Hello list,

I am troubled by a problem I face in my thesis: In a difference-in-differences approach, can I use an unbalanced (asymmetric) number of pre and post periods to the treatment? That is, if I were using a t-5 through t+5 years in my estimation, can I instead use t-1 through t+5? What are the implications? I fear including all the way back to t-5 adds a lot of irrelevant noise into my sample measure.

Thank you.

Eben