The “Good” and the “Bad”: Identity-Based Transference and Countertransference in Co-Led Group Therapy

Co-leadership, in which two clinicians jointly facilitate a therapy group, is a widely used and valued format in group psychotherapy (Rabin, 1967; Yalom & Leszcz, 2005). Advantages of co-leadership include the ability to have dual points of view within the group, divide clinical roles, model healthy interpersonal relationships for members, and the ability for a novice leader to co-lead with and learn from a more experienced leader (Miles & Kivlighan, 2010; Yalom & Leszcz, 2005). Yet co-leadership also introduces a distinct clinical challenge: when co-leaders differ in race, gender, culture, or other social identities, those differences become powerful organizers of members’ own transference. Consequently, how co-leaders are paired, and how they prepare to work together across differences, warrants careful consideration.

These differences in co-leader perception can become consequential when members experience one leader as exceptionally competent, trustworthy or caring, while the other as deficient or untrustworthy. Such idealization and devaluation can function as, “intrapsychic defenses, interpersonal patterns, and collective forces shaping group dynamics” (van Paassen, 2026). Importantly, these processes do not occur in isolation but are embedded within the broader cultural and intersectional contexts that shape how group members perceive, and relate to, their co-leaders (Dalal, 2022; van Paassen, 2026).

One way in which cultural values shape the expression of idealization-devaluation is through expectations surrounding power and authority. “Power-distance” refers to the extent to which unequal distributions of power are accepted and expected within a society, with higher power-distance cultures placing greater emphasis on hierarchy and deference to authority, and lower power-distance cultures emphasizing equality and the ability to question authority (Hofstede et al., 2010). Consequently, members from higher power-distance cultures may idealize leaders more fully and with less questioning, casting the leader as an all-knowing expert. Devaluation of the group leader, on the other hand, may be expressed more passively through absences or somatic complaints (Hofstede et al., 2010; van Paassen, 2026). Conversely, members from lower power-distance cultures may idealize more conditionally, regard them as skilled peers, and express devaluation more directly (van Paassen, 2026). These patterns, however, describe broader cultural tendencies rather than predictions about any individual group member.

Identity as a Lens for Transference

Comas-Díaz and Jacobsen (1991) introduced the concept of ethnocultural transference to describe how a client’s reactions to a therapist are shaped not only by their own personal history but also by their therapist’s perceived race, ethnicity, and culture, and the client’s own lived experience of those identities within the larger society. Group therapy intensifies this process. Because a therapy group functions as a social microcosm, the dynamics of power, privilege, and oppression that members encounter outside the group tend to be reenacted within it (Abernethy, 2024).

Culture, however, is only one dimension. An intersectional perspective (Crenshaw, 1989) recognizes that each member occupies a distinct dimension across multiple social identities, including but not limited to race, gender, class, sexual orientation, national origin. Blackwell (2023)locates these processes in the social unconscious, arguing that colonial templates, in which the colonizer is idealized and the colonized is devalued, remain active in diverse group settings. For example, members may unconsciously attribute greater competence, credibility, or authority to a White co-leader, while questioning or overlooking similar contributions from a co-leader of color, thereby reproducing historically entrenched hierarchies of racial power. Conversely, members whose experiences of marginalization have fostered distrust of dominant institutions may perceive a White co-leader as representative of oppressive authority while idealizing a co-leader of color as someone more likely to understand their lived experiences. Nayak (2022)adds the concept of “compound injuries,” referring to the cumulative and intersecting effects of marginalization across multiple social identities. For instance, a woman of color whose experiences of racial and gender-based discrimination has shaped her expectations of being dismissed or overlooked may interpret a group leader’s failure to acknowledge her contributions through the lens of these earlier experiences, thereby reactivating feelings of exclusion. In response, she may perceive the leader as dismissive or untrustworthy, while idealizing a co-leader whom she experiences as more attentive or affirming.

Splitting within Co-Leadership

McWilliams (2011) described splitting as a defense in which people unconsciously sort others and experiences into all-good and all-bad categories. Idealization and devaluation are central manifestations of splitting, a primitive defense that protects the self from the anxiety associated with ambivalence (Klein 1946; Kernberg, 1975, van Paassen, 2026). Within the group setting, this can appear on an individual level (i.e., co-leaders and/or other group members are idealized and devalued), interpersonal level (when members form opposing subgroups that align with one leader while rejecting the other, or perceive certain fellow members as all-good and other members as all-bad) and collective level (the group itself as idealized or devalued). Consequently, the therapeutic task is for group leaders to help members integrate these polarized representations by exploring the anxieties and interpersonal pressures that make it difficult to hold contradictory feelings towards the same individual. For example, when a group member or co-leader is consistently idealized, other members may hesitate to express frustration or disagreement for fear of disrupting the status-quo or risking disapproval from the group. To alleviate the split, leaders can help the group examine what makes certain feelings difficult to express and model the capacity to tolerate disagreement without becoming defensive or withdrawing from the relationship. Over time, this may help members recognize that appreciation and disappointment can coexist within the same relationship, subsequently strengthening group members’ capacity to tolerate ambivalence (van Paassen, 2026).

