Dear authors,

I have read with interest your well-designed study. I understand the aim of your study is the test the effects of normative feedback in an applied setting rather than disentangle the conditions under which it works in a laboratory setting. Nevertheless, if I have not misread some of the work done, it would be interesting and relevant perhaps to address the following questions :

- this feedback is expected to work among those with perceived peer drinking behavior being discrepant from actual peer drinking behavior. So instead of focusing on high risk drinkers, why not focus the analysis on people with the largest discrepancies? That's where the strategy should have most impact (we may expect a correlation between high risk drinkers and this discrepancy, but that is unlikely to approach 1)
- providing social reference is expected to have a positive impact if the reference is in the desired direction, but it may also have an opposite impact if it is in the undesired direction (I.e. you reveal that moderate drinkers drink less than what their peers do). Have you examined whether your intervention has had a positive effects on those with discrepancies in one direction but a negative effects on those with discrepancies in the other?
- if the principle works, it only works after successful manipulation. Have you examined whether your manipulation has had an effect in discrepancies between perceived peer drinking behavior and actual peer drinking behavior pre-post? If not, it says something about the succes of your manipulation rather than the potency of the mechanism.

Thanks for considering these thoughts.

Best wishes,
Marijn de Bruin