Mutuality in Psychotherapy: A Meta-Analysis and Meta-Synthesis

This article presents a systematic review of mutuality in psychotherapy, including meta-analysis of quantitative and metasynthesis of qualitative studies. A search with specified keyword combinations yielded 21 studies, including 10 quantitative studies with 1,071 participants and 11 qualitative studies with 81 participants. Researchers calculated effect sizes, conducted homogeneity tests, and assessed potential variables moderating the relationship between mutuality and therapeutic variables from quantitative studies; they analyzed qualitative studies to identify and synthesize themes related to mutuality in psychotherapy. Meta-analysis showed a large weighted mean effect size with a statistically significant overall relationship between mutuality and therapeutic variables (r = .51, 95% CI [0.37, 0.66], p < .001). The relationship between mutuality and session quality was strongest of the six relationships analyzed (r = .70, 95% CI [0.43, 0.97], p < .001). Qualitative metasynthesis of studies produced six themes: (a) lack of mutuality/strategies for disconnection, (b) cocreated relational process, (c) metacommunication and misunderstanding, (d) therapist congruence/being real, (e) mutual impact and client agency, and (f) asymmetric role power and boundaries. These findings suggest that mutuality is worthy of further research in psychotherapy, particularly in relation to its strong relationship with session quality. Mutualidad en psicoterapia: un metanálisis y meta-síntesis Este artículo presenta una revisión sistemática de la mutualidad en psicoterapia, que incluye metaanálisis de estudios cuantitativos y metasíntesis de estudios cualitativos. Una búsqueda con las combinaciones de palabras clave especificadas produjeron 21 estudios, incluidos 10 estudios cuantitativos con 1,071 participantes y 11 estudios cualitativos con 81 participantes. Investigadores tamaños de efecto calculados, pruebas de homogeneidad realizadas y variables potenciales evaluadas que moderan la relación entre la mutualidad y las variables terapéuticas de los estudios cuantitativos; analizaron estudios cualitativos para identificar y sintetizar temas relacionados con la mutualidad en psicoterapia. El metanálisis mostró un tamaño de efecto medio ponderado grande con una relación general estadísticamente significativa entre las variables de mutualidad y terapéuticas (r = .51, IC 95% [0.37, 0.66], p < .001). La relación entre la mutualidad y la calidad de la sesión fue la más sólida de las seis relaciones analizadas (r = .70, IC del 95% [0.43, 0.97], p < .001). La metasíntesis cualitativa de los estudios produjo seis temas: (a) falta de reciprocidad / estrategias para desconexión, (b) proceso relacional co-creado, (c) metacomunicación y malentendido, (d) congruencia terapeuta / ser real, (e) impacto mutuo y agencia del cliente y (f) poder de rol asimétrico y límites. Estos hallazgos sugieren que la mutualidad es digna de mayor investigación en psicoterapia, particularmente en relación con su fuerte relación con la calidad de la sesión.

Across all the major schools of psychotherapy, the client-therapist relationship has been shown to be an effective contributor to positive change (Norcross, 2011). This is true for a range of therapeutic relationship variables, including empathy (Elliott, Bohart, Watson, & Greenberg, 2011), positive regard and affirmation (Farber & Doolin, 2011), genuineness (Kolden, Klein, Wang, & Austin, 2011), and therapeutic alliance in adults (Hovarth, Del Re, Flückiger, & Symonds, 2011) and in adolescents (Shirk & Karver, 2011). The association between these psychotherapeutic relationship variables and psychotherapy outcome in clients is considered, by and large, to be empirically supported.
Following Rogers' (1957) classic conditions for therapy, research looking at the therapeutic relationship variables has often focused on a conception of the therapeutic relationship based on elements that are seen as "inside" the therapist or the therapist's actions (e.g., empathy, positive regard, affirmation from the therapist for the client). The measurement of these relational elements may be from the client, observer, or therapist perspectives, with research generally supporting a decreasing correlation with therapeutic outcome aligned with this or-der, where the client's perspective on the relationship (highlighted in Rogers' sixth condition of client perception) is a better indicator of success than therapists' own perceptions (Norcross, 2011), particularly when measured early in the course of therapy (Elliott et al., 2011;Hovarth et al., 2011). Rather than use ratings of the therapeutic relationship variables "in" the therapist, other conceptions have looked at variables between the therapist and client (e.g., working alliance). In this case, however, the working alliance has been conceptualized broader than the relational qualities to include technical aspects, which approach the "content" realm, such as goals and tasks, not just bond. Admittedly, goal identification and task agreement between therapist and client often flows from a good bond, but nevertheless the conception of working alliance is not simply a perception of the relationship between the people. In contrast, the idea of mutuality in psychotherapy considers the elements of the relationship that are not in the therapist or client alone but are shared between the client and therapist. Mutuality is an emergent property that is the product of coexperiencing of therapeutic attitudinal qualities in the relationship between client and therapist.
We have stated above that mutuality cannot simply be reduced to the experiencing of individual characteristics of the therapist and client that are inside either person. Mutuality emerges as a distinct dimension of the relationship as a result of the therapist and client coexperiencing specific relational attitudinal qualities. These attitudinal qualities include, but may not be restricted to, mutual empathy, mutual positive regard or a mutual perception of the realness of the other. Hence, mutuality might be associated with the "climate" of the person-to-person relationship. This contrasts with those more specific features of the person-to-person relationship such as the agreement on goals and tasks. Consequently, it is conceivable that there is mutuality emergent from the coexperiencing of empathy, positive regard, and realness while there is also disagreement on the specific goals and the tasks. In this sense, except for the bond dimension, the alliance characterizes one aspect of the emergent relationship but not one that directly leads to the development of mutuality. However, to address this last point substantively would require further empirical study to explore the association between measures of task and goal agreement and mutuality. Having defined what we mean by mutuality, we now turn to considering the issue of attempting to measure it.

