Based on conversation analysis (CA) of video-recorded therapy sessions, the article explicates a particular interactional project of positively evaluating client-reported behavior in psychotherapy. The analysis focuses on the therapist's actions that convey a positive evaluation of client-reported behavior that represents therapeutic progress. First, the data analysis revealed three components that constitute the evaluation project: discourse marker, compliment, and account. Second, the article shows that participants orient towards the observed evaluation project, both as a unified whole and as a combination of discrete and separate interactional turns. The article suggests that this evaluation project functions as a tool for achieving the institutional goal of reinforcing therapeutically desired behaviors. The empirical findings are discussed in relation to the Stocks of Interactional Knowledge, described in handbooks on dialectical behavior therapy (the specific setting in which the data were collected). (Professional-client interaction, adolescent, psychotherapy, evaluation, complimenting, conversation analysis)*

Background. Disasters are believed to have large effects on the mental health of adolescents but the lack of prospective pre- and post-disaster data on affected and control populations have limited our knowledge on the validity of these claims. We examined the medium-term, 12 months' effects of a severe disaster on the mental health of adolescents, and compared them to effects after 5 months.

Method. A café fire in The Netherlands injured 250 adolescents and killed 14. We obtained data 15 months before and 12 months after the disaster about behavioural and emotional problems (using the Youth Self-Report) and substance misuse, in 124 students of an affected school of whom 31 were present at the fire (response 77·5%) and 830 other students (56·4%); mean age at baseline, 13·8 years.

Results. We found differences between students from the affected school and others for excessive use of alcohol (odds ratio 3·42, 95% confidence interval 2·00–5·85, p<0·0001), but not for behavioural and emotional problems and use of other substances. Effects had decreased compared to those after 5 months.

Conclusions. In the long run, the effects of disaster decrease regarding self-reported behavioural and emotional problems, but they remain regarding alcohol misuse among those present at the disaster, and their peers.