There's a number of ways to try and avoid response bias - the most obvious would be to (a) work out roughly how many people are listening to AM radio at all, (b

There's a number of ways to try and avoid response bias - the most obvious would be to (a) work out roughly how many people are listening to AM radio at all, (b) go find some representative sample of people who are listening to AM radio and ask them what they're listening to, (c) scale your results proportionately. Nielsen specialises in this sort of research.