e.g., Hardy et al., 2019; Swift & Greenberg, 2012). Here are several relational behaviors that therapists should avoid according to the Task Forces review of that research (Norcross & Lambert, 2019):

1. ♦ Confrontations. Controlled research trials, particularly in the addictions field, consistently find that a confrontational style proves ineffective. In one review (Miller et al., 2003), confrontation was ineffective in all 12 identified trials. And yet it persists. By contrast, expressing empathy, rolling with resistance, developing discrepancy, and supporting self-efficacy characteristic of motivational interviewing have demonstrated large effects in a small number of sessions (Lundahl et al., 2013).

2. ♦ Negative processes. Client reports and research studies converge in warning therapists to avoid comments or behaviors that are experienced by clients as hostile, pejorative, critical, rejecting, or blaming (Binder & Strupp, 1997). Therapists who attack a client’s dysfunctional thoughts or relational patterns need, repeatedly, to distinguish between attacking the person versus their behavior. When negative processes ensue, then repairing alliance ruptures is amongst the most easily applied skills and strongest relationship behaviors documented in psychotherapy (Eubanks et al., 2019; Table 21.1).

3. ♦ Assumptions. Psychotherapists who assume or intuit their client’s perceptions of relationship satisfaction and treatment success frequently misjudge these aspects. By contrast, therapists who formally measure and respectfully inquire about their client’s perceptions, via feedback or ROM, frequently enhance the alliance and prevent premature termination (Lambert et al., 2019).

4. ♦ Therapist-centricity. A recurrent lesson from process–outcome research and the associated meta-analyses is that the client’s perspective on the therapy relationship best predicts outcome. Psychotherapy that relies on the therapist’s observational perspective alone, while valuable, does not predict outcome as accurately. Therefore, privileging and monitoring the patient’s experience of the relationship prove central.

5. ♦ Rigidity. By inflexibly and excessively structuring treatment, psychotherapists risk empathic failures and inattentiveness to clients’ experiences. Such a therapist is likely to overlook a breach in the relationship and mistakenly assume they have not contributed to that breach. Dogmatic reliance on particular relational or therapy methods, incompatible with the client, imperils treatment (Ackerman & Hilsenroth, 2003).

6. ♦ Cultural arrogance. Arrogant impositions of therapists’ cultural beliefs in terms of gender, race/ethnicity, sexual orientation, and other intersecting dimensions of identity are culturally insensitive and demonstrably less effective (Soto et al., 2019). By contrast, therapists’ expressing cultural humility, offering adapted treatments, and emphasizing cultural responsiveness markedly improve client engagement, retention, and eventual treatment outcome.

We can optimize therapy relationships by simultaneously using what works and studiously avoiding what does not work.

Cautions and Caveats

In valuing the therapeutic relationship, it becomes deceptively easy to overplay the impact of that relationship in treatment (Norcross & Lambert, 2014). “It’s all the relationship” is a frequent (and inaccurate) refrain among many trainees and some practitioners. The Task Force repeatedly urged restraint and balance in disseminating its meta-analytic findings, pointedly concluding that “The therapy relationship acts in concert with treatment methods, patient characteristics, and other practitioner qualities in determining effectiveness; a comprehensive understanding of effective (and ineffective) psychotherapy will consider all of these determinants and how they work together to produce benefit” (Norcross & Lambert, 2019, p. 632).

Historically, the polarized psychotherapy community rarely found the middle ground in valuing the therapy relationship in context; contemporarily, most psychotherapists avoid the unproductive schism of relationship versus method and frame the therapeutic relationship in comprehensive, scientific contexts. Multiple lines of evidence suggest that humans (including patients) hold two major dimensions of social perception, often called warmth and competence (Eisenbruch & Krasnow, 2022). The relational warmth of the therapist is typically prioritized by patients over therapist competence (e.g., Swift & Callahan, 2010). Fortunately, the choice is not binary or exclusive. The most effective therapists regularly manifest both relational warmth and technical competence (Castonguay & Hill, 2017; Seewald & Rief, 2022).

The research on these relational elements features additional cautions. The meta-analytic results in Table 21.1 probably underestimate the true effect of the relationship due to the responsiveness problem (Kramer & Stiles, 2015; Stiles et al., 1998). It is a problem for researchers but a boon to practitioners, who flexibly adjust the amount and timing of relational behaviors in psychotherapy to fit the unique individual and singular context. Effective psychotherapists responsively provide varying levels of relationship elements in different cases and, within the same case, at different moments. This responsiveness tends to confound attempts to find naturalistically observed linear relations of outcome with therapist behaviors (e.g., empathy, ROM). As a consequence, the reported statistical association between therapy relationship and outcome cannot always be trusted and tends to be lower than it actually is. By being clinically attuned and flexible, psychotherapists ironically make it more difficult in research studies to discern what works (Norcross & Lambert, 2019).

