1) Policy should be based on scientific evidence -> negative effects of peer aggregation (deviant groups) should be acknowledged (not just mean effects discussed - who got worse / better / stayed the same)

2) Identify validated intervention principles that avoid peer aggregation (family-based, multi-systemic, integrate settings, etc.)

3) Reduce harm when aggregation cannot be avoided (minimize time in such groups, reduce unstructured interactions, have well-trained facilitators)