i think rossin is one of them. c-span: why would that be? >> guest: i am not exactly sure probably is methodology. i'm not exactly sure about that. but in general, in terms of methodology the like live interview polls. so if a person uses a kind of rolling you don't know who is on the phone, so they don't like those sorts of polls. they don't like internet polling. and these are all valid reasons for not doing this. so there are some that you can kind of look at for reference and then try to find someone who has called live and see if that works out, but there's some that we just don't use. c-span: is very stable base that you will not use? and other words if they only had 500 people call on the phone, does that trip certain things versus 1200 people that they talk to in person? >> guest: we kind of one over 1,000. you want -- and when you look at who is called, hopefully they've broken down by kind of race, age, ideology. and you want it to be somewhat representative. it isn't always perfect, but close to a representative of the population