The Measurement of Mutuality
Like with the classic conceptualizations of therapeutic relationship variables (e.g., Rogerian conditions, working alliance) whether viewed within a person or between them, mutuality may be measured from different perspectives. Ideally, any emergent relational element between people would include measures from both sides and as concerns mutuality, an integration of those sides to measure the emergent property itself. Murphy and Cramer (2014) explored the mutuality hypothesis in this way, showing that when the relationship variables were being experienced positively from both sides, it will be a better predictor of generic distress than when either the therapist or the client experiences a specific relationship variable from their own perspective.
Other studies that could be argued to have assessed mutuality are those that measured coexperiencing of the therapeutic bond such as Saunders, Howard, and Orlinsky (1989) and Saunders (1999Saunders ( , 2000. In each of these studies measures of the relationship variables as experienced on both sides were analyzed, even if they were not presented in the integrative way that is most consistent with the emergent definition above. Further to this, studies of mutuality include those that have looked at the real relationship such as Gelso et al. (2005) and Kelley, Gelso, Fuertes, Marmarosh, and Lanier (2010) when they consider the real relationship (Gelso, 2009) that is mutually experienced. However, in these studies of the real relationship, measures of realness were constructed as they are perceived from sometimes either one or both sides. This raises an important question as to whether the measures of mutuality can be taken as they are perceived, rather than actually thought to be present. That is, we have identified some studies that seem to measure mutuality as the emergence of coexperiencing while other studies have measured mutuality as the perception by client and/or therapist reports on coexperiencing of specific relationship variables.
There are some recent studies using novel statistical approaches to analyze dyadic research. For example, Kivlighan, Kline, Gelso, and Hill (2017) looked at the congruence for ratings of working alliance and real relationship scores and linked these to session evaluation scores. However, the analysis did not report on congruence between clients and therapists but rather looked at each individual within the dyad.
A recent study Rubel, Bar-Kalifa, Atzil-Slonim, Schmidt, and Lutz (2018) has looked at the congruence of therapeutic bond perceptions within therapeutic dyads and showed that when clients and therapists were in agreement on their bond scale ratings that higher bond ratings predicted lower levels of next-session symptoms. This approach to looking at within-dyad perceptions of the therapeutic relationship is similar to that used by Murphy and Cramer (2014). The study is, in our view, measuring the presence of mutuality and consequently suggests that therapeutic relationship research is advancing toward recognizing the concept of mutuality as a significant therapeutic relationship variable.
These studies, together with others considering mutual empathy, unconditional positive regard and genuineness (Murphy & Cramer, 2014), mutual caring (Halstead, Wagner, Vivero, & Ferkol, 2002), and mutual love (Kahane, 2001) are considered to provide the basis for an inquiry into the construct of mutuality as an emerging and potentially important therapeutic relationship variable that is as yet underresearched and not consistently articulated in the literature. Given the growing interest in the potential beneficial effects of mutuality in psychotherapy on therapeutic variables as described above, the authors set out to conduct a systematic review of the current literature in order to evaluate this relationship.

Research Questions
Specifically, the authors aimed to answer the following questions about mutuality in psychotherapy studies in relation to therapeutic variables, referring to both session quality and treatment effectiveness: 1. How is mutuality defined, understood, and/or measured? 2. What themes are found in qualitative studies?
3. What is the relationship between mutuality and therapeutic variables? (a) If based on client-perceived mutuality? (b) If based on therapist-perceived mutuality? (c) If only between based on the association between mutuality and treatment effectiveness? (d) If only based on the association between mutuality and session quality? 4. What potential variables moderate the associations?

