This study will test whether making plans of action for situations that trigger anger is effective in reducing anger and aggression in adolescents with behavioural problems.

Participants will be assigned to three groups: a control group and two experimental groups. Participants assigned to the experimental groups will receive either one general anger trigger or a list of specific anger triggers in addition to a list of strategies that can be used to manage anger. Participants will be instructed to link the triggers with the strategies, thus creating action plans with an if-then structure. Participants assigned to the control group will receive the same lists. However, they will receive different instructions which will ask them to select separately the most encountered triggers and the most useful strategies.

It is expected that making plans will reduce the anger and aggression of participants. It is also expected that the reduction will be larger for participants with low violent intentions, low callous-unemotional traits and low impulsivity.

This study aims to test whether aggression in adolescents can be reduced using implementation intentions. This technique involves identifying critical situations that might trigger the unwanted behaviour and making a plan to engage in an alternative course of action when those triggers are present. Implementation intentions formation has been found to be effective in reducing smoking uptake and consultations for emergency contraception in this age group.

A randomised controlled trial will be conducted. Students from 11 to 17 years old attending special schools for children with behavioural problems or referred for anger issues will be randomised to three conditions: attention control, specific triggers and general trigger. Participants in the attention control and specific triggers conditions will receive the same Volitional Help Sheet (VHS), composed of a list of 10 anger triggers (if statements) and a list of 10 anger management strategies (then statements). However, the instructions will be different. Participants in the attention control condition will be instructed to select the triggers that they encounter more often and the strategies that they consider more useful. In contrast, participants in the specific triggers condition will have to link each specific trigger that they encounter often with a specific strategy, thus creating action plans with an if-then structure. Participants assigned to the general trigger condition will also be instructed to make plans linking the trigger with a strategy, but the VHS they will receive will contain only one general trigger: "If I am getting angry".

Measures of anger and reactive physical aggression, reactive relational aggression, proactive physical aggression and proactive relation aggression will be collected at baseline, one month and six months after intervention. In addition, measures of negative urgency, violent intentions and callous-unemotional traits will be collected once.

Both intervention groups are hypothesised to reduce anger and reactive aggression in comparison to the control group. It is hypothesised that the reduction in anger acts as a mediator for the reduction of reactive aggression. It is also hypothesised that violent intentions, callous-unemotional traits and negative urgency act as moderators of effectiveness.

The effects of the intervention will be analysed on anger, on the aggression total score and on the four types of aggression separately using Multivariate Analysis of Covariance. In order to test whether the effect of the interventions on reactive aggression at follow-up is mediated by anger, a mediation analysis will be conducted using the PROCESS macro (Hayes, 2013). In the analysis, the intervention conditions (dummy coded) will be entered as the independent variable, anger as the mediator variable, reactive aggression at follow-up as the dependent variable, and reactive aggression at baseline as the covariate. The analysis will asses the strength of the direct and indirect effects of the interventions on reactive aggression via anger. In order to test whether the effect of the interventions on aggression at follow-up is moderated by violent intentions, callous-unemotional traits and negative urgency, a moderated regression analysis will be conducted using the PROCESS macro (Hayes, 2013). In the analysis, the intervention conditions (dummy coded) will be entered as the independent variable, violent intentions, callous-unemotional traits and negative urgency as the moderator variables, aggression at follow-up as the dependent variable, and aggression at baseline as covariates.

Inclusion Criteria:

• Attending a school for children with Social, Emotional and Mental Health needs.
• Being referred by school staff due to anger issues or behavioural problems.

Exclusion Criteria:

• Having severe or profound learning disabilities.

The control condition is active and has a similar format to the intervention conditions to avoid participants knowing which condition they have been assigned to. Outcomes are assessed by self-report.