Yikes, that sounds like a great way to bias your data away from controversial opinions about weed. That would be like taking an exit poll of only people wearing lots of political apparel.