Countertransference within the Co-Leader Alliance

Members do not solely maintain the split between good and bad objects. Countertransference, or the therapist’s emotions and behavioral reactions shaped by their own history, identities, and vulnerabilities, often determines whether a split is understood or enacted. In co-led groups, one leader may readily come to hold the “good” position and the other the “bad,” mirroring the split within the group itself (Wuhrmann, 1999). The idealized leader faces two common pitfalls: colluding with admiration and/or deflection. While the former reinforces dependency, the latter can leave members feeling rejected and undermine their trust altogether. Meanwhile, the devalued leader may become defensive or withdrawn, which can escalate conflict or abandon the group ultimately.

Atieno Okech and Kline (2006) found that, as part of the group therapy process, co-leaders carry ongoing internal questions about themselves, their co-leaders, and the group. Chang-Caffaro and Caffaro (2018) noted that these internal questions and reactions are shaped by ethnic, cultural, gender, religious, and socioeconomic differences that frequently operate outside awareness, making leaders less inclined to discuss differences at precisely the moments when it matters most clinically. They argue that transference and countertransference arising from identity-based differences need to be worked through toward a relationship of equal power and status if equity and social justice are to become group norms.

Clinical Vignette

The following vignette is drawn from my experience as a trainee. Details about group members and colleagues have been omitted, altered, or generalized to protect confidentiality.

During my doctoral training, I co-led several interpersonal process groups. As a South Asian cisgender woman in my early 20s, I was often paired with fellow trainees who were White cisgender men ranging in age from their mid-20s to late 30s. Across groups, a consistent pattern emerged. Men in the group tended to direct their questions and comments to my co-leader. When I posed a question, they often answered while making eye contact with him, as though he had asked it. When we offered similar process interpretations within minutes of each other, members readily took up the interpretation only when it came from him, while mine passed through the room unacknowledged.

In other groups, a different pattern unfolded. Several Black men responded to my co-leader’s interventions with visible frustration, at times challenging or dismissing his observations outright. At the same time, they turned toward me in search of alliance by seeking my eye contact, inviting my agreement, or framing their remarks as though I would understand their experiences in a way he could not.

In retrospect, each group had organized a layered split around our intersecting identities. Along the axis of gender, male members’ transference positioned my co-leader as the credible authority and me as peripheral, likely reenacting broader social patterns about whose voice carries weight. Along the axis of race, some members positioned him as a representative of White authority they had reason to distrust, and me as a fellow person of color who might recognize their experiences of marginalization. Depending on which identity was most salient to a given member, one of us was idealized and the other devalued.

Our countertransference at times reinforced this split. When my interventions went unacknowledged, I felt invisible and began to question my competence. This experience is consistent with projective identification, in which the devalued leader comes to feel the very inadequacy projected onto them. Gradually, I offered fewer interpretations and deferred to my co-leader, which confirmed the group’s perception of me as secondary. At the same time, I felt drawn toward members who sought alliance with me. Being seen was a relief, which made it tempting to accept the role of the “good object” rather than explore what that alliance meant for them. Our status as trainees added a further layer: each of us was wary of appearing less competent to the other and to our supervisors, which made the pattern harder to name.

When we brought these observations to supervision and began debriefing more deliberately, we were able to approach the pattern as clinical material rather than as a reflection of our abilities. With our supervisors’ encouragement, we began naming the dynamic in session. The initial response was a striking silence, and at times members expressed frustration or distress. Remaining with that discomfort, however, allowed the group to explore the broader relational and social patterns underlying members’ reactions. Naming the dynamic did not resolve questions of race, power, and privilege. Instead, it transformed them from something enacted into something the group could discuss openly, and members gradually engaged with both of us more fully.

Conclusion and Implications for Practice

Identity-based transference and countertransference are expected byproducts of co-leadership in group therapy settings. Understanding these processes can offer clinical insight into members’ early attachment patterns and the collective forces shaping how they navigate vulnerability, power, and belonging. The goal is not to eliminate idealization and devaluation, but to help the group arrive at a “good enough” perception of its leaders, one that holds both strengths and limitations, fosters self-acceptance, and builds resilience in the face of disappointment (van Paassen, 2026). Although no single approach will fit every group, the following practices offer a flexible framework for this work:

1. Talk about identity before group begins: Co-leaders can benefit from explicit pre-group conversations about their identities and positionalities, including whether those overlap or diverge, and how members may perceive them.

2. Name differences early within the group: During the initial sessions, co-leaders can acknowledge visible differences both between themselves and among members and invite reflection on how these may shape the group. Proactively preparing the group to address cultural ruptures establishes a norm that identity-based transference can be spoken about rather than acted out.

3. Debrief regularly: A 10-15 minute debrief after every session allows co-leaders to track emerging dynamics and countertransferences. Because successful countertransference management is associated with better treatment outcomes (Hayes et al., 2018), the debrief is where much of that management happens.

4. Help members integrate polarized feelings: Once a split becomes visible, co-leaders can invite curiosity rather than correction. Helping members tolerate mixed feelings toward both leaders moves them away from rigid all-or-none thinking and builds a capacity for integration that can extend beyond the group setting.

5. Model repair within the co-leader alliance: When co-leaders acknowledge the dynamic openly and support one another in the room, they demonstrate that difference and connection can coexist (Chang-Caffaro & Caffaro, 2018). Rupture repair is linked to better outcomes (Eubanks et al., 2018), and a stance of cultural humility helps both leaders remain open rather than defensive.

6. Seek supervision and consultation: Identity-based transference often activates co-leaders’ own histories, and trainees may hesitate to raise it for fear of appearing less competent. Supervisors can lower this barrier by asking directly about identity dynamics in the group and within the co-leader pairing, and by being thoughtful about how pairings are assigned and supported.