Search Procedures
For the present study, the authors searched for qualitative and quantitative studies relating to bona fide psychotherapy. Bona fide psychotherapies are those carried out by trained professionals or those training to practice a given approach to psychotherapy. The search was targeted at studies that had identified a measure of mutuality (predictor variable) and also a measure of reported outcome or session quality (criterion variable). Potentially relevant published studies were found through a comprehensive search of four databases (PsycINFO, PsycARTICLES, SocINDEX, Academic Search Complete). Additionally, WorldCat was used to identify unpublished dissertations and theses. To search the databases, the authors used the following Boolean search protocol: "mutuality OR reciprocity OR 'mutual empathy' OR 'mutual affirmation' OR 'mutual support' OR 'mutual relatedness' OR 'real relationship' OR 'Therapeutic Bond Scale' AND outcome OR progress OR 'session quality' OR effectiveness AND counseling OR therapy OR psychotherapy." The authors acknowledge that there is a thin line between the terms used in the search above and other terms that were not included, which may have led to similar constructs (e.g., Mearns & Cooper, 2005). In total, the authors identified 505 articles through the four above-listed databases and 33 theses/ dissertations through WorldCat. The search was conducted between November 2014 and March 2015. Additionally, references of identified studies were searched to determine additional studies for inclusion. Only studies written in English were considered. Figure 1 provides a flow diagram of the search procedures.

Selection Procedures
The initial inclusion criteria were kept intentionally broad to allow for both qualitative and quantitative studies examining mutuality and psychotherapy outcome in adults in either individual or group interventions. For the purpose of the present selection, outcome referred to treatment effectiveness measured by symptom, session quality or progress, or overall quality of life measures. The authors read the titles of the identified articles, then moved onto abstracts and/or full text as needed to determine the acceptability of the studies for inclusion. After the initial screening of records, 35 studies were considered for the current study, but 14 were eliminated for methodological reasons, including lack of a clear measure of the predictor and/or criterion variable (e.g., use of the Therapeutic Bond Scale (TBS) only as a working alliance measure) or not being correlational in design (e.g., Tantillo & Sanftner, 2003), leaving 10 quantitative and 11 qualitative studies for inclusion in the present study. The 10 quantitative studies met the following inclusion criteria: (a) written in English, (b) provided sufficient data to compute effect sizes, (c) reported a measure of mutuality and session quality and/or treatment effectiveness, and (d) used either cross-sectional or longitudinal correlational design.

Sample of Studies
The final pool of studies to be analyzed included 21 studies, consisting of 10 correlational and 11 qualitative studies. All correlational studies were published and included a total of 1,071 participants. The average sample size was approximately 107 participants per study with a standard deviation of around 64 participants. Of the 10 correlational studies, one reported a change in R 2 as the test statistic for the primary analysis of mutuality, which was also the only  study to construct a single measure of mutuality incorporating both client and therapist perspectives (Murphy & Cramer, 2014). Most used separate measures from both therapist and client and some only one of the two perspectives to evaluate mutuality. Five of the qualitative studies were not published. There were 81 participants across the 11 qualitative studies with an average of about seven participants per study and a standard deviation of about five participants. The qualitative studies were all concerned with exploring either or both the therapist and client experience of being in a therapeutic relationship; however, they did not necessarily explicitly set out to explore experiences of mutuality as a number of studies presented mutuality as a research finding rather than the initial focus of the study.

Coding Procedures
For each of the 21 studies meeting our inclusion criteria, 18 variables, including year of publication, sample size and characteristics, instrumentation, and study design, were coded (Table 1 presents examples of variables consistently available across studies).
Two authors independently coded each study. For all studies, variations existed in what was coded for the variables by each reviewer. Typically, these differences were superficial and did not represent disagreement. In other words, the coders reported, accurate, but different, information. For example, for sample size, one coder may have emphasized the full sample while another emphasized the sample relevant for the effect size related to mutuality; or for sample age, one may have reported the range rather than impute the average age when only the range was given. For studies where there was conflicting information provided, the original coders and an additional coder reached consensus as to the best information to include. For the correlational studies, two coders reexamined the original articles along with the coding sheets to increase accuracy in the data entry process required to conduct the statistical analyses. Overall, coders showed high reliability on the variables coded for the analyses. For the eight variables for which a kappa coefficient was calculated, all yielded perfect reliability (kappa ϭ 1.00) except for sample size and dependent variable (kappa ϭ .84 and kappa ϭ .81, respectively), in which differences were resolved through joint discussion.

