Marianne Welmers

General Discussion 141 CHAPTER 6 weaker alliances. These findings imply that some therapists by nature may be more successful in building strong alliances with family members than others. However, it is important to note here that specific preferred behaviors associated with personality traits are not necessarily unchangeable: the extent to which personality traits are activated can vary across situations (Hoekstra & De Fruyt, 2014). An important next step to shed more light on the effect of therapist personality is to investigate what (behavioral) mechanisms explain the association between therapist personality and alliance. Nevertheless, training therapists to become more aware of typical conscientious or extravert behaviors and to apply alternative behaviorsmore beneficial to the alliance, may support highly extraverted and conscientious therapists in being more attuned in their communication with family members. Finally, observational measures of therapists’ behavioral contributions to the alliance (Chapter 3) indicated that therapists were mainly observed to show engagement and emotional connection behaviors, promoting alliances with individual family members. Their investment in family members’ safety and the family’s shared sense of purpose, specific systemic aspects of the alliance, were scarce. This may be due to the complexity of building systemic alliances: multiple family members with different behaviors are simultaneously involved, requiring more meta-reflection on the therapeutic process and careful guidance of within-family interactions during a treatment session (Escudero & Friedlander, 2017; Rober, 2017). However, the finding that therapists reported similar differences in family members’ alliances as compared to observer- and family members’ self-reports (Chapter 4), indicate therapists’ abilities to detect discrepancies between their alliances with different family members. This is an important starting point for another systemic aspect of alliances in family treatment: balancing different family members’ alliances over the course of treatment. Directions for Future Research Is It Really the Alliance? Causality Considerations Although several methodological features strengthen the interpretability of the research findings of this dissertation (e.g., the use of multi-level models, multiple informants, and multiple time points), it is important to note that the study design does not provide evidence for a causal relation between the alliance and treatment outcome. One reason for this is that the presented studies only provided data on end-of-treatment (meta-analysis in Chapter 2) or follow up (Chapters 2 and 5) measures of outcome. This raises the question whether a strong alliance predicts more problem improvement, or rather, a stronger alliance is an effect of problem improvement during treatment (Crits-