Nor has the research generated a definitive list of what works in the therapy relationship. We have neither completed the search nor exhausted the relationship behaviors associated with therapy success. Insufficient controlled research exists to draw conclusions at this juncture on many other relationship behaviors advocated by practitioners.

As the evidence base of therapist relationship behaviors develop, we will know more about their effectiveness for particular circumstances and conditions. A case in point is the meta-analysis on collecting client feedback or conducting ROM. The evidence is clear that adding formal feedback/ROM helps clinicians effectively treat patients at risk for deterioration and that adding some form of clinical support tools to assist clinicians boosts its effectiveness. But for most of these relational elements, we do not yet know for whom and when they prove effective.

The strength of the therapy relationship also depends in some instances on the client’s principal disorder. The moderator analyses occasionally find some relationship elements less efficacious with some disorders, usually substance abuse, severe anxiety, and eating disorders. Most moderator analyses usually find the relationship equally effective across disorders, but that conclusion may be due to the relatively small number of studies for any single disorder and the resulting low statistical power to find actual differences. And, of course, it gets more complicated as patients typically present with multiple, comorbid disorders.

Finally, we emphasize that, with a couple of exceptions (collecting feedback, repairing alliance ruptures), the meta-analyses reported the association and prediction of the relationship element to psychotherapy outcome. These were overwhelmingly correlational designs, showing that more of, say, collaboration, emotional expression, and positive regard were associated with improved patient success. Of the relationship behaviors reviewed in this chapter, only two (client feedback/ROM, alliance ruptures) have addressed disaggregation by means of RCTs and only one (alliance in individual therapy; Del Re et al., 2012) by other statistical means. And it turns out the evidence is strong that it is the therapist that is important: therapists who generally form stronger alliances generally have better outcomes, but not vice versa (Wampold et al., 2012). It is largely the therapist’s contribution, not the patient’s, that relates to therapy outcome (Baldwin et al., 2007; Wampold & Imel, 2015). Put differently, for most of these relationship elements, we know with certainty that they characterize, positively correlate with, and predict successful psychotherapy. But that does not necessarily mean that they are therapist contributions. Another type of causal linkage is still needed.

Box 21.1 Therapy and Training Practices

Review of the research evidence led the APA Interdivisional Task Force (Norcross & Lambert, 2019) to advance several recommendations for clinical practice and training. Practitioners are encouraged to:

1. ♦ make the creation and cultivation of the therapy relationship a primary aim of treatment. This is especially true for relationship elements found to be demonstrably and probably effective.

2. ♦ assess relational behaviors (e.g., alliance, empathy, cohesion) vis-à-vis cut-off scores on popular clinical measures in ways that lead to more positive outcomes.

3. ♦ assess and responsively attune psychotherapy to clients’ cultural identities (broadly defined).

4. ♦ monitor patients’ satisfaction with the therapy relationship, comfort with responsiveness efforts, and response to treatment. Such monitoring leads to increased opportunities to re-establish collaboration, improve the relationship, modify technical strategies, and investigate factors external to therapy that may be hindering its effects.

5. ♦ use concurrently evidence-based relationships and evidence-based treatments adapted to the whole patient as that is likely to generate the best outcomes in psychotherapy.

In turn, training programs are encouraged to:

1. ♦ provide competency-based training in the demonstrably and probably effective elements of the therapy relationship.

2. ♦ train students in assessing and honoring clients’ cultural heritages, values, and beliefs in ways that enhance the therapeutic relationship and inform treatment adaptations.

3. ♦ develop criteria for assessing the adequacy of training in evidence-based therapy relationships and responsiveness.

Conclusion

How to improve the outcomes of psychological treatments? Follow the scientific evidence; follow what contributes to treatment outcome as reviewed in this chapter. Begin by leveraging the patient’s resources and self-healing capacities; create and cultivate a therapy relationship characterized by these effective elements; avoid use of ineffective and discredited relational behaviors; responsively personalize to the patient’s characteristics, personality, and worldviews. That’s evidence-based therapy relationships.

All treatment, all health care, all methods are embedded within a relational context. Not only is there a deep synergy between a treatment method and a therapeutic relationship, but one does not exist without the other. This point was convincingly made decades ago in Winnicott’s observation that there is no such thing as a baby without a mother. Effective psychotherapy cannot, and does not, exist without a relationship.

The future of mental health services portends the integration of science and service, of the instrumental and the interpersonal, of the technical and the relational in the EBP tradition (Norcross et al., 2016). We can imagine few practices in mental health that can confidently boast that they seamlessly integrate “the best available research with clinical expertise in the context of patient characteristics, culture, and preferences” (APA Task Force, 2006, p. 273) as well as cultivating and customizing the powerful therapy relationship. As Carl Rogers (1980) compellingly demonstrated, there is no inherent tension between a relational approach and a scientific one. Science can and should inform us about what works in psychotherapy – be it relational or otherwise.