Data Analysis
Qualitative data analysis. Metasynthesis is a qualitative method used to analyze multiple qualitative studies identified through systematic review. Metasynthesis can be carried out in three different ways: first, it can be conducted to analyze multiple studies carried out by a single author over a period of time on a specific topic; second, it can be used to analyze the findings of studies from a number of researchers in a specific field; third, it may be utilized to quantitatively analyze the findings of a number of qualitative studies (Sandelowski et al., 1997). The approach in the current study was to analyze the findings of a number of research studies carried out by multiple research teams on the topic of mutuality in the therapeutic relationship. We analyzed the qualitative studies to identify the main findings in each study. These findings were then further analyzed as a process of synthesis, which led to the creation of six themes. A further cross check was carried out by a second researcher to verify that all findings of the original studies were subsumed into one of the six themes emergent from the synthesis.
Quantitative data analysis. To address Research Question (RQ) 3, the authors used the correlation coefficient r for effect size estimates. Accepted convention suggests the following concerning the interpretation of effect size estimates: values between 0.1 and 0.3 (small effect size), 0.4 and 0.5 (medium effect size), and greater than 0.5 (large effect size; Cohen, 1988). For the analyses, a combined study-level effect size was computed for studies reporting multiple outcomes by averaging the provided effect sizes (Berkeljon & Baldwin, 2009). Specifically, the authors generated one effect size for each study addressing RQ3(a) and RQ3(b), averaging effect sizes when more than one was given by the original study authors. Likewise, the effect sizes computed for research questions 3a and 3b (where both were found in the same study) were averaged to generate the overall study-level effect size in order to address RQ3. The authors addressed RQ3(c) and RQ3(d) by only including study-level findings that addressed those questions, again, following the same averaging procedures de- scribed above when more than one relevant result was provided in the original studies. Moreover, the authors corrected for measurement error attenuating effect size estimates, using independent and dependent variable reliability measures (i.e., Cronbach's alpha), utilizing the following formula: r/͌(rxx)(ryy). Reliability measures were averaged when more than one was given and not corrected for attenuation when none was given. The authors further corrected for sample size variation, by weighting each study proportional to its sample size, specifically to the inverse of the variance (Hedges & Olkin, 1985). The aggregate effect sizes across studies (weighted mean effect size) were computed using the random effects model as recommended by the National Research Council (1992). Additionally, homogeneity tests using the Q statistic were conducted to determine the potential for moderator variables. Researchers ran inverse variance weighted one-way analyses of variance for the analyses of categorical variables and inverse variance weighted regression analyses for continuous variables using the mixed-effects model. All analyses were conducted using meta-analysis macros for SPSS (Wilson, 2017).

Mutuality Definitions and Measurement
In relation to RQ1 mutuality was often defined differently in the qualitative studies but with greater similarity in the correlational studies. In many of the qualitative studies, mutuality was not measured or defined explicitly but emerged as an important finding within the study. In the correlational studies, six used the Real Relationship Inventory (RRI) or a derivative, which comes in therapist (Gelso et al., 2005) and client (Kelley et al., 2010) versions, or an extraction of the patient and therapist scores from the Comprehensive Scale of Psychotherapy Session Constructs, which also measures real relationship. Two studies used the TBS (Saunders et al., 1989), which included Reciprocal Empathy and Mutual Affirmation subscales, not just the Working Alliance subscale, which distinguishes the mutuality construct from working alliance. One study measured reciprocal intimacy on the Therapy Session Reports (Saunders, 1999). A final study used the Barrett-Lennard Relationship Inventory to measure the mutual experiencing of empathy, unconditional positive regard, and genuineness. To create a measure of mutuality, the interaction between client and therapist ratings was used to predict therapeutic progress (Murphy & Cramer, 2014). See Table 1 for mutuality measurements used by each study instrument and Table 2 for more information on the measurements such as reliabilities and number of items.
Consequently, we have proposed a working definition for mutuality within the psychotherapy relationship that is as follows: mutuality in psychotherapy refers to the coexperiencing by client and therapist of key therapeutic relationship variables, especially mutual empathy, positive regard, or a mutual perception of the relationship being "real." It refers to the mutual bond, self-disclosure, resonance, and liking between client and therapist as perceived by either or both parties.
Session evaluation was largely consistent across studies. Typically this was a one-or two-item measure of the quality of the session evaluated from client and/or therapist perspective. Outcome measures varied across nearly all studies but consisted of typical outcome measures, such as the Symptom Checklist-90-R (e.g., Marmarosh et al., 2009) or Clinical Outcome in Routine Evaluation-Outcome Measure (e.g., Murphy & Cramer, 2014).

