Marianne Welmers

Chapter 2 46 Table 5 Results of Moderator Analyses based on Three-level Mixed Effects Models for Split Alliances and Treatment Outcome Moderator # Studies # ES Mean r (SE) Study Quality 5 17 .839 (.282)* Sample Characteristics Problem type 5 17 drug abuse youth 2 8 -.137 (.093) eating disorders youth 1 2 .179 (.157) mixed problem types 2 7 .308 (.105)* Average age youth 5 17 .970 (.898) % Male youth 5 17 .327 (.274) % Male adult 3 13 .046 (.175) % Noncaucasian 4 15 -.106 (.215) Treatment characteristics Treatment model 5 17 FBT 1 2 .177 (.307) FFT 2 4 .179 (.220) other 2 11 .018 (.211) Alliance characteristics Alliance rater 5 17 observer 4 12 .062 (.127) youth + parent averaged or added 1 5 .274 (.244) Outcome characteristics Outcome domain 5 17 youth symptom severity or functioning 2 7 .229 (.171) retention 3 12 .024 (.139) Outcome rater 5 17 youth 1 5 .274 (.385) therapist 1 2 .022 (.390) objectified measure 2 8 .046 (.281) youth and objectified measure combined 1 2 .177 (.423) Note . ES = effect size; CI = confidence interval; σ 2 level2 = variance between effect sizes (within studies); σ 2 level3 = variance between effect sizes (between studies); a. The number of studies reflects the number of independent samples. * p < .05, ** p < .01, *** p < .001.