their socioeconomic background. The original model was specified using backwards selection. Due to the gender difference in self-harming behavior (15), secondary analyses were performed to investigate the interaction effects between gender and the other significant exposure variables.
To investigate whether the relationship between early childhood hyperactivity difficulties and DSH is explained through peer difficulties, a mediation analysis was performed. For a variable to be suitable as a mediator
Therese A. Evald,
Scandinavian Journal of Child and Adolescent Psychiatry and Psychology , 176–188