Qualitative Themes
Six themes emerged from the 11 qualitative studies. Table 3 provides methodological information and an accounting of the themes.
Evidence of all six themes could be found in a majority of the studies. The researchers identified the six themes by selecting them from the major findings reported within the 11 qualitative studies. First, two researchers identified all the major themes reported in the 11 studies. Each of these was then separately considered in the context of the existing literature that had identified mutuality as a key therapeutic relationship construct. Using the literature, each individual finding was then determined to either be related to the construct of mutuality or was discarded. Next, a third researcher independently "cross-examined" the list of findings generated by the first researcher's analysis. The three researchers acted in an adversary role, much like Elliott (2002) has proposed within the hermeneutic single case efficacy design model. The researchers defended their decisions and the final list was agreed upon when the researchers had presented sufficient evidence to justify the inclusion of the theme and/or were checked through rereading of the studies. The themes follow.
Theme 1: Lack of mutuality; strategies of disconnection. Nonmutual relationships are often expected by clients and a cause of mental distress through a loss of agency (Gagerman, 2004;Murphy, Cramer, & Joseph, 2012;Tickle & Murphy, 2014). The expectation of nonmutual relationships stems from a feeling of not being worthy of connection (Gagerman, 2004), where being vulnerable (Binder, Moltu, Hummelsund, & Holgersen, 2011) leads to clients' strategies of disconnection (Tickle & Murphy, 2014). Strategies of disconnection are employed to maintain a nonmutual position within the relationship by not engaging fully in the relationship. This might, for example, involve withholding one's true emotional and inner world from the scrutiny and control of the other person, while presenting an appearance of being in the relationship (Miller, 1988). When a client perceives the therapist as intrusive, through unrequested self-disclosure (Wells, 1994), not recognizing client preference for personal distance (Halstead et al., 2002), or recognition of the client as an autonomous person, the client disconnects from the relationship. Etherington (2011) argues that such disconnection results in feeling ashamed and then internalizing the sense of shame. Feeling "pissed off," "humiliated," "scared," "offended," and "ashamed" are also some of the comments recorded by Wells (1994) in response to unsolicited therapists' disclosure. Therapists reported that being taskorientated with clients can trigger their own experience of disconnection from more emotional involvement in the relationship and, in turn, that of their client's (Halstead et al., 2002). An empathic attunement to the pace and rhythm of the client's need for connection is vital to avoid disconnections (Etherington, 2011) and this empathic attunement balances the wishedfor therapist's interest against the feared intrusion of adult power for young clients (Binder et al., 2011).
Theme 2: Cocreated relational process. Tickle and  suggest that strategies of connection are cocreated in the way that Gagerman (2004) argues that empathy is necessarily mutual. The client's and therapist's strategies of disconnection similarly are cocreated in the relationship. Mutuality is seen as a process (Tickle & Murphy, 2014) of collaborative cocreation of meaning making (Etherington, 2011) and as an emergent product of a sensitivity to the rhythmic flow of mutual relating (Halstead et al., 2002). Mutuality can also be a process that mediates other relationship variables (Binder et al., 2011), an intersubjective and relational knowing (Gagerman, 2004), and an experience that influences and is influenced by therapist's self-disclosure (Wells, 1994). Theme 3: Metacommunication; misunderstanding. Given that mutuality is, in part, viewed as a shared subjective experience (Tickle & Murphy, 2014), the relationship needs to exist in a shared meaning system (Gagerman, 2004). As the therapeutic relationship seems beset with misunderstanding (Mearns & Cooper, 2005), mutuality can be present when client and therapist engage in metacommunication (Rennie, 1998) about misunderstandings. Metacommunication is when a client and therapist talk about talk itself; this might refere to any element of an expression between the two, including the content, tone, focus, or perceived intentions. It is the client's perception of therapist intention which is paramount in experiencing the presence of therapeutic relationship conditions, including facilitative therapist self-disclosure (Wells, 1994). As a counterexample, the meaning of termination processes intended by therapists for collaboration and to release client agency can be experienced by clients as "you must end" (Etherington, 2011). Metacommunication about how the client experiences the therapist's interventions as too direct or how they feel swamped by negative emotions were found to increase trust in the relationship (Binder et al., 2011). Halstead et al. (2002) found that clients also sometimes initiate metacommunication, increasing the therapist's awareness of themselves in the client's world.
Theme 4: Therapist congruence; being real. According to Rogers and Truax (1967), therapist's congruence includes the therapist being fully themselves and a willingness to be known and to be present to the client; it was this realness that makes metacommunication possi- ble. Clients' misperceptions about the termination review process stemmed from their belief that therapists had a hidden-role agenda (Etherington, 2011). Studies described how the therapist's experience of being real is made available to the client (Tickle &  and that willingness to be known reveals a person behind the role without a hidden agenda (Binder et al., 2011), letting clients be most valuably impacted (Etherington, 2011) and safe to connect (Binder et al., 2011). Wells (1994 found that the therapists' revelations of their realness through self disclosure could equalize the relationship, but without the sense of role boundaries, this mutuality was burdensome. A cautionary note is that some self-disclosures can be unhelpful to the therapeutic process and great care and thought should always be given to making self-disclosures. Theme 5: Mutual impact; client agency. A willingness to be present and authentic involves the possibility of being impacted by the client (Tickle & Murphy, 2014), revealing the therapist as equally human and also capable of growth. Binder et al. (2011) found that teenagers wanted their therapists to show their realness. Shared humanity rather than merging is a recognition of each other's autonomy, which helps clients feel safe while intimately connected (Binder et al., 2011). This mutual impact promotes client agency as they perceive the impact they have on the therapist (Binder et al., 2011;Tickle & Murphy, 2014) and the therapeutic process itself (Binder et al., 2011;Etherington, 2011). This reconnects the client's sense of being worthy of connection and relational efficacy, which is then experienced outside of the therapeutic relationship (Gagerman, 2004;Tickle & Murphy, 2014).
Theme 6: Asymmetric role power; boundaries. Self-disclosure has sometimes been assumed to mean mutuality. Wells' (1994) study reported that perceived mutuality was a boundary intrusion where roles become symmetrical through therapist self-disclosure. However, all other studies explicitly describe how the asymmetric nature of the therapeutic relationship does not limit the potential for mutual connection. Clients and therapists can experience mutuality specifically because the care that therapists offer clients does not have to be reciprocated by clients, that is, clients are not indebted to their therapists and clients have no requirement to reciprocate a caring attitude. Some therapists report an experience of care or compassion or any other variable with no strings attached and it is this unconditionality that creates the experience of mutuality. Clients can trust the therapist to maintain professional boundaries (Halstead et al., 2002). Mutuality is not about being in equal need of care or being equally responsible to care for the other. Rather, it is the purpose of the relationship, which is the same for both, that is, the client's well-being. Mutuality arises out of the therapist's capacity to maintain the integrity of this asymmetry of roles and needs. This role difference relates to the second and the third necessary and sufficient conditions of therapy (Rogers, 1957). Mutuality sometimes describes the joint focus and mutual attention that is on the client and the client may not have to fear the therapist taking up space (Binder et al., 2011;Tickle & Murphy, 2014). That is, the asymmetry in roles does not mitigate against the mutual creation of the therapeutic conditions; rather, it facilitates it. When there is the possibility of sharing a mutual subjective experience of the other, the relationship is experienced as equal from inside the relationship. If the therapist disconnects as a real person without self-acceptance of their persistent subjective experience, incongruently hiding behind the role of therapist, this nonmutual relating risks the client also disconnecting, thus putting the client into the isolation of nonmutual relating.

