Marianne Welmers

Chapter 5 122 For alliances with youth we found that observer rated alliances at T1 were strongly and (trend-) significantly associated with youth internalizing ( r = .578, p = .067), externalizing ( r = .647, p = . 042), and total behavior problems ( r = .689, p = .029) after controlling for baseline CBCL scores, indicating that stronger early treatment observed alliances were associated with less improvement of youth behavior problems from baseline to 18 months follow up. At T2, this association was in the opposite direction and large yet not significant for internalizing ( r = -.932, p = .118) and total problems ( r = -.733, p = .238). Associations of self-reported youth alliances at T1 with CBCL scores were small and not significant, except for the partial correlation with youth externalizing behavior problems ( r = -.613, p = .072), indicating a trend towards more improvement of externalizing youth behavior problems 18 months after treatment when children and adolescents perceived their early alliances with the therapist as stronger. At T2, youth self-reports of the alliance were largely associated with improvement of youth behavior problems. Here, only the association with youth total behavior problems was significant ( r = -.945, p = .027). It should be noted, however, that the number of youth self-reports on alliance at T2 was extremely small ( n = 5). Unbalanced Alliances Results from zero order and partial correlation tests between unbalanced alliances, shared sense of purpose, and youth behavior problems at follow up are reported in Table 5 and Table 6 respectively. At T1, both tests showed that unbalanced observed alliances were significantly associated with youth problems at follow up. In contrast to our hypothesis, results indicated that larger differences at the starting phase of treatment were associated with less youth externalizing problems ( r = -.411, p = .032) and total problems ( r = -.415, p = .031), also when controlling for baseline youth externalizing problems ( r = -.564, p = .005). The association between unbalanced therapist reported alliances and youth total behavior problems at follow up was moderate and significant, after controlling for baseline CBCL scores it was trend-significant ( r = -.369, p = .055). These findings indicate that when differences between familymembers in their observed early treatment alliances with the therapist were larger, there was more improvement of youth externalizing and total behavior problems at 18 months post-treatment. At T2, associations between unbalanced observed alliances and youth behavior problems at follow up after controlling for baseline were in the expected direction, indicating that smaller differences between family members in their observed mid-treatment alliances with the therapist were associated with more improvement of youth behavior problems. However, these correlations at T2 were rather small and failed to reach significance.