Hester van Eeren

| Chapter 5 5 | 108 Subsample: Youth without a court order Of the 697 adolescents in the study sample, 370 (168 MST; 202 FFT) had no court order before beginning the intervention. Of adolescents who had completed MST, 61.5% were male and 90.3% born in the Netherlands. For FFT, 52.3% of the adolescents were male and 97.4% born in the Netherlands (for an extensive comparison of the treatment arms, see Table IV in Supplemental Material). Comparing the treatment groups within this subsample on baseline characteristics showed significant differences in age, externalizing and total behavioral problems measured with the CBCL, parenting stress, country of birth, level of education, previous treatment, engagement in school or work, and previous police contact (Table IV in Supplemental Material). Balance assessment Before the PS application, the largest imbalances—standardized bias higher than the accepted .25—were found for age, externalizing problems on the CBCL, level of education, previous treatment, and having had police contact before treatment (Table IV in Supplemental Material). After PS application, balance was found when all covariates except for the total score of behavioral problems measured by the CBCL were selected for the PS estimation. Before inspecting balance, 11 MST and 29 FFT cases were removed for which there was no overlap on the PS scores. Except for the standardized bias of the level of education of the adolescent, values of the standardized bias after PS application were lower than .25 (Table IV in Supplemental Material). Values for the standardized bias for the missing indicator variables were also lower than .25 (Table V in Supplemental Material). The variance ratios of the continuous variables, except for parenting stress, were within the boundaries defined by the 2.5 th and 97.5 th percentiles of the F-distribution in the weighted sample. Thus, except for parenting stress, balance can be assumed given these values (Table VI in Supplemental Material). The five-number summaries show roughly equally distributed continuous variables between the treatment groups (Table VI in Supplemental Material). In the complete case sample of 210 adolescents without court orders (104 MST; 106 FFT), balance was achieved when the variables of age, internalizing and externalizing problems measured with the CBCL, parenting stress, gender, country of birth, previous treatment, engagement in school or work, court order, police contacts, and employment status of the primary caregiver were included in the PS estimation. Except for the level of education of the adolescent, all standardized bias values were lower than .25 after PS application. Because balance was inspected in the sample with overlapping PS scores, 6 MST and 5 FFT cases were removed from this sample. Analysis of treatment effect In the subsample of adolescents without a court order, MST and FFT differed significantly in terms of externalizing problem behavior. Multisystemic Therapy resulted in lower scores on externalizing problem behavior than FFT (CBCL: -3.24; 95% CI -5.97 – -.39, YSR: -3.33; 95% CI -5.81 – -.86), with a medium effect size of d = 0.32 and d = 0.34, respectively. The differences (RR and RD) between MST and FFT on the three secondary outcomes were insignificant (Table 3).