"I think the essential thing if you want to be a good statistician, as opposed to being a mathematician, is to talk to people and find out what they're doing and why they're doing it."