I don't always formally sample participants But when I do I make sure that my population is fully enumerated, randomly sample from that population, randomly assign those individuals to my treatments, note who dropped out, replace those individuals rando