Overall Association Between Mutuality and Therapeutic Variables
Analysis of the 10 correlational studies showed a statistically significant large overall effect size for the relationship between mutuality and session evaluation and/or outcome (RQ3), r ϭ .51, 95% CI [.036, 0.66], p Ͻ .001). Additionally, a test for homogeneity of effect size distribution showed evidence of heterogeneity, Q(9) ϭ 49.25, p Ͻ .001, suggesting the presence of moderators (moderator analysis to follow below).
In addition to looking at the overall relationship, separate meta-analyses were run investigating studies addressing client-perceived mutuality (RQ3[a]), therapist-perceived mutuality (RQ3[b]), relationship of mutuality specifically to session quality (RQ3[c]), and specifically to treatment outcome (RQ3[d]).
Client-perceived mutuality. Eight studies included measurements of mutuality from the client's perspective, which showed a medium effect size, r ϭ .46, 95% CI [0.30, 0.62], p Ͻ .001. A test of homogeneity, again, demonstrated significant heterogeneity, Q(7) ϭ 37.00, p Ͻ .001, suggesting the presence of study-level variables accounting for between-study effect size variability.
Therapist-perceived mutuality (RQ3[b]). The authors followed the same procedures and found similar results for RQ3(b) as the previous question. Specifically, six studies included measurements of mutuality from the therapist's perspective, showing a medium effect size, r ϭ .48, 95% CI [0.34, 0.62], p Ͻ .001. While much smaller than previous analyses, significant heterogeneity was found for this study also, Q(5) ϭ 12.24, p ϭ .031, suggesting the presence of moderators.
Session quality (RQ3[c]). Five studies included evaluations of session quality, showing a large effect size, r ϭ .70, 95% CI [0.43, 0.97], p Ͻ .001). A test of homogeneity of effect size distribution, again, demonstrated significant heterogeneity, Q(4) ϭ 49.65, p Ͻ .001, indicating that effect size variability between studies may be accounted for by study characteristics.

Treatment outcome (RQ3[d]).
All 10 studies included one or more measures of treatment outcome, which showed a medium effect size, r ϭ .42, 95% CI [0.27, 0.58], p Ͻ .001. Evidence of significant heterogeneity was also found, Q(9) ϭ 49.94, p Ͻ .001, suggesting the presence of moderators affecting the association between mutuality and outcome.
The authors ran an additional analysis for RQ3(d; mutuality and treatment outcome) as one of the studies (Saunders, 1999) showed that almost all of the positive findings in that study were found when clients were seen for 2-8 sessions rather than other lengths of time. In both the working alliance (Hovarth et al., 2011) and mutuality literature, there is an emphasis on allied viewpoints by client and therapist early in therapy. The overall effect size for treatment effectiveness reported in Saunders' (1999) study was the outlier among the group, where their specific finding related to early perceptions of mutuality is more in line with the rest of the literature. In this analysis, the overall effect size increased only minimally, again, showing a medium effect size, r ϭ .47, 95% CI [0.36, 0.59], p Ͻ .001. Evidence of significant heterogeneity was, again, also found, Q(9) ϭ 29.33, p ϭ .001, indicating the presence of moderators in this second analysis of the relationship of mutuality with treatment outcome. However, the confidence interval and the Q statistic are both much smaller, suggesting a closer alignment between the studies when only the early therapy measurements are used from Saunders (1999).

Potential Moderators
Because of the evidence for heterogeneity found in all of the above analyses, implying the presence of moderator variables, the authors conducted moderator analyses to identify studylevel variables influencing the relationship between the predictor and criterion variables. However, all variables assessed-client mean age, percent female, percent White, publication year, instrumentation-with only two exceptions, showed no evidence of moderating the relationship between mutuality and therapeutic variables. Out of the 30 moderator analyses (5 for each of the 6 meta-analyses run), the two variables that reached significance were: (a) client mean age on the association between clientperceived mutuality and therapeutic variables (B ϭ 0.02, Z ϭ 2.15, p ϭ .031), and (b) publication year on the association between therapist-perceived mutuality and therapeutic variables (B ϭ Ϫ0.02, Z ϭ Ϫ2.01, p ϭ .045). In other words, for (a) the effect sizes were larger when researchers measured mutuality as perceived by the clients in the older clients, and for (b) the effect sizes were larger when researchers measured mutuality as perceived by the therapist in the older studies.

Discussion
This meta-analysis and metasynthesis suggests promising results and encourages more research on the role of mutuality in the field of psychotherapy. Both qualitative and correlational studies show consistently positive relationships between mutuality and therapeutic variables. While analyzed from a relatively small literature with limitations as explained below, this systematic review does suggest that mutuality might someday be considered for the list of empirically validated elements of relational "evidenced-based responsiveness" as synthesized by Norcross (2011) and colleagues.
At the same time, there are many limitations. First, there is a relatively small but growing literature on mutuality in relation to the size of the literature on psychotherapy or on other process constructs (e.g., empathy or working alliance). Second, very few of the 21 studies measured mutuality from both sides of the relationship rather than a perception of reciprocity and mutuality between the client and therapist from one side.
Additionally, until now there was no one agreed-upon definition of mutuality and therefore no clear sense about what research should constitute studies on mutuality. In other words, the central concept of mutuality as introduced in the introduction and then defined and described in the studies or those potentially excluded from this review through the keyword combinations shows inconsistencies. Nevertheless a goal of this systematic review was to offer clarity in the definition, which, though modest, has somewhat been obtained. Mutuality can be defined as an emergent property of the therapeutic relationship between the client and the therapist that is the result of coexperiencing person-to-person qualities such as mutual empathy, positive regard, realness, affirmation, reciprocal affirmation, bond, and real relationship.
With regard to the meta-analytic procedures, the "criterion" or outcome-related variables included in relation to mutuality varied from study to study. Combining statistics with such disparate concepts results in a more impressionistic view of the size of association between mutuality and session quality or outcome rather than one where tests of homogeneity or subsequent moderator analyses are more revealing. This is apparent in the studies using therapeutic bond or real relationship scales where, overall, it might be said that the items can give an impression of either the therapist's or client's sense of mutuality in the relationship; they did not always look at the congruence between these two perspectives. Consequently, we recommend that future studies of therapeutic relationship variables should always collect data from both the client's and therapist's perspectives and calculate the level of congruency between these points of view. New emerging statistical methods for dyadic research make this relatively straightforward and can be used to compute scores for mutuality and its association with outcomes.
Likewise, the session quality measure of one or two items is weaker than the longer, validated outcome measures. A main benefit of this meta-analysis is to provide an overview of a literature to suggest potential areas where additional research is warranted and encourage more sophisticated analyses to infer mean effect sizes with less heterogenous concepts and findings.
Nevertheless, overall, the association of mutuality with therapeutic variables as defined in this meta-analysis appears to be substantial. In particular, the effect size on session quality is noteworthy. An effect size this large (r ϭ .70) suggests that approximately half of the variance of session quality may be attributable to perceptions of mutuality. The very idea of session quality appears to be wrapped up in this sense of reciprocity and the attunement in the assessment of realness between both client and therapist.
The lack of homogeneity in the effect size distribution between the studies is significant. This is partly a function of the small sample of studies meta-analyzed in the present study (N ϭ 10), but also may be a function of difficulty extracting accurate information from the studies to assess for moderation. In the case of age, gender, and ethnicity variables, estimates were entered when exact information was not provided. For example, "typically White and female" was entered as 85% but may in actuality mean only a mode of White, suggesting a percentage less than 50%. Likewise, other factors were not available from the majority of studies or were available with nondescript language (e.g., diverse therapy orientations) and thus estimation was not attempted, given the high likelihood of inaccuracy. Therefore, it is likely that other variables are responsible for the moderation. Clear reporting of information that could be used for moderator analysis in future metaanalyses will be important in future research.
Qualitative studies have highlighted several key features of the mutuality construct. They suggest that mutuality is least likely when clients and therapists disconnect from one another in therapeutic relationships. Mutuality is made possible through a cocreated relational process that involves metacommunication, especially concerning misunderstandings and disconnec-tions; this can be achieved through the therapist and client being real and present to one another. Therapists may be able to be fully present (but not indiscriminately or excessively selfdisclosing) by allowing clients to see the impact they have on the therapist, giving clients a chance to experience their agency. All this appears possible and, in part, attributable to the asymmetrical structure of the therapeutic relationship.
At the same time, there are some limitations of the qualitative studies reviewed. First, like with the measures in the correlational studies, the definition of mutuality varies in the qualitative studies, providing a limitation to the transferability even as it did to the generalizability of the quantitative studies. Additionally, in all but one study, mutuality was not the main focus of the research and instead was a finding within a study with another focus. The results of this systematic review suggest that researchers should focus on mutuality intentionally as the subject of qualitative study. Also, while small sample sizes are not considered an issue for qualitative research per se, the authors suggest that further metasynthesis research should be carried out in the future after there has been a larger sample of qualitative studies conducted to increase our capacity to generalize further and make firmer recommendations for practice.
These findings and those from the metaanalysis present some practice implications. For example, therapists and clients seem able to experience the therapeutic relationship as transcending the negative aspects of role power difference as beneficial, suggesting that, similar to Simonds and Spokes' (2017) study, some degree of disciplined self-disclosure of immediacy or personal nature may be warranted. Further qualitative research is needed to extend our understanding of experiences of mutuality related to role power, agency, and client impacts on the therapist.
Likewise, as the effect size of therapistperceived mutuality (r ϭ .48) is nearly identical to that of client-perceived mutuality (r ϭ .46) with therapeutic variables, yet is typically smaller than client perceptions with other empirically validated relational variables like empathy or working alliance (e.g., Cornelius-White, 2002;Norcross, 2011), therapist perceptions of mutuality may be better to use than these conventional measures when client feedback is not formally sought or available. In practice, an implication may be that a therapist's own informal evaluation of their mutuality may be a more reliable indicator of how well therapy is going than their own informal perception of empathy or positive regard for the client.

Conclusion
To conclude, the authors suggest that mutuality is an important variable worthy of further consideration within the field of psychotherapy research. This study has demonstrated that, as a relationship variable, mutuality has shown moderate-to-large effect sizes in its contribution to therapeutic variables in bona fide psychotherapy studies. Researchers and practitioners should consider including measures of mutuality in their process outcome research and practice to increase the body of literature in this area and potentially benefit client welfare. Mutuality offers a new direction for psychotherapy relationship research-one in which role power structures are reconsidered and client agency and impacts can be foregrounded. Further quantitative and qualitative research is needed to prioritize